Tutorial 5: Custom Augmentations

In this tutorial, we will train a model on chest X-ray images in a self-supervised manner. In self-supervised learning, X-ray images can pose some problems: They are often more than eight bits deep which makes them incompatible with certain standard torchvision transforms such as, for example, random-resized cropping. Additionally, some augmentations which are often used in self-supervised learning are ineffective on X-ray images. For example, applying color jitter to an X-ray image with a single color channel does not make sense.

We will show how to address these problems and how to train a ResNet-18 with MoCo on a set of 16-bit X-ray images in TIFF format.

The original dataset this tutorial is based on can be found on Kaggle. These images are in the DICOM format. For simplicity and efficiency reasons, we randomly selected ~4000 images from the above dataset, resized them such that the maximum of the width and height of each image is no larger than 512, and converted them to the 16-bit TIFF format. To do so, we used ImageMagick which is preinstalled on most Linux systems.

mogrify -path path/to/new/dataset -resize 512x512 -format tiff "*.dicom"

Imports

Import the Python frameworks we need for this tutorial.

import copy
import os

import matplotlib.pyplot as plt
import numpy as np
import pandas
import pytorch_lightning as pl
import torch
import torch.nn as nn
import torchvision
from PIL import Image
from sklearn.neighbors import NearestNeighbors
from sklearn.preprocessing import normalize

from lightly.data import LightlyDataset
from lightly.loss import NTXentLoss
from lightly.models.modules.heads import MoCoProjectionHead
from lightly.models.utils import (
    batch_shuffle,
    batch_unshuffle,
    deactivate_requires_grad,
    update_momentum,
)
from lightly.transforms.multi_view_transform import MultiViewTransform

Configuration

Let’s set the configuration parameters for our experiments.

We will use eight workers to fetch the data from disc and a batch size of 128. The input size of the images is set to 128. With these settings, the training requires 2.5GB of GPU memory.

num_workers = 8
batch_size = 128
input_size = 128
seed = 1
max_epochs = 50

Let’s set the seed for our experiments.

pl.seed_everything(seed)

1

Set the path to our dataset.

path_to_data = "/datasets/vinbigdata/train_small"

Setup custom data augmentations

The key to working with 16-bit X-ray images is to convert them to 8-bit images which are compatible with the torchvision augmentations without creating harmful artifacts. A good way to do so, is to use histogram normalization as described in this paper about Covid-19 prognosis.

Let’s write an augmentation, which takes as input a numpy array with 16-bit input depth and returns a histogram normalized 8-bit PIL image.

class HistogramNormalize:
    """Performs histogram normalization on numpy array and returns 8-bit image.

    Code was taken and adapted from Facebook:
    https://github.com/facebookresearch/CovidPrognosis

    """

    def __init__(self, number_bins: int = 256):
        self.number_bins = number_bins

    def __call__(self, image: np.array) -> Image:
        # Get the image histogram.
        image_histogram, bins = np.histogram(
            image.flatten(), self.number_bins, density=True
        )
        cdf = image_histogram.cumsum()  # cumulative distribution function
        cdf = 255 * cdf / cdf[-1]  # normalize

        # Use linear interpolation of cdf to find new pixel values.
        image_equalized = np.interp(image.flatten(), bins[:-1], cdf)
        return Image.fromarray(image_equalized.reshape(image.shape))

Since we can’t use color jitter on X-ray images, let’s replace it and add some Gaussian noise instead. It’s easiest to apply this after the image has been converted to a PyTorch tensor.

class GaussianNoise:
    """Applies random Gaussian noise to a tensor.

    The intensity of the noise is dependent on the mean of the pixel values.
    See https://arxiv.org/pdf/2101.04909.pdf for more information.

    """

    def __call__(self, sample: torch.Tensor) -> torch.Tensor:
        mu = sample.mean()
        snr = np.random.randint(low=4, high=8)
        sigma = mu / snr
        noise = torch.normal(torch.zeros(sample.shape), sigma)
        return sample + noise

Now that we have implemented our custom augmentations, we can combine them with available augmentations from the torchvision library to get to the same set of augmentations as used in the aforementioned paper. Make sure, that the first augmentation is the histogram normalization, and that the Gaussian noise is applied after converting the image to a tensor.

Note that we also transform the image from grayscale to RGB by simply repeating the single color channel three times. The reason for this is that our ResNet expects a three color channel input. This step can be skipped if a different backbone network is used.

# Compose the custom augmentations with available augmentations.
view_transform = torchvision.transforms.Compose(
    [
        HistogramNormalize(),
        torchvision.transforms.Grayscale(num_output_channels=3),
        torchvision.transforms.RandomResizedCrop(size=input_size, scale=(0.2, 1.0)),
        torchvision.transforms.RandomHorizontalFlip(p=0.5),
        torchvision.transforms.RandomVerticalFlip(p=0.5),
        torchvision.transforms.GaussianBlur(21),
        torchvision.transforms.ToTensor(),
        GaussianNoise(),
    ]
)

# Create a multiview transform that returns two different augmentations of each image.
transform = MultiViewTransform(transforms=[view_transform, view_transform])

Let’s take a look at what our augmentation pipeline does to an image! We plot the original image on the left and two random augmentations on the right.

example_image_name = "55e8e3db7309febee415515d06418171.tiff"
example_image_path = os.path.join(path_to_data, example_image_name)
example_image = np.array(Image.open(example_image_path))

# Torch transform returns a 3 x W x H image, we only show one color channel.
augmented_image_1 = view_transform(example_image).numpy()[0]
augmented_image_2 = view_transform(example_image).numpy()[0]

fig, axs = plt.subplots(1, 3)

axs[0].imshow(example_image)
axs[0].set_axis_off()
axs[0].set_title("Original Image")

axs[1].imshow(augmented_image_1)
axs[1].set_axis_off()

axs[2].imshow(augmented_image_2)
axs[2].set_axis_off()

Setup dataset and dataloader

We create a dataset which loads the images in the input directory. Since the input images are 16 bits deep, we need to overwrite the image loader such that it doesn’t convert the images to RGB (and hence to 8-bit) automatically.

Note

The LightlyDataset uses a torchvision dataset underneath, which in turn uses an image loader which transforms the input image to an 8-bit RGB image. If a 16-bit grayscale image is loaded that way, all pixel values above 255 are simply clamped. Therefore, we overwrite the default image loader with our custom one.

def tiff_loader(f):
    """Loads a 16-bit tiff image and returns it as a numpy array."""
    with open(f, "rb") as f:
        image = Image.open(f)
        return np.array(image)


# Create the dataset with the custom transform and overwrite the image loader.
dataset_train = LightlyDataset(input_dir=path_to_data, transform=transform)
dataset_train.dataset.loader = tiff_loader

# Setup the dataloader for training.
dataloader_train = torch.utils.data.DataLoader(
    dataset_train,
    batch_size=batch_size,
    shuffle=True,
    drop_last=True,
    num_workers=num_workers,
)

Create the MoCo model

Using the building blocks provided by lightly we can write our MoCo model. We implement it as a PyTorch Lightning module. For the criterion, we use the NTXentLoss which should always be used with MoCo.

MoCo also requires a memory bank - we set its size to 4096 which is approximately the size of the input dataset. The temperature parameter of the loss is set to 0.1. This smoothens the cross entropy term in the loss function.

The choice of the optimizer is left to the user. Here, we go with simple stochastic gradient descent with momentum.

class MoCoModel(pl.LightningModule):
    def __init__(self):
        super().__init__()

        # Create a ResNet backbone and remove the classification head.
        resnet = torchvision.models.resnet18()
        self.backbone = nn.Sequential(
            *list(resnet.children())[:-1],
        )

        # The backbone has output dimension 512 which also defines the size of
        # the hidden dimension. We select 128 for the output dimension.
        self.projection_head = MoCoProjectionHead(512, 512, 128)

        # Add the momentum network.
        self.backbone_momentum = copy.deepcopy(self.backbone)
        self.projection_head_momentum = copy.deepcopy(self.projection_head)
        deactivate_requires_grad(self.backbone_momentum)
        deactivate_requires_grad(self.projection_head_momentum)

        # Create the loss function with memory bank.
        self.criterion = NTXentLoss(temperature=0.1, memory_bank_size=(4096, 128))

    def training_step(self, batch, batch_idx):
        (x_q, x_k), _, _ = batch

        # Momentum update
        update_momentum(self.backbone, self.backbone_momentum, 0.99)
        update_momentum(self.projection_head, self.projection_head_momentum, 0.99)

        # Get the queries.
        q = self.backbone(x_q).flatten(start_dim=1)
        q = self.projection_head(q)

        # Get the keys.
        k, shuffle = batch_shuffle(x_k)
        k = self.backbone_momentum(k).flatten(start_dim=1)
        k = self.projection_head_momentum(k)
        k = batch_unshuffle(k, shuffle)

        loss = self.criterion(q, k)
        self.log("train_loss_ssl", loss)
        return loss

    def configure_optimizers(self):
        # Use SGD optimizer with momentum and weight decay.
        optim = torch.optim.SGD(
            self.parameters(),
            lr=0.1,
            momentum=0.9,
            weight_decay=1e-4,
        )
        scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optim, max_epochs)
        return [optim], [scheduler]

Train MoCo with custom augmentations

Training the self-supervised model is now very easy. We can create a new MoCoModel instance and pass it to the PyTorch Lightning trainer.

model = MoCoModel()

trainer = pl.Trainer(
    max_epochs=max_epochs,
    devices=1,
    accelerator="gpu",
    precision=16,
)
trainer.fit(model, dataloader_train)

/datasets/actions-runner/core_gpu_runner_01/hostedtoolcache/Python/3.10.13/x64/lib/python3.10/site-packages/lightning_fabric/plugins/environments/slurm.py:165: PossibleUserWarning: The `srun` command is available on your system but is not used. HINT: If your intention is to run Lightning on SLURM, prepend your python command with `srun` like so: srun python /datasets/actions-runner/core_gpu_runner_01/_work/li ...
  rank_zero_warn(
/datasets/actions-runner/core_gpu_runner_01/hostedtoolcache/Python/3.10.13/x64/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py:1609: PossibleUserWarning: The number of training batches (32) is smaller than the logging interval Trainer(log_every_n_steps=50). Set a lower value for log_every_n_steps if you want to see logs for the training epoch.
  rank_zero_warn(

Training: 0it [00:00, ?it/s]
Training:   0%|          | 0/32 [00:00<?, ?it/s]
Epoch 0:   0%|          | 0/32 [00:00<?, ?it/s]
Epoch 0:   3%|▎         | 1/32 [00:03<01:58,  3.82s/it]
Epoch 0:   3%|▎         | 1/32 [00:03<01:58,  3.83s/it, loss=0.523, v_num=2]
Epoch 0:   6%|▋         | 2/32 [00:03<00:58,  1.93s/it, loss=0.523, v_num=2]
Epoch 0:   6%|▋         | 2/32 [00:03<00:58,  1.94s/it, loss=2.69, v_num=2]
Epoch 0:   9%|▉         | 3/32 [00:03<00:37,  1.31s/it, loss=2.69, v_num=2]
Epoch 0:   9%|▉         | 3/32 [00:03<00:37,  1.31s/it, loss=3.63, v_num=2]
Epoch 0:  12%|█▎        | 4/32 [00:03<00:27,  1.01it/s, loss=3.63, v_num=2]
Epoch 0:  12%|█▎        | 4/32 [00:03<00:27,  1.01it/s, loss=4.19, v_num=2]
Epoch 0:  16%|█▌        | 5/32 [00:04<00:21,  1.25it/s, loss=4.19, v_num=2]
Epoch 0:  16%|█▌        | 5/32 [00:04<00:21,  1.25it/s, loss=4.57, v_num=2]
Epoch 0:  19%|█▉        | 6/32 [00:04<00:17,  1.48it/s, loss=4.57, v_num=2]
Epoch 0:  19%|█▉        | 6/32 [00:04<00:17,  1.48it/s, loss=4.85, v_num=2]
Epoch 0:  22%|██▏       | 7/32 [00:04<00:14,  1.71it/s, loss=4.85, v_num=2]
Epoch 0:  22%|██▏       | 7/32 [00:04<00:14,  1.71it/s, loss=5.06, v_num=2]
Epoch 0:  25%|██▌       | 8/32 [00:04<00:12,  1.94it/s, loss=5.06, v_num=2]
Epoch 0:  25%|██▌       | 8/32 [00:04<00:12,  1.94it/s, loss=5.22, v_num=2]
Epoch 0:  28%|██▊       | 9/32 [00:06<00:17,  1.32it/s, loss=5.22, v_num=2]
Epoch 0:  28%|██▊       | 9/32 [00:06<00:17,  1.32it/s, loss=5.34, v_num=2]
Epoch 0:  31%|███▏      | 10/32 [00:06<00:15,  1.46it/s, loss=5.34, v_num=2]
Epoch 0:  31%|███▏      | 10/32 [00:06<00:15,  1.46it/s, loss=5.43, v_num=2]
Epoch 0:  34%|███▍      | 11/32 [00:06<00:13,  1.60it/s, loss=5.43, v_num=2]
Epoch 0:  34%|███▍      | 11/32 [00:06<00:13,  1.60it/s, loss=5.48, v_num=2]
Epoch 0:  38%|███▊      | 12/32 [00:06<00:11,  1.73it/s, loss=5.48, v_num=2]
Epoch 0:  38%|███▊      | 12/32 [00:06<00:11,  1.73it/s, loss=5.52, v_num=2]
Epoch 0:  41%|████      | 13/32 [00:06<00:10,  1.86it/s, loss=5.52, v_num=2]
Epoch 0:  41%|████      | 13/32 [00:06<00:10,  1.86it/s, loss=5.54, v_num=2]
Epoch 0:  44%|████▍     | 14/32 [00:07<00:09,  1.99it/s, loss=5.54, v_num=2]
Epoch 0:  44%|████▍     | 14/32 [00:07<00:09,  1.99it/s, loss=5.55, v_num=2]
Epoch 0:  47%|████▋     | 15/32 [00:07<00:08,  2.12it/s, loss=5.55, v_num=2]
Epoch 0:  47%|████▋     | 15/32 [00:07<00:08,  2.12it/s, loss=5.56, v_num=2]
Epoch 0:  50%|█████     | 16/32 [00:07<00:07,  2.25it/s, loss=5.56, v_num=2]
Epoch 0:  50%|█████     | 16/32 [00:07<00:07,  2.25it/s, loss=5.56, v_num=2]
Epoch 0:  53%|█████▎    | 17/32 [00:09<00:08,  1.73it/s, loss=5.56, v_num=2]
Epoch 0:  53%|█████▎    | 17/32 [00:09<00:08,  1.73it/s, loss=5.56, v_num=2]
Epoch 0:  56%|█████▋    | 18/32 [00:09<00:07,  1.82it/s, loss=5.56, v_num=2]
Epoch 0:  56%|█████▋    | 18/32 [00:09<00:07,  1.82it/s, loss=5.56, v_num=2]
Epoch 0:  59%|█████▉    | 19/32 [00:09<00:06,  1.91it/s, loss=5.56, v_num=2]
Epoch 0:  59%|█████▉    | 19/32 [00:09<00:06,  1.91it/s, loss=5.56, v_num=2]
Epoch 0:  62%|██████▎   | 20/32 [00:09<00:05,  2.00it/s, loss=5.56, v_num=2]
Epoch 0:  62%|██████▎   | 20/32 [00:09<00:05,  2.00it/s, loss=5.56, v_num=2]
Epoch 0:  66%|██████▌   | 21/32 [00:10<00:05,  2.10it/s, loss=5.56, v_num=2]
Epoch 0:  66%|██████▌   | 21/32 [00:10<00:05,  2.10it/s, loss=5.81, v_num=2]
Epoch 0:  69%|██████▉   | 22/32 [00:10<00:04,  2.18it/s, loss=5.81, v_num=2]
Epoch 0:  69%|██████▉   | 22/32 [00:10<00:04,  2.18it/s, loss=5.85, v_num=2]
Epoch 0:  72%|███████▏  | 23/32 [00:10<00:03,  2.27it/s, loss=5.85, v_num=2]
Epoch 0:  72%|███████▏  | 23/32 [00:10<00:03,  2.27it/s, loss=5.85, v_num=2]
Epoch 0:  75%|███████▌  | 24/32 [00:10<00:03,  2.36it/s, loss=5.85, v_num=2]
Epoch 0:  75%|███████▌  | 24/32 [00:10<00:03,  2.36it/s, loss=5.84, v_num=2]
Epoch 0:  78%|███████▊  | 25/32 [00:12<00:03,  1.94it/s, loss=5.84, v_num=2]
Epoch 0:  78%|███████▊  | 25/32 [00:12<00:03,  1.94it/s, loss=5.82, v_num=2]
Epoch 0:  81%|████████▏ | 26/32 [00:12<00:02,  2.01it/s, loss=5.82, v_num=2]
Epoch 0:  81%|████████▏ | 26/32 [00:12<00:02,  2.01it/s, loss=5.79, v_num=2]
Epoch 0:  84%|████████▍ | 27/32 [00:13<00:02,  2.08it/s, loss=5.79, v_num=2]
Epoch 0:  84%|████████▍ | 27/32 [00:13<00:02,  2.08it/s, loss=5.76, v_num=2]
Epoch 0:  88%|████████▊ | 28/32 [00:13<00:01,  2.15it/s, loss=5.76, v_num=2]
Epoch 0:  88%|████████▊ | 28/32 [00:13<00:01,  2.15it/s, loss=5.73, v_num=2]
Epoch 0:  91%|█████████ | 29/32 [00:13<00:01,  2.22it/s, loss=5.73, v_num=2]
Epoch 0:  91%|█████████ | 29/32 [00:13<00:01,  2.22it/s, loss=5.71, v_num=2]
Epoch 0:  94%|█████████▍| 30/32 [00:13<00:00,  2.28it/s, loss=5.71, v_num=2]
Epoch 0:  94%|█████████▍| 30/32 [00:13<00:00,  2.28it/s, loss=5.68, v_num=2]
Epoch 0:  97%|█████████▋| 31/32 [00:13<00:00,  2.35it/s, loss=5.68, v_num=2]
Epoch 0:  97%|█████████▋| 31/32 [00:13<00:00,  2.35it/s, loss=5.67, v_num=2]
Epoch 0: 100%|██████████| 32/32 [00:13<00:00,  2.42it/s, loss=5.67, v_num=2]
Epoch 0: 100%|██████████| 32/32 [00:13<00:00,  2.42it/s, loss=5.66, v_num=2]
Epoch 0: 100%|██████████| 32/32 [00:13<00:00,  2.42it/s, loss=5.66, v_num=2]
Epoch 0:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.66, v_num=2]
Epoch 1:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.66, v_num=2]
Epoch 1:   3%|▎         | 1/32 [00:03<01:53,  3.66s/it, loss=5.66, v_num=2]
Epoch 1:   3%|▎         | 1/32 [00:03<01:53,  3.66s/it, loss=5.65, v_num=2]
Epoch 1:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=5.65, v_num=2]
Epoch 1:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=5.66, v_num=2]
Epoch 1:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=5.66, v_num=2]
Epoch 1:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=5.66, v_num=2]
Epoch 1:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=5.66, v_num=2]
Epoch 1:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=5.67, v_num=2]
Epoch 1:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=5.67, v_num=2]
Epoch 1:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=5.68, v_num=2]
Epoch 1:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=5.68, v_num=2]
Epoch 1:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=5.69, v_num=2]
Epoch 1:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=5.69, v_num=2]
Epoch 1:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=5.7, v_num=2]
Epoch 1:  25%|██▌       | 8/32 [00:03<00:11,  2.00it/s, loss=5.7, v_num=2]
Epoch 1:  25%|██▌       | 8/32 [00:03<00:11,  2.00it/s, loss=5.7, v_num=2]
Epoch 1:  28%|██▊       | 9/32 [00:06<00:16,  1.39it/s, loss=5.7, v_num=2]
Epoch 1:  28%|██▊       | 9/32 [00:06<00:16,  1.39it/s, loss=5.71, v_num=2]
Epoch 1:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=5.71, v_num=2]
Epoch 1:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=5.71, v_num=2]
Epoch 1:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=5.71, v_num=2]
Epoch 1:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=5.72, v_num=2]
Epoch 1:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=5.72, v_num=2]
Epoch 1:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=5.72, v_num=2]
Epoch 1:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=5.72, v_num=2]
Epoch 1:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=5.72, v_num=2]
Epoch 1:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=5.72, v_num=2]
Epoch 1:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=5.72, v_num=2]
Epoch 1:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=5.72, v_num=2]
Epoch 1:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=5.72, v_num=2]
Epoch 1:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=5.72, v_num=2]
Epoch 1:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=5.72, v_num=2]
Epoch 1:  53%|█████▎    | 17/32 [00:09<00:08,  1.82it/s, loss=5.72, v_num=2]
Epoch 1:  53%|█████▎    | 17/32 [00:09<00:08,  1.82it/s, loss=5.71, v_num=2]
Epoch 1:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=5.71, v_num=2]
Epoch 1:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=5.71, v_num=2]
Epoch 1:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=5.71, v_num=2]
Epoch 1:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=5.71, v_num=2]
Epoch 1:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=5.71, v_num=2]
Epoch 1:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=5.71, v_num=2]
Epoch 1:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=5.71, v_num=2]
Epoch 1:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=5.71, v_num=2]
Epoch 1:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=5.71, v_num=2]
Epoch 1:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=5.7, v_num=2]
Epoch 1:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=5.7, v_num=2]
Epoch 1:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=5.7, v_num=2]
Epoch 1:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=5.7, v_num=2]
Epoch 1:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=5.7, v_num=2]
Epoch 1:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=5.7, v_num=2]
Epoch 1:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=5.71, v_num=2]
Epoch 1:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=5.71, v_num=2]
Epoch 1:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=5.71, v_num=2]
Epoch 1:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=5.71, v_num=2]
Epoch 1:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=5.71, v_num=2]
Epoch 1:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=5.71, v_num=2]
Epoch 1:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=5.71, v_num=2]
Epoch 1:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=5.71, v_num=2]
Epoch 1:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=5.71, v_num=2]
Epoch 1:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=5.71, v_num=2]
Epoch 1:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=5.72, v_num=2]
Epoch 1:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=5.72, v_num=2]
Epoch 1:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=5.72, v_num=2]
Epoch 1: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=5.72, v_num=2]
Epoch 1: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=5.73, v_num=2]
Epoch 1: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=5.73, v_num=2]
Epoch 1:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.73, v_num=2]
Epoch 2:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.73, v_num=2]
Epoch 2:   3%|▎         | 1/32 [00:03<01:56,  3.74s/it, loss=5.73, v_num=2]
Epoch 2:   3%|▎         | 1/32 [00:03<01:56,  3.74s/it, loss=5.73, v_num=2]
Epoch 2:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=5.73, v_num=2]
Epoch 2:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=5.74, v_num=2]
Epoch 2:   9%|▉         | 3/32 [00:03<00:37,  1.28s/it, loss=5.74, v_num=2]
Epoch 2:   9%|▉         | 3/32 [00:03<00:37,  1.28s/it, loss=5.75, v_num=2]
Epoch 2:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=5.75, v_num=2]
Epoch 2:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=5.75, v_num=2]
Epoch 2:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=5.75, v_num=2]
Epoch 2:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=5.76, v_num=2]
Epoch 2:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=5.76, v_num=2]
Epoch 2:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=5.77, v_num=2]
Epoch 2:  22%|██▏       | 7/32 [00:04<00:14,  1.75it/s, loss=5.77, v_num=2]
Epoch 2:  22%|██▏       | 7/32 [00:04<00:14,  1.75it/s, loss=5.78, v_num=2]
Epoch 2:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=5.78, v_num=2]
Epoch 2:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=5.78, v_num=2]
Epoch 2:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=5.78, v_num=2]
Epoch 2:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=5.79, v_num=2]
Epoch 2:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=5.79, v_num=2]
Epoch 2:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=5.8, v_num=2]
Epoch 2:  34%|███▍      | 11/32 [00:06<00:12,  1.63it/s, loss=5.8, v_num=2]
Epoch 2:  34%|███▍      | 11/32 [00:06<00:12,  1.63it/s, loss=5.81, v_num=2]
Epoch 2:  38%|███▊      | 12/32 [00:06<00:11,  1.77it/s, loss=5.81, v_num=2]
Epoch 2:  38%|███▊      | 12/32 [00:06<00:11,  1.77it/s, loss=5.82, v_num=2]
Epoch 2:  41%|████      | 13/32 [00:06<00:09,  1.91it/s, loss=5.82, v_num=2]
Epoch 2:  41%|████      | 13/32 [00:06<00:09,  1.91it/s, loss=5.83, v_num=2]
Epoch 2:  44%|████▍     | 14/32 [00:06<00:08,  2.04it/s, loss=5.83, v_num=2]
Epoch 2:  44%|████▍     | 14/32 [00:06<00:08,  2.04it/s, loss=5.83, v_num=2]
Epoch 2:  47%|████▋     | 15/32 [00:06<00:07,  2.17it/s, loss=5.83, v_num=2]
Epoch 2:  47%|████▋     | 15/32 [00:06<00:07,  2.17it/s, loss=5.84, v_num=2]
Epoch 2:  50%|█████     | 16/32 [00:06<00:06,  2.30it/s, loss=5.84, v_num=2]
Epoch 2:  50%|█████     | 16/32 [00:06<00:06,  2.30it/s, loss=5.85, v_num=2]
Epoch 2:  53%|█████▎    | 17/32 [00:09<00:08,  1.78it/s, loss=5.85, v_num=2]
Epoch 2:  53%|█████▎    | 17/32 [00:09<00:08,  1.78it/s, loss=5.86, v_num=2]
Epoch 2:  56%|█████▋    | 18/32 [00:09<00:07,  1.88it/s, loss=5.86, v_num=2]
Epoch 2:  56%|█████▋    | 18/32 [00:09<00:07,  1.88it/s, loss=5.87, v_num=2]
Epoch 2:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=5.87, v_num=2]
Epoch 2:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=5.88, v_num=2]
Epoch 2:  62%|██████▎   | 20/32 [00:09<00:05,  2.07it/s, loss=5.88, v_num=2]
Epoch 2:  62%|██████▎   | 20/32 [00:09<00:05,  2.07it/s, loss=5.89, v_num=2]
Epoch 2:  66%|██████▌   | 21/32 [00:09<00:05,  2.16it/s, loss=5.89, v_num=2]
Epoch 2:  66%|██████▌   | 21/32 [00:09<00:05,  2.16it/s, loss=5.89, v_num=2]
Epoch 2:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=5.89, v_num=2]
Epoch 2:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=5.9, v_num=2]
Epoch 2:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=5.9, v_num=2]
Epoch 2:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=5.91, v_num=2]
Epoch 2:  75%|███████▌  | 24/32 [00:09<00:03,  2.44it/s, loss=5.91, v_num=2]
Epoch 2:  75%|███████▌  | 24/32 [00:09<00:03,  2.44it/s, loss=5.92, v_num=2]
Epoch 2:  78%|███████▊  | 25/32 [00:12<00:03,  2.02it/s, loss=5.92, v_num=2]
Epoch 2:  78%|███████▊  | 25/32 [00:12<00:03,  2.02it/s, loss=5.93, v_num=2]
Epoch 2:  81%|████████▏ | 26/32 [00:12<00:02,  2.09it/s, loss=5.93, v_num=2]
Epoch 2:  81%|████████▏ | 26/32 [00:12<00:02,  2.09it/s, loss=5.93, v_num=2]
Epoch 2:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=5.93, v_num=2]
Epoch 2:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=5.94, v_num=2]
Epoch 2:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=5.94, v_num=2]
Epoch 2:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=5.95, v_num=2]
Epoch 2:  91%|█████████ | 29/32 [00:12<00:01,  2.31it/s, loss=5.95, v_num=2]
Epoch 2:  91%|█████████ | 29/32 [00:12<00:01,  2.31it/s, loss=5.95, v_num=2]
Epoch 2:  94%|█████████▍| 30/32 [00:12<00:00,  2.38it/s, loss=5.95, v_num=2]
Epoch 2:  94%|█████████▍| 30/32 [00:12<00:00,  2.38it/s, loss=5.96, v_num=2]
Epoch 2:  97%|█████████▋| 31/32 [00:12<00:00,  2.45it/s, loss=5.96, v_num=2]
Epoch 2:  97%|█████████▋| 31/32 [00:12<00:00,  2.45it/s, loss=5.97, v_num=2]
Epoch 2: 100%|██████████| 32/32 [00:12<00:00,  2.52it/s, loss=5.97, v_num=2]
Epoch 2: 100%|██████████| 32/32 [00:12<00:00,  2.52it/s, loss=5.98, v_num=2]
Epoch 2: 100%|██████████| 32/32 [00:12<00:00,  2.52it/s, loss=5.98, v_num=2]
Epoch 2:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.98, v_num=2]
Epoch 3:   0%|          | 0/32 [00:00<?, ?it/s, loss=5.98, v_num=2]
Epoch 3:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=5.98, v_num=2]
Epoch 3:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=5.98, v_num=2]
Epoch 3:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=5.98, v_num=2]
Epoch 3:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=5.99, v_num=2]
Epoch 3:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=5.99, v_num=2]
Epoch 3:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=5.99, v_num=2]
Epoch 3:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=5.99, v_num=2]
Epoch 3:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6, v_num=2]
Epoch 3:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6, v_num=2]
Epoch 3:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.01, v_num=2]
Epoch 3:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.01, v_num=2]
Epoch 3:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.01, v_num=2]
Epoch 3:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.01, v_num=2]
Epoch 3:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.02, v_num=2]
Epoch 3:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.02, v_num=2]
Epoch 3:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.02, v_num=2]
Epoch 3:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.02, v_num=2]
Epoch 3:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.03, v_num=2]
Epoch 3:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.03, v_num=2]
Epoch 3:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.03, v_num=2]
Epoch 3:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.03, v_num=2]
Epoch 3:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.04, v_num=2]
Epoch 3:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.04, v_num=2]
Epoch 3:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.04, v_num=2]
Epoch 3:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.04, v_num=2]
Epoch 3:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.05, v_num=2]
Epoch 3:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.05, v_num=2]
Epoch 3:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.05, v_num=2]
Epoch 3:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.05, v_num=2]
Epoch 3:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.06, v_num=2]
Epoch 3:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.06, v_num=2]
Epoch 3:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.06, v_num=2]
Epoch 3:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.06, v_num=2]
Epoch 3:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.06, v_num=2]
Epoch 3:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.06, v_num=2]
Epoch 3:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.07, v_num=2]
Epoch 3:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.07, v_num=2]
Epoch 3:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.07, v_num=2]
Epoch 3:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.07, v_num=2]
Epoch 3:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.07, v_num=2]
Epoch 3:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.07, v_num=2]
Epoch 3:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.08, v_num=2]
Epoch 3:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.08, v_num=2]
Epoch 3:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.08, v_num=2]
Epoch 3:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.08, v_num=2]
Epoch 3:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.08, v_num=2]
Epoch 3:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.08, v_num=2]
Epoch 3:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.09, v_num=2]
Epoch 3:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.09, v_num=2]
Epoch 3:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.09, v_num=2]
Epoch 3:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.09, v_num=2]
Epoch 3:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.09, v_num=2]
Epoch 3:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.09, v_num=2]
Epoch 3:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.1, v_num=2]
Epoch 3:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.1, v_num=2]
Epoch 3:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.1, v_num=2]
Epoch 3:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.1, v_num=2]
Epoch 3:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.1, v_num=2]
Epoch 3:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.1, v_num=2]
Epoch 3:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.1, v_num=2]
Epoch 3:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.1, v_num=2]
Epoch 3:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.11, v_num=2]
Epoch 3: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.11, v_num=2]
Epoch 3: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.11, v_num=2]
Epoch 3: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.11, v_num=2]
Epoch 3:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.11, v_num=2]
Epoch 4:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.11, v_num=2]
Epoch 4:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=6.11, v_num=2]
Epoch 4:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=6.11, v_num=2]
Epoch 4:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=6.11, v_num=2]
Epoch 4:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=6.11, v_num=2]
Epoch 4:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.11, v_num=2]
Epoch 4:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.12, v_num=2]
Epoch 4:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=6.12, v_num=2]
Epoch 4:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=6.12, v_num=2]
Epoch 4:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.12, v_num=2]
Epoch 4:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.12, v_num=2]
Epoch 4:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=6.12, v_num=2]
Epoch 4:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=6.12, v_num=2]
Epoch 4:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=6.12, v_num=2]
Epoch 4:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=6.13, v_num=2]
Epoch 4:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=6.13, v_num=2]
Epoch 4:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=6.13, v_num=2]
Epoch 4:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.13, v_num=2]
Epoch 4:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.13, v_num=2]
Epoch 4:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.13, v_num=2]
Epoch 4:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.14, v_num=2]
Epoch 4:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.14, v_num=2]
Epoch 4:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.14, v_num=2]
Epoch 4:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.14, v_num=2]
Epoch 4:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.14, v_num=2]
Epoch 4:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.14, v_num=2]
Epoch 4:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.14, v_num=2]
Epoch 4:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.14, v_num=2]
Epoch 4:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.15, v_num=2]
Epoch 4:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.15, v_num=2]
Epoch 4:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.15, v_num=2]
Epoch 4:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.15, v_num=2]
Epoch 4:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.15, v_num=2]
Epoch 4:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=6.15, v_num=2]
Epoch 4:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=6.15, v_num=2]
Epoch 4:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.15, v_num=2]
Epoch 4:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.16, v_num=2]
Epoch 4:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.16, v_num=2]
Epoch 4:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.16, v_num=2]
Epoch 4:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.16, v_num=2]
Epoch 4:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.16, v_num=2]
Epoch 4:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=6.16, v_num=2]
Epoch 4:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=6.16, v_num=2]
Epoch 4:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.16, v_num=2]
Epoch 4:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.17, v_num=2]
Epoch 4:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.17, v_num=2]
Epoch 4:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.17, v_num=2]
Epoch 4:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=6.17, v_num=2]
Epoch 4:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=6.17, v_num=2]
Epoch 4:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.17, v_num=2]
Epoch 4:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.17, v_num=2]
Epoch 4:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.17, v_num=2]
Epoch 4:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.18, v_num=2]
Epoch 4:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.18, v_num=2]
Epoch 4:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.18, v_num=2]
Epoch 4:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.18, v_num=2]
Epoch 4:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.18, v_num=2]
Epoch 4:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.18, v_num=2]
Epoch 4:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.18, v_num=2]
Epoch 4:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.18, v_num=2]
Epoch 4:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.19, v_num=2]
Epoch 4:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.19, v_num=2]
Epoch 4:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.19, v_num=2]
Epoch 4: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.19, v_num=2]
Epoch 4: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.19, v_num=2]
Epoch 4: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.19, v_num=2]
Epoch 4:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.19, v_num=2]
Epoch 5:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.19, v_num=2]
Epoch 5:   3%|▎         | 1/32 [00:03<01:54,  3.68s/it, loss=6.19, v_num=2]
Epoch 5:   3%|▎         | 1/32 [00:03<01:54,  3.68s/it, loss=6.19, v_num=2]
Epoch 5:   6%|▋         | 2/32 [00:03<00:55,  1.87s/it, loss=6.19, v_num=2]
Epoch 5:   6%|▋         | 2/32 [00:03<00:55,  1.87s/it, loss=6.19, v_num=2]
Epoch 5:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.19, v_num=2]
Epoch 5:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.2, v_num=2]
Epoch 5:  12%|█▎        | 4/32 [00:03<00:27,  1.04it/s, loss=6.2, v_num=2]
Epoch 5:  12%|█▎        | 4/32 [00:03<00:27,  1.04it/s, loss=6.2, v_num=2]
Epoch 5:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.2, v_num=2]
Epoch 5:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.2, v_num=2]
Epoch 5:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.2, v_num=2]
Epoch 5:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.2, v_num=2]
Epoch 5:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=6.2, v_num=2]
Epoch 5:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=6.2, v_num=2]
Epoch 5:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.2, v_num=2]
Epoch 5:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.21, v_num=2]
Epoch 5:  28%|██▊       | 9/32 [00:06<00:16,  1.35it/s, loss=6.21, v_num=2]
Epoch 5:  28%|██▊       | 9/32 [00:06<00:16,  1.35it/s, loss=6.21, v_num=2]
Epoch 5:  31%|███▏      | 10/32 [00:06<00:14,  1.49it/s, loss=6.21, v_num=2]
Epoch 5:  31%|███▏      | 10/32 [00:06<00:14,  1.49it/s, loss=6.21, v_num=2]
Epoch 5:  34%|███▍      | 11/32 [00:06<00:12,  1.63it/s, loss=6.21, v_num=2]
Epoch 5:  34%|███▍      | 11/32 [00:06<00:12,  1.63it/s, loss=6.21, v_num=2]
Epoch 5:  38%|███▊      | 12/32 [00:06<00:11,  1.77it/s, loss=6.21, v_num=2]
Epoch 5:  38%|███▊      | 12/32 [00:06<00:11,  1.77it/s, loss=6.22, v_num=2]
Epoch 5:  41%|████      | 13/32 [00:06<00:09,  1.90it/s, loss=6.22, v_num=2]
Epoch 5:  41%|████      | 13/32 [00:06<00:09,  1.90it/s, loss=6.22, v_num=2]
Epoch 5:  44%|████▍     | 14/32 [00:06<00:08,  2.03it/s, loss=6.22, v_num=2]
Epoch 5:  44%|████▍     | 14/32 [00:06<00:08,  2.03it/s, loss=6.22, v_num=2]
Epoch 5:  47%|████▋     | 15/32 [00:06<00:07,  2.16it/s, loss=6.22, v_num=2]
Epoch 5:  47%|████▋     | 15/32 [00:06<00:07,  2.16it/s, loss=6.22, v_num=2]
Epoch 5:  50%|█████     | 16/32 [00:06<00:06,  2.29it/s, loss=6.22, v_num=2]
Epoch 5:  50%|█████     | 16/32 [00:06<00:06,  2.29it/s, loss=6.22, v_num=2]
Epoch 5:  53%|█████▎    | 17/32 [00:11<00:09,  1.52it/s, loss=6.22, v_num=2]
Epoch 5:  53%|█████▎    | 17/32 [00:11<00:09,  1.52it/s, loss=6.22, v_num=2]
Epoch 5:  56%|█████▋    | 18/32 [00:11<00:08,  1.59it/s, loss=6.22, v_num=2]
Epoch 5:  56%|█████▋    | 18/32 [00:11<00:08,  1.59it/s, loss=6.23, v_num=2]
Epoch 5:  59%|█████▉    | 19/32 [00:11<00:07,  1.66it/s, loss=6.23, v_num=2]
Epoch 5:  59%|█████▉    | 19/32 [00:11<00:07,  1.66it/s, loss=6.23, v_num=2]
Epoch 5:  62%|██████▎   | 20/32 [00:11<00:06,  1.74it/s, loss=6.23, v_num=2]
Epoch 5:  62%|██████▎   | 20/32 [00:11<00:06,  1.74it/s, loss=6.23, v_num=2]
Epoch 5:  66%|██████▌   | 21/32 [00:11<00:06,  1.81it/s, loss=6.23, v_num=2]
Epoch 5:  66%|██████▌   | 21/32 [00:11<00:06,  1.81it/s, loss=6.23, v_num=2]
Epoch 5:  69%|██████▉   | 22/32 [00:11<00:05,  1.89it/s, loss=6.23, v_num=2]
Epoch 5:  69%|██████▉   | 22/32 [00:11<00:05,  1.89it/s, loss=6.23, v_num=2]
Epoch 5:  72%|███████▏  | 23/32 [00:11<00:04,  1.96it/s, loss=6.23, v_num=2]
Epoch 5:  72%|███████▏  | 23/32 [00:11<00:04,  1.96it/s, loss=6.23, v_num=2]
Epoch 5:  75%|███████▌  | 24/32 [00:11<00:03,  2.03it/s, loss=6.23, v_num=2]
Epoch 5:  75%|███████▌  | 24/32 [00:11<00:03,  2.03it/s, loss=6.23, v_num=2]
Epoch 5:  78%|███████▊  | 25/32 [00:15<00:04,  1.62it/s, loss=6.23, v_num=2]
Epoch 5:  78%|███████▊  | 25/32 [00:15<00:04,  1.62it/s, loss=6.23, v_num=2]
Epoch 5:  81%|████████▏ | 26/32 [00:15<00:03,  1.66it/s, loss=6.23, v_num=2]
Epoch 5:  81%|████████▏ | 26/32 [00:15<00:03,  1.66it/s, loss=6.23, v_num=2]
Epoch 5:  84%|████████▍ | 27/32 [00:15<00:02,  1.72it/s, loss=6.23, v_num=2]
Epoch 5:  84%|████████▍ | 27/32 [00:15<00:02,  1.72it/s, loss=6.23, v_num=2]
Epoch 5:  88%|████████▊ | 28/32 [00:15<00:02,  1.77it/s, loss=6.23, v_num=2]
Epoch 5:  88%|████████▊ | 28/32 [00:15<00:02,  1.77it/s, loss=6.23, v_num=2]
Epoch 5:  91%|█████████ | 29/32 [00:15<00:01,  1.83it/s, loss=6.23, v_num=2]
Epoch 5:  91%|█████████ | 29/32 [00:15<00:01,  1.83it/s, loss=6.23, v_num=2]
Epoch 5:  94%|█████████▍| 30/32 [00:15<00:01,  1.88it/s, loss=6.23, v_num=2]
Epoch 5:  94%|█████████▍| 30/32 [00:15<00:01,  1.88it/s, loss=6.23, v_num=2]
Epoch 5:  97%|█████████▋| 31/32 [00:15<00:00,  1.94it/s, loss=6.23, v_num=2]
Epoch 5:  97%|█████████▋| 31/32 [00:15<00:00,  1.94it/s, loss=6.23, v_num=2]
Epoch 5: 100%|██████████| 32/32 [00:16<00:00,  1.99it/s, loss=6.23, v_num=2]
Epoch 5: 100%|██████████| 32/32 [00:16<00:00,  1.99it/s, loss=6.23, v_num=2]
Epoch 5: 100%|██████████| 32/32 [00:16<00:00,  1.99it/s, loss=6.23, v_num=2]
Epoch 5:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.23, v_num=2]
Epoch 6:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.23, v_num=2]
Epoch 6:   3%|▎         | 1/32 [00:05<02:36,  5.06s/it, loss=6.23, v_num=2]
Epoch 6:   3%|▎         | 1/32 [00:05<02:37,  5.07s/it, loss=6.23, v_num=2]
Epoch 6:   6%|▋         | 2/32 [00:05<01:16,  2.56s/it, loss=6.23, v_num=2]
Epoch 6:   6%|▋         | 2/32 [00:05<01:16,  2.56s/it, loss=6.23, v_num=2]
Epoch 6:   9%|▉         | 3/32 [00:05<00:50,  1.73s/it, loss=6.23, v_num=2]
Epoch 6:   9%|▉         | 3/32 [00:05<00:50,  1.73s/it, loss=6.23, v_num=2]
Epoch 6:  12%|█▎        | 4/32 [00:05<00:36,  1.32s/it, loss=6.23, v_num=2]
Epoch 6:  12%|█▎        | 4/32 [00:05<00:36,  1.32s/it, loss=6.24, v_num=2]
Epoch 6:  16%|█▌        | 5/32 [00:05<00:28,  1.06s/it, loss=6.24, v_num=2]
Epoch 6:  16%|█▌        | 5/32 [00:05<00:28,  1.06s/it, loss=6.24, v_num=2]
Epoch 6:  19%|█▉        | 6/32 [00:05<00:23,  1.12it/s, loss=6.24, v_num=2]
Epoch 6:  19%|█▉        | 6/32 [00:05<00:23,  1.12it/s, loss=6.24, v_num=2]
Epoch 6:  22%|██▏       | 7/32 [00:05<00:19,  1.29it/s, loss=6.24, v_num=2]
Epoch 6:  22%|██▏       | 7/32 [00:05<00:19,  1.29it/s, loss=6.24, v_num=2]
Epoch 6:  25%|██▌       | 8/32 [00:05<00:16,  1.45it/s, loss=6.24, v_num=2]
Epoch 6:  25%|██▌       | 8/32 [00:05<00:16,  1.45it/s, loss=6.24, v_num=2]
Epoch 6:  28%|██▊       | 9/32 [00:08<00:22,  1.03it/s, loss=6.24, v_num=2]
Epoch 6:  28%|██▊       | 9/32 [00:08<00:22,  1.03it/s, loss=6.24, v_num=2]
Epoch 6:  31%|███▏      | 10/32 [00:08<00:19,  1.13it/s, loss=6.24, v_num=2]
Epoch 6:  31%|███▏      | 10/32 [00:08<00:19,  1.13it/s, loss=6.24, v_num=2]
Epoch 6:  34%|███▍      | 11/32 [00:08<00:16,  1.24it/s, loss=6.24, v_num=2]
Epoch 6:  34%|███▍      | 11/32 [00:08<00:16,  1.24it/s, loss=6.24, v_num=2]
Epoch 6:  38%|███▊      | 12/32 [00:09<00:15,  1.33it/s, loss=6.24, v_num=2]
Epoch 6:  38%|███▊      | 12/32 [00:09<00:15,  1.33it/s, loss=6.24, v_num=2]
Epoch 6:  41%|████      | 13/32 [00:09<00:13,  1.43it/s, loss=6.24, v_num=2]
Epoch 6:  41%|████      | 13/32 [00:09<00:13,  1.43it/s, loss=6.25, v_num=2]
Epoch 6:  44%|████▍     | 14/32 [00:09<00:11,  1.53it/s, loss=6.25, v_num=2]
Epoch 6:  44%|████▍     | 14/32 [00:09<00:11,  1.53it/s, loss=6.25, v_num=2]
Epoch 6:  47%|████▋     | 15/32 [00:09<00:10,  1.63it/s, loss=6.25, v_num=2]
Epoch 6:  47%|████▋     | 15/32 [00:09<00:10,  1.63it/s, loss=6.25, v_num=2]
Epoch 6:  50%|█████     | 16/32 [00:09<00:09,  1.73it/s, loss=6.25, v_num=2]
Epoch 6:  50%|█████     | 16/32 [00:09<00:09,  1.73it/s, loss=6.25, v_num=2]
Epoch 6:  53%|█████▎    | 17/32 [00:11<00:10,  1.43it/s, loss=6.25, v_num=2]
Epoch 6:  53%|█████▎    | 17/32 [00:11<00:10,  1.43it/s, loss=6.25, v_num=2]
Epoch 6:  56%|█████▋    | 18/32 [00:11<00:09,  1.50it/s, loss=6.25, v_num=2]
Epoch 6:  56%|█████▋    | 18/32 [00:11<00:09,  1.50it/s, loss=6.26, v_num=2]
Epoch 6:  59%|█████▉    | 19/32 [00:12<00:08,  1.58it/s, loss=6.26, v_num=2]
Epoch 6:  59%|█████▉    | 19/32 [00:12<00:08,  1.58it/s, loss=6.26, v_num=2]
Epoch 6:  62%|██████▎   | 20/32 [00:12<00:07,  1.66it/s, loss=6.26, v_num=2]
Epoch 6:  62%|██████▎   | 20/32 [00:12<00:07,  1.66it/s, loss=6.26, v_num=2]
Epoch 6:  66%|██████▌   | 21/32 [00:12<00:06,  1.73it/s, loss=6.26, v_num=2]
Epoch 6:  66%|██████▌   | 21/32 [00:12<00:06,  1.73it/s, loss=6.26, v_num=2]
Epoch 6:  69%|██████▉   | 22/32 [00:12<00:05,  1.81it/s, loss=6.26, v_num=2]
Epoch 6:  69%|██████▉   | 22/32 [00:12<00:05,  1.81it/s, loss=6.27, v_num=2]
Epoch 6:  72%|███████▏  | 23/32 [00:12<00:04,  1.88it/s, loss=6.27, v_num=2]
Epoch 6:  72%|███████▏  | 23/32 [00:12<00:04,  1.88it/s, loss=6.27, v_num=2]
Epoch 6:  75%|███████▌  | 24/32 [00:12<00:04,  1.96it/s, loss=6.27, v_num=2]
Epoch 6:  75%|███████▌  | 24/32 [00:12<00:04,  1.95it/s, loss=6.27, v_num=2]
Epoch 6:  78%|███████▊  | 25/32 [00:15<00:04,  1.66it/s, loss=6.27, v_num=2]
Epoch 6:  78%|███████▊  | 25/32 [00:15<00:04,  1.66it/s, loss=6.27, v_num=2]
Epoch 6:  81%|████████▏ | 26/32 [00:15<00:03,  1.72it/s, loss=6.27, v_num=2]
Epoch 6:  81%|████████▏ | 26/32 [00:15<00:03,  1.72it/s, loss=6.28, v_num=2]
Epoch 6:  84%|████████▍ | 27/32 [00:15<00:02,  1.78it/s, loss=6.28, v_num=2]
Epoch 6:  84%|████████▍ | 27/32 [00:15<00:02,  1.78it/s, loss=6.28, v_num=2]
Epoch 6:  88%|████████▊ | 28/32 [00:15<00:02,  1.84it/s, loss=6.28, v_num=2]
Epoch 6:  88%|████████▊ | 28/32 [00:15<00:02,  1.84it/s, loss=6.28, v_num=2]
Epoch 6:  91%|█████████ | 29/32 [00:15<00:01,  1.90it/s, loss=6.28, v_num=2]
Epoch 6:  91%|█████████ | 29/32 [00:15<00:01,  1.90it/s, loss=6.28, v_num=2]
Epoch 6:  94%|█████████▍| 30/32 [00:15<00:01,  1.96it/s, loss=6.28, v_num=2]
Epoch 6:  94%|█████████▍| 30/32 [00:15<00:01,  1.96it/s, loss=6.29, v_num=2]
Epoch 6:  97%|█████████▋| 31/32 [00:15<00:00,  2.02it/s, loss=6.29, v_num=2]
Epoch 6:  97%|█████████▋| 31/32 [00:15<00:00,  2.02it/s, loss=6.29, v_num=2]
Epoch 6: 100%|██████████| 32/32 [00:15<00:00,  2.08it/s, loss=6.29, v_num=2]
Epoch 6: 100%|██████████| 32/32 [00:15<00:00,  2.08it/s, loss=6.29, v_num=2]
Epoch 6: 100%|██████████| 32/32 [00:15<00:00,  2.08it/s, loss=6.29, v_num=2]
Epoch 6:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.29, v_num=2]
Epoch 7:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.29, v_num=2]
Epoch 7:   3%|▎         | 1/32 [00:05<02:59,  5.80s/it, loss=6.29, v_num=2]
Epoch 7:   3%|▎         | 1/32 [00:05<02:59,  5.80s/it, loss=6.29, v_num=2]
Epoch 7:   6%|▋         | 2/32 [00:05<01:28,  2.94s/it, loss=6.29, v_num=2]
Epoch 7:   6%|▋         | 2/32 [00:05<01:28,  2.94s/it, loss=6.29, v_num=2]
Epoch 7:   9%|▉         | 3/32 [00:05<00:57,  1.98s/it, loss=6.29, v_num=2]
Epoch 7:   9%|▉         | 3/32 [00:05<00:57,  1.98s/it, loss=6.3, v_num=2]
Epoch 7:  12%|█▎        | 4/32 [00:05<00:41,  1.50s/it, loss=6.3, v_num=2]
Epoch 7:  12%|█▎        | 4/32 [00:05<00:41,  1.50s/it, loss=6.3, v_num=2]
Epoch 7:  16%|█▌        | 5/32 [00:06<00:32,  1.21s/it, loss=6.3, v_num=2]
Epoch 7:  16%|█▌        | 5/32 [00:06<00:32,  1.21s/it, loss=6.3, v_num=2]
Epoch 7:  19%|█▉        | 6/32 [00:06<00:26,  1.02s/it, loss=6.3, v_num=2]
Epoch 7:  19%|█▉        | 6/32 [00:06<00:26,  1.02s/it, loss=6.3, v_num=2]
Epoch 7:  22%|██▏       | 7/32 [00:06<00:22,  1.13it/s, loss=6.3, v_num=2]
Epoch 7:  22%|██▏       | 7/32 [00:06<00:22,  1.13it/s, loss=6.3, v_num=2]
Epoch 7:  25%|██▌       | 8/32 [00:06<00:18,  1.28it/s, loss=6.3, v_num=2]
Epoch 7:  25%|██▌       | 8/32 [00:06<00:18,  1.28it/s, loss=6.31, v_num=2]
Epoch 7:  28%|██▊       | 9/32 [00:10<00:25,  1.12s/it, loss=6.31, v_num=2]
Epoch 7:  28%|██▊       | 9/32 [00:10<00:25,  1.12s/it, loss=6.31, v_num=2]
Epoch 7:  31%|███▏      | 10/32 [00:10<00:22,  1.02s/it, loss=6.31, v_num=2]
Epoch 7:  31%|███▏      | 10/32 [00:10<00:22,  1.02s/it, loss=6.31, v_num=2]
Epoch 7:  34%|███▍      | 11/32 [00:10<00:19,  1.07it/s, loss=6.31, v_num=2]
Epoch 7:  34%|███▍      | 11/32 [00:10<00:19,  1.07it/s, loss=6.31, v_num=2]
Epoch 7:  38%|███▊      | 12/32 [00:10<00:17,  1.16it/s, loss=6.31, v_num=2]
Epoch 7:  38%|███▊      | 12/32 [00:10<00:17,  1.16it/s, loss=6.31, v_num=2]
Epoch 7:  41%|████      | 13/32 [00:10<00:15,  1.25it/s, loss=6.31, v_num=2]
Epoch 7:  41%|████      | 13/32 [00:10<00:15,  1.25it/s, loss=6.31, v_num=2]
Epoch 7:  44%|████▍     | 14/32 [00:10<00:13,  1.34it/s, loss=6.31, v_num=2]
Epoch 7:  44%|████▍     | 14/32 [00:10<00:13,  1.34it/s, loss=6.31, v_num=2]
Epoch 7:  47%|████▋     | 15/32 [00:10<00:11,  1.43it/s, loss=6.31, v_num=2]
Epoch 7:  47%|████▋     | 15/32 [00:10<00:11,  1.43it/s, loss=6.31, v_num=2]
Epoch 7:  50%|█████     | 16/32 [00:10<00:10,  1.51it/s, loss=6.31, v_num=2]
Epoch 7:  50%|█████     | 16/32 [00:10<00:10,  1.51it/s, loss=6.32, v_num=2]
Epoch 7:  53%|█████▎    | 17/32 [00:13<00:12,  1.23it/s, loss=6.32, v_num=2]
Epoch 7:  53%|█████▎    | 17/32 [00:13<00:12,  1.23it/s, loss=6.32, v_num=2]
Epoch 7:  56%|█████▋    | 18/32 [00:13<00:10,  1.30it/s, loss=6.32, v_num=2]
Epoch 7:  56%|█████▋    | 18/32 [00:13<00:10,  1.30it/s, loss=6.32, v_num=2]
Epoch 7:  59%|█████▉    | 19/32 [00:13<00:09,  1.36it/s, loss=6.32, v_num=2]
Epoch 7:  59%|█████▉    | 19/32 [00:13<00:09,  1.36it/s, loss=6.32, v_num=2]
Epoch 7:  62%|██████▎   | 20/32 [00:13<00:08,  1.43it/s, loss=6.32, v_num=2]
Epoch 7:  62%|██████▎   | 20/32 [00:13<00:08,  1.43it/s, loss=6.32, v_num=2]
Epoch 7:  66%|██████▌   | 21/32 [00:14<00:07,  1.49it/s, loss=6.32, v_num=2]
Epoch 7:  66%|██████▌   | 21/32 [00:14<00:07,  1.49it/s, loss=6.33, v_num=2]
Epoch 7:  69%|██████▉   | 22/32 [00:14<00:06,  1.56it/s, loss=6.33, v_num=2]
Epoch 7:  69%|██████▉   | 22/32 [00:14<00:06,  1.56it/s, loss=6.33, v_num=2]
Epoch 7:  72%|███████▏  | 23/32 [00:14<00:05,  1.62it/s, loss=6.33, v_num=2]
Epoch 7:  72%|███████▏  | 23/32 [00:14<00:05,  1.62it/s, loss=6.33, v_num=2]
Epoch 7:  75%|███████▌  | 24/32 [00:14<00:04,  1.69it/s, loss=6.33, v_num=2]
Epoch 7:  75%|███████▌  | 24/32 [00:14<00:04,  1.68it/s, loss=6.33, v_num=2]
Epoch 7:  78%|███████▊  | 25/32 [00:17<00:04,  1.45it/s, loss=6.33, v_num=2]
Epoch 7:  78%|███████▊  | 25/32 [00:17<00:04,  1.45it/s, loss=6.34, v_num=2]
Epoch 7:  81%|████████▏ | 26/32 [00:17<00:04,  1.49it/s, loss=6.34, v_num=2]
Epoch 7:  81%|████████▏ | 26/32 [00:17<00:04,  1.49it/s, loss=6.34, v_num=2]
Epoch 7:  84%|████████▍ | 27/32 [00:17<00:03,  1.54it/s, loss=6.34, v_num=2]
Epoch 7:  84%|████████▍ | 27/32 [00:17<00:03,  1.54it/s, loss=6.35, v_num=2]
Epoch 7:  88%|████████▊ | 28/32 [00:17<00:02,  1.59it/s, loss=6.35, v_num=2]
Epoch 7:  88%|████████▊ | 28/32 [00:17<00:02,  1.59it/s, loss=6.35, v_num=2]
Epoch 7:  91%|█████████ | 29/32 [00:17<00:01,  1.65it/s, loss=6.35, v_num=2]
Epoch 7:  91%|█████████ | 29/32 [00:17<00:01,  1.65it/s, loss=6.35, v_num=2]
Epoch 7:  94%|█████████▍| 30/32 [00:17<00:01,  1.70it/s, loss=6.35, v_num=2]
Epoch 7:  94%|█████████▍| 30/32 [00:17<00:01,  1.70it/s, loss=6.36, v_num=2]
Epoch 7:  97%|█████████▋| 31/32 [00:17<00:00,  1.75it/s, loss=6.36, v_num=2]
Epoch 7:  97%|█████████▋| 31/32 [00:17<00:00,  1.75it/s, loss=6.36, v_num=2]
Epoch 7: 100%|██████████| 32/32 [00:17<00:00,  1.80it/s, loss=6.36, v_num=2]
Epoch 7: 100%|██████████| 32/32 [00:17<00:00,  1.80it/s, loss=6.37, v_num=2]
Epoch 7: 100%|██████████| 32/32 [00:17<00:00,  1.80it/s, loss=6.37, v_num=2]
Epoch 7:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.37, v_num=2]
Epoch 8:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.37, v_num=2]
Epoch 8:   3%|▎         | 1/32 [00:04<02:12,  4.27s/it, loss=6.37, v_num=2]
Epoch 8:   3%|▎         | 1/32 [00:04<02:12,  4.27s/it, loss=6.37, v_num=2]
Epoch 8:   6%|▋         | 2/32 [00:04<01:04,  2.16s/it, loss=6.37, v_num=2]
Epoch 8:   6%|▋         | 2/32 [00:04<01:04,  2.16s/it, loss=6.38, v_num=2]
Epoch 8:   9%|▉         | 3/32 [00:04<00:42,  1.46s/it, loss=6.38, v_num=2]
Epoch 8:   9%|▉         | 3/32 [00:04<00:42,  1.46s/it, loss=6.38, v_num=2]
Epoch 8:  12%|█▎        | 4/32 [00:04<00:30,  1.10s/it, loss=6.38, v_num=2]
Epoch 8:  12%|█▎        | 4/32 [00:04<00:30,  1.10s/it, loss=6.38, v_num=2]
Epoch 8:  16%|█▌        | 5/32 [00:04<00:24,  1.12it/s, loss=6.38, v_num=2]
Epoch 8:  16%|█▌        | 5/32 [00:04<00:24,  1.12it/s, loss=6.39, v_num=2]
Epoch 8:  19%|█▉        | 6/32 [00:04<00:19,  1.33it/s, loss=6.39, v_num=2]
Epoch 8:  19%|█▉        | 6/32 [00:04<00:19,  1.33it/s, loss=6.39, v_num=2]
Epoch 8:  22%|██▏       | 7/32 [00:04<00:16,  1.53it/s, loss=6.39, v_num=2]
Epoch 8:  22%|██▏       | 7/32 [00:04<00:16,  1.53it/s, loss=6.4, v_num=2]
Epoch 8:  25%|██▌       | 8/32 [00:04<00:13,  1.73it/s, loss=6.4, v_num=2]
Epoch 8:  25%|██▌       | 8/32 [00:04<00:13,  1.73it/s, loss=6.4, v_num=2]
Epoch 8:  28%|██▊       | 9/32 [00:07<00:19,  1.19it/s, loss=6.4, v_num=2]
Epoch 8:  28%|██▊       | 9/32 [00:07<00:19,  1.19it/s, loss=6.4, v_num=2]
Epoch 8:  31%|███▏      | 10/32 [00:07<00:16,  1.32it/s, loss=6.4, v_num=2]
Epoch 8:  31%|███▏      | 10/32 [00:07<00:16,  1.32it/s, loss=6.41, v_num=2]
Epoch 8:  34%|███▍      | 11/32 [00:07<00:14,  1.44it/s, loss=6.41, v_num=2]
Epoch 8:  34%|███▍      | 11/32 [00:07<00:14,  1.44it/s, loss=6.41, v_num=2]
Epoch 8:  38%|███▊      | 12/32 [00:07<00:12,  1.56it/s, loss=6.41, v_num=2]
Epoch 8:  38%|███▊      | 12/32 [00:07<00:12,  1.56it/s, loss=6.41, v_num=2]
Epoch 8:  41%|████      | 13/32 [00:07<00:11,  1.68it/s, loss=6.41, v_num=2]
Epoch 8:  41%|████      | 13/32 [00:07<00:11,  1.68it/s, loss=6.41, v_num=2]
Epoch 8:  44%|████▍     | 14/32 [00:07<00:10,  1.80it/s, loss=6.41, v_num=2]
Epoch 8:  44%|████▍     | 14/32 [00:07<00:10,  1.80it/s, loss=6.41, v_num=2]
Epoch 8:  47%|████▋     | 15/32 [00:07<00:08,  1.92it/s, loss=6.41, v_num=2]
Epoch 8:  47%|████▋     | 15/32 [00:07<00:08,  1.92it/s, loss=6.41, v_num=2]
Epoch 8:  50%|█████     | 16/32 [00:07<00:07,  2.03it/s, loss=6.41, v_num=2]
Epoch 8:  50%|█████     | 16/32 [00:07<00:07,  2.03it/s, loss=6.41, v_num=2]
Epoch 8:  53%|█████▎    | 17/32 [00:10<00:09,  1.56it/s, loss=6.41, v_num=2]
Epoch 8:  53%|█████▎    | 17/32 [00:10<00:09,  1.56it/s, loss=6.41, v_num=2]
Epoch 8:  56%|█████▋    | 18/32 [00:10<00:08,  1.64it/s, loss=6.41, v_num=2]
Epoch 8:  56%|█████▋    | 18/32 [00:10<00:08,  1.64it/s, loss=6.42, v_num=2]
Epoch 8:  59%|█████▉    | 19/32 [00:11<00:07,  1.73it/s, loss=6.42, v_num=2]
Epoch 8:  59%|█████▉    | 19/32 [00:11<00:07,  1.73it/s, loss=6.42, v_num=2]
Epoch 8:  62%|██████▎   | 20/32 [00:11<00:06,  1.81it/s, loss=6.42, v_num=2]
Epoch 8:  62%|██████▎   | 20/32 [00:11<00:06,  1.81it/s, loss=6.42, v_num=2]
Epoch 8:  66%|██████▌   | 21/32 [00:11<00:05,  1.89it/s, loss=6.42, v_num=2]
Epoch 8:  66%|██████▌   | 21/32 [00:11<00:05,  1.89it/s, loss=6.42, v_num=2]
Epoch 8:  69%|██████▉   | 22/32 [00:11<00:05,  1.97it/s, loss=6.42, v_num=2]
Epoch 8:  69%|██████▉   | 22/32 [00:11<00:05,  1.97it/s, loss=6.42, v_num=2]
Epoch 8:  72%|███████▏  | 23/32 [00:11<00:04,  2.05it/s, loss=6.42, v_num=2]
Epoch 8:  72%|███████▏  | 23/32 [00:11<00:04,  2.05it/s, loss=6.42, v_num=2]
Epoch 8:  75%|███████▌  | 24/32 [00:11<00:03,  2.13it/s, loss=6.42, v_num=2]
Epoch 8:  75%|███████▌  | 24/32 [00:11<00:03,  2.12it/s, loss=6.41, v_num=2]
Epoch 8:  78%|███████▊  | 25/32 [00:14<00:03,  1.78it/s, loss=6.41, v_num=2]
Epoch 8:  78%|███████▊  | 25/32 [00:14<00:03,  1.78it/s, loss=6.41, v_num=2]
Epoch 8:  81%|████████▏ | 26/32 [00:14<00:03,  1.85it/s, loss=6.41, v_num=2]
Epoch 8:  81%|████████▏ | 26/32 [00:14<00:03,  1.85it/s, loss=6.41, v_num=2]
Epoch 8:  84%|████████▍ | 27/32 [00:14<00:02,  1.91it/s, loss=6.41, v_num=2]
Epoch 8:  84%|████████▍ | 27/32 [00:14<00:02,  1.91it/s, loss=6.41, v_num=2]
Epoch 8:  88%|████████▊ | 28/32 [00:14<00:02,  1.98it/s, loss=6.41, v_num=2]
Epoch 8:  88%|████████▊ | 28/32 [00:14<00:02,  1.98it/s, loss=6.41, v_num=2]
Epoch 8:  91%|█████████ | 29/32 [00:14<00:01,  2.04it/s, loss=6.41, v_num=2]
Epoch 8:  91%|█████████ | 29/32 [00:14<00:01,  2.04it/s, loss=6.41, v_num=2]
Epoch 8:  94%|█████████▍| 30/32 [00:14<00:00,  2.10it/s, loss=6.41, v_num=2]
Epoch 8:  94%|█████████▍| 30/32 [00:14<00:00,  2.10it/s, loss=6.41, v_num=2]
Epoch 8:  97%|█████████▋| 31/32 [00:14<00:00,  2.17it/s, loss=6.41, v_num=2]
Epoch 8:  97%|█████████▋| 31/32 [00:14<00:00,  2.17it/s, loss=6.41, v_num=2]
Epoch 8: 100%|██████████| 32/32 [00:14<00:00,  2.23it/s, loss=6.41, v_num=2]
Epoch 8: 100%|██████████| 32/32 [00:14<00:00,  2.23it/s, loss=6.41, v_num=2]
Epoch 8: 100%|██████████| 32/32 [00:14<00:00,  2.23it/s, loss=6.41, v_num=2]
Epoch 8:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.41, v_num=2]
Epoch 9:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.41, v_num=2]
Epoch 9:   3%|▎         | 1/32 [00:03<02:02,  3.96s/it, loss=6.41, v_num=2]
Epoch 9:   3%|▎         | 1/32 [00:03<02:02,  3.96s/it, loss=6.41, v_num=2]
Epoch 9:   6%|▋         | 2/32 [00:04<01:00,  2.00s/it, loss=6.41, v_num=2]
Epoch 9:   6%|▋         | 2/32 [00:04<01:00,  2.00s/it, loss=6.41, v_num=2]
Epoch 9:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.41, v_num=2]
Epoch 9:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.41, v_num=2]
Epoch 9:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.41, v_num=2]
Epoch 9:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.4, v_num=2]
Epoch 9:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.4, v_num=2]
Epoch 9:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.4, v_num=2]
Epoch 9:  19%|█▉        | 6/32 [00:04<00:18,  1.43it/s, loss=6.4, v_num=2]
Epoch 9:  19%|█▉        | 6/32 [00:04<00:18,  1.42it/s, loss=6.4, v_num=2]
Epoch 9:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.4, v_num=2]
Epoch 9:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.4, v_num=2]
Epoch 9:  25%|██▌       | 8/32 [00:04<00:12,  1.86it/s, loss=6.4, v_num=2]
Epoch 9:  25%|██▌       | 8/32 [00:04<00:12,  1.86it/s, loss=6.4, v_num=2]
Epoch 9:  28%|██▊       | 9/32 [00:07<00:19,  1.16it/s, loss=6.4, v_num=2]
Epoch 9:  28%|██▊       | 9/32 [00:07<00:19,  1.16it/s, loss=6.4, v_num=2]
Epoch 9:  31%|███▏      | 10/32 [00:07<00:17,  1.28it/s, loss=6.4, v_num=2]
Epoch 9:  31%|███▏      | 10/32 [00:07<00:17,  1.28it/s, loss=6.4, v_num=2]
Epoch 9:  34%|███▍      | 11/32 [00:08<00:15,  1.36it/s, loss=6.4, v_num=2]
Epoch 9:  34%|███▍      | 11/32 [00:08<00:15,  1.36it/s, loss=6.4, v_num=2]
Epoch 9:  38%|███▊      | 12/32 [00:08<00:13,  1.47it/s, loss=6.4, v_num=2]
Epoch 9:  38%|███▊      | 12/32 [00:08<00:13,  1.47it/s, loss=6.4, v_num=2]
Epoch 9:  41%|████      | 13/32 [00:08<00:12,  1.58it/s, loss=6.4, v_num=2]
Epoch 9:  41%|████      | 13/32 [00:08<00:12,  1.58it/s, loss=6.4, v_num=2]
Epoch 9:  44%|████▍     | 14/32 [00:08<00:10,  1.69it/s, loss=6.4, v_num=2]
Epoch 9:  44%|████▍     | 14/32 [00:08<00:10,  1.69it/s, loss=6.4, v_num=2]
Epoch 9:  47%|████▋     | 15/32 [00:08<00:09,  1.79it/s, loss=6.4, v_num=2]
Epoch 9:  47%|████▋     | 15/32 [00:08<00:09,  1.79it/s, loss=6.4, v_num=2]
Epoch 9:  50%|█████     | 16/32 [00:08<00:08,  1.90it/s, loss=6.4, v_num=2]
Epoch 9:  50%|█████     | 16/32 [00:08<00:08,  1.90it/s, loss=6.4, v_num=2]
Epoch 9:  53%|█████▎    | 17/32 [00:11<00:10,  1.49it/s, loss=6.4, v_num=2]
Epoch 9:  53%|█████▎    | 17/32 [00:11<00:10,  1.49it/s, loss=6.4, v_num=2]
Epoch 9:  56%|█████▋    | 18/32 [00:11<00:09,  1.54it/s, loss=6.4, v_num=2]
Epoch 9:  56%|█████▋    | 18/32 [00:11<00:09,  1.54it/s, loss=6.41, v_num=2]
Epoch 9:  59%|█████▉    | 19/32 [00:11<00:08,  1.62it/s, loss=6.41, v_num=2]
Epoch 9:  59%|█████▉    | 19/32 [00:11<00:08,  1.62it/s, loss=6.41, v_num=2]
Epoch 9:  62%|██████▎   | 20/32 [00:11<00:07,  1.69it/s, loss=6.41, v_num=2]
Epoch 9:  62%|██████▎   | 20/32 [00:11<00:07,  1.69it/s, loss=6.41, v_num=2]
Epoch 9:  66%|██████▌   | 21/32 [00:11<00:06,  1.77it/s, loss=6.41, v_num=2]
Epoch 9:  66%|██████▌   | 21/32 [00:11<00:06,  1.77it/s, loss=6.41, v_num=2]
Epoch 9:  69%|██████▉   | 22/32 [00:11<00:05,  1.84it/s, loss=6.41, v_num=2]
Epoch 9:  69%|██████▉   | 22/32 [00:11<00:05,  1.84it/s, loss=6.41, v_num=2]
Epoch 9:  72%|███████▏  | 23/32 [00:11<00:04,  1.92it/s, loss=6.41, v_num=2]
Epoch 9:  72%|███████▏  | 23/32 [00:11<00:04,  1.92it/s, loss=6.42, v_num=2]
Epoch 9:  75%|███████▌  | 24/32 [00:12<00:04,  1.99it/s, loss=6.42, v_num=2]
Epoch 9:  75%|███████▌  | 24/32 [00:12<00:04,  1.99it/s, loss=6.42, v_num=2]
Epoch 9:  78%|███████▊  | 25/32 [00:15<00:04,  1.64it/s, loss=6.42, v_num=2]
Epoch 9:  78%|███████▊  | 25/32 [00:15<00:04,  1.64it/s, loss=6.42, v_num=2]
Epoch 9:  81%|████████▏ | 26/32 [00:15<00:03,  1.67it/s, loss=6.42, v_num=2]
Epoch 9:  81%|████████▏ | 26/32 [00:15<00:03,  1.67it/s, loss=6.42, v_num=2]
Epoch 9:  84%|████████▍ | 27/32 [00:15<00:02,  1.73it/s, loss=6.42, v_num=2]
Epoch 9:  84%|████████▍ | 27/32 [00:15<00:02,  1.73it/s, loss=6.42, v_num=2]
Epoch 9:  88%|████████▊ | 28/32 [00:15<00:02,  1.79it/s, loss=6.42, v_num=2]
Epoch 9:  88%|████████▊ | 28/32 [00:15<00:02,  1.79it/s, loss=6.43, v_num=2]
Epoch 9:  91%|█████████ | 29/32 [00:15<00:01,  1.85it/s, loss=6.43, v_num=2]
Epoch 9:  91%|█████████ | 29/32 [00:15<00:01,  1.85it/s, loss=6.43, v_num=2]
Epoch 9:  94%|█████████▍| 30/32 [00:15<00:01,  1.91it/s, loss=6.43, v_num=2]
Epoch 9:  94%|█████████▍| 30/32 [00:15<00:01,  1.91it/s, loss=6.43, v_num=2]
Epoch 9:  97%|█████████▋| 31/32 [00:15<00:00,  1.96it/s, loss=6.43, v_num=2]
Epoch 9:  97%|█████████▋| 31/32 [00:15<00:00,  1.96it/s, loss=6.43, v_num=2]
Epoch 9: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.43, v_num=2]
Epoch 9: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.44, v_num=2]
Epoch 9: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.44, v_num=2]
Epoch 9:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.44, v_num=2]
Epoch 10:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.44, v_num=2]
Epoch 10:   3%|▎         | 1/32 [00:04<02:17,  4.43s/it, loss=6.44, v_num=2]
Epoch 10:   3%|▎         | 1/32 [00:04<02:17,  4.43s/it, loss=6.44, v_num=2]
Epoch 10:   6%|▋         | 2/32 [00:04<01:07,  2.24s/it, loss=6.44, v_num=2]
Epoch 10:   6%|▋         | 2/32 [00:04<01:07,  2.24s/it, loss=6.44, v_num=2]
Epoch 10:   9%|▉         | 3/32 [00:04<00:43,  1.51s/it, loss=6.44, v_num=2]
Epoch 10:   9%|▉         | 3/32 [00:04<00:43,  1.51s/it, loss=6.44, v_num=2]
Epoch 10:  12%|█▎        | 4/32 [00:04<00:32,  1.15s/it, loss=6.44, v_num=2]
Epoch 10:  12%|█▎        | 4/32 [00:04<00:32,  1.15s/it, loss=6.44, v_num=2]
Epoch 10:  16%|█▌        | 5/32 [00:04<00:25,  1.08it/s, loss=6.44, v_num=2]
Epoch 10:  16%|█▌        | 5/32 [00:04<00:25,  1.08it/s, loss=6.45, v_num=2]
Epoch 10:  19%|█▉        | 6/32 [00:04<00:20,  1.28it/s, loss=6.45, v_num=2]
Epoch 10:  19%|█▉        | 6/32 [00:04<00:20,  1.28it/s, loss=6.45, v_num=2]
Epoch 10:  22%|██▏       | 7/32 [00:04<00:16,  1.48it/s, loss=6.45, v_num=2]
Epoch 10:  22%|██▏       | 7/32 [00:04<00:16,  1.48it/s, loss=6.45, v_num=2]
Epoch 10:  25%|██▌       | 8/32 [00:04<00:14,  1.67it/s, loss=6.45, v_num=2]
Epoch 10:  25%|██▌       | 8/32 [00:04<00:14,  1.67it/s, loss=6.45, v_num=2]
Epoch 10:  28%|██▊       | 9/32 [00:07<00:19,  1.19it/s, loss=6.45, v_num=2]
Epoch 10:  28%|██▊       | 9/32 [00:07<00:19,  1.19it/s, loss=6.46, v_num=2]
Epoch 10:  31%|███▏      | 10/32 [00:07<00:16,  1.32it/s, loss=6.46, v_num=2]
Epoch 10:  31%|███▏      | 10/32 [00:07<00:16,  1.32it/s, loss=6.46, v_num=2]
Epoch 10:  34%|███▍      | 11/32 [00:07<00:14,  1.44it/s, loss=6.46, v_num=2]
Epoch 10:  34%|███▍      | 11/32 [00:07<00:14,  1.44it/s, loss=6.46, v_num=2]
Epoch 10:  38%|███▊      | 12/32 [00:07<00:12,  1.56it/s, loss=6.46, v_num=2]
Epoch 10:  38%|███▊      | 12/32 [00:07<00:12,  1.56it/s, loss=6.46, v_num=2]
Epoch 10:  41%|████      | 13/32 [00:07<00:11,  1.68it/s, loss=6.46, v_num=2]
Epoch 10:  41%|████      | 13/32 [00:07<00:11,  1.68it/s, loss=6.47, v_num=2]
Epoch 10:  44%|████▍     | 14/32 [00:07<00:10,  1.80it/s, loss=6.47, v_num=2]
Epoch 10:  44%|████▍     | 14/32 [00:07<00:10,  1.80it/s, loss=6.47, v_num=2]
Epoch 10:  47%|████▋     | 15/32 [00:07<00:08,  1.91it/s, loss=6.47, v_num=2]
Epoch 10:  47%|████▋     | 15/32 [00:07<00:08,  1.91it/s, loss=6.47, v_num=2]
Epoch 10:  50%|█████     | 16/32 [00:07<00:07,  2.03it/s, loss=6.47, v_num=2]
Epoch 10:  50%|█████     | 16/32 [00:07<00:07,  2.03it/s, loss=6.47, v_num=2]
Epoch 10:  53%|█████▎    | 17/32 [00:10<00:09,  1.61it/s, loss=6.47, v_num=2]
Epoch 10:  53%|█████▎    | 17/32 [00:10<00:09,  1.61it/s, loss=6.48, v_num=2]
Epoch 10:  56%|█████▋    | 18/32 [00:10<00:08,  1.70it/s, loss=6.48, v_num=2]
Epoch 10:  56%|█████▋    | 18/32 [00:10<00:08,  1.69it/s, loss=6.48, v_num=2]
Epoch 10:  59%|█████▉    | 19/32 [00:10<00:07,  1.78it/s, loss=6.48, v_num=2]
Epoch 10:  59%|█████▉    | 19/32 [00:10<00:07,  1.78it/s, loss=6.48, v_num=2]
Epoch 10:  62%|██████▎   | 20/32 [00:10<00:06,  1.87it/s, loss=6.48, v_num=2]
Epoch 10:  62%|██████▎   | 20/32 [00:10<00:06,  1.87it/s, loss=6.48, v_num=2]
Epoch 10:  66%|██████▌   | 21/32 [00:10<00:05,  1.93it/s, loss=6.48, v_num=2]
Epoch 10:  66%|██████▌   | 21/32 [00:10<00:05,  1.93it/s, loss=6.48, v_num=2]
Epoch 10:  69%|██████▉   | 22/32 [00:10<00:04,  2.01it/s, loss=6.48, v_num=2]
Epoch 10:  69%|██████▉   | 22/32 [00:10<00:04,  2.01it/s, loss=6.49, v_num=2]
Epoch 10:  72%|███████▏  | 23/32 [00:11<00:04,  2.09it/s, loss=6.49, v_num=2]
Epoch 10:  72%|███████▏  | 23/32 [00:11<00:04,  2.09it/s, loss=6.49, v_num=2]
Epoch 10:  75%|███████▌  | 24/32 [00:11<00:03,  2.17it/s, loss=6.49, v_num=2]
Epoch 10:  75%|███████▌  | 24/32 [00:11<00:03,  2.17it/s, loss=6.49, v_num=2]
Epoch 10:  78%|███████▊  | 25/32 [00:14<00:04,  1.67it/s, loss=6.49, v_num=2]
Epoch 10:  78%|███████▊  | 25/32 [00:14<00:04,  1.67it/s, loss=6.49, v_num=2]
Epoch 10:  81%|████████▏ | 26/32 [00:15<00:03,  1.73it/s, loss=6.49, v_num=2]
Epoch 10:  81%|████████▏ | 26/32 [00:15<00:03,  1.73it/s, loss=6.5, v_num=2]
Epoch 10:  84%|████████▍ | 27/32 [00:15<00:02,  1.79it/s, loss=6.5, v_num=2]
Epoch 10:  84%|████████▍ | 27/32 [00:15<00:02,  1.79it/s, loss=6.5, v_num=2]
Epoch 10:  88%|████████▊ | 28/32 [00:15<00:02,  1.85it/s, loss=6.5, v_num=2]
Epoch 10:  88%|████████▊ | 28/32 [00:15<00:02,  1.85it/s, loss=6.5, v_num=2]
Epoch 10:  91%|█████████ | 29/32 [00:15<00:01,  1.89it/s, loss=6.5, v_num=2]
Epoch 10:  91%|█████████ | 29/32 [00:15<00:01,  1.89it/s, loss=6.5, v_num=2]
Epoch 10:  94%|█████████▍| 30/32 [00:15<00:01,  1.95it/s, loss=6.5, v_num=2]
Epoch 10:  94%|█████████▍| 30/32 [00:15<00:01,  1.95it/s, loss=6.51, v_num=2]
Epoch 10:  97%|█████████▋| 31/32 [00:15<00:00,  2.00it/s, loss=6.51, v_num=2]
Epoch 10:  97%|█████████▋| 31/32 [00:15<00:00,  2.00it/s, loss=6.51, v_num=2]
Epoch 10: 100%|██████████| 32/32 [00:15<00:00,  2.06it/s, loss=6.51, v_num=2]
Epoch 10: 100%|██████████| 32/32 [00:15<00:00,  2.06it/s, loss=6.51, v_num=2]
Epoch 10: 100%|██████████| 32/32 [00:15<00:00,  2.06it/s, loss=6.51, v_num=2]
Epoch 10:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.51, v_num=2]
Epoch 11:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.51, v_num=2]
Epoch 11:   3%|▎         | 1/32 [00:05<02:56,  5.71s/it, loss=6.51, v_num=2]
Epoch 11:   3%|▎         | 1/32 [00:05<02:56,  5.71s/it, loss=6.52, v_num=2]
Epoch 11:   6%|▋         | 2/32 [00:05<01:26,  2.89s/it, loss=6.52, v_num=2]
Epoch 11:   6%|▋         | 2/32 [00:05<01:26,  2.89s/it, loss=6.52, v_num=2]
Epoch 11:   9%|▉         | 3/32 [00:05<00:56,  1.94s/it, loss=6.52, v_num=2]
Epoch 11:   9%|▉         | 3/32 [00:05<00:56,  1.94s/it, loss=6.52, v_num=2]
Epoch 11:  12%|█▎        | 4/32 [00:05<00:41,  1.47s/it, loss=6.52, v_num=2]
Epoch 11:  12%|█▎        | 4/32 [00:05<00:41,  1.47s/it, loss=6.53, v_num=2]
Epoch 11:  16%|█▌        | 5/32 [00:05<00:32,  1.19s/it, loss=6.53, v_num=2]
Epoch 11:  16%|█▌        | 5/32 [00:05<00:32,  1.19s/it, loss=6.53, v_num=2]
Epoch 11:  19%|█▉        | 6/32 [00:05<00:25,  1.00it/s, loss=6.53, v_num=2]
Epoch 11:  19%|█▉        | 6/32 [00:05<00:25,  1.00it/s, loss=6.53, v_num=2]
Epoch 11:  22%|██▏       | 7/32 [00:06<00:21,  1.16it/s, loss=6.53, v_num=2]
Epoch 11:  22%|██▏       | 7/32 [00:06<00:21,  1.16it/s, loss=6.54, v_num=2]
Epoch 11:  25%|██▌       | 8/32 [00:06<00:18,  1.31it/s, loss=6.54, v_num=2]
Epoch 11:  25%|██▌       | 8/32 [00:06<00:18,  1.31it/s, loss=6.54, v_num=2]
Epoch 11:  28%|██▊       | 9/32 [00:09<00:23,  1.00s/it, loss=6.54, v_num=2]
Epoch 11:  28%|██▊       | 9/32 [00:09<00:23,  1.00s/it, loss=6.55, v_num=2]
Epoch 11:  31%|███▏      | 10/32 [00:09<00:19,  1.10it/s, loss=6.55, v_num=2]
Epoch 11:  31%|███▏      | 10/32 [00:09<00:19,  1.10it/s, loss=6.55, v_num=2]
Epoch 11:  34%|███▍      | 11/32 [00:09<00:17,  1.20it/s, loss=6.55, v_num=2]
Epoch 11:  34%|███▍      | 11/32 [00:09<00:17,  1.20it/s, loss=6.55, v_num=2]
Epoch 11:  38%|███▊      | 12/32 [00:09<00:15,  1.30it/s, loss=6.55, v_num=2]
Epoch 11:  38%|███▊      | 12/32 [00:09<00:15,  1.30it/s, loss=6.56, v_num=2]
Epoch 11:  41%|████      | 13/32 [00:09<00:13,  1.40it/s, loss=6.56, v_num=2]
Epoch 11:  41%|████      | 13/32 [00:09<00:13,  1.40it/s, loss=6.56, v_num=2]
Epoch 11:  44%|████▍     | 14/32 [00:09<00:12,  1.50it/s, loss=6.56, v_num=2]
Epoch 11:  44%|████▍     | 14/32 [00:09<00:12,  1.50it/s, loss=6.56, v_num=2]
Epoch 11:  47%|████▋     | 15/32 [00:09<00:10,  1.59it/s, loss=6.56, v_num=2]
Epoch 11:  47%|████▋     | 15/32 [00:09<00:10,  1.59it/s, loss=6.57, v_num=2]
Epoch 11:  50%|█████     | 16/32 [00:09<00:09,  1.69it/s, loss=6.57, v_num=2]
Epoch 11:  50%|█████     | 16/32 [00:09<00:09,  1.69it/s, loss=6.57, v_num=2]
Epoch 11:  53%|█████▎    | 17/32 [00:12<00:10,  1.37it/s, loss=6.57, v_num=2]
Epoch 11:  53%|█████▎    | 17/32 [00:12<00:10,  1.37it/s, loss=6.57, v_num=2]
Epoch 11:  56%|█████▋    | 18/32 [00:12<00:09,  1.42it/s, loss=6.57, v_num=2]
Epoch 11:  56%|█████▋    | 18/32 [00:12<00:09,  1.42it/s, loss=6.58, v_num=2]
Epoch 11:  59%|█████▉    | 19/32 [00:12<00:08,  1.49it/s, loss=6.58, v_num=2]
Epoch 11:  59%|█████▉    | 19/32 [00:12<00:08,  1.49it/s, loss=6.58, v_num=2]
Epoch 11:  62%|██████▎   | 20/32 [00:12<00:07,  1.56it/s, loss=6.58, v_num=2]
Epoch 11:  62%|██████▎   | 20/32 [00:12<00:07,  1.56it/s, loss=6.58, v_num=2]
Epoch 11:  66%|██████▌   | 21/32 [00:12<00:06,  1.63it/s, loss=6.58, v_num=2]
Epoch 11:  66%|██████▌   | 21/32 [00:12<00:06,  1.63it/s, loss=6.58, v_num=2]
Epoch 11:  69%|██████▉   | 22/32 [00:12<00:05,  1.70it/s, loss=6.58, v_num=2]
Epoch 11:  69%|██████▉   | 22/32 [00:12<00:05,  1.70it/s, loss=6.58, v_num=2]
Epoch 11:  72%|███████▏  | 23/32 [00:12<00:05,  1.77it/s, loss=6.58, v_num=2]
Epoch 11:  72%|███████▏  | 23/32 [00:12<00:05,  1.77it/s, loss=6.59, v_num=2]
Epoch 11:  75%|███████▌  | 24/32 [00:13<00:04,  1.84it/s, loss=6.59, v_num=2]
Epoch 11:  75%|███████▌  | 24/32 [00:13<00:04,  1.84it/s, loss=6.59, v_num=2]
Epoch 11:  78%|███████▊  | 25/32 [00:15<00:04,  1.60it/s, loss=6.59, v_num=2]
Epoch 11:  78%|███████▊  | 25/32 [00:15<00:04,  1.60it/s, loss=6.59, v_num=2]
Epoch 11:  81%|████████▏ | 26/32 [00:15<00:03,  1.65it/s, loss=6.59, v_num=2]
Epoch 11:  81%|████████▏ | 26/32 [00:15<00:03,  1.65it/s, loss=6.59, v_num=2]
Epoch 11:  84%|████████▍ | 27/32 [00:15<00:02,  1.71it/s, loss=6.59, v_num=2]
Epoch 11:  84%|████████▍ | 27/32 [00:15<00:02,  1.71it/s, loss=6.59, v_num=2]
Epoch 11:  88%|████████▊ | 28/32 [00:15<00:02,  1.77it/s, loss=6.59, v_num=2]
Epoch 11:  88%|████████▊ | 28/32 [00:15<00:02,  1.77it/s, loss=6.59, v_num=2]
Epoch 11:  91%|█████████ | 29/32 [00:15<00:01,  1.83it/s, loss=6.59, v_num=2]
Epoch 11:  91%|█████████ | 29/32 [00:15<00:01,  1.83it/s, loss=6.59, v_num=2]
Epoch 11:  94%|█████████▍| 30/32 [00:15<00:01,  1.89it/s, loss=6.59, v_num=2]
Epoch 11:  94%|█████████▍| 30/32 [00:15<00:01,  1.89it/s, loss=6.6, v_num=2]
Epoch 11:  97%|█████████▋| 31/32 [00:15<00:00,  1.94it/s, loss=6.6, v_num=2]
Epoch 11:  97%|█████████▋| 31/32 [00:15<00:00,  1.94it/s, loss=6.6, v_num=2]
Epoch 11: 100%|██████████| 32/32 [00:15<00:00,  2.00it/s, loss=6.6, v_num=2]
Epoch 11: 100%|██████████| 32/32 [00:15<00:00,  2.00it/s, loss=6.6, v_num=2]
Epoch 11: 100%|██████████| 32/32 [00:15<00:00,  2.00it/s, loss=6.6, v_num=2]
Epoch 11:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.6, v_num=2]
Epoch 12:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.6, v_num=2]
Epoch 12:   3%|▎         | 1/32 [00:03<01:59,  3.86s/it, loss=6.6, v_num=2]
Epoch 12:   3%|▎         | 1/32 [00:03<01:59,  3.87s/it, loss=6.6, v_num=2]
Epoch 12:   6%|▋         | 2/32 [00:03<00:58,  1.96s/it, loss=6.6, v_num=2]
Epoch 12:   6%|▋         | 2/32 [00:03<00:58,  1.96s/it, loss=6.6, v_num=2]
Epoch 12:   9%|▉         | 3/32 [00:03<00:38,  1.33s/it, loss=6.6, v_num=2]
Epoch 12:   9%|▉         | 3/32 [00:03<00:38,  1.33s/it, loss=6.6, v_num=2]
Epoch 12:  12%|█▎        | 4/32 [00:04<00:28,  1.01s/it, loss=6.6, v_num=2]
Epoch 12:  12%|█▎        | 4/32 [00:04<00:28,  1.01s/it, loss=6.6, v_num=2]
Epoch 12:  16%|█▌        | 5/32 [00:04<00:22,  1.22it/s, loss=6.6, v_num=2]
Epoch 12:  16%|█▌        | 5/32 [00:04<00:22,  1.22it/s, loss=6.6, v_num=2]
Epoch 12:  19%|█▉        | 6/32 [00:04<00:18,  1.44it/s, loss=6.6, v_num=2]
Epoch 12:  19%|█▉        | 6/32 [00:04<00:18,  1.44it/s, loss=6.6, v_num=2]
Epoch 12:  22%|██▏       | 7/32 [00:04<00:15,  1.66it/s, loss=6.6, v_num=2]
Epoch 12:  22%|██▏       | 7/32 [00:04<00:15,  1.66it/s, loss=6.61, v_num=2]
Epoch 12:  25%|██▌       | 8/32 [00:04<00:12,  1.88it/s, loss=6.61, v_num=2]
Epoch 12:  25%|██▌       | 8/32 [00:04<00:12,  1.88it/s, loss=6.61, v_num=2]
Epoch 12:  28%|██▊       | 9/32 [00:07<00:18,  1.25it/s, loss=6.61, v_num=2]
Epoch 12:  28%|██▊       | 9/32 [00:07<00:18,  1.25it/s, loss=6.61, v_num=2]
Epoch 12:  31%|███▏      | 10/32 [00:07<00:16,  1.37it/s, loss=6.61, v_num=2]
Epoch 12:  31%|███▏      | 10/32 [00:07<00:16,  1.37it/s, loss=6.61, v_num=2]
Epoch 12:  34%|███▍      | 11/32 [00:07<00:14,  1.48it/s, loss=6.61, v_num=2]
Epoch 12:  34%|███▍      | 11/32 [00:07<00:14,  1.48it/s, loss=6.61, v_num=2]
Epoch 12:  38%|███▊      | 12/32 [00:07<00:12,  1.61it/s, loss=6.61, v_num=2]
Epoch 12:  38%|███▊      | 12/32 [00:07<00:12,  1.61it/s, loss=6.61, v_num=2]
Epoch 12:  41%|████      | 13/32 [00:07<00:10,  1.73it/s, loss=6.61, v_num=2]
Epoch 12:  41%|████      | 13/32 [00:07<00:10,  1.73it/s, loss=6.61, v_num=2]
Epoch 12:  44%|████▍     | 14/32 [00:07<00:09,  1.85it/s, loss=6.61, v_num=2]
Epoch 12:  44%|████▍     | 14/32 [00:07<00:09,  1.85it/s, loss=6.62, v_num=2]
Epoch 12:  47%|████▋     | 15/32 [00:07<00:08,  1.97it/s, loss=6.62, v_num=2]
Epoch 12:  47%|████▋     | 15/32 [00:07<00:08,  1.97it/s, loss=6.62, v_num=2]
Epoch 12:  50%|█████     | 16/32 [00:07<00:07,  2.09it/s, loss=6.62, v_num=2]
Epoch 12:  50%|█████     | 16/32 [00:07<00:07,  2.09it/s, loss=6.62, v_num=2]
Epoch 12:  53%|█████▎    | 17/32 [00:10<00:09,  1.64it/s, loss=6.62, v_num=2]
Epoch 12:  53%|█████▎    | 17/32 [00:10<00:09,  1.64it/s, loss=6.62, v_num=2]
Epoch 12:  56%|█████▋    | 18/32 [00:10<00:08,  1.73it/s, loss=6.62, v_num=2]
Epoch 12:  56%|█████▋    | 18/32 [00:10<00:08,  1.73it/s, loss=6.62, v_num=2]
Epoch 12:  59%|█████▉    | 19/32 [00:10<00:07,  1.81it/s, loss=6.62, v_num=2]
Epoch 12:  59%|█████▉    | 19/32 [00:10<00:07,  1.81it/s, loss=6.63, v_num=2]
Epoch 12:  62%|██████▎   | 20/32 [00:10<00:06,  1.89it/s, loss=6.63, v_num=2]
Epoch 12:  62%|██████▎   | 20/32 [00:10<00:06,  1.89it/s, loss=6.63, v_num=2]
Epoch 12:  66%|██████▌   | 21/32 [00:10<00:05,  1.98it/s, loss=6.63, v_num=2]
Epoch 12:  66%|██████▌   | 21/32 [00:10<00:05,  1.98it/s, loss=6.63, v_num=2]
Epoch 12:  69%|██████▉   | 22/32 [00:10<00:04,  2.06it/s, loss=6.63, v_num=2]
Epoch 12:  69%|██████▉   | 22/32 [00:10<00:04,  2.06it/s, loss=6.63, v_num=2]
Epoch 12:  72%|███████▏  | 23/32 [00:10<00:04,  2.15it/s, loss=6.63, v_num=2]
Epoch 12:  72%|███████▏  | 23/32 [00:10<00:04,  2.15it/s, loss=6.63, v_num=2]
Epoch 12:  75%|███████▌  | 24/32 [00:10<00:03,  2.23it/s, loss=6.63, v_num=2]
Epoch 12:  75%|███████▌  | 24/32 [00:10<00:03,  2.23it/s, loss=6.64, v_num=2]
Epoch 12:  78%|███████▊  | 25/32 [00:13<00:03,  1.83it/s, loss=6.64, v_num=2]
Epoch 12:  78%|███████▊  | 25/32 [00:13<00:03,  1.83it/s, loss=6.64, v_num=2]
Epoch 12:  81%|████████▏ | 26/32 [00:13<00:03,  1.89it/s, loss=6.64, v_num=2]
Epoch 12:  81%|████████▏ | 26/32 [00:13<00:03,  1.89it/s, loss=6.64, v_num=2]
Epoch 12:  84%|████████▍ | 27/32 [00:13<00:02,  1.93it/s, loss=6.64, v_num=2]
Epoch 12:  84%|████████▍ | 27/32 [00:13<00:02,  1.93it/s, loss=6.64, v_num=2]
Epoch 12:  88%|████████▊ | 28/32 [00:14<00:02,  2.00it/s, loss=6.64, v_num=2]
Epoch 12:  88%|████████▊ | 28/32 [00:14<00:02,  2.00it/s, loss=6.65, v_num=2]
Epoch 12:  91%|█████████ | 29/32 [00:14<00:01,  2.06it/s, loss=6.65, v_num=2]
Epoch 12:  91%|█████████ | 29/32 [00:14<00:01,  2.06it/s, loss=6.65, v_num=2]
Epoch 12:  94%|█████████▍| 30/32 [00:14<00:00,  2.12it/s, loss=6.65, v_num=2]
Epoch 12:  94%|█████████▍| 30/32 [00:14<00:00,  2.12it/s, loss=6.65, v_num=2]
Epoch 12:  97%|█████████▋| 31/32 [00:14<00:00,  2.19it/s, loss=6.65, v_num=2]
Epoch 12:  97%|█████████▋| 31/32 [00:14<00:00,  2.19it/s, loss=6.65, v_num=2]
Epoch 12: 100%|██████████| 32/32 [00:14<00:00,  2.25it/s, loss=6.65, v_num=2]
Epoch 12: 100%|██████████| 32/32 [00:14<00:00,  2.25it/s, loss=6.66, v_num=2]
Epoch 12: 100%|██████████| 32/32 [00:14<00:00,  2.25it/s, loss=6.66, v_num=2]
Epoch 12:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.66, v_num=2]
Epoch 13:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.66, v_num=2]
Epoch 13:   3%|▎         | 1/32 [00:03<02:03,  3.99s/it, loss=6.66, v_num=2]
Epoch 13:   3%|▎         | 1/32 [00:03<02:03,  3.99s/it, loss=6.66, v_num=2]
Epoch 13:   6%|▋         | 2/32 [00:04<01:00,  2.02s/it, loss=6.66, v_num=2]
Epoch 13:   6%|▋         | 2/32 [00:04<01:00,  2.02s/it, loss=6.66, v_num=2]
Epoch 13:   9%|▉         | 3/32 [00:04<00:39,  1.37s/it, loss=6.66, v_num=2]
Epoch 13:   9%|▉         | 3/32 [00:04<00:39,  1.37s/it, loss=6.66, v_num=2]
Epoch 13:  12%|█▎        | 4/32 [00:04<00:29,  1.04s/it, loss=6.66, v_num=2]
Epoch 13:  12%|█▎        | 4/32 [00:04<00:29,  1.04s/it, loss=6.67, v_num=2]
Epoch 13:  16%|█▌        | 5/32 [00:04<00:22,  1.19it/s, loss=6.67, v_num=2]
Epoch 13:  16%|█▌        | 5/32 [00:04<00:22,  1.19it/s, loss=6.67, v_num=2]
Epoch 13:  19%|█▉        | 6/32 [00:04<00:18,  1.41it/s, loss=6.67, v_num=2]
Epoch 13:  19%|█▉        | 6/32 [00:04<00:18,  1.41it/s, loss=6.67, v_num=2]
Epoch 13:  22%|██▏       | 7/32 [00:04<00:15,  1.63it/s, loss=6.67, v_num=2]
Epoch 13:  22%|██▏       | 7/32 [00:04<00:15,  1.63it/s, loss=6.67, v_num=2]
Epoch 13:  25%|██▌       | 8/32 [00:04<00:13,  1.84it/s, loss=6.67, v_num=2]
Epoch 13:  25%|██▌       | 8/32 [00:04<00:13,  1.84it/s, loss=6.68, v_num=2]
Epoch 13:  28%|██▊       | 9/32 [00:07<00:18,  1.26it/s, loss=6.68, v_num=2]
Epoch 13:  28%|██▊       | 9/32 [00:07<00:18,  1.26it/s, loss=6.68, v_num=2]
Epoch 13:  31%|███▏      | 10/32 [00:07<00:15,  1.39it/s, loss=6.68, v_num=2]
Epoch 13:  31%|███▏      | 10/32 [00:07<00:15,  1.39it/s, loss=6.68, v_num=2]
Epoch 13:  34%|███▍      | 11/32 [00:07<00:13,  1.52it/s, loss=6.68, v_num=2]
Epoch 13:  34%|███▍      | 11/32 [00:07<00:13,  1.52it/s, loss=6.69, v_num=2]
Epoch 13:  38%|███▊      | 12/32 [00:07<00:12,  1.65it/s, loss=6.69, v_num=2]
Epoch 13:  38%|███▊      | 12/32 [00:07<00:12,  1.65it/s, loss=6.69, v_num=2]
Epoch 13:  41%|████      | 13/32 [00:07<00:10,  1.77it/s, loss=6.69, v_num=2]
Epoch 13:  41%|████      | 13/32 [00:07<00:10,  1.77it/s, loss=6.69, v_num=2]
Epoch 13:  44%|████▍     | 14/32 [00:07<00:09,  1.90it/s, loss=6.69, v_num=2]
Epoch 13:  44%|████▍     | 14/32 [00:07<00:09,  1.90it/s, loss=6.7, v_num=2]
Epoch 13:  47%|████▋     | 15/32 [00:07<00:08,  2.02it/s, loss=6.7, v_num=2]
Epoch 13:  47%|████▋     | 15/32 [00:07<00:08,  2.02it/s, loss=6.7, v_num=2]
Epoch 13:  50%|█████     | 16/32 [00:07<00:07,  2.14it/s, loss=6.7, v_num=2]
Epoch 13:  50%|█████     | 16/32 [00:07<00:07,  2.14it/s, loss=6.7, v_num=2]
Epoch 13:  53%|█████▎    | 17/32 [00:10<00:09,  1.66it/s, loss=6.7, v_num=2]
Epoch 13:  53%|█████▎    | 17/32 [00:10<00:09,  1.66it/s, loss=6.71, v_num=2]
Epoch 13:  56%|█████▋    | 18/32 [00:10<00:07,  1.75it/s, loss=6.71, v_num=2]
Epoch 13:  56%|█████▋    | 18/32 [00:10<00:07,  1.75it/s, loss=6.71, v_num=2]
Epoch 13:  59%|█████▉    | 19/32 [00:10<00:07,  1.84it/s, loss=6.71, v_num=2]
Epoch 13:  59%|█████▉    | 19/32 [00:10<00:07,  1.84it/s, loss=6.71, v_num=2]
Epoch 13:  62%|██████▎   | 20/32 [00:10<00:06,  1.93it/s, loss=6.71, v_num=2]
Epoch 13:  62%|██████▎   | 20/32 [00:10<00:06,  1.93it/s, loss=6.72, v_num=2]
Epoch 13:  66%|██████▌   | 21/32 [00:10<00:05,  2.02it/s, loss=6.72, v_num=2]
Epoch 13:  66%|██████▌   | 21/32 [00:10<00:05,  2.02it/s, loss=6.72, v_num=2]
Epoch 13:  69%|██████▉   | 22/32 [00:10<00:04,  2.10it/s, loss=6.72, v_num=2]
Epoch 13:  69%|██████▉   | 22/32 [00:10<00:04,  2.10it/s, loss=6.73, v_num=2]
Epoch 13:  72%|███████▏  | 23/32 [00:10<00:04,  2.19it/s, loss=6.73, v_num=2]
Epoch 13:  72%|███████▏  | 23/32 [00:10<00:04,  2.19it/s, loss=6.73, v_num=2]
Epoch 13:  75%|███████▌  | 24/32 [00:10<00:03,  2.27it/s, loss=6.73, v_num=2]
Epoch 13:  75%|███████▌  | 24/32 [00:10<00:03,  2.27it/s, loss=6.73, v_num=2]
Epoch 13:  78%|███████▊  | 25/32 [00:13<00:03,  1.88it/s, loss=6.73, v_num=2]
Epoch 13:  78%|███████▊  | 25/32 [00:13<00:03,  1.88it/s, loss=6.74, v_num=2]
Epoch 13:  81%|████████▏ | 26/32 [00:13<00:03,  1.95it/s, loss=6.74, v_num=2]
Epoch 13:  81%|████████▏ | 26/32 [00:13<00:03,  1.95it/s, loss=6.74, v_num=2]
Epoch 13:  84%|████████▍ | 27/32 [00:13<00:02,  2.02it/s, loss=6.74, v_num=2]
Epoch 13:  84%|████████▍ | 27/32 [00:13<00:02,  2.02it/s, loss=6.75, v_num=2]
Epoch 13:  88%|████████▊ | 28/32 [00:13<00:01,  2.08it/s, loss=6.75, v_num=2]
Epoch 13:  88%|████████▊ | 28/32 [00:13<00:01,  2.08it/s, loss=6.75, v_num=2]
Epoch 13:  91%|█████████ | 29/32 [00:13<00:01,  2.15it/s, loss=6.75, v_num=2]
Epoch 13:  91%|█████████ | 29/32 [00:13<00:01,  2.15it/s, loss=6.75, v_num=2]
Epoch 13:  94%|█████████▍| 30/32 [00:13<00:00,  2.22it/s, loss=6.75, v_num=2]
Epoch 13:  94%|█████████▍| 30/32 [00:13<00:00,  2.22it/s, loss=6.76, v_num=2]
Epoch 13:  97%|█████████▋| 31/32 [00:13<00:00,  2.28it/s, loss=6.76, v_num=2]
Epoch 13:  97%|█████████▋| 31/32 [00:13<00:00,  2.28it/s, loss=6.76, v_num=2]
Epoch 13: 100%|██████████| 32/32 [00:13<00:00,  2.35it/s, loss=6.76, v_num=2]
Epoch 13: 100%|██████████| 32/32 [00:13<00:00,  2.35it/s, loss=6.77, v_num=2]
Epoch 13: 100%|██████████| 32/32 [00:13<00:00,  2.35it/s, loss=6.77, v_num=2]
Epoch 13:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.77, v_num=2]
Epoch 14:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.77, v_num=2]
Epoch 14:   3%|▎         | 1/32 [00:04<02:25,  4.70s/it, loss=6.77, v_num=2]
Epoch 14:   3%|▎         | 1/32 [00:04<02:25,  4.70s/it, loss=6.77, v_num=2]
Epoch 14:   6%|▋         | 2/32 [00:04<01:11,  2.38s/it, loss=6.77, v_num=2]
Epoch 14:   6%|▋         | 2/32 [00:04<01:11,  2.38s/it, loss=6.77, v_num=2]
Epoch 14:   9%|▉         | 3/32 [00:04<00:46,  1.61s/it, loss=6.77, v_num=2]
Epoch 14:   9%|▉         | 3/32 [00:04<00:46,  1.61s/it, loss=6.78, v_num=2]
Epoch 14:  12%|█▎        | 4/32 [00:04<00:34,  1.22s/it, loss=6.78, v_num=2]
Epoch 14:  12%|█▎        | 4/32 [00:04<00:34,  1.22s/it, loss=6.78, v_num=2]
Epoch 14:  16%|█▌        | 5/32 [00:04<00:26,  1.01it/s, loss=6.78, v_num=2]
Epoch 14:  16%|█▌        | 5/32 [00:04<00:26,  1.01it/s, loss=6.78, v_num=2]
Epoch 14:  19%|█▉        | 6/32 [00:04<00:21,  1.20it/s, loss=6.78, v_num=2]
Epoch 14:  19%|█▉        | 6/32 [00:04<00:21,  1.20it/s, loss=6.79, v_num=2]
Epoch 14:  22%|██▏       | 7/32 [00:05<00:18,  1.38it/s, loss=6.79, v_num=2]
Epoch 14:  22%|██▏       | 7/32 [00:05<00:18,  1.38it/s, loss=6.79, v_num=2]
Epoch 14:  25%|██▌       | 8/32 [00:05<00:15,  1.56it/s, loss=6.79, v_num=2]
Epoch 14:  25%|██▌       | 8/32 [00:05<00:15,  1.56it/s, loss=6.79, v_num=2]
Epoch 14:  28%|██▊       | 9/32 [00:08<00:21,  1.05it/s, loss=6.79, v_num=2]
Epoch 14:  28%|██▊       | 9/32 [00:08<00:21,  1.05it/s, loss=6.8, v_num=2]
Epoch 14:  31%|███▏      | 10/32 [00:08<00:18,  1.16it/s, loss=6.8, v_num=2]
Epoch 14:  31%|███▏      | 10/32 [00:08<00:18,  1.16it/s, loss=6.8, v_num=2]
Epoch 14:  34%|███▍      | 11/32 [00:08<00:16,  1.27it/s, loss=6.8, v_num=2]
Epoch 14:  34%|███▍      | 11/32 [00:08<00:16,  1.27it/s, loss=6.8, v_num=2]
Epoch 14:  38%|███▊      | 12/32 [00:08<00:14,  1.38it/s, loss=6.8, v_num=2]
Epoch 14:  38%|███▊      | 12/32 [00:08<00:14,  1.38it/s, loss=6.81, v_num=2]
Epoch 14:  41%|████      | 13/32 [00:08<00:12,  1.48it/s, loss=6.81, v_num=2]
Epoch 14:  41%|████      | 13/32 [00:08<00:12,  1.48it/s, loss=6.81, v_num=2]
Epoch 14:  44%|████▍     | 14/32 [00:08<00:11,  1.59it/s, loss=6.81, v_num=2]
Epoch 14:  44%|████▍     | 14/32 [00:08<00:11,  1.59it/s, loss=6.81, v_num=2]
Epoch 14:  47%|████▋     | 15/32 [00:08<00:10,  1.69it/s, loss=6.81, v_num=2]
Epoch 14:  47%|████▋     | 15/32 [00:08<00:10,  1.69it/s, loss=6.81, v_num=2]
Epoch 14:  50%|█████     | 16/32 [00:08<00:08,  1.79it/s, loss=6.81, v_num=2]
Epoch 14:  50%|█████     | 16/32 [00:08<00:08,  1.79it/s, loss=6.82, v_num=2]
Epoch 14:  53%|█████▎    | 17/32 [00:11<00:10,  1.44it/s, loss=6.82, v_num=2]
Epoch 14:  53%|█████▎    | 17/32 [00:11<00:10,  1.44it/s, loss=6.82, v_num=2]
Epoch 14:  56%|█████▋    | 18/32 [00:11<00:09,  1.51it/s, loss=6.82, v_num=2]
Epoch 14:  56%|█████▋    | 18/32 [00:11<00:09,  1.51it/s, loss=6.82, v_num=2]
Epoch 14:  59%|█████▉    | 19/32 [00:12<00:08,  1.58it/s, loss=6.82, v_num=2]
Epoch 14:  59%|█████▉    | 19/32 [00:12<00:08,  1.58it/s, loss=6.83, v_num=2]
Epoch 14:  62%|██████▎   | 20/32 [00:12<00:07,  1.66it/s, loss=6.83, v_num=2]
Epoch 14:  62%|██████▎   | 20/32 [00:12<00:07,  1.66it/s, loss=6.83, v_num=2]
Epoch 14:  66%|██████▌   | 21/32 [00:12<00:06,  1.73it/s, loss=6.83, v_num=2]
Epoch 14:  66%|██████▌   | 21/32 [00:12<00:06,  1.73it/s, loss=6.83, v_num=2]
Epoch 14:  69%|██████▉   | 22/32 [00:12<00:05,  1.81it/s, loss=6.83, v_num=2]
Epoch 14:  69%|██████▉   | 22/32 [00:12<00:05,  1.81it/s, loss=6.83, v_num=2]
Epoch 14:  72%|███████▏  | 23/32 [00:12<00:04,  1.88it/s, loss=6.83, v_num=2]
Epoch 14:  72%|███████▏  | 23/32 [00:12<00:04,  1.88it/s, loss=6.84, v_num=2]
Epoch 14:  75%|███████▌  | 24/32 [00:12<00:04,  1.95it/s, loss=6.84, v_num=2]
Epoch 14:  75%|███████▌  | 24/32 [00:12<00:04,  1.95it/s, loss=6.84, v_num=2]
Epoch 14:  78%|███████▊  | 25/32 [00:15<00:04,  1.65it/s, loss=6.84, v_num=2]
Epoch 14:  78%|███████▊  | 25/32 [00:15<00:04,  1.65it/s, loss=6.84, v_num=2]
Epoch 14:  81%|████████▏ | 26/32 [00:15<00:03,  1.69it/s, loss=6.84, v_num=2]
Epoch 14:  81%|████████▏ | 26/32 [00:15<00:03,  1.69it/s, loss=6.84, v_num=2]
Epoch 14:  84%|████████▍ | 27/32 [00:15<00:02,  1.74it/s, loss=6.84, v_num=2]
Epoch 14:  84%|████████▍ | 27/32 [00:15<00:02,  1.74it/s, loss=6.85, v_num=2]
Epoch 14:  88%|████████▊ | 28/32 [00:15<00:02,  1.80it/s, loss=6.85, v_num=2]
Epoch 14:  88%|████████▊ | 28/32 [00:15<00:02,  1.80it/s, loss=6.85, v_num=2]
Epoch 14:  91%|█████████ | 29/32 [00:15<00:01,  1.86it/s, loss=6.85, v_num=2]
Epoch 14:  91%|█████████ | 29/32 [00:15<00:01,  1.86it/s, loss=6.85, v_num=2]
Epoch 14:  94%|█████████▍| 30/32 [00:15<00:01,  1.92it/s, loss=6.85, v_num=2]
Epoch 14:  94%|█████████▍| 30/32 [00:15<00:01,  1.92it/s, loss=6.85, v_num=2]
Epoch 14:  97%|█████████▋| 31/32 [00:15<00:00,  1.98it/s, loss=6.85, v_num=2]
Epoch 14:  97%|█████████▋| 31/32 [00:15<00:00,  1.98it/s, loss=6.86, v_num=2]
Epoch 14: 100%|██████████| 32/32 [00:15<00:00,  2.04it/s, loss=6.86, v_num=2]
Epoch 14: 100%|██████████| 32/32 [00:15<00:00,  2.04it/s, loss=6.86, v_num=2]
Epoch 14: 100%|██████████| 32/32 [00:15<00:00,  2.04it/s, loss=6.86, v_num=2]
Epoch 14:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.86, v_num=2]
Epoch 15:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.86, v_num=2]
Epoch 15:   3%|▎         | 1/32 [00:03<01:58,  3.82s/it, loss=6.86, v_num=2]
Epoch 15:   3%|▎         | 1/32 [00:03<01:58,  3.82s/it, loss=6.86, v_num=2]
Epoch 15:   6%|▋         | 2/32 [00:03<00:58,  1.94s/it, loss=6.86, v_num=2]
Epoch 15:   6%|▋         | 2/32 [00:03<00:58,  1.94s/it, loss=6.86, v_num=2]
Epoch 15:   9%|▉         | 3/32 [00:03<00:38,  1.31s/it, loss=6.86, v_num=2]
Epoch 15:   9%|▉         | 3/32 [00:03<00:38,  1.31s/it, loss=6.86, v_num=2]
Epoch 15:  12%|█▎        | 4/32 [00:04<00:28,  1.00s/it, loss=6.86, v_num=2]
Epoch 15:  12%|█▎        | 4/32 [00:04<00:28,  1.00s/it, loss=6.87, v_num=2]
Epoch 15:  16%|█▌        | 5/32 [00:04<00:21,  1.23it/s, loss=6.87, v_num=2]
Epoch 15:  16%|█▌        | 5/32 [00:04<00:21,  1.23it/s, loss=6.87, v_num=2]
Epoch 15:  19%|█▉        | 6/32 [00:04<00:17,  1.45it/s, loss=6.87, v_num=2]
Epoch 15:  19%|█▉        | 6/32 [00:04<00:17,  1.45it/s, loss=6.87, v_num=2]
Epoch 15:  22%|██▏       | 7/32 [00:04<00:14,  1.67it/s, loss=6.87, v_num=2]
Epoch 15:  22%|██▏       | 7/32 [00:04<00:14,  1.67it/s, loss=6.87, v_num=2]
Epoch 15:  25%|██▌       | 8/32 [00:04<00:12,  1.89it/s, loss=6.87, v_num=2]
Epoch 15:  25%|██▌       | 8/32 [00:04<00:12,  1.89it/s, loss=6.88, v_num=2]
Epoch 15:  28%|██▊       | 9/32 [00:07<00:18,  1.27it/s, loss=6.88, v_num=2]
Epoch 15:  28%|██▊       | 9/32 [00:07<00:18,  1.27it/s, loss=6.88, v_num=2]
Epoch 15:  31%|███▏      | 10/32 [00:07<00:15,  1.40it/s, loss=6.88, v_num=2]
Epoch 15:  31%|███▏      | 10/32 [00:07<00:15,  1.40it/s, loss=6.88, v_num=2]
Epoch 15:  34%|███▍      | 11/32 [00:07<00:13,  1.52it/s, loss=6.88, v_num=2]
Epoch 15:  34%|███▍      | 11/32 [00:07<00:13,  1.52it/s, loss=6.88, v_num=2]
Epoch 15:  38%|███▊      | 12/32 [00:07<00:12,  1.65it/s, loss=6.88, v_num=2]
Epoch 15:  38%|███▊      | 12/32 [00:07<00:12,  1.65it/s, loss=6.88, v_num=2]
Epoch 15:  41%|████      | 13/32 [00:07<00:10,  1.76it/s, loss=6.88, v_num=2]
Epoch 15:  41%|████      | 13/32 [00:07<00:10,  1.76it/s, loss=6.89, v_num=2]
Epoch 15:  44%|████▍     | 14/32 [00:07<00:09,  1.88it/s, loss=6.89, v_num=2]
Epoch 15:  44%|████▍     | 14/32 [00:07<00:09,  1.88it/s, loss=6.89, v_num=2]
Epoch 15:  47%|████▋     | 15/32 [00:07<00:08,  1.99it/s, loss=6.89, v_num=2]
Epoch 15:  47%|████▋     | 15/32 [00:07<00:08,  1.99it/s, loss=6.89, v_num=2]
Epoch 15:  50%|█████     | 16/32 [00:07<00:07,  2.11it/s, loss=6.89, v_num=2]
Epoch 15:  50%|█████     | 16/32 [00:07<00:07,  2.11it/s, loss=6.89, v_num=2]
Epoch 15:  53%|█████▎    | 17/32 [00:11<00:10,  1.47it/s, loss=6.89, v_num=2]
Epoch 15:  53%|█████▎    | 17/32 [00:11<00:10,  1.47it/s, loss=6.9, v_num=2]
Epoch 15:  56%|█████▋    | 18/32 [00:11<00:09,  1.55it/s, loss=6.9, v_num=2]
Epoch 15:  56%|█████▋    | 18/32 [00:11<00:09,  1.55it/s, loss=6.9, v_num=2]
Epoch 15:  59%|█████▉    | 19/32 [00:11<00:07,  1.63it/s, loss=6.9, v_num=2]
Epoch 15:  59%|█████▉    | 19/32 [00:11<00:07,  1.63it/s, loss=6.9, v_num=2]
Epoch 15:  62%|██████▎   | 20/32 [00:11<00:07,  1.70it/s, loss=6.9, v_num=2]
Epoch 15:  62%|██████▎   | 20/32 [00:11<00:07,  1.70it/s, loss=6.9, v_num=2]
Epoch 15:  66%|██████▌   | 21/32 [00:11<00:06,  1.77it/s, loss=6.9, v_num=2]
Epoch 15:  66%|██████▌   | 21/32 [00:11<00:06,  1.77it/s, loss=6.91, v_num=2]
Epoch 15:  69%|██████▉   | 22/32 [00:11<00:05,  1.84it/s, loss=6.91, v_num=2]
Epoch 15:  69%|██████▉   | 22/32 [00:11<00:05,  1.84it/s, loss=6.91, v_num=2]
Epoch 15:  72%|███████▏  | 23/32 [00:12<00:04,  1.91it/s, loss=6.91, v_num=2]
Epoch 15:  72%|███████▏  | 23/32 [00:12<00:04,  1.91it/s, loss=6.91, v_num=2]
Epoch 15:  75%|███████▌  | 24/32 [00:12<00:04,  1.99it/s, loss=6.91, v_num=2]
Epoch 15:  75%|███████▌  | 24/32 [00:12<00:04,  1.99it/s, loss=6.91, v_num=2]
Epoch 15:  78%|███████▊  | 25/32 [00:15<00:04,  1.63it/s, loss=6.91, v_num=2]
Epoch 15:  78%|███████▊  | 25/32 [00:15<00:04,  1.63it/s, loss=6.92, v_num=2]
Epoch 15:  81%|████████▏ | 26/32 [00:15<00:03,  1.67it/s, loss=6.92, v_num=2]
Epoch 15:  81%|████████▏ | 26/32 [00:15<00:03,  1.67it/s, loss=6.92, v_num=2]
Epoch 15:  84%|████████▍ | 27/32 [00:15<00:02,  1.73it/s, loss=6.92, v_num=2]
Epoch 15:  84%|████████▍ | 27/32 [00:15<00:02,  1.73it/s, loss=6.92, v_num=2]
Epoch 15:  88%|████████▊ | 28/32 [00:15<00:02,  1.79it/s, loss=6.92, v_num=2]
Epoch 15:  88%|████████▊ | 28/32 [00:15<00:02,  1.79it/s, loss=6.92, v_num=2]
Epoch 15:  91%|█████████ | 29/32 [00:15<00:01,  1.85it/s, loss=6.92, v_num=2]
Epoch 15:  91%|█████████ | 29/32 [00:15<00:01,  1.85it/s, loss=6.92, v_num=2]
Epoch 15:  94%|█████████▍| 30/32 [00:15<00:01,  1.90it/s, loss=6.92, v_num=2]
Epoch 15:  94%|█████████▍| 30/32 [00:15<00:01,  1.90it/s, loss=6.93, v_num=2]
Epoch 15:  97%|█████████▋| 31/32 [00:15<00:00,  1.96it/s, loss=6.93, v_num=2]
Epoch 15:  97%|█████████▋| 31/32 [00:15<00:00,  1.96it/s, loss=6.93, v_num=2]
Epoch 15: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.93, v_num=2]
Epoch 15: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.93, v_num=2]
Epoch 15: 100%|██████████| 32/32 [00:15<00:00,  2.02it/s, loss=6.93, v_num=2]
Epoch 15:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.93, v_num=2]
Epoch 16:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.93, v_num=2]
Epoch 16:   3%|▎         | 1/32 [00:04<02:06,  4.09s/it, loss=6.93, v_num=2]
Epoch 16:   3%|▎         | 1/32 [00:04<02:06,  4.09s/it, loss=6.93, v_num=2]
Epoch 16:   6%|▋         | 2/32 [00:04<01:02,  2.07s/it, loss=6.93, v_num=2]
Epoch 16:   6%|▋         | 2/32 [00:04<01:02,  2.08s/it, loss=6.94, v_num=2]
Epoch 16:   9%|▉         | 3/32 [00:04<00:44,  1.52s/it, loss=6.94, v_num=2]
Epoch 16:   9%|▉         | 3/32 [00:04<00:44,  1.52s/it, loss=6.94, v_num=2]
Epoch 16:  12%|█▎        | 4/32 [00:04<00:32,  1.15s/it, loss=6.94, v_num=2]
Epoch 16:  12%|█▎        | 4/32 [00:04<00:32,  1.16s/it, loss=6.94, v_num=2]
Epoch 16:  16%|█▌        | 5/32 [00:04<00:25,  1.07it/s, loss=6.94, v_num=2]
Epoch 16:  16%|█▌        | 5/32 [00:04<00:25,  1.07it/s, loss=6.94, v_num=2]
Epoch 16:  19%|█▉        | 6/32 [00:04<00:20,  1.27it/s, loss=6.94, v_num=2]
Epoch 16:  19%|█▉        | 6/32 [00:04<00:20,  1.27it/s, loss=6.94, v_num=2]
Epoch 16:  22%|██▏       | 7/32 [00:04<00:17,  1.46it/s, loss=6.94, v_num=2]
Epoch 16:  22%|██▏       | 7/32 [00:04<00:17,  1.46it/s, loss=6.95, v_num=2]
Epoch 16:  25%|██▌       | 8/32 [00:04<00:14,  1.65it/s, loss=6.95, v_num=2]
Epoch 16:  25%|██▌       | 8/32 [00:04<00:14,  1.65it/s, loss=6.95, v_num=2]
Epoch 16:  28%|██▊       | 9/32 [00:07<00:18,  1.22it/s, loss=6.95, v_num=2]
Epoch 16:  28%|██▊       | 9/32 [00:07<00:18,  1.22it/s, loss=6.95, v_num=2]
Epoch 16:  31%|███▏      | 10/32 [00:07<00:16,  1.35it/s, loss=6.95, v_num=2]
Epoch 16:  31%|███▏      | 10/32 [00:07<00:16,  1.35it/s, loss=6.95, v_num=2]
Epoch 16:  34%|███▍      | 11/32 [00:07<00:15,  1.38it/s, loss=6.95, v_num=2]
Epoch 16:  34%|███▍      | 11/32 [00:07<00:15,  1.38it/s, loss=6.95, v_num=2]
Epoch 16:  38%|███▊      | 12/32 [00:07<00:13,  1.50it/s, loss=6.95, v_num=2]
Epoch 16:  38%|███▊      | 12/32 [00:07<00:13,  1.50it/s, loss=6.96, v_num=2]
Epoch 16:  41%|████      | 13/32 [00:08<00:11,  1.62it/s, loss=6.96, v_num=2]
Epoch 16:  41%|████      | 13/32 [00:08<00:11,  1.62it/s, loss=6.96, v_num=2]
Epoch 16:  44%|████▍     | 14/32 [00:08<00:10,  1.73it/s, loss=6.96, v_num=2]
Epoch 16:  44%|████▍     | 14/32 [00:08<00:10,  1.73it/s, loss=6.96, v_num=2]
Epoch 16:  47%|████▋     | 15/32 [00:08<00:09,  1.84it/s, loss=6.96, v_num=2]
Epoch 16:  47%|████▋     | 15/32 [00:08<00:09,  1.84it/s, loss=6.96, v_num=2]
Epoch 16:  50%|█████     | 16/32 [00:08<00:08,  1.95it/s, loss=6.96, v_num=2]
Epoch 16:  50%|█████     | 16/32 [00:08<00:08,  1.95it/s, loss=6.97, v_num=2]
Epoch 16:  53%|█████▎    | 17/32 [00:10<00:09,  1.61it/s, loss=6.97, v_num=2]
Epoch 16:  53%|█████▎    | 17/32 [00:10<00:09,  1.61it/s, loss=6.97, v_num=2]
Epoch 16:  56%|█████▋    | 18/32 [00:10<00:08,  1.69it/s, loss=6.97, v_num=2]
Epoch 16:  56%|█████▋    | 18/32 [00:10<00:08,  1.69it/s, loss=6.97, v_num=2]
Epoch 16:  59%|█████▉    | 19/32 [00:11<00:07,  1.72it/s, loss=6.97, v_num=2]
Epoch 16:  59%|█████▉    | 19/32 [00:11<00:07,  1.72it/s, loss=6.97, v_num=2]
Epoch 16:  62%|██████▎   | 20/32 [00:11<00:06,  1.80it/s, loss=6.97, v_num=2]
Epoch 16:  62%|██████▎   | 20/32 [00:11<00:06,  1.80it/s, loss=6.97, v_num=2]
Epoch 16:  66%|██████▌   | 21/32 [00:11<00:05,  1.88it/s, loss=6.97, v_num=2]
Epoch 16:  66%|██████▌   | 21/32 [00:11<00:05,  1.88it/s, loss=6.98, v_num=2]
Epoch 16:  69%|██████▉   | 22/32 [00:11<00:05,  1.96it/s, loss=6.98, v_num=2]
Epoch 16:  69%|██████▉   | 22/32 [00:11<00:05,  1.96it/s, loss=6.98, v_num=2]
Epoch 16:  72%|███████▏  | 23/32 [00:11<00:04,  2.04it/s, loss=6.98, v_num=2]
Epoch 16:  72%|███████▏  | 23/32 [00:11<00:04,  2.04it/s, loss=6.98, v_num=2]
Epoch 16:  75%|███████▌  | 24/32 [00:11<00:03,  2.12it/s, loss=6.98, v_num=2]
Epoch 16:  75%|███████▌  | 24/32 [00:11<00:03,  2.12it/s, loss=6.98, v_num=2]
Epoch 16:  78%|███████▊  | 25/32 [00:14<00:03,  1.75it/s, loss=6.98, v_num=2]
Epoch 16:  78%|███████▊  | 25/32 [00:14<00:03,  1.75it/s, loss=6.98, v_num=2]
Epoch 16:  81%|████████▏ | 26/32 [00:14<00:03,  1.81it/s, loss=6.98, v_num=2]
Epoch 16:  81%|████████▏ | 26/32 [00:14<00:03,  1.81it/s, loss=6.99, v_num=2]
Epoch 16:  84%|████████▍ | 27/32 [00:14<00:02,  1.83it/s, loss=6.99, v_num=2]
Epoch 16:  84%|████████▍ | 27/32 [00:14<00:02,  1.83it/s, loss=6.99, v_num=2]
Epoch 16:  88%|████████▊ | 28/32 [00:14<00:02,  1.89it/s, loss=6.99, v_num=2]
Epoch 16:  88%|████████▊ | 28/32 [00:14<00:02,  1.89it/s, loss=6.99, v_num=2]
Epoch 16:  91%|█████████ | 29/32 [00:14<00:01,  1.95it/s, loss=6.99, v_num=2]
Epoch 16:  91%|█████████ | 29/32 [00:14<00:01,  1.95it/s, loss=6.99, v_num=2]
Epoch 16:  94%|█████████▍| 30/32 [00:14<00:00,  2.01it/s, loss=6.99, v_num=2]
Epoch 16:  94%|█████████▍| 30/32 [00:14<00:00,  2.01it/s, loss=6.99, v_num=2]
Epoch 16:  97%|█████████▋| 31/32 [00:14<00:00,  2.07it/s, loss=6.99, v_num=2]
Epoch 16:  97%|█████████▋| 31/32 [00:14<00:00,  2.07it/s, loss=7, v_num=2]
Epoch 16: 100%|██████████| 32/32 [00:15<00:00,  2.13it/s, loss=7, v_num=2]
Epoch 16: 100%|██████████| 32/32 [00:15<00:00,  2.13it/s, loss=7, v_num=2]
Epoch 16: 100%|██████████| 32/32 [00:15<00:00,  2.13it/s, loss=7, v_num=2]
Epoch 16:   0%|          | 0/32 [00:00<?, ?it/s, loss=7, v_num=2]
Epoch 17:   0%|          | 0/32 [00:00<?, ?it/s, loss=7, v_num=2]
Epoch 17:   3%|▎         | 1/32 [00:03<01:58,  3.83s/it, loss=7, v_num=2]
Epoch 17:   3%|▎         | 1/32 [00:03<01:58,  3.83s/it, loss=7, v_num=2]
Epoch 17:   6%|▋         | 2/32 [00:03<00:58,  1.94s/it, loss=7, v_num=2]
Epoch 17:   6%|▋         | 2/32 [00:03<00:58,  1.94s/it, loss=7, v_num=2]
Epoch 17:   9%|▉         | 3/32 [00:03<00:38,  1.31s/it, loss=7, v_num=2]
Epoch 17:   9%|▉         | 3/32 [00:03<00:38,  1.31s/it, loss=7, v_num=2]
Epoch 17:  12%|█▎        | 4/32 [00:03<00:27,  1.00it/s, loss=7, v_num=2]
Epoch 17:  12%|█▎        | 4/32 [00:03<00:27,  1.00it/s, loss=7.01, v_num=2]
Epoch 17:  16%|█▌        | 5/32 [00:04<00:21,  1.24it/s, loss=7.01, v_num=2]
Epoch 17:  16%|█▌        | 5/32 [00:04<00:21,  1.24it/s, loss=7.01, v_num=2]
Epoch 17:  19%|█▉        | 6/32 [00:04<00:17,  1.47it/s, loss=7.01, v_num=2]
Epoch 17:  19%|█▉        | 6/32 [00:04<00:17,  1.47it/s, loss=7.01, v_num=2]
Epoch 17:  22%|██▏       | 7/32 [00:04<00:14,  1.69it/s, loss=7.01, v_num=2]
Epoch 17:  22%|██▏       | 7/32 [00:04<00:14,  1.69it/s, loss=7.01, v_num=2]
Epoch 17:  25%|██▌       | 8/32 [00:04<00:12,  1.91it/s, loss=7.01, v_num=2]
Epoch 17:  25%|██▌       | 8/32 [00:04<00:12,  1.91it/s, loss=7.01, v_num=2]
Epoch 17:  28%|██▊       | 9/32 [00:06<00:17,  1.35it/s, loss=7.01, v_num=2]
Epoch 17:  28%|██▊       | 9/32 [00:06<00:17,  1.35it/s, loss=7.01, v_num=2]
Epoch 17:  31%|███▏      | 10/32 [00:06<00:14,  1.47it/s, loss=7.01, v_num=2]
Epoch 17:  31%|███▏      | 10/32 [00:06<00:14,  1.47it/s, loss=7.02, v_num=2]
Epoch 17:  34%|███▍      | 11/32 [00:06<00:13,  1.59it/s, loss=7.02, v_num=2]
Epoch 17:  34%|███▍      | 11/32 [00:06<00:13,  1.59it/s, loss=7.02, v_num=2]
Epoch 17:  38%|███▊      | 12/32 [00:06<00:11,  1.72it/s, loss=7.02, v_num=2]
Epoch 17:  38%|███▊      | 12/32 [00:06<00:11,  1.72it/s, loss=7.02, v_num=2]
Epoch 17:  41%|████      | 13/32 [00:07<00:10,  1.85it/s, loss=7.02, v_num=2]
Epoch 17:  41%|████      | 13/32 [00:07<00:10,  1.85it/s, loss=7.02, v_num=2]
Epoch 17:  44%|████▍     | 14/32 [00:07<00:09,  1.98it/s, loss=7.02, v_num=2]
Epoch 17:  44%|████▍     | 14/32 [00:07<00:09,  1.98it/s, loss=7.02, v_num=2]
Epoch 17:  47%|████▋     | 15/32 [00:07<00:08,  2.11it/s, loss=7.02, v_num=2]
Epoch 17:  47%|████▋     | 15/32 [00:07<00:08,  2.11it/s, loss=7.02, v_num=2]
Epoch 17:  50%|█████     | 16/32 [00:07<00:07,  2.23it/s, loss=7.02, v_num=2]
Epoch 17:  50%|█████     | 16/32 [00:07<00:07,  2.23it/s, loss=7.02, v_num=2]
Epoch 17:  53%|█████▎    | 17/32 [00:09<00:08,  1.78it/s, loss=7.02, v_num=2]
Epoch 17:  53%|█████▎    | 17/32 [00:09<00:08,  1.78it/s, loss=7.03, v_num=2]
Epoch 17:  56%|█████▋    | 18/32 [00:09<00:07,  1.84it/s, loss=7.03, v_num=2]
Epoch 17:  56%|█████▋    | 18/32 [00:09<00:07,  1.84it/s, loss=7.03, v_num=2]
Epoch 17:  59%|█████▉    | 19/32 [00:09<00:06,  1.93it/s, loss=7.03, v_num=2]
Epoch 17:  59%|█████▉    | 19/32 [00:09<00:06,  1.93it/s, loss=7.03, v_num=2]
Epoch 17:  62%|██████▎   | 20/32 [00:10<00:06,  2.00it/s, loss=7.03, v_num=2]
Epoch 17:  62%|██████▎   | 20/32 [00:10<00:06,  2.00it/s, loss=7.03, v_num=2]
Epoch 17:  66%|██████▌   | 21/32 [00:10<00:05,  2.09it/s, loss=7.03, v_num=2]
Epoch 17:  66%|██████▌   | 21/32 [00:10<00:05,  2.09it/s, loss=7.03, v_num=2]
Epoch 17:  69%|██████▉   | 22/32 [00:10<00:04,  2.18it/s, loss=7.03, v_num=2]
Epoch 17:  69%|██████▉   | 22/32 [00:10<00:04,  2.18it/s, loss=7.03, v_num=2]
Epoch 17:  72%|███████▏  | 23/32 [00:10<00:03,  2.26it/s, loss=7.03, v_num=2]
Epoch 17:  72%|███████▏  | 23/32 [00:10<00:03,  2.26it/s, loss=7.03, v_num=2]
Epoch 17:  75%|███████▌  | 24/32 [00:10<00:03,  2.35it/s, loss=7.03, v_num=2]
Epoch 17:  75%|███████▌  | 24/32 [00:10<00:03,  2.35it/s, loss=7.03, v_num=2]
Epoch 17:  78%|███████▊  | 25/32 [00:12<00:03,  1.95it/s, loss=7.03, v_num=2]
Epoch 17:  78%|███████▊  | 25/32 [00:12<00:03,  1.95it/s, loss=7.03, v_num=2]
Epoch 17:  81%|████████▏ | 26/32 [00:13<00:03,  1.99it/s, loss=7.03, v_num=2]
Epoch 17:  81%|████████▏ | 26/32 [00:13<00:03,  1.99it/s, loss=7.04, v_num=2]
Epoch 17:  84%|████████▍ | 27/32 [00:13<00:02,  2.06it/s, loss=7.04, v_num=2]
Epoch 17:  84%|████████▍ | 27/32 [00:13<00:02,  2.06it/s, loss=7.04, v_num=2]
Epoch 17:  88%|████████▊ | 28/32 [00:13<00:01,  2.11it/s, loss=7.04, v_num=2]
Epoch 17:  88%|████████▊ | 28/32 [00:13<00:01,  2.11it/s, loss=7.04, v_num=2]
Epoch 17:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.04, v_num=2]
Epoch 17:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.04, v_num=2]
Epoch 17:  94%|█████████▍| 30/32 [00:13<00:00,  2.24it/s, loss=7.04, v_num=2]
Epoch 17:  94%|█████████▍| 30/32 [00:13<00:00,  2.24it/s, loss=7.04, v_num=2]
Epoch 17:  97%|█████████▋| 31/32 [00:13<00:00,  2.30it/s, loss=7.04, v_num=2]
Epoch 17:  97%|█████████▋| 31/32 [00:13<00:00,  2.30it/s, loss=7.04, v_num=2]
Epoch 17: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.04, v_num=2]
Epoch 17: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.04, v_num=2]
Epoch 17: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.04, v_num=2]
Epoch 17:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.04, v_num=2]
Epoch 18:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.04, v_num=2]
Epoch 18:   3%|▎         | 1/32 [00:04<02:08,  4.16s/it, loss=7.04, v_num=2]
Epoch 18:   3%|▎         | 1/32 [00:04<02:08,  4.16s/it, loss=7.04, v_num=2]
Epoch 18:   6%|▋         | 2/32 [00:04<01:03,  2.11s/it, loss=7.04, v_num=2]
Epoch 18:   6%|▋         | 2/32 [00:04<01:03,  2.11s/it, loss=7.04, v_num=2]
Epoch 18:   9%|▉         | 3/32 [00:04<00:41,  1.42s/it, loss=7.04, v_num=2]
Epoch 18:   9%|▉         | 3/32 [00:04<00:41,  1.42s/it, loss=7.04, v_num=2]
Epoch 18:  12%|█▎        | 4/32 [00:04<00:30,  1.08s/it, loss=7.04, v_num=2]
Epoch 18:  12%|█▎        | 4/32 [00:04<00:30,  1.08s/it, loss=7.04, v_num=2]
Epoch 18:  16%|█▌        | 5/32 [00:04<00:23,  1.15it/s, loss=7.04, v_num=2]
Epoch 18:  16%|█▌        | 5/32 [00:04<00:23,  1.15it/s, loss=7.04, v_num=2]
Epoch 18:  19%|█▉        | 6/32 [00:04<00:19,  1.36it/s, loss=7.04, v_num=2]
Epoch 18:  19%|█▉        | 6/32 [00:04<00:19,  1.36it/s, loss=7.04, v_num=2]
Epoch 18:  22%|██▏       | 7/32 [00:04<00:15,  1.57it/s, loss=7.04, v_num=2]
Epoch 18:  22%|██▏       | 7/32 [00:04<00:15,  1.57it/s, loss=7.04, v_num=2]
Epoch 18:  25%|██▌       | 8/32 [00:04<00:13,  1.78it/s, loss=7.04, v_num=2]
Epoch 18:  25%|██▌       | 8/32 [00:04<00:13,  1.78it/s, loss=7.04, v_num=2]
Epoch 18:  28%|██▊       | 9/32 [00:07<00:18,  1.25it/s, loss=7.04, v_num=2]
Epoch 18:  28%|██▊       | 9/32 [00:07<00:18,  1.25it/s, loss=7.04, v_num=2]
Epoch 18:  31%|███▏      | 10/32 [00:07<00:16,  1.37it/s, loss=7.04, v_num=2]
Epoch 18:  31%|███▏      | 10/32 [00:07<00:16,  1.37it/s, loss=7.04, v_num=2]
Epoch 18:  34%|███▍      | 11/32 [00:07<00:13,  1.50it/s, loss=7.04, v_num=2]
Epoch 18:  34%|███▍      | 11/32 [00:07<00:13,  1.50it/s, loss=7.04, v_num=2]
Epoch 18:  38%|███▊      | 12/32 [00:07<00:12,  1.63it/s, loss=7.04, v_num=2]
Epoch 18:  38%|███▊      | 12/32 [00:07<00:12,  1.63it/s, loss=7.04, v_num=2]
Epoch 18:  41%|████      | 13/32 [00:07<00:10,  1.75it/s, loss=7.04, v_num=2]
Epoch 18:  41%|████      | 13/32 [00:07<00:10,  1.75it/s, loss=7.04, v_num=2]
Epoch 18:  44%|████▍     | 14/32 [00:07<00:09,  1.87it/s, loss=7.04, v_num=2]
Epoch 18:  44%|████▍     | 14/32 [00:07<00:09,  1.87it/s, loss=7.04, v_num=2]
Epoch 18:  47%|████▋     | 15/32 [00:07<00:08,  1.99it/s, loss=7.04, v_num=2]
Epoch 18:  47%|████▋     | 15/32 [00:07<00:08,  1.99it/s, loss=7.04, v_num=2]
Epoch 18:  50%|█████     | 16/32 [00:07<00:07,  2.11it/s, loss=7.04, v_num=2]
Epoch 18:  50%|█████     | 16/32 [00:07<00:07,  2.11it/s, loss=7.04, v_num=2]
Epoch 18:  53%|█████▎    | 17/32 [00:10<00:08,  1.69it/s, loss=7.04, v_num=2]
Epoch 18:  53%|█████▎    | 17/32 [00:10<00:08,  1.69it/s, loss=7.04, v_num=2]
Epoch 18:  56%|█████▋    | 18/32 [00:10<00:07,  1.76it/s, loss=7.04, v_num=2]
Epoch 18:  56%|█████▋    | 18/32 [00:10<00:07,  1.76it/s, loss=7.04, v_num=2]
Epoch 18:  59%|█████▉    | 19/32 [00:10<00:07,  1.85it/s, loss=7.04, v_num=2]
Epoch 18:  59%|█████▉    | 19/32 [00:10<00:07,  1.85it/s, loss=7.03, v_num=2]
Epoch 18:  62%|██████▎   | 20/32 [00:10<00:06,  1.94it/s, loss=7.03, v_num=2]
Epoch 18:  62%|██████▎   | 20/32 [00:10<00:06,  1.94it/s, loss=7.03, v_num=2]
Epoch 18:  66%|██████▌   | 21/32 [00:10<00:05,  2.03it/s, loss=7.03, v_num=2]
Epoch 18:  66%|██████▌   | 21/32 [00:10<00:05,  2.03it/s, loss=7.03, v_num=2]
Epoch 18:  69%|██████▉   | 22/32 [00:10<00:04,  2.11it/s, loss=7.03, v_num=2]
Epoch 18:  69%|██████▉   | 22/32 [00:10<00:04,  2.11it/s, loss=7.03, v_num=2]
Epoch 18:  72%|███████▏  | 23/32 [00:10<00:04,  2.20it/s, loss=7.03, v_num=2]
Epoch 18:  72%|███████▏  | 23/32 [00:10<00:04,  2.20it/s, loss=7.03, v_num=2]
Epoch 18:  75%|███████▌  | 24/32 [00:10<00:03,  2.28it/s, loss=7.03, v_num=2]
Epoch 18:  75%|███████▌  | 24/32 [00:10<00:03,  2.28it/s, loss=7.02, v_num=2]
Epoch 18:  78%|███████▊  | 25/32 [00:13<00:03,  1.91it/s, loss=7.02, v_num=2]
Epoch 18:  78%|███████▊  | 25/32 [00:13<00:03,  1.91it/s, loss=7.02, v_num=2]
Epoch 18:  81%|████████▏ | 26/32 [00:13<00:03,  1.97it/s, loss=7.02, v_num=2]
Epoch 18:  81%|████████▏ | 26/32 [00:13<00:03,  1.97it/s, loss=7.02, v_num=2]
Epoch 18:  84%|████████▍ | 27/32 [00:13<00:02,  2.03it/s, loss=7.02, v_num=2]
Epoch 18:  84%|████████▍ | 27/32 [00:13<00:02,  2.03it/s, loss=7.01, v_num=2]
Epoch 18:  88%|████████▊ | 28/32 [00:13<00:01,  2.10it/s, loss=7.01, v_num=2]
Epoch 18:  88%|████████▊ | 28/32 [00:13<00:01,  2.10it/s, loss=7.01, v_num=2]
Epoch 18:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.01, v_num=2]
Epoch 18:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.01, v_num=2]
Epoch 18:  94%|█████████▍| 30/32 [00:13<00:00,  2.23it/s, loss=7.01, v_num=2]
Epoch 18:  94%|█████████▍| 30/32 [00:13<00:00,  2.23it/s, loss=7, v_num=2]
Epoch 18:  97%|█████████▋| 31/32 [00:13<00:00,  2.30it/s, loss=7, v_num=2]
Epoch 18:  97%|█████████▋| 31/32 [00:13<00:00,  2.30it/s, loss=7, v_num=2]
Epoch 18: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7, v_num=2]
Epoch 18: 100%|██████████| 32/32 [00:13<00:00,  2.36it/s, loss=6.99, v_num=2]
Epoch 18: 100%|██████████| 32/32 [00:13<00:00,  2.36it/s, loss=6.99, v_num=2]
Epoch 18:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.99, v_num=2]
Epoch 19:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.99, v_num=2]
Epoch 19:   3%|▎         | 1/32 [00:03<02:02,  3.94s/it, loss=6.99, v_num=2]
Epoch 19:   3%|▎         | 1/32 [00:03<02:02,  3.94s/it, loss=6.99, v_num=2]
Epoch 19:   6%|▋         | 2/32 [00:03<00:59,  2.00s/it, loss=6.99, v_num=2]
Epoch 19:   6%|▋         | 2/32 [00:03<00:59,  2.00s/it, loss=6.98, v_num=2]
Epoch 19:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.98, v_num=2]
Epoch 19:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.98, v_num=2]
Epoch 19:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.98, v_num=2]
Epoch 19:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.97, v_num=2]
Epoch 19:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.97, v_num=2]
Epoch 19:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.97, v_num=2]
Epoch 19:  19%|█▉        | 6/32 [00:04<00:18,  1.42it/s, loss=6.97, v_num=2]
Epoch 19:  19%|█▉        | 6/32 [00:04<00:18,  1.42it/s, loss=6.96, v_num=2]
Epoch 19:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.96, v_num=2]
Epoch 19:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.95, v_num=2]
Epoch 19:  25%|██▌       | 8/32 [00:04<00:12,  1.86it/s, loss=6.95, v_num=2]
Epoch 19:  25%|██▌       | 8/32 [00:04<00:12,  1.85it/s, loss=6.95, v_num=2]
Epoch 19:  28%|██▊       | 9/32 [00:06<00:17,  1.30it/s, loss=6.95, v_num=2]
Epoch 19:  28%|██▊       | 9/32 [00:06<00:17,  1.30it/s, loss=6.94, v_num=2]
Epoch 19:  31%|███▏      | 10/32 [00:06<00:15,  1.44it/s, loss=6.94, v_num=2]
Epoch 19:  31%|███▏      | 10/32 [00:06<00:15,  1.44it/s, loss=6.93, v_num=2]
Epoch 19:  34%|███▍      | 11/32 [00:07<00:13,  1.57it/s, loss=6.93, v_num=2]
Epoch 19:  34%|███▍      | 11/32 [00:07<00:13,  1.57it/s, loss=6.93, v_num=2]
Epoch 19:  38%|███▊      | 12/32 [00:07<00:11,  1.69it/s, loss=6.93, v_num=2]
Epoch 19:  38%|███▊      | 12/32 [00:07<00:11,  1.69it/s, loss=6.92, v_num=2]
Epoch 19:  41%|████      | 13/32 [00:07<00:10,  1.82it/s, loss=6.92, v_num=2]
Epoch 19:  41%|████      | 13/32 [00:07<00:10,  1.82it/s, loss=6.91, v_num=2]
Epoch 19:  44%|████▍     | 14/32 [00:07<00:09,  1.95it/s, loss=6.91, v_num=2]
Epoch 19:  44%|████▍     | 14/32 [00:07<00:09,  1.95it/s, loss=6.91, v_num=2]
Epoch 19:  47%|████▋     | 15/32 [00:07<00:08,  2.07it/s, loss=6.91, v_num=2]
Epoch 19:  47%|████▋     | 15/32 [00:07<00:08,  2.07it/s, loss=6.9, v_num=2]
Epoch 19:  50%|█████     | 16/32 [00:07<00:07,  2.20it/s, loss=6.9, v_num=2]
Epoch 19:  50%|█████     | 16/32 [00:07<00:07,  2.20it/s, loss=6.9, v_num=2]
Epoch 19:  53%|█████▎    | 17/32 [00:10<00:08,  1.68it/s, loss=6.9, v_num=2]
Epoch 19:  53%|█████▎    | 17/32 [00:10<00:08,  1.68it/s, loss=6.89, v_num=2]
Epoch 19:  56%|█████▋    | 18/32 [00:10<00:07,  1.77it/s, loss=6.89, v_num=2]
Epoch 19:  56%|█████▋    | 18/32 [00:10<00:07,  1.77it/s, loss=6.89, v_num=2]
Epoch 19:  59%|█████▉    | 19/32 [00:10<00:06,  1.86it/s, loss=6.89, v_num=2]
Epoch 19:  59%|█████▉    | 19/32 [00:10<00:06,  1.86it/s, loss=6.88, v_num=2]
Epoch 19:  62%|██████▎   | 20/32 [00:10<00:06,  1.95it/s, loss=6.88, v_num=2]
Epoch 19:  62%|██████▎   | 20/32 [00:10<00:06,  1.95it/s, loss=6.88, v_num=2]
Epoch 19:  66%|██████▌   | 21/32 [00:10<00:05,  2.04it/s, loss=6.88, v_num=2]
Epoch 19:  66%|██████▌   | 21/32 [00:10<00:05,  2.04it/s, loss=6.87, v_num=2]
Epoch 19:  69%|██████▉   | 22/32 [00:10<00:04,  2.13it/s, loss=6.87, v_num=2]
Epoch 19:  69%|██████▉   | 22/32 [00:10<00:04,  2.13it/s, loss=6.87, v_num=2]
Epoch 19:  72%|███████▏  | 23/32 [00:10<00:04,  2.21it/s, loss=6.87, v_num=2]
Epoch 19:  72%|███████▏  | 23/32 [00:10<00:04,  2.21it/s, loss=6.86, v_num=2]
Epoch 19:  75%|███████▌  | 24/32 [00:10<00:03,  2.30it/s, loss=6.86, v_num=2]
Epoch 19:  75%|███████▌  | 24/32 [00:10<00:03,  2.30it/s, loss=6.86, v_num=2]
Epoch 19:  78%|███████▊  | 25/32 [00:13<00:03,  1.87it/s, loss=6.86, v_num=2]
Epoch 19:  78%|███████▊  | 25/32 [00:13<00:03,  1.87it/s, loss=6.85, v_num=2]
Epoch 19:  81%|████████▏ | 26/32 [00:13<00:03,  1.94it/s, loss=6.85, v_num=2]
Epoch 19:  81%|████████▏ | 26/32 [00:13<00:03,  1.94it/s, loss=6.85, v_num=2]
Epoch 19:  84%|████████▍ | 27/32 [00:13<00:02,  2.00it/s, loss=6.85, v_num=2]
Epoch 19:  84%|████████▍ | 27/32 [00:13<00:02,  2.00it/s, loss=6.85, v_num=2]
Epoch 19:  88%|████████▊ | 28/32 [00:13<00:01,  2.07it/s, loss=6.85, v_num=2]
Epoch 19:  88%|████████▊ | 28/32 [00:13<00:01,  2.07it/s, loss=6.85, v_num=2]
Epoch 19:  91%|█████████ | 29/32 [00:13<00:01,  2.14it/s, loss=6.85, v_num=2]
Epoch 19:  91%|█████████ | 29/32 [00:13<00:01,  2.14it/s, loss=6.84, v_num=2]
Epoch 19:  94%|█████████▍| 30/32 [00:13<00:00,  2.20it/s, loss=6.84, v_num=2]
Epoch 19:  94%|█████████▍| 30/32 [00:13<00:00,  2.20it/s, loss=6.84, v_num=2]
Epoch 19:  97%|█████████▋| 31/32 [00:13<00:00,  2.27it/s, loss=6.84, v_num=2]
Epoch 19:  97%|█████████▋| 31/32 [00:13<00:00,  2.27it/s, loss=6.84, v_num=2]
Epoch 19: 100%|██████████| 32/32 [00:13<00:00,  2.33it/s, loss=6.84, v_num=2]
Epoch 19: 100%|██████████| 32/32 [00:13<00:00,  2.33it/s, loss=6.84, v_num=2]
Epoch 19: 100%|██████████| 32/32 [00:13<00:00,  2.33it/s, loss=6.84, v_num=2]
Epoch 19:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.84, v_num=2]
Epoch 20:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.84, v_num=2]
Epoch 20:   3%|▎         | 1/32 [00:03<02:02,  3.95s/it, loss=6.84, v_num=2]
Epoch 20:   3%|▎         | 1/32 [00:03<02:02,  3.95s/it, loss=6.84, v_num=2]
Epoch 20:   6%|▋         | 2/32 [00:04<01:00,  2.00s/it, loss=6.84, v_num=2]
Epoch 20:   6%|▋         | 2/32 [00:04<01:00,  2.00s/it, loss=6.83, v_num=2]
Epoch 20:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.83, v_num=2]
Epoch 20:   9%|▉         | 3/32 [00:04<00:39,  1.35s/it, loss=6.83, v_num=2]
Epoch 20:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.83, v_num=2]
Epoch 20:  12%|█▎        | 4/32 [00:04<00:28,  1.03s/it, loss=6.83, v_num=2]
Epoch 20:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.83, v_num=2]
Epoch 20:  16%|█▌        | 5/32 [00:04<00:22,  1.20it/s, loss=6.83, v_num=2]
Epoch 20:  19%|█▉        | 6/32 [00:04<00:18,  1.43it/s, loss=6.83, v_num=2]
Epoch 20:  19%|█▉        | 6/32 [00:04<00:18,  1.43it/s, loss=6.83, v_num=2]
Epoch 20:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.83, v_num=2]
Epoch 20:  22%|██▏       | 7/32 [00:04<00:15,  1.64it/s, loss=6.83, v_num=2]
Epoch 20:  25%|██▌       | 8/32 [00:04<00:12,  1.86it/s, loss=6.83, v_num=2]
Epoch 20:  25%|██▌       | 8/32 [00:04<00:12,  1.86it/s, loss=6.83, v_num=2]
Epoch 20:  28%|██▊       | 9/32 [00:06<00:17,  1.29it/s, loss=6.83, v_num=2]
Epoch 20:  28%|██▊       | 9/32 [00:06<00:17,  1.29it/s, loss=6.83, v_num=2]
Epoch 20:  31%|███▏      | 10/32 [00:07<00:15,  1.41it/s, loss=6.83, v_num=2]
Epoch 20:  31%|███▏      | 10/32 [00:07<00:15,  1.41it/s, loss=6.83, v_num=2]
Epoch 20:  34%|███▍      | 11/32 [00:07<00:13,  1.54it/s, loss=6.83, v_num=2]
Epoch 20:  34%|███▍      | 11/32 [00:07<00:13,  1.54it/s, loss=6.83, v_num=2]
Epoch 20:  38%|███▊      | 12/32 [00:07<00:11,  1.67it/s, loss=6.83, v_num=2]
Epoch 20:  38%|███▊      | 12/32 [00:07<00:11,  1.67it/s, loss=6.83, v_num=2]
Epoch 20:  41%|████      | 13/32 [00:07<00:10,  1.80it/s, loss=6.83, v_num=2]
Epoch 20:  41%|████      | 13/32 [00:07<00:10,  1.80it/s, loss=6.83, v_num=2]
Epoch 20:  44%|████▍     | 14/32 [00:07<00:09,  1.93it/s, loss=6.83, v_num=2]
Epoch 20:  44%|████▍     | 14/32 [00:07<00:09,  1.93it/s, loss=6.83, v_num=2]
Epoch 20:  47%|████▋     | 15/32 [00:07<00:08,  2.05it/s, loss=6.83, v_num=2]
Epoch 20:  47%|████▋     | 15/32 [00:07<00:08,  2.05it/s, loss=6.83, v_num=2]
Epoch 20:  50%|█████     | 16/32 [00:07<00:07,  2.17it/s, loss=6.83, v_num=2]
Epoch 20:  50%|█████     | 16/32 [00:07<00:07,  2.17it/s, loss=6.83, v_num=2]
Epoch 20:  53%|█████▎    | 17/32 [00:09<00:08,  1.72it/s, loss=6.83, v_num=2]
Epoch 20:  53%|█████▎    | 17/32 [00:09<00:08,  1.72it/s, loss=6.84, v_num=2]
Epoch 20:  56%|█████▋    | 18/32 [00:09<00:07,  1.81it/s, loss=6.84, v_num=2]
Epoch 20:  56%|█████▋    | 18/32 [00:09<00:07,  1.81it/s, loss=6.84, v_num=2]
Epoch 20:  59%|█████▉    | 19/32 [00:10<00:06,  1.90it/s, loss=6.84, v_num=2]
Epoch 20:  59%|█████▉    | 19/32 [00:10<00:06,  1.90it/s, loss=6.84, v_num=2]
Epoch 20:  62%|██████▎   | 20/32 [00:10<00:06,  1.99it/s, loss=6.84, v_num=2]
Epoch 20:  62%|██████▎   | 20/32 [00:10<00:06,  1.99it/s, loss=6.84, v_num=2]
Epoch 20:  66%|██████▌   | 21/32 [00:10<00:05,  2.08it/s, loss=6.84, v_num=2]
Epoch 20:  66%|██████▌   | 21/32 [00:10<00:05,  2.08it/s, loss=6.85, v_num=2]
Epoch 20:  69%|██████▉   | 22/32 [00:10<00:04,  2.15it/s, loss=6.85, v_num=2]
Epoch 20:  69%|██████▉   | 22/32 [00:10<00:04,  2.15it/s, loss=6.85, v_num=2]
Epoch 20:  72%|███████▏  | 23/32 [00:10<00:04,  2.24it/s, loss=6.85, v_num=2]
Epoch 20:  72%|███████▏  | 23/32 [00:10<00:04,  2.24it/s, loss=6.85, v_num=2]
Epoch 20:  75%|███████▌  | 24/32 [00:10<00:03,  2.32it/s, loss=6.85, v_num=2]
Epoch 20:  75%|███████▌  | 24/32 [00:10<00:03,  2.32it/s, loss=6.85, v_num=2]
Epoch 20:  78%|███████▊  | 25/32 [00:12<00:03,  1.95it/s, loss=6.85, v_num=2]
Epoch 20:  78%|███████▊  | 25/32 [00:12<00:03,  1.95it/s, loss=6.85, v_num=2]
Epoch 20:  81%|████████▏ | 26/32 [00:12<00:02,  2.02it/s, loss=6.85, v_num=2]
Epoch 20:  81%|████████▏ | 26/32 [00:12<00:02,  2.02it/s, loss=6.85, v_num=2]
Epoch 20:  84%|████████▍ | 27/32 [00:12<00:02,  2.09it/s, loss=6.85, v_num=2]
Epoch 20:  84%|████████▍ | 27/32 [00:12<00:02,  2.09it/s, loss=6.85, v_num=2]
Epoch 20:  88%|████████▊ | 28/32 [00:12<00:01,  2.16it/s, loss=6.85, v_num=2]
Epoch 20:  88%|████████▊ | 28/32 [00:12<00:01,  2.16it/s, loss=6.86, v_num=2]
Epoch 20:  91%|█████████ | 29/32 [00:12<00:01,  2.23it/s, loss=6.86, v_num=2]
Epoch 20:  91%|█████████ | 29/32 [00:12<00:01,  2.23it/s, loss=6.86, v_num=2]
Epoch 20:  94%|█████████▍| 30/32 [00:13<00:00,  2.25it/s, loss=6.86, v_num=2]
Epoch 20:  94%|█████████▍| 30/32 [00:13<00:00,  2.25it/s, loss=6.86, v_num=2]
Epoch 20:  97%|█████████▋| 31/32 [00:13<00:00,  2.32it/s, loss=6.86, v_num=2]
Epoch 20:  97%|█████████▋| 31/32 [00:13<00:00,  2.32it/s, loss=6.87, v_num=2]
Epoch 20: 100%|██████████| 32/32 [00:13<00:00,  2.39it/s, loss=6.87, v_num=2]
Epoch 20: 100%|██████████| 32/32 [00:13<00:00,  2.39it/s, loss=6.87, v_num=2]
Epoch 20: 100%|██████████| 32/32 [00:13<00:00,  2.39it/s, loss=6.87, v_num=2]
Epoch 20:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.87, v_num=2]
Epoch 21:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.87, v_num=2]
Epoch 21:   3%|▎         | 1/32 [00:04<02:06,  4.08s/it, loss=6.87, v_num=2]
Epoch 21:   3%|▎         | 1/32 [00:04<02:06,  4.08s/it, loss=6.87, v_num=2]
Epoch 21:   6%|▋         | 2/32 [00:04<01:03,  2.11s/it, loss=6.87, v_num=2]
Epoch 21:   6%|▋         | 2/32 [00:04<01:03,  2.11s/it, loss=6.88, v_num=2]
Epoch 21:   9%|▉         | 3/32 [00:04<00:41,  1.43s/it, loss=6.88, v_num=2]
Epoch 21:   9%|▉         | 3/32 [00:04<00:41,  1.43s/it, loss=6.88, v_num=2]
Epoch 21:  12%|█▎        | 4/32 [00:04<00:30,  1.08s/it, loss=6.88, v_num=2]
Epoch 21:  12%|█▎        | 4/32 [00:04<00:30,  1.08s/it, loss=6.88, v_num=2]
Epoch 21:  16%|█▌        | 5/32 [00:04<00:23,  1.14it/s, loss=6.88, v_num=2]
Epoch 21:  16%|█▌        | 5/32 [00:04<00:23,  1.14it/s, loss=6.89, v_num=2]
Epoch 21:  19%|█▉        | 6/32 [00:04<00:19,  1.36it/s, loss=6.89, v_num=2]
Epoch 21:  19%|█▉        | 6/32 [00:04<00:19,  1.36it/s, loss=6.89, v_num=2]
Epoch 21:  22%|██▏       | 7/32 [00:04<00:15,  1.57it/s, loss=6.89, v_num=2]
Epoch 21:  22%|██▏       | 7/32 [00:04<00:15,  1.57it/s, loss=6.9, v_num=2]
Epoch 21:  25%|██▌       | 8/32 [00:04<00:13,  1.77it/s, loss=6.9, v_num=2]
Epoch 21:  25%|██▌       | 8/32 [00:04<00:13,  1.77it/s, loss=6.91, v_num=2]
Epoch 21:  28%|██▊       | 9/32 [00:07<00:18,  1.27it/s, loss=6.91, v_num=2]
Epoch 21:  28%|██▊       | 9/32 [00:07<00:18,  1.27it/s, loss=6.91, v_num=2]
Epoch 21:  31%|███▏      | 10/32 [00:07<00:15,  1.38it/s, loss=6.91, v_num=2]
Epoch 21:  31%|███▏      | 10/32 [00:07<00:15,  1.38it/s, loss=6.91, v_num=2]
Epoch 21:  34%|███▍      | 11/32 [00:07<00:13,  1.51it/s, loss=6.91, v_num=2]
Epoch 21:  34%|███▍      | 11/32 [00:07<00:13,  1.51it/s, loss=6.92, v_num=2]
Epoch 21:  38%|███▊      | 12/32 [00:07<00:12,  1.63it/s, loss=6.92, v_num=2]
Epoch 21:  38%|███▊      | 12/32 [00:07<00:12,  1.63it/s, loss=6.92, v_num=2]
Epoch 21:  41%|████      | 13/32 [00:07<00:10,  1.76it/s, loss=6.92, v_num=2]
Epoch 21:  41%|████      | 13/32 [00:07<00:10,  1.76it/s, loss=6.93, v_num=2]
Epoch 21:  44%|████▍     | 14/32 [00:07<00:09,  1.88it/s, loss=6.93, v_num=2]
Epoch 21:  44%|████▍     | 14/32 [00:07<00:09,  1.88it/s, loss=6.94, v_num=2]
Epoch 21:  47%|████▋     | 15/32 [00:07<00:08,  2.00it/s, loss=6.94, v_num=2]
Epoch 21:  47%|████▋     | 15/32 [00:07<00:08,  2.00it/s, loss=6.95, v_num=2]
Epoch 21:  50%|█████     | 16/32 [00:07<00:07,  2.12it/s, loss=6.95, v_num=2]
Epoch 21:  50%|█████     | 16/32 [00:07<00:07,  2.12it/s, loss=6.95, v_num=2]
Epoch 21:  53%|█████▎    | 17/32 [00:09<00:08,  1.70it/s, loss=6.95, v_num=2]
Epoch 21:  53%|█████▎    | 17/32 [00:09<00:08,  1.70it/s, loss=6.96, v_num=2]
Epoch 21:  56%|█████▋    | 18/32 [00:10<00:07,  1.77it/s, loss=6.96, v_num=2]
Epoch 21:  56%|█████▋    | 18/32 [00:10<00:07,  1.77it/s, loss=6.96, v_num=2]
Epoch 21:  59%|█████▉    | 19/32 [00:10<00:07,  1.84it/s, loss=6.96, v_num=2]
Epoch 21:  59%|█████▉    | 19/32 [00:10<00:07,  1.84it/s, loss=6.97, v_num=2]
Epoch 21:  62%|██████▎   | 20/32 [00:10<00:06,  1.93it/s, loss=6.97, v_num=2]
Epoch 21:  62%|██████▎   | 20/32 [00:10<00:06,  1.93it/s, loss=6.97, v_num=2]
Epoch 21:  66%|██████▌   | 21/32 [00:10<00:05,  2.02it/s, loss=6.97, v_num=2]
Epoch 21:  66%|██████▌   | 21/32 [00:10<00:05,  2.02it/s, loss=6.98, v_num=2]
Epoch 21:  69%|██████▉   | 22/32 [00:10<00:04,  2.10it/s, loss=6.98, v_num=2]
Epoch 21:  69%|██████▉   | 22/32 [00:10<00:04,  2.10it/s, loss=6.99, v_num=2]
Epoch 21:  72%|███████▏  | 23/32 [00:10<00:04,  2.19it/s, loss=6.99, v_num=2]
Epoch 21:  72%|███████▏  | 23/32 [00:10<00:04,  2.19it/s, loss=7, v_num=2]
Epoch 21:  75%|███████▌  | 24/32 [00:10<00:03,  2.28it/s, loss=7, v_num=2]
Epoch 21:  75%|███████▌  | 24/32 [00:10<00:03,  2.28it/s, loss=7, v_num=2]
Epoch 21:  78%|███████▊  | 25/32 [00:12<00:03,  1.94it/s, loss=7, v_num=2]
Epoch 21:  78%|███████▊  | 25/32 [00:12<00:03,  1.94it/s, loss=7.01, v_num=2]
Epoch 21:  81%|████████▏ | 26/32 [00:13<00:03,  1.98it/s, loss=7.01, v_num=2]
Epoch 21:  81%|████████▏ | 26/32 [00:13<00:03,  1.98it/s, loss=7.01, v_num=2]
Epoch 21:  84%|████████▍ | 27/32 [00:13<00:02,  2.04it/s, loss=7.01, v_num=2]
Epoch 21:  84%|████████▍ | 27/32 [00:13<00:02,  2.04it/s, loss=7.02, v_num=2]
Epoch 21:  88%|████████▊ | 28/32 [00:13<00:01,  2.11it/s, loss=7.02, v_num=2]
Epoch 21:  88%|████████▊ | 28/32 [00:13<00:01,  2.11it/s, loss=7.02, v_num=2]
Epoch 21:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.02, v_num=2]
Epoch 21:  91%|█████████ | 29/32 [00:13<00:01,  2.17it/s, loss=7.03, v_num=2]
Epoch 21:  94%|█████████▍| 30/32 [00:13<00:00,  2.24it/s, loss=7.03, v_num=2]
Epoch 21:  94%|█████████▍| 30/32 [00:13<00:00,  2.24it/s, loss=7.04, v_num=2]
Epoch 21:  97%|█████████▋| 31/32 [00:13<00:00,  2.31it/s, loss=7.04, v_num=2]
Epoch 21:  97%|█████████▋| 31/32 [00:13<00:00,  2.30it/s, loss=7.04, v_num=2]
Epoch 21: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.04, v_num=2]
Epoch 21: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.05, v_num=2]
Epoch 21: 100%|██████████| 32/32 [00:13<00:00,  2.37it/s, loss=7.05, v_num=2]
Epoch 21:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.05, v_num=2]
Epoch 22:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.05, v_num=2]
Epoch 22:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.05, v_num=2]
Epoch 22:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.06, v_num=2]
Epoch 22:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.06, v_num=2]
Epoch 22:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.07, v_num=2]
Epoch 22:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.07, v_num=2]
Epoch 22:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.08, v_num=2]
Epoch 22:  12%|█▎        | 4/32 [00:03<00:27,  1.02it/s, loss=7.08, v_num=2]
Epoch 22:  12%|█▎        | 4/32 [00:03<00:27,  1.02it/s, loss=7.08, v_num=2]
Epoch 22:  16%|█▌        | 5/32 [00:03<00:21,  1.26it/s, loss=7.08, v_num=2]
Epoch 22:  16%|█▌        | 5/32 [00:03<00:21,  1.26it/s, loss=7.09, v_num=2]
Epoch 22:  19%|█▉        | 6/32 [00:04<00:17,  1.49it/s, loss=7.09, v_num=2]
Epoch 22:  19%|█▉        | 6/32 [00:04<00:17,  1.49it/s, loss=7.1, v_num=2]
Epoch 22:  22%|██▏       | 7/32 [00:04<00:14,  1.72it/s, loss=7.1, v_num=2]
Epoch 22:  22%|██▏       | 7/32 [00:04<00:14,  1.72it/s, loss=7.1, v_num=2]
Epoch 22:  25%|██▌       | 8/32 [00:04<00:12,  1.94it/s, loss=7.1, v_num=2]
Epoch 22:  25%|██▌       | 8/32 [00:04<00:12,  1.94it/s, loss=7.12, v_num=2]
Epoch 22:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.12, v_num=2]
Epoch 22:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.13, v_num=2]
Epoch 22:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.13, v_num=2]
Epoch 22:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.13, v_num=2]
Epoch 22:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.13, v_num=2]
Epoch 22:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.14, v_num=2]
Epoch 22:  38%|███▊      | 12/32 [00:06<00:11,  1.73it/s, loss=7.14, v_num=2]
Epoch 22:  38%|███▊      | 12/32 [00:06<00:11,  1.73it/s, loss=7.16, v_num=2]
Epoch 22:  41%|████      | 13/32 [00:07<00:10,  1.86it/s, loss=7.16, v_num=2]
Epoch 22:  41%|████      | 13/32 [00:07<00:10,  1.86it/s, loss=7.17, v_num=2]
Epoch 22:  44%|████▍     | 14/32 [00:07<00:09,  1.99it/s, loss=7.17, v_num=2]
Epoch 22:  44%|████▍     | 14/32 [00:07<00:09,  1.98it/s, loss=7.18, v_num=2]
Epoch 22:  47%|████▋     | 15/32 [00:07<00:08,  2.11it/s, loss=7.18, v_num=2]
Epoch 22:  47%|████▋     | 15/32 [00:07<00:08,  2.11it/s, loss=7.2, v_num=2]
Epoch 22:  50%|█████     | 16/32 [00:07<00:07,  2.24it/s, loss=7.2, v_num=2]
Epoch 22:  50%|█████     | 16/32 [00:07<00:07,  2.24it/s, loss=7.21, v_num=2]
Epoch 22:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.21, v_num=2]
Epoch 22:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.22, v_num=2]
Epoch 22:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.22, v_num=2]
Epoch 22:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.23, v_num=2]
Epoch 22:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.23, v_num=2]
Epoch 22:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.24, v_num=2]
Epoch 22:  62%|██████▎   | 20/32 [00:09<00:05,  2.03it/s, loss=7.24, v_num=2]
Epoch 22:  62%|██████▎   | 20/32 [00:09<00:05,  2.03it/s, loss=7.26, v_num=2]
Epoch 22:  66%|██████▌   | 21/32 [00:09<00:05,  2.12it/s, loss=7.26, v_num=2]
Epoch 22:  66%|██████▌   | 21/32 [00:09<00:05,  2.12it/s, loss=7.27, v_num=2]
Epoch 22:  69%|██████▉   | 22/32 [00:09<00:04,  2.21it/s, loss=7.27, v_num=2]
Epoch 22:  69%|██████▉   | 22/32 [00:09<00:04,  2.21it/s, loss=7.27, v_num=2]
Epoch 22:  72%|███████▏  | 23/32 [00:09<00:03,  2.30it/s, loss=7.27, v_num=2]
Epoch 22:  72%|███████▏  | 23/32 [00:10<00:03,  2.30it/s, loss=7.28, v_num=2]
Epoch 22:  75%|███████▌  | 24/32 [00:10<00:03,  2.39it/s, loss=7.28, v_num=2]
Epoch 22:  75%|███████▌  | 24/32 [00:10<00:03,  2.39it/s, loss=7.31, v_num=2]
Epoch 22:  78%|███████▊  | 25/32 [00:12<00:03,  2.01it/s, loss=7.31, v_num=2]
Epoch 22:  78%|███████▊  | 25/32 [00:12<00:03,  2.01it/s, loss=7.33, v_num=2]
Epoch 22:  81%|████████▏ | 26/32 [00:12<00:02,  2.08it/s, loss=7.33, v_num=2]
Epoch 22:  81%|████████▏ | 26/32 [00:12<00:02,  2.08it/s, loss=7.35, v_num=2]
Epoch 22:  84%|████████▍ | 27/32 [00:12<00:02,  2.15it/s, loss=7.35, v_num=2]
Epoch 22:  84%|████████▍ | 27/32 [00:12<00:02,  2.15it/s, loss=7.36, v_num=2]
Epoch 22:  88%|████████▊ | 28/32 [00:12<00:01,  2.18it/s, loss=7.36, v_num=2]
Epoch 22:  88%|████████▊ | 28/32 [00:12<00:01,  2.18it/s, loss=7.38, v_num=2]
Epoch 22:  91%|█████████ | 29/32 [00:12<00:01,  2.24it/s, loss=7.38, v_num=2]
Epoch 22:  91%|█████████ | 29/32 [00:12<00:01,  2.24it/s, loss=7.39, v_num=2]
Epoch 22:  94%|█████████▍| 30/32 [00:12<00:00,  2.31it/s, loss=7.39, v_num=2]
Epoch 22:  94%|█████████▍| 30/32 [00:12<00:00,  2.31it/s, loss=7.41, v_num=2]
Epoch 22:  97%|█████████▋| 31/32 [00:13<00:00,  2.38it/s, loss=7.41, v_num=2]
Epoch 22:  97%|█████████▋| 31/32 [00:13<00:00,  2.38it/s, loss=7.42, v_num=2]
Epoch 22: 100%|██████████| 32/32 [00:13<00:00,  2.45it/s, loss=7.42, v_num=2]
Epoch 22: 100%|██████████| 32/32 [00:13<00:00,  2.45it/s, loss=7.43, v_num=2]
Epoch 22: 100%|██████████| 32/32 [00:13<00:00,  2.45it/s, loss=7.43, v_num=2]
Epoch 22:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.43, v_num=2]
Epoch 23:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.43, v_num=2]
Epoch 23:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.43, v_num=2]
Epoch 23:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.45, v_num=2]
Epoch 23:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.45, v_num=2]
Epoch 23:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.46, v_num=2]
Epoch 23:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.46, v_num=2]
Epoch 23:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.47, v_num=2]
Epoch 23:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.47, v_num=2]
Epoch 23:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.49, v_num=2]
Epoch 23:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.49, v_num=2]
Epoch 23:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.51, v_num=2]
Epoch 23:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.51, v_num=2]
Epoch 23:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.52, v_num=2]
Epoch 23:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.52, v_num=2]
Epoch 23:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.53, v_num=2]
Epoch 23:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.53, v_num=2]
Epoch 23:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.54, v_num=2]
Epoch 23:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.54, v_num=2]
Epoch 23:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.56, v_num=2]
Epoch 23:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.56, v_num=2]
Epoch 23:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.58, v_num=2]
Epoch 23:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.58, v_num=2]
Epoch 23:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.6, v_num=2]
Epoch 23:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.6, v_num=2]
Epoch 23:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.61, v_num=2]
Epoch 23:  41%|████      | 13/32 [00:06<00:10,  1.89it/s, loss=7.61, v_num=2]
Epoch 23:  41%|████      | 13/32 [00:06<00:10,  1.89it/s, loss=7.62, v_num=2]
Epoch 23:  44%|████▍     | 14/32 [00:06<00:08,  2.02it/s, loss=7.62, v_num=2]
Epoch 23:  44%|████▍     | 14/32 [00:06<00:08,  2.02it/s, loss=7.63, v_num=2]
Epoch 23:  47%|████▋     | 15/32 [00:06<00:07,  2.15it/s, loss=7.63, v_num=2]
Epoch 23:  47%|████▋     | 15/32 [00:06<00:07,  2.15it/s, loss=7.64, v_num=2]
Epoch 23:  50%|█████     | 16/32 [00:07<00:07,  2.28it/s, loss=7.64, v_num=2]
Epoch 23:  50%|█████     | 16/32 [00:07<00:07,  2.28it/s, loss=7.66, v_num=2]
Epoch 23:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.66, v_num=2]
Epoch 23:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.67, v_num=2]
Epoch 23:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.67, v_num=2]
Epoch 23:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.68, v_num=2]
Epoch 23:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.68, v_num=2]
Epoch 23:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.69, v_num=2]
Epoch 23:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.69, v_num=2]
Epoch 23:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.71, v_num=2]
Epoch 23:  66%|██████▌   | 21/32 [00:09<00:05,  2.14it/s, loss=7.71, v_num=2]
Epoch 23:  66%|██████▌   | 21/32 [00:09<00:05,  2.14it/s, loss=7.72, v_num=2]
Epoch 23:  69%|██████▉   | 22/32 [00:09<00:04,  2.23it/s, loss=7.72, v_num=2]
Epoch 23:  69%|██████▉   | 22/32 [00:09<00:04,  2.23it/s, loss=7.74, v_num=2]
Epoch 23:  72%|███████▏  | 23/32 [00:09<00:03,  2.32it/s, loss=7.74, v_num=2]
Epoch 23:  72%|███████▏  | 23/32 [00:09<00:03,  2.32it/s, loss=7.75, v_num=2]
Epoch 23:  75%|███████▌  | 24/32 [00:09<00:03,  2.41it/s, loss=7.75, v_num=2]
Epoch 23:  75%|███████▌  | 24/32 [00:09<00:03,  2.41it/s, loss=7.75, v_num=2]
Epoch 23:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.75, v_num=2]
Epoch 23:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.76, v_num=2]
Epoch 23:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.76, v_num=2]
Epoch 23:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.77, v_num=2]
Epoch 23:  84%|████████▍ | 27/32 [00:12<00:02,  2.16it/s, loss=7.77, v_num=2]
Epoch 23:  84%|████████▍ | 27/32 [00:12<00:02,  2.16it/s, loss=7.78, v_num=2]
Epoch 23:  88%|████████▊ | 28/32 [00:12<00:01,  2.23it/s, loss=7.78, v_num=2]
Epoch 23:  88%|████████▊ | 28/32 [00:12<00:01,  2.23it/s, loss=7.79, v_num=2]
Epoch 23:  91%|█████████ | 29/32 [00:12<00:01,  2.28it/s, loss=7.79, v_num=2]
Epoch 23:  91%|█████████ | 29/32 [00:12<00:01,  2.28it/s, loss=7.79, v_num=2]
Epoch 23:  94%|█████████▍| 30/32 [00:12<00:00,  2.35it/s, loss=7.79, v_num=2]
Epoch 23:  94%|█████████▍| 30/32 [00:12<00:00,  2.35it/s, loss=7.8, v_num=2]
Epoch 23:  97%|█████████▋| 31/32 [00:12<00:00,  2.42it/s, loss=7.8, v_num=2]
Epoch 23:  97%|█████████▋| 31/32 [00:12<00:00,  2.42it/s, loss=7.8, v_num=2]
Epoch 23: 100%|██████████| 32/32 [00:12<00:00,  2.49it/s, loss=7.8, v_num=2]
Epoch 23: 100%|██████████| 32/32 [00:12<00:00,  2.49it/s, loss=7.8, v_num=2]
Epoch 23: 100%|██████████| 32/32 [00:12<00:00,  2.49it/s, loss=7.8, v_num=2]
Epoch 23:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.8, v_num=2]
Epoch 24:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.8, v_num=2]
Epoch 24:   3%|▎         | 1/32 [00:03<01:53,  3.65s/it, loss=7.8, v_num=2]
Epoch 24:   3%|▎         | 1/32 [00:03<01:53,  3.65s/it, loss=7.8, v_num=2]
Epoch 24:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=7.8, v_num=2]
Epoch 24:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=7.8, v_num=2]
Epoch 24:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.8, v_num=2]
Epoch 24:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.8, v_num=2]
Epoch 24:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.8, v_num=2]
Epoch 24:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.8, v_num=2]
Epoch 24:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7.8, v_num=2]
Epoch 24:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7.81, v_num=2]
Epoch 24:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7.81, v_num=2]
Epoch 24:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7.81, v_num=2]
Epoch 24:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.81, v_num=2]
Epoch 24:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.8, v_num=2]
Epoch 24:  25%|██▌       | 8/32 [00:03<00:11,  2.00it/s, loss=7.8, v_num=2]
Epoch 24:  25%|██▌       | 8/32 [00:03<00:11,  2.00it/s, loss=7.8, v_num=2]
Epoch 24:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.8, v_num=2]
Epoch 24:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.8, v_num=2]
Epoch 24:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.8, v_num=2]
Epoch 24:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.8, v_num=2]
Epoch 24:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.8, v_num=2]
Epoch 24:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.79, v_num=2]
Epoch 24:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.79, v_num=2]
Epoch 24:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.79, v_num=2]
Epoch 24:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.79, v_num=2]
Epoch 24:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.79, v_num=2]
Epoch 24:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.79, v_num=2]
Epoch 24:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.79, v_num=2]
Epoch 24:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.79, v_num=2]
Epoch 24:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.79, v_num=2]
Epoch 24:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.79, v_num=2]
Epoch 24:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.78, v_num=2]
Epoch 24:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.78, v_num=2]
Epoch 24:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.78, v_num=2]
Epoch 24:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.78, v_num=2]
Epoch 24:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.77, v_num=2]
Epoch 24:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.77, v_num=2]
Epoch 24:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.76, v_num=2]
Epoch 24:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.76, v_num=2]
Epoch 24:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.76, v_num=2]
Epoch 24:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.76, v_num=2]
Epoch 24:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.76, v_num=2]
Epoch 24:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.76, v_num=2]
Epoch 24:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.75, v_num=2]
Epoch 24:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.75, v_num=2]
Epoch 24:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.75, v_num=2]
Epoch 24:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.75, v_num=2]
Epoch 24:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.74, v_num=2]
Epoch 24:  78%|███████▊  | 25/32 [00:12<00:03,  2.05it/s, loss=7.74, v_num=2]
Epoch 24:  78%|███████▊  | 25/32 [00:12<00:03,  2.05it/s, loss=7.73, v_num=2]
Epoch 24:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.73, v_num=2]
Epoch 24:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.72, v_num=2]
Epoch 24:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.72, v_num=2]
Epoch 24:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.72, v_num=2]
Epoch 24:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.72, v_num=2]
Epoch 24:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.71, v_num=2]
Epoch 24:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.71, v_num=2]
Epoch 24:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.7, v_num=2]
Epoch 24:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.7, v_num=2]
Epoch 24:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.69, v_num=2]
Epoch 24:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.69, v_num=2]
Epoch 24:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.68, v_num=2]
Epoch 24: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.68, v_num=2]
Epoch 24: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.67, v_num=2]
Epoch 24: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.67, v_num=2]
Epoch 24:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.67, v_num=2]
Epoch 25:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.67, v_num=2]
Epoch 25:   3%|▎         | 1/32 [00:03<01:55,  3.74s/it, loss=7.67, v_num=2]
Epoch 25:   3%|▎         | 1/32 [00:03<01:56,  3.74s/it, loss=7.66, v_num=2]
Epoch 25:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=7.66, v_num=2]
Epoch 25:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=7.65, v_num=2]
Epoch 25:   9%|▉         | 3/32 [00:03<00:37,  1.28s/it, loss=7.65, v_num=2]
Epoch 25:   9%|▉         | 3/32 [00:03<00:37,  1.28s/it, loss=7.64, v_num=2]
Epoch 25:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=7.64, v_num=2]
Epoch 25:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=7.63, v_num=2]
Epoch 25:  16%|█▌        | 5/32 [00:03<00:21,  1.27it/s, loss=7.63, v_num=2]
Epoch 25:  16%|█▌        | 5/32 [00:03<00:21,  1.27it/s, loss=7.61, v_num=2]
Epoch 25:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=7.61, v_num=2]
Epoch 25:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=7.6, v_num=2]
Epoch 25:  22%|██▏       | 7/32 [00:04<00:14,  1.74it/s, loss=7.6, v_num=2]
Epoch 25:  22%|██▏       | 7/32 [00:04<00:14,  1.74it/s, loss=7.59, v_num=2]
Epoch 25:  25%|██▌       | 8/32 [00:04<00:12,  1.97it/s, loss=7.59, v_num=2]
Epoch 25:  25%|██▌       | 8/32 [00:04<00:12,  1.96it/s, loss=7.58, v_num=2]
Epoch 25:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=7.58, v_num=2]
Epoch 25:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=7.56, v_num=2]
Epoch 25:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=7.56, v_num=2]
Epoch 25:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=7.54, v_num=2]
Epoch 25:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=7.54, v_num=2]
Epoch 25:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=7.53, v_num=2]
Epoch 25:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.53, v_num=2]
Epoch 25:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.51, v_num=2]
Epoch 25:  41%|████      | 13/32 [00:06<00:09,  1.91it/s, loss=7.51, v_num=2]
Epoch 25:  41%|████      | 13/32 [00:06<00:09,  1.91it/s, loss=7.5, v_num=2]
Epoch 25:  44%|████▍     | 14/32 [00:06<00:08,  2.04it/s, loss=7.5, v_num=2]
Epoch 25:  44%|████▍     | 14/32 [00:06<00:08,  2.04it/s, loss=7.49, v_num=2]
Epoch 25:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=7.49, v_num=2]
Epoch 25:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=7.48, v_num=2]
Epoch 25:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=7.48, v_num=2]
Epoch 25:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=7.46, v_num=2]
Epoch 25:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.46, v_num=2]
Epoch 25:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.45, v_num=2]
Epoch 25:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.45, v_num=2]
Epoch 25:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.44, v_num=2]
Epoch 25:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.44, v_num=2]
Epoch 25:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.43, v_num=2]
Epoch 25:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.43, v_num=2]
Epoch 25:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.42, v_num=2]
Epoch 25:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=7.42, v_num=2]
Epoch 25:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=7.41, v_num=2]
Epoch 25:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=7.41, v_num=2]
Epoch 25:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=7.39, v_num=2]
Epoch 25:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=7.39, v_num=2]
Epoch 25:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=7.38, v_num=2]
Epoch 25:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.38, v_num=2]
Epoch 25:  75%|███████▌  | 24/32 [00:09<00:03,  2.44it/s, loss=7.37, v_num=2]
Epoch 25:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.37, v_num=2]
Epoch 25:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.36, v_num=2]
Epoch 25:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.36, v_num=2]
Epoch 25:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.36, v_num=2]
Epoch 25:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.36, v_num=2]
Epoch 25:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.35, v_num=2]
Epoch 25:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=7.35, v_num=2]
Epoch 25:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=7.34, v_num=2]
Epoch 25:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.34, v_num=2]
Epoch 25:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.33, v_num=2]
Epoch 25:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.33, v_num=2]
Epoch 25:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.32, v_num=2]
Epoch 25:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.32, v_num=2]
Epoch 25:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.31, v_num=2]
Epoch 25: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.31, v_num=2]
Epoch 25: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.3, v_num=2]
Epoch 25: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.3, v_num=2]
Epoch 25:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.3, v_num=2]
Epoch 26:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.3, v_num=2]
Epoch 26:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=7.3, v_num=2]
Epoch 26:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=7.29, v_num=2]
Epoch 26:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.29, v_num=2]
Epoch 26:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.29, v_num=2]
Epoch 26:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.29, v_num=2]
Epoch 26:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.28, v_num=2]
Epoch 26:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.28, v_num=2]
Epoch 26:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.27, v_num=2]
Epoch 26:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.27, v_num=2]
Epoch 26:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.26, v_num=2]
Epoch 26:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.26, v_num=2]
Epoch 26:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.25, v_num=2]
Epoch 26:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.25, v_num=2]
Epoch 26:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.25, v_num=2]
Epoch 26:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=7.25, v_num=2]
Epoch 26:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=7.25, v_num=2]
Epoch 26:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.25, v_num=2]
Epoch 26:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.24, v_num=2]
Epoch 26:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.24, v_num=2]
Epoch 26:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.23, v_num=2]
Epoch 26:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.23, v_num=2]
Epoch 26:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.23, v_num=2]
Epoch 26:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.23, v_num=2]
Epoch 26:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.24, v_num=2]
Epoch 26:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.24, v_num=2]
Epoch 26:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.23, v_num=2]
Epoch 26:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.23, v_num=2]
Epoch 26:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.23, v_num=2]
Epoch 26:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.23, v_num=2]
Epoch 26:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.23, v_num=2]
Epoch 26:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.23, v_num=2]
Epoch 26:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.23, v_num=2]
Epoch 26:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.23, v_num=2]
Epoch 26:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.24, v_num=2]
Epoch 26:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.24, v_num=2]
Epoch 26:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.25, v_num=2]
Epoch 26:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.25, v_num=2]
Epoch 26:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.25, v_num=2]
Epoch 26:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.25, v_num=2]
Epoch 26:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.26, v_num=2]
Epoch 26:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.26, v_num=2]
Epoch 26:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.27, v_num=2]
Epoch 26:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.27, v_num=2]
Epoch 26:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.28, v_num=2]
Epoch 26:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.28, v_num=2]
Epoch 26:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.28, v_num=2]
Epoch 26:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.28, v_num=2]
Epoch 26:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.28, v_num=2]
Epoch 26:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.28, v_num=2]
Epoch 26:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.28, v_num=2]
Epoch 26:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=7.28, v_num=2]
Epoch 26:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=7.28, v_num=2]
Epoch 26:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.28, v_num=2]
Epoch 26:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.29, v_num=2]
Epoch 26:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.29, v_num=2]
Epoch 26:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.3, v_num=2]
Epoch 26:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.3, v_num=2]
Epoch 26:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.3, v_num=2]
Epoch 26:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=7.3, v_num=2]
Epoch 26:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=7.3, v_num=2]
Epoch 26:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.3, v_num=2]
Epoch 26:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.3, v_num=2]
Epoch 26: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.3, v_num=2]
Epoch 26: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.29, v_num=2]
Epoch 26: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.29, v_num=2]
Epoch 26:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.29, v_num=2]
Epoch 27:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.29, v_num=2]
Epoch 27:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=7.29, v_num=2]
Epoch 27:   3%|▎         | 1/32 [00:03<01:53,  3.67s/it, loss=7.28, v_num=2]
Epoch 27:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.28, v_num=2]
Epoch 27:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.28, v_num=2]
Epoch 27:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.28, v_num=2]
Epoch 27:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.28, v_num=2]
Epoch 27:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.28, v_num=2]
Epoch 27:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.29, v_num=2]
Epoch 27:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.29, v_num=2]
Epoch 27:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.27, v_num=2]
Epoch 27:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.27, v_num=2]
Epoch 27:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.26, v_num=2]
Epoch 27:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.26, v_num=2]
Epoch 27:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.28, v_num=2]
Epoch 27:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=7.28, v_num=2]
Epoch 27:  25%|██▌       | 8/32 [00:04<00:12,  2.00it/s, loss=7.26, v_num=2]
Epoch 27:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.26, v_num=2]
Epoch 27:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.25, v_num=2]
Epoch 27:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.25, v_num=2]
Epoch 27:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.23, v_num=2]
Epoch 27:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.23, v_num=2]
Epoch 27:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.24, v_num=2]
Epoch 27:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.24, v_num=2]
Epoch 27:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.25, v_num=2]
Epoch 27:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.25, v_num=2]
Epoch 27:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.26, v_num=2]
Epoch 27:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.26, v_num=2]
Epoch 27:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.27, v_num=2]
Epoch 27:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.27, v_num=2]
Epoch 27:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.26, v_num=2]
Epoch 27:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.26, v_num=2]
Epoch 27:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.26, v_num=2]
Epoch 27:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.26, v_num=2]
Epoch 27:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.31, v_num=2]
Epoch 27:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.31, v_num=2]
Epoch 27:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.32, v_num=2]
Epoch 27:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.32, v_num=2]
Epoch 27:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.33, v_num=2]
Epoch 27:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.33, v_num=2]
Epoch 27:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.38, v_num=2]
Epoch 27:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.38, v_num=2]
Epoch 27:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.38, v_num=2]
Epoch 27:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=7.38, v_num=2]
Epoch 27:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=7.38, v_num=2]
Epoch 27:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.38, v_num=2]
Epoch 27:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.4, v_num=2]
Epoch 27:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.4, v_num=2]
Epoch 27:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.41, v_num=2]
Epoch 27:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.41, v_num=2]
Epoch 27:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.44, v_num=2]
Epoch 27:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.44, v_num=2]
Epoch 27:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.46, v_num=2]
Epoch 27:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.46, v_num=2]
Epoch 27:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.46, v_num=2]
Epoch 27:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.46, v_num=2]
Epoch 27:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.48, v_num=2]
Epoch 27:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.48, v_num=2]
Epoch 27:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.49, v_num=2]
Epoch 27:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.49, v_num=2]
Epoch 27:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.52, v_num=2]
Epoch 27:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.52, v_num=2]
Epoch 27:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.53, v_num=2]
Epoch 27: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.53, v_num=2]
Epoch 27: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.55, v_num=2]
Epoch 27: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.55, v_num=2]
Epoch 27:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.55, v_num=2]
Epoch 28:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.55, v_num=2]
Epoch 28:   3%|▎         | 1/32 [00:03<01:53,  3.68s/it, loss=7.55, v_num=2]
Epoch 28:   3%|▎         | 1/32 [00:03<01:53,  3.68s/it, loss=7.56, v_num=2]
Epoch 28:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.56, v_num=2]
Epoch 28:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.58, v_num=2]
Epoch 28:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.58, v_num=2]
Epoch 28:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.61, v_num=2]
Epoch 28:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.61, v_num=2]
Epoch 28:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.63, v_num=2]
Epoch 28:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.63, v_num=2]
Epoch 28:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.6, v_num=2]
Epoch 28:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.6, v_num=2]
Epoch 28:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.61, v_num=2]
Epoch 28:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.61, v_num=2]
Epoch 28:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.62, v_num=2]
Epoch 28:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.62, v_num=2]
Epoch 28:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.6, v_num=2]
Epoch 28:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.6, v_num=2]
Epoch 28:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.63, v_num=2]
Epoch 28:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.63, v_num=2]
Epoch 28:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.65, v_num=2]
Epoch 28:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.65, v_num=2]
Epoch 28:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.67, v_num=2]
Epoch 28:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.67, v_num=2]
Epoch 28:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.67, v_num=2]
Epoch 28:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.67, v_num=2]
Epoch 28:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.67, v_num=2]
Epoch 28:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.67, v_num=2]
Epoch 28:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.67, v_num=2]
Epoch 28:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.67, v_num=2]
Epoch 28:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.67, v_num=2]
Epoch 28:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.67, v_num=2]
Epoch 28:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.68, v_num=2]
Epoch 28:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.68, v_num=2]
Epoch 28:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.69, v_num=2]
Epoch 28:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.69, v_num=2]
Epoch 28:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.69, v_num=2]
Epoch 28:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.69, v_num=2]
Epoch 28:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.69, v_num=2]
Epoch 28:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.69, v_num=2]
Epoch 28:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.69, v_num=2]
Epoch 28:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.69, v_num=2]
Epoch 28:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.68, v_num=2]
Epoch 28:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.68, v_num=2]
Epoch 28:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.68, v_num=2]
Epoch 28:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.68, v_num=2]
Epoch 28:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.66, v_num=2]
Epoch 28:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.66, v_num=2]
Epoch 28:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.65, v_num=2]
Epoch 28:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.65, v_num=2]
Epoch 28:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.64, v_num=2]
Epoch 28:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.64, v_num=2]
Epoch 28:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.63, v_num=2]
Epoch 28:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.63, v_num=2]
Epoch 28:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.62, v_num=2]
Epoch 28:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=7.62, v_num=2]
Epoch 28:  88%|████████▊ | 28/32 [00:12<00:01,  2.24it/s, loss=7.61, v_num=2]
Epoch 28:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.61, v_num=2]
Epoch 28:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.59, v_num=2]
Epoch 28:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.59, v_num=2]
Epoch 28:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.57, v_num=2]
Epoch 28:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.57, v_num=2]
Epoch 28:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.53, v_num=2]
Epoch 28: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.53, v_num=2]
Epoch 28: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.52, v_num=2]
Epoch 28: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.52, v_num=2]
Epoch 28:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.52, v_num=2]
Epoch 29:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.52, v_num=2]
Epoch 29:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.52, v_num=2]
Epoch 29:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.51, v_num=2]
Epoch 29:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.51, v_num=2]
Epoch 29:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.49, v_num=2]
Epoch 29:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.49, v_num=2]
Epoch 29:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.47, v_num=2]
Epoch 29:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.47, v_num=2]
Epoch 29:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.46, v_num=2]
Epoch 29:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.46, v_num=2]
Epoch 29:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.43, v_num=2]
Epoch 29:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.43, v_num=2]
Epoch 29:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.41, v_num=2]
Epoch 29:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.41, v_num=2]
Epoch 29:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.4, v_num=2]
Epoch 29:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.4, v_num=2]
Epoch 29:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.38, v_num=2]
Epoch 29:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.38, v_num=2]
Epoch 29:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.36, v_num=2]
Epoch 29:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.36, v_num=2]
Epoch 29:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.34, v_num=2]
Epoch 29:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.34, v_num=2]
Epoch 29:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.32, v_num=2]
Epoch 29:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.32, v_num=2]
Epoch 29:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.31, v_num=2]
Epoch 29:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=7.31, v_num=2]
Epoch 29:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=7.3, v_num=2]
Epoch 29:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=7.3, v_num=2]
Epoch 29:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=7.28, v_num=2]
Epoch 29:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.28, v_num=2]
Epoch 29:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.26, v_num=2]
Epoch 29:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.26, v_num=2]
Epoch 29:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.25, v_num=2]
Epoch 29:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.25, v_num=2]
Epoch 29:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.23, v_num=2]
Epoch 29:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.23, v_num=2]
Epoch 29:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.22, v_num=2]
Epoch 29:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.22, v_num=2]
Epoch 29:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.21, v_num=2]
Epoch 29:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.21, v_num=2]
Epoch 29:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.2, v_num=2]
Epoch 29:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.2, v_num=2]
Epoch 29:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.19, v_num=2]
Epoch 29:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=7.19, v_num=2]
Epoch 29:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=7.18, v_num=2]
Epoch 29:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.18, v_num=2]
Epoch 29:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.17, v_num=2]
Epoch 29:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.17, v_num=2]
Epoch 29:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.16, v_num=2]
Epoch 29:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.16, v_num=2]
Epoch 29:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.15, v_num=2]
Epoch 29:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.15, v_num=2]
Epoch 29:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.14, v_num=2]
Epoch 29:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.14, v_num=2]
Epoch 29:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=7.12, v_num=2]
Epoch 29:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.12, v_num=2]
Epoch 29:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.11, v_num=2]
Epoch 29:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.11, v_num=2]
Epoch 29:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.1, v_num=2]
Epoch 29:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.1, v_num=2]
Epoch 29:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.09, v_num=2]
Epoch 29:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.09, v_num=2]
Epoch 29:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.09, v_num=2]
Epoch 29: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.09, v_num=2]
Epoch 29: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.07, v_num=2]
Epoch 29: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.07, v_num=2]
Epoch 29:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.07, v_num=2]
Epoch 30:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.07, v_num=2]
Epoch 30:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.07, v_num=2]
Epoch 30:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.06, v_num=2]
Epoch 30:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.06, v_num=2]
Epoch 30:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.05, v_num=2]
Epoch 30:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.05, v_num=2]
Epoch 30:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.05, v_num=2]
Epoch 30:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.05, v_num=2]
Epoch 30:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.05, v_num=2]
Epoch 30:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.05, v_num=2]
Epoch 30:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.05, v_num=2]
Epoch 30:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.05, v_num=2]
Epoch 30:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.04, v_num=2]
Epoch 30:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.04, v_num=2]
Epoch 30:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.03, v_num=2]
Epoch 30:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.03, v_num=2]
Epoch 30:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.02, v_num=2]
Epoch 30:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.02, v_num=2]
Epoch 30:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.01, v_num=2]
Epoch 30:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.01, v_num=2]
Epoch 30:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7, v_num=2]
Epoch 30:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7, v_num=2]
Epoch 30:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7, v_num=2]
Epoch 30:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7, v_num=2]
Epoch 30:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.99, v_num=2]
Epoch 30:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.99, v_num=2]
Epoch 30:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.99, v_num=2]
Epoch 30:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.99, v_num=2]
Epoch 30:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.98, v_num=2]
Epoch 30:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.98, v_num=2]
Epoch 30:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.98, v_num=2]
Epoch 30:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.98, v_num=2]
Epoch 30:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.98, v_num=2]
Epoch 30:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.98, v_num=2]
Epoch 30:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.98, v_num=2]
Epoch 30:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.98, v_num=2]
Epoch 30:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.97, v_num=2]
Epoch 30:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.97, v_num=2]
Epoch 30:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.96, v_num=2]
Epoch 30:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.96, v_num=2]
Epoch 30:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.95, v_num=2]
Epoch 30:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.95, v_num=2]
Epoch 30:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.95, v_num=2]
Epoch 30:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.95, v_num=2]
Epoch 30:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.95, v_num=2]
Epoch 30:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.95, v_num=2]
Epoch 30:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.95, v_num=2]
Epoch 30:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.95, v_num=2]
Epoch 30:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.94, v_num=2]
Epoch 30:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.94, v_num=2]
Epoch 30:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.95, v_num=2]
Epoch 30:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.95, v_num=2]
Epoch 30:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.94, v_num=2]
Epoch 30:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.94, v_num=2]
Epoch 30:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.94, v_num=2]
Epoch 30:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.94, v_num=2]
Epoch 30:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.94, v_num=2]
Epoch 30:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.94, v_num=2]
Epoch 30:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.94, v_num=2]
Epoch 30:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.94, v_num=2]
Epoch 30:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.93, v_num=2]
Epoch 30:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.93, v_num=2]
Epoch 30:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.92, v_num=2]
Epoch 30: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.92, v_num=2]
Epoch 30: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.92, v_num=2]
Epoch 30: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.92, v_num=2]
Epoch 30:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.92, v_num=2]
Epoch 31:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.92, v_num=2]
Epoch 31:   3%|▎         | 1/32 [00:03<01:55,  3.73s/it, loss=6.92, v_num=2]
Epoch 31:   3%|▎         | 1/32 [00:03<01:55,  3.73s/it, loss=6.92, v_num=2]
Epoch 31:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=6.92, v_num=2]
Epoch 31:   6%|▋         | 2/32 [00:03<00:56,  1.89s/it, loss=6.92, v_num=2]
Epoch 31:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.92, v_num=2]
Epoch 31:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.91, v_num=2]
Epoch 31:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=6.91, v_num=2]
Epoch 31:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=6.91, v_num=2]
Epoch 31:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.91, v_num=2]
Epoch 31:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.9, v_num=2]
Epoch 31:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=6.9, v_num=2]
Epoch 31:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=6.91, v_num=2]
Epoch 31:  22%|██▏       | 7/32 [00:04<00:14,  1.75it/s, loss=6.91, v_num=2]
Epoch 31:  22%|██▏       | 7/32 [00:04<00:14,  1.74it/s, loss=6.91, v_num=2]
Epoch 31:  25%|██▌       | 8/32 [00:04<00:12,  1.97it/s, loss=6.91, v_num=2]
Epoch 31:  25%|██▌       | 8/32 [00:04<00:12,  1.97it/s, loss=6.92, v_num=2]
Epoch 31:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.92, v_num=2]
Epoch 31:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.9, v_num=2]
Epoch 31:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.9, v_num=2]
Epoch 31:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.9, v_num=2]
Epoch 31:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.9, v_num=2]
Epoch 31:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.9, v_num=2]
Epoch 31:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.9, v_num=2]
Epoch 31:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.89, v_num=2]
Epoch 31:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.89, v_num=2]
Epoch 31:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.88, v_num=2]
Epoch 31:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.88, v_num=2]
Epoch 31:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.88, v_num=2]
Epoch 31:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.88, v_num=2]
Epoch 31:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.88, v_num=2]
Epoch 31:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.88, v_num=2]
Epoch 31:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.88, v_num=2]
Epoch 31:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.88, v_num=2]
Epoch 31:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.87, v_num=2]
Epoch 31:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.87, v_num=2]
Epoch 31:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.87, v_num=2]
Epoch 31:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.87, v_num=2]
Epoch 31:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.87, v_num=2]
Epoch 31:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.87, v_num=2]
Epoch 31:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.87, v_num=2]
Epoch 31:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=6.87, v_num=2]
Epoch 31:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=6.87, v_num=2]
Epoch 31:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.87, v_num=2]
Epoch 31:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.86, v_num=2]
Epoch 31:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.86, v_num=2]
Epoch 31:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.86, v_num=2]
Epoch 31:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=6.86, v_num=2]
Epoch 31:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=6.86, v_num=2]
Epoch 31:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.86, v_num=2]
Epoch 31:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.86, v_num=2]
Epoch 31:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.86, v_num=2]
Epoch 31:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.85, v_num=2]
Epoch 31:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.85, v_num=2]
Epoch 31:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.85, v_num=2]
Epoch 31:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.85, v_num=2]
Epoch 31:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.84, v_num=2]
Epoch 31:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.84, v_num=2]
Epoch 31:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.85, v_num=2]
Epoch 31:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.85, v_num=2]
Epoch 31:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.85, v_num=2]
Epoch 31:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=6.85, v_num=2]
Epoch 31:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=6.84, v_num=2]
Epoch 31: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.84, v_num=2]
Epoch 31: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.86, v_num=2]
Epoch 31: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=6.86, v_num=2]
Epoch 31:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.86, v_num=2]
Epoch 32:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.86, v_num=2]
Epoch 32:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=6.86, v_num=2]
Epoch 32:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=6.85, v_num=2]
Epoch 32:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.85, v_num=2]
Epoch 32:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.85, v_num=2]
Epoch 32:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.85, v_num=2]
Epoch 32:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.86, v_num=2]
Epoch 32:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.86, v_num=2]
Epoch 32:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.85, v_num=2]
Epoch 32:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.85, v_num=2]
Epoch 32:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.85, v_num=2]
Epoch 32:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=6.85, v_num=2]
Epoch 32:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=6.85, v_num=2]
Epoch 32:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.85, v_num=2]
Epoch 32:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.85, v_num=2]
Epoch 32:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.85, v_num=2]
Epoch 32:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.85, v_num=2]
Epoch 32:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.85, v_num=2]
Epoch 32:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.85, v_num=2]
Epoch 32:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.85, v_num=2]
Epoch 32:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.85, v_num=2]
Epoch 32:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.85, v_num=2]
Epoch 32:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.85, v_num=2]
Epoch 32:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.85, v_num=2]
Epoch 32:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.85, v_num=2]
Epoch 32:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.85, v_num=2]
Epoch 32:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.85, v_num=2]
Epoch 32:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.85, v_num=2]
Epoch 32:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.85, v_num=2]
Epoch 32:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.85, v_num=2]
Epoch 32:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.85, v_num=2]
Epoch 32:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.85, v_num=2]
Epoch 32:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.86, v_num=2]
Epoch 32:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.86, v_num=2]
Epoch 32:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.85, v_num=2]
Epoch 32:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.85, v_num=2]
Epoch 32:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.85, v_num=2]
Epoch 32:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.85, v_num=2]
Epoch 32:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.85, v_num=2]
Epoch 32:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.85, v_num=2]
Epoch 32:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.85, v_num=2]
Epoch 32:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.85, v_num=2]
Epoch 32:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.85, v_num=2]
Epoch 32:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.85, v_num=2]
Epoch 32:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.85, v_num=2]
Epoch 32:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.85, v_num=2]
Epoch 32:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.84, v_num=2]
Epoch 32:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.84, v_num=2]
Epoch 32:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.85, v_num=2]
Epoch 32:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.85, v_num=2]
Epoch 32:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.85, v_num=2]
Epoch 32:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.85, v_num=2]
Epoch 32:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.85, v_num=2]
Epoch 32:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.85, v_num=2]
Epoch 32:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.85, v_num=2]
Epoch 32:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.85, v_num=2]
Epoch 32:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.85, v_num=2]
Epoch 32:  91%|█████████ | 29/32 [00:12<00:01,  2.34it/s, loss=6.85, v_num=2]
Epoch 32:  91%|█████████ | 29/32 [00:12<00:01,  2.34it/s, loss=6.86, v_num=2]
Epoch 32:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.86, v_num=2]
Epoch 32:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.86, v_num=2]
Epoch 32:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.86, v_num=2]
Epoch 32:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.85, v_num=2]
Epoch 32: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.85, v_num=2]
Epoch 32: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.85, v_num=2]
Epoch 32: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.85, v_num=2]
Epoch 32:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.85, v_num=2]
Epoch 33:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.85, v_num=2]
Epoch 33:   3%|▎         | 1/32 [00:03<01:55,  3.72s/it, loss=6.85, v_num=2]
Epoch 33:   3%|▎         | 1/32 [00:03<01:55,  3.72s/it, loss=6.86, v_num=2]
Epoch 33:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=6.86, v_num=2]
Epoch 33:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=6.86, v_num=2]
Epoch 33:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.86, v_num=2]
Epoch 33:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.86, v_num=2]
Epoch 33:  12%|█▎        | 4/32 [00:03<00:27,  1.04it/s, loss=6.86, v_num=2]
Epoch 33:  12%|█▎        | 4/32 [00:03<00:27,  1.03it/s, loss=6.86, v_num=2]
Epoch 33:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.86, v_num=2]
Epoch 33:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.85, v_num=2]
Epoch 33:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=6.85, v_num=2]
Epoch 33:  19%|█▉        | 6/32 [00:03<00:17,  1.51it/s, loss=6.85, v_num=2]
Epoch 33:  22%|██▏       | 7/32 [00:04<00:14,  1.75it/s, loss=6.85, v_num=2]
Epoch 33:  22%|██▏       | 7/32 [00:04<00:14,  1.75it/s, loss=6.85, v_num=2]
Epoch 33:  25%|██▌       | 8/32 [00:04<00:12,  1.97it/s, loss=6.85, v_num=2]
Epoch 33:  25%|██▌       | 8/32 [00:04<00:12,  1.97it/s, loss=6.85, v_num=2]
Epoch 33:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.85, v_num=2]
Epoch 33:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.84, v_num=2]
Epoch 33:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.84, v_num=2]
Epoch 33:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=6.84, v_num=2]
Epoch 33:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.84, v_num=2]
Epoch 33:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=6.85, v_num=2]
Epoch 33:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.85, v_num=2]
Epoch 33:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.84, v_num=2]
Epoch 33:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.84, v_num=2]
Epoch 33:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.85, v_num=2]
Epoch 33:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.85, v_num=2]
Epoch 33:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.85, v_num=2]
Epoch 33:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.85, v_num=2]
Epoch 33:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=6.84, v_num=2]
Epoch 33:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.84, v_num=2]
Epoch 33:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=6.85, v_num=2]
Epoch 33:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.85, v_num=2]
Epoch 33:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.85, v_num=2]
Epoch 33:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.85, v_num=2]
Epoch 33:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.85, v_num=2]
Epoch 33:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.85, v_num=2]
Epoch 33:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.85, v_num=2]
Epoch 33:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.85, v_num=2]
Epoch 33:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.85, v_num=2]
Epoch 33:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.85, v_num=2]
Epoch 33:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.84, v_num=2]
Epoch 33:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.84, v_num=2]
Epoch 33:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.85, v_num=2]
Epoch 33:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.85, v_num=2]
Epoch 33:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.85, v_num=2]
Epoch 33:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.85, v_num=2]
Epoch 33:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.84, v_num=2]
Epoch 33:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.84, v_num=2]
Epoch 33:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.85, v_num=2]
Epoch 33:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.85, v_num=2]
Epoch 33:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.84, v_num=2]
Epoch 33:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.84, v_num=2]
Epoch 33:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.84, v_num=2]
Epoch 33:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.84, v_num=2]
Epoch 33:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.84, v_num=2]
Epoch 33:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.84, v_num=2]
Epoch 33:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.84, v_num=2]
Epoch 33:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.84, v_num=2]
Epoch 33:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.85, v_num=2]
Epoch 33:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.85, v_num=2]
Epoch 33:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.84, v_num=2]
Epoch 33: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.84, v_num=2]
Epoch 33: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.84, v_num=2]
Epoch 33: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.84, v_num=2]
Epoch 33:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.84, v_num=2]
Epoch 34:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.84, v_num=2]
Epoch 34:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=6.84, v_num=2]
Epoch 34:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=6.84, v_num=2]
Epoch 34:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.84, v_num=2]
Epoch 34:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.85, v_num=2]
Epoch 34:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.85, v_num=2]
Epoch 34:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.85, v_num=2]
Epoch 34:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.85, v_num=2]
Epoch 34:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.85, v_num=2]
Epoch 34:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.85, v_num=2]
Epoch 34:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=6.85, v_num=2]
Epoch 34:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=6.85, v_num=2]
Epoch 34:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=6.85, v_num=2]
Epoch 34:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.85, v_num=2]
Epoch 34:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.85, v_num=2]
Epoch 34:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.85, v_num=2]
Epoch 34:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.85, v_num=2]
Epoch 34:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.85, v_num=2]
Epoch 34:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.85, v_num=2]
Epoch 34:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.85, v_num=2]
Epoch 34:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.85, v_num=2]
Epoch 34:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.85, v_num=2]
Epoch 34:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.85, v_num=2]
Epoch 34:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.85, v_num=2]
Epoch 34:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.85, v_num=2]
Epoch 34:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.85, v_num=2]
Epoch 34:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.85, v_num=2]
Epoch 34:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.85, v_num=2]
Epoch 34:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.86, v_num=2]
Epoch 34:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.86, v_num=2]
Epoch 34:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.86, v_num=2]
Epoch 34:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.86, v_num=2]
Epoch 34:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.88, v_num=2]
Epoch 34:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.88, v_num=2]
Epoch 34:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.88, v_num=2]
Epoch 34:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.88, v_num=2]
Epoch 34:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.87, v_num=2]
Epoch 34:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.87, v_num=2]
Epoch 34:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.87, v_num=2]
Epoch 34:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.87, v_num=2]
Epoch 34:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.88, v_num=2]
Epoch 34:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.88, v_num=2]
Epoch 34:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.88, v_num=2]
Epoch 34:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.88, v_num=2]
Epoch 34:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.89, v_num=2]
Epoch 34:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.89, v_num=2]
Epoch 34:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.88, v_num=2]
Epoch 34:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.88, v_num=2]
Epoch 34:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.88, v_num=2]
Epoch 34:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.88, v_num=2]
Epoch 34:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.87, v_num=2]
Epoch 34:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.87, v_num=2]
Epoch 34:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.88, v_num=2]
Epoch 34:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.88, v_num=2]
Epoch 34:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.87, v_num=2]
Epoch 34:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.87, v_num=2]
Epoch 34:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.88, v_num=2]
Epoch 34:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.88, v_num=2]
Epoch 34:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.87, v_num=2]
Epoch 34:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.87, v_num=2]
Epoch 34:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.88, v_num=2]
Epoch 34:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.88, v_num=2]
Epoch 34:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.88, v_num=2]
Epoch 34: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.88, v_num=2]
Epoch 34: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.88, v_num=2]
Epoch 34: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.88, v_num=2]
Epoch 34:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.88, v_num=2]
Epoch 35:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.88, v_num=2]
Epoch 35:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=6.88, v_num=2]
Epoch 35:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=6.89, v_num=2]
Epoch 35:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.89, v_num=2]
Epoch 35:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.89, v_num=2]
Epoch 35:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.89, v_num=2]
Epoch 35:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.89, v_num=2]
Epoch 35:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.89, v_num=2]
Epoch 35:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.89, v_num=2]
Epoch 35:  16%|█▌        | 5/32 [00:03<00:21,  1.29it/s, loss=6.89, v_num=2]
Epoch 35:  16%|█▌        | 5/32 [00:03<00:21,  1.29it/s, loss=6.89, v_num=2]
Epoch 35:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.89, v_num=2]
Epoch 35:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.89, v_num=2]
Epoch 35:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.89, v_num=2]
Epoch 35:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.89, v_num=2]
Epoch 35:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.89, v_num=2]
Epoch 35:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.89, v_num=2]
Epoch 35:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.89, v_num=2]
Epoch 35:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.89, v_num=2]
Epoch 35:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.89, v_num=2]
Epoch 35:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.88, v_num=2]
Epoch 35:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.88, v_num=2]
Epoch 35:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.89, v_num=2]
Epoch 35:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.89, v_num=2]
Epoch 35:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.89, v_num=2]
Epoch 35:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.89, v_num=2]
Epoch 35:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.9, v_num=2]
Epoch 35:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.9, v_num=2]
Epoch 35:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.9, v_num=2]
Epoch 35:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.9, v_num=2]
Epoch 35:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.9, v_num=2]
Epoch 35:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.9, v_num=2]
Epoch 35:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.9, v_num=2]
Epoch 35:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.9, v_num=2]
Epoch 35:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.91, v_num=2]
Epoch 35:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.91, v_num=2]
Epoch 35:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.91, v_num=2]
Epoch 35:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.91, v_num=2]
Epoch 35:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.91, v_num=2]
Epoch 35:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.91, v_num=2]
Epoch 35:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.92, v_num=2]
Epoch 35:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.92, v_num=2]
Epoch 35:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.91, v_num=2]
Epoch 35:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.91, v_num=2]
Epoch 35:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=6.9, v_num=2]
Epoch 35:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.9, v_num=2]
Epoch 35:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.9, v_num=2]
Epoch 35:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.9, v_num=2]
Epoch 35:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.9, v_num=2]
Epoch 35:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.9, v_num=2]
Epoch 35:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.9, v_num=2]
Epoch 35:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.9, v_num=2]
Epoch 35:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.9, v_num=2]
Epoch 35:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.9, v_num=2]
Epoch 35:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.9, v_num=2]
Epoch 35:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.9, v_num=2]
Epoch 35:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.9, v_num=2]
Epoch 35:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.9, v_num=2]
Epoch 35:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.91, v_num=2]
Epoch 35:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.91, v_num=2]
Epoch 35:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.9, v_num=2]
Epoch 35:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.9, v_num=2]
Epoch 35:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.9, v_num=2]
Epoch 35: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.9, v_num=2]
Epoch 35: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.89, v_num=2]
Epoch 35: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.89, v_num=2]
Epoch 35:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.89, v_num=2]
Epoch 36:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.89, v_num=2]
Epoch 36:   3%|▎         | 1/32 [00:03<01:52,  3.63s/it, loss=6.89, v_num=2]
Epoch 36:   3%|▎         | 1/32 [00:03<01:52,  3.63s/it, loss=6.89, v_num=2]
Epoch 36:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=6.89, v_num=2]
Epoch 36:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=6.89, v_num=2]
Epoch 36:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=6.89, v_num=2]
Epoch 36:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=6.89, v_num=2]
Epoch 36:  12%|█▎        | 4/32 [00:03<00:26,  1.06it/s, loss=6.89, v_num=2]
Epoch 36:  12%|█▎        | 4/32 [00:03<00:26,  1.06it/s, loss=6.89, v_num=2]
Epoch 36:  16%|█▌        | 5/32 [00:03<00:20,  1.31it/s, loss=6.89, v_num=2]
Epoch 36:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=6.89, v_num=2]
Epoch 36:  19%|█▉        | 6/32 [00:03<00:16,  1.55it/s, loss=6.89, v_num=2]
Epoch 36:  19%|█▉        | 6/32 [00:03<00:16,  1.55it/s, loss=6.89, v_num=2]
Epoch 36:  22%|██▏       | 7/32 [00:03<00:14,  1.78it/s, loss=6.89, v_num=2]
Epoch 36:  22%|██▏       | 7/32 [00:03<00:14,  1.78it/s, loss=6.89, v_num=2]
Epoch 36:  25%|██▌       | 8/32 [00:03<00:11,  2.02it/s, loss=6.89, v_num=2]
Epoch 36:  25%|██▌       | 8/32 [00:03<00:11,  2.01it/s, loss=6.88, v_num=2]
Epoch 36:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.88, v_num=2]
Epoch 36:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.89, v_num=2]
Epoch 36:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.89, v_num=2]
Epoch 36:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.9, v_num=2]
Epoch 36:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.9, v_num=2]
Epoch 36:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.9, v_num=2]
Epoch 36:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.9, v_num=2]
Epoch 36:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=6.9, v_num=2]
Epoch 36:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.9, v_num=2]
Epoch 36:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=6.91, v_num=2]
Epoch 36:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.91, v_num=2]
Epoch 36:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=6.91, v_num=2]
Epoch 36:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.91, v_num=2]
Epoch 36:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=6.9, v_num=2]
Epoch 36:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.9, v_num=2]
Epoch 36:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=6.91, v_num=2]
Epoch 36:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.91, v_num=2]
Epoch 36:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=6.9, v_num=2]
Epoch 36:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=6.9, v_num=2]
Epoch 36:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.9, v_num=2]
Epoch 36:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.9, v_num=2]
Epoch 36:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.9, v_num=2]
Epoch 36:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.9, v_num=2]
Epoch 36:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=6.91, v_num=2]
Epoch 36:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.91, v_num=2]
Epoch 36:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.91, v_num=2]
Epoch 36:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.91, v_num=2]
Epoch 36:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.91, v_num=2]
Epoch 36:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.91, v_num=2]
Epoch 36:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=6.91, v_num=2]
Epoch 36:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.91, v_num=2]
Epoch 36:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.92, v_num=2]
Epoch 36:  78%|███████▊  | 25/32 [00:12<00:03,  2.05it/s, loss=6.92, v_num=2]
Epoch 36:  78%|███████▊  | 25/32 [00:12<00:03,  2.05it/s, loss=6.91, v_num=2]
Epoch 36:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.91, v_num=2]
Epoch 36:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=6.91, v_num=2]
Epoch 36:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.91, v_num=2]
Epoch 36:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.91, v_num=2]
Epoch 36:  88%|████████▊ | 28/32 [00:12<00:01,  2.27it/s, loss=6.91, v_num=2]
Epoch 36:  88%|████████▊ | 28/32 [00:12<00:01,  2.27it/s, loss=6.91, v_num=2]
Epoch 36:  91%|█████████ | 29/32 [00:12<00:01,  2.34it/s, loss=6.91, v_num=2]
Epoch 36:  91%|█████████ | 29/32 [00:12<00:01,  2.34it/s, loss=6.9, v_num=2]
Epoch 36:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.9, v_num=2]
Epoch 36:  94%|█████████▍| 30/32 [00:12<00:00,  2.41it/s, loss=6.9, v_num=2]
Epoch 36:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.9, v_num=2]
Epoch 36:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=6.9, v_num=2]
Epoch 36: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.9, v_num=2]
Epoch 36: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.9, v_num=2]
Epoch 36: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.9, v_num=2]
Epoch 36:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.9, v_num=2]
Epoch 37:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.9, v_num=2]
Epoch 37:   3%|▎         | 1/32 [00:03<01:55,  3.72s/it, loss=6.9, v_num=2]
Epoch 37:   3%|▎         | 1/32 [00:03<01:55,  3.72s/it, loss=6.89, v_num=2]
Epoch 37:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=6.89, v_num=2]
Epoch 37:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=6.88, v_num=2]
Epoch 37:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.88, v_num=2]
Epoch 37:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.89, v_num=2]
Epoch 37:  12%|█▎        | 4/32 [00:03<00:27,  1.04it/s, loss=6.89, v_num=2]
Epoch 37:  12%|█▎        | 4/32 [00:03<00:27,  1.04it/s, loss=6.89, v_num=2]
Epoch 37:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.89, v_num=2]
Epoch 37:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.89, v_num=2]
Epoch 37:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.89, v_num=2]
Epoch 37:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.9, v_num=2]
Epoch 37:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=6.9, v_num=2]
Epoch 37:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=6.9, v_num=2]
Epoch 37:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.9, v_num=2]
Epoch 37:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.89, v_num=2]
Epoch 37:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=6.89, v_num=2]
Epoch 37:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=6.89, v_num=2]
Epoch 37:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=6.89, v_num=2]
Epoch 37:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=6.88, v_num=2]
Epoch 37:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=6.88, v_num=2]
Epoch 37:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=6.88, v_num=2]
Epoch 37:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=6.88, v_num=2]
Epoch 37:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=6.88, v_num=2]
Epoch 37:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.88, v_num=2]
Epoch 37:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=6.89, v_num=2]
Epoch 37:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.89, v_num=2]
Epoch 37:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=6.89, v_num=2]
Epoch 37:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=6.89, v_num=2]
Epoch 37:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=6.89, v_num=2]
Epoch 37:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=6.89, v_num=2]
Epoch 37:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=6.89, v_num=2]
Epoch 37:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=6.89, v_num=2]
Epoch 37:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=6.89, v_num=2]
Epoch 37:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.89, v_num=2]
Epoch 37:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=6.88, v_num=2]
Epoch 37:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=6.88, v_num=2]
Epoch 37:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=6.89, v_num=2]
Epoch 37:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.89, v_num=2]
Epoch 37:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=6.89, v_num=2]
Epoch 37:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=6.89, v_num=2]
Epoch 37:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=6.89, v_num=2]
Epoch 37:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=6.89, v_num=2]
Epoch 37:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=6.88, v_num=2]
Epoch 37:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=6.88, v_num=2]
Epoch 37:  72%|███████▏  | 23/32 [00:09<00:03,  2.35it/s, loss=6.87, v_num=2]
Epoch 37:  75%|███████▌  | 24/32 [00:09<00:03,  2.44it/s, loss=6.87, v_num=2]
Epoch 37:  75%|███████▌  | 24/32 [00:09<00:03,  2.44it/s, loss=6.87, v_num=2]
Epoch 37:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.87, v_num=2]
Epoch 37:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.87, v_num=2]
Epoch 37:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=6.87, v_num=2]
Epoch 37:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=6.87, v_num=2]
Epoch 37:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=6.87, v_num=2]
Epoch 37:  84%|████████▍ | 27/32 [00:12<00:02,  2.17it/s, loss=6.87, v_num=2]
Epoch 37:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.87, v_num=2]
Epoch 37:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.87, v_num=2]
Epoch 37:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.87, v_num=2]
Epoch 37:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.87, v_num=2]
Epoch 37:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.87, v_num=2]
Epoch 37:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.87, v_num=2]
Epoch 37:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=6.87, v_num=2]
Epoch 37:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=6.87, v_num=2]
Epoch 37: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=6.87, v_num=2]
Epoch 37: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=6.87, v_num=2]
Epoch 37: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=6.87, v_num=2]
Epoch 37:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.87, v_num=2]
Epoch 38:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.87, v_num=2]
Epoch 38:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=6.87, v_num=2]
Epoch 38:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=6.87, v_num=2]
Epoch 38:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.87, v_num=2]
Epoch 38:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.87, v_num=2]
Epoch 38:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.87, v_num=2]
Epoch 38:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=6.87, v_num=2]
Epoch 38:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.87, v_num=2]
Epoch 38:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.88, v_num=2]
Epoch 38:  16%|█▌        | 5/32 [00:03<00:21,  1.29it/s, loss=6.88, v_num=2]
Epoch 38:  16%|█▌        | 5/32 [00:03<00:21,  1.29it/s, loss=6.88, v_num=2]
Epoch 38:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.88, v_num=2]
Epoch 38:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.88, v_num=2]
Epoch 38:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.88, v_num=2]
Epoch 38:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.87, v_num=2]
Epoch 38:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.87, v_num=2]
Epoch 38:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=6.88, v_num=2]
Epoch 38:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.88, v_num=2]
Epoch 38:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=6.88, v_num=2]
Epoch 38:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.88, v_num=2]
Epoch 38:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.88, v_num=2]
Epoch 38:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.88, v_num=2]
Epoch 38:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.88, v_num=2]
Epoch 38:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.88, v_num=2]
Epoch 38:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.89, v_num=2]
Epoch 38:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.89, v_num=2]
Epoch 38:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.88, v_num=2]
Epoch 38:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.88, v_num=2]
Epoch 38:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.88, v_num=2]
Epoch 38:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.88, v_num=2]
Epoch 38:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.88, v_num=2]
Epoch 38:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.88, v_num=2]
Epoch 38:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.88, v_num=2]
Epoch 38:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.88, v_num=2]
Epoch 38:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.88, v_num=2]
Epoch 38:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.88, v_num=2]
Epoch 38:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.9, v_num=2]
Epoch 38:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.9, v_num=2]
Epoch 38:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=6.9, v_num=2]
Epoch 38:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.9, v_num=2]
Epoch 38:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.89, v_num=2]
Epoch 38:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=6.89, v_num=2]
Epoch 38:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=6.89, v_num=2]
Epoch 38:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.89, v_num=2]
Epoch 38:  69%|██████▉   | 22/32 [00:09<00:04,  2.27it/s, loss=6.89, v_num=2]
Epoch 38:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.89, v_num=2]
Epoch 38:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=6.89, v_num=2]
Epoch 38:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.89, v_num=2]
Epoch 38:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=6.88, v_num=2]
Epoch 38:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.88, v_num=2]
Epoch 38:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=6.88, v_num=2]
Epoch 38:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.88, v_num=2]
Epoch 38:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.88, v_num=2]
Epoch 38:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.88, v_num=2]
Epoch 38:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=6.89, v_num=2]
Epoch 38:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.89, v_num=2]
Epoch 38:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=6.9, v_num=2]
Epoch 38:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.9, v_num=2]
Epoch 38:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=6.9, v_num=2]
Epoch 38:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.9, v_num=2]
Epoch 38:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=6.91, v_num=2]
Epoch 38:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.91, v_num=2]
Epoch 38:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.91, v_num=2]
Epoch 38: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.91, v_num=2]
Epoch 38: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.91, v_num=2]
Epoch 38: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=6.91, v_num=2]
Epoch 38:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.91, v_num=2]
Epoch 39:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.91, v_num=2]
Epoch 39:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=6.91, v_num=2]
Epoch 39:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=6.91, v_num=2]
Epoch 39:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.91, v_num=2]
Epoch 39:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=6.91, v_num=2]
Epoch 39:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.91, v_num=2]
Epoch 39:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=6.93, v_num=2]
Epoch 39:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.93, v_num=2]
Epoch 39:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=6.93, v_num=2]
Epoch 39:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.93, v_num=2]
Epoch 39:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=6.93, v_num=2]
Epoch 39:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.93, v_num=2]
Epoch 39:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=6.92, v_num=2]
Epoch 39:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.92, v_num=2]
Epoch 39:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=6.93, v_num=2]
Epoch 39:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.93, v_num=2]
Epoch 39:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=6.93, v_num=2]
Epoch 39:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.93, v_num=2]
Epoch 39:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=6.94, v_num=2]
Epoch 39:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.94, v_num=2]
Epoch 39:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=6.94, v_num=2]
Epoch 39:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.94, v_num=2]
Epoch 39:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=6.95, v_num=2]
Epoch 39:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.95, v_num=2]
Epoch 39:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=6.96, v_num=2]
Epoch 39:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.96, v_num=2]
Epoch 39:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=6.96, v_num=2]
Epoch 39:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.96, v_num=2]
Epoch 39:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=6.96, v_num=2]
Epoch 39:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.96, v_num=2]
Epoch 39:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=6.96, v_num=2]
Epoch 39:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.96, v_num=2]
Epoch 39:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=6.96, v_num=2]
Epoch 39:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.96, v_num=2]
Epoch 39:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=6.96, v_num=2]
Epoch 39:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.96, v_num=2]
Epoch 39:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=6.96, v_num=2]
Epoch 39:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.96, v_num=2]
Epoch 39:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=6.96, v_num=2]
Epoch 39:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.96, v_num=2]
Epoch 39:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=6.97, v_num=2]
Epoch 39:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.97, v_num=2]
Epoch 39:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=6.97, v_num=2]
Epoch 39:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.97, v_num=2]
Epoch 39:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=6.97, v_num=2]
Epoch 39:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.97, v_num=2]
Epoch 39:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=6.97, v_num=2]
Epoch 39:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.97, v_num=2]
Epoch 39:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=6.97, v_num=2]
Epoch 39:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.97, v_num=2]
Epoch 39:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=6.98, v_num=2]
Epoch 39:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.98, v_num=2]
Epoch 39:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=6.98, v_num=2]
Epoch 39:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.98, v_num=2]
Epoch 39:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=6.98, v_num=2]
Epoch 39:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.98, v_num=2]
Epoch 39:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=6.98, v_num=2]
Epoch 39:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.98, v_num=2]
Epoch 39:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=6.99, v_num=2]
Epoch 39:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.99, v_num=2]
Epoch 39:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=6.98, v_num=2]
Epoch 39:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.98, v_num=2]
Epoch 39:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=6.99, v_num=2]
Epoch 39: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.99, v_num=2]
Epoch 39: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.98, v_num=2]
Epoch 39: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=6.98, v_num=2]
Epoch 39:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.98, v_num=2]
Epoch 40:   0%|          | 0/32 [00:00<?, ?it/s, loss=6.98, v_num=2]
Epoch 40:   3%|▎         | 1/32 [00:03<01:53,  3.65s/it, loss=6.98, v_num=2]
Epoch 40:   3%|▎         | 1/32 [00:03<01:53,  3.65s/it, loss=6.99, v_num=2]
Epoch 40:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=6.99, v_num=2]
Epoch 40:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=6.99, v_num=2]
Epoch 40:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=6.99, v_num=2]
Epoch 40:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7, v_num=2]
Epoch 40:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7, v_num=2]
Epoch 40:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7, v_num=2]
Epoch 40:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7, v_num=2]
Epoch 40:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7, v_num=2]
Epoch 40:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7, v_num=2]
Epoch 40:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7.01, v_num=2]
Epoch 40:  22%|██▏       | 7/32 [00:03<00:14,  1.78it/s, loss=7.01, v_num=2]
Epoch 40:  22%|██▏       | 7/32 [00:03<00:14,  1.77it/s, loss=7.01, v_num=2]
Epoch 40:  25%|██▌       | 8/32 [00:03<00:11,  2.01it/s, loss=7.01, v_num=2]
Epoch 40:  25%|██▌       | 8/32 [00:03<00:11,  2.00it/s, loss=7.01, v_num=2]
Epoch 40:  28%|██▊       | 9/32 [00:06<00:16,  1.39it/s, loss=7.01, v_num=2]
Epoch 40:  28%|██▊       | 9/32 [00:06<00:16,  1.39it/s, loss=7.02, v_num=2]
Epoch 40:  31%|███▏      | 10/32 [00:06<00:14,  1.53it/s, loss=7.02, v_num=2]
Epoch 40:  31%|███▏      | 10/32 [00:06<00:14,  1.53it/s, loss=7.02, v_num=2]
Epoch 40:  34%|███▍      | 11/32 [00:06<00:12,  1.67it/s, loss=7.02, v_num=2]
Epoch 40:  34%|███▍      | 11/32 [00:06<00:12,  1.67it/s, loss=7.02, v_num=2]
Epoch 40:  38%|███▊      | 12/32 [00:06<00:11,  1.81it/s, loss=7.02, v_num=2]
Epoch 40:  38%|███▊      | 12/32 [00:06<00:11,  1.81it/s, loss=7.02, v_num=2]
Epoch 40:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.02, v_num=2]
Epoch 40:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.03, v_num=2]
Epoch 40:  44%|████▍     | 14/32 [00:06<00:08,  2.08it/s, loss=7.03, v_num=2]
Epoch 40:  44%|████▍     | 14/32 [00:06<00:08,  2.08it/s, loss=7.03, v_num=2]
Epoch 40:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.03, v_num=2]
Epoch 40:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.03, v_num=2]
Epoch 40:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.03, v_num=2]
Epoch 40:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.03, v_num=2]
Epoch 40:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.03, v_num=2]
Epoch 40:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.04, v_num=2]
Epoch 40:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.04, v_num=2]
Epoch 40:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.04, v_num=2]
Epoch 40:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.04, v_num=2]
Epoch 40:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.04, v_num=2]
Epoch 40:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.04, v_num=2]
Epoch 40:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.04, v_num=2]
Epoch 40:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.04, v_num=2]
Epoch 40:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.04, v_num=2]
Epoch 40:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.04, v_num=2]
Epoch 40:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.04, v_num=2]
Epoch 40:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.04, v_num=2]
Epoch 40:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.04, v_num=2]
Epoch 40:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.04, v_num=2]
Epoch 40:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.04, v_num=2]
Epoch 40:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.04, v_num=2]
Epoch 40:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.04, v_num=2]
Epoch 40:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=7.04, v_num=2]
Epoch 40:  81%|████████▏ | 26/32 [00:12<00:02,  2.12it/s, loss=7.04, v_num=2]
Epoch 40:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.04, v_num=2]
Epoch 40:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.05, v_num=2]
Epoch 40:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.05, v_num=2]
Epoch 40:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.05, v_num=2]
Epoch 40:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.05, v_num=2]
Epoch 40:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.05, v_num=2]
Epoch 40:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.05, v_num=2]
Epoch 40:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.05, v_num=2]
Epoch 40:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.05, v_num=2]
Epoch 40:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.05, v_num=2]
Epoch 40: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.05, v_num=2]
Epoch 40: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.06, v_num=2]
Epoch 40: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.06, v_num=2]
Epoch 40:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.06, v_num=2]
Epoch 41:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.06, v_num=2]
Epoch 41:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.06, v_num=2]
Epoch 41:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.05, v_num=2]
Epoch 41:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.05, v_num=2]
Epoch 41:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.05, v_num=2]
Epoch 41:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.05, v_num=2]
Epoch 41:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.05, v_num=2]
Epoch 41:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.05, v_num=2]
Epoch 41:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.06, v_num=2]
Epoch 41:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.06, v_num=2]
Epoch 41:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.05, v_num=2]
Epoch 41:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.05, v_num=2]
Epoch 41:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.05, v_num=2]
Epoch 41:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.05, v_num=2]
Epoch 41:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.05, v_num=2]
Epoch 41:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.05, v_num=2]
Epoch 41:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.06, v_num=2]
Epoch 41:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.06, v_num=2]
Epoch 41:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.07, v_num=2]
Epoch 41:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.07, v_num=2]
Epoch 41:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.07, v_num=2]
Epoch 41:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.07, v_num=2]
Epoch 41:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.08, v_num=2]
Epoch 41:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.08, v_num=2]
Epoch 41:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.07, v_num=2]
Epoch 41:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.07, v_num=2]
Epoch 41:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.08, v_num=2]
Epoch 41:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.08, v_num=2]
Epoch 41:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.09, v_num=2]
Epoch 41:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.09, v_num=2]
Epoch 41:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.09, v_num=2]
Epoch 41:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.09, v_num=2]
Epoch 41:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.09, v_num=2]
Epoch 41:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.09, v_num=2]
Epoch 41:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.09, v_num=2]
Epoch 41:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.09, v_num=2]
Epoch 41:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.09, v_num=2]
Epoch 41:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.09, v_num=2]
Epoch 41:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.09, v_num=2]
Epoch 41:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.09, v_num=2]
Epoch 41:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.1, v_num=2]
Epoch 41:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.1, v_num=2]
Epoch 41:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.1, v_num=2]
Epoch 41:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.1, v_num=2]
Epoch 41:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.11, v_num=2]
Epoch 41:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.11, v_num=2]
Epoch 41:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.11, v_num=2]
Epoch 41:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.11, v_num=2]
Epoch 41:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.11, v_num=2]
Epoch 41:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.11, v_num=2]
Epoch 41:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.11, v_num=2]
Epoch 41:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.11, v_num=2]
Epoch 41:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.12, v_num=2]
Epoch 41:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.12, v_num=2]
Epoch 41:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.12, v_num=2]
Epoch 41:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.12, v_num=2]
Epoch 41:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.12, v_num=2]
Epoch 41:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.12, v_num=2]
Epoch 41:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.11, v_num=2]
Epoch 41:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.11, v_num=2]
Epoch 41:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.11, v_num=2]
Epoch 41:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.11, v_num=2]
Epoch 41:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.11, v_num=2]
Epoch 41: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.11, v_num=2]
Epoch 41: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.12, v_num=2]
Epoch 41: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.12, v_num=2]
Epoch 41:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.12, v_num=2]
Epoch 42:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.12, v_num=2]
Epoch 42:   3%|▎         | 1/32 [00:03<01:55,  3.71s/it, loss=7.12, v_num=2]
Epoch 42:   3%|▎         | 1/32 [00:03<01:55,  3.71s/it, loss=7.12, v_num=2]
Epoch 42:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.12, v_num=2]
Epoch 42:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.12, v_num=2]
Epoch 42:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.12, v_num=2]
Epoch 42:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.11, v_num=2]
Epoch 42:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.11, v_num=2]
Epoch 42:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.11, v_num=2]
Epoch 42:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.11, v_num=2]
Epoch 42:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.11, v_num=2]
Epoch 42:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.11, v_num=2]
Epoch 42:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.12, v_num=2]
Epoch 42:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.12, v_num=2]
Epoch 42:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.12, v_num=2]
Epoch 42:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.12, v_num=2]
Epoch 42:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.13, v_num=2]
Epoch 42:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=7.13, v_num=2]
Epoch 42:  28%|██▊       | 9/32 [00:06<00:16,  1.36it/s, loss=7.13, v_num=2]
Epoch 42:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=7.13, v_num=2]
Epoch 42:  31%|███▏      | 10/32 [00:06<00:14,  1.50it/s, loss=7.13, v_num=2]
Epoch 42:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=7.13, v_num=2]
Epoch 42:  34%|███▍      | 11/32 [00:06<00:12,  1.64it/s, loss=7.13, v_num=2]
Epoch 42:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.13, v_num=2]
Epoch 42:  38%|███▊      | 12/32 [00:06<00:11,  1.78it/s, loss=7.13, v_num=2]
Epoch 42:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=7.13, v_num=2]
Epoch 42:  41%|████      | 13/32 [00:06<00:09,  1.91it/s, loss=7.13, v_num=2]
Epoch 42:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=7.13, v_num=2]
Epoch 42:  44%|████▍     | 14/32 [00:06<00:08,  2.05it/s, loss=7.13, v_num=2]
Epoch 42:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=7.13, v_num=2]
Epoch 42:  47%|████▋     | 15/32 [00:06<00:07,  2.18it/s, loss=7.14, v_num=2]
Epoch 42:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=7.14, v_num=2]
Epoch 42:  50%|█████     | 16/32 [00:06<00:06,  2.31it/s, loss=7.14, v_num=2]
Epoch 42:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.14, v_num=2]
Epoch 42:  53%|█████▎    | 17/32 [00:09<00:08,  1.79it/s, loss=7.15, v_num=2]
Epoch 42:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.15, v_num=2]
Epoch 42:  56%|█████▋    | 18/32 [00:09<00:07,  1.89it/s, loss=7.15, v_num=2]
Epoch 42:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.15, v_num=2]
Epoch 42:  59%|█████▉    | 19/32 [00:09<00:06,  1.98it/s, loss=7.16, v_num=2]
Epoch 42:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.16, v_num=2]
Epoch 42:  62%|██████▎   | 20/32 [00:09<00:05,  2.08it/s, loss=7.16, v_num=2]
Epoch 42:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=7.16, v_num=2]
Epoch 42:  66%|██████▌   | 21/32 [00:09<00:05,  2.17it/s, loss=7.16, v_num=2]
Epoch 42:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=7.16, v_num=2]
Epoch 42:  69%|██████▉   | 22/32 [00:09<00:04,  2.26it/s, loss=7.17, v_num=2]
Epoch 42:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.17, v_num=2]
Epoch 42:  72%|███████▏  | 23/32 [00:09<00:03,  2.36it/s, loss=7.17, v_num=2]
Epoch 42:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.17, v_num=2]
Epoch 42:  75%|███████▌  | 24/32 [00:09<00:03,  2.45it/s, loss=7.18, v_num=2]
Epoch 42:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.18, v_num=2]
Epoch 42:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.18, v_num=2]
Epoch 42:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.18, v_num=2]
Epoch 42:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.18, v_num=2]
Epoch 42:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.18, v_num=2]
Epoch 42:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.18, v_num=2]
Epoch 42:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.18, v_num=2]
Epoch 42:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.18, v_num=2]
Epoch 42:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.18, v_num=2]
Epoch 42:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.18, v_num=2]
Epoch 42:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.18, v_num=2]
Epoch 42:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.19, v_num=2]
Epoch 42:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.19, v_num=2]
Epoch 42:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.19, v_num=2]
Epoch 42: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.19, v_num=2]
Epoch 42: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.2, v_num=2]
Epoch 42: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.2, v_num=2]
Epoch 42:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.2, v_num=2]
Epoch 43:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.2, v_num=2]
Epoch 43:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.2, v_num=2]
Epoch 43:   3%|▎         | 1/32 [00:03<01:55,  3.71s/it, loss=7.2, v_num=2]
Epoch 43:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.2, v_num=2]
Epoch 43:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.2, v_num=2]
Epoch 43:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.2, v_num=2]
Epoch 43:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.2, v_num=2]
Epoch 43:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.2, v_num=2]
Epoch 43:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.2, v_num=2]
Epoch 43:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.2, v_num=2]
Epoch 43:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.21, v_num=2]
Epoch 43:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.21, v_num=2]
Epoch 43:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.22, v_num=2]
Epoch 43:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.22, v_num=2]
Epoch 43:  22%|██▏       | 7/32 [00:03<00:14,  1.75it/s, loss=7.21, v_num=2]
Epoch 43:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.21, v_num=2]
Epoch 43:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.21, v_num=2]
Epoch 43:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.21, v_num=2]
Epoch 43:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.21, v_num=2]
Epoch 43:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.21, v_num=2]
Epoch 43:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.21, v_num=2]
Epoch 43:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.21, v_num=2]
Epoch 43:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.21, v_num=2]
Epoch 43:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.21, v_num=2]
Epoch 43:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.21, v_num=2]
Epoch 43:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=7.21, v_num=2]
Epoch 43:  41%|████      | 13/32 [00:06<00:09,  1.92it/s, loss=7.21, v_num=2]
Epoch 43:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.21, v_num=2]
Epoch 43:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.22, v_num=2]
Epoch 43:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.22, v_num=2]
Epoch 43:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.23, v_num=2]
Epoch 43:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.23, v_num=2]
Epoch 43:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.23, v_num=2]
Epoch 43:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.23, v_num=2]
Epoch 43:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.23, v_num=2]
Epoch 43:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.23, v_num=2]
Epoch 43:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.24, v_num=2]
Epoch 43:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.24, v_num=2]
Epoch 43:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.23, v_num=2]
Epoch 43:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.23, v_num=2]
Epoch 43:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.23, v_num=2]
Epoch 43:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.23, v_num=2]
Epoch 43:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.24, v_num=2]
Epoch 43:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.24, v_num=2]
Epoch 43:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.25, v_num=2]
Epoch 43:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.25, v_num=2]
Epoch 43:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.25, v_num=2]
Epoch 43:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.25, v_num=2]
Epoch 43:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.26, v_num=2]
Epoch 43:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.26, v_num=2]
Epoch 43:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.25, v_num=2]
Epoch 43:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.25, v_num=2]
Epoch 43:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.25, v_num=2]
Epoch 43:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.25, v_num=2]
Epoch 43:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.25, v_num=2]
Epoch 43:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.25, v_num=2]
Epoch 43:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.26, v_num=2]
Epoch 43:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.26, v_num=2]
Epoch 43:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.27, v_num=2]
Epoch 43:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.27, v_num=2]
Epoch 43:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.27, v_num=2]
Epoch 43:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.27, v_num=2]
Epoch 43:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.27, v_num=2]
Epoch 43: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.27, v_num=2]
Epoch 43: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.27, v_num=2]
Epoch 43: 100%|██████████| 32/32 [00:12<00:00,  2.54it/s, loss=7.27, v_num=2]
Epoch 43:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.27, v_num=2]
Epoch 44:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.27, v_num=2]
Epoch 44:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=7.27, v_num=2]
Epoch 44:   3%|▎         | 1/32 [00:03<01:54,  3.69s/it, loss=7.28, v_num=2]
Epoch 44:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.28, v_num=2]
Epoch 44:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.28, v_num=2]
Epoch 44:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.28, v_num=2]
Epoch 44:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.27, v_num=2]
Epoch 44:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.27, v_num=2]
Epoch 44:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.27, v_num=2]
Epoch 44:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.27, v_num=2]
Epoch 44:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.28, v_num=2]
Epoch 44:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=7.28, v_num=2]
Epoch 44:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=7.29, v_num=2]
Epoch 44:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.29, v_num=2]
Epoch 44:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.29, v_num=2]
Epoch 44:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.29, v_num=2]
Epoch 44:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.29, v_num=2]
Epoch 44:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.29, v_num=2]
Epoch 44:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.3, v_num=2]
Epoch 44:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.3, v_num=2]
Epoch 44:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.3, v_num=2]
Epoch 44:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.3, v_num=2]
Epoch 44:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.3, v_num=2]
Epoch 44:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.3, v_num=2]
Epoch 44:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.3, v_num=2]
Epoch 44:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.3, v_num=2]
Epoch 44:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.3, v_num=2]
Epoch 44:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.3, v_num=2]
Epoch 44:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.3, v_num=2]
Epoch 44:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.3, v_num=2]
Epoch 44:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.31, v_num=2]
Epoch 44:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.31, v_num=2]
Epoch 44:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.31, v_num=2]
Epoch 44:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.31, v_num=2]
Epoch 44:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.31, v_num=2]
Epoch 44:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.31, v_num=2]
Epoch 44:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.32, v_num=2]
Epoch 44:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.32, v_num=2]
Epoch 44:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.32, v_num=2]
Epoch 44:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.32, v_num=2]
Epoch 44:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.32, v_num=2]
Epoch 44:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.32, v_num=2]
Epoch 44:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.32, v_num=2]
Epoch 44:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.32, v_num=2]
Epoch 44:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.32, v_num=2]
Epoch 44:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.32, v_num=2]
Epoch 44:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.32, v_num=2]
Epoch 44:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.32, v_num=2]
Epoch 44:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.33, v_num=2]
Epoch 44:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.33, v_num=2]
Epoch 44:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.33, v_num=2]
Epoch 44:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.33, v_num=2]
Epoch 44:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.34, v_num=2]
Epoch 44:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.34, v_num=2]
Epoch 44:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.34, v_num=2]
Epoch 44:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.34, v_num=2]
Epoch 44:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.34, v_num=2]
Epoch 44:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.34, v_num=2]
Epoch 44:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.34, v_num=2]
Epoch 44:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.34, v_num=2]
Epoch 44:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.34, v_num=2]
Epoch 44:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.34, v_num=2]
Epoch 44:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.34, v_num=2]
Epoch 44: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.34, v_num=2]
Epoch 44: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.33, v_num=2]
Epoch 44: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.33, v_num=2]
Epoch 44:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.33, v_num=2]
Epoch 45:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.33, v_num=2]
Epoch 45:   3%|▎         | 1/32 [00:03<01:54,  3.68s/it, loss=7.33, v_num=2]
Epoch 45:   3%|▎         | 1/32 [00:03<01:54,  3.68s/it, loss=7.33, v_num=2]
Epoch 45:   6%|▋         | 2/32 [00:03<00:55,  1.86s/it, loss=7.33, v_num=2]
Epoch 45:   6%|▋         | 2/32 [00:03<00:55,  1.87s/it, loss=7.34, v_num=2]
Epoch 45:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.34, v_num=2]
Epoch 45:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.34, v_num=2]
Epoch 45:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.34, v_num=2]
Epoch 45:  12%|█▎        | 4/32 [00:03<00:26,  1.05it/s, loss=7.34, v_num=2]
Epoch 45:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.34, v_num=2]
Epoch 45:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.34, v_num=2]
Epoch 45:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.34, v_num=2]
Epoch 45:  19%|█▉        | 6/32 [00:03<00:16,  1.53it/s, loss=7.35, v_num=2]
Epoch 45:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.35, v_num=2]
Epoch 45:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.35, v_num=2]
Epoch 45:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.35, v_num=2]
Epoch 45:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.34, v_num=2]
Epoch 45:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.34, v_num=2]
Epoch 45:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.35, v_num=2]
Epoch 45:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.35, v_num=2]
Epoch 45:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.35, v_num=2]
Epoch 45:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.35, v_num=2]
Epoch 45:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.35, v_num=2]
Epoch 45:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.35, v_num=2]
Epoch 45:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.35, v_num=2]
Epoch 45:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.35, v_num=2]
Epoch 45:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.34, v_num=2]
Epoch 45:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.34, v_num=2]
Epoch 45:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.34, v_num=2]
Epoch 45:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.34, v_num=2]
Epoch 45:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.35, v_num=2]
Epoch 45:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.35, v_num=2]
Epoch 45:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.35, v_num=2]
Epoch 45:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.35, v_num=2]
Epoch 45:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.35, v_num=2]
Epoch 45:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.35, v_num=2]
Epoch 45:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.34, v_num=2]
Epoch 45:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.34, v_num=2]
Epoch 45:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.35, v_num=2]
Epoch 45:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.35, v_num=2]
Epoch 45:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.36, v_num=2]
Epoch 45:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.36, v_num=2]
Epoch 45:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.36, v_num=2]
Epoch 45:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.36, v_num=2]
Epoch 45:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.37, v_num=2]
Epoch 45:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.37, v_num=2]
Epoch 45:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.37, v_num=2]
Epoch 45:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.37, v_num=2]
Epoch 45:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.37, v_num=2]
Epoch 45:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.37, v_num=2]
Epoch 45:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.38, v_num=2]
Epoch 45:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.38, v_num=2]
Epoch 45:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.38, v_num=2]
Epoch 45:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.38, v_num=2]
Epoch 45:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.38, v_num=2]
Epoch 45:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.38, v_num=2]
Epoch 45:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.38, v_num=2]
Epoch 45:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.38, v_num=2]
Epoch 45:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.39, v_num=2]
Epoch 45:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.39, v_num=2]
Epoch 45:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.39, v_num=2]
Epoch 45:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.39, v_num=2]
Epoch 45:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.39, v_num=2]
Epoch 45: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.39, v_num=2]
Epoch 45: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.4, v_num=2]
Epoch 45: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.4, v_num=2]
Epoch 45:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.4, v_num=2]
Epoch 46:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.4, v_num=2]
Epoch 46:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.4, v_num=2]
Epoch 46:   3%|▎         | 1/32 [00:03<01:54,  3.70s/it, loss=7.4, v_num=2]
Epoch 46:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.4, v_num=2]
Epoch 46:   6%|▋         | 2/32 [00:03<00:56,  1.87s/it, loss=7.41, v_num=2]
Epoch 46:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.41, v_num=2]
Epoch 46:   9%|▉         | 3/32 [00:03<00:36,  1.26s/it, loss=7.41, v_num=2]
Epoch 46:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.41, v_num=2]
Epoch 46:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.42, v_num=2]
Epoch 46:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.42, v_num=2]
Epoch 46:  16%|█▌        | 5/32 [00:03<00:20,  1.29it/s, loss=7.42, v_num=2]
Epoch 46:  19%|█▉        | 6/32 [00:03<00:17,  1.53it/s, loss=7.42, v_num=2]
Epoch 46:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.42, v_num=2]
Epoch 46:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.42, v_num=2]
Epoch 46:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.43, v_num=2]
Epoch 46:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.43, v_num=2]
Epoch 46:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.43, v_num=2]
Epoch 46:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.43, v_num=2]
Epoch 46:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.43, v_num=2]
Epoch 46:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.43, v_num=2]
Epoch 46:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.42, v_num=2]
Epoch 46:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.42, v_num=2]
Epoch 46:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.42, v_num=2]
Epoch 46:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.42, v_num=2]
Epoch 46:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.42, v_num=2]
Epoch 46:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.42, v_num=2]
Epoch 46:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.42, v_num=2]
Epoch 46:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.42, v_num=2]
Epoch 46:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.42, v_num=2]
Epoch 46:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.42, v_num=2]
Epoch 46:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.43, v_num=2]
Epoch 46:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.43, v_num=2]
Epoch 46:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.43, v_num=2]
Epoch 46:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.43, v_num=2]
Epoch 46:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.43, v_num=2]
Epoch 46:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.43, v_num=2]
Epoch 46:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.43, v_num=2]
Epoch 46:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.43, v_num=2]
Epoch 46:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.43, v_num=2]
Epoch 46:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.43, v_num=2]
Epoch 46:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.43, v_num=2]
Epoch 46:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.43, v_num=2]
Epoch 46:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.43, v_num=2]
Epoch 46:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.43, v_num=2]
Epoch 46:  69%|██████▉   | 22/32 [00:09<00:04,  2.29it/s, loss=7.42, v_num=2]
Epoch 46:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.42, v_num=2]
Epoch 46:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.43, v_num=2]
Epoch 46:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.43, v_num=2]
Epoch 46:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.42, v_num=2]
Epoch 46:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.42, v_num=2]
Epoch 46:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.42, v_num=2]
Epoch 46:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.42, v_num=2]
Epoch 46:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.43, v_num=2]
Epoch 46:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.43, v_num=2]
Epoch 46:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.43, v_num=2]
Epoch 46:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.43, v_num=2]
Epoch 46:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.43, v_num=2]
Epoch 46:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.43, v_num=2]
Epoch 46:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.43, v_num=2]
Epoch 46:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.43, v_num=2]
Epoch 46:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.43, v_num=2]
Epoch 46:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.43, v_num=2]
Epoch 46:  97%|█████████▋| 31/32 [00:12<00:00,  2.48it/s, loss=7.44, v_num=2]
Epoch 46: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.44, v_num=2]
Epoch 46: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.44, v_num=2]
Epoch 46: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.44, v_num=2]
Epoch 46:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.44, v_num=2]
Epoch 47:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.44, v_num=2]
Epoch 47:   3%|▎         | 1/32 [00:03<01:52,  3.64s/it, loss=7.44, v_num=2]
Epoch 47:   3%|▎         | 1/32 [00:03<01:52,  3.64s/it, loss=7.44, v_num=2]
Epoch 47:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=7.44, v_num=2]
Epoch 47:   6%|▋         | 2/32 [00:03<00:55,  1.85s/it, loss=7.45, v_num=2]
Epoch 47:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.45, v_num=2]
Epoch 47:   9%|▉         | 3/32 [00:03<00:36,  1.25s/it, loss=7.44, v_num=2]
Epoch 47:  12%|█▎        | 4/32 [00:03<00:26,  1.06it/s, loss=7.44, v_num=2]
Epoch 47:  12%|█▎        | 4/32 [00:03<00:26,  1.06it/s, loss=7.44, v_num=2]
Epoch 47:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7.44, v_num=2]
Epoch 47:  16%|█▌        | 5/32 [00:03<00:20,  1.30it/s, loss=7.44, v_num=2]
Epoch 47:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7.44, v_num=2]
Epoch 47:  19%|█▉        | 6/32 [00:03<00:16,  1.54it/s, loss=7.44, v_num=2]
Epoch 47:  22%|██▏       | 7/32 [00:03<00:14,  1.78it/s, loss=7.44, v_num=2]
Epoch 47:  22%|██▏       | 7/32 [00:03<00:14,  1.78it/s, loss=7.45, v_num=2]
Epoch 47:  25%|██▌       | 8/32 [00:03<00:11,  2.01it/s, loss=7.45, v_num=2]
Epoch 47:  25%|██▌       | 8/32 [00:03<00:11,  2.01it/s, loss=7.45, v_num=2]
Epoch 47:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.45, v_num=2]
Epoch 47:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.45, v_num=2]
Epoch 47:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.45, v_num=2]
Epoch 47:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.46, v_num=2]
Epoch 47:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.46, v_num=2]
Epoch 47:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.46, v_num=2]
Epoch 47:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.46, v_num=2]
Epoch 47:  38%|███▊      | 12/32 [00:06<00:11,  1.80it/s, loss=7.46, v_num=2]
Epoch 47:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.46, v_num=2]
Epoch 47:  41%|████      | 13/32 [00:06<00:09,  1.94it/s, loss=7.46, v_num=2]
Epoch 47:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.46, v_num=2]
Epoch 47:  44%|████▍     | 14/32 [00:06<00:08,  2.07it/s, loss=7.46, v_num=2]
Epoch 47:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.46, v_num=2]
Epoch 47:  47%|████▋     | 15/32 [00:06<00:07,  2.21it/s, loss=7.46, v_num=2]
Epoch 47:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.46, v_num=2]
Epoch 47:  50%|█████     | 16/32 [00:06<00:06,  2.34it/s, loss=7.46, v_num=2]
Epoch 47:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.46, v_num=2]
Epoch 47:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.46, v_num=2]
Epoch 47:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.46, v_num=2]
Epoch 47:  56%|█████▋    | 18/32 [00:09<00:07,  1.91it/s, loss=7.46, v_num=2]
Epoch 47:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.46, v_num=2]
Epoch 47:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.46, v_num=2]
Epoch 47:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.46, v_num=2]
Epoch 47:  62%|██████▎   | 20/32 [00:09<00:05,  2.10it/s, loss=7.46, v_num=2]
Epoch 47:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.46, v_num=2]
Epoch 47:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.46, v_num=2]
Epoch 47:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.46, v_num=2]
Epoch 47:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.46, v_num=2]
Epoch 47:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.46, v_num=2]
Epoch 47:  72%|███████▏  | 23/32 [00:09<00:03,  2.38it/s, loss=7.46, v_num=2]
Epoch 47:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.46, v_num=2]
Epoch 47:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.46, v_num=2]
Epoch 47:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.46, v_num=2]
Epoch 47:  78%|███████▊  | 25/32 [00:12<00:03,  2.04it/s, loss=7.46, v_num=2]
Epoch 47:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.46, v_num=2]
Epoch 47:  81%|████████▏ | 26/32 [00:12<00:02,  2.11it/s, loss=7.46, v_num=2]
Epoch 47:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.46, v_num=2]
Epoch 47:  84%|████████▍ | 27/32 [00:12<00:02,  2.19it/s, loss=7.46, v_num=2]
Epoch 47:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.46, v_num=2]
Epoch 47:  88%|████████▊ | 28/32 [00:12<00:01,  2.26it/s, loss=7.46, v_num=2]
Epoch 47:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.46, v_num=2]
Epoch 47:  91%|█████████ | 29/32 [00:12<00:01,  2.33it/s, loss=7.46, v_num=2]
Epoch 47:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.46, v_num=2]
Epoch 47:  94%|█████████▍| 30/32 [00:12<00:00,  2.40it/s, loss=7.46, v_num=2]
Epoch 47:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.46, v_num=2]
Epoch 47:  97%|█████████▋| 31/32 [00:12<00:00,  2.47it/s, loss=7.46, v_num=2]
Epoch 47: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.46, v_num=2]
Epoch 47: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.46, v_num=2]
Epoch 47: 100%|██████████| 32/32 [00:12<00:00,  2.55it/s, loss=7.46, v_num=2]
Epoch 47:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.46, v_num=2]
Epoch 48:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.46, v_num=2]
Epoch 48:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.46, v_num=2]
Epoch 48:   3%|▎         | 1/32 [00:03<01:54,  3.71s/it, loss=7.46, v_num=2]
Epoch 48:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.46, v_num=2]
Epoch 48:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.46, v_num=2]
Epoch 48:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.46, v_num=2]
Epoch 48:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.46, v_num=2]
Epoch 48:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.46, v_num=2]
Epoch 48:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.46, v_num=2]
Epoch 48:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.46, v_num=2]
Epoch 48:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.46, v_num=2]
Epoch 48:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.46, v_num=2]
Epoch 48:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.46, v_num=2]
Epoch 48:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.46, v_num=2]
Epoch 48:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.46, v_num=2]
Epoch 48:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.46, v_num=2]
Epoch 48:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.46, v_num=2]
Epoch 48:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.46, v_num=2]
Epoch 48:  28%|██▊       | 9/32 [00:06<00:16,  1.38it/s, loss=7.46, v_num=2]
Epoch 48:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.46, v_num=2]
Epoch 48:  31%|███▏      | 10/32 [00:06<00:14,  1.52it/s, loss=7.46, v_num=2]
Epoch 48:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.46, v_num=2]
Epoch 48:  34%|███▍      | 11/32 [00:06<00:12,  1.66it/s, loss=7.46, v_num=2]
Epoch 48:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.46, v_num=2]
Epoch 48:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.46, v_num=2]
Epoch 48:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.46, v_num=2]
Epoch 48:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.46, v_num=2]
Epoch 48:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.46, v_num=2]
Epoch 48:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.47, v_num=2]
Epoch 48:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.47, v_num=2]
Epoch 48:  47%|████▋     | 15/32 [00:06<00:07,  2.20it/s, loss=7.46, v_num=2]
Epoch 48:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.46, v_num=2]
Epoch 48:  50%|█████     | 16/32 [00:06<00:06,  2.33it/s, loss=7.46, v_num=2]
Epoch 48:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.46, v_num=2]
Epoch 48:  53%|█████▎    | 17/32 [00:09<00:08,  1.81it/s, loss=7.47, v_num=2]
Epoch 48:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.47, v_num=2]
Epoch 48:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.46, v_num=2]
Epoch 48:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.46, v_num=2]
Epoch 48:  59%|█████▉    | 19/32 [00:09<00:06,  2.00it/s, loss=7.46, v_num=2]
Epoch 48:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.46, v_num=2]
Epoch 48:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.46, v_num=2]
Epoch 48:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.46, v_num=2]
Epoch 48:  66%|██████▌   | 21/32 [00:09<00:05,  2.19it/s, loss=7.46, v_num=2]
Epoch 48:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.46, v_num=2]
Epoch 48:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.46, v_num=2]
Epoch 48:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.46, v_num=2]
Epoch 48:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.47, v_num=2]
Epoch 48:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.47, v_num=2]
Epoch 48:  75%|███████▌  | 24/32 [00:09<00:03,  2.47it/s, loss=7.47, v_num=2]
Epoch 48:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.47, v_num=2]
Epoch 48:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.47, v_num=2]
Epoch 48:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.47, v_num=2]
Epoch 48:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.47, v_num=2]
Epoch 48:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.47, v_num=2]
Epoch 48:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.47, v_num=2]
Epoch 48:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.47, v_num=2]
Epoch 48:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.47, v_num=2]
Epoch 48:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.47, v_num=2]
Epoch 48:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.48, v_num=2]
Epoch 48:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.48, v_num=2]
Epoch 48:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.48, v_num=2]
Epoch 48:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.48, v_num=2]
Epoch 48:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.49, v_num=2]
Epoch 48: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 48: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 48: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 48:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.49, v_num=2]
Epoch 49:   0%|          | 0/32 [00:00<?, ?it/s, loss=7.49, v_num=2]
Epoch 49:   3%|▎         | 1/32 [00:03<01:55,  3.71s/it, loss=7.49, v_num=2]
Epoch 49:   3%|▎         | 1/32 [00:03<01:55,  3.71s/it, loss=7.49, v_num=2]
Epoch 49:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.49, v_num=2]
Epoch 49:   6%|▋         | 2/32 [00:03<00:56,  1.88s/it, loss=7.49, v_num=2]
Epoch 49:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.49, v_num=2]
Epoch 49:   9%|▉         | 3/32 [00:03<00:36,  1.27s/it, loss=7.49, v_num=2]
Epoch 49:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.49, v_num=2]
Epoch 49:  12%|█▎        | 4/32 [00:03<00:26,  1.04it/s, loss=7.49, v_num=2]
Epoch 49:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.49, v_num=2]
Epoch 49:  16%|█▌        | 5/32 [00:03<00:21,  1.28it/s, loss=7.49, v_num=2]
Epoch 49:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.49, v_num=2]
Epoch 49:  19%|█▉        | 6/32 [00:03<00:17,  1.52it/s, loss=7.5, v_num=2]
Epoch 49:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.5, v_num=2]
Epoch 49:  22%|██▏       | 7/32 [00:03<00:14,  1.76it/s, loss=7.51, v_num=2]
Epoch 49:  25%|██▌       | 8/32 [00:04<00:12,  1.99it/s, loss=7.51, v_num=2]
Epoch 49:  25%|██▌       | 8/32 [00:04<00:12,  1.98it/s, loss=7.5, v_num=2]
Epoch 49:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.5, v_num=2]
Epoch 49:  28%|██▊       | 9/32 [00:06<00:16,  1.37it/s, loss=7.5, v_num=2]
Epoch 49:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.5, v_num=2]
Epoch 49:  31%|███▏      | 10/32 [00:06<00:14,  1.51it/s, loss=7.5, v_num=2]
Epoch 49:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.5, v_num=2]
Epoch 49:  34%|███▍      | 11/32 [00:06<00:12,  1.65it/s, loss=7.49, v_num=2]
Epoch 49:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.49, v_num=2]
Epoch 49:  38%|███▊      | 12/32 [00:06<00:11,  1.79it/s, loss=7.5, v_num=2]
Epoch 49:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.5, v_num=2]
Epoch 49:  41%|████      | 13/32 [00:06<00:09,  1.93it/s, loss=7.5, v_num=2]
Epoch 49:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.5, v_num=2]
Epoch 49:  44%|████▍     | 14/32 [00:06<00:08,  2.06it/s, loss=7.5, v_num=2]
Epoch 49:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.5, v_num=2]
Epoch 49:  47%|████▋     | 15/32 [00:06<00:07,  2.19it/s, loss=7.5, v_num=2]
Epoch 49:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.5, v_num=2]
Epoch 49:  50%|█████     | 16/32 [00:06<00:06,  2.32it/s, loss=7.5, v_num=2]
Epoch 49:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.5, v_num=2]
Epoch 49:  53%|█████▎    | 17/32 [00:09<00:08,  1.80it/s, loss=7.5, v_num=2]
Epoch 49:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.5, v_num=2]
Epoch 49:  56%|█████▋    | 18/32 [00:09<00:07,  1.90it/s, loss=7.49, v_num=2]
Epoch 49:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.49, v_num=2]
Epoch 49:  59%|█████▉    | 19/32 [00:09<00:06,  1.99it/s, loss=7.49, v_num=2]
Epoch 49:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.49, v_num=2]
Epoch 49:  62%|██████▎   | 20/32 [00:09<00:05,  2.09it/s, loss=7.49, v_num=2]
Epoch 49:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.49, v_num=2]
Epoch 49:  66%|██████▌   | 21/32 [00:09<00:05,  2.18it/s, loss=7.49, v_num=2]
Epoch 49:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.49, v_num=2]
Epoch 49:  69%|██████▉   | 22/32 [00:09<00:04,  2.28it/s, loss=7.49, v_num=2]
Epoch 49:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.49, v_num=2]
Epoch 49:  72%|███████▏  | 23/32 [00:09<00:03,  2.37it/s, loss=7.49, v_num=2]
Epoch 49:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.49, v_num=2]
Epoch 49:  75%|███████▌  | 24/32 [00:09<00:03,  2.46it/s, loss=7.49, v_num=2]
Epoch 49:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.49, v_num=2]
Epoch 49:  78%|███████▊  | 25/32 [00:12<00:03,  2.03it/s, loss=7.48, v_num=2]
Epoch 49:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.48, v_num=2]
Epoch 49:  81%|████████▏ | 26/32 [00:12<00:02,  2.10it/s, loss=7.48, v_num=2]
Epoch 49:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.48, v_num=2]
Epoch 49:  84%|████████▍ | 27/32 [00:12<00:02,  2.18it/s, loss=7.48, v_num=2]
Epoch 49:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.48, v_num=2]
Epoch 49:  88%|████████▊ | 28/32 [00:12<00:01,  2.25it/s, loss=7.48, v_num=2]
Epoch 49:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.48, v_num=2]
Epoch 49:  91%|█████████ | 29/32 [00:12<00:01,  2.32it/s, loss=7.48, v_num=2]
Epoch 49:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.48, v_num=2]
Epoch 49:  94%|█████████▍| 30/32 [00:12<00:00,  2.39it/s, loss=7.48, v_num=2]
Epoch 49:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.48, v_num=2]
Epoch 49:  97%|█████████▋| 31/32 [00:12<00:00,  2.46it/s, loss=7.49, v_num=2]
Epoch 49: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 49: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 49: 100%|██████████| 32/32 [00:12<00:00,  2.53it/s, loss=7.49, v_num=2]
Epoch 49: 100%|██████████| 32/32 [00:12<00:00,  2.47it/s, loss=7.49, v_num=2]

Evaluate the results

It’s always a good idea to evaluate how good the learned representations really are. How to do this depends on the available data and metdata. Luckily, in our case, we have annotations of critical findings on the X-ray images. We can use this information to see, whether images with similar annotations are grouped together.

We start by getting a vector representation of each image in the dataset. For this, we create a new dataloader. This time, we can pass the transform directly to the dataset.

# test transforms differ from training transforms as they do not introduce
# additional noise
test_transforms = torchvision.transforms.Compose(
    [
        HistogramNormalize(),
        torchvision.transforms.Grayscale(num_output_channels=3),
        torchvision.transforms.Resize(input_size),
        torchvision.transforms.ToTensor(),
    ]
)

# Create the dataset and overwrite the image loader as before.
dataset_test = LightlyDataset(input_dir=path_to_data, transform=test_transforms)
dataset_test.dataset.loader = tiff_loader

# Create the test dataloader.
dataloader_test = torch.utils.data.DataLoader(
    dataset_test, batch_size=1, shuffle=False, drop_last=False, num_workers=num_workers
)


# Next, we add a small helper function to generate embeddings of our images
def generate_embeddings(model, dataloader):
    """Generates representations for all images in the dataloader"""

    embeddings = []
    filenames = []
    with torch.no_grad():
        for img, label, fnames in dataloader:
            img = img.to(model.device)
            emb = model.backbone(img).flatten(start_dim=1)
            embeddings.append(emb)
            filenames.extend(fnames)

    embeddings = torch.cat(embeddings, 0)
    embeddings = normalize(embeddings)
    return embeddings, filenames


# Generate the embeddings (remember to put the model in eval mode).
model.eval()
embeddings, fnames = generate_embeddings(model, dataloader_test)

Now, we can use the embeddings to search for nearest neighbors.

We choose three example images. For each example image, we find 50 nearest neighbors. Then, we plot the critical findings in the example image (dark blue) and the distribution of the critical findings in the nearest neighbor images (light blue) as bar plots.

# Transform the original bounding box annotations to multiclass labels.
fnames = [fname.split(".")[0] for fname in fnames]

df = pandas.read_csv("/datasets/vinbigdata/train.csv")
classes = list(np.unique(df.class_name))
filenames = list(np.unique(df.image_id))

# Iterate over all bounding boxes and add a one-hot label if an image contains
# a bounding box of a given class, after that, the array "multilabels" will
# contain a row for every image in the input dataset and each row of the
# array contains a one-hot vector of critical findings for this image.
multilabels = np.zeros((len(dataset_test.get_filenames()), len(classes)))
for filename, label in zip(df.image_id, df.class_name):
    try:
        i = fnames.index(filename.split(".")[0])
        j = classes.index(label)
        multilabels[i, j] = 1.0
    except Exception:
        pass


def plot_knn_multilabels(
    embeddings, multilabels, samples_idx, filenames, n_neighbors=50
):
    """Plots multiple rows of random images with their nearest neighbors"""
    # Let0s look at the nearest neighbors for some samples using the sklearn library.
    nbrs = NearestNeighbors(n_neighbors=n_neighbors).fit(embeddings)
    _, indices = nbrs.kneighbors(embeddings)

    # Position the bars.
    bar_width = 0.4
    r1 = np.arange(multilabels.shape[1])
    r2 = r1 + bar_width

    # Loop through our randomly picked samples.
    for idx in samples_idx:
        fig = plt.figure()

        bars1 = multilabels[idx]
        bars2 = np.mean(multilabels[indices[idx]], axis=0)

        plt.title(filenames[idx])
        plt.bar(r1, bars1, color="steelblue", edgecolor="black", width=bar_width)
        plt.bar(r2, bars2, color="lightsteelblue", edgecolor="black", width=bar_width)
        plt.xticks(0.5 * (r1 + r2), classes, rotation=90)
        plt.tight_layout()


# Plot the distribution of the multilabels of the k nearest neighbors of
# the three example images at indices 4111, 3340, and 1796.
k = 20
plot_knn_multilabels(embeddings, multilabels, [4111, 3340, 1796], fnames, n_neighbors=k)

• 
• 
•