Downstream Analysis with flashscenic#
This tutorial demonstrates how to use flashscenic for gene regulatory network (GRN) analysis and downstream biological interpretation. We use the Immune_ALL_human dataset from the scIB benchmark (Luecken et al. 2022): 33,506 cells across 16 immune and hematopoietic cell types from 10 batches.
What you’ll learn:
Running the flashscenic pipeline (one function call)
Visualizing regulon activity with UMAP – and why it naturally handles batch effects
Identifying cell-type-specific regulons with Regulon Specificity Scores (RSS)
Validating results against known immune biology (PAX5 in B cells, SPI1 in monocytes, etc.)
Differential regulon activity analysis
Exploring TF-target regulatory networks
Requirements: flashscenic[tutorial] (includes scanpy, matplotlib, seaborn, networkx)
GPU: A CUDA-capable GPU is recommended. The pipeline takes ~2 minutes on a modern GPU; downstream analyses are CPU-only.
1. Setup and Data Loading#
import numpy as np
import pandas as pd
import scanpy as sc
import matplotlib.pyplot as plt
import seaborn as sns
import anndata
import warnings
warnings.filterwarnings('ignore')
import flashscenic as fs
sc.settings.set_figure_params(dpi=100, frameon=False, figsize=(4, 4))
plt.rcParams['figure.dpi'] = 100
Load the dataset#
We use the Immune_ALL_human dataset from the scIB benchmark, containing PBMC and bone marrow cells from 5 studies (10 batches) and 16 annotated cell types spanning lymphoid, myeloid, erythroid, and progenitor populations.
Download from: https://figshare.com/ndownloader/files/25717328
DATA_PATH = '../experiments/data/Immune_ALL_human.h5ad'
adata = sc.read_h5ad(DATA_PATH)
print(f"Dataset: {adata.shape[0]:,} cells x {adata.shape[1]:,} genes")
print(f"Cell types: {adata.obs['final_annotation'].nunique()}")
print(f"Batches: {adata.obs['batch'].nunique()}")
print()
print(adata.obs['final_annotation'].value_counts())
Dataset: 33,506 cells x 12,303 genes
Cell types: 16
Batches: 10
final_annotation
CD4+ T cells 11011
CD14+ Monocytes 6483
CD20+ B cells 2873
NKT cells 2745
NK cells 2294
CD8+ T cells 2183
Erythrocytes 1502
Monocyte-derived dendritic cells 1012
CD16+ Monocytes 997
HSPCs 473
Erythroid progenitors 463
Plasmacytoid dendritic cells 436
Monocyte progenitors 428
Megakaryocyte progenitors 270
CD10+ B cells 207
Plasma cells 129
Name: count, dtype: int64
Preprocessing#
flashscenic needs a dense expression matrix with no zero-variance genes. We also filter to genes expressed in at least 10 cells.
from scipy.sparse import issparse
# Filter genes: keep genes expressed in > 10 cells
if issparse(adata.X):
cells_per_gene = np.array((adata.X > 0).sum(axis=0)).flatten()
else:
cells_per_gene = (adata.X > 0).sum(axis=0)
gene_mask = cells_per_gene > 10
adata = adata[:, gene_mask].copy()
print(f"After gene filter (>10 cells): {adata.shape}")
# Densify sparse matrix
if issparse(adata.X):
adata.X = adata.X.toarray()
# Remove zero-variance genes
gene_var = np.var(adata.X, axis=0)
nonzero_mask = gene_var > 0
n_removed = (~nonzero_mask).sum()
if n_removed > 0:
print(f"Removing {n_removed} zero-variance genes")
adata = adata[:, nonzero_mask].copy()
print(f"Final shape: {adata.shape}")
gene_names = adata.var_names.tolist()
exp_matrix = adata.X.astype(np.float32)
After gene filter (>10 cells): (33506, 12282)
Final shape: (33506, 12282)
Compute PCA baseline#
We compute a PCA-based UMAP first so we can compare it against the AUCell-based UMAP later.
sc.tl.pca(adata, n_comps=30)
sc.pp.neighbors(adata, n_neighbors=15, use_rep='X_pca')
sc.tl.umap(adata)
adata.obsm['X_umap_pca'] = adata.obsm['X_umap'].copy()
2. Running flashscenic#
The entire SCENIC pipeline – GRN inference (RegDiffusion), module filtering, cisTarget pruning, and AUCell scoring – runs in a single function call.
result = fs.run_flashscenic(
exp_matrix, gene_names,
species='human',
device='cuda',
seed=42,
verbose=True,
)
# Unpack results
auc_scores = result['auc_scores'] # (n_cells, n_regulons)
regulon_names = result['regulon_names'] # list of str
regulons = result['regulons'] # list of dicts
regulon_adj = result['regulon_adj'] # (n_regulons, n_genes) binary
print(f"\nFound {len(regulon_names)} regulons")
print(f"AUCell scores shape: {auc_scores.shape}")
print(f"Example regulons: {regulon_names[:10]}")
[flashscenic] Step 0/5: Preparing resources...
Downloading scenic/human/v10 resources to flashscenic_data/
[Human TF list (hg38)] Already cached: allTFs_hg38.txt
[Human 500bp/100bp ranking database (v10)] Already cached: hg38_500bp_up_100bp_down_full_tx_v10_clust.genes_vs_motifs.rankings.feather
[Human 10kbp ranking database (v10)] Already cached: hg38_10kbp_up_10kbp_down_full_tx_v10_clust.genes_vs_motifs.rankings.feather
[Human motif annotations (v10)] Already cached: motifs-v10nr_clust-nr.hgnc-m0.001-o0.0.tbl
Download complete.
[flashscenic] Step 1/5: Running RegDiffusion GRN inference (33506 cells, 12282 genes, 1000 steps)...
0%| | 0/1000 [00:00<?, ?it/s]
Training loss: 1.576, Change on Adj: -0.020: 0%| | 0/1000 [00:02<?, ?it/s]
Training loss: 1.576, Change on Adj: -0.020: 0%| | 1/1000 [00:02<34:39, 2.08s/it]
Training loss: 1.298, Change on Adj: -0.018: 0%| | 1/1000 [00:03<34:39, 2.08s/it]
Training loss: 1.298, Change on Adj: -0.018: 0%| | 2/1000 [00:03<26:03, 1.57s/it]
Training loss: 1.140, Change on Adj: -0.016: 0%| | 2/1000 [00:03<26:03, 1.57s/it]
Training loss: 1.070, Change on Adj: -0.015: 0%| | 2/1000 [00:03<26:03, 1.57s/it]
Training loss: 1.070, Change on Adj: -0.015: 0%| | 4/1000 [00:03<10:35, 1.57it/s]
Training loss: 1.039, Change on Adj: -0.015: 0%| | 4/1000 [00:03<10:35, 1.57it/s]
Training loss: 1.042, Change on Adj: -0.014: 0%| | 4/1000 [00:03<10:35, 1.57it/s]
Training loss: 1.042, Change on Adj: -0.014: 1%| | 6/1000 [00:03<06:12, 2.67it/s]
Training loss: 1.061, Change on Adj: -0.014: 1%| | 6/1000 [00:03<06:12, 2.67it/s]
Training loss: 1.075, Change on Adj: -0.014: 1%| | 6/1000 [00:03<06:12, 2.67it/s]
Training loss: 1.075, Change on Adj: -0.014: 1%| | 8/1000 [00:03<04:12, 3.93it/s]
Training loss: 1.065, Change on Adj: -0.014: 1%| | 8/1000 [00:03<04:12, 3.93it/s]
Training loss: 1.057, Change on Adj: -0.015: 1%| | 8/1000 [00:03<04:12, 3.93it/s]
Training loss: 1.057, Change on Adj: -0.015: 1%| | 10/1000 [00:03<03:07, 5.27it/s]
Training loss: 1.055, Change on Adj: -0.015: 1%| | 10/1000 [00:04<03:07, 5.27it/s]
Training loss: 1.029, Change on Adj: -0.015: 1%| | 10/1000 [00:04<03:07, 5.27it/s]
Training loss: 1.029, Change on Adj: -0.015: 1%| | 12/1000 [00:04<02:30, 6.56it/s]
Training loss: 1.004, Change on Adj: -0.014: 1%| | 12/1000 [00:04<02:30, 6.56it/s]
Training loss: 0.984, Change on Adj: -0.014: 1%| | 12/1000 [00:04<02:30, 6.56it/s]
Training loss: 0.984, Change on Adj: -0.014: 1%|▏ | 14/1000 [00:04<02:08, 7.67it/s]
Training loss: 0.966, Change on Adj: -0.014: 1%|▏ | 14/1000 [00:04<02:08, 7.67it/s]
Training loss: 0.954, Change on Adj: -0.014: 1%|▏ | 14/1000 [00:04<02:08, 7.67it/s]
Training loss: 0.954, Change on Adj: -0.014: 2%|▏ | 16/1000 [00:04<01:54, 8.62it/s]
Training loss: 0.952, Change on Adj: -0.013: 2%|▏ | 16/1000 [00:04<01:54, 8.62it/s]
Training loss: 0.940, Change on Adj: -0.013: 2%|▏ | 16/1000 [00:04<01:54, 8.62it/s]
Training loss: 0.940, Change on Adj: -0.013: 2%|▏ | 18/1000 [00:04<01:44, 9.41it/s]
Training loss: 0.941, Change on Adj: -0.013: 2%|▏ | 18/1000 [00:04<01:44, 9.41it/s]
Training loss: 0.945, Change on Adj: -0.013: 2%|▏ | 18/1000 [00:04<01:44, 9.41it/s]
Training loss: 0.945, Change on Adj: -0.013: 2%|▏ | 20/1000 [00:04<01:37, 10.03it/s]
Training loss: 0.942, Change on Adj: -0.012: 2%|▏ | 20/1000 [00:04<01:37, 10.03it/s]
Training loss: 0.931, Change on Adj: -0.012: 2%|▏ | 20/1000 [00:04<01:37, 10.03it/s]
Training loss: 0.931, Change on Adj: -0.012: 2%|▏ | 22/1000 [00:04<01:33, 10.51it/s]
Training loss: 0.926, Change on Adj: -0.012: 2%|▏ | 22/1000 [00:05<01:33, 10.51it/s]
Training loss: 0.920, Change on Adj: -0.012: 2%|▏ | 22/1000 [00:05<01:33, 10.51it/s]
Training loss: 0.920, Change on Adj: -0.012: 2%|▏ | 24/1000 [00:05<01:29, 10.86it/s]
Training loss: 0.912, Change on Adj: -0.012: 2%|▏ | 24/1000 [00:05<01:29, 10.86it/s]
Training loss: 0.897, Change on Adj: -0.012: 2%|▏ | 24/1000 [00:05<01:29, 10.86it/s]
Training loss: 0.897, Change on Adj: -0.012: 3%|▎ | 26/1000 [00:05<01:27, 11.12it/s]
Training loss: 0.893, Change on Adj: -0.012: 3%|▎ | 26/1000 [00:05<01:27, 11.12it/s]
Training loss: 0.887, Change on Adj: -0.012: 3%|▎ | 26/1000 [00:05<01:27, 11.12it/s]
Training loss: 0.887, Change on Adj: -0.012: 3%|▎ | 28/1000 [00:05<01:26, 11.30it/s]
Training loss: 0.880, Change on Adj: -0.012: 3%|▎ | 28/1000 [00:05<01:26, 11.30it/s]
Training loss: 0.868, Change on Adj: -0.012: 3%|▎ | 28/1000 [00:05<01:26, 11.30it/s]
Training loss: 0.868, Change on Adj: -0.012: 3%|▎ | 30/1000 [00:05<01:24, 11.43it/s]
Training loss: 0.856, Change on Adj: -0.012: 3%|▎ | 30/1000 [00:05<01:24, 11.43it/s]
Training loss: 0.858, Change on Adj: -0.012: 3%|▎ | 30/1000 [00:05<01:24, 11.43it/s]
Training loss: 0.858, Change on Adj: -0.012: 3%|▎ | 32/1000 [00:05<01:23, 11.53it/s]
Training loss: 0.850, Change on Adj: -0.012: 3%|▎ | 32/1000 [00:05<01:23, 11.53it/s]
Training loss: 0.853, Change on Adj: -0.012: 3%|▎ | 32/1000 [00:05<01:23, 11.53it/s]
Training loss: 0.853, Change on Adj: -0.012: 3%|▎ | 34/1000 [00:05<01:23, 11.59it/s]
Training loss: 0.850, Change on Adj: -0.012: 3%|▎ | 34/1000 [00:06<01:23, 11.59it/s]
Training loss: 0.830, Change on Adj: -0.012: 3%|▎ | 34/1000 [00:06<01:23, 11.59it/s]
Training loss: 0.830, Change on Adj: -0.012: 4%|▎ | 36/1000 [00:06<01:22, 11.64it/s]
Training loss: 0.821, Change on Adj: -0.012: 4%|▎ | 36/1000 [00:06<01:22, 11.64it/s]
Training loss: 0.814, Change on Adj: -0.012: 4%|▎ | 36/1000 [00:06<01:22, 11.64it/s]
Training loss: 0.814, Change on Adj: -0.012: 4%|▍ | 38/1000 [00:06<01:22, 11.67it/s]
Training loss: 0.815, Change on Adj: -0.012: 4%|▍ | 38/1000 [00:06<01:22, 11.67it/s]
Training loss: 0.806, Change on Adj: -0.012: 4%|▍ | 38/1000 [00:06<01:22, 11.67it/s]
Training loss: 0.806, Change on Adj: -0.012: 4%|▍ | 40/1000 [00:06<01:22, 11.69it/s]
Training loss: 0.789, Change on Adj: -0.012: 4%|▍ | 40/1000 [00:06<01:22, 11.69it/s]
Training loss: 0.798, Change on Adj: -0.012: 4%|▍ | 40/1000 [00:06<01:22, 11.69it/s]
Training loss: 0.798, Change on Adj: -0.012: 4%|▍ | 42/1000 [00:06<01:21, 11.71it/s]
Training loss: 0.787, Change on Adj: -0.012: 4%|▍ | 42/1000 [00:06<01:21, 11.71it/s]
Training loss: 0.772, Change on Adj: -0.012: 4%|▍ | 42/1000 [00:06<01:21, 11.71it/s]
Training loss: 0.772, Change on Adj: -0.012: 4%|▍ | 44/1000 [00:06<01:21, 11.74it/s]
Training loss: 0.780, Change on Adj: -0.012: 4%|▍ | 44/1000 [00:06<01:21, 11.74it/s]
Training loss: 0.775, Change on Adj: -0.012: 4%|▍ | 44/1000 [00:06<01:21, 11.74it/s]
Training loss: 0.775, Change on Adj: -0.012: 5%|▍ | 46/1000 [00:06<01:21, 11.75it/s]
Training loss: 0.763, Change on Adj: -0.012: 5%|▍ | 46/1000 [00:07<01:21, 11.75it/s]
Training loss: 0.754, Change on Adj: -0.012: 5%|▍ | 46/1000 [00:07<01:21, 11.75it/s]
Training loss: 0.754, Change on Adj: -0.012: 5%|▍ | 48/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.745, Change on Adj: -0.012: 5%|▍ | 48/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.744, Change on Adj: -0.012: 5%|▍ | 48/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.744, Change on Adj: -0.012: 5%|▌ | 50/1000 [00:07<01:20, 11.77it/s]
Training loss: 0.732, Change on Adj: -0.012: 5%|▌ | 50/1000 [00:07<01:20, 11.77it/s]
Training loss: 0.721, Change on Adj: -0.012: 5%|▌ | 50/1000 [00:07<01:20, 11.77it/s]
Training loss: 0.721, Change on Adj: -0.012: 5%|▌ | 52/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.712, Change on Adj: -0.012: 5%|▌ | 52/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.716, Change on Adj: -0.012: 5%|▌ | 52/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.716, Change on Adj: -0.012: 5%|▌ | 54/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.719, Change on Adj: -0.012: 5%|▌ | 54/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.717, Change on Adj: -0.012: 5%|▌ | 54/1000 [00:07<01:20, 11.76it/s]
Training loss: 0.717, Change on Adj: -0.012: 6%|▌ | 56/1000 [00:07<01:20, 11.78it/s]
Training loss: 0.699, Change on Adj: -0.012: 6%|▌ | 56/1000 [00:07<01:20, 11.78it/s]
Training loss: 0.678, Change on Adj: -0.012: 6%|▌ | 56/1000 [00:08<01:20, 11.78it/s]
Training loss: 0.678, Change on Adj: -0.012: 6%|▌ | 58/1000 [00:08<01:19, 11.78it/s]
Training loss: 0.666, Change on Adj: -0.012: 6%|▌ | 58/1000 [00:08<01:19, 11.78it/s]
Training loss: 0.674, Change on Adj: -0.012: 6%|▌ | 58/1000 [00:08<01:19, 11.78it/s]
Training loss: 0.674, Change on Adj: -0.012: 6%|▌ | 60/1000 [00:08<01:19, 11.79it/s]
Training loss: 0.666, Change on Adj: -0.012: 6%|▌ | 60/1000 [00:08<01:19, 11.79it/s]
Training loss: 0.670, Change on Adj: -0.012: 6%|▌ | 60/1000 [00:08<01:19, 11.79it/s]
Training loss: 0.670, Change on Adj: -0.012: 6%|▌ | 62/1000 [00:08<01:19, 11.80it/s]
Training loss: 0.685, Change on Adj: -0.012: 6%|▌ | 62/1000 [00:08<01:19, 11.80it/s]
Training loss: 0.652, Change on Adj: -0.012: 6%|▌ | 62/1000 [00:08<01:19, 11.80it/s]
Training loss: 0.652, Change on Adj: -0.012: 6%|▋ | 64/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.636, Change on Adj: -0.012: 6%|▋ | 64/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.636, Change on Adj: -0.012: 6%|▋ | 64/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.636, Change on Adj: -0.012: 7%|▋ | 66/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.654, Change on Adj: -0.012: 7%|▋ | 66/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.611, Change on Adj: -0.012: 7%|▋ | 66/1000 [00:08<01:19, 11.81it/s]
Training loss: 0.611, Change on Adj: -0.012: 7%|▋ | 68/1000 [00:08<01:18, 11.81it/s]
Training loss: 0.608, Change on Adj: -0.012: 7%|▋ | 68/1000 [00:08<01:18, 11.81it/s]
Training loss: 0.623, Change on Adj: -0.012: 7%|▋ | 68/1000 [00:09<01:18, 11.81it/s]
Training loss: 0.623, Change on Adj: -0.012: 7%|▋ | 70/1000 [00:09<01:18, 11.81it/s]
Training loss: 0.585, Change on Adj: -0.012: 7%|▋ | 70/1000 [00:09<01:18, 11.81it/s]
Training loss: 0.600, Change on Adj: -0.012: 7%|▋ | 70/1000 [00:09<01:18, 11.81it/s]
Training loss: 0.600, Change on Adj: -0.012: 7%|▋ | 72/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.585, Change on Adj: -0.012: 7%|▋ | 72/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.586, Change on Adj: -0.012: 7%|▋ | 72/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.586, Change on Adj: -0.012: 7%|▋ | 74/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.568, Change on Adj: -0.012: 7%|▋ | 74/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.568, Change on Adj: -0.012: 7%|▋ | 74/1000 [00:09<01:18, 11.82it/s]
Training loss: 0.568, Change on Adj: -0.012: 8%|▊ | 76/1000 [00:09<01:18, 11.83it/s]
Training loss: 0.546, Change on Adj: -0.012: 8%|▊ | 76/1000 [00:09<01:18, 11.83it/s]
Training loss: 0.569, Change on Adj: -0.012: 8%|▊ | 76/1000 [00:09<01:18, 11.83it/s]
Training loss: 0.569, Change on Adj: -0.012: 8%|▊ | 78/1000 [00:09<01:17, 11.82it/s]
Training loss: 0.563, Change on Adj: -0.012: 8%|▊ | 78/1000 [00:09<01:17, 11.82it/s]
Training loss: 0.536, Change on Adj: -0.012: 8%|▊ | 78/1000 [00:09<01:17, 11.82it/s]
Training loss: 0.536, Change on Adj: -0.012: 8%|▊ | 80/1000 [00:09<01:17, 11.83it/s]
Training loss: 0.538, Change on Adj: -0.012: 8%|▊ | 80/1000 [00:09<01:17, 11.83it/s]
Training loss: 0.527, Change on Adj: -0.012: 8%|▊ | 80/1000 [00:10<01:17, 11.83it/s]
Training loss: 0.527, Change on Adj: -0.012: 8%|▊ | 82/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.536, Change on Adj: -0.012: 8%|▊ | 82/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.527, Change on Adj: -0.012: 8%|▊ | 82/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.527, Change on Adj: -0.012: 8%|▊ | 84/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.501, Change on Adj: -0.012: 8%|▊ | 84/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.506, Change on Adj: -0.012: 8%|▊ | 84/1000 [00:10<01:17, 11.82it/s]
Training loss: 0.506, Change on Adj: -0.012: 9%|▊ | 86/1000 [00:10<01:17, 11.83it/s]
Training loss: 0.493, Change on Adj: -0.012: 9%|▊ | 86/1000 [00:10<01:17, 11.83it/s]
Training loss: 0.513, Change on Adj: -0.012: 9%|▊ | 86/1000 [00:10<01:17, 11.83it/s]
Training loss: 0.513, Change on Adj: -0.012: 9%|▉ | 88/1000 [00:10<01:17, 11.80it/s]
Training loss: 0.522, Change on Adj: -0.012: 9%|▉ | 88/1000 [00:10<01:17, 11.80it/s]
Training loss: 0.496, Change on Adj: -0.012: 9%|▉ | 88/1000 [00:10<01:17, 11.80it/s]
Training loss: 0.496, Change on Adj: -0.012: 9%|▉ | 90/1000 [00:10<01:17, 11.74it/s]
Training loss: 0.507, Change on Adj: -0.012: 9%|▉ | 90/1000 [00:10<01:17, 11.74it/s]
Training loss: 0.467, Change on Adj: -0.012: 9%|▉ | 90/1000 [00:10<01:17, 11.74it/s]
Training loss: 0.467, Change on Adj: -0.012: 9%|▉ | 92/1000 [00:10<01:17, 11.76it/s]
Training loss: 0.446, Change on Adj: -0.012: 9%|▉ | 92/1000 [00:10<01:17, 11.76it/s]
Training loss: 0.437, Change on Adj: -0.012: 9%|▉ | 92/1000 [00:11<01:17, 11.76it/s]
Training loss: 0.437, Change on Adj: -0.012: 9%|▉ | 94/1000 [00:11<01:16, 11.78it/s]
Training loss: 0.442, Change on Adj: -0.012: 9%|▉ | 94/1000 [00:11<01:16, 11.78it/s]
Training loss: 0.478, Change on Adj: -0.012: 9%|▉ | 94/1000 [00:11<01:16, 11.78it/s]
Training loss: 0.478, Change on Adj: -0.012: 10%|▉ | 96/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.440, Change on Adj: -0.012: 10%|▉ | 96/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.451, Change on Adj: -0.013: 10%|▉ | 96/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.451, Change on Adj: -0.013: 10%|▉ | 98/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.450, Change on Adj: -0.012: 10%|▉ | 98/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.416, Change on Adj: -0.013: 10%|▉ | 98/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.416, Change on Adj: -0.013: 10%|█ | 100/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.412, Change on Adj: -0.013: 10%|█ | 100/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.400, Change on Adj: -0.012: 10%|█ | 100/1000 [00:11<01:16, 11.75it/s]
Training loss: 0.400, Change on Adj: -0.012: 10%|█ | 102/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.411, Change on Adj: -0.012: 10%|█ | 102/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.398, Change on Adj: -0.013: 10%|█ | 102/1000 [00:11<01:16, 11.76it/s]
Training loss: 0.398, Change on Adj: -0.013: 10%|█ | 104/1000 [00:11<01:16, 11.78it/s]
Training loss: 0.421, Change on Adj: -0.013: 10%|█ | 104/1000 [00:11<01:16, 11.78it/s]
Training loss: 0.372, Change on Adj: -0.013: 10%|█ | 104/1000 [00:12<01:16, 11.78it/s]
Training loss: 0.372, Change on Adj: -0.013: 11%|█ | 106/1000 [00:12<01:16, 11.66it/s]
Training loss: 0.416, Change on Adj: -0.012: 11%|█ | 106/1000 [00:12<01:16, 11.66it/s]
Training loss: 0.383, Change on Adj: -0.012: 11%|█ | 106/1000 [00:12<01:16, 11.66it/s]
Training loss: 0.383, Change on Adj: -0.012: 11%|█ | 108/1000 [00:12<01:17, 11.57it/s]
Training loss: 0.396, Change on Adj: -0.012: 11%|█ | 108/1000 [00:12<01:17, 11.57it/s]
Training loss: 0.347, Change on Adj: -0.012: 11%|█ | 108/1000 [00:12<01:17, 11.57it/s]
Training loss: 0.347, Change on Adj: -0.012: 11%|█ | 110/1000 [00:12<01:16, 11.63it/s]
Training loss: 0.368, Change on Adj: -0.012: 11%|█ | 110/1000 [00:12<01:16, 11.63it/s]
Training loss: 0.373, Change on Adj: -0.012: 11%|█ | 110/1000 [00:12<01:16, 11.63it/s]
Training loss: 0.373, Change on Adj: -0.012: 11%|█ | 112/1000 [00:12<01:16, 11.58it/s]
Training loss: 0.395, Change on Adj: -0.012: 11%|█ | 112/1000 [00:12<01:16, 11.58it/s]
Training loss: 0.387, Change on Adj: -0.012: 11%|█ | 112/1000 [00:12<01:16, 11.58it/s]
Training loss: 0.387, Change on Adj: -0.012: 11%|█▏ | 114/1000 [00:12<01:16, 11.65it/s]
Training loss: 0.330, Change on Adj: -0.012: 11%|█▏ | 114/1000 [00:12<01:16, 11.65it/s]
Training loss: 0.342, Change on Adj: -0.012: 11%|█▏ | 114/1000 [00:12<01:16, 11.65it/s]
Training loss: 0.342, Change on Adj: -0.012: 12%|█▏ | 116/1000 [00:12<01:15, 11.71it/s]
Training loss: 0.331, Change on Adj: -0.012: 12%|█▏ | 116/1000 [00:13<01:15, 11.71it/s]
Training loss: 0.335, Change on Adj: -0.012: 12%|█▏ | 116/1000 [00:13<01:15, 11.71it/s]
Training loss: 0.335, Change on Adj: -0.012: 12%|█▏ | 118/1000 [00:13<01:15, 11.74it/s]
Training loss: 0.320, Change on Adj: -0.012: 12%|█▏ | 118/1000 [00:13<01:15, 11.74it/s]
Training loss: 0.355, Change on Adj: -0.012: 12%|█▏ | 118/1000 [00:13<01:15, 11.74it/s]
Training loss: 0.355, Change on Adj: -0.012: 12%|█▏ | 120/1000 [00:13<01:14, 11.76it/s]
Training loss: 0.308, Change on Adj: -0.012: 12%|█▏ | 120/1000 [00:13<01:14, 11.76it/s]
Training loss: 0.306, Change on Adj: -0.012: 12%|█▏ | 120/1000 [00:13<01:14, 11.76it/s]
Training loss: 0.306, Change on Adj: -0.012: 12%|█▏ | 122/1000 [00:13<01:14, 11.79it/s]
Training loss: 0.331, Change on Adj: -0.012: 12%|█▏ | 122/1000 [00:13<01:14, 11.79it/s]
Training loss: 0.308, Change on Adj: -0.012: 12%|█▏ | 122/1000 [00:13<01:14, 11.79it/s]
Training loss: 0.308, Change on Adj: -0.012: 12%|█▏ | 124/1000 [00:13<01:14, 11.80it/s]
Training loss: 0.309, Change on Adj: -0.012: 12%|█▏ | 124/1000 [00:13<01:14, 11.80it/s]
Training loss: 0.290, Change on Adj: -0.012: 12%|█▏ | 124/1000 [00:13<01:14, 11.80it/s]
Training loss: 0.290, Change on Adj: -0.012: 13%|█▎ | 126/1000 [00:13<01:13, 11.81it/s]
Training loss: 0.330, Change on Adj: -0.012: 13%|█▎ | 126/1000 [00:13<01:13, 11.81it/s]
Training loss: 0.271, Change on Adj: -0.012: 13%|█▎ | 126/1000 [00:13<01:13, 11.81it/s]
Training loss: 0.271, Change on Adj: -0.012: 13%|█▎ | 128/1000 [00:13<01:16, 11.37it/s]
Training loss: 0.304, Change on Adj: -0.012: 13%|█▎ | 128/1000 [00:14<01:16, 11.37it/s]
Training loss: 0.305, Change on Adj: -0.012: 13%|█▎ | 128/1000 [00:14<01:16, 11.37it/s]
Training loss: 0.305, Change on Adj: -0.012: 13%|█▎ | 130/1000 [00:14<01:22, 10.53it/s]
Training loss: 0.301, Change on Adj: -0.012: 13%|█▎ | 130/1000 [00:14<01:22, 10.53it/s]
Training loss: 0.304, Change on Adj: -0.012: 13%|█▎ | 130/1000 [00:14<01:22, 10.53it/s]
Training loss: 0.304, Change on Adj: -0.012: 13%|█▎ | 132/1000 [00:14<01:24, 10.30it/s]
Training loss: 0.297, Change on Adj: -0.012: 13%|█▎ | 132/1000 [00:14<01:24, 10.30it/s]
Training loss: 0.278, Change on Adj: -0.012: 13%|█▎ | 132/1000 [00:14<01:24, 10.30it/s]
Training loss: 0.278, Change on Adj: -0.012: 13%|█▎ | 134/1000 [00:14<01:23, 10.31it/s]
Training loss: 0.283, Change on Adj: -0.012: 13%|█▎ | 134/1000 [00:14<01:23, 10.31it/s]
Training loss: 0.277, Change on Adj: -0.012: 13%|█▎ | 134/1000 [00:14<01:23, 10.31it/s]
Training loss: 0.277, Change on Adj: -0.012: 14%|█▎ | 136/1000 [00:14<01:24, 10.17it/s]
Training loss: 0.296, Change on Adj: -0.012: 14%|█▎ | 136/1000 [00:14<01:24, 10.17it/s]
Training loss: 0.342, Change on Adj: -0.012: 14%|█▎ | 136/1000 [00:14<01:24, 10.17it/s]
Training loss: 0.342, Change on Adj: -0.012: 14%|█▍ | 138/1000 [00:14<01:23, 10.26it/s]
Training loss: 0.271, Change on Adj: -0.012: 14%|█▍ | 138/1000 [00:15<01:23, 10.26it/s]
Training loss: 0.288, Change on Adj: -0.012: 14%|█▍ | 138/1000 [00:15<01:23, 10.26it/s]
Training loss: 0.288, Change on Adj: -0.012: 14%|█▍ | 140/1000 [00:15<01:24, 10.20it/s]
Training loss: 0.266, Change on Adj: -0.012: 14%|█▍ | 140/1000 [00:15<01:24, 10.20it/s]
Training loss: 0.231, Change on Adj: -0.012: 14%|█▍ | 140/1000 [00:15<01:24, 10.20it/s]
Training loss: 0.231, Change on Adj: -0.012: 14%|█▍ | 142/1000 [00:15<01:25, 9.99it/s]
Training loss: 0.252, Change on Adj: -0.011: 14%|█▍ | 142/1000 [00:15<01:25, 9.99it/s]
Training loss: 0.325, Change on Adj: -0.011: 14%|█▍ | 142/1000 [00:15<01:25, 9.99it/s]
Training loss: 0.325, Change on Adj: -0.011: 14%|█▍ | 144/1000 [00:15<01:22, 10.32it/s]
Training loss: 0.258, Change on Adj: -0.011: 14%|█▍ | 144/1000 [00:15<01:22, 10.32it/s]
Training loss: 0.285, Change on Adj: -0.012: 14%|█▍ | 144/1000 [00:15<01:22, 10.32it/s]
Training loss: 0.285, Change on Adj: -0.012: 15%|█▍ | 146/1000 [00:15<01:19, 10.73it/s]
Training loss: 0.298, Change on Adj: -0.012: 15%|█▍ | 146/1000 [00:15<01:19, 10.73it/s]
Training loss: 0.284, Change on Adj: -0.012: 15%|█▍ | 146/1000 [00:15<01:19, 10.73it/s]
Training loss: 0.284, Change on Adj: -0.012: 15%|█▍ | 148/1000 [00:15<01:17, 11.04it/s]
Training loss: 0.327, Change on Adj: -0.012: 15%|█▍ | 148/1000 [00:15<01:17, 11.04it/s]
Training loss: 0.274, Change on Adj: -0.012: 15%|█▍ | 148/1000 [00:16<01:17, 11.04it/s]
Training loss: 0.274, Change on Adj: -0.012: 15%|█▌ | 150/1000 [00:16<01:15, 11.27it/s]
Training loss: 0.241, Change on Adj: -0.012: 15%|█▌ | 150/1000 [00:16<01:15, 11.27it/s]
Training loss: 0.286, Change on Adj: -0.012: 15%|█▌ | 150/1000 [00:16<01:15, 11.27it/s]
Training loss: 0.286, Change on Adj: -0.012: 15%|█▌ | 152/1000 [00:16<01:14, 11.43it/s]
Training loss: 0.270, Change on Adj: -0.012: 15%|█▌ | 152/1000 [00:16<01:14, 11.43it/s]
Training loss: 0.303, Change on Adj: -0.012: 15%|█▌ | 152/1000 [00:16<01:14, 11.43it/s]
Training loss: 0.303, Change on Adj: -0.012: 15%|█▌ | 154/1000 [00:16<01:13, 11.55it/s]
Training loss: 0.326, Change on Adj: -0.012: 15%|█▌ | 154/1000 [00:16<01:13, 11.55it/s]
Training loss: 0.216, Change on Adj: -0.012: 15%|█▌ | 154/1000 [00:16<01:13, 11.55it/s]
Training loss: 0.216, Change on Adj: -0.012: 16%|█▌ | 156/1000 [00:16<01:12, 11.63it/s]
Training loss: 0.315, Change on Adj: -0.011: 16%|█▌ | 156/1000 [00:16<01:12, 11.63it/s]
Training loss: 0.254, Change on Adj: -0.011: 16%|█▌ | 156/1000 [00:16<01:12, 11.63it/s]
Training loss: 0.254, Change on Adj: -0.011: 16%|█▌ | 158/1000 [00:16<01:11, 11.70it/s]
Training loss: 0.243, Change on Adj: -0.011: 16%|█▌ | 158/1000 [00:16<01:11, 11.70it/s]
Training loss: 0.231, Change on Adj: -0.011: 16%|█▌ | 158/1000 [00:16<01:11, 11.70it/s]
Training loss: 0.231, Change on Adj: -0.011: 16%|█▌ | 160/1000 [00:16<01:11, 11.75it/s]
Training loss: 0.255, Change on Adj: -0.011: 16%|█▌ | 160/1000 [00:17<01:11, 11.75it/s]
Training loss: 0.265, Change on Adj: -0.011: 16%|█▌ | 160/1000 [00:17<01:11, 11.75it/s]
Training loss: 0.265, Change on Adj: -0.011: 16%|█▌ | 162/1000 [00:17<01:11, 11.78it/s]
Training loss: 0.298, Change on Adj: -0.011: 16%|█▌ | 162/1000 [00:17<01:11, 11.78it/s]
Training loss: 0.273, Change on Adj: -0.011: 16%|█▌ | 162/1000 [00:17<01:11, 11.78it/s]
Training loss: 0.273, Change on Adj: -0.011: 16%|█▋ | 164/1000 [00:17<01:10, 11.80it/s]
Training loss: 0.288, Change on Adj: -0.011: 16%|█▋ | 164/1000 [00:17<01:10, 11.80it/s]
Training loss: 0.291, Change on Adj: -0.011: 16%|█▋ | 164/1000 [00:17<01:10, 11.80it/s]
Training loss: 0.291, Change on Adj: -0.011: 17%|█▋ | 166/1000 [00:17<01:10, 11.82it/s]
Training loss: 0.276, Change on Adj: -0.011: 17%|█▋ | 166/1000 [00:17<01:10, 11.82it/s]
Training loss: 0.217, Change on Adj: -0.011: 17%|█▋ | 166/1000 [00:17<01:10, 11.82it/s]
Training loss: 0.217, Change on Adj: -0.011: 17%|█▋ | 168/1000 [00:17<01:10, 11.83it/s]
Training loss: 0.312, Change on Adj: -0.011: 17%|█▋ | 168/1000 [00:17<01:10, 11.83it/s]
Training loss: 0.247, Change on Adj: -0.011: 17%|█▋ | 168/1000 [00:17<01:10, 11.83it/s]
Training loss: 0.247, Change on Adj: -0.011: 17%|█▋ | 170/1000 [00:17<01:10, 11.84it/s]
Training loss: 0.251, Change on Adj: -0.011: 17%|█▋ | 170/1000 [00:17<01:10, 11.84it/s]
Training loss: 0.258, Change on Adj: -0.011: 17%|█▋ | 170/1000 [00:17<01:10, 11.84it/s]
Training loss: 0.258, Change on Adj: -0.011: 17%|█▋ | 172/1000 [00:17<01:09, 11.85it/s]
Training loss: 0.262, Change on Adj: -0.010: 17%|█▋ | 172/1000 [00:18<01:09, 11.85it/s]
Training loss: 0.317, Change on Adj: -0.011: 17%|█▋ | 172/1000 [00:18<01:09, 11.85it/s]
Training loss: 0.317, Change on Adj: -0.011: 17%|█▋ | 174/1000 [00:18<01:09, 11.84it/s]
Training loss: 0.233, Change on Adj: -0.011: 17%|█▋ | 174/1000 [00:18<01:09, 11.84it/s]
Training loss: 0.287, Change on Adj: -0.011: 17%|█▋ | 174/1000 [00:18<01:09, 11.84it/s]
Training loss: 0.287, Change on Adj: -0.011: 18%|█▊ | 176/1000 [00:18<01:11, 11.55it/s]
Training loss: 0.258, Change on Adj: -0.010: 18%|█▊ | 176/1000 [00:18<01:11, 11.55it/s]
Training loss: 0.250, Change on Adj: -0.010: 18%|█▊ | 176/1000 [00:18<01:11, 11.55it/s]
Training loss: 0.250, Change on Adj: -0.010: 18%|█▊ | 178/1000 [00:18<01:14, 11.00it/s]
Training loss: 0.210, Change on Adj: -0.011: 18%|█▊ | 178/1000 [00:18<01:14, 11.00it/s]
Training loss: 0.259, Change on Adj: -0.011: 18%|█▊ | 178/1000 [00:18<01:14, 11.00it/s]
Training loss: 0.259, Change on Adj: -0.011: 18%|█▊ | 180/1000 [00:18<01:17, 10.63it/s]
Training loss: 0.244, Change on Adj: -0.011: 18%|█▊ | 180/1000 [00:18<01:17, 10.63it/s]
Training loss: 0.237, Change on Adj: -0.010: 18%|█▊ | 180/1000 [00:18<01:17, 10.63it/s]
Training loss: 0.237, Change on Adj: -0.010: 18%|█▊ | 182/1000 [00:18<01:16, 10.63it/s]
Training loss: 0.243, Change on Adj: -0.010: 18%|█▊ | 182/1000 [00:18<01:16, 10.63it/s]
Training loss: 0.238, Change on Adj: -0.010: 18%|█▊ | 182/1000 [00:19<01:16, 10.63it/s]
Training loss: 0.238, Change on Adj: -0.010: 18%|█▊ | 184/1000 [00:19<01:18, 10.38it/s]
Training loss: 0.224, Change on Adj: -0.010: 18%|█▊ | 184/1000 [00:19<01:18, 10.38it/s]
Training loss: 0.275, Change on Adj: -0.010: 18%|█▊ | 184/1000 [00:19<01:18, 10.38it/s]
Training loss: 0.275, Change on Adj: -0.010: 19%|█▊ | 186/1000 [00:19<01:19, 10.21it/s]
Training loss: 0.314, Change on Adj: -0.010: 19%|█▊ | 186/1000 [00:19<01:19, 10.21it/s]
Training loss: 0.250, Change on Adj: -0.010: 19%|█▊ | 186/1000 [00:19<01:19, 10.21it/s]
Training loss: 0.250, Change on Adj: -0.010: 19%|█▉ | 188/1000 [00:19<01:20, 10.10it/s]
Training loss: 0.211, Change on Adj: -0.010: 19%|█▉ | 188/1000 [00:19<01:20, 10.10it/s]
Training loss: 0.253, Change on Adj: -0.010: 19%|█▉ | 188/1000 [00:19<01:20, 10.10it/s]
Training loss: 0.253, Change on Adj: -0.010: 19%|█▉ | 190/1000 [00:19<01:20, 10.02it/s]
Training loss: 0.260, Change on Adj: -0.010: 19%|█▉ | 190/1000 [00:19<01:20, 10.02it/s]
Training loss: 0.284, Change on Adj: -0.010: 19%|█▉ | 190/1000 [00:19<01:20, 10.02it/s]
Training loss: 0.284, Change on Adj: -0.010: 19%|█▉ | 192/1000 [00:19<01:21, 9.96it/s]
Training loss: 0.276, Change on Adj: -0.010: 19%|█▉ | 192/1000 [00:19<01:21, 9.96it/s]
Training loss: 0.276, Change on Adj: -0.010: 19%|█▉ | 193/1000 [00:20<01:21, 9.94it/s]
Training loss: 0.242, Change on Adj: -0.010: 19%|█▉ | 193/1000 [00:20<01:21, 9.94it/s]
Training loss: 0.242, Change on Adj: -0.010: 19%|█▉ | 194/1000 [00:20<01:21, 9.92it/s]
Training loss: 0.246, Change on Adj: -0.009: 19%|█▉ | 194/1000 [00:20<01:21, 9.92it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|█▉ | 195/1000 [00:20<01:21, 9.89it/s]
Training loss: 0.212, Change on Adj: -0.009: 20%|█▉ | 195/1000 [00:20<01:21, 9.89it/s]
Training loss: 0.212, Change on Adj: -0.009: 20%|█▉ | 196/1000 [00:20<01:21, 9.87it/s]
Training loss: 0.267, Change on Adj: -0.009: 20%|█▉ | 196/1000 [00:20<01:21, 9.87it/s]
Training loss: 0.267, Change on Adj: -0.009: 20%|█▉ | 197/1000 [00:20<01:21, 9.85it/s]
Training loss: 0.210, Change on Adj: -0.009: 20%|█▉ | 197/1000 [00:20<01:21, 9.85it/s]
Training loss: 0.210, Change on Adj: -0.009: 20%|█▉ | 198/1000 [00:20<01:21, 9.84it/s]
Training loss: 0.220, Change on Adj: -0.009: 20%|█▉ | 198/1000 [00:20<01:21, 9.84it/s]
Training loss: 0.220, Change on Adj: -0.009: 20%|█▉ | 199/1000 [00:20<01:21, 9.83it/s]
Training loss: 0.244, Change on Adj: -0.009: 20%|█▉ | 199/1000 [00:20<01:21, 9.83it/s]
Training loss: 0.244, Change on Adj: -0.009: 20%|██ | 200/1000 [00:20<01:21, 9.83it/s]
Training loss: 0.271, Change on Adj: -0.009: 20%|██ | 200/1000 [00:20<01:21, 9.83it/s]
Training loss: 0.271, Change on Adj: -0.009: 20%|██ | 201/1000 [00:20<01:21, 9.81it/s]
Training loss: 0.228, Change on Adj: -0.009: 20%|██ | 201/1000 [00:20<01:21, 9.81it/s]
Training loss: 0.228, Change on Adj: -0.009: 20%|██ | 202/1000 [00:20<01:21, 9.81it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|██ | 202/1000 [00:21<01:21, 9.81it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|██ | 203/1000 [00:21<01:21, 9.80it/s]
Training loss: 0.215, Change on Adj: -0.009: 20%|██ | 203/1000 [00:21<01:21, 9.80it/s]
Training loss: 0.215, Change on Adj: -0.009: 20%|██ | 204/1000 [00:21<01:21, 9.79it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|██ | 204/1000 [00:21<01:21, 9.79it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|██ | 205/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.246, Change on Adj: -0.009: 20%|██ | 205/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.246, Change on Adj: -0.009: 21%|██ | 206/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.289, Change on Adj: -0.009: 21%|██ | 206/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.289, Change on Adj: -0.009: 21%|██ | 207/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.238, Change on Adj: -0.009: 21%|██ | 207/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.246, Change on Adj: -0.009: 21%|██ | 207/1000 [00:21<01:21, 9.78it/s]
Training loss: 0.246, Change on Adj: -0.009: 21%|██ | 209/1000 [00:21<01:14, 10.63it/s]
Training loss: 0.240, Change on Adj: -0.009: 21%|██ | 209/1000 [00:21<01:14, 10.63it/s]
Training loss: 0.271, Change on Adj: -0.009: 21%|██ | 209/1000 [00:21<01:14, 10.63it/s]
Training loss: 0.271, Change on Adj: -0.009: 21%|██ | 211/1000 [00:21<01:11, 11.09it/s]
Training loss: 0.202, Change on Adj: -0.009: 21%|██ | 211/1000 [00:21<01:11, 11.09it/s]
Training loss: 0.276, Change on Adj: -0.009: 21%|██ | 211/1000 [00:21<01:11, 11.09it/s]
Training loss: 0.276, Change on Adj: -0.009: 21%|██▏ | 213/1000 [00:21<01:09, 11.35it/s]
Training loss: 0.234, Change on Adj: -0.009: 21%|██▏ | 213/1000 [00:22<01:09, 11.35it/s]
Training loss: 0.199, Change on Adj: -0.009: 21%|██▏ | 213/1000 [00:22<01:09, 11.35it/s]
Training loss: 0.199, Change on Adj: -0.009: 22%|██▏ | 215/1000 [00:22<01:08, 11.51it/s]
Training loss: 0.217, Change on Adj: -0.009: 22%|██▏ | 215/1000 [00:22<01:08, 11.51it/s]
Training loss: 0.213, Change on Adj: -0.009: 22%|██▏ | 215/1000 [00:22<01:08, 11.51it/s]
Training loss: 0.213, Change on Adj: -0.009: 22%|██▏ | 217/1000 [00:22<01:07, 11.62it/s]
Training loss: 0.263, Change on Adj: -0.009: 22%|██▏ | 217/1000 [00:22<01:07, 11.62it/s]
Training loss: 0.223, Change on Adj: -0.008: 22%|██▏ | 217/1000 [00:22<01:07, 11.62it/s]
Training loss: 0.223, Change on Adj: -0.008: 22%|██▏ | 219/1000 [00:22<01:06, 11.69it/s]
Training loss: 0.199, Change on Adj: -0.008: 22%|██▏ | 219/1000 [00:22<01:06, 11.69it/s]
Training loss: 0.231, Change on Adj: -0.008: 22%|██▏ | 219/1000 [00:22<01:06, 11.69it/s]
Training loss: 0.231, Change on Adj: -0.008: 22%|██▏ | 221/1000 [00:22<01:06, 11.65it/s]
Training loss: 0.232, Change on Adj: -0.008: 22%|██▏ | 221/1000 [00:22<01:06, 11.65it/s]
Training loss: 0.284, Change on Adj: -0.009: 22%|██▏ | 221/1000 [00:22<01:06, 11.65it/s]
Training loss: 0.284, Change on Adj: -0.009: 22%|██▏ | 223/1000 [00:22<01:06, 11.71it/s]
Training loss: 0.304, Change on Adj: -0.009: 22%|██▏ | 223/1000 [00:22<01:06, 11.71it/s]
Training loss: 0.217, Change on Adj: -0.009: 22%|██▏ | 223/1000 [00:22<01:06, 11.71it/s]
Training loss: 0.217, Change on Adj: -0.009: 22%|██▎ | 225/1000 [00:22<01:05, 11.75it/s]
Training loss: 0.230, Change on Adj: -0.008: 22%|██▎ | 225/1000 [00:23<01:05, 11.75it/s]
Training loss: 0.239, Change on Adj: -0.008: 22%|██▎ | 225/1000 [00:23<01:05, 11.75it/s]
Training loss: 0.239, Change on Adj: -0.008: 23%|██▎ | 227/1000 [00:23<01:05, 11.79it/s]
Training loss: 0.248, Change on Adj: -0.008: 23%|██▎ | 227/1000 [00:23<01:05, 11.79it/s]
Training loss: 0.243, Change on Adj: -0.008: 23%|██▎ | 227/1000 [00:23<01:05, 11.79it/s]
Training loss: 0.243, Change on Adj: -0.008: 23%|██▎ | 229/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.260, Change on Adj: -0.008: 23%|██▎ | 229/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.270, Change on Adj: -0.008: 23%|██▎ | 229/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.270, Change on Adj: -0.008: 23%|██▎ | 231/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.240, Change on Adj: -0.008: 23%|██▎ | 231/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.246, Change on Adj: -0.008: 23%|██▎ | 231/1000 [00:23<01:05, 11.82it/s]
Training loss: 0.246, Change on Adj: -0.008: 23%|██▎ | 233/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.294, Change on Adj: -0.008: 23%|██▎ | 233/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.244, Change on Adj: -0.008: 23%|██▎ | 233/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.244, Change on Adj: -0.008: 24%|██▎ | 235/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.257, Change on Adj: -0.008: 24%|██▎ | 235/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.255, Change on Adj: -0.008: 24%|██▎ | 235/1000 [00:23<01:04, 11.84it/s]
Training loss: 0.255, Change on Adj: -0.008: 24%|██▎ | 237/1000 [00:23<01:04, 11.74it/s]
Training loss: 0.260, Change on Adj: -0.008: 24%|██▎ | 237/1000 [00:24<01:04, 11.74it/s]
Training loss: 0.273, Change on Adj: -0.008: 24%|██▎ | 237/1000 [00:24<01:04, 11.74it/s]
Training loss: 0.273, Change on Adj: -0.008: 24%|██▍ | 239/1000 [00:24<01:04, 11.77it/s]
Training loss: 0.271, Change on Adj: -0.008: 24%|██▍ | 239/1000 [00:24<01:04, 11.77it/s]
Training loss: 0.249, Change on Adj: -0.008: 24%|██▍ | 239/1000 [00:24<01:04, 11.77it/s]
Training loss: 0.249, Change on Adj: -0.008: 24%|██▍ | 241/1000 [00:24<01:04, 11.80it/s]
Training loss: 0.251, Change on Adj: -0.008: 24%|██▍ | 241/1000 [00:24<01:04, 11.80it/s]
Training loss: 0.211, Change on Adj: -0.008: 24%|██▍ | 241/1000 [00:24<01:04, 11.80it/s]
Training loss: 0.211, Change on Adj: -0.008: 24%|██▍ | 243/1000 [00:24<01:04, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.008: 24%|██▍ | 243/1000 [00:24<01:04, 11.81it/s]
Training loss: 0.266, Change on Adj: -0.008: 24%|██▍ | 243/1000 [00:24<01:04, 11.81it/s]
Training loss: 0.266, Change on Adj: -0.008: 24%|██▍ | 245/1000 [00:24<01:03, 11.82it/s]
Training loss: 0.192, Change on Adj: -0.008: 24%|██▍ | 245/1000 [00:24<01:03, 11.82it/s]
Training loss: 0.230, Change on Adj: -0.008: 24%|██▍ | 245/1000 [00:24<01:03, 11.82it/s]
Training loss: 0.230, Change on Adj: -0.008: 25%|██▍ | 247/1000 [00:24<01:03, 11.84it/s]
Training loss: 0.262, Change on Adj: -0.008: 25%|██▍ | 247/1000 [00:24<01:03, 11.84it/s]
Training loss: 0.222, Change on Adj: -0.008: 25%|██▍ | 247/1000 [00:24<01:03, 11.84it/s]
Training loss: 0.222, Change on Adj: -0.008: 25%|██▍ | 249/1000 [00:24<01:03, 11.84it/s]
Training loss: 0.248, Change on Adj: -0.008: 25%|██▍ | 249/1000 [00:25<01:03, 11.84it/s]
Training loss: 0.290, Change on Adj: -0.008: 25%|██▍ | 249/1000 [00:25<01:03, 11.84it/s]
Training loss: 0.290, Change on Adj: -0.008: 25%|██▌ | 251/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.226, Change on Adj: -0.008: 25%|██▌ | 251/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.250, Change on Adj: -0.008: 25%|██▌ | 251/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.250, Change on Adj: -0.008: 25%|██▌ | 253/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.197, Change on Adj: -0.008: 25%|██▌ | 253/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.251, Change on Adj: -0.008: 25%|██▌ | 253/1000 [00:25<01:03, 11.85it/s]
Training loss: 0.251, Change on Adj: -0.008: 26%|██▌ | 255/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.260, Change on Adj: -0.008: 26%|██▌ | 255/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.259, Change on Adj: -0.007: 26%|██▌ | 255/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.259, Change on Adj: -0.007: 26%|██▌ | 257/1000 [00:25<01:02, 11.84it/s]
Training loss: 0.171, Change on Adj: -0.007: 26%|██▌ | 257/1000 [00:25<01:02, 11.84it/s]
Training loss: 0.222, Change on Adj: -0.007: 26%|██▌ | 257/1000 [00:25<01:02, 11.84it/s]
Training loss: 0.222, Change on Adj: -0.007: 26%|██▌ | 259/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.225, Change on Adj: -0.007: 26%|██▌ | 259/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.203, Change on Adj: -0.007: 26%|██▌ | 259/1000 [00:25<01:02, 11.83it/s]
Training loss: 0.203, Change on Adj: -0.007: 26%|██▌ | 261/1000 [00:25<01:02, 11.85it/s]
Training loss: 0.257, Change on Adj: -0.007: 26%|██▌ | 261/1000 [00:26<01:02, 11.85it/s]
Training loss: 0.233, Change on Adj: -0.007: 26%|██▌ | 261/1000 [00:26<01:02, 11.85it/s]
Training loss: 0.233, Change on Adj: -0.007: 26%|██▋ | 263/1000 [00:26<01:02, 11.85it/s]
Training loss: 0.240, Change on Adj: -0.007: 26%|██▋ | 263/1000 [00:26<01:02, 11.85it/s]
Training loss: 0.259, Change on Adj: -0.007: 26%|██▋ | 263/1000 [00:26<01:02, 11.85it/s]
Training loss: 0.259, Change on Adj: -0.007: 26%|██▋ | 265/1000 [00:26<01:01, 11.86it/s]
Training loss: 0.230, Change on Adj: -0.007: 26%|██▋ | 265/1000 [00:26<01:01, 11.86it/s]
Training loss: 0.224, Change on Adj: -0.007: 26%|██▋ | 265/1000 [00:26<01:01, 11.86it/s]
Training loss: 0.224, Change on Adj: -0.007: 27%|██▋ | 267/1000 [00:26<01:01, 11.87it/s]
Training loss: 0.222, Change on Adj: -0.007: 27%|██▋ | 267/1000 [00:26<01:01, 11.87it/s]
Training loss: 0.270, Change on Adj: -0.007: 27%|██▋ | 267/1000 [00:26<01:01, 11.87it/s]
Training loss: 0.270, Change on Adj: -0.007: 27%|██▋ | 269/1000 [00:26<01:02, 11.71it/s]
Training loss: 0.225, Change on Adj: -0.007: 27%|██▋ | 269/1000 [00:26<01:02, 11.71it/s]
Training loss: 0.225, Change on Adj: -0.007: 27%|██▋ | 269/1000 [00:26<01:02, 11.71it/s]
Training loss: 0.225, Change on Adj: -0.007: 27%|██▋ | 271/1000 [00:26<01:05, 11.06it/s]
Training loss: 0.202, Change on Adj: -0.007: 27%|██▋ | 271/1000 [00:26<01:05, 11.06it/s]
Training loss: 0.252, Change on Adj: -0.007: 27%|██▋ | 271/1000 [00:27<01:05, 11.06it/s]
Training loss: 0.252, Change on Adj: -0.007: 27%|██▋ | 273/1000 [00:27<01:08, 10.56it/s]
Training loss: 0.260, Change on Adj: -0.007: 27%|██▋ | 273/1000 [00:27<01:08, 10.56it/s]
Training loss: 0.227, Change on Adj: -0.007: 27%|██▋ | 273/1000 [00:27<01:08, 10.56it/s]
Training loss: 0.227, Change on Adj: -0.007: 28%|██▊ | 275/1000 [00:27<01:10, 10.28it/s]
Training loss: 0.234, Change on Adj: -0.007: 28%|██▊ | 275/1000 [00:27<01:10, 10.28it/s]
Training loss: 0.227, Change on Adj: -0.007: 28%|██▊ | 275/1000 [00:27<01:10, 10.28it/s]
Training loss: 0.227, Change on Adj: -0.007: 28%|██▊ | 277/1000 [00:27<01:11, 10.10it/s]
Training loss: 0.198, Change on Adj: -0.007: 28%|██▊ | 277/1000 [00:27<01:11, 10.10it/s]
Training loss: 0.257, Change on Adj: -0.007: 28%|██▊ | 277/1000 [00:27<01:11, 10.10it/s]
Training loss: 0.257, Change on Adj: -0.007: 28%|██▊ | 279/1000 [00:27<01:12, 9.97it/s]
Training loss: 0.227, Change on Adj: -0.007: 28%|██▊ | 279/1000 [00:27<01:12, 9.97it/s]
Training loss: 0.240, Change on Adj: -0.007: 28%|██▊ | 279/1000 [00:27<01:12, 9.97it/s]
Training loss: 0.240, Change on Adj: -0.007: 28%|██▊ | 281/1000 [00:27<01:12, 9.87it/s]
Training loss: 0.282, Change on Adj: -0.007: 28%|██▊ | 281/1000 [00:28<01:12, 9.87it/s]
Training loss: 0.282, Change on Adj: -0.007: 28%|██▊ | 282/1000 [00:28<01:12, 9.84it/s]
Training loss: 0.237, Change on Adj: -0.007: 28%|██▊ | 282/1000 [00:28<01:12, 9.84it/s]
Training loss: 0.237, Change on Adj: -0.007: 28%|██▊ | 283/1000 [00:28<01:13, 9.79it/s]
Training loss: 0.233, Change on Adj: -0.007: 28%|██▊ | 283/1000 [00:28<01:13, 9.79it/s]
Training loss: 0.233, Change on Adj: -0.007: 28%|██▊ | 284/1000 [00:28<01:13, 9.72it/s]
Training loss: 0.205, Change on Adj: -0.006: 28%|██▊ | 284/1000 [00:28<01:13, 9.72it/s]
Training loss: 0.205, Change on Adj: -0.006: 28%|██▊ | 285/1000 [00:28<01:13, 9.68it/s]
Training loss: 0.232, Change on Adj: -0.006: 28%|██▊ | 285/1000 [00:28<01:13, 9.68it/s]
Training loss: 0.232, Change on Adj: -0.006: 29%|██▊ | 286/1000 [00:28<01:13, 9.67it/s]
Training loss: 0.248, Change on Adj: -0.006: 29%|██▊ | 286/1000 [00:28<01:13, 9.67it/s]
Training loss: 0.248, Change on Adj: -0.006: 29%|██▊ | 287/1000 [00:28<01:13, 9.64it/s]
Training loss: 0.223, Change on Adj: -0.006: 29%|██▊ | 287/1000 [00:28<01:13, 9.64it/s]
Training loss: 0.223, Change on Adj: -0.006: 29%|██▉ | 288/1000 [00:28<01:13, 9.62it/s]
Training loss: 0.225, Change on Adj: -0.006: 29%|██▉ | 288/1000 [00:28<01:13, 9.62it/s]
Training loss: 0.225, Change on Adj: -0.006: 29%|██▉ | 289/1000 [00:28<01:13, 9.65it/s]
Training loss: 0.240, Change on Adj: -0.006: 29%|██▉ | 289/1000 [00:28<01:13, 9.65it/s]
Training loss: 0.240, Change on Adj: -0.006: 29%|██▉ | 290/1000 [00:28<01:13, 9.63it/s]
Training loss: 0.198, Change on Adj: -0.006: 29%|██▉ | 290/1000 [00:28<01:13, 9.63it/s]
Training loss: 0.198, Change on Adj: -0.006: 29%|██▉ | 291/1000 [00:28<01:13, 9.60it/s]
Training loss: 0.247, Change on Adj: -0.006: 29%|██▉ | 291/1000 [00:29<01:13, 9.60it/s]
Training loss: 0.247, Change on Adj: -0.006: 29%|██▉ | 292/1000 [00:29<01:13, 9.58it/s]
Training loss: 0.229, Change on Adj: -0.006: 29%|██▉ | 292/1000 [00:29<01:13, 9.58it/s]
Training loss: 0.229, Change on Adj: -0.006: 29%|██▉ | 293/1000 [00:29<01:13, 9.60it/s]
Training loss: 0.244, Change on Adj: -0.006: 29%|██▉ | 293/1000 [00:29<01:13, 9.60it/s]
Training loss: 0.244, Change on Adj: -0.006: 29%|██▉ | 294/1000 [00:29<01:13, 9.59it/s]
Training loss: 0.259, Change on Adj: -0.006: 29%|██▉ | 294/1000 [00:29<01:13, 9.59it/s]
Training loss: 0.259, Change on Adj: -0.006: 30%|██▉ | 295/1000 [00:29<01:13, 9.59it/s]
Training loss: 0.180, Change on Adj: -0.006: 30%|██▉ | 295/1000 [00:29<01:13, 9.59it/s]
Training loss: 0.180, Change on Adj: -0.006: 30%|██▉ | 296/1000 [00:29<01:13, 9.63it/s]
Training loss: 0.235, Change on Adj: -0.006: 30%|██▉ | 296/1000 [00:29<01:13, 9.63it/s]
Training loss: 0.235, Change on Adj: -0.006: 30%|██▉ | 297/1000 [00:29<01:13, 9.62it/s]
Training loss: 0.233, Change on Adj: -0.006: 30%|██▉ | 297/1000 [00:29<01:13, 9.62it/s]
Training loss: 0.233, Change on Adj: -0.006: 30%|██▉ | 298/1000 [00:29<01:13, 9.60it/s]
Training loss: 0.263, Change on Adj: -0.006: 30%|██▉ | 298/1000 [00:29<01:13, 9.60it/s]
Training loss: 0.263, Change on Adj: -0.006: 30%|██▉ | 299/1000 [00:29<01:13, 9.57it/s]
Training loss: 0.236, Change on Adj: -0.006: 30%|██▉ | 299/1000 [00:29<01:13, 9.57it/s]
Training loss: 0.236, Change on Adj: -0.006: 30%|███ | 300/1000 [00:29<01:13, 9.58it/s]
Training loss: 0.248, Change on Adj: -0.006: 30%|███ | 300/1000 [00:30<01:13, 9.58it/s]
Training loss: 0.248, Change on Adj: -0.006: 30%|███ | 301/1000 [00:30<01:12, 9.59it/s]
Training loss: 0.222, Change on Adj: -0.006: 30%|███ | 301/1000 [00:30<01:12, 9.59it/s]
Training loss: 0.222, Change on Adj: -0.006: 30%|███ | 302/1000 [00:30<01:12, 9.60it/s]
Training loss: 0.305, Change on Adj: -0.006: 30%|███ | 302/1000 [00:30<01:12, 9.60it/s]
Training loss: 0.305, Change on Adj: -0.006: 30%|███ | 303/1000 [00:30<01:12, 9.62it/s]
Training loss: 0.276, Change on Adj: -0.006: 30%|███ | 303/1000 [00:30<01:12, 9.62it/s]
Training loss: 0.276, Change on Adj: -0.006: 30%|███ | 304/1000 [00:30<01:12, 9.56it/s]
Training loss: 0.269, Change on Adj: -0.006: 30%|███ | 304/1000 [00:30<01:12, 9.56it/s]
Training loss: 0.269, Change on Adj: -0.006: 30%|███ | 305/1000 [00:30<01:12, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.006: 30%|███ | 305/1000 [00:30<01:12, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.006: 31%|███ | 306/1000 [00:30<01:12, 9.54it/s]
Training loss: 0.219, Change on Adj: -0.006: 31%|███ | 306/1000 [00:30<01:12, 9.54it/s]
Training loss: 0.219, Change on Adj: -0.006: 31%|███ | 307/1000 [00:30<01:12, 9.55it/s]
Training loss: 0.241, Change on Adj: -0.005: 31%|███ | 307/1000 [00:30<01:12, 9.55it/s]
Training loss: 0.241, Change on Adj: -0.005: 31%|███ | 308/1000 [00:30<01:12, 9.57it/s]
Training loss: 0.225, Change on Adj: -0.005: 31%|███ | 308/1000 [00:30<01:12, 9.57it/s]
Training loss: 0.225, Change on Adj: -0.005: 31%|███ | 309/1000 [00:30<01:12, 9.54it/s]
Training loss: 0.259, Change on Adj: -0.005: 31%|███ | 309/1000 [00:30<01:12, 9.54it/s]
Training loss: 0.259, Change on Adj: -0.005: 31%|███ | 310/1000 [00:30<01:11, 9.60it/s]
Training loss: 0.226, Change on Adj: -0.005: 31%|███ | 310/1000 [00:31<01:11, 9.60it/s]
Training loss: 0.226, Change on Adj: -0.005: 31%|███ | 311/1000 [00:31<01:11, 9.61it/s]
Training loss: 0.189, Change on Adj: -0.005: 31%|███ | 311/1000 [00:31<01:11, 9.61it/s]
Training loss: 0.189, Change on Adj: -0.005: 31%|███ | 312/1000 [00:31<01:11, 9.61it/s]
Training loss: 0.275, Change on Adj: -0.005: 31%|███ | 312/1000 [00:31<01:11, 9.61it/s]
Training loss: 0.275, Change on Adj: -0.005: 31%|███▏ | 313/1000 [00:31<01:11, 9.59it/s]
Training loss: 0.255, Change on Adj: -0.005: 31%|███▏ | 313/1000 [00:31<01:11, 9.59it/s]
Training loss: 0.255, Change on Adj: -0.005: 31%|███▏ | 314/1000 [00:31<01:11, 9.56it/s]
Training loss: 0.255, Change on Adj: -0.005: 31%|███▏ | 314/1000 [00:31<01:11, 9.56it/s]
Training loss: 0.255, Change on Adj: -0.005: 32%|███▏ | 315/1000 [00:31<01:11, 9.55it/s]
Training loss: 0.221, Change on Adj: -0.005: 32%|███▏ | 315/1000 [00:31<01:11, 9.55it/s]
Training loss: 0.221, Change on Adj: -0.005: 32%|███▏ | 316/1000 [00:31<01:11, 9.53it/s]
Training loss: 0.252, Change on Adj: -0.005: 32%|███▏ | 316/1000 [00:31<01:11, 9.53it/s]
Training loss: 0.252, Change on Adj: -0.005: 32%|███▏ | 317/1000 [00:31<01:11, 9.57it/s]
Training loss: 0.196, Change on Adj: -0.005: 32%|███▏ | 317/1000 [00:31<01:11, 9.57it/s]
Training loss: 0.196, Change on Adj: -0.005: 32%|███▏ | 318/1000 [00:31<01:11, 9.57it/s]
Training loss: 0.250, Change on Adj: -0.005: 32%|███▏ | 318/1000 [00:31<01:11, 9.57it/s]
Training loss: 0.198, Change on Adj: -0.005: 32%|███▏ | 318/1000 [00:31<01:11, 9.57it/s]
Training loss: 0.198, Change on Adj: -0.005: 32%|███▏ | 320/1000 [00:31<01:04, 10.49it/s]
Training loss: 0.205, Change on Adj: -0.005: 32%|███▏ | 320/1000 [00:32<01:04, 10.49it/s]
Training loss: 0.282, Change on Adj: -0.005: 32%|███▏ | 320/1000 [00:32<01:04, 10.49it/s]
Training loss: 0.282, Change on Adj: -0.005: 32%|███▏ | 322/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.220, Change on Adj: -0.005: 32%|███▏ | 322/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.231, Change on Adj: -0.005: 32%|███▏ | 322/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.231, Change on Adj: -0.005: 32%|███▏ | 324/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.217, Change on Adj: -0.005: 32%|███▏ | 324/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.271, Change on Adj: -0.005: 32%|███▏ | 324/1000 [00:32<01:01, 11.00it/s]
Training loss: 0.271, Change on Adj: -0.005: 33%|███▎ | 326/1000 [00:32<01:04, 10.49it/s]
Training loss: 0.236, Change on Adj: -0.005: 33%|███▎ | 326/1000 [00:32<01:04, 10.49it/s]
Training loss: 0.263, Change on Adj: -0.005: 33%|███▎ | 326/1000 [00:32<01:04, 10.49it/s]
Training loss: 0.263, Change on Adj: -0.005: 33%|███▎ | 328/1000 [00:32<01:05, 10.20it/s]
Training loss: 0.216, Change on Adj: -0.005: 33%|███▎ | 328/1000 [00:32<01:05, 10.20it/s]
Training loss: 0.237, Change on Adj: -0.005: 33%|███▎ | 328/1000 [00:32<01:05, 10.20it/s]
Training loss: 0.237, Change on Adj: -0.005: 33%|███▎ | 330/1000 [00:32<01:06, 10.02it/s]
Training loss: 0.227, Change on Adj: -0.004: 33%|███▎ | 330/1000 [00:33<01:06, 10.02it/s]
Training loss: 0.223, Change on Adj: -0.004: 33%|███▎ | 330/1000 [00:33<01:06, 10.02it/s]
Training loss: 0.223, Change on Adj: -0.004: 33%|███▎ | 332/1000 [00:33<01:07, 9.89it/s]
Training loss: 0.203, Change on Adj: -0.004: 33%|███▎ | 332/1000 [00:33<01:07, 9.89it/s]
Training loss: 0.203, Change on Adj: -0.004: 33%|███▎ | 333/1000 [00:33<01:07, 9.84it/s]
Training loss: 0.243, Change on Adj: -0.004: 33%|███▎ | 333/1000 [00:33<01:07, 9.84it/s]
Training loss: 0.243, Change on Adj: -0.004: 33%|███▎ | 334/1000 [00:33<01:08, 9.79it/s]
Training loss: 0.223, Change on Adj: -0.004: 33%|███▎ | 334/1000 [00:33<01:08, 9.79it/s]
Training loss: 0.223, Change on Adj: -0.004: 34%|███▎ | 335/1000 [00:33<01:08, 9.73it/s]
Training loss: 0.220, Change on Adj: -0.004: 34%|███▎ | 335/1000 [00:33<01:08, 9.73it/s]
Training loss: 0.220, Change on Adj: -0.004: 34%|███▎ | 336/1000 [00:33<01:08, 9.69it/s]
Training loss: 0.263, Change on Adj: -0.004: 34%|███▎ | 336/1000 [00:33<01:08, 9.69it/s]
Training loss: 0.263, Change on Adj: -0.004: 34%|███▎ | 337/1000 [00:33<01:08, 9.70it/s]
Training loss: 0.223, Change on Adj: -0.004: 34%|███▎ | 337/1000 [00:33<01:08, 9.70it/s]
Training loss: 0.223, Change on Adj: -0.004: 34%|███▍ | 338/1000 [00:33<01:08, 9.64it/s]
Training loss: 0.243, Change on Adj: -0.004: 34%|███▍ | 338/1000 [00:33<01:08, 9.64it/s]
Training loss: 0.243, Change on Adj: -0.004: 34%|███▍ | 339/1000 [00:33<01:08, 9.66it/s]
Training loss: 0.240, Change on Adj: -0.004: 34%|███▍ | 339/1000 [00:33<01:08, 9.66it/s]
Training loss: 0.240, Change on Adj: -0.004: 34%|███▍ | 340/1000 [00:33<01:08, 9.67it/s]
Training loss: 0.258, Change on Adj: -0.004: 34%|███▍ | 340/1000 [00:34<01:08, 9.67it/s]
Training loss: 0.258, Change on Adj: -0.004: 34%|███▍ | 341/1000 [00:34<01:08, 9.66it/s]
Training loss: 0.226, Change on Adj: -0.004: 34%|███▍ | 341/1000 [00:34<01:08, 9.66it/s]
Training loss: 0.226, Change on Adj: -0.004: 34%|███▍ | 342/1000 [00:34<01:07, 9.68it/s]
Training loss: 0.242, Change on Adj: -0.004: 34%|███▍ | 342/1000 [00:34<01:07, 9.68it/s]
Training loss: 0.242, Change on Adj: -0.004: 34%|███▍ | 343/1000 [00:34<01:07, 9.66it/s]
Training loss: 0.298, Change on Adj: -0.004: 34%|███▍ | 343/1000 [00:34<01:07, 9.66it/s]
Training loss: 0.298, Change on Adj: -0.004: 34%|███▍ | 344/1000 [00:34<01:07, 9.68it/s]
Training loss: 0.285, Change on Adj: -0.004: 34%|███▍ | 344/1000 [00:34<01:07, 9.68it/s]
Training loss: 0.285, Change on Adj: -0.004: 34%|███▍ | 345/1000 [00:34<01:07, 9.66it/s]
Training loss: 0.179, Change on Adj: -0.004: 34%|███▍ | 345/1000 [00:34<01:07, 9.66it/s]
Training loss: 0.179, Change on Adj: -0.004: 35%|███▍ | 346/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.206, Change on Adj: -0.004: 35%|███▍ | 346/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.206, Change on Adj: -0.004: 35%|███▍ | 347/1000 [00:34<01:07, 9.64it/s]
Training loss: 0.270, Change on Adj: -0.004: 35%|███▍ | 347/1000 [00:34<01:07, 9.64it/s]
Training loss: 0.270, Change on Adj: -0.004: 35%|███▍ | 348/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.220, Change on Adj: -0.004: 35%|███▍ | 348/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.220, Change on Adj: -0.004: 35%|███▍ | 349/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.280, Change on Adj: -0.004: 35%|███▍ | 349/1000 [00:34<01:07, 9.63it/s]
Training loss: 0.280, Change on Adj: -0.004: 35%|███▌ | 350/1000 [00:34<01:07, 9.60it/s]
Training loss: 0.178, Change on Adj: -0.004: 35%|███▌ | 350/1000 [00:35<01:07, 9.60it/s]
Training loss: 0.178, Change on Adj: -0.004: 35%|███▌ | 351/1000 [00:35<01:07, 9.62it/s]
Training loss: 0.240, Change on Adj: -0.004: 35%|███▌ | 351/1000 [00:35<01:07, 9.62it/s]
Training loss: 0.240, Change on Adj: -0.004: 35%|███▌ | 352/1000 [00:35<01:07, 9.59it/s]
Training loss: 0.242, Change on Adj: -0.004: 35%|███▌ | 352/1000 [00:35<01:07, 9.59it/s]
Training loss: 0.242, Change on Adj: -0.004: 35%|███▌ | 353/1000 [00:35<01:07, 9.61it/s]
Training loss: 0.280, Change on Adj: -0.004: 35%|███▌ | 353/1000 [00:35<01:07, 9.61it/s]
Training loss: 0.280, Change on Adj: -0.004: 35%|███▌ | 354/1000 [00:35<01:07, 9.58it/s]
Training loss: 0.185, Change on Adj: -0.004: 35%|███▌ | 354/1000 [00:35<01:07, 9.58it/s]
Training loss: 0.185, Change on Adj: -0.004: 36%|███▌ | 355/1000 [00:35<01:07, 9.61it/s]
Training loss: 0.203, Change on Adj: -0.004: 36%|███▌ | 355/1000 [00:35<01:07, 9.61it/s]
Training loss: 0.203, Change on Adj: -0.004: 36%|███▌ | 356/1000 [00:35<01:07, 9.57it/s]
Training loss: 0.287, Change on Adj: -0.003: 36%|███▌ | 356/1000 [00:35<01:07, 9.57it/s]
Training loss: 0.287, Change on Adj: -0.003: 36%|███▌ | 357/1000 [00:35<01:07, 9.59it/s]
Training loss: 0.227, Change on Adj: -0.003: 36%|███▌ | 357/1000 [00:35<01:07, 9.59it/s]
Training loss: 0.227, Change on Adj: -0.003: 36%|███▌ | 358/1000 [00:35<01:06, 9.63it/s]
Training loss: 0.239, Change on Adj: -0.003: 36%|███▌ | 358/1000 [00:35<01:06, 9.63it/s]
Training loss: 0.239, Change on Adj: -0.003: 36%|███▌ | 359/1000 [00:35<01:06, 9.60it/s]
Training loss: 0.238, Change on Adj: -0.003: 36%|███▌ | 359/1000 [00:36<01:06, 9.60it/s]
Training loss: 0.238, Change on Adj: -0.003: 36%|███▌ | 360/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.003: 36%|███▌ | 360/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.003: 36%|███▌ | 361/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.255, Change on Adj: -0.003: 36%|███▌ | 361/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.255, Change on Adj: -0.003: 36%|███▌ | 362/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.184, Change on Adj: -0.003: 36%|███▌ | 362/1000 [00:36<01:06, 9.58it/s]
Training loss: 0.184, Change on Adj: -0.003: 36%|███▋ | 363/1000 [00:36<01:06, 9.53it/s]
Training loss: 0.256, Change on Adj: -0.003: 36%|███▋ | 363/1000 [00:36<01:06, 9.53it/s]
Training loss: 0.256, Change on Adj: -0.003: 36%|███▋ | 364/1000 [00:36<01:06, 9.53it/s]
Training loss: 0.259, Change on Adj: -0.003: 36%|███▋ | 364/1000 [00:36<01:06, 9.53it/s]
Training loss: 0.259, Change on Adj: -0.003: 36%|███▋ | 365/1000 [00:36<01:06, 9.57it/s]
Training loss: 0.247, Change on Adj: -0.003: 36%|███▋ | 365/1000 [00:36<01:06, 9.57it/s]
Training loss: 0.247, Change on Adj: -0.003: 37%|███▋ | 366/1000 [00:36<01:06, 9.56it/s]
Training loss: 0.254, Change on Adj: -0.003: 37%|███▋ | 366/1000 [00:36<01:06, 9.56it/s]
Training loss: 0.254, Change on Adj: -0.003: 37%|███▋ | 367/1000 [00:36<01:06, 9.57it/s]
Training loss: 0.192, Change on Adj: -0.003: 37%|███▋ | 367/1000 [00:36<01:06, 9.57it/s]
Training loss: 0.192, Change on Adj: -0.003: 37%|███▋ | 368/1000 [00:36<01:05, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.003: 37%|███▋ | 368/1000 [00:36<01:05, 9.58it/s]
Training loss: 0.253, Change on Adj: -0.003: 37%|███▋ | 369/1000 [00:36<01:05, 9.60it/s]
Training loss: 0.258, Change on Adj: -0.003: 37%|███▋ | 369/1000 [00:37<01:05, 9.60it/s]
Training loss: 0.258, Change on Adj: -0.003: 37%|███▋ | 370/1000 [00:37<01:05, 9.61it/s]
Training loss: 0.244, Change on Adj: -0.003: 37%|███▋ | 370/1000 [00:37<01:05, 9.61it/s]
Training loss: 0.244, Change on Adj: -0.003: 37%|███▋ | 371/1000 [00:37<01:05, 9.61it/s]
Training loss: 0.201, Change on Adj: -0.003: 37%|███▋ | 371/1000 [00:37<01:05, 9.61it/s]
Training loss: 0.201, Change on Adj: -0.003: 37%|███▋ | 372/1000 [00:37<01:04, 9.68it/s]
Training loss: 0.221, Change on Adj: -0.003: 37%|███▋ | 372/1000 [00:37<01:04, 9.68it/s]
Training loss: 0.221, Change on Adj: -0.003: 37%|███▋ | 373/1000 [00:37<01:04, 9.70it/s]
Training loss: 0.210, Change on Adj: -0.003: 37%|███▋ | 373/1000 [00:37<01:04, 9.70it/s]
Training loss: 0.210, Change on Adj: -0.003: 37%|███▋ | 374/1000 [00:37<01:04, 9.67it/s]
Training loss: 0.207, Change on Adj: -0.003: 37%|███▋ | 374/1000 [00:37<01:04, 9.67it/s]
Training loss: 0.207, Change on Adj: -0.003: 38%|███▊ | 375/1000 [00:37<01:04, 9.66it/s]
Training loss: 0.232, Change on Adj: -0.003: 38%|███▊ | 375/1000 [00:37<01:04, 9.66it/s]
Training loss: 0.232, Change on Adj: -0.003: 38%|███▊ | 376/1000 [00:37<01:04, 9.64it/s]
Training loss: 0.280, Change on Adj: -0.003: 38%|███▊ | 376/1000 [00:37<01:04, 9.64it/s]
Training loss: 0.280, Change on Adj: -0.003: 38%|███▊ | 377/1000 [00:37<01:04, 9.66it/s]
Training loss: 0.227, Change on Adj: -0.003: 38%|███▊ | 377/1000 [00:37<01:04, 9.66it/s]
Training loss: 0.227, Change on Adj: -0.003: 38%|███▊ | 378/1000 [00:37<01:04, 9.64it/s]
Training loss: 0.217, Change on Adj: -0.003: 38%|███▊ | 378/1000 [00:38<01:04, 9.64it/s]
Training loss: 0.217, Change on Adj: -0.003: 38%|███▊ | 379/1000 [00:38<01:04, 9.63it/s]
Training loss: 0.209, Change on Adj: -0.003: 38%|███▊ | 379/1000 [00:38<01:04, 9.63it/s]
Training loss: 0.209, Change on Adj: -0.003: 38%|███▊ | 380/1000 [00:38<01:04, 9.62it/s]
Training loss: 0.205, Change on Adj: -0.003: 38%|███▊ | 380/1000 [00:38<01:04, 9.62it/s]
Training loss: 0.205, Change on Adj: -0.003: 38%|███▊ | 381/1000 [00:38<01:04, 9.63it/s]
Training loss: 0.231, Change on Adj: -0.003: 38%|███▊ | 381/1000 [00:38<01:04, 9.63it/s]
Training loss: 0.231, Change on Adj: -0.003: 38%|███▊ | 382/1000 [00:38<01:04, 9.58it/s]
Training loss: 0.252, Change on Adj: -0.003: 38%|███▊ | 382/1000 [00:38<01:04, 9.58it/s]
Training loss: 0.252, Change on Adj: -0.003: 38%|███▊ | 383/1000 [00:38<01:04, 9.59it/s]
Training loss: 0.199, Change on Adj: -0.003: 38%|███▊ | 383/1000 [00:38<01:04, 9.59it/s]
Training loss: 0.199, Change on Adj: -0.003: 38%|███▊ | 384/1000 [00:38<01:03, 9.65it/s]
Training loss: 0.230, Change on Adj: -0.003: 38%|███▊ | 384/1000 [00:38<01:03, 9.65it/s]
Training loss: 0.230, Change on Adj: -0.003: 38%|███▊ | 385/1000 [00:38<01:04, 9.58it/s]
Training loss: 0.221, Change on Adj: -0.003: 38%|███▊ | 385/1000 [00:38<01:04, 9.58it/s]
Training loss: 0.221, Change on Adj: -0.003: 39%|███▊ | 386/1000 [00:38<01:03, 9.60it/s]
Training loss: 0.237, Change on Adj: -0.003: 39%|███▊ | 386/1000 [00:38<01:03, 9.60it/s]
Training loss: 0.237, Change on Adj: -0.003: 39%|███▊ | 387/1000 [00:38<01:03, 9.59it/s]
Training loss: 0.196, Change on Adj: -0.003: 39%|███▊ | 387/1000 [00:38<01:03, 9.59it/s]
Training loss: 0.196, Change on Adj: -0.003: 39%|███▉ | 388/1000 [00:38<01:03, 9.56it/s]
Training loss: 0.181, Change on Adj: -0.003: 39%|███▉ | 388/1000 [00:39<01:03, 9.56it/s]
Training loss: 0.181, Change on Adj: -0.003: 39%|███▉ | 389/1000 [00:39<01:03, 9.55it/s]
Training loss: 0.235, Change on Adj: -0.003: 39%|███▉ | 389/1000 [00:39<01:03, 9.55it/s]
Training loss: 0.235, Change on Adj: -0.003: 39%|███▉ | 390/1000 [00:39<01:03, 9.60it/s]
Training loss: 0.226, Change on Adj: -0.003: 39%|███▉ | 390/1000 [00:39<01:03, 9.60it/s]
Training loss: 0.226, Change on Adj: -0.003: 39%|███▉ | 391/1000 [00:39<01:03, 9.61it/s]
Training loss: 0.231, Change on Adj: -0.003: 39%|███▉ | 391/1000 [00:39<01:03, 9.61it/s]
Training loss: 0.231, Change on Adj: -0.003: 39%|███▉ | 392/1000 [00:39<01:03, 9.60it/s]
Training loss: 0.218, Change on Adj: -0.003: 39%|███▉ | 392/1000 [00:39<01:03, 9.60it/s]
Training loss: 0.218, Change on Adj: -0.003: 39%|███▉ | 393/1000 [00:39<01:03, 9.55it/s]
Training loss: 0.286, Change on Adj: -0.003: 39%|███▉ | 393/1000 [00:39<01:03, 9.55it/s]
Training loss: 0.286, Change on Adj: -0.003: 39%|███▉ | 394/1000 [00:39<01:03, 9.52it/s]
Training loss: 0.248, Change on Adj: -0.003: 39%|███▉ | 394/1000 [00:39<01:03, 9.52it/s]
Training loss: 0.248, Change on Adj: -0.003: 40%|███▉ | 395/1000 [00:39<01:03, 9.50it/s]
Training loss: 0.279, Change on Adj: -0.002: 40%|███▉ | 395/1000 [00:39<01:03, 9.50it/s]
Training loss: 0.279, Change on Adj: -0.002: 40%|███▉ | 396/1000 [00:39<01:03, 9.53it/s]
Training loss: 0.191, Change on Adj: -0.002: 40%|███▉ | 396/1000 [00:39<01:03, 9.53it/s]
Training loss: 0.191, Change on Adj: -0.002: 40%|███▉ | 397/1000 [00:39<01:03, 9.54it/s]
Training loss: 0.199, Change on Adj: -0.002: 40%|███▉ | 397/1000 [00:40<01:03, 9.54it/s]
Training loss: 0.199, Change on Adj: -0.002: 40%|███▉ | 398/1000 [00:40<01:03, 9.54it/s]
Training loss: 0.254, Change on Adj: -0.002: 40%|███▉ | 398/1000 [00:40<01:03, 9.54it/s]
Training loss: 0.254, Change on Adj: -0.002: 40%|███▉ | 399/1000 [00:40<01:03, 9.49it/s]
Training loss: 0.198, Change on Adj: -0.002: 40%|███▉ | 399/1000 [00:40<01:03, 9.49it/s]
Training loss: 0.198, Change on Adj: -0.002: 40%|████ | 400/1000 [00:40<01:02, 9.56it/s]
Training loss: 0.188, Change on Adj: -0.002: 40%|████ | 400/1000 [00:40<01:02, 9.56it/s]
Training loss: 0.188, Change on Adj: -0.002: 40%|████ | 401/1000 [00:40<01:02, 9.57it/s]
Training loss: 0.191, Change on Adj: -0.002: 40%|████ | 401/1000 [00:40<01:02, 9.57it/s]
Training loss: 0.191, Change on Adj: -0.002: 40%|████ | 402/1000 [00:40<01:02, 9.56it/s]
Training loss: 0.216, Change on Adj: -0.002: 40%|████ | 402/1000 [00:40<01:02, 9.56it/s]
Training loss: 0.216, Change on Adj: -0.002: 40%|████ | 403/1000 [00:40<01:02, 9.55it/s]
Training loss: 0.278, Change on Adj: -0.002: 40%|████ | 403/1000 [00:40<01:02, 9.55it/s]
Training loss: 0.278, Change on Adj: -0.002: 40%|████ | 404/1000 [00:40<01:02, 9.60it/s]
Training loss: 0.245, Change on Adj: -0.002: 40%|████ | 404/1000 [00:40<01:02, 9.60it/s]
Training loss: 0.245, Change on Adj: -0.002: 40%|████ | 405/1000 [00:40<01:02, 9.58it/s]
Training loss: 0.255, Change on Adj: -0.002: 40%|████ | 405/1000 [00:40<01:02, 9.58it/s]
Training loss: 0.255, Change on Adj: -0.002: 41%|████ | 406/1000 [00:40<01:02, 9.57it/s]
Training loss: 0.208, Change on Adj: -0.002: 41%|████ | 406/1000 [00:40<01:02, 9.57it/s]
Training loss: 0.208, Change on Adj: -0.002: 41%|████ | 407/1000 [00:40<01:02, 9.55it/s]
Training loss: 0.216, Change on Adj: -0.002: 41%|████ | 407/1000 [00:41<01:02, 9.55it/s]
Training loss: 0.216, Change on Adj: -0.002: 41%|████ | 408/1000 [00:41<01:02, 9.53it/s]
Training loss: 0.192, Change on Adj: -0.002: 41%|████ | 408/1000 [00:41<01:02, 9.53it/s]
Training loss: 0.192, Change on Adj: -0.002: 41%|████ | 409/1000 [00:41<01:01, 9.58it/s]
Training loss: 0.204, Change on Adj: -0.002: 41%|████ | 409/1000 [00:41<01:01, 9.58it/s]
Training loss: 0.204, Change on Adj: -0.002: 41%|████ | 410/1000 [00:41<01:01, 9.56it/s]
Training loss: 0.224, Change on Adj: -0.002: 41%|████ | 410/1000 [00:41<01:01, 9.56it/s]
Training loss: 0.224, Change on Adj: -0.002: 41%|████ | 411/1000 [00:41<01:01, 9.56it/s]
Training loss: 0.227, Change on Adj: -0.002: 41%|████ | 411/1000 [00:41<01:01, 9.56it/s]
Training loss: 0.227, Change on Adj: -0.002: 41%|████ | 412/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.199, Change on Adj: -0.002: 41%|████ | 412/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.199, Change on Adj: -0.002: 41%|████▏ | 413/1000 [00:41<01:01, 9.57it/s]
Training loss: 0.221, Change on Adj: -0.002: 41%|████▏ | 413/1000 [00:41<01:01, 9.57it/s]
Training loss: 0.221, Change on Adj: -0.002: 41%|████▏ | 414/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.205, Change on Adj: -0.002: 41%|████▏ | 414/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.205, Change on Adj: -0.002: 42%|████▏ | 415/1000 [00:41<01:01, 9.49it/s]
Training loss: 0.267, Change on Adj: -0.002: 42%|████▏ | 415/1000 [00:41<01:01, 9.49it/s]
Training loss: 0.267, Change on Adj: -0.002: 42%|████▏ | 416/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.246, Change on Adj: -0.002: 42%|████▏ | 416/1000 [00:41<01:01, 9.53it/s]
Training loss: 0.246, Change on Adj: -0.002: 42%|████▏ | 417/1000 [00:41<01:00, 9.58it/s]
Training loss: 0.260, Change on Adj: -0.002: 42%|████▏ | 417/1000 [00:42<01:00, 9.58it/s]
Training loss: 0.260, Change on Adj: -0.002: 42%|████▏ | 418/1000 [00:42<01:01, 9.54it/s]
Training loss: 0.215, Change on Adj: -0.002: 42%|████▏ | 418/1000 [00:42<01:01, 9.54it/s]
Training loss: 0.215, Change on Adj: -0.002: 42%|████▏ | 419/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.215, Change on Adj: -0.002: 42%|████▏ | 419/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.215, Change on Adj: -0.002: 42%|████▏ | 420/1000 [00:42<01:00, 9.54it/s]
Training loss: 0.203, Change on Adj: -0.002: 42%|████▏ | 420/1000 [00:42<01:00, 9.54it/s]
Training loss: 0.203, Change on Adj: -0.002: 42%|████▏ | 421/1000 [00:42<01:00, 9.50it/s]
Training loss: 0.230, Change on Adj: -0.002: 42%|████▏ | 421/1000 [00:42<01:00, 9.50it/s]
Training loss: 0.230, Change on Adj: -0.002: 42%|████▏ | 422/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.246, Change on Adj: -0.002: 42%|████▏ | 422/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.246, Change on Adj: -0.002: 42%|████▏ | 423/1000 [00:42<01:00, 9.57it/s]
Training loss: 0.174, Change on Adj: -0.002: 42%|████▏ | 423/1000 [00:42<01:00, 9.57it/s]
Training loss: 0.174, Change on Adj: -0.002: 42%|████▏ | 424/1000 [00:42<01:00, 9.58it/s]
Training loss: 0.236, Change on Adj: -0.002: 42%|████▏ | 424/1000 [00:42<01:00, 9.58it/s]
Training loss: 0.236, Change on Adj: -0.002: 42%|████▎ | 425/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.277, Change on Adj: -0.002: 42%|████▎ | 425/1000 [00:42<01:00, 9.53it/s]
Training loss: 0.277, Change on Adj: -0.002: 43%|████▎ | 426/1000 [00:42<01:00, 9.51it/s]
Training loss: 0.286, Change on Adj: -0.002: 43%|████▎ | 426/1000 [00:43<01:00, 9.51it/s]
Training loss: 0.286, Change on Adj: -0.002: 43%|████▎ | 427/1000 [00:43<01:00, 9.51it/s]
Training loss: 0.222, Change on Adj: -0.002: 43%|████▎ | 427/1000 [00:43<01:00, 9.51it/s]
Training loss: 0.222, Change on Adj: -0.002: 43%|████▎ | 428/1000 [00:43<01:00, 9.49it/s]
Training loss: 0.266, Change on Adj: -0.002: 43%|████▎ | 428/1000 [00:43<01:00, 9.49it/s]
Training loss: 0.266, Change on Adj: -0.002: 43%|████▎ | 429/1000 [00:43<00:59, 9.52it/s]
Training loss: 0.237, Change on Adj: -0.002: 43%|████▎ | 429/1000 [00:43<00:59, 9.52it/s]
Training loss: 0.237, Change on Adj: -0.002: 43%|████▎ | 430/1000 [00:43<01:00, 9.43it/s]
Training loss: 0.313, Change on Adj: -0.002: 43%|████▎ | 430/1000 [00:43<01:00, 9.43it/s]
Training loss: 0.313, Change on Adj: -0.002: 43%|████▎ | 431/1000 [00:43<01:00, 9.45it/s]
Training loss: 0.278, Change on Adj: -0.002: 43%|████▎ | 431/1000 [00:43<01:00, 9.45it/s]
Training loss: 0.278, Change on Adj: -0.002: 43%|████▎ | 432/1000 [00:43<01:00, 9.46it/s]
Training loss: 0.232, Change on Adj: -0.002: 43%|████▎ | 432/1000 [00:43<01:00, 9.46it/s]
Training loss: 0.232, Change on Adj: -0.002: 43%|████▎ | 433/1000 [00:43<00:59, 9.53it/s]
Training loss: 0.257, Change on Adj: -0.002: 43%|████▎ | 433/1000 [00:43<00:59, 9.53it/s]
Training loss: 0.257, Change on Adj: -0.002: 43%|████▎ | 434/1000 [00:43<00:59, 9.50it/s]
Training loss: 0.218, Change on Adj: -0.002: 43%|████▎ | 434/1000 [00:43<00:59, 9.50it/s]
Training loss: 0.218, Change on Adj: -0.002: 44%|████▎ | 435/1000 [00:43<00:59, 9.54it/s]
Training loss: 0.257, Change on Adj: -0.002: 44%|████▎ | 435/1000 [00:43<00:59, 9.54it/s]
Training loss: 0.257, Change on Adj: -0.002: 44%|████▎ | 436/1000 [00:43<00:59, 9.55it/s]
Training loss: 0.219, Change on Adj: -0.002: 44%|████▎ | 436/1000 [00:44<00:59, 9.55it/s]
Training loss: 0.219, Change on Adj: -0.002: 44%|████▎ | 437/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.215, Change on Adj: -0.002: 44%|████▎ | 437/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.215, Change on Adj: -0.002: 44%|████▍ | 438/1000 [00:44<00:58, 9.53it/s]
Training loss: 0.174, Change on Adj: -0.002: 44%|████▍ | 438/1000 [00:44<00:58, 9.53it/s]
Training loss: 0.174, Change on Adj: -0.002: 44%|████▍ | 439/1000 [00:44<00:58, 9.54it/s]
Training loss: 0.190, Change on Adj: -0.002: 44%|████▍ | 439/1000 [00:44<00:58, 9.54it/s]
Training loss: 0.190, Change on Adj: -0.002: 44%|████▍ | 440/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.236, Change on Adj: -0.002: 44%|████▍ | 440/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.236, Change on Adj: -0.002: 44%|████▍ | 441/1000 [00:44<00:58, 9.57it/s]
Training loss: 0.245, Change on Adj: -0.002: 44%|████▍ | 441/1000 [00:44<00:58, 9.57it/s]
Training loss: 0.245, Change on Adj: -0.002: 44%|████▍ | 442/1000 [00:44<00:58, 9.58it/s]
Training loss: 0.248, Change on Adj: -0.002: 44%|████▍ | 442/1000 [00:44<00:58, 9.58it/s]
Training loss: 0.248, Change on Adj: -0.002: 44%|████▍ | 443/1000 [00:44<00:58, 9.54it/s]
Training loss: 0.218, Change on Adj: -0.002: 44%|████▍ | 443/1000 [00:44<00:58, 9.54it/s]
Training loss: 0.218, Change on Adj: -0.002: 44%|████▍ | 444/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.250, Change on Adj: -0.002: 44%|████▍ | 444/1000 [00:44<00:58, 9.55it/s]
Training loss: 0.250, Change on Adj: -0.002: 44%|████▍ | 445/1000 [00:44<00:58, 9.52it/s]
Training loss: 0.212, Change on Adj: -0.002: 44%|████▍ | 445/1000 [00:45<00:58, 9.52it/s]
Training loss: 0.212, Change on Adj: -0.002: 45%|████▍ | 446/1000 [00:45<00:58, 9.52it/s]
Training loss: 0.237, Change on Adj: -0.002: 45%|████▍ | 446/1000 [00:45<00:58, 9.52it/s]
Training loss: 0.237, Change on Adj: -0.002: 45%|████▍ | 447/1000 [00:45<00:58, 9.51it/s]
Training loss: 0.250, Change on Adj: -0.002: 45%|████▍ | 447/1000 [00:45<00:58, 9.51it/s]
Training loss: 0.250, Change on Adj: -0.002: 45%|████▍ | 448/1000 [00:45<00:57, 9.53it/s]
Training loss: 0.202, Change on Adj: -0.002: 45%|████▍ | 448/1000 [00:45<00:57, 9.53it/s]
Training loss: 0.202, Change on Adj: -0.002: 45%|████▍ | 449/1000 [00:45<00:58, 9.50it/s]
Training loss: 0.245, Change on Adj: -0.002: 45%|████▍ | 449/1000 [00:45<00:58, 9.50it/s]
Training loss: 0.245, Change on Adj: -0.002: 45%|████▌ | 450/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.205, Change on Adj: -0.002: 45%|████▌ | 450/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.205, Change on Adj: -0.002: 45%|████▌ | 451/1000 [00:45<00:57, 9.50it/s]
Training loss: 0.245, Change on Adj: -0.002: 45%|████▌ | 451/1000 [00:45<00:57, 9.50it/s]
Training loss: 0.245, Change on Adj: -0.002: 45%|████▌ | 452/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.218, Change on Adj: -0.002: 45%|████▌ | 452/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.218, Change on Adj: -0.002: 45%|████▌ | 453/1000 [00:45<00:57, 9.53it/s]
Training loss: 0.188, Change on Adj: -0.002: 45%|████▌ | 453/1000 [00:45<00:57, 9.53it/s]
Training loss: 0.188, Change on Adj: -0.002: 45%|████▌ | 454/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.234, Change on Adj: -0.002: 45%|████▌ | 454/1000 [00:45<00:57, 9.51it/s]
Training loss: 0.234, Change on Adj: -0.002: 46%|████▌ | 455/1000 [00:45<00:57, 9.52it/s]
Training loss: 0.229, Change on Adj: -0.002: 46%|████▌ | 455/1000 [00:46<00:57, 9.52it/s]
Training loss: 0.229, Change on Adj: -0.002: 46%|████▌ | 456/1000 [00:46<00:57, 9.53it/s]
Training loss: 0.181, Change on Adj: -0.002: 46%|████▌ | 456/1000 [00:46<00:57, 9.53it/s]
Training loss: 0.181, Change on Adj: -0.002: 46%|████▌ | 457/1000 [00:46<00:56, 9.53it/s]
Training loss: 0.251, Change on Adj: -0.002: 46%|████▌ | 457/1000 [00:46<00:56, 9.53it/s]
Training loss: 0.251, Change on Adj: -0.002: 46%|████▌ | 458/1000 [00:46<00:57, 9.50it/s]
Training loss: 0.205, Change on Adj: -0.002: 46%|████▌ | 458/1000 [00:46<00:57, 9.50it/s]
Training loss: 0.205, Change on Adj: -0.002: 46%|████▌ | 459/1000 [00:46<00:56, 9.50it/s]
Training loss: 0.234, Change on Adj: -0.002: 46%|████▌ | 459/1000 [00:46<00:56, 9.50it/s]
Training loss: 0.234, Change on Adj: -0.002: 46%|████▌ | 460/1000 [00:46<00:56, 9.49it/s]
Training loss: 0.255, Change on Adj: -0.002: 46%|████▌ | 460/1000 [00:46<00:56, 9.49it/s]
Training loss: 0.255, Change on Adj: -0.002: 46%|████▌ | 461/1000 [00:46<00:56, 9.49it/s]
Training loss: 0.203, Change on Adj: -0.002: 46%|████▌ | 461/1000 [00:46<00:56, 9.49it/s]
Training loss: 0.203, Change on Adj: -0.002: 46%|████▌ | 462/1000 [00:46<00:56, 9.47it/s]
Training loss: 0.267, Change on Adj: -0.002: 46%|████▌ | 462/1000 [00:46<00:56, 9.47it/s]
Training loss: 0.267, Change on Adj: -0.002: 46%|████▋ | 463/1000 [00:46<00:56, 9.51it/s]
Training loss: 0.209, Change on Adj: -0.002: 46%|████▋ | 463/1000 [00:46<00:56, 9.51it/s]
Training loss: 0.209, Change on Adj: -0.002: 46%|████▋ | 464/1000 [00:46<00:56, 9.50it/s]
Training loss: 0.250, Change on Adj: -0.002: 46%|████▋ | 464/1000 [00:47<00:56, 9.50it/s]
Training loss: 0.250, Change on Adj: -0.002: 46%|████▋ | 465/1000 [00:47<00:56, 9.51it/s]
Training loss: 0.263, Change on Adj: -0.002: 46%|████▋ | 465/1000 [00:47<00:56, 9.51it/s]
Training loss: 0.263, Change on Adj: -0.002: 47%|████▋ | 466/1000 [00:47<00:56, 9.52it/s]
Training loss: 0.292, Change on Adj: -0.002: 47%|████▋ | 466/1000 [00:47<00:56, 9.52it/s]
Training loss: 0.292, Change on Adj: -0.002: 47%|████▋ | 467/1000 [00:47<00:55, 9.53it/s]
Training loss: 0.251, Change on Adj: -0.002: 47%|████▋ | 467/1000 [00:47<00:55, 9.53it/s]
Training loss: 0.251, Change on Adj: -0.002: 47%|████▋ | 468/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.187, Change on Adj: -0.002: 47%|████▋ | 468/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.187, Change on Adj: -0.002: 47%|████▋ | 469/1000 [00:47<00:55, 9.53it/s]
Training loss: 0.193, Change on Adj: -0.002: 47%|████▋ | 469/1000 [00:47<00:55, 9.53it/s]
Training loss: 0.193, Change on Adj: -0.002: 47%|████▋ | 470/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.210, Change on Adj: -0.002: 47%|████▋ | 470/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.210, Change on Adj: -0.002: 47%|████▋ | 471/1000 [00:47<00:55, 9.55it/s]
Training loss: 0.273, Change on Adj: -0.002: 47%|████▋ | 471/1000 [00:47<00:55, 9.55it/s]
Training loss: 0.273, Change on Adj: -0.002: 47%|████▋ | 472/1000 [00:47<00:55, 9.54it/s]
Training loss: 0.247, Change on Adj: -0.002: 47%|████▋ | 472/1000 [00:47<00:55, 9.54it/s]
Training loss: 0.247, Change on Adj: -0.002: 47%|████▋ | 473/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.225, Change on Adj: -0.002: 47%|████▋ | 473/1000 [00:47<00:55, 9.52it/s]
Training loss: 0.225, Change on Adj: -0.002: 47%|████▋ | 474/1000 [00:47<00:55, 9.48it/s]
Training loss: 0.185, Change on Adj: -0.002: 47%|████▋ | 474/1000 [00:48<00:55, 9.48it/s]
Training loss: 0.185, Change on Adj: -0.002: 48%|████▊ | 475/1000 [00:48<00:55, 9.52it/s]
Training loss: 0.220, Change on Adj: -0.002: 48%|████▊ | 475/1000 [00:48<00:55, 9.52it/s]
Training loss: 0.220, Change on Adj: -0.002: 48%|████▊ | 476/1000 [00:48<00:55, 9.52it/s]
Training loss: 0.243, Change on Adj: -0.002: 48%|████▊ | 476/1000 [00:48<00:55, 9.52it/s]
Training loss: 0.243, Change on Adj: -0.002: 48%|████▊ | 477/1000 [00:48<00:54, 9.51it/s]
Training loss: 0.252, Change on Adj: -0.002: 48%|████▊ | 477/1000 [00:48<00:54, 9.51it/s]
Training loss: 0.252, Change on Adj: -0.002: 48%|████▊ | 478/1000 [00:48<00:54, 9.51it/s]
Training loss: 0.213, Change on Adj: -0.002: 48%|████▊ | 478/1000 [00:48<00:54, 9.51it/s]
Training loss: 0.213, Change on Adj: -0.002: 48%|████▊ | 479/1000 [00:48<00:54, 9.52it/s]
Training loss: 0.259, Change on Adj: -0.002: 48%|████▊ | 479/1000 [00:48<00:54, 9.52it/s]
Training loss: 0.259, Change on Adj: -0.002: 48%|████▊ | 480/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.254, Change on Adj: -0.002: 48%|████▊ | 480/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.254, Change on Adj: -0.002: 48%|████▊ | 481/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.270, Change on Adj: -0.002: 48%|████▊ | 481/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.270, Change on Adj: -0.002: 48%|████▊ | 482/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.267, Change on Adj: -0.001: 48%|████▊ | 482/1000 [00:48<00:54, 9.53it/s]
Training loss: 0.267, Change on Adj: -0.001: 48%|████▊ | 483/1000 [00:48<00:54, 9.55it/s]
Training loss: 0.245, Change on Adj: -0.001: 48%|████▊ | 483/1000 [00:49<00:54, 9.55it/s]
Training loss: 0.245, Change on Adj: -0.001: 48%|████▊ | 484/1000 [00:49<00:54, 9.55it/s]
Training loss: 0.212, Change on Adj: -0.001: 48%|████▊ | 484/1000 [00:49<00:54, 9.55it/s]
Training loss: 0.212, Change on Adj: -0.001: 48%|████▊ | 485/1000 [00:49<00:54, 9.52it/s]
Training loss: 0.241, Change on Adj: -0.001: 48%|████▊ | 485/1000 [00:49<00:54, 9.52it/s]
Training loss: 0.241, Change on Adj: -0.001: 49%|████▊ | 486/1000 [00:49<00:53, 9.53it/s]
Training loss: 0.283, Change on Adj: -0.001: 49%|████▊ | 486/1000 [00:49<00:53, 9.53it/s]
Training loss: 0.283, Change on Adj: -0.001: 49%|████▊ | 487/1000 [00:49<00:53, 9.55it/s]
Training loss: 0.226, Change on Adj: -0.001: 49%|████▊ | 487/1000 [00:49<00:53, 9.55it/s]
Training loss: 0.226, Change on Adj: -0.001: 49%|████▉ | 488/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.225, Change on Adj: -0.001: 49%|████▉ | 488/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.225, Change on Adj: -0.001: 49%|████▉ | 489/1000 [00:49<00:53, 9.53it/s]
Training loss: 0.262, Change on Adj: -0.001: 49%|████▉ | 489/1000 [00:49<00:53, 9.53it/s]
Training loss: 0.262, Change on Adj: -0.001: 49%|████▉ | 490/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.249, Change on Adj: -0.001: 49%|████▉ | 490/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.249, Change on Adj: -0.001: 49%|████▉ | 491/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.192, Change on Adj: -0.002: 49%|████▉ | 491/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.192, Change on Adj: -0.002: 49%|████▉ | 492/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.251, Change on Adj: -0.002: 49%|████▉ | 492/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.251, Change on Adj: -0.002: 49%|████▉ | 493/1000 [00:49<00:53, 9.52it/s]
Training loss: 0.249, Change on Adj: -0.002: 49%|████▉ | 493/1000 [00:50<00:53, 9.52it/s]
Training loss: 0.249, Change on Adj: -0.002: 49%|████▉ | 494/1000 [00:50<00:53, 9.51it/s]
Training loss: 0.224, Change on Adj: -0.001: 49%|████▉ | 494/1000 [00:50<00:53, 9.51it/s]
Training loss: 0.224, Change on Adj: -0.001: 50%|████▉ | 495/1000 [00:50<00:53, 9.49it/s]
Training loss: 0.204, Change on Adj: -0.001: 50%|████▉ | 495/1000 [00:50<00:53, 9.49it/s]
Training loss: 0.204, Change on Adj: -0.001: 50%|████▉ | 496/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.221, Change on Adj: -0.001: 50%|████▉ | 496/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.221, Change on Adj: -0.001: 50%|████▉ | 497/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.196, Change on Adj: -0.001: 50%|████▉ | 497/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.196, Change on Adj: -0.001: 50%|████▉ | 498/1000 [00:50<00:52, 9.50it/s]
Training loss: 0.236, Change on Adj: -0.001: 50%|████▉ | 498/1000 [00:50<00:52, 9.50it/s]
Training loss: 0.236, Change on Adj: -0.001: 50%|████▉ | 499/1000 [00:50<00:52, 9.46it/s]
Training loss: 0.217, Change on Adj: -0.001: 50%|████▉ | 499/1000 [00:50<00:52, 9.46it/s]
Training loss: 0.217, Change on Adj: -0.001: 50%|█████ | 500/1000 [00:50<00:52, 9.48it/s]
Training loss: 0.270, Change on Adj: -0.001: 50%|█████ | 500/1000 [00:50<00:52, 9.48it/s]
Training loss: 0.270, Change on Adj: -0.001: 50%|█████ | 501/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.229, Change on Adj: -0.001: 50%|█████ | 501/1000 [00:50<00:52, 9.51it/s]
Training loss: 0.229, Change on Adj: -0.001: 50%|█████ | 502/1000 [00:50<00:52, 9.50it/s]
Training loss: 0.197, Change on Adj: -0.001: 50%|█████ | 502/1000 [00:51<00:52, 9.50it/s]
Training loss: 0.197, Change on Adj: -0.001: 50%|█████ | 503/1000 [00:51<00:52, 9.49it/s]
Training loss: 0.268, Change on Adj: -0.001: 50%|█████ | 503/1000 [00:51<00:52, 9.49it/s]
Training loss: 0.268, Change on Adj: -0.001: 50%|█████ | 504/1000 [00:51<00:52, 9.46it/s]
Training loss: 0.196, Change on Adj: -0.002: 50%|█████ | 504/1000 [00:51<00:52, 9.46it/s]
Training loss: 0.196, Change on Adj: -0.002: 50%|█████ | 505/1000 [00:51<00:52, 9.45it/s]
Training loss: 0.236, Change on Adj: -0.002: 50%|█████ | 505/1000 [00:51<00:52, 9.45it/s]
Training loss: 0.236, Change on Adj: -0.002: 51%|█████ | 506/1000 [00:51<00:52, 9.47it/s]
Training loss: 0.188, Change on Adj: -0.002: 51%|█████ | 506/1000 [00:51<00:52, 9.47it/s]
Training loss: 0.188, Change on Adj: -0.002: 51%|█████ | 507/1000 [00:51<00:52, 9.47it/s]
Training loss: 0.247, Change on Adj: -0.001: 51%|█████ | 507/1000 [00:51<00:52, 9.47it/s]
Training loss: 0.247, Change on Adj: -0.001: 51%|█████ | 508/1000 [00:51<00:52, 9.45it/s]
Training loss: 0.238, Change on Adj: -0.002: 51%|█████ | 508/1000 [00:51<00:52, 9.45it/s]
Training loss: 0.238, Change on Adj: -0.002: 51%|█████ | 509/1000 [00:51<00:51, 9.52it/s]
Training loss: 0.263, Change on Adj: -0.001: 51%|█████ | 509/1000 [00:51<00:51, 9.52it/s]
Training loss: 0.263, Change on Adj: -0.001: 51%|█████ | 510/1000 [00:51<00:51, 9.54it/s]
Training loss: 0.266, Change on Adj: -0.001: 51%|█████ | 510/1000 [00:51<00:51, 9.54it/s]
Training loss: 0.266, Change on Adj: -0.001: 51%|█████ | 511/1000 [00:51<00:51, 9.48it/s]
Training loss: 0.202, Change on Adj: -0.002: 51%|█████ | 511/1000 [00:51<00:51, 9.48it/s]
Training loss: 0.202, Change on Adj: -0.002: 51%|█████ | 512/1000 [00:51<00:51, 9.44it/s]
Training loss: 0.237, Change on Adj: -0.002: 51%|█████ | 512/1000 [00:52<00:51, 9.44it/s]
Training loss: 0.237, Change on Adj: -0.002: 51%|█████▏ | 513/1000 [00:52<00:51, 9.45it/s]
Training loss: 0.259, Change on Adj: -0.002: 51%|█████▏ | 513/1000 [00:52<00:51, 9.45it/s]
Training loss: 0.259, Change on Adj: -0.002: 51%|█████▏ | 514/1000 [00:52<00:51, 9.46it/s]
Training loss: 0.314, Change on Adj: -0.001: 51%|█████▏ | 514/1000 [00:52<00:51, 9.46it/s]
Training loss: 0.314, Change on Adj: -0.001: 52%|█████▏ | 515/1000 [00:52<00:50, 9.51it/s]
Training loss: 0.212, Change on Adj: -0.001: 52%|█████▏ | 515/1000 [00:52<00:50, 9.51it/s]
Training loss: 0.212, Change on Adj: -0.001: 52%|█████▏ | 516/1000 [00:52<00:50, 9.52it/s]
Training loss: 0.227, Change on Adj: -0.001: 52%|█████▏ | 516/1000 [00:52<00:50, 9.52it/s]
Training loss: 0.227, Change on Adj: -0.001: 52%|█████▏ | 517/1000 [00:52<00:50, 9.49it/s]
Training loss: 0.224, Change on Adj: -0.001: 52%|█████▏ | 517/1000 [00:52<00:50, 9.49it/s]
Training loss: 0.224, Change on Adj: -0.001: 52%|█████▏ | 518/1000 [00:52<00:51, 9.44it/s]
Training loss: 0.250, Change on Adj: -0.001: 52%|█████▏ | 518/1000 [00:52<00:51, 9.44it/s]
Training loss: 0.250, Change on Adj: -0.001: 52%|█████▏ | 519/1000 [00:52<00:51, 9.43it/s]
Training loss: 0.218, Change on Adj: -0.001: 52%|█████▏ | 519/1000 [00:52<00:51, 9.43it/s]
Training loss: 0.218, Change on Adj: -0.001: 52%|█████▏ | 520/1000 [00:52<00:50, 9.43it/s]
Training loss: 0.193, Change on Adj: -0.001: 52%|█████▏ | 520/1000 [00:52<00:50, 9.43it/s]
Training loss: 0.193, Change on Adj: -0.001: 52%|█████▏ | 521/1000 [00:52<00:50, 9.43it/s]
Training loss: 0.209, Change on Adj: -0.001: 52%|█████▏ | 521/1000 [00:53<00:50, 9.43it/s]
Training loss: 0.209, Change on Adj: -0.001: 52%|█████▏ | 522/1000 [00:53<00:50, 9.44it/s]
Training loss: 0.234, Change on Adj: -0.001: 52%|█████▏ | 522/1000 [00:53<00:50, 9.44it/s]
Training loss: 0.234, Change on Adj: -0.001: 52%|█████▏ | 523/1000 [00:53<00:50, 9.45it/s]
Training loss: 0.229, Change on Adj: -0.001: 52%|█████▏ | 523/1000 [00:53<00:50, 9.45it/s]
Training loss: 0.229, Change on Adj: -0.001: 52%|█████▏ | 524/1000 [00:53<00:50, 9.45it/s]
Training loss: 0.238, Change on Adj: -0.001: 52%|█████▏ | 524/1000 [00:53<00:50, 9.45it/s]
Training loss: 0.238, Change on Adj: -0.001: 52%|█████▎ | 525/1000 [00:53<00:50, 9.42it/s]
Training loss: 0.240, Change on Adj: -0.001: 52%|█████▎ | 525/1000 [00:53<00:50, 9.42it/s]
Training loss: 0.240, Change on Adj: -0.001: 53%|█████▎ | 526/1000 [00:53<00:50, 9.44it/s]
Training loss: 0.221, Change on Adj: -0.001: 53%|█████▎ | 526/1000 [00:53<00:50, 9.44it/s]
Training loss: 0.221, Change on Adj: -0.001: 53%|█████▎ | 527/1000 [00:53<00:50, 9.43it/s]
Training loss: 0.170, Change on Adj: -0.001: 53%|█████▎ | 527/1000 [00:53<00:50, 9.43it/s]
Training loss: 0.170, Change on Adj: -0.001: 53%|█████▎ | 528/1000 [00:53<00:50, 9.40it/s]
Training loss: 0.211, Change on Adj: -0.001: 53%|█████▎ | 528/1000 [00:53<00:50, 9.40it/s]
Training loss: 0.211, Change on Adj: -0.001: 53%|█████▎ | 529/1000 [00:53<00:49, 9.42it/s]
Training loss: 0.195, Change on Adj: -0.001: 53%|█████▎ | 529/1000 [00:53<00:49, 9.42it/s]
Training loss: 0.195, Change on Adj: -0.001: 53%|█████▎ | 530/1000 [00:53<00:49, 9.45it/s]
Training loss: 0.212, Change on Adj: -0.002: 53%|█████▎ | 530/1000 [00:53<00:49, 9.45it/s]
Training loss: 0.212, Change on Adj: -0.002: 53%|█████▎ | 531/1000 [00:53<00:49, 9.47it/s]
Training loss: 0.241, Change on Adj: -0.001: 53%|█████▎ | 531/1000 [00:54<00:49, 9.47it/s]
Training loss: 0.241, Change on Adj: -0.001: 53%|█████▎ | 532/1000 [00:54<00:49, 9.44it/s]
Training loss: 0.253, Change on Adj: -0.001: 53%|█████▎ | 532/1000 [00:54<00:49, 9.44it/s]
Training loss: 0.253, Change on Adj: -0.001: 53%|█████▎ | 533/1000 [00:54<00:49, 9.43it/s]
Training loss: 0.258, Change on Adj: -0.001: 53%|█████▎ | 533/1000 [00:54<00:49, 9.43it/s]
Training loss: 0.258, Change on Adj: -0.001: 53%|█████▎ | 534/1000 [00:54<00:49, 9.42it/s]
Training loss: 0.227, Change on Adj: -0.001: 53%|█████▎ | 534/1000 [00:54<00:49, 9.42it/s]
Training loss: 0.227, Change on Adj: -0.001: 54%|█████▎ | 535/1000 [00:54<00:49, 9.44it/s]
Training loss: 0.200, Change on Adj: -0.001: 54%|█████▎ | 535/1000 [00:54<00:49, 9.44it/s]
Training loss: 0.200, Change on Adj: -0.001: 54%|█████▎ | 536/1000 [00:54<00:49, 9.45it/s]
Training loss: 0.184, Change on Adj: -0.001: 54%|█████▎ | 536/1000 [00:54<00:49, 9.45it/s]
Training loss: 0.184, Change on Adj: -0.001: 54%|█████▎ | 537/1000 [00:54<00:49, 9.43it/s]
Training loss: 0.208, Change on Adj: -0.002: 54%|█████▎ | 537/1000 [00:54<00:49, 9.43it/s]
Training loss: 0.208, Change on Adj: -0.002: 54%|█████▍ | 538/1000 [00:54<00:48, 9.45it/s]
Training loss: 0.206, Change on Adj: -0.002: 54%|█████▍ | 538/1000 [00:54<00:48, 9.45it/s]
Training loss: 0.206, Change on Adj: -0.002: 54%|█████▍ | 539/1000 [00:54<00:48, 9.47it/s]
Training loss: 0.255, Change on Adj: -0.002: 54%|█████▍ | 539/1000 [00:54<00:48, 9.47it/s]
Training loss: 0.255, Change on Adj: -0.002: 54%|█████▍ | 540/1000 [00:54<00:48, 9.46it/s]
Training loss: 0.240, Change on Adj: -0.001: 54%|█████▍ | 540/1000 [00:55<00:48, 9.46it/s]
Training loss: 0.240, Change on Adj: -0.001: 54%|█████▍ | 541/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.216, Change on Adj: -0.002: 54%|█████▍ | 541/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.216, Change on Adj: -0.002: 54%|█████▍ | 542/1000 [00:55<00:48, 9.45it/s]
Training loss: 0.252, Change on Adj: -0.001: 54%|█████▍ | 542/1000 [00:55<00:48, 9.45it/s]
Training loss: 0.252, Change on Adj: -0.001: 54%|█████▍ | 543/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.169, Change on Adj: -0.001: 54%|█████▍ | 543/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.169, Change on Adj: -0.001: 54%|█████▍ | 544/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.228, Change on Adj: -0.001: 54%|█████▍ | 544/1000 [00:55<00:48, 9.43it/s]
Training loss: 0.228, Change on Adj: -0.001: 55%|█████▍ | 545/1000 [00:55<00:48, 9.41it/s]
Training loss: 0.242, Change on Adj: -0.001: 55%|█████▍ | 545/1000 [00:55<00:48, 9.41it/s]
Training loss: 0.242, Change on Adj: -0.001: 55%|█████▍ | 546/1000 [00:55<00:47, 9.46it/s]
Training loss: 0.247, Change on Adj: -0.001: 55%|█████▍ | 546/1000 [00:55<00:47, 9.46it/s]
Training loss: 0.247, Change on Adj: -0.001: 55%|█████▍ | 547/1000 [00:55<00:48, 9.42it/s]
Training loss: 0.282, Change on Adj: -0.001: 55%|█████▍ | 547/1000 [00:55<00:48, 9.42it/s]
Training loss: 0.282, Change on Adj: -0.001: 55%|█████▍ | 548/1000 [00:55<00:47, 9.44it/s]
Training loss: 0.230, Change on Adj: -0.001: 55%|█████▍ | 548/1000 [00:55<00:47, 9.44it/s]
Training loss: 0.230, Change on Adj: -0.001: 55%|█████▍ | 549/1000 [00:55<00:47, 9.46it/s]
Training loss: 0.180, Change on Adj: -0.001: 55%|█████▍ | 549/1000 [00:56<00:47, 9.46it/s]
Training loss: 0.180, Change on Adj: -0.001: 55%|█████▌ | 550/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.214, Change on Adj: -0.001: 55%|█████▌ | 550/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.214, Change on Adj: -0.001: 55%|█████▌ | 551/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.235, Change on Adj: -0.001: 55%|█████▌ | 551/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.235, Change on Adj: -0.001: 55%|█████▌ | 552/1000 [00:56<00:47, 9.45it/s]
Training loss: 0.231, Change on Adj: -0.001: 55%|█████▌ | 552/1000 [00:56<00:47, 9.45it/s]
Training loss: 0.231, Change on Adj: -0.001: 55%|█████▌ | 553/1000 [00:56<00:47, 9.45it/s]
Training loss: 0.237, Change on Adj: -0.001: 55%|█████▌ | 553/1000 [00:56<00:47, 9.45it/s]
Training loss: 0.237, Change on Adj: -0.001: 55%|█████▌ | 554/1000 [00:56<00:47, 9.42it/s]
Training loss: 0.211, Change on Adj: -0.001: 55%|█████▌ | 554/1000 [00:56<00:47, 9.42it/s]
Training loss: 0.211, Change on Adj: -0.001: 56%|█████▌ | 555/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.238, Change on Adj: -0.001: 56%|█████▌ | 555/1000 [00:56<00:47, 9.44it/s]
Training loss: 0.238, Change on Adj: -0.001: 56%|█████▌ | 556/1000 [00:56<00:46, 9.45it/s]
Training loss: 0.237, Change on Adj: -0.001: 56%|█████▌ | 556/1000 [00:56<00:46, 9.45it/s]
Training loss: 0.237, Change on Adj: -0.001: 56%|█████▌ | 557/1000 [00:56<00:46, 9.47it/s]
Training loss: 0.201, Change on Adj: -0.001: 56%|█████▌ | 557/1000 [00:56<00:46, 9.47it/s]
Training loss: 0.201, Change on Adj: -0.001: 56%|█████▌ | 558/1000 [00:56<00:46, 9.48it/s]
Training loss: 0.207, Change on Adj: -0.001: 56%|█████▌ | 558/1000 [00:56<00:46, 9.48it/s]
Training loss: 0.207, Change on Adj: -0.001: 56%|█████▌ | 559/1000 [00:56<00:46, 9.47it/s]
Training loss: 0.282, Change on Adj: -0.001: 56%|█████▌ | 559/1000 [00:57<00:46, 9.47it/s]
Training loss: 0.282, Change on Adj: -0.001: 56%|█████▌ | 560/1000 [00:57<00:46, 9.49it/s]
Training loss: 0.279, Change on Adj: -0.001: 56%|█████▌ | 560/1000 [00:57<00:46, 9.49it/s]
Training loss: 0.279, Change on Adj: -0.001: 56%|█████▌ | 561/1000 [00:57<00:46, 9.47it/s]
Training loss: 0.171, Change on Adj: -0.001: 56%|█████▌ | 561/1000 [00:57<00:46, 9.47it/s]
Training loss: 0.171, Change on Adj: -0.001: 56%|█████▌ | 562/1000 [00:57<00:46, 9.46it/s]
Training loss: 0.219, Change on Adj: -0.001: 56%|█████▌ | 562/1000 [00:57<00:46, 9.46it/s]
Training loss: 0.219, Change on Adj: -0.001: 56%|█████▋ | 563/1000 [00:57<00:46, 9.44it/s]
Training loss: 0.250, Change on Adj: -0.001: 56%|█████▋ | 563/1000 [00:57<00:46, 9.44it/s]
Training loss: 0.250, Change on Adj: -0.001: 56%|█████▋ | 564/1000 [00:57<00:46, 9.46it/s]
Training loss: 0.224, Change on Adj: -0.001: 56%|█████▋ | 564/1000 [00:57<00:46, 9.46it/s]
Training loss: 0.224, Change on Adj: -0.001: 56%|█████▋ | 565/1000 [00:57<00:46, 9.44it/s]
Training loss: 0.209, Change on Adj: -0.001: 56%|█████▋ | 565/1000 [00:57<00:46, 9.44it/s]
Training loss: 0.258, Change on Adj: -0.001: 56%|█████▋ | 565/1000 [00:57<00:46, 9.44it/s]
Training loss: 0.258, Change on Adj: -0.001: 57%|█████▋ | 567/1000 [00:57<00:41, 10.33it/s]
Training loss: 0.216, Change on Adj: -0.001: 57%|█████▋ | 567/1000 [00:57<00:41, 10.33it/s]
Training loss: 0.176, Change on Adj: -0.001: 57%|█████▋ | 567/1000 [00:57<00:41, 10.33it/s]
Training loss: 0.176, Change on Adj: -0.001: 57%|█████▋ | 569/1000 [00:57<00:39, 10.84it/s]
Training loss: 0.206, Change on Adj: -0.001: 57%|█████▋ | 569/1000 [00:58<00:39, 10.84it/s]
Training loss: 0.200, Change on Adj: -0.001: 57%|█████▋ | 569/1000 [00:58<00:39, 10.84it/s]
Training loss: 0.200, Change on Adj: -0.001: 57%|█████▋ | 571/1000 [00:58<00:38, 11.13it/s]
Training loss: 0.186, Change on Adj: -0.001: 57%|█████▋ | 571/1000 [00:58<00:38, 11.13it/s]
Training loss: 0.256, Change on Adj: -0.001: 57%|█████▋ | 571/1000 [00:58<00:38, 11.13it/s]
Training loss: 0.256, Change on Adj: -0.001: 57%|█████▋ | 573/1000 [00:58<00:37, 11.31it/s]
Training loss: 0.207, Change on Adj: -0.001: 57%|█████▋ | 573/1000 [00:58<00:37, 11.31it/s]
Training loss: 0.204, Change on Adj: -0.001: 57%|█████▋ | 573/1000 [00:58<00:37, 11.31it/s]
Training loss: 0.204, Change on Adj: -0.001: 57%|█████▊ | 575/1000 [00:58<00:37, 11.45it/s]
Training loss: 0.218, Change on Adj: -0.001: 57%|█████▊ | 575/1000 [00:58<00:37, 11.45it/s]
Training loss: 0.271, Change on Adj: -0.001: 57%|█████▊ | 575/1000 [00:58<00:37, 11.45it/s]
Training loss: 0.271, Change on Adj: -0.001: 58%|█████▊ | 577/1000 [00:58<00:36, 11.53it/s]
Training loss: 0.250, Change on Adj: -0.001: 58%|█████▊ | 577/1000 [00:58<00:36, 11.53it/s]
Training loss: 0.226, Change on Adj: -0.001: 58%|█████▊ | 577/1000 [00:58<00:36, 11.53it/s]
Training loss: 0.226, Change on Adj: -0.001: 58%|█████▊ | 579/1000 [00:58<00:36, 11.59it/s]
Training loss: 0.205, Change on Adj: -0.001: 58%|█████▊ | 579/1000 [00:58<00:36, 11.59it/s]
Training loss: 0.182, Change on Adj: -0.001: 58%|█████▊ | 579/1000 [00:58<00:36, 11.59it/s]
Training loss: 0.182, Change on Adj: -0.001: 58%|█████▊ | 581/1000 [00:58<00:36, 11.63it/s]
Training loss: 0.247, Change on Adj: -0.001: 58%|█████▊ | 581/1000 [00:59<00:36, 11.63it/s]
Training loss: 0.274, Change on Adj: -0.001: 58%|█████▊ | 581/1000 [00:59<00:36, 11.63it/s]
Training loss: 0.274, Change on Adj: -0.001: 58%|█████▊ | 583/1000 [00:59<00:35, 11.66it/s]
Training loss: 0.221, Change on Adj: -0.001: 58%|█████▊ | 583/1000 [00:59<00:35, 11.66it/s]
Training loss: 0.217, Change on Adj: -0.001: 58%|█████▊ | 583/1000 [00:59<00:35, 11.66it/s]
Training loss: 0.217, Change on Adj: -0.001: 58%|█████▊ | 585/1000 [00:59<00:35, 11.68it/s]
Training loss: 0.218, Change on Adj: -0.001: 58%|█████▊ | 585/1000 [00:59<00:35, 11.68it/s]
Training loss: 0.259, Change on Adj: -0.001: 58%|█████▊ | 585/1000 [00:59<00:35, 11.68it/s]
Training loss: 0.259, Change on Adj: -0.001: 59%|█████▊ | 587/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.191, Change on Adj: -0.001: 59%|█████▊ | 587/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.212, Change on Adj: -0.001: 59%|█████▊ | 587/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.212, Change on Adj: -0.001: 59%|█████▉ | 589/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.209, Change on Adj: -0.001: 59%|█████▉ | 589/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.258, Change on Adj: -0.001: 59%|█████▉ | 589/1000 [00:59<00:35, 11.69it/s]
Training loss: 0.258, Change on Adj: -0.001: 59%|█████▉ | 591/1000 [00:59<00:34, 11.71it/s]
Training loss: 0.252, Change on Adj: -0.001: 59%|█████▉ | 591/1000 [00:59<00:34, 11.71it/s]
Training loss: 0.207, Change on Adj: -0.001: 59%|█████▉ | 591/1000 [00:59<00:34, 11.71it/s]
Training loss: 0.207, Change on Adj: -0.001: 59%|█████▉ | 593/1000 [00:59<00:34, 11.71it/s]
Training loss: 0.186, Change on Adj: -0.001: 59%|█████▉ | 593/1000 [01:00<00:34, 11.71it/s]
Training loss: 0.201, Change on Adj: -0.001: 59%|█████▉ | 593/1000 [01:00<00:34, 11.71it/s]
Training loss: 0.201, Change on Adj: -0.001: 60%|█████▉ | 595/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.279, Change on Adj: -0.001: 60%|█████▉ | 595/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.226, Change on Adj: -0.001: 60%|█████▉ | 595/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.226, Change on Adj: -0.001: 60%|█████▉ | 597/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.210, Change on Adj: -0.001: 60%|█████▉ | 597/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.238, Change on Adj: -0.001: 60%|█████▉ | 597/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.238, Change on Adj: -0.001: 60%|█████▉ | 599/1000 [01:00<00:34, 11.71it/s]
Training loss: 0.248, Change on Adj: -0.001: 60%|█████▉ | 599/1000 [01:00<00:34, 11.71it/s]
Training loss: 0.237, Change on Adj: -0.001: 60%|█████▉ | 599/1000 [01:00<00:34, 11.71it/s]
Training loss: 0.237, Change on Adj: -0.001: 60%|██████ | 601/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.211, Change on Adj: -0.001: 60%|██████ | 601/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.252, Change on Adj: -0.001: 60%|██████ | 601/1000 [01:00<00:34, 11.72it/s]
Training loss: 0.252, Change on Adj: -0.001: 60%|██████ | 603/1000 [01:00<00:33, 11.72it/s]
Training loss: 0.193, Change on Adj: -0.001: 60%|██████ | 603/1000 [01:00<00:33, 11.72it/s]
Training loss: 0.233, Change on Adj: -0.001: 60%|██████ | 603/1000 [01:01<00:33, 11.72it/s]
Training loss: 0.233, Change on Adj: -0.001: 60%|██████ | 605/1000 [01:01<00:35, 11.16it/s]
Training loss: 0.208, Change on Adj: -0.001: 60%|██████ | 605/1000 [01:01<00:35, 11.16it/s]
Training loss: 0.266, Change on Adj: -0.001: 60%|██████ | 605/1000 [01:01<00:35, 11.16it/s]
Training loss: 0.266, Change on Adj: -0.001: 61%|██████ | 607/1000 [01:01<00:34, 11.33it/s]
Training loss: 0.222, Change on Adj: -0.001: 61%|██████ | 607/1000 [01:01<00:34, 11.33it/s]
Training loss: 0.184, Change on Adj: -0.001: 61%|██████ | 607/1000 [01:01<00:34, 11.33it/s]
Training loss: 0.184, Change on Adj: -0.001: 61%|██████ | 609/1000 [01:01<00:34, 11.44it/s]
Training loss: 0.225, Change on Adj: -0.001: 61%|██████ | 609/1000 [01:01<00:34, 11.44it/s]
Training loss: 0.215, Change on Adj: -0.001: 61%|██████ | 609/1000 [01:01<00:34, 11.44it/s]
Training loss: 0.215, Change on Adj: -0.001: 61%|██████ | 611/1000 [01:01<00:33, 11.53it/s]
Training loss: 0.226, Change on Adj: -0.001: 61%|██████ | 611/1000 [01:01<00:33, 11.53it/s]
Training loss: 0.216, Change on Adj: -0.001: 61%|██████ | 611/1000 [01:01<00:33, 11.53it/s]
Training loss: 0.216, Change on Adj: -0.001: 61%|██████▏ | 613/1000 [01:01<00:33, 11.59it/s]
Training loss: 0.250, Change on Adj: -0.001: 61%|██████▏ | 613/1000 [01:01<00:33, 11.59it/s]
Training loss: 0.253, Change on Adj: -0.001: 61%|██████▏ | 613/1000 [01:01<00:33, 11.59it/s]
Training loss: 0.253, Change on Adj: -0.001: 62%|██████▏ | 615/1000 [01:01<00:33, 11.62it/s]
Training loss: 0.190, Change on Adj: -0.001: 62%|██████▏ | 615/1000 [01:01<00:33, 11.62it/s]
Training loss: 0.268, Change on Adj: -0.001: 62%|██████▏ | 615/1000 [01:02<00:33, 11.62it/s]
Training loss: 0.268, Change on Adj: -0.001: 62%|██████▏ | 617/1000 [01:02<00:32, 11.64it/s]
Training loss: 0.259, Change on Adj: -0.001: 62%|██████▏ | 617/1000 [01:02<00:32, 11.64it/s]
Training loss: 0.236, Change on Adj: -0.001: 62%|██████▏ | 617/1000 [01:02<00:32, 11.64it/s]
Training loss: 0.236, Change on Adj: -0.001: 62%|██████▏ | 619/1000 [01:02<00:32, 11.67it/s]
Training loss: 0.219, Change on Adj: -0.001: 62%|██████▏ | 619/1000 [01:02<00:32, 11.67it/s]
Training loss: 0.252, Change on Adj: -0.001: 62%|██████▏ | 619/1000 [01:02<00:32, 11.67it/s]
Training loss: 0.252, Change on Adj: -0.001: 62%|██████▏ | 621/1000 [01:02<00:32, 11.69it/s]
Training loss: 0.248, Change on Adj: -0.001: 62%|██████▏ | 621/1000 [01:02<00:32, 11.69it/s]
Training loss: 0.198, Change on Adj: -0.001: 62%|██████▏ | 621/1000 [01:02<00:32, 11.69it/s]
Training loss: 0.198, Change on Adj: -0.001: 62%|██████▏ | 623/1000 [01:02<00:32, 11.70it/s]
Training loss: 0.269, Change on Adj: -0.001: 62%|██████▏ | 623/1000 [01:02<00:32, 11.70it/s]
Training loss: 0.220, Change on Adj: -0.001: 62%|██████▏ | 623/1000 [01:02<00:32, 11.70it/s]
Training loss: 0.220, Change on Adj: -0.001: 62%|██████▎ | 625/1000 [01:02<00:32, 11.72it/s]
Training loss: 0.197, Change on Adj: -0.001: 62%|██████▎ | 625/1000 [01:02<00:32, 11.72it/s]
Training loss: 0.266, Change on Adj: -0.001: 62%|██████▎ | 625/1000 [01:02<00:32, 11.72it/s]
Training loss: 0.266, Change on Adj: -0.001: 63%|██████▎ | 627/1000 [01:02<00:31, 11.72it/s]
Training loss: 0.237, Change on Adj: -0.001: 63%|██████▎ | 627/1000 [01:02<00:31, 11.72it/s]
Training loss: 0.212, Change on Adj: -0.001: 63%|██████▎ | 627/1000 [01:03<00:31, 11.72it/s]
Training loss: 0.212, Change on Adj: -0.001: 63%|██████▎ | 629/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.222, Change on Adj: -0.001: 63%|██████▎ | 629/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.221, Change on Adj: -0.001: 63%|██████▎ | 629/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.221, Change on Adj: -0.001: 63%|██████▎ | 631/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.207, Change on Adj: -0.001: 63%|██████▎ | 631/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.211, Change on Adj: -0.001: 63%|██████▎ | 631/1000 [01:03<00:31, 11.73it/s]
Training loss: 0.211, Change on Adj: -0.001: 63%|██████▎ | 633/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.259, Change on Adj: -0.001: 63%|██████▎ | 633/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.209, Change on Adj: -0.001: 63%|██████▎ | 633/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.209, Change on Adj: -0.001: 64%|██████▎ | 635/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.232, Change on Adj: -0.001: 64%|██████▎ | 635/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.216, Change on Adj: -0.001: 64%|██████▎ | 635/1000 [01:03<00:31, 11.74it/s]
Training loss: 0.216, Change on Adj: -0.001: 64%|██████▎ | 637/1000 [01:03<00:30, 11.75it/s]
Training loss: 0.211, Change on Adj: -0.001: 64%|██████▎ | 637/1000 [01:03<00:30, 11.75it/s]
Training loss: 0.247, Change on Adj: -0.001: 64%|██████▎ | 637/1000 [01:03<00:30, 11.75it/s]
Training loss: 0.247, Change on Adj: -0.001: 64%|██████▍ | 639/1000 [01:03<00:30, 11.75it/s]
Training loss: 0.203, Change on Adj: -0.001: 64%|██████▍ | 639/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.235, Change on Adj: -0.001: 64%|██████▍ | 639/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.235, Change on Adj: -0.001: 64%|██████▍ | 641/1000 [01:04<00:30, 11.74it/s]
Training loss: 0.229, Change on Adj: -0.001: 64%|██████▍ | 641/1000 [01:04<00:30, 11.74it/s]
Training loss: 0.241, Change on Adj: -0.001: 64%|██████▍ | 641/1000 [01:04<00:30, 11.74it/s]
Training loss: 0.241, Change on Adj: -0.001: 64%|██████▍ | 643/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.235, Change on Adj: -0.001: 64%|██████▍ | 643/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.210, Change on Adj: -0.001: 64%|██████▍ | 643/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.210, Change on Adj: -0.001: 64%|██████▍ | 645/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.214, Change on Adj: -0.001: 64%|██████▍ | 645/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.233, Change on Adj: -0.001: 64%|██████▍ | 645/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.233, Change on Adj: -0.001: 65%|██████▍ | 647/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.229, Change on Adj: -0.001: 65%|██████▍ | 647/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.246, Change on Adj: -0.001: 65%|██████▍ | 647/1000 [01:04<00:30, 11.75it/s]
Training loss: 0.246, Change on Adj: -0.001: 65%|██████▍ | 649/1000 [01:04<00:29, 11.75it/s]
Training loss: 0.207, Change on Adj: -0.001: 65%|██████▍ | 649/1000 [01:04<00:29, 11.75it/s]
Training loss: 0.215, Change on Adj: -0.001: 65%|██████▍ | 649/1000 [01:04<00:29, 11.75it/s]
Training loss: 0.215, Change on Adj: -0.001: 65%|██████▌ | 651/1000 [01:04<00:29, 11.76it/s]
Training loss: 0.233, Change on Adj: -0.001: 65%|██████▌ | 651/1000 [01:05<00:29, 11.76it/s]
Training loss: 0.261, Change on Adj: -0.001: 65%|██████▌ | 651/1000 [01:05<00:29, 11.76it/s]
Training loss: 0.261, Change on Adj: -0.001: 65%|██████▌ | 653/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.236, Change on Adj: -0.001: 65%|██████▌ | 653/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.271, Change on Adj: -0.001: 65%|██████▌ | 653/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.271, Change on Adj: -0.001: 66%|██████▌ | 655/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.209, Change on Adj: -0.001: 66%|██████▌ | 655/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.285, Change on Adj: -0.001: 66%|██████▌ | 655/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.285, Change on Adj: -0.001: 66%|██████▌ | 657/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.232, Change on Adj: -0.001: 66%|██████▌ | 657/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.259, Change on Adj: -0.001: 66%|██████▌ | 657/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.259, Change on Adj: -0.001: 66%|██████▌ | 659/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.187, Change on Adj: -0.001: 66%|██████▌ | 659/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.245, Change on Adj: -0.001: 66%|██████▌ | 659/1000 [01:05<00:29, 11.75it/s]
Training loss: 0.245, Change on Adj: -0.001: 66%|██████▌ | 661/1000 [01:05<00:28, 11.75it/s]
Training loss: 0.193, Change on Adj: -0.001: 66%|██████▌ | 661/1000 [01:05<00:28, 11.75it/s]
Training loss: 0.200, Change on Adj: -0.001: 66%|██████▌ | 661/1000 [01:05<00:28, 11.75it/s]
Training loss: 0.200, Change on Adj: -0.001: 66%|██████▋ | 663/1000 [01:05<00:28, 11.75it/s]
Training loss: 0.225, Change on Adj: -0.001: 66%|██████▋ | 663/1000 [01:06<00:28, 11.75it/s]
Training loss: 0.232, Change on Adj: -0.001: 66%|██████▋ | 663/1000 [01:06<00:28, 11.75it/s]
Training loss: 0.232, Change on Adj: -0.001: 66%|██████▋ | 665/1000 [01:06<00:28, 11.76it/s]
Training loss: 0.204, Change on Adj: -0.001: 66%|██████▋ | 665/1000 [01:06<00:28, 11.76it/s]
Training loss: 0.229, Change on Adj: -0.001: 66%|██████▋ | 665/1000 [01:06<00:28, 11.76it/s]
Training loss: 0.229, Change on Adj: -0.001: 67%|██████▋ | 667/1000 [01:06<00:28, 11.75it/s]
Training loss: 0.252, Change on Adj: -0.001: 67%|██████▋ | 667/1000 [01:06<00:28, 11.75it/s]
Training loss: 0.201, Change on Adj: -0.001: 67%|██████▋ | 667/1000 [01:06<00:28, 11.75it/s]
Training loss: 0.201, Change on Adj: -0.001: 67%|██████▋ | 669/1000 [01:06<00:28, 11.77it/s]
Training loss: 0.219, Change on Adj: -0.001: 67%|██████▋ | 669/1000 [01:06<00:28, 11.77it/s]
Training loss: 0.244, Change on Adj: -0.001: 67%|██████▋ | 669/1000 [01:06<00:28, 11.77it/s]
Training loss: 0.244, Change on Adj: -0.001: 67%|██████▋ | 671/1000 [01:06<00:27, 11.76it/s]
Training loss: 0.227, Change on Adj: -0.001: 67%|██████▋ | 671/1000 [01:06<00:27, 11.76it/s]
Training loss: 0.172, Change on Adj: -0.001: 67%|██████▋ | 671/1000 [01:06<00:27, 11.76it/s]
Training loss: 0.172, Change on Adj: -0.001: 67%|██████▋ | 673/1000 [01:06<00:27, 11.77it/s]
Training loss: 0.268, Change on Adj: -0.001: 67%|██████▋ | 673/1000 [01:06<00:27, 11.77it/s]
Training loss: 0.242, Change on Adj: -0.001: 67%|██████▋ | 673/1000 [01:06<00:27, 11.77it/s]
Training loss: 0.242, Change on Adj: -0.001: 68%|██████▊ | 675/1000 [01:06<00:27, 11.76it/s]
Training loss: 0.224, Change on Adj: -0.001: 68%|██████▊ | 675/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.211, Change on Adj: -0.001: 68%|██████▊ | 675/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.211, Change on Adj: -0.001: 68%|██████▊ | 677/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.164, Change on Adj: -0.001: 68%|██████▊ | 677/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.203, Change on Adj: -0.001: 68%|██████▊ | 677/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.203, Change on Adj: -0.001: 68%|██████▊ | 679/1000 [01:07<00:27, 11.77it/s]
Training loss: 0.247, Change on Adj: -0.001: 68%|██████▊ | 679/1000 [01:07<00:27, 11.77it/s]
Training loss: 0.294, Change on Adj: -0.001: 68%|██████▊ | 679/1000 [01:07<00:27, 11.77it/s]
Training loss: 0.294, Change on Adj: -0.001: 68%|██████▊ | 681/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.262, Change on Adj: -0.001: 68%|██████▊ | 681/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.251, Change on Adj: -0.001: 68%|██████▊ | 681/1000 [01:07<00:27, 11.76it/s]
Training loss: 0.251, Change on Adj: -0.001: 68%|██████▊ | 683/1000 [01:07<00:26, 11.76it/s]
Training loss: 0.232, Change on Adj: -0.001: 68%|██████▊ | 683/1000 [01:07<00:26, 11.76it/s]
Training loss: 0.254, Change on Adj: -0.001: 68%|██████▊ | 683/1000 [01:07<00:26, 11.76it/s]
Training loss: 0.254, Change on Adj: -0.001: 68%|██████▊ | 685/1000 [01:07<00:26, 11.76it/s]
Training loss: 0.191, Change on Adj: -0.000: 68%|██████▊ | 685/1000 [01:07<00:26, 11.76it/s]
Training loss: 0.203, Change on Adj: -0.000: 68%|██████▊ | 685/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.203, Change on Adj: -0.000: 69%|██████▊ | 687/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.207, Change on Adj: -0.001: 69%|██████▊ | 687/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.227, Change on Adj: -0.001: 69%|██████▊ | 687/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.227, Change on Adj: -0.001: 69%|██████▉ | 689/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.274, Change on Adj: -0.001: 69%|██████▉ | 689/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.220, Change on Adj: -0.001: 69%|██████▉ | 689/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.220, Change on Adj: -0.001: 69%|██████▉ | 691/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.262, Change on Adj: -0.001: 69%|██████▉ | 691/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.255, Change on Adj: -0.001: 69%|██████▉ | 691/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.255, Change on Adj: -0.001: 69%|██████▉ | 693/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.233, Change on Adj: -0.001: 69%|██████▉ | 693/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.228, Change on Adj: -0.001: 69%|██████▉ | 693/1000 [01:08<00:26, 11.76it/s]
Training loss: 0.228, Change on Adj: -0.001: 70%|██████▉ | 695/1000 [01:08<00:25, 11.75it/s]
Training loss: 0.227, Change on Adj: -0.001: 70%|██████▉ | 695/1000 [01:08<00:25, 11.75it/s]
Training loss: 0.225, Change on Adj: -0.001: 70%|██████▉ | 695/1000 [01:08<00:25, 11.75it/s]
Training loss: 0.225, Change on Adj: -0.001: 70%|██████▉ | 697/1000 [01:08<00:25, 11.76it/s]
Training loss: 0.239, Change on Adj: -0.000: 70%|██████▉ | 697/1000 [01:08<00:25, 11.76it/s]
Training loss: 0.239, Change on Adj: -0.000: 70%|██████▉ | 697/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.239, Change on Adj: -0.000: 70%|██████▉ | 699/1000 [01:09<00:25, 11.75it/s]
Training loss: 0.194, Change on Adj: -0.000: 70%|██████▉ | 699/1000 [01:09<00:25, 11.75it/s]
Training loss: 0.221, Change on Adj: -0.000: 70%|██████▉ | 699/1000 [01:09<00:25, 11.75it/s]
Training loss: 0.221, Change on Adj: -0.000: 70%|███████ | 701/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.198, Change on Adj: -0.000: 70%|███████ | 701/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.204, Change on Adj: -0.000: 70%|███████ | 701/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.204, Change on Adj: -0.000: 70%|███████ | 703/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.217, Change on Adj: -0.000: 70%|███████ | 703/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.206, Change on Adj: -0.000: 70%|███████ | 703/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.206, Change on Adj: -0.000: 70%|███████ | 705/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.211, Change on Adj: -0.000: 70%|███████ | 705/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.227, Change on Adj: -0.000: 70%|███████ | 705/1000 [01:09<00:25, 11.76it/s]
Training loss: 0.227, Change on Adj: -0.000: 71%|███████ | 707/1000 [01:09<00:24, 11.76it/s]
Training loss: 0.225, Change on Adj: -0.000: 71%|███████ | 707/1000 [01:09<00:24, 11.76it/s]
Training loss: 0.223, Change on Adj: -0.001: 71%|███████ | 707/1000 [01:09<00:24, 11.76it/s]
Training loss: 0.223, Change on Adj: -0.001: 71%|███████ | 709/1000 [01:09<00:24, 11.77it/s]
Training loss: 0.233, Change on Adj: -0.001: 71%|███████ | 709/1000 [01:09<00:24, 11.77it/s]
Training loss: 0.263, Change on Adj: -0.001: 71%|███████ | 709/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.263, Change on Adj: -0.001: 71%|███████ | 711/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.260, Change on Adj: -0.001: 71%|███████ | 711/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.264, Change on Adj: -0.001: 71%|███████ | 711/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.264, Change on Adj: -0.001: 71%|███████▏ | 713/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.238, Change on Adj: -0.001: 71%|███████▏ | 713/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.258, Change on Adj: -0.001: 71%|███████▏ | 713/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.258, Change on Adj: -0.001: 72%|███████▏ | 715/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.223, Change on Adj: -0.001: 72%|███████▏ | 715/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.237, Change on Adj: -0.001: 72%|███████▏ | 715/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.237, Change on Adj: -0.001: 72%|███████▏ | 717/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.259, Change on Adj: -0.001: 72%|███████▏ | 717/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.254, Change on Adj: -0.000: 72%|███████▏ | 717/1000 [01:10<00:24, 11.77it/s]
Training loss: 0.254, Change on Adj: -0.000: 72%|███████▏ | 719/1000 [01:10<00:23, 11.77it/s]
Training loss: 0.239, Change on Adj: -0.000: 72%|███████▏ | 719/1000 [01:10<00:23, 11.77it/s]
Training loss: 0.245, Change on Adj: -0.000: 72%|███████▏ | 719/1000 [01:10<00:23, 11.77it/s]
Training loss: 0.245, Change on Adj: -0.000: 72%|███████▏ | 721/1000 [01:10<00:23, 11.77it/s]
Training loss: 0.254, Change on Adj: -0.000: 72%|███████▏ | 721/1000 [01:10<00:23, 11.77it/s]
Training loss: 0.250, Change on Adj: -0.000: 72%|███████▏ | 721/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.250, Change on Adj: -0.000: 72%|███████▏ | 723/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.230, Change on Adj: -0.000: 72%|███████▏ | 723/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.226, Change on Adj: -0.000: 72%|███████▏ | 723/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.226, Change on Adj: -0.000: 72%|███████▎ | 725/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.237, Change on Adj: -0.000: 72%|███████▎ | 725/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.251, Change on Adj: -0.000: 72%|███████▎ | 725/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.251, Change on Adj: -0.000: 73%|███████▎ | 727/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.208, Change on Adj: -0.000: 73%|███████▎ | 727/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.211, Change on Adj: -0.000: 73%|███████▎ | 727/1000 [01:11<00:23, 11.76it/s]
Training loss: 0.211, Change on Adj: -0.000: 73%|███████▎ | 729/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.206, Change on Adj: -0.000: 73%|███████▎ | 729/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.324, Change on Adj: -0.001: 73%|███████▎ | 729/1000 [01:11<00:23, 11.77it/s]
Training loss: 0.324, Change on Adj: -0.001: 73%|███████▎ | 731/1000 [01:11<00:22, 11.76it/s]
Training loss: 0.210, Change on Adj: -0.001: 73%|███████▎ | 731/1000 [01:11<00:22, 11.76it/s]
Training loss: 0.163, Change on Adj: -0.001: 73%|███████▎ | 731/1000 [01:11<00:22, 11.76it/s]
Training loss: 0.163, Change on Adj: -0.001: 73%|███████▎ | 733/1000 [01:11<00:22, 11.77it/s]
Training loss: 0.227, Change on Adj: -0.001: 73%|███████▎ | 733/1000 [01:12<00:22, 11.77it/s]
Training loss: 0.247, Change on Adj: -0.001: 73%|███████▎ | 733/1000 [01:12<00:22, 11.77it/s]
Training loss: 0.247, Change on Adj: -0.001: 74%|███████▎ | 735/1000 [01:12<00:22, 11.77it/s]
Training loss: 0.216, Change on Adj: -0.001: 74%|███████▎ | 735/1000 [01:12<00:22, 11.77it/s]
Training loss: 0.227, Change on Adj: -0.001: 74%|███████▎ | 735/1000 [01:12<00:22, 11.77it/s]
Training loss: 0.227, Change on Adj: -0.001: 74%|███████▎ | 737/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.246, Change on Adj: -0.001: 74%|███████▎ | 737/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.248, Change on Adj: -0.001: 74%|███████▎ | 737/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.248, Change on Adj: -0.001: 74%|███████▍ | 739/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.270, Change on Adj: -0.001: 74%|███████▍ | 739/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.264, Change on Adj: -0.000: 74%|███████▍ | 739/1000 [01:12<00:22, 11.78it/s]
Training loss: 0.264, Change on Adj: -0.000: 74%|███████▍ | 741/1000 [01:12<00:21, 11.78it/s]
Training loss: 0.204, Change on Adj: -0.000: 74%|███████▍ | 741/1000 [01:12<00:21, 11.78it/s]
Training loss: 0.200, Change on Adj: -0.000: 74%|███████▍ | 741/1000 [01:12<00:21, 11.78it/s]
Training loss: 0.200, Change on Adj: -0.000: 74%|███████▍ | 743/1000 [01:12<00:21, 11.77it/s]
Training loss: 0.234, Change on Adj: -0.000: 74%|███████▍ | 743/1000 [01:12<00:21, 11.77it/s]
Training loss: 0.204, Change on Adj: -0.000: 74%|███████▍ | 743/1000 [01:12<00:21, 11.77it/s]
Training loss: 0.204, Change on Adj: -0.000: 74%|███████▍ | 745/1000 [01:12<00:21, 11.77it/s]
Training loss: 0.212, Change on Adj: -0.000: 74%|███████▍ | 745/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.231, Change on Adj: -0.000: 74%|███████▍ | 745/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.231, Change on Adj: -0.000: 75%|███████▍ | 747/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.212, Change on Adj: -0.001: 75%|███████▍ | 747/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.191, Change on Adj: -0.000: 75%|███████▍ | 747/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.191, Change on Adj: -0.000: 75%|███████▍ | 749/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.278, Change on Adj: -0.000: 75%|███████▍ | 749/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.194, Change on Adj: -0.001: 75%|███████▍ | 749/1000 [01:13<00:21, 11.77it/s]
Training loss: 0.194, Change on Adj: -0.001: 75%|███████▌ | 751/1000 [01:13<00:21, 11.78it/s]
Training loss: 0.191, Change on Adj: -0.001: 75%|███████▌ | 751/1000 [01:13<00:21, 11.78it/s]
Training loss: 0.210, Change on Adj: -0.001: 75%|███████▌ | 751/1000 [01:13<00:21, 11.78it/s]
Training loss: 0.210, Change on Adj: -0.001: 75%|███████▌ | 753/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.248, Change on Adj: -0.001: 75%|███████▌ | 753/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.264, Change on Adj: -0.001: 75%|███████▌ | 753/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.264, Change on Adj: -0.001: 76%|███████▌ | 755/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.252, Change on Adj: -0.001: 76%|███████▌ | 755/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.272, Change on Adj: -0.001: 76%|███████▌ | 755/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.272, Change on Adj: -0.001: 76%|███████▌ | 757/1000 [01:13<00:20, 11.78it/s]
Training loss: 0.243, Change on Adj: -0.001: 76%|███████▌ | 757/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.248, Change on Adj: -0.001: 76%|███████▌ | 757/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.248, Change on Adj: -0.001: 76%|███████▌ | 759/1000 [01:14<00:20, 11.77it/s]
Training loss: 0.208, Change on Adj: -0.001: 76%|███████▌ | 759/1000 [01:14<00:20, 11.77it/s]
Training loss: 0.213, Change on Adj: -0.000: 76%|███████▌ | 759/1000 [01:14<00:20, 11.77it/s]
Training loss: 0.213, Change on Adj: -0.000: 76%|███████▌ | 761/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.261, Change on Adj: -0.000: 76%|███████▌ | 761/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.230, Change on Adj: -0.000: 76%|███████▌ | 761/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.230, Change on Adj: -0.000: 76%|███████▋ | 763/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.223, Change on Adj: -0.000: 76%|███████▋ | 763/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.216, Change on Adj: -0.000: 76%|███████▋ | 763/1000 [01:14<00:20, 11.78it/s]
Training loss: 0.216, Change on Adj: -0.000: 76%|███████▋ | 765/1000 [01:14<00:19, 11.78it/s]
Training loss: 0.201, Change on Adj: -0.000: 76%|███████▋ | 765/1000 [01:14<00:19, 11.78it/s]
Training loss: 0.261, Change on Adj: -0.000: 76%|███████▋ | 765/1000 [01:14<00:19, 11.78it/s]
Training loss: 0.261, Change on Adj: -0.000: 77%|███████▋ | 767/1000 [01:14<00:19, 11.79it/s]
Training loss: 0.252, Change on Adj: -0.000: 77%|███████▋ | 767/1000 [01:14<00:19, 11.79it/s]
Training loss: 0.237, Change on Adj: -0.000: 77%|███████▋ | 767/1000 [01:14<00:19, 11.79it/s]
Training loss: 0.237, Change on Adj: -0.000: 77%|███████▋ | 769/1000 [01:14<00:19, 11.78it/s]
Training loss: 0.190, Change on Adj: -0.000: 77%|███████▋ | 769/1000 [01:15<00:19, 11.78it/s]
Training loss: 0.270, Change on Adj: -0.000: 77%|███████▋ | 769/1000 [01:15<00:19, 11.78it/s]
Training loss: 0.270, Change on Adj: -0.000: 77%|███████▋ | 771/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.242, Change on Adj: -0.001: 77%|███████▋ | 771/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.173, Change on Adj: -0.001: 77%|███████▋ | 771/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.173, Change on Adj: -0.001: 77%|███████▋ | 773/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.172, Change on Adj: -0.001: 77%|███████▋ | 773/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.209, Change on Adj: -0.001: 77%|███████▋ | 773/1000 [01:15<00:19, 11.79it/s]
Training loss: 0.209, Change on Adj: -0.001: 78%|███████▊ | 775/1000 [01:15<00:19, 11.77it/s]
Training loss: 0.236, Change on Adj: -0.001: 78%|███████▊ | 775/1000 [01:15<00:19, 11.77it/s]
Training loss: 0.212, Change on Adj: -0.001: 78%|███████▊ | 775/1000 [01:15<00:19, 11.77it/s]
Training loss: 0.212, Change on Adj: -0.001: 78%|███████▊ | 777/1000 [01:15<00:18, 11.78it/s]
Training loss: 0.217, Change on Adj: -0.001: 78%|███████▊ | 777/1000 [01:15<00:18, 11.78it/s]
Training loss: 0.181, Change on Adj: -0.001: 78%|███████▊ | 777/1000 [01:15<00:18, 11.78it/s]
Training loss: 0.181, Change on Adj: -0.001: 78%|███████▊ | 779/1000 [01:15<00:18, 11.77it/s]
Training loss: 0.196, Change on Adj: -0.001: 78%|███████▊ | 779/1000 [01:15<00:18, 11.77it/s]
Training loss: 0.216, Change on Adj: -0.001: 78%|███████▊ | 779/1000 [01:15<00:18, 11.77it/s]
Training loss: 0.216, Change on Adj: -0.001: 78%|███████▊ | 781/1000 [01:15<00:18, 11.78it/s]
Training loss: 0.236, Change on Adj: -0.001: 78%|███████▊ | 781/1000 [01:16<00:18, 11.78it/s]
Training loss: 0.171, Change on Adj: -0.001: 78%|███████▊ | 781/1000 [01:16<00:18, 11.78it/s]
Training loss: 0.171, Change on Adj: -0.001: 78%|███████▊ | 783/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.246, Change on Adj: -0.001: 78%|███████▊ | 783/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.227, Change on Adj: -0.001: 78%|███████▊ | 783/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.227, Change on Adj: -0.001: 78%|███████▊ | 785/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.243, Change on Adj: -0.001: 78%|███████▊ | 785/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.264, Change on Adj: -0.001: 78%|███████▊ | 785/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.264, Change on Adj: -0.001: 79%|███████▊ | 787/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.275, Change on Adj: -0.001: 79%|███████▊ | 787/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.233, Change on Adj: -0.001: 79%|███████▊ | 787/1000 [01:16<00:18, 11.79it/s]
Training loss: 0.233, Change on Adj: -0.001: 79%|███████▉ | 789/1000 [01:16<00:17, 11.79it/s]
Training loss: 0.253, Change on Adj: -0.001: 79%|███████▉ | 789/1000 [01:16<00:17, 11.79it/s]
Training loss: 0.249, Change on Adj: -0.001: 79%|███████▉ | 789/1000 [01:16<00:17, 11.79it/s]
Training loss: 0.249, Change on Adj: -0.001: 79%|███████▉ | 791/1000 [01:16<00:17, 11.79it/s]
Training loss: 0.251, Change on Adj: -0.001: 79%|███████▉ | 791/1000 [01:16<00:17, 11.79it/s]
Training loss: 0.220, Change on Adj: -0.001: 79%|███████▉ | 791/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.220, Change on Adj: -0.001: 79%|███████▉ | 793/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.224, Change on Adj: -0.000: 79%|███████▉ | 793/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.188, Change on Adj: -0.000: 79%|███████▉ | 793/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.188, Change on Adj: -0.000: 80%|███████▉ | 795/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.252, Change on Adj: -0.000: 80%|███████▉ | 795/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.234, Change on Adj: -0.000: 80%|███████▉ | 795/1000 [01:17<00:17, 11.79it/s]
Training loss: 0.234, Change on Adj: -0.000: 80%|███████▉ | 797/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.255, Change on Adj: -0.000: 80%|███████▉ | 797/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.215, Change on Adj: -0.000: 80%|███████▉ | 797/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.215, Change on Adj: -0.000: 80%|███████▉ | 799/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.196, Change on Adj: -0.000: 80%|███████▉ | 799/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.238, Change on Adj: -0.000: 80%|███████▉ | 799/1000 [01:17<00:17, 11.80it/s]
Training loss: 0.238, Change on Adj: -0.000: 80%|████████ | 801/1000 [01:17<00:16, 11.80it/s]
Training loss: 0.236, Change on Adj: -0.001: 80%|████████ | 801/1000 [01:17<00:16, 11.80it/s]
Training loss: 0.235, Change on Adj: -0.001: 80%|████████ | 801/1000 [01:17<00:16, 11.80it/s]
Training loss: 0.235, Change on Adj: -0.001: 80%|████████ | 803/1000 [01:17<00:16, 11.80it/s]
Training loss: 0.246, Change on Adj: -0.001: 80%|████████ | 803/1000 [01:17<00:16, 11.80it/s]
Training loss: 0.182, Change on Adj: -0.001: 80%|████████ | 803/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.182, Change on Adj: -0.001: 80%|████████ | 805/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.189, Change on Adj: -0.001: 80%|████████ | 805/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.284, Change on Adj: -0.001: 80%|████████ | 805/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.284, Change on Adj: -0.001: 81%|████████ | 807/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.201, Change on Adj: -0.001: 81%|████████ | 807/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.230, Change on Adj: -0.001: 81%|████████ | 807/1000 [01:18<00:16, 11.80it/s]
Training loss: 0.230, Change on Adj: -0.001: 81%|████████ | 809/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.244, Change on Adj: -0.001: 81%|████████ | 809/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.208, Change on Adj: -0.001: 81%|████████ | 809/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.208, Change on Adj: -0.001: 81%|████████ | 811/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.189, Change on Adj: -0.001: 81%|████████ | 811/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.228, Change on Adj: -0.001: 81%|████████ | 811/1000 [01:18<00:16, 11.79it/s]
Training loss: 0.228, Change on Adj: -0.001: 81%|████████▏ | 813/1000 [01:18<00:15, 11.80it/s]
Training loss: 0.240, Change on Adj: -0.000: 81%|████████▏ | 813/1000 [01:18<00:15, 11.80it/s]
Training loss: 0.235, Change on Adj: -0.001: 81%|████████▏ | 813/1000 [01:18<00:15, 11.80it/s]
Training loss: 0.235, Change on Adj: -0.001: 82%|████████▏ | 815/1000 [01:18<00:15, 11.80it/s]
Training loss: 0.221, Change on Adj: -0.001: 82%|████████▏ | 815/1000 [01:18<00:15, 11.80it/s]
Training loss: 0.217, Change on Adj: -0.000: 82%|████████▏ | 815/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.217, Change on Adj: -0.000: 82%|████████▏ | 817/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.196, Change on Adj: -0.000: 82%|████████▏ | 817/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.249, Change on Adj: -0.000: 82%|████████▏ | 817/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.249, Change on Adj: -0.000: 82%|████████▏ | 819/1000 [01:19<00:15, 11.79it/s]
Training loss: 0.204, Change on Adj: -0.000: 82%|████████▏ | 819/1000 [01:19<00:15, 11.79it/s]
Training loss: 0.216, Change on Adj: -0.000: 82%|████████▏ | 819/1000 [01:19<00:15, 11.79it/s]
Training loss: 0.216, Change on Adj: -0.000: 82%|████████▏ | 821/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.276, Change on Adj: -0.000: 82%|████████▏ | 821/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.276, Change on Adj: -0.000: 82%|████████▏ | 821/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.276, Change on Adj: -0.000: 82%|████████▏ | 823/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.224, Change on Adj: -0.000: 82%|████████▏ | 823/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.213, Change on Adj: -0.000: 82%|████████▏ | 823/1000 [01:19<00:15, 11.80it/s]
Training loss: 0.213, Change on Adj: -0.000: 82%|████████▎ | 825/1000 [01:19<00:14, 11.79it/s]
Training loss: 0.187, Change on Adj: -0.000: 82%|████████▎ | 825/1000 [01:19<00:14, 11.79it/s]
Training loss: 0.238, Change on Adj: -0.000: 82%|████████▎ | 825/1000 [01:19<00:14, 11.79it/s]
Training loss: 0.238, Change on Adj: -0.000: 83%|████████▎ | 827/1000 [01:19<00:14, 11.80it/s]
Training loss: 0.206, Change on Adj: -0.000: 83%|████████▎ | 827/1000 [01:19<00:14, 11.80it/s]
Training loss: 0.243, Change on Adj: -0.000: 83%|████████▎ | 827/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.243, Change on Adj: -0.000: 83%|████████▎ | 829/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.222, Change on Adj: -0.001: 83%|████████▎ | 829/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.222, Change on Adj: -0.000: 83%|████████▎ | 829/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.222, Change on Adj: -0.000: 83%|████████▎ | 831/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.246, Change on Adj: -0.000: 83%|████████▎ | 831/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.264, Change on Adj: -0.000: 83%|████████▎ | 831/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.264, Change on Adj: -0.000: 83%|████████▎ | 833/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.211, Change on Adj: -0.000: 83%|████████▎ | 833/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.252, Change on Adj: -0.000: 83%|████████▎ | 833/1000 [01:20<00:14, 11.80it/s]
Training loss: 0.252, Change on Adj: -0.000: 84%|████████▎ | 835/1000 [01:20<00:13, 11.80it/s]
Training loss: 0.215, Change on Adj: -0.000: 84%|████████▎ | 835/1000 [01:20<00:13, 11.80it/s]
Training loss: 0.196, Change on Adj: -0.000: 84%|████████▎ | 835/1000 [01:20<00:13, 11.80it/s]
Training loss: 0.196, Change on Adj: -0.000: 84%|████████▎ | 837/1000 [01:20<00:13, 11.81it/s]
Training loss: 0.241, Change on Adj: -0.000: 84%|████████▎ | 837/1000 [01:20<00:13, 11.81it/s]
Training loss: 0.184, Change on Adj: -0.000: 84%|████████▎ | 837/1000 [01:20<00:13, 11.81it/s]
Training loss: 0.184, Change on Adj: -0.000: 84%|████████▍ | 839/1000 [01:20<00:13, 11.80it/s]
Training loss: 0.192, Change on Adj: -0.000: 84%|████████▍ | 839/1000 [01:20<00:13, 11.80it/s]
Training loss: 0.208, Change on Adj: -0.000: 84%|████████▍ | 839/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.208, Change on Adj: -0.000: 84%|████████▍ | 841/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.239, Change on Adj: -0.000: 84%|████████▍ | 841/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.265, Change on Adj: -0.000: 84%|████████▍ | 841/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.265, Change on Adj: -0.000: 84%|████████▍ | 843/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.171, Change on Adj: -0.000: 84%|████████▍ | 843/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.281, Change on Adj: -0.000: 84%|████████▍ | 843/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.281, Change on Adj: -0.000: 84%|████████▍ | 845/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.284, Change on Adj: -0.000: 84%|████████▍ | 845/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.253, Change on Adj: -0.000: 84%|████████▍ | 845/1000 [01:21<00:13, 11.80it/s]
Training loss: 0.253, Change on Adj: -0.000: 85%|████████▍ | 847/1000 [01:21<00:12, 11.80it/s]
Training loss: 0.207, Change on Adj: -0.000: 85%|████████▍ | 847/1000 [01:21<00:12, 11.80it/s]
Training loss: 0.219, Change on Adj: -0.000: 85%|████████▍ | 847/1000 [01:21<00:12, 11.80it/s]
Training loss: 0.219, Change on Adj: -0.000: 85%|████████▍ | 849/1000 [01:21<00:12, 11.81it/s]
Training loss: 0.228, Change on Adj: -0.000: 85%|████████▍ | 849/1000 [01:21<00:12, 11.81it/s]
Training loss: 0.201, Change on Adj: -0.000: 85%|████████▍ | 849/1000 [01:21<00:12, 11.81it/s]
Training loss: 0.201, Change on Adj: -0.000: 85%|████████▌ | 851/1000 [01:21<00:12, 11.82it/s]
Training loss: 0.199, Change on Adj: -0.000: 85%|████████▌ | 851/1000 [01:22<00:12, 11.82it/s]
Training loss: 0.245, Change on Adj: -0.000: 85%|████████▌ | 851/1000 [01:22<00:12, 11.82it/s]
Training loss: 0.245, Change on Adj: -0.000: 85%|████████▌ | 853/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.199, Change on Adj: -0.000: 85%|████████▌ | 853/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.215, Change on Adj: -0.000: 85%|████████▌ | 853/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.215, Change on Adj: -0.000: 86%|████████▌ | 855/1000 [01:22<00:12, 11.80it/s]
Training loss: 0.180, Change on Adj: -0.000: 86%|████████▌ | 855/1000 [01:22<00:12, 11.80it/s]
Training loss: 0.258, Change on Adj: -0.000: 86%|████████▌ | 855/1000 [01:22<00:12, 11.80it/s]
Training loss: 0.258, Change on Adj: -0.000: 86%|████████▌ | 857/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.297, Change on Adj: -0.000: 86%|████████▌ | 857/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.194, Change on Adj: -0.000: 86%|████████▌ | 857/1000 [01:22<00:12, 11.81it/s]
Training loss: 0.194, Change on Adj: -0.000: 86%|████████▌ | 859/1000 [01:22<00:11, 11.80it/s]
Training loss: 0.203, Change on Adj: -0.000: 86%|████████▌ | 859/1000 [01:22<00:11, 11.80it/s]
Training loss: 0.231, Change on Adj: -0.000: 86%|████████▌ | 859/1000 [01:22<00:11, 11.80it/s]
Training loss: 0.231, Change on Adj: -0.000: 86%|████████▌ | 861/1000 [01:22<00:11, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.000: 86%|████████▌ | 861/1000 [01:22<00:11, 11.81it/s]
Training loss: 0.255, Change on Adj: -0.000: 86%|████████▌ | 861/1000 [01:22<00:11, 11.81it/s]
Training loss: 0.255, Change on Adj: -0.000: 86%|████████▋ | 863/1000 [01:22<00:11, 11.81it/s]
Training loss: 0.218, Change on Adj: -0.000: 86%|████████▋ | 863/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.266, Change on Adj: -0.000: 86%|████████▋ | 863/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.266, Change on Adj: -0.000: 86%|████████▋ | 865/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.159, Change on Adj: -0.000: 86%|████████▋ | 865/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.182, Change on Adj: -0.000: 86%|████████▋ | 865/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.182, Change on Adj: -0.000: 87%|████████▋ | 867/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.193, Change on Adj: -0.000: 87%|████████▋ | 867/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.231, Change on Adj: -0.000: 87%|████████▋ | 867/1000 [01:23<00:11, 11.81it/s]
Training loss: 0.231, Change on Adj: -0.000: 87%|████████▋ | 869/1000 [01:23<00:11, 11.80it/s]
Training loss: 0.288, Change on Adj: -0.000: 87%|████████▋ | 869/1000 [01:23<00:11, 11.80it/s]
Training loss: 0.248, Change on Adj: -0.000: 87%|████████▋ | 869/1000 [01:23<00:11, 11.80it/s]
Training loss: 0.248, Change on Adj: -0.000: 87%|████████▋ | 871/1000 [01:23<00:10, 11.81it/s]
Training loss: 0.265, Change on Adj: -0.000: 87%|████████▋ | 871/1000 [01:23<00:10, 11.81it/s]
Training loss: 0.228, Change on Adj: -0.000: 87%|████████▋ | 871/1000 [01:23<00:10, 11.81it/s]
Training loss: 0.228, Change on Adj: -0.000: 87%|████████▋ | 873/1000 [01:23<00:10, 11.82it/s]
Training loss: 0.216, Change on Adj: -0.000: 87%|████████▋ | 873/1000 [01:23<00:10, 11.82it/s]
Training loss: 0.243, Change on Adj: -0.000: 87%|████████▋ | 873/1000 [01:23<00:10, 11.82it/s]
Training loss: 0.243, Change on Adj: -0.000: 88%|████████▊ | 875/1000 [01:23<00:10, 11.81it/s]
Training loss: 0.225, Change on Adj: -0.000: 88%|████████▊ | 875/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.174, Change on Adj: -0.000: 88%|████████▊ | 875/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.174, Change on Adj: -0.000: 88%|████████▊ | 877/1000 [01:24<00:10, 11.82it/s]
Training loss: 0.244, Change on Adj: -0.000: 88%|████████▊ | 877/1000 [01:24<00:10, 11.82it/s]
Training loss: 0.192, Change on Adj: -0.000: 88%|████████▊ | 877/1000 [01:24<00:10, 11.82it/s]
Training loss: 0.192, Change on Adj: -0.000: 88%|████████▊ | 879/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.234, Change on Adj: -0.000: 88%|████████▊ | 879/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.202, Change on Adj: -0.000: 88%|████████▊ | 879/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.202, Change on Adj: -0.000: 88%|████████▊ | 881/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.167, Change on Adj: -0.000: 88%|████████▊ | 881/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.000: 88%|████████▊ | 881/1000 [01:24<00:10, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.000: 88%|████████▊ | 883/1000 [01:24<00:09, 11.81it/s]
Training loss: 0.203, Change on Adj: -0.000: 88%|████████▊ | 883/1000 [01:24<00:09, 11.81it/s]
Training loss: 0.264, Change on Adj: -0.000: 88%|████████▊ | 883/1000 [01:24<00:09, 11.81it/s]
Training loss: 0.264, Change on Adj: -0.000: 88%|████████▊ | 885/1000 [01:24<00:09, 11.80it/s]
Training loss: 0.247, Change on Adj: -0.001: 88%|████████▊ | 885/1000 [01:24<00:09, 11.80it/s]
Training loss: 0.286, Change on Adj: -0.000: 88%|████████▊ | 885/1000 [01:24<00:09, 11.80it/s]
Training loss: 0.286, Change on Adj: -0.000: 89%|████████▊ | 887/1000 [01:24<00:09, 11.80it/s]
Training loss: 0.242, Change on Adj: -0.000: 89%|████████▊ | 887/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.247, Change on Adj: -0.000: 89%|████████▊ | 887/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.247, Change on Adj: -0.000: 89%|████████▉ | 889/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.248, Change on Adj: -0.000: 89%|████████▉ | 889/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.248, Change on Adj: -0.000: 89%|████████▉ | 889/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.248, Change on Adj: -0.000: 89%|████████▉ | 891/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.245, Change on Adj: -0.000: 89%|████████▉ | 891/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.174, Change on Adj: -0.000: 89%|████████▉ | 891/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.174, Change on Adj: -0.000: 89%|████████▉ | 893/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.286, Change on Adj: -0.000: 89%|████████▉ | 893/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.270, Change on Adj: 0.000: 89%|████████▉ | 893/1000 [01:25<00:09, 11.80it/s]
Training loss: 0.270, Change on Adj: 0.000: 90%|████████▉ | 895/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.230, Change on Adj: 0.000: 90%|████████▉ | 895/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.194, Change on Adj: 0.000: 90%|████████▉ | 895/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.194, Change on Adj: 0.000: 90%|████████▉ | 897/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.257, Change on Adj: 0.000: 90%|████████▉ | 897/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.000: 90%|████████▉ | 897/1000 [01:25<00:08, 11.81it/s]
Training loss: 0.245, Change on Adj: -0.000: 90%|████████▉ | 899/1000 [01:25<00:08, 11.82it/s]
Training loss: 0.234, Change on Adj: -0.000: 90%|████████▉ | 899/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.269, Change on Adj: -0.000: 90%|████████▉ | 899/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.269, Change on Adj: -0.000: 90%|█████████ | 901/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.190, Change on Adj: -0.000: 90%|█████████ | 901/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.207, Change on Adj: -0.000: 90%|█████████ | 901/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.207, Change on Adj: -0.000: 90%|█████████ | 903/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.216, Change on Adj: -0.000: 90%|█████████ | 903/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 90%|█████████ | 903/1000 [01:26<00:08, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 90%|█████████ | 905/1000 [01:26<00:08, 11.81it/s]
Training loss: 0.259, Change on Adj: -0.000: 90%|█████████ | 905/1000 [01:26<00:08, 11.81it/s]
Training loss: 0.265, Change on Adj: -0.000: 90%|█████████ | 905/1000 [01:26<00:08, 11.81it/s]
Training loss: 0.265, Change on Adj: -0.000: 91%|█████████ | 907/1000 [01:26<00:07, 11.82it/s]
Training loss: 0.237, Change on Adj: -0.000: 91%|█████████ | 907/1000 [01:26<00:07, 11.82it/s]
Training loss: 0.190, Change on Adj: -0.000: 91%|█████████ | 907/1000 [01:26<00:07, 11.82it/s]
Training loss: 0.190, Change on Adj: -0.000: 91%|█████████ | 909/1000 [01:26<00:07, 11.81it/s]
Training loss: 0.242, Change on Adj: -0.000: 91%|█████████ | 909/1000 [01:26<00:07, 11.81it/s]
Training loss: 0.191, Change on Adj: -0.000: 91%|█████████ | 909/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.191, Change on Adj: -0.000: 91%|█████████ | 911/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.199, Change on Adj: -0.000: 91%|█████████ | 911/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.262, Change on Adj: -0.000: 91%|█████████ | 911/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.262, Change on Adj: -0.000: 91%|█████████▏| 913/1000 [01:27<00:07, 11.80it/s]
Training loss: 0.218, Change on Adj: -0.000: 91%|█████████▏| 913/1000 [01:27<00:07, 11.80it/s]
Training loss: 0.265, Change on Adj: -0.000: 91%|█████████▏| 913/1000 [01:27<00:07, 11.80it/s]
Training loss: 0.265, Change on Adj: -0.000: 92%|█████████▏| 915/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.208, Change on Adj: -0.000: 92%|█████████▏| 915/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.187, Change on Adj: -0.000: 92%|█████████▏| 915/1000 [01:27<00:07, 11.81it/s]
Training loss: 0.187, Change on Adj: -0.000: 92%|█████████▏| 917/1000 [01:27<00:07, 11.82it/s]
Training loss: 0.186, Change on Adj: -0.000: 92%|█████████▏| 917/1000 [01:27<00:07, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 92%|█████████▏| 917/1000 [01:27<00:07, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 92%|█████████▏| 919/1000 [01:27<00:06, 11.81it/s]
Training loss: 0.234, Change on Adj: -0.000: 92%|█████████▏| 919/1000 [01:27<00:06, 11.81it/s]
Training loss: 0.231, Change on Adj: -0.000: 92%|█████████▏| 919/1000 [01:27<00:06, 11.81it/s]
Training loss: 0.231, Change on Adj: -0.000: 92%|█████████▏| 921/1000 [01:27<00:06, 11.82it/s]
Training loss: 0.256, Change on Adj: -0.000: 92%|█████████▏| 921/1000 [01:27<00:06, 11.82it/s]
Training loss: 0.291, Change on Adj: -0.000: 92%|█████████▏| 921/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.291, Change on Adj: -0.000: 92%|█████████▏| 923/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.222, Change on Adj: -0.000: 92%|█████████▏| 923/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.236, Change on Adj: -0.000: 92%|█████████▏| 923/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.236, Change on Adj: -0.000: 92%|█████████▎| 925/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.190, Change on Adj: -0.000: 92%|█████████▎| 925/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.230, Change on Adj: -0.000: 92%|█████████▎| 925/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.230, Change on Adj: -0.000: 93%|█████████▎| 927/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.237, Change on Adj: 0.000: 93%|█████████▎| 927/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.199, Change on Adj: 0.000: 93%|█████████▎| 927/1000 [01:28<00:06, 11.82it/s]
Training loss: 0.199, Change on Adj: 0.000: 93%|█████████▎| 929/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.184, Change on Adj: -0.000: 93%|█████████▎| 929/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.228, Change on Adj: -0.000: 93%|█████████▎| 929/1000 [01:28<00:06, 11.81it/s]
Training loss: 0.228, Change on Adj: -0.000: 93%|█████████▎| 931/1000 [01:28<00:05, 11.82it/s]
Training loss: 0.246, Change on Adj: -0.000: 93%|█████████▎| 931/1000 [01:28<00:05, 11.82it/s]
Training loss: 0.281, Change on Adj: -0.000: 93%|█████████▎| 931/1000 [01:28<00:05, 11.82it/s]
Training loss: 0.281, Change on Adj: -0.000: 93%|█████████▎| 933/1000 [01:28<00:05, 11.81it/s]
Training loss: 0.217, Change on Adj: -0.000: 93%|█████████▎| 933/1000 [01:28<00:05, 11.81it/s]
Training loss: 0.210, Change on Adj: -0.000: 93%|█████████▎| 933/1000 [01:29<00:05, 11.81it/s]
Training loss: 0.210, Change on Adj: -0.000: 94%|█████████▎| 935/1000 [01:29<00:05, 11.78it/s]
Training loss: 0.238, Change on Adj: -0.000: 94%|█████████▎| 935/1000 [01:29<00:05, 11.78it/s]
Training loss: 0.277, Change on Adj: -0.000: 94%|█████████▎| 935/1000 [01:29<00:05, 11.78it/s]
Training loss: 0.277, Change on Adj: -0.000: 94%|█████████▎| 937/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.242, Change on Adj: -0.000: 94%|█████████▎| 937/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.235, Change on Adj: -0.000: 94%|█████████▎| 937/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.235, Change on Adj: -0.000: 94%|█████████▍| 939/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.226, Change on Adj: -0.000: 94%|█████████▍| 939/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.232, Change on Adj: -0.000: 94%|█████████▍| 939/1000 [01:29<00:05, 11.79it/s]
Training loss: 0.232, Change on Adj: -0.000: 94%|█████████▍| 941/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.208, Change on Adj: -0.000: 94%|█████████▍| 941/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.224, Change on Adj: -0.000: 94%|█████████▍| 941/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.224, Change on Adj: -0.000: 94%|█████████▍| 943/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.288, Change on Adj: -0.000: 94%|█████████▍| 943/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.247, Change on Adj: 0.000: 94%|█████████▍| 943/1000 [01:29<00:04, 11.80it/s]
Training loss: 0.247, Change on Adj: 0.000: 94%|█████████▍| 945/1000 [01:29<00:04, 11.81it/s]
Training loss: 0.236, Change on Adj: 0.000: 94%|█████████▍| 945/1000 [01:29<00:04, 11.81it/s]
Training loss: 0.209, Change on Adj: 0.000: 94%|█████████▍| 945/1000 [01:30<00:04, 11.81it/s]
Training loss: 0.209, Change on Adj: 0.000: 95%|█████████▍| 947/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.218, Change on Adj: -0.000: 95%|█████████▍| 947/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.226, Change on Adj: -0.000: 95%|█████████▍| 947/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.226, Change on Adj: -0.000: 95%|█████████▍| 949/1000 [01:30<00:04, 11.81it/s]
Training loss: 0.272, Change on Adj: -0.000: 95%|█████████▍| 949/1000 [01:30<00:04, 11.81it/s]
Training loss: 0.210, Change on Adj: -0.000: 95%|█████████▍| 949/1000 [01:30<00:04, 11.81it/s]
Training loss: 0.210, Change on Adj: -0.000: 95%|█████████▌| 951/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.200, Change on Adj: -0.000: 95%|█████████▌| 951/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.221, Change on Adj: -0.000: 95%|█████████▌| 951/1000 [01:30<00:04, 11.82it/s]
Training loss: 0.221, Change on Adj: -0.000: 95%|█████████▌| 953/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.171, Change on Adj: -0.000: 95%|█████████▌| 953/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.220, Change on Adj: -0.000: 95%|█████████▌| 953/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.220, Change on Adj: -0.000: 96%|█████████▌| 955/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.192, Change on Adj: -0.000: 96%|█████████▌| 955/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.220, Change on Adj: -0.001: 96%|█████████▌| 955/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.220, Change on Adj: -0.001: 96%|█████████▌| 957/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.230, Change on Adj: -0.000: 96%|█████████▌| 957/1000 [01:30<00:03, 11.82it/s]
Training loss: 0.309, Change on Adj: -0.000: 96%|█████████▌| 957/1000 [01:31<00:03, 11.82it/s]
Training loss: 0.309, Change on Adj: -0.000: 96%|█████████▌| 959/1000 [01:31<00:03, 11.82it/s]
Training loss: 0.239, Change on Adj: -0.000: 96%|█████████▌| 959/1000 [01:31<00:03, 11.82it/s]
Training loss: 0.257, Change on Adj: -0.000: 96%|█████████▌| 959/1000 [01:31<00:03, 11.82it/s]
Training loss: 0.257, Change on Adj: -0.000: 96%|█████████▌| 961/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.200, Change on Adj: -0.000: 96%|█████████▌| 961/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.208, Change on Adj: -0.000: 96%|█████████▌| 961/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.208, Change on Adj: -0.000: 96%|█████████▋| 963/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.213, Change on Adj: -0.000: 96%|█████████▋| 963/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.260, Change on Adj: 0.000: 96%|█████████▋| 963/1000 [01:31<00:03, 11.83it/s]
Training loss: 0.260, Change on Adj: 0.000: 96%|█████████▋| 965/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.232, Change on Adj: 0.000: 96%|█████████▋| 965/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.241, Change on Adj: 0.000: 96%|█████████▋| 965/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.241, Change on Adj: 0.000: 97%|█████████▋| 967/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.260, Change on Adj: 0.000: 97%|█████████▋| 967/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.209, Change on Adj: -0.000: 97%|█████████▋| 967/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.209, Change on Adj: -0.000: 97%|█████████▋| 969/1000 [01:31<00:02, 11.83it/s]
Training loss: 0.214, Change on Adj: -0.000: 97%|█████████▋| 969/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.257, Change on Adj: -0.000: 97%|█████████▋| 969/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.257, Change on Adj: -0.000: 97%|█████████▋| 971/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.195, Change on Adj: -0.000: 97%|█████████▋| 971/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.233, Change on Adj: -0.000: 97%|█████████▋| 971/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.233, Change on Adj: -0.000: 97%|█████████▋| 973/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.260, Change on Adj: -0.000: 97%|█████████▋| 973/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.220, Change on Adj: -0.000: 97%|█████████▋| 973/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.220, Change on Adj: -0.000: 98%|█████████▊| 975/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.205, Change on Adj: -0.000: 98%|█████████▊| 975/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.210, Change on Adj: -0.000: 98%|█████████▊| 975/1000 [01:32<00:02, 11.83it/s]
Training loss: 0.210, Change on Adj: -0.000: 98%|█████████▊| 977/1000 [01:32<00:01, 11.83it/s]
Training loss: 0.239, Change on Adj: -0.000: 98%|█████████▊| 977/1000 [01:32<00:01, 11.83it/s]
Training loss: 0.215, Change on Adj: -0.000: 98%|█████████▊| 977/1000 [01:32<00:01, 11.83it/s]
Training loss: 0.215, Change on Adj: -0.000: 98%|█████████▊| 979/1000 [01:32<00:01, 11.82it/s]
Training loss: 0.270, Change on Adj: -0.000: 98%|█████████▊| 979/1000 [01:32<00:01, 11.82it/s]
Training loss: 0.198, Change on Adj: -0.000: 98%|█████████▊| 979/1000 [01:32<00:01, 11.82it/s]
Training loss: 0.198, Change on Adj: -0.000: 98%|█████████▊| 981/1000 [01:32<00:01, 11.83it/s]
Training loss: 0.256, Change on Adj: -0.000: 98%|█████████▊| 981/1000 [01:33<00:01, 11.83it/s]
Training loss: 0.218, Change on Adj: -0.000: 98%|█████████▊| 981/1000 [01:33<00:01, 11.83it/s]
Training loss: 0.218, Change on Adj: -0.000: 98%|█████████▊| 983/1000 [01:33<00:01, 11.83it/s]
Training loss: 0.199, Change on Adj: -0.000: 98%|█████████▊| 983/1000 [01:33<00:01, 11.83it/s]
Training loss: 0.230, Change on Adj: -0.000: 98%|█████████▊| 983/1000 [01:33<00:01, 11.83it/s]
Training loss: 0.230, Change on Adj: -0.000: 98%|█████████▊| 985/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.205, Change on Adj: -0.000: 98%|█████████▊| 985/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.258, Change on Adj: -0.000: 98%|█████████▊| 985/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.258, Change on Adj: -0.000: 99%|█████████▊| 987/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.205, Change on Adj: -0.000: 99%|█████████▊| 987/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 99%|█████████▊| 987/1000 [01:33<00:01, 11.82it/s]
Training loss: 0.233, Change on Adj: -0.000: 99%|█████████▉| 989/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.204, Change on Adj: -0.000: 99%|█████████▉| 989/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.250, Change on Adj: -0.000: 99%|█████████▉| 989/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.250, Change on Adj: -0.000: 99%|█████████▉| 991/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.258, Change on Adj: -0.000: 99%|█████████▉| 991/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.175, Change on Adj: -0.000: 99%|█████████▉| 991/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.175, Change on Adj: -0.000: 99%|█████████▉| 993/1000 [01:33<00:00, 11.83it/s]
Training loss: 0.255, Change on Adj: -0.000: 99%|█████████▉| 993/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.242, Change on Adj: -0.000: 99%|█████████▉| 993/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.242, Change on Adj: -0.000: 100%|█████████▉| 995/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.217, Change on Adj: -0.000: 100%|█████████▉| 995/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.198, Change on Adj: -0.000: 100%|█████████▉| 995/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.198, Change on Adj: -0.000: 100%|█████████▉| 997/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.226, Change on Adj: -0.000: 100%|█████████▉| 997/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.245, Change on Adj: -0.000: 100%|█████████▉| 997/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.245, Change on Adj: -0.000: 100%|█████████▉| 999/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.196, Change on Adj: -0.000: 100%|█████████▉| 999/1000 [01:34<00:00, 11.83it/s]
Training loss: 0.196, Change on Adj: -0.000: 100%|██████████| 1000/1000 [01:34<00:00, 10.58it/s]
[flashscenic] Adjacency matrix: (12282, 12282)
[flashscenic] Step 2/5: Filtering to known TFs...
[flashscenic] 1376 TFs found, sparsified at threshold=1.5
[flashscenic] Step 3/5: Creating and filtering modules...
[flashscenic] 5 module types: ['top50', 'pct75', 'top5pertarget', 'top10pertarget', 'top50pertarget']
[flashscenic] top50: 741 TF modules
[flashscenic] pct75: 427 TF modules
[flashscenic] top5pertarget: 218 TF modules
[flashscenic] top10pertarget: 317 TF modules
[flashscenic] top50pertarget: 594 TF modules
[flashscenic] 2297 total modules across all types
[flashscenic] Deduplicated: 2297 → 2296 unique modules
[flashscenic] Step 4/5: Running cisTarget pruning (2 databases)...
Loaded database 'hg38_500bp_up_100bp_down_full_tx_v10_clust.genes_vs_motifs.rankings': 5876 motifs × 27015 genes
Loaded database 'hg38_10kbp_up_10kbp_down_full_tx_v10_clust.genes_vs_motifs.rankings': 5876 motifs × 27090 genes
Loaded 2 databases
Reading annotation file: flashscenic_data/motifs-v10nr_clust-nr.hgnc-m0.001-o0.0.tbl
Column indices: motif_id=0, gene_name=5, similarity_qvalue=6, ortho_identity=9, description=12
Read 253096 rows, 253096 passed filters
Loaded 177084 motif annotations (27072 unique motifs, 1605 unique TFs)
Loaded motif annotations (filter_for_annotation=True)
Pruning with database 1/2: hg38_500bp_up_100bp_down_full_tx_v10_clust.genes_vs_motifs.rankings
Pruning with database 2/2: hg38_10kbp_up_10kbp_down_full_tx_v10_clust.genes_vs_motifs.rankings
Total regulons before merging: 1761
Total regulons after merging: 1107
Total regulons after TF merging: 192
[flashscenic] 192 regulons after pruning
[flashscenic] Step 5/5: Computing AUCell scores...
[flashscenic] Done! 192 regulons, AUCell scores shape: (33506, 192)
Found 192 regulons
AUCell scores shape: (33506, 192)
Example regulons: ['E2F2(+)', 'RUNX3(+)', 'TAL1(+)', 'JUN(+)', 'NFIA(+)', 'RFX5(+)', 'ZBTB7B(+)', 'ELK4(+)', 'ATF3(+)', 'FOSL2(+)']
# Store AUCell scores in AnnData for downstream scanpy functions
adata.obsm['X_aucell'] = auc_scores
# Create a dedicated AnnData for regulon activities
# (scanpy's rank_genes_groups, dotplot, etc. expect features in .X)
adata_regulons = anndata.AnnData(
X=auc_scores,
obs=adata.obs.copy(),
)
adata_regulons.var_names = pd.Index(regulon_names)
3. AUCell UMAP vs Gene Expression UMAP#
A key advantage of regulon-based analysis: cells are represented by their regulatory program activity rather than raw gene expression. This often produces cleaner separation of cell types and naturally reduces batch effects, since regulon activity is more biologically stable across experimental conditions.
# Compute neighbors and UMAP on AUCell space
sc.pp.neighbors(adata, use_rep='X_aucell', n_neighbors=15,
metric='correlation', key_added='aucell_neighbors')
sc.tl.umap(adata, neighbors_key='aucell_neighbors')
adata.obsm['X_umap_aucell'] = adata.obsm['X_umap'].copy()
fig, axes = plt.subplots(1, 2, figsize=(14, 5))
# PCA UMAP
adata.obsm['X_umap'] = adata.obsm['X_umap_pca']
sc.pl.umap(adata, color='final_annotation', ax=axes[0], show=False,
title='Gene Expression PCA', legend_loc='none', frameon=False)
# AUCell UMAP
adata.obsm['X_umap'] = adata.obsm['X_umap_aucell']
sc.pl.umap(adata, color='final_annotation', ax=axes[1], show=False,
title='AUCell Regulon Activity', frameon=False,
legend_loc='right margin')
plt.tight_layout()
plt.show()
4. Batch Effect Comparison#
Batch effects are a major challenge in scRNA-seq. Here we show that the AUCell-based UMAP naturally mixes batches while separating cell types – without any explicit batch correction.
fig, axes = plt.subplots(2, 2, figsize=(16, 14))
# Row 1: Colored by cell type
adata.obsm['X_umap'] = adata.obsm['X_umap_pca']
sc.pl.umap(adata, color='final_annotation', ax=axes[0, 0], show=False,
title='PCA UMAP - Cell Types', legend_loc='none', frameon=False)
adata.obsm['X_umap'] = adata.obsm['X_umap_aucell']
sc.pl.umap(adata, color='final_annotation', ax=axes[0, 1], show=False,
title='AUCell UMAP - Cell Types', legend_loc='none', frameon=False)
# Row 2: Colored by batch
adata.obsm['X_umap'] = adata.obsm['X_umap_pca']
sc.pl.umap(adata, color='batch', ax=axes[1, 0], show=False,
title='PCA UMAP - Batch', legend_loc='none', frameon=False)
adata.obsm['X_umap'] = adata.obsm['X_umap_aucell']
sc.pl.umap(adata, color='batch', ax=axes[1, 1], show=False,
title='AUCell UMAP - Batch', legend_loc='none', frameon=False)
plt.tight_layout()
plt.show()
5. Regulon Activity Heatmap#
A z-score normalized heatmap of regulon activity reveals which transcription factor programs define each cell type. We select the top 5 most active regulons per cell type.
# Mean AUCell score per cell type
auc_df = pd.DataFrame(auc_scores, columns=regulon_names, index=adata.obs_names)
auc_df['cell_type'] = adata.obs['final_annotation'].values
mean_auc = auc_df.groupby('cell_type')[regulon_names].mean()
# Select top 5 regulons per cell type
top_n = 5
top_regulons = set()
for ct in mean_auc.index:
top = mean_auc.loc[ct].nlargest(top_n).index.tolist()
top_regulons.update(top)
top_regulons = sorted(top_regulons)
print(f"Selected {len(top_regulons)} unique regulons across {mean_auc.shape[0]} cell types")
Selected 14 unique regulons across 16 cell types
# Z-score normalize
mean_sub = mean_auc[top_regulons]
zscore_df = (mean_sub - mean_sub.mean()) / mean_sub.std()
# Simplify regulon names for display (e.g., "PAX5(+)" -> "PAX5")
display_names = [r.split('(')[0] if '(' in r else r for r in zscore_df.columns]
zscore_display = zscore_df.copy()
zscore_display.columns = display_names
g = sns.clustermap(
zscore_display,
cmap='RdBu_r', center=0, vmin=-2, vmax=3,
figsize=(max(14, len(top_regulons) * 0.4), max(8, len(mean_auc) * 0.5)),
row_cluster=True, col_cluster=True,
xticklabels=True, yticklabels=True,
linewidths=0.5, dendrogram_ratio=0.1,
cbar_kws={'label': 'Z-score'},
)
g.ax_heatmap.set_xlabel('Regulons')
g.ax_heatmap.set_ylabel('Cell Types')
plt.show()
6. Regulon Specificity Scores (RSS)#
The Regulon Specificity Score (RSS) quantifies how specific each regulon is to each cell type using Jensen-Shannon divergence (Suo et al. 2018). An RSS close to 1 means the regulon is exclusively active in that cell type.
flashscenic provides regulon_specificity_scores() for this.
rss_result = fs.regulon_specificity_scores(
auc_scores, adata.obs['final_annotation'].values, regulon_names
)
rss = pd.DataFrame(
rss_result['rss'],
index=rss_result['cell_types'],
columns=rss_result['regulon_names'],
)
print(f"RSS matrix: {rss.shape[0]} cell types x {rss.shape[1]} regulons")
RSS matrix: 16 cell types x 192 regulons
def plot_rss(rss_df, cell_type, ax, top_n=5):
"""Plot RSS for a single cell type, highlighting top regulons."""
scores = rss_df.loc[cell_type].sort_values(ascending=False)
ax.plot(range(len(scores)), scores.values, '.', color='lightgray',
markersize=3, alpha=0.5)
top = scores.head(top_n)
top_idx = [scores.index.tolist().index(r) for r in top.index]
ax.scatter(top_idx, top.values, color='#e74c3c', s=30, zorder=5)
for idx, name in zip(top_idx, top.index):
tf_name = name.split('(')[0] if '(' in name else name
ax.annotate(tf_name, (idx, scores[name]), fontsize=7,
textcoords='offset points', xytext=(5, 2))
ax.set_title(cell_type, fontsize=9)
ax.set_ylabel('RSS')
ax.set_xticks([])
cell_types = sorted(adata.obs['final_annotation'].unique())
n_cols = 4
n_rows = (len(cell_types) + n_cols - 1) // n_cols
fig, axes = plt.subplots(n_rows, n_cols, figsize=(16, n_rows * 3))
axes_flat = axes.flatten()
for i, ct in enumerate(cell_types):
plot_rss(rss, ct, axes_flat[i], top_n=5)
for j in range(i + 1, len(axes_flat)):
axes_flat[j].set_visible(False)
plt.suptitle('Regulon Specificity Scores by Cell Type', fontsize=14, y=1.02)
plt.tight_layout()
plt.show()
7. Known Biology Validation#
A critical sanity check: do the discovered regulons match known biology? The immune dataset contains well-characterized cell types with known master regulators:
TF |
Expected Cell Type |
Role |
|---|---|---|
PAX5 |
B cells |
B cell lineage commitment |
SPI1 (PU.1) |
Monocytes |
Myeloid differentiation |
GATA1 |
Erythrocytes |
Erythroid commitment |
TBX21 (T-bet) |
NK cells, CD8+ T cells |
Th1/cytotoxic response |
EOMES |
NK cells, NKT cells |
Effector lymphocyte function |
IRF4 |
Plasma cells, DCs |
Plasma cell differentiation |
IRF8 |
pDCs, Monocytes |
DC and monocyte development |
CEBPA |
Monocytes, HSPCs |
Myeloid progenitor differentiation |
TCF7 |
T cells |
T cell identity/stemness |
# Check which known TFs were found as regulons
known_tfs = ['PAX5', 'SPI1', 'GATA1', 'TBX21', 'EOMES', 'IRF4', 'IRF8',
'CEBPA', 'TCF7']
tf_to_regulon = {}
for tf in known_tfs:
matches = [r for r in regulon_names if r.startswith(tf + '(') or r == tf]
if matches:
tf_to_regulon[tf] = matches[0]
found_tfs = list(tf_to_regulon.keys())
print(f"Found {len(found_tfs)}/{len(known_tfs)} known marker TFs as regulons:")
for tf, reg in tf_to_regulon.items():
idx = regulon_names.index(reg)
n_genes = len(regulons[idx]['genes'])
print(f" {tf} -> {reg} ({n_genes} target genes)")
Found 7/9 known marker TFs as regulons:
SPI1 -> SPI1(+) (1468 target genes)
GATA1 -> GATA1(+) (315 target genes)
TBX21 -> TBX21(+) (122 target genes)
EOMES -> EOMES(+) (112 target genes)
IRF4 -> IRF4(+) (23 target genes)
IRF8 -> IRF8(+) (835 target genes)
CEBPA -> CEBPA(+) (408 target genes)
# Feature plots: regulon activity on the AUCell UMAP
adata.obsm['X_umap'] = adata.obsm['X_umap_aucell']
for tf, reg in tf_to_regulon.items():
idx = regulon_names.index(reg)
adata.obs[f'regulon_{tf}'] = auc_scores[:, idx]
n_tfs = len(found_tfs)
n_cols = min(5, n_tfs)
n_rows = (n_tfs + n_cols - 1) // n_cols
fig, axes = plt.subplots(n_rows, n_cols, figsize=(4 * n_cols, 3.5 * n_rows))
axes_flat = np.array(axes).flatten()
for i, tf in enumerate(found_tfs):
sc.pl.umap(adata, color=f'regulon_{tf}', ax=axes_flat[i], show=False,
title=f'{tf}', frameon=False, color_map='YlOrRd', vmin=0)
for j in range(i + 1, len(axes_flat)):
axes_flat[j].set_visible(False)
plt.suptitle('Known TF Regulon Activity on AUCell UMAP', fontsize=14, y=1.02)
plt.tight_layout()
plt.show()
# Violin plots: regulon activity by cell type
fig, axes = plt.subplots(len(found_tfs), 1,
figsize=(12, 3 * len(found_tfs)))
if len(found_tfs) == 1:
axes = [axes]
for i, tf in enumerate(found_tfs):
sc.pl.violin(adata, keys=f'regulon_{tf}', groupby='final_annotation',
ax=axes[i], show=False, rotation=45, stripplot=False)
axes[i].set_title(f'{tf} regulon activity by cell type')
axes[i].set_xlabel('')
plt.tight_layout()
plt.show()
8. Binary Regulon Activity#
AUCell scores are continuous, but it’s often useful to binarize them – classifying each regulon as “on” or “off” in each cell. We threshold at the 75th percentile of each regulon’s AUCell score distribution. This lets us ask: what fraction of cells in each type have a given regulon active?
# Binarize AUCell scores
threshold_pct = 75
thresholds = np.percentile(auc_scores, threshold_pct, axis=0)
binary_auc = (auc_scores > thresholds[np.newaxis, :]).astype(np.float32)
print(f"Mean fraction active per regulon: {binary_auc.mean(axis=0).mean():.3f}")
Mean fraction active per regulon: 0.232
# Fraction-active heatmap per cell type
binary_df = pd.DataFrame(binary_auc, columns=regulon_names, index=adata.obs_names)
binary_df['cell_type'] = adata.obs['final_annotation'].values
pct_active = binary_df.groupby('cell_type')[regulon_names].mean()
# Subset to the top regulons from section 5
pct_sub = pct_active[top_regulons]
pct_display = pct_sub.copy()
pct_display.columns = [r.split('(')[0] if '(' in r else r for r in pct_sub.columns]
g = sns.clustermap(
pct_display,
cmap='YlOrRd', vmin=0, vmax=1,
figsize=(max(14, len(top_regulons) * 0.4), max(8, len(cell_types) * 0.5)),
row_cluster=True, col_cluster=True,
xticklabels=True, yticklabels=True,
linewidths=0.5,
cbar_kws={'label': 'Fraction Active'},
)
plt.show()
9. Dot Plot#
Scanpy’s dot plot combines two dimensions: dot size shows the fraction of cells with a regulon active, and dot color shows the mean activity level. We use the top 3 regulons per cell type by RSS.
# Select top 3 regulons per cell type by RSS
top_rss_regulons = []
for ct in sorted(rss.index):
top3 = rss.loc[ct].nlargest(3).index.tolist()
top_rss_regulons.extend(top3)
top_rss_regulons = list(dict.fromkeys(top_rss_regulons)) # deduplicate
# Create subset with simplified names
adata_dot = adata_regulons[:, top_rss_regulons].copy()
adata_dot.var_names = pd.Index(
[r.split('(')[0] if '(' in r else r for r in top_rss_regulons]
)
sc.pl.dotplot(
adata_dot,
var_names=adata_dot.var_names.tolist(),
groupby='final_annotation',
standard_scale='var',
title='Regulon Activity (top 3 RSS per cell type)',
figsize=(max(12, len(top_rss_regulons) * 0.4), 6),
)
plt.show()
10. Differential Regulon Activity#
We can use scanpy’s rank_genes_groups on the AUCell scores to find regulons that are statistically differentially active between cell types. This is analogous to differential expression analysis, but for regulatory programs.
# Find differentially active regulons per cell type
sc.tl.rank_genes_groups(adata_regulons, groupby='final_annotation',
method='wilcoxon', use_raw=False)
sc.pl.rank_genes_groups(adata_regulons, n_genes=10, sharey=False,
fontsize=8, figsize=(16, 4))
plt.show()
# Focused comparison: CD4+ T cells vs CD8+ T cells
sc.tl.rank_genes_groups(adata_regulons, groupby='final_annotation',
groups=['CD4+ T cells'], reference='CD8+ T cells',
method='wilcoxon', use_raw=False, key_added='cd4_vs_cd8')
result_df = sc.get.rank_genes_groups_df(
adata_regulons, group='CD4+ T cells', key='cd4_vs_cd8'
)
result_df['tf'] = result_df['names'].apply(
lambda x: x.split('(')[0] if '(' in x else x
)
print("Top 10 regulons enriched in CD4+ T cells vs CD8+ T cells:")
result_df.head(10)[['tf', 'names', 'scores', 'pvals_adj', 'logfoldchanges']]
Top 10 regulons enriched in CD4+ T cells vs CD8+ T cells:
| tf | names | scores | pvals_adj | logfoldchanges | |
|---|---|---|---|---|---|
| 0 | MAF | MAF(+) | 23.021828 | 5.410760e-115 | 0.457621 |
| 1 | FLI1 | FLI1(+) | 19.457815 | 6.005364e-83 | 0.143068 |
| 2 | ZNF467 | ZNF467(+) | 16.820351 | 2.374377e-62 | 0.220426 |
| 3 | IRF2 | IRF2(+) | 16.530512 | 2.832068e-60 | 0.149056 |
| 4 | NFKB2 | NFKB2(+) | 16.312944 | 9.581768e-59 | 0.228235 |
| 5 | HOXB2 | HOXB2(+) | 14.680341 | 9.730587e-48 | 0.153528 |
| 6 | BCL6 | BCL6(+) | 14.352610 | 1.094430e-45 | 0.126546 |
| 7 | IRF5 | IRF5(+) | 14.165979 | 1.428279e-44 | 0.132637 |
| 8 | NFATC2 | NFATC2(+) | 13.659850 | 1.539023e-41 | 0.164607 |
| 9 | RELA | RELA(+) | 13.239033 | 4.358992e-39 | 0.175988 |
11. TF-Target Network Visualization#
For any regulon of interest, we can visualize its target genes as a network graph. The TF is shown at the center with edges to target genes, sized by expression correlation.
import networkx as nx
def plot_regulon_network(regulon_dict, exp_matrix, gene_names, ax=None,
max_targets=30):
"""Plot a TF-target network for a single regulon."""
tf = regulon_dict['tf']
targets = regulon_dict['genes'][:max_targets]
G = nx.DiGraph()
G.add_node(tf, node_type='tf')
tf_expr = None
if tf in gene_names:
tf_expr = exp_matrix[:, gene_names.index(tf)]
for target in targets:
corr = 0.1
if tf_expr is not None and target in gene_names:
c = np.corrcoef(tf_expr, exp_matrix[:, gene_names.index(target)])[0, 1]
corr = max(0.05, c) if not np.isnan(c) else 0.1
G.add_node(target, node_type='target')
G.add_edge(tf, target, weight=corr)
if ax is None:
fig, ax = plt.subplots(figsize=(10, 10))
pos = nx.spring_layout(G, k=2, seed=42)
tf_nodes = [n for n, d in G.nodes(data=True) if d.get('node_type') == 'tf']
target_nodes = [n for n, d in G.nodes(data=True)
if d.get('node_type') == 'target']
nx.draw_networkx_nodes(G, pos, nodelist=tf_nodes, node_color='#e74c3c',
node_size=800, ax=ax)
nx.draw_networkx_nodes(G, pos, nodelist=target_nodes, node_color='#3498db',
node_size=200, alpha=0.7, ax=ax)
edges = G.edges(data=True)
widths = [d['weight'] * 3 for _, _, d in edges]
nx.draw_networkx_edges(G, pos, width=widths, alpha=0.5,
edge_color='gray', arrows=True, ax=ax)
nx.draw_networkx_labels(G, pos, font_size=7, ax=ax)
n_total = len(regulon_dict['genes'])
shown = min(max_targets, n_total)
ax.set_title(f"{tf} regulon ({n_total} targets, showing {shown})")
ax.axis('off')
# Plot networks for a few interesting regulons
interesting_tfs = [tf for tf in ['PAX5', 'SPI1', 'GATA1'] if tf in tf_to_regulon]
if interesting_tfs:
fig, axes = plt.subplots(1, len(interesting_tfs),
figsize=(8 * len(interesting_tfs), 8))
if len(interesting_tfs) == 1:
axes = [axes]
for i, tf in enumerate(interesting_tfs):
reg_name = tf_to_regulon[tf]
reg_idx = regulon_names.index(reg_name)
plot_regulon_network(regulons[reg_idx], exp_matrix, gene_names,
ax=axes[i])
plt.tight_layout()
plt.show()
12. Regulon Co-Activity Analysis#
Regulons don’t act in isolation – TFs form regulatory circuits and cascades. A correlation heatmap of regulon activities reveals which TF programs are co-activated (suggesting cooperative regulation) or mutually exclusive (suggesting alternative cell fate programs).
# Pearson correlation of regulon AUCell scores (top regulons)
auc_top = auc_df[top_regulons]
corr_matrix = auc_top.corr()
# Simplified names
display = {r: r.split('(')[0] if '(' in r else r for r in top_regulons}
corr_display = corr_matrix.copy()
corr_display.index = [display[r] for r in corr_display.index]
corr_display.columns = [display[r] for r in corr_display.columns]
g = sns.clustermap(
corr_display,
cmap='RdBu_r', center=0, vmin=-0.5, vmax=0.5,
figsize=(max(12, len(top_regulons) * 0.35),
max(12, len(top_regulons) * 0.35)),
linewidths=0.5,
xticklabels=True, yticklabels=True,
dendrogram_ratio=0.1,
cbar_kws={'label': 'Pearson r'},
)
plt.show()
Summary#
Analysis |
Key Insight |
|---|---|
AUCell UMAP |
Regulon-based embedding naturally handles batch effects |
Regulon heatmap |
Z-score visualization reveals cell-type-defining TF programs |
RSS scores |
Principled identification of cell-type-specific regulons |
Biology validation |
Known TF-cell type associations are recovered |
Binary activity |
Fraction-active analysis shows regulatory penetrance |
Dot plot |
Combined view of activity strength and breadth |
Differential activity |
Statistical identification of TFs distinguishing cell types |
TF-target networks |
Visualization of regulatory targets for any TF |
Co-activity analysis |
Discovery of cooperative and antagonistic TF programs |
Next Steps#
Custom TF lists: Pass
tf_list_pathto use curated TF lists (e.g., Lambert 2018)Multi-species: Use
species='mouse'for mouse dataParameter tuning: Adjust
module_k,pruning_nes_threshold, etc. – see the Pipeline GuideIntegration comparison: Compare AUCell embedding against Harmony, scVI, or Scanorama