Ruben_tito's workspace
Runs
128
Name
1 visualized
State
Notes
Tags
Created
Runtime
Batch size
Best iteration
Best metric
Test/FullVal - Retrieval precision
Document Pages
DOCUMENT CLS tokens
Model
Dataset
OCR tokens
Optimizer/Scheduler
Optimizer/Type
Optimizer/lr
Initial weights
Retrieval Module
Retrieval Loss
Retrieval Loss weight
Model Params
Model Trainable Params
Retrieval Loss weight
Test/Batch Retrieval loss
Test/FullVal - Retrieval loss
Train/Avg. Retrieval loss
Train/Batch Retrieval loss
Val/Batch Retrieval loss
Val/FullVal - Retrieval loss
Crashed
Local
inference
layout_ct5_small
layout_on
t5_doccvqa
14m 46s
2
-
-
-
3
10
layout_ct5_small
t5_doccvqa
300
Linear
AdamW
0.0002
/SSD2/DocVQA_T5_Weights/LayoutT5/layoutt5_small_finetune/best.ckpt
true
CE
1
62
62
1
-
-
-
-
-
-
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
19h 10m 58s
8
32001
0.44611
0.9049
2
10
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
1
226
226
1
0.000055098
0.3061
0.1382
0.040721
3.20647
0.2497
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
18h 32m 1s
8
32000
0.44701
0.9025
2
10
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0
226
226
0
0.0000030696
0.3084
0.1391
0.12518
1.56386
0.2274
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
18h 47m 10s
8
31000
0.45064
0.9021
2
25
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
0
0.3425
0.1512
0.00011004
3.16646
0.2735
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
18h 26m 53s
8
32001
0.42693
0.9108
2
5
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
4.1723e-7
0.2722
0.1335
0.011557
1.82691
0.2171
Failed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
17h 3m 55s
8
29000
0.37042
-
2
2
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
-
-
0.1365
0.0071224
0.0691
0.1975
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
18h 23m 41s
8
32000
0.35335
0.8989
2
1
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
0.0024785
0.238
0.1349
0.041522
1.48963
0.1928
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
18h 27m 44s
8
32001
0.44859
0.9021
2
10
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
8.0466e-7
0.3198
0.1386
0.11973
1.64961
0.2398
Crashed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
1d 20h 48m 50s
8
12000
0.46319
-
20
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/doccvqa_2Pgs_50CLS_from_long_pretrain_RetModule_CEw025__finetune_decoder_20Pgs/t5_doccvqa_layout_ct5_base/models/model_5000.ckpt
true
CE
0.25
242
128
0.25
-
-
2.3661
0.24794
0.66045
3.0315
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
1d 30m 25s
8
1000
0.46886
-
20
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/doccvqa_2Pgs_50CLS_from_long_pretrain_RetModule_CEw025_2nd/t5_doccvqa_layout_ct5_base/best.ckpt
true
CE
0.25
242
128
0.25
-
-
2.794
1.37289
3.66826
3.1126
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
1d 4h 7m 10s
8
-
-
-
20
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
242
128
0.25
-
-
-
-
-
-
Failed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
1d 4h 6m 55s
8
7000
0.018644
-
20
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
242
128
0.25
-
-
5.9035
3.57767
5.29829
6.9039
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
23h 6m 27s
8
7000
0.018259
-
20
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
242
128
0.25
-
-
5.8823
3.36253
7.48352
7.8909
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
19h 8m 16s
8
32001
0.49721
0.8913
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
226
226
0.25
0.000021844
0.3874
0.2086
0.000069662
2.81684
0.2774
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
19h 12m 5s
8
32001
0.44027
0.9009
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
-
true
CE
0.25
226
226
0.25
5.9605e-8
0.3795
0.1654
0.018437
2.90162
0.2667
Crashed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
7m 8s
8
-
-
-
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
226
226
0.25
-
-
-
-
-
-
Crashed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
7m 8s
8
-
-
-
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
226
226
0.25
-
-
0.5578
0.6011
0.26902
-
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
19h 1m 3s
8
-
-
-
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
226
226
0.25
-
-
-
-
-
-
Finished
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
6h 9m 32s
8
10000
0.42009
-
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long_cont/t5_ocr_idl_layout_ct5_base/models/model_129000.ckpt
true
CE
0.25
226
226
0.25
-
-
0.2904
0.0059505
0.083585
0.2789
Crashed
DAG-A40
layout_ct5_base
layout_on
t5_doccvqa
train+inference
8h 5m 38s
8
12000
0.40657
-
2
50
layout_ct5_base
t5_doccvqa
1024
Linear
AdamW
0.0002
/data/users/rperez/pythia_save/ocr_idl_1Pgs_50CLS_pretrain_long/t5_ocr_idl_layout_ct5_base/models/model_50000_mod.ckpt
true
CE
0.25
226
226
0.25
-
-
0.2761
0.011122
0.18823
0.2812
1-20
of 128