ohashi56225
commited on
Commit
•
8b742bc
1
Parent(s):
fc5444e
Add pretrained checkpoints
Browse files- coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/004.pt +3 -0
- coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/args.json +26 -0
- coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/log.json +52 -0
- coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/011.pt +3 -0
- coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/args.json +26 -0
- coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/log.json +102 -0
- coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/009.pt +3 -0
- coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/args.json +26 -0
- coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/log.json +67 -0
coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/004.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:692bec3c3fa7b1c4c4c14dca42d9c5347fb0d9f0ee2478b1a50cb8b2acdc4ed8
|
3 |
+
size 1659607609
|
coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/args.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"train_name_prefix": null,
|
3 |
+
"dataset_name": "coco",
|
4 |
+
"rinna_gpt_name": "gpt_medium",
|
5 |
+
"clip_model_name": "en_clip_b32",
|
6 |
+
"pretrained_path": null,
|
7 |
+
"datasets_dpath": "./data",
|
8 |
+
"checkpoints_dpath": "./checkpoints",
|
9 |
+
"epochs": 10,
|
10 |
+
"per_gpu_train_batch_size": 24,
|
11 |
+
"per_gpu_eval_batch_size": 24,
|
12 |
+
"lr": 2e-05,
|
13 |
+
"warmup_steps": 5000,
|
14 |
+
"save_every": 1,
|
15 |
+
"mapping_type": "transformer",
|
16 |
+
"prefix_length": 10,
|
17 |
+
"prefix_length_clip": 10,
|
18 |
+
"only_prefix": false,
|
19 |
+
"num_layers": 8,
|
20 |
+
"n_gpu": 2,
|
21 |
+
"prefix_dim": 512,
|
22 |
+
"train_batch_size": 48,
|
23 |
+
"eval_batch_size": 48,
|
24 |
+
"train_data_fpath": "./data/coco/processed-en_clip_b32/train.pkl",
|
25 |
+
"valid_data_fpath": "./data/coco/processed-en_clip_b32/valid.pkl"
|
26 |
+
}
|
coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/log.json
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[
|
2 |
+
{
|
3 |
+
"epoch": 0,
|
4 |
+
"train_avg_loss": 2.900748385220834,
|
5 |
+
"valid_avg_loss": 2.0124718794857497
|
6 |
+
},
|
7 |
+
{
|
8 |
+
"epoch": 1,
|
9 |
+
"train_avg_loss": 1.929572620044265,
|
10 |
+
"valid_avg_loss": 1.842193490397321
|
11 |
+
},
|
12 |
+
{
|
13 |
+
"epoch": 2,
|
14 |
+
"train_avg_loss": 1.755735797382433,
|
15 |
+
"valid_avg_loss": 1.7596544592920011
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 3,
|
19 |
+
"train_avg_loss": 1.630243831397732,
|
20 |
+
"valid_avg_loss": 1.7320620456751246
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 4,
|
24 |
+
"train_avg_loss": 1.5287480419480448,
|
25 |
+
"valid_avg_loss": 1.7225063604159947
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"epoch": 5,
|
29 |
+
"train_avg_loss": 1.4382578874774836,
|
30 |
+
"valid_avg_loss": 1.7327922234570023
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 6,
|
34 |
+
"train_avg_loss": 1.3579212888226693,
|
35 |
+
"valid_avg_loss": 1.7398730559940756
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"epoch": 7,
|
39 |
+
"train_avg_loss": 1.286986698702419,
|
40 |
+
"valid_avg_loss": 1.7643859573524363
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"epoch": 8,
|
44 |
+
"train_avg_loss": 1.229390325198684,
|
45 |
+
"valid_avg_loss": 1.7796538959454447
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 9,
|
49 |
+
"train_avg_loss": 1.1848171936080776,
|
50 |
+
"valid_avg_loss": 1.796674854999041
|
51 |
+
}
|
52 |
+
]
|
coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/011.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbb6f5ed8ca6c04885ffd0a418816b8d715fe1a9330d2696961c0c0e77df1c3f
|
3 |
+
size 1659601962
|
coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/args.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"train_name_prefix": "coco_based",
|
3 |
+
"dataset_name": "sfcoco2022",
|
4 |
+
"rinna_gpt_name": "gpt_medium",
|
5 |
+
"clip_model_name": "en_clip_b32",
|
6 |
+
"pretrained_path": "checkpoints/coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/004.pt",
|
7 |
+
"datasets_dpath": "./data",
|
8 |
+
"checkpoints_dpath": "./checkpoints",
|
9 |
+
"epochs": 20,
|
10 |
+
"per_gpu_train_batch_size": 8,
|
11 |
+
"per_gpu_eval_batch_size": 4,
|
12 |
+
"lr": 2e-05,
|
13 |
+
"warmup_steps": 5000,
|
14 |
+
"save_every": 2,
|
15 |
+
"mapping_type": "transformer",
|
16 |
+
"prefix_length": 10,
|
17 |
+
"prefix_length_clip": 10,
|
18 |
+
"only_prefix": false,
|
19 |
+
"num_layers": 8,
|
20 |
+
"n_gpu": 1,
|
21 |
+
"prefix_dim": 512,
|
22 |
+
"train_batch_size": 8,
|
23 |
+
"eval_batch_size": 4,
|
24 |
+
"train_data_fpath": "./data/sfcoco2022/processed-en_clip_b32/train.pkl",
|
25 |
+
"valid_data_fpath": "./data/sfcoco2022/processed-en_clip_b32/valid.pkl"
|
26 |
+
}
|
coco_based-sfcoco2022-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/log.json
ADDED
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[
|
2 |
+
{
|
3 |
+
"epoch": 0,
|
4 |
+
"train_avg_loss": 5.23631593082728,
|
5 |
+
"valid_avg_loss": 4.589402989907698
|
6 |
+
},
|
7 |
+
{
|
8 |
+
"epoch": 1,
|
9 |
+
"train_avg_loss": 4.692692512876532,
|
10 |
+
"valid_avg_loss": 4.057069377465681
|
11 |
+
},
|
12 |
+
{
|
13 |
+
"epoch": 2,
|
14 |
+
"train_avg_loss": 4.1836722861515,
|
15 |
+
"valid_avg_loss": 3.802787184715271
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 3,
|
19 |
+
"train_avg_loss": 3.912685916664895,
|
20 |
+
"valid_avg_loss": 3.6272812431508843
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 4,
|
24 |
+
"train_avg_loss": 3.71119554123182,
|
25 |
+
"valid_avg_loss": 3.5032514225352895
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"epoch": 5,
|
29 |
+
"train_avg_loss": 3.5524264828542647,
|
30 |
+
"valid_avg_loss": 3.416551037268205
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 6,
|
34 |
+
"train_avg_loss": 3.4106314369801725,
|
35 |
+
"valid_avg_loss": 3.3739624565297905
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"epoch": 7,
|
39 |
+
"train_avg_loss": 3.2732532479789818,
|
40 |
+
"valid_avg_loss": 3.311254208738154
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"epoch": 8,
|
44 |
+
"train_avg_loss": 3.1427170078406172,
|
45 |
+
"valid_avg_loss": 3.307594797827981
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 9,
|
49 |
+
"train_avg_loss": 3.010394993792759,
|
50 |
+
"valid_avg_loss": 3.2557569200342353
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 10,
|
54 |
+
"train_avg_loss": 2.878009672914998,
|
55 |
+
"valid_avg_loss": 3.2686923525550147
|
56 |
+
},
|
57 |
+
{
|
58 |
+
"epoch": 11,
|
59 |
+
"train_avg_loss": 2.7372213776191967,
|
60 |
+
"valid_avg_loss": 3.2398276220668447
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 12,
|
64 |
+
"train_avg_loss": 2.5947886558061235,
|
65 |
+
"valid_avg_loss": 3.308401259509
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 13,
|
69 |
+
"train_avg_loss": 2.436928727653589,
|
70 |
+
"valid_avg_loss": 3.3593532172116367
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"epoch": 14,
|
74 |
+
"train_avg_loss": 2.2813261600022905,
|
75 |
+
"valid_avg_loss": 3.457464749162847
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 15,
|
79 |
+
"train_avg_loss": 2.141249016429601,
|
80 |
+
"valid_avg_loss": 3.4138374870473687
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 16,
|
84 |
+
"train_avg_loss": 1.9144119077853943,
|
85 |
+
"valid_avg_loss": 3.4829627058722754
|
86 |
+
},
|
87 |
+
{
|
88 |
+
"epoch": 17,
|
89 |
+
"train_avg_loss": 1.757168180487129,
|
90 |
+
"valid_avg_loss": 3.509248224171725
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 18,
|
94 |
+
"train_avg_loss": 1.572956797781955,
|
95 |
+
"valid_avg_loss": 3.5913555296984585
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"epoch": 19,
|
99 |
+
"train_avg_loss": 1.3907193682166967,
|
100 |
+
"valid_avg_loss": 3.651268948208202
|
101 |
+
}
|
102 |
+
]
|
coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/009.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf67194444f24c8de46efdaa3e74f4a3475e10554deaa2aae1fd168e72493761
|
3 |
+
size 1659602257
|
coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/args.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"train_name_prefix": "coco_based",
|
3 |
+
"dataset_name": "sfcoco_merged",
|
4 |
+
"rinna_gpt_name": "gpt_medium",
|
5 |
+
"clip_model_name": "en_clip_b32",
|
6 |
+
"pretrained_path": "checkpoints/coco-gpt_medium-en_clip_b32-transformer-finetune-ep10-bs48-lr2e-05/004.pt",
|
7 |
+
"datasets_dpath": "./data",
|
8 |
+
"checkpoints_dpath": "./checkpoints",
|
9 |
+
"epochs": 20,
|
10 |
+
"per_gpu_train_batch_size": 8,
|
11 |
+
"per_gpu_eval_batch_size": 4,
|
12 |
+
"lr": 2e-05,
|
13 |
+
"warmup_steps": 5000,
|
14 |
+
"save_every": 2,
|
15 |
+
"mapping_type": "transformer",
|
16 |
+
"prefix_length": 10,
|
17 |
+
"prefix_length_clip": 10,
|
18 |
+
"only_prefix": false,
|
19 |
+
"num_layers": 8,
|
20 |
+
"n_gpu": 1,
|
21 |
+
"prefix_dim": 512,
|
22 |
+
"train_batch_size": 8,
|
23 |
+
"eval_batch_size": 4,
|
24 |
+
"train_data_fpath": "./data/sfcoco_merged/processed-en_clip_b32/train.pkl",
|
25 |
+
"valid_data_fpath": "./data/sfcoco_merged/processed-en_clip_b32/valid.pkl"
|
26 |
+
}
|
coco_based-sfcoco_merged-gpt_medium-en_clip_b32-transformer-finetune-ep20-bs8-lr2e-05/log.json
ADDED
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[
|
2 |
+
{
|
3 |
+
"epoch": 0,
|
4 |
+
"train_avg_loss": 4.911076001559986,
|
5 |
+
"valid_avg_loss": 4.273226414408002
|
6 |
+
},
|
7 |
+
{
|
8 |
+
"epoch": 1,
|
9 |
+
"train_avg_loss": 4.039783334732055,
|
10 |
+
"valid_avg_loss": 3.7189245564596995
|
11 |
+
},
|
12 |
+
{
|
13 |
+
"epoch": 2,
|
14 |
+
"train_avg_loss": 3.6596964639775895,
|
15 |
+
"valid_avg_loss": 3.472972523598444
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 3,
|
19 |
+
"train_avg_loss": 3.440858439838185,
|
20 |
+
"valid_avg_loss": 3.344558346839178
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 4,
|
24 |
+
"train_avg_loss": 3.2580301088445327,
|
25 |
+
"valid_avg_loss": 3.2396208558763777
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"epoch": 5,
|
29 |
+
"train_avg_loss": 3.100968130897073,
|
30 |
+
"valid_avg_loss": 3.191791926111494
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 6,
|
34 |
+
"train_avg_loss": 2.943284174975227,
|
35 |
+
"valid_avg_loss": 3.1369220245452154
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"epoch": 7,
|
39 |
+
"train_avg_loss": 2.7977304584839766,
|
40 |
+
"valid_avg_loss": 3.0809161152158464
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"epoch": 8,
|
44 |
+
"train_avg_loss": 2.6307832703870884,
|
45 |
+
"valid_avg_loss": 3.0518643458684287
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 9,
|
49 |
+
"train_avg_loss": 2.455103006783654,
|
50 |
+
"valid_avg_loss": 3.0716028610865274
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 10,
|
54 |
+
"train_avg_loss": 2.2595722352757175,
|
55 |
+
"valid_avg_loss": 3.074081716083345
|
56 |
+
},
|
57 |
+
{
|
58 |
+
"epoch": 11,
|
59 |
+
"train_avg_loss": 2.0651670189464792,
|
60 |
+
"valid_avg_loss": 3.1168776012602306
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 12,
|
64 |
+
"train_avg_loss": 1.8646753367255715,
|
65 |
+
"valid_avg_loss": 3.179236167953128
|
66 |
+
}
|
67 |
+
]
|