Upload 3 files
Browse fileschange version to submission 3
- pytorch_model.bin +1 -1
- train_args.json +32 -0
- trainer_state.json +37 -0
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3785732668
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:820cb1a44e5a3d0304ef31ea105fc140bcb2865dc4d257a030d7fea72e557668
|
| 3 |
size 3785732668
|
train_args.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "../drive/MyDrive/NICE/model/0422_train_encoder",
|
| 3 |
+
"model_name_or_path": "../drive/MyDrive/NICE/model/cap_best/OFA-huge-caption",
|
| 4 |
+
"train_caption_file": "./data/train_caption.jsonl",
|
| 5 |
+
"train_image_file": "../drive/MyDrive/NICE/train_image.tsv",
|
| 6 |
+
"test_caption_file": null,
|
| 7 |
+
"test_image_file": null,
|
| 8 |
+
"freeze_encoder": false,
|
| 9 |
+
"freeze_word_embed": false,
|
| 10 |
+
"num_train_epochs": 30,
|
| 11 |
+
"max_steps": -1,
|
| 12 |
+
"per_device_train_batch_size": 12,
|
| 13 |
+
"per_device_eval_batch_size": 256,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"max_seq_length": 150,
|
| 16 |
+
"logging_steps": 5000,
|
| 17 |
+
"save_steps": 5000,
|
| 18 |
+
"save_total_limit": 2,
|
| 19 |
+
"lr_scheduler_type": "cosine",
|
| 20 |
+
"warmup_steps": 2000,
|
| 21 |
+
"warmup_ratio": 0,
|
| 22 |
+
"gradient_accumulation_steps": 1,
|
| 23 |
+
"optim": "adamw_torch",
|
| 24 |
+
"seed": 69,
|
| 25 |
+
"fp16": false,
|
| 26 |
+
"no_cuda": false,
|
| 27 |
+
"dataloader_num_workers": 8,
|
| 28 |
+
"save_strategy": "steps",
|
| 29 |
+
"weight_decay": 0,
|
| 30 |
+
"max_grad_norm": 1.0,
|
| 31 |
+
"remove_unused_columns": false
|
| 32 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 30.0,
|
| 5 |
+
"global_step": 12510,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 11.99,
|
| 12 |
+
"learning_rate": 4.060393227846204e-05,
|
| 13 |
+
"loss": 0.7845,
|
| 14 |
+
"step": 5000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 23.98,
|
| 18 |
+
"learning_rate": 6.712477093047967e-06,
|
| 19 |
+
"loss": 0.0208,
|
| 20 |
+
"step": 10000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 30.0,
|
| 24 |
+
"step": 12510,
|
| 25 |
+
"total_flos": 1.1203460780602368e+17,
|
| 26 |
+
"train_loss": 0.32232193622848304,
|
| 27 |
+
"train_runtime": 17163.0194,
|
| 28 |
+
"train_samples_per_second": 8.74,
|
| 29 |
+
"train_steps_per_second": 0.729
|
| 30 |
+
}
|
| 31 |
+
],
|
| 32 |
+
"max_steps": 12510,
|
| 33 |
+
"num_train_epochs": 30,
|
| 34 |
+
"total_flos": 1.1203460780602368e+17,
|
| 35 |
+
"trial_name": null,
|
| 36 |
+
"trial_params": null
|
| 37 |
+
}
|