Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/trainer_state.json +484 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/trainer_state.json +297 -0
- ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/trainer_state.json +403 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/trainer_state.json +583 -0
- ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/trainer_state.json +477 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/trainer_state.json +290 -0
- ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/trainer_state.json +572 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/tokenizer_config.json +55 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/trainer_state.json +371 -0
- ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/vocab.txt +0 -0
- ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/added_tokens.json +3 -0
- ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/bpe.codes +0 -0
- ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/special_tokens_map.json +9 -0
- ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/tokenizer_config.json +55 -0
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/trainer_state.json
ADDED
|
@@ -0,0 +1,484 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 752,
|
| 3 |
+
"best_metric": 0.8325866772541457,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752",
|
| 5 |
+
"epoch": 7.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1316,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.13351134846461948,
|
| 14 |
+
"grad_norm": 7.965886116027832,
|
| 15 |
+
"learning_rate": 1.276595744680851e-06,
|
| 16 |
+
"loss": 4.4191,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.26702269692923897,
|
| 21 |
+
"grad_norm": 7.028569221496582,
|
| 22 |
+
"learning_rate": 2.6063829787234047e-06,
|
| 23 |
+
"loss": 4.427,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.40053404539385845,
|
| 28 |
+
"grad_norm": 6.050756454467773,
|
| 29 |
+
"learning_rate": 3.936170212765958e-06,
|
| 30 |
+
"loss": 4.3877,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5340453938584779,
|
| 35 |
+
"grad_norm": 6.718469142913818,
|
| 36 |
+
"learning_rate": 5.265957446808511e-06,
|
| 37 |
+
"loss": 4.3441,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6675567423230975,
|
| 42 |
+
"grad_norm": 7.3162431716918945,
|
| 43 |
+
"learning_rate": 6.595744680851064e-06,
|
| 44 |
+
"loss": 4.3093,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8010680907877169,
|
| 49 |
+
"grad_norm": 8.02818489074707,
|
| 50 |
+
"learning_rate": 7.925531914893617e-06,
|
| 51 |
+
"loss": 4.2785,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.9345794392523364,
|
| 56 |
+
"grad_norm": 10.621535301208496,
|
| 57 |
+
"learning_rate": 9.255319148936171e-06,
|
| 58 |
+
"loss": 4.2058,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5049627791563276,
|
| 64 |
+
"eval_loss": 1.0008918046951294,
|
| 65 |
+
"eval_macro_f1": 0.47173060704844394,
|
| 66 |
+
"eval_runtime": 1.5369,
|
| 67 |
+
"eval_samples_per_second": 524.442,
|
| 68 |
+
"eval_steps_per_second": 33.184,
|
| 69 |
+
"eval_weighted_f1": 0.47341133961111737,
|
| 70 |
+
"step": 188
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.0640854472630175,
|
| 74 |
+
"grad_norm": 21.895902633666992,
|
| 75 |
+
"learning_rate": 1.0585106382978725e-05,
|
| 76 |
+
"loss": 3.7407,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.1975967957276368,
|
| 81 |
+
"grad_norm": 40.787078857421875,
|
| 82 |
+
"learning_rate": 1.1914893617021277e-05,
|
| 83 |
+
"loss": 3.605,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.3311081441922563,
|
| 88 |
+
"grad_norm": 26.384397506713867,
|
| 89 |
+
"learning_rate": 1.3244680851063832e-05,
|
| 90 |
+
"loss": 3.0976,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.4646194926568759,
|
| 95 |
+
"grad_norm": 23.437131881713867,
|
| 96 |
+
"learning_rate": 1.4574468085106384e-05,
|
| 97 |
+
"loss": 2.7655,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.5981308411214954,
|
| 102 |
+
"grad_norm": 45.37485122680664,
|
| 103 |
+
"learning_rate": 1.590425531914894e-05,
|
| 104 |
+
"loss": 2.7698,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.7316421895861147,
|
| 109 |
+
"grad_norm": 34.840240478515625,
|
| 110 |
+
"learning_rate": 1.723404255319149e-05,
|
| 111 |
+
"loss": 2.3689,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.8651535380507345,
|
| 116 |
+
"grad_norm": 42.31020736694336,
|
| 117 |
+
"learning_rate": 1.8563829787234043e-05,
|
| 118 |
+
"loss": 2.2223,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.9986648865153538,
|
| 123 |
+
"grad_norm": 31.220808029174805,
|
| 124 |
+
"learning_rate": 1.9893617021276595e-05,
|
| 125 |
+
"loss": 2.3008,
|
| 126 |
+
"step": 375
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 2.0,
|
| 130 |
+
"eval_accuracy": 0.8200992555831266,
|
| 131 |
+
"eval_loss": 0.48810315132141113,
|
| 132 |
+
"eval_macro_f1": 0.8206753318318988,
|
| 133 |
+
"eval_runtime": 1.5516,
|
| 134 |
+
"eval_samples_per_second": 519.458,
|
| 135 |
+
"eval_steps_per_second": 32.869,
|
| 136 |
+
"eval_weighted_f1": 0.8205190687800941,
|
| 137 |
+
"step": 376
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.128170894526035,
|
| 141 |
+
"grad_norm": 24.27152442932129,
|
| 142 |
+
"learning_rate": 1.986406619385343e-05,
|
| 143 |
+
"loss": 1.5853,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.2616822429906542,
|
| 148 |
+
"grad_norm": 36.713199615478516,
|
| 149 |
+
"learning_rate": 1.971631205673759e-05,
|
| 150 |
+
"loss": 1.5087,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.3951935914552736,
|
| 155 |
+
"grad_norm": 27.86497688293457,
|
| 156 |
+
"learning_rate": 1.956855791962175e-05,
|
| 157 |
+
"loss": 1.6186,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.5287049399198933,
|
| 162 |
+
"grad_norm": 29.326784133911133,
|
| 163 |
+
"learning_rate": 1.9420803782505913e-05,
|
| 164 |
+
"loss": 1.3477,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.6622162883845126,
|
| 169 |
+
"grad_norm": 27.314132690429688,
|
| 170 |
+
"learning_rate": 1.927304964539007e-05,
|
| 171 |
+
"loss": 1.7717,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.795727636849132,
|
| 176 |
+
"grad_norm": 46.23450469970703,
|
| 177 |
+
"learning_rate": 1.9125295508274234e-05,
|
| 178 |
+
"loss": 1.7067,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.9292389853137517,
|
| 183 |
+
"grad_norm": 34.22685241699219,
|
| 184 |
+
"learning_rate": 1.8977541371158394e-05,
|
| 185 |
+
"loss": 1.5755,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 3.0,
|
| 190 |
+
"eval_accuracy": 0.8325062034739454,
|
| 191 |
+
"eval_loss": 0.45100587606430054,
|
| 192 |
+
"eval_macro_f1": 0.8323040471599615,
|
| 193 |
+
"eval_runtime": 1.5357,
|
| 194 |
+
"eval_samples_per_second": 524.854,
|
| 195 |
+
"eval_steps_per_second": 33.21,
|
| 196 |
+
"eval_weighted_f1": 0.8320199824637069,
|
| 197 |
+
"step": 564
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.0587449933244324,
|
| 201 |
+
"grad_norm": 12.387439727783203,
|
| 202 |
+
"learning_rate": 1.8829787234042557e-05,
|
| 203 |
+
"loss": 1.2998,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.192256341789052,
|
| 208 |
+
"grad_norm": 18.85713005065918,
|
| 209 |
+
"learning_rate": 1.8682033096926714e-05,
|
| 210 |
+
"loss": 1.0312,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.3257676902536715,
|
| 215 |
+
"grad_norm": 69.58992767333984,
|
| 216 |
+
"learning_rate": 1.8534278959810878e-05,
|
| 217 |
+
"loss": 1.0381,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.4592790387182912,
|
| 222 |
+
"grad_norm": 108.8010025024414,
|
| 223 |
+
"learning_rate": 1.8386524822695038e-05,
|
| 224 |
+
"loss": 1.0332,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.5927903871829105,
|
| 229 |
+
"grad_norm": 9.588434219360352,
|
| 230 |
+
"learning_rate": 1.8238770685579198e-05,
|
| 231 |
+
"loss": 1.0001,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.72630173564753,
|
| 236 |
+
"grad_norm": 35.762901306152344,
|
| 237 |
+
"learning_rate": 1.8091016548463358e-05,
|
| 238 |
+
"loss": 0.946,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.8598130841121496,
|
| 243 |
+
"grad_norm": 45.67814254760742,
|
| 244 |
+
"learning_rate": 1.7943262411347518e-05,
|
| 245 |
+
"loss": 1.0035,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.993324432576769,
|
| 250 |
+
"grad_norm": 62.99699401855469,
|
| 251 |
+
"learning_rate": 1.779550827423168e-05,
|
| 252 |
+
"loss": 1.0334,
|
| 253 |
+
"step": 750
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 4.0,
|
| 257 |
+
"eval_accuracy": 0.8325062034739454,
|
| 258 |
+
"eval_loss": 0.4956439435482025,
|
| 259 |
+
"eval_macro_f1": 0.8325866772541457,
|
| 260 |
+
"eval_runtime": 1.5554,
|
| 261 |
+
"eval_samples_per_second": 518.194,
|
| 262 |
+
"eval_steps_per_second": 32.789,
|
| 263 |
+
"eval_weighted_f1": 0.8322920075812518,
|
| 264 |
+
"step": 752
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.12283044058745,
|
| 268 |
+
"grad_norm": 33.03057861328125,
|
| 269 |
+
"learning_rate": 1.7647754137115842e-05,
|
| 270 |
+
"loss": 0.622,
|
| 271 |
+
"step": 775
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.25634178905207,
|
| 275 |
+
"grad_norm": 8.428254127502441,
|
| 276 |
+
"learning_rate": 1.7500000000000002e-05,
|
| 277 |
+
"loss": 0.6731,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.389853137516689,
|
| 282 |
+
"grad_norm": 17.192779541015625,
|
| 283 |
+
"learning_rate": 1.7352245862884162e-05,
|
| 284 |
+
"loss": 0.6761,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.5233644859813085,
|
| 289 |
+
"grad_norm": 71.76996612548828,
|
| 290 |
+
"learning_rate": 1.7204491725768322e-05,
|
| 291 |
+
"loss": 0.8275,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 4.656875834445928,
|
| 296 |
+
"grad_norm": 50.8429069519043,
|
| 297 |
+
"learning_rate": 1.7056737588652482e-05,
|
| 298 |
+
"loss": 0.7105,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.790387182910547,
|
| 303 |
+
"grad_norm": 39.10743713378906,
|
| 304 |
+
"learning_rate": 1.6908983451536646e-05,
|
| 305 |
+
"loss": 0.7741,
|
| 306 |
+
"step": 900
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 4.923898531375167,
|
| 310 |
+
"grad_norm": 96.48683166503906,
|
| 311 |
+
"learning_rate": 1.6761229314420806e-05,
|
| 312 |
+
"loss": 0.6811,
|
| 313 |
+
"step": 925
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 5.0,
|
| 317 |
+
"eval_accuracy": 0.8126550868486352,
|
| 318 |
+
"eval_loss": 0.601965606212616,
|
| 319 |
+
"eval_macro_f1": 0.8130126007621471,
|
| 320 |
+
"eval_runtime": 1.4975,
|
| 321 |
+
"eval_samples_per_second": 538.229,
|
| 322 |
+
"eval_steps_per_second": 34.057,
|
| 323 |
+
"eval_weighted_f1": 0.8127456442696498,
|
| 324 |
+
"step": 940
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 5.0534045393858475,
|
| 328 |
+
"grad_norm": 2.6056230068206787,
|
| 329 |
+
"learning_rate": 1.6613475177304966e-05,
|
| 330 |
+
"loss": 0.6366,
|
| 331 |
+
"step": 950
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.186915887850467,
|
| 335 |
+
"grad_norm": 40.955055236816406,
|
| 336 |
+
"learning_rate": 1.6465721040189126e-05,
|
| 337 |
+
"loss": 0.5218,
|
| 338 |
+
"step": 975
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.320427236315087,
|
| 342 |
+
"grad_norm": 144.09292602539062,
|
| 343 |
+
"learning_rate": 1.6317966903073286e-05,
|
| 344 |
+
"loss": 0.4191,
|
| 345 |
+
"step": 1000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.453938584779706,
|
| 349 |
+
"grad_norm": 55.89153289794922,
|
| 350 |
+
"learning_rate": 1.6170212765957446e-05,
|
| 351 |
+
"loss": 0.5517,
|
| 352 |
+
"step": 1025
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 5.587449933244326,
|
| 356 |
+
"grad_norm": 56.29103088378906,
|
| 357 |
+
"learning_rate": 1.602245862884161e-05,
|
| 358 |
+
"loss": 0.6066,
|
| 359 |
+
"step": 1050
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.720961281708945,
|
| 363 |
+
"grad_norm": 66.94448852539062,
|
| 364 |
+
"learning_rate": 1.587470449172577e-05,
|
| 365 |
+
"loss": 0.6073,
|
| 366 |
+
"step": 1075
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 5.854472630173564,
|
| 370 |
+
"grad_norm": 3.7524659633636475,
|
| 371 |
+
"learning_rate": 1.572695035460993e-05,
|
| 372 |
+
"loss": 0.3546,
|
| 373 |
+
"step": 1100
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 5.9879839786381845,
|
| 377 |
+
"grad_norm": 0.8753046989440918,
|
| 378 |
+
"learning_rate": 1.557919621749409e-05,
|
| 379 |
+
"loss": 0.4768,
|
| 380 |
+
"step": 1125
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 6.0,
|
| 384 |
+
"eval_accuracy": 0.8163771712158809,
|
| 385 |
+
"eval_loss": 0.7267618775367737,
|
| 386 |
+
"eval_macro_f1": 0.8151945520366572,
|
| 387 |
+
"eval_runtime": 1.6257,
|
| 388 |
+
"eval_samples_per_second": 495.785,
|
| 389 |
+
"eval_steps_per_second": 31.371,
|
| 390 |
+
"eval_weighted_f1": 0.8151067947332803,
|
| 391 |
+
"step": 1128
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 6.117489986648865,
|
| 395 |
+
"grad_norm": 16.276411056518555,
|
| 396 |
+
"learning_rate": 1.543144208037825e-05,
|
| 397 |
+
"loss": 0.3285,
|
| 398 |
+
"step": 1150
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 6.251001335113485,
|
| 402 |
+
"grad_norm": 10.489091873168945,
|
| 403 |
+
"learning_rate": 1.5283687943262414e-05,
|
| 404 |
+
"loss": 0.377,
|
| 405 |
+
"step": 1175
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 6.384512683578104,
|
| 409 |
+
"grad_norm": 0.765251874923706,
|
| 410 |
+
"learning_rate": 1.5135933806146574e-05,
|
| 411 |
+
"loss": 0.3017,
|
| 412 |
+
"step": 1200
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 6.518024032042724,
|
| 416 |
+
"grad_norm": 190.42669677734375,
|
| 417 |
+
"learning_rate": 1.4988179669030734e-05,
|
| 418 |
+
"loss": 0.4522,
|
| 419 |
+
"step": 1225
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 6.651535380507343,
|
| 423 |
+
"grad_norm": 68.99852752685547,
|
| 424 |
+
"learning_rate": 1.4840425531914894e-05,
|
| 425 |
+
"loss": 0.3354,
|
| 426 |
+
"step": 1250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 6.785046728971962,
|
| 430 |
+
"grad_norm": 13.840676307678223,
|
| 431 |
+
"learning_rate": 1.4692671394799056e-05,
|
| 432 |
+
"loss": 0.5567,
|
| 433 |
+
"step": 1275
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 6.9185580774365825,
|
| 437 |
+
"grad_norm": 3.273326873779297,
|
| 438 |
+
"learning_rate": 1.4544917257683215e-05,
|
| 439 |
+
"loss": 0.3121,
|
| 440 |
+
"step": 1300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 7.0,
|
| 444 |
+
"eval_accuracy": 0.8325062034739454,
|
| 445 |
+
"eval_loss": 0.7452908754348755,
|
| 446 |
+
"eval_macro_f1": 0.8313875579858673,
|
| 447 |
+
"eval_runtime": 1.5845,
|
| 448 |
+
"eval_samples_per_second": 508.69,
|
| 449 |
+
"eval_steps_per_second": 32.188,
|
| 450 |
+
"eval_weighted_f1": 0.8312589228278212,
|
| 451 |
+
"step": 1316
|
| 452 |
+
}
|
| 453 |
+
],
|
| 454 |
+
"logging_steps": 25,
|
| 455 |
+
"max_steps": 3760,
|
| 456 |
+
"num_input_tokens_seen": 0,
|
| 457 |
+
"num_train_epochs": 20,
|
| 458 |
+
"save_steps": 500,
|
| 459 |
+
"stateful_callbacks": {
|
| 460 |
+
"EarlyStoppingCallback": {
|
| 461 |
+
"args": {
|
| 462 |
+
"early_stopping_patience": 3,
|
| 463 |
+
"early_stopping_threshold": 0.0
|
| 464 |
+
},
|
| 465 |
+
"attributes": {
|
| 466 |
+
"early_stopping_patience_counter": 3
|
| 467 |
+
}
|
| 468 |
+
},
|
| 469 |
+
"TrainerControl": {
|
| 470 |
+
"args": {
|
| 471 |
+
"should_epoch_stop": false,
|
| 472 |
+
"should_evaluate": false,
|
| 473 |
+
"should_log": false,
|
| 474 |
+
"should_save": true,
|
| 475 |
+
"should_training_stop": true
|
| 476 |
+
},
|
| 477 |
+
"attributes": {}
|
| 478 |
+
}
|
| 479 |
+
},
|
| 480 |
+
"total_flos": 0.0,
|
| 481 |
+
"train_batch_size": 8,
|
| 482 |
+
"trial_name": null,
|
| 483 |
+
"trial_params": null
|
| 484 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-1316/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/trainer_state.json
ADDED
|
@@ -0,0 +1,297 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 752,
|
| 3 |
+
"best_metric": 0.8325866772541457,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752",
|
| 5 |
+
"epoch": 4.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 752,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.13351134846461948,
|
| 14 |
+
"grad_norm": 7.965886116027832,
|
| 15 |
+
"learning_rate": 1.276595744680851e-06,
|
| 16 |
+
"loss": 4.4191,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.26702269692923897,
|
| 21 |
+
"grad_norm": 7.028569221496582,
|
| 22 |
+
"learning_rate": 2.6063829787234047e-06,
|
| 23 |
+
"loss": 4.427,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.40053404539385845,
|
| 28 |
+
"grad_norm": 6.050756454467773,
|
| 29 |
+
"learning_rate": 3.936170212765958e-06,
|
| 30 |
+
"loss": 4.3877,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5340453938584779,
|
| 35 |
+
"grad_norm": 6.718469142913818,
|
| 36 |
+
"learning_rate": 5.265957446808511e-06,
|
| 37 |
+
"loss": 4.3441,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6675567423230975,
|
| 42 |
+
"grad_norm": 7.3162431716918945,
|
| 43 |
+
"learning_rate": 6.595744680851064e-06,
|
| 44 |
+
"loss": 4.3093,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8010680907877169,
|
| 49 |
+
"grad_norm": 8.02818489074707,
|
| 50 |
+
"learning_rate": 7.925531914893617e-06,
|
| 51 |
+
"loss": 4.2785,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.9345794392523364,
|
| 56 |
+
"grad_norm": 10.621535301208496,
|
| 57 |
+
"learning_rate": 9.255319148936171e-06,
|
| 58 |
+
"loss": 4.2058,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5049627791563276,
|
| 64 |
+
"eval_loss": 1.0008918046951294,
|
| 65 |
+
"eval_macro_f1": 0.47173060704844394,
|
| 66 |
+
"eval_runtime": 1.5369,
|
| 67 |
+
"eval_samples_per_second": 524.442,
|
| 68 |
+
"eval_steps_per_second": 33.184,
|
| 69 |
+
"eval_weighted_f1": 0.47341133961111737,
|
| 70 |
+
"step": 188
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.0640854472630175,
|
| 74 |
+
"grad_norm": 21.895902633666992,
|
| 75 |
+
"learning_rate": 1.0585106382978725e-05,
|
| 76 |
+
"loss": 3.7407,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.1975967957276368,
|
| 81 |
+
"grad_norm": 40.787078857421875,
|
| 82 |
+
"learning_rate": 1.1914893617021277e-05,
|
| 83 |
+
"loss": 3.605,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.3311081441922563,
|
| 88 |
+
"grad_norm": 26.384397506713867,
|
| 89 |
+
"learning_rate": 1.3244680851063832e-05,
|
| 90 |
+
"loss": 3.0976,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.4646194926568759,
|
| 95 |
+
"grad_norm": 23.437131881713867,
|
| 96 |
+
"learning_rate": 1.4574468085106384e-05,
|
| 97 |
+
"loss": 2.7655,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.5981308411214954,
|
| 102 |
+
"grad_norm": 45.37485122680664,
|
| 103 |
+
"learning_rate": 1.590425531914894e-05,
|
| 104 |
+
"loss": 2.7698,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.7316421895861147,
|
| 109 |
+
"grad_norm": 34.840240478515625,
|
| 110 |
+
"learning_rate": 1.723404255319149e-05,
|
| 111 |
+
"loss": 2.3689,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.8651535380507345,
|
| 116 |
+
"grad_norm": 42.31020736694336,
|
| 117 |
+
"learning_rate": 1.8563829787234043e-05,
|
| 118 |
+
"loss": 2.2223,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.9986648865153538,
|
| 123 |
+
"grad_norm": 31.220808029174805,
|
| 124 |
+
"learning_rate": 1.9893617021276595e-05,
|
| 125 |
+
"loss": 2.3008,
|
| 126 |
+
"step": 375
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 2.0,
|
| 130 |
+
"eval_accuracy": 0.8200992555831266,
|
| 131 |
+
"eval_loss": 0.48810315132141113,
|
| 132 |
+
"eval_macro_f1": 0.8206753318318988,
|
| 133 |
+
"eval_runtime": 1.5516,
|
| 134 |
+
"eval_samples_per_second": 519.458,
|
| 135 |
+
"eval_steps_per_second": 32.869,
|
| 136 |
+
"eval_weighted_f1": 0.8205190687800941,
|
| 137 |
+
"step": 376
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.128170894526035,
|
| 141 |
+
"grad_norm": 24.27152442932129,
|
| 142 |
+
"learning_rate": 1.986406619385343e-05,
|
| 143 |
+
"loss": 1.5853,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.2616822429906542,
|
| 148 |
+
"grad_norm": 36.713199615478516,
|
| 149 |
+
"learning_rate": 1.971631205673759e-05,
|
| 150 |
+
"loss": 1.5087,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.3951935914552736,
|
| 155 |
+
"grad_norm": 27.86497688293457,
|
| 156 |
+
"learning_rate": 1.956855791962175e-05,
|
| 157 |
+
"loss": 1.6186,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.5287049399198933,
|
| 162 |
+
"grad_norm": 29.326784133911133,
|
| 163 |
+
"learning_rate": 1.9420803782505913e-05,
|
| 164 |
+
"loss": 1.3477,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.6622162883845126,
|
| 169 |
+
"grad_norm": 27.314132690429688,
|
| 170 |
+
"learning_rate": 1.927304964539007e-05,
|
| 171 |
+
"loss": 1.7717,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.795727636849132,
|
| 176 |
+
"grad_norm": 46.23450469970703,
|
| 177 |
+
"learning_rate": 1.9125295508274234e-05,
|
| 178 |
+
"loss": 1.7067,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.9292389853137517,
|
| 183 |
+
"grad_norm": 34.22685241699219,
|
| 184 |
+
"learning_rate": 1.8977541371158394e-05,
|
| 185 |
+
"loss": 1.5755,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 3.0,
|
| 190 |
+
"eval_accuracy": 0.8325062034739454,
|
| 191 |
+
"eval_loss": 0.45100587606430054,
|
| 192 |
+
"eval_macro_f1": 0.8323040471599615,
|
| 193 |
+
"eval_runtime": 1.5357,
|
| 194 |
+
"eval_samples_per_second": 524.854,
|
| 195 |
+
"eval_steps_per_second": 33.21,
|
| 196 |
+
"eval_weighted_f1": 0.8320199824637069,
|
| 197 |
+
"step": 564
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.0587449933244324,
|
| 201 |
+
"grad_norm": 12.387439727783203,
|
| 202 |
+
"learning_rate": 1.8829787234042557e-05,
|
| 203 |
+
"loss": 1.2998,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.192256341789052,
|
| 208 |
+
"grad_norm": 18.85713005065918,
|
| 209 |
+
"learning_rate": 1.8682033096926714e-05,
|
| 210 |
+
"loss": 1.0312,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.3257676902536715,
|
| 215 |
+
"grad_norm": 69.58992767333984,
|
| 216 |
+
"learning_rate": 1.8534278959810878e-05,
|
| 217 |
+
"loss": 1.0381,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.4592790387182912,
|
| 222 |
+
"grad_norm": 108.8010025024414,
|
| 223 |
+
"learning_rate": 1.8386524822695038e-05,
|
| 224 |
+
"loss": 1.0332,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.5927903871829105,
|
| 229 |
+
"grad_norm": 9.588434219360352,
|
| 230 |
+
"learning_rate": 1.8238770685579198e-05,
|
| 231 |
+
"loss": 1.0001,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.72630173564753,
|
| 236 |
+
"grad_norm": 35.762901306152344,
|
| 237 |
+
"learning_rate": 1.8091016548463358e-05,
|
| 238 |
+
"loss": 0.946,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.8598130841121496,
|
| 243 |
+
"grad_norm": 45.67814254760742,
|
| 244 |
+
"learning_rate": 1.7943262411347518e-05,
|
| 245 |
+
"loss": 1.0035,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.993324432576769,
|
| 250 |
+
"grad_norm": 62.99699401855469,
|
| 251 |
+
"learning_rate": 1.779550827423168e-05,
|
| 252 |
+
"loss": 1.0334,
|
| 253 |
+
"step": 750
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 4.0,
|
| 257 |
+
"eval_accuracy": 0.8325062034739454,
|
| 258 |
+
"eval_loss": 0.4956439435482025,
|
| 259 |
+
"eval_macro_f1": 0.8325866772541457,
|
| 260 |
+
"eval_runtime": 1.5554,
|
| 261 |
+
"eval_samples_per_second": 518.194,
|
| 262 |
+
"eval_steps_per_second": 32.789,
|
| 263 |
+
"eval_weighted_f1": 0.8322920075812518,
|
| 264 |
+
"step": 752
|
| 265 |
+
}
|
| 266 |
+
],
|
| 267 |
+
"logging_steps": 25,
|
| 268 |
+
"max_steps": 3760,
|
| 269 |
+
"num_input_tokens_seen": 0,
|
| 270 |
+
"num_train_epochs": 20,
|
| 271 |
+
"save_steps": 500,
|
| 272 |
+
"stateful_callbacks": {
|
| 273 |
+
"EarlyStoppingCallback": {
|
| 274 |
+
"args": {
|
| 275 |
+
"early_stopping_patience": 3,
|
| 276 |
+
"early_stopping_threshold": 0.0
|
| 277 |
+
},
|
| 278 |
+
"attributes": {
|
| 279 |
+
"early_stopping_patience_counter": 0
|
| 280 |
+
}
|
| 281 |
+
},
|
| 282 |
+
"TrainerControl": {
|
| 283 |
+
"args": {
|
| 284 |
+
"should_epoch_stop": false,
|
| 285 |
+
"should_evaluate": false,
|
| 286 |
+
"should_log": false,
|
| 287 |
+
"should_save": true,
|
| 288 |
+
"should_training_stop": false
|
| 289 |
+
},
|
| 290 |
+
"attributes": {}
|
| 291 |
+
}
|
| 292 |
+
},
|
| 293 |
+
"total_flos": 0.0,
|
| 294 |
+
"train_batch_size": 8,
|
| 295 |
+
"trial_name": null,
|
| 296 |
+
"trial_params": null
|
| 297 |
+
}
|
ood-vifactcheck-phobert-topic-domain-04-s42-20260814T114226Z-00ca7096/trainer/checkpoint-752/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/trainer_state.json
ADDED
|
@@ -0,0 +1,403 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 1050,
|
| 3 |
+
"best_metric": 0.8572722694614372,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050",
|
| 5 |
+
"epoch": 6.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1050,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.1430615164520744,
|
| 14 |
+
"grad_norm": 7.234732627868652,
|
| 15 |
+
"learning_rate": 1.3714285714285717e-06,
|
| 16 |
+
"loss": 4.4028,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.2861230329041488,
|
| 21 |
+
"grad_norm": 6.111971855163574,
|
| 22 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 23 |
+
"loss": 4.4214,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.4291845493562232,
|
| 28 |
+
"grad_norm": 6.735830783843994,
|
| 29 |
+
"learning_rate": 4.228571428571429e-06,
|
| 30 |
+
"loss": 4.3905,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5722460658082976,
|
| 35 |
+
"grad_norm": 6.769274711608887,
|
| 36 |
+
"learning_rate": 5.6571428571428576e-06,
|
| 37 |
+
"loss": 4.4068,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.7153075822603719,
|
| 42 |
+
"grad_norm": 9.92778205871582,
|
| 43 |
+
"learning_rate": 7.085714285714286e-06,
|
| 44 |
+
"loss": 4.3451,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8583690987124464,
|
| 49 |
+
"grad_norm": 8.201742172241211,
|
| 50 |
+
"learning_rate": 8.514285714285715e-06,
|
| 51 |
+
"loss": 4.2443,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 1.0,
|
| 56 |
+
"grad_norm": 11.032033920288086,
|
| 57 |
+
"learning_rate": 9.942857142857145e-06,
|
| 58 |
+
"loss": 4.0712,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5876010781671159,
|
| 64 |
+
"eval_loss": 0.9521220922470093,
|
| 65 |
+
"eval_macro_f1": 0.5740345789141216,
|
| 66 |
+
"eval_runtime": 1.4608,
|
| 67 |
+
"eval_samples_per_second": 507.927,
|
| 68 |
+
"eval_steps_per_second": 32.173,
|
| 69 |
+
"eval_weighted_f1": 0.5774423182822452,
|
| 70 |
+
"step": 175
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.1430615164520743,
|
| 74 |
+
"grad_norm": 19.701786041259766,
|
| 75 |
+
"learning_rate": 1.1371428571428573e-05,
|
| 76 |
+
"loss": 3.7206,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.2861230329041489,
|
| 81 |
+
"grad_norm": 33.54481887817383,
|
| 82 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 83 |
+
"loss": 3.4389,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.4291845493562232,
|
| 88 |
+
"grad_norm": 25.37718963623047,
|
| 89 |
+
"learning_rate": 1.422857142857143e-05,
|
| 90 |
+
"loss": 3.0285,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.5722460658082977,
|
| 95 |
+
"grad_norm": 23.7049617767334,
|
| 96 |
+
"learning_rate": 1.5657142857142856e-05,
|
| 97 |
+
"loss": 2.5281,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.715307582260372,
|
| 102 |
+
"grad_norm": 34.06275177001953,
|
| 103 |
+
"learning_rate": 1.708571428571429e-05,
|
| 104 |
+
"loss": 2.4805,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.8583690987124464,
|
| 109 |
+
"grad_norm": 34.69179153442383,
|
| 110 |
+
"learning_rate": 1.8514285714285716e-05,
|
| 111 |
+
"loss": 2.0691,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.0,
|
| 116 |
+
"grad_norm": 12.415122032165527,
|
| 117 |
+
"learning_rate": 1.9942857142857142e-05,
|
| 118 |
+
"loss": 2.0887,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.0,
|
| 123 |
+
"eval_accuracy": 0.7991913746630728,
|
| 124 |
+
"eval_loss": 0.5084040760993958,
|
| 125 |
+
"eval_macro_f1": 0.7966981355469368,
|
| 126 |
+
"eval_runtime": 1.4012,
|
| 127 |
+
"eval_samples_per_second": 529.561,
|
| 128 |
+
"eval_steps_per_second": 33.544,
|
| 129 |
+
"eval_weighted_f1": 0.7971849691058734,
|
| 130 |
+
"step": 350
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.1430615164520743,
|
| 134 |
+
"grad_norm": 30.174409866333008,
|
| 135 |
+
"learning_rate": 1.984761904761905e-05,
|
| 136 |
+
"loss": 1.592,
|
| 137 |
+
"step": 375
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.2861230329041486,
|
| 141 |
+
"grad_norm": 38.79440689086914,
|
| 142 |
+
"learning_rate": 1.968888888888889e-05,
|
| 143 |
+
"loss": 1.5487,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.429184549356223,
|
| 148 |
+
"grad_norm": 52.02887725830078,
|
| 149 |
+
"learning_rate": 1.953015873015873e-05,
|
| 150 |
+
"loss": 1.8353,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.5722460658082977,
|
| 155 |
+
"grad_norm": 36.64469528198242,
|
| 156 |
+
"learning_rate": 1.937142857142857e-05,
|
| 157 |
+
"loss": 1.7984,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.715307582260372,
|
| 162 |
+
"grad_norm": 44.32551956176758,
|
| 163 |
+
"learning_rate": 1.9212698412698414e-05,
|
| 164 |
+
"loss": 1.4184,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.8583690987124464,
|
| 169 |
+
"grad_norm": 29.049049377441406,
|
| 170 |
+
"learning_rate": 1.9053968253968257e-05,
|
| 171 |
+
"loss": 1.5536,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 3.0,
|
| 176 |
+
"grad_norm": 23.547271728515625,
|
| 177 |
+
"learning_rate": 1.8895238095238096e-05,
|
| 178 |
+
"loss": 1.4959,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 3.0,
|
| 183 |
+
"eval_accuracy": 0.8261455525606469,
|
| 184 |
+
"eval_loss": 0.5115092992782593,
|
| 185 |
+
"eval_macro_f1": 0.8247734932614236,
|
| 186 |
+
"eval_runtime": 1.4826,
|
| 187 |
+
"eval_samples_per_second": 500.472,
|
| 188 |
+
"eval_steps_per_second": 31.701,
|
| 189 |
+
"eval_weighted_f1": 0.8253364622584316,
|
| 190 |
+
"step": 525
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 3.1430615164520743,
|
| 194 |
+
"grad_norm": 41.176551818847656,
|
| 195 |
+
"learning_rate": 1.8736507936507936e-05,
|
| 196 |
+
"loss": 1.1319,
|
| 197 |
+
"step": 550
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.2861230329041486,
|
| 201 |
+
"grad_norm": 69.8743667602539,
|
| 202 |
+
"learning_rate": 1.857777777777778e-05,
|
| 203 |
+
"loss": 1.1486,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.429184549356223,
|
| 208 |
+
"grad_norm": 95.68185424804688,
|
| 209 |
+
"learning_rate": 1.8419047619047622e-05,
|
| 210 |
+
"loss": 1.0582,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.5722460658082977,
|
| 215 |
+
"grad_norm": 26.72191047668457,
|
| 216 |
+
"learning_rate": 1.8260317460317462e-05,
|
| 217 |
+
"loss": 1.0875,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.715307582260372,
|
| 222 |
+
"grad_norm": 35.186214447021484,
|
| 223 |
+
"learning_rate": 1.81015873015873e-05,
|
| 224 |
+
"loss": 0.9114,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.8583690987124464,
|
| 229 |
+
"grad_norm": 21.028076171875,
|
| 230 |
+
"learning_rate": 1.7942857142857144e-05,
|
| 231 |
+
"loss": 1.0831,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 4.0,
|
| 236 |
+
"grad_norm": 74.53148651123047,
|
| 237 |
+
"learning_rate": 1.7784126984126987e-05,
|
| 238 |
+
"loss": 0.9946,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 4.0,
|
| 243 |
+
"eval_accuracy": 0.8126684636118598,
|
| 244 |
+
"eval_loss": 0.525178074836731,
|
| 245 |
+
"eval_macro_f1": 0.8126293393650315,
|
| 246 |
+
"eval_runtime": 1.6612,
|
| 247 |
+
"eval_samples_per_second": 446.661,
|
| 248 |
+
"eval_steps_per_second": 28.293,
|
| 249 |
+
"eval_weighted_f1": 0.8125059465433461,
|
| 250 |
+
"step": 700
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 4.143061516452074,
|
| 254 |
+
"grad_norm": 12.17982006072998,
|
| 255 |
+
"learning_rate": 1.7625396825396827e-05,
|
| 256 |
+
"loss": 0.535,
|
| 257 |
+
"step": 725
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.286123032904149,
|
| 261 |
+
"grad_norm": 29.905244827270508,
|
| 262 |
+
"learning_rate": 1.7466666666666667e-05,
|
| 263 |
+
"loss": 0.8032,
|
| 264 |
+
"step": 750
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.429184549356223,
|
| 268 |
+
"grad_norm": 63.210670471191406,
|
| 269 |
+
"learning_rate": 1.730793650793651e-05,
|
| 270 |
+
"loss": 0.756,
|
| 271 |
+
"step": 775
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.572246065808297,
|
| 275 |
+
"grad_norm": 47.123619079589844,
|
| 276 |
+
"learning_rate": 1.7149206349206353e-05,
|
| 277 |
+
"loss": 0.8204,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.715307582260372,
|
| 282 |
+
"grad_norm": 28.559167861938477,
|
| 283 |
+
"learning_rate": 1.6990476190476192e-05,
|
| 284 |
+
"loss": 0.8945,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.858369098712446,
|
| 289 |
+
"grad_norm": 55.45006561279297,
|
| 290 |
+
"learning_rate": 1.6831746031746032e-05,
|
| 291 |
+
"loss": 0.7219,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.0,
|
| 296 |
+
"grad_norm": 40.08330154418945,
|
| 297 |
+
"learning_rate": 1.6673015873015875e-05,
|
| 298 |
+
"loss": 0.7728,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 5.0,
|
| 303 |
+
"eval_accuracy": 0.8261455525606469,
|
| 304 |
+
"eval_loss": 0.5565875768661499,
|
| 305 |
+
"eval_macro_f1": 0.824687481830339,
|
| 306 |
+
"eval_runtime": 1.4901,
|
| 307 |
+
"eval_samples_per_second": 497.948,
|
| 308 |
+
"eval_steps_per_second": 31.541,
|
| 309 |
+
"eval_weighted_f1": 0.824909647320121,
|
| 310 |
+
"step": 875
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.143061516452074,
|
| 314 |
+
"grad_norm": 7.73151159286499,
|
| 315 |
+
"learning_rate": 1.6514285714285714e-05,
|
| 316 |
+
"loss": 0.72,
|
| 317 |
+
"step": 900
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 5.286123032904149,
|
| 321 |
+
"grad_norm": 7.84256649017334,
|
| 322 |
+
"learning_rate": 1.6355555555555557e-05,
|
| 323 |
+
"loss": 0.6782,
|
| 324 |
+
"step": 925
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 5.429184549356223,
|
| 328 |
+
"grad_norm": 21.195018768310547,
|
| 329 |
+
"learning_rate": 1.6196825396825397e-05,
|
| 330 |
+
"loss": 0.5648,
|
| 331 |
+
"step": 950
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.572246065808297,
|
| 335 |
+
"grad_norm": 2.495529890060425,
|
| 336 |
+
"learning_rate": 1.603809523809524e-05,
|
| 337 |
+
"loss": 0.4648,
|
| 338 |
+
"step": 975
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.715307582260372,
|
| 342 |
+
"grad_norm": 13.436622619628906,
|
| 343 |
+
"learning_rate": 1.587936507936508e-05,
|
| 344 |
+
"loss": 0.3205,
|
| 345 |
+
"step": 1000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.858369098712446,
|
| 349 |
+
"grad_norm": 8.621908187866211,
|
| 350 |
+
"learning_rate": 1.5720634920634923e-05,
|
| 351 |
+
"loss": 0.4704,
|
| 352 |
+
"step": 1025
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 6.0,
|
| 356 |
+
"grad_norm": 50.76936721801758,
|
| 357 |
+
"learning_rate": 1.5561904761904762e-05,
|
| 358 |
+
"loss": 0.5928,
|
| 359 |
+
"step": 1050
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 6.0,
|
| 363 |
+
"eval_accuracy": 0.8571428571428571,
|
| 364 |
+
"eval_loss": 0.6029855608940125,
|
| 365 |
+
"eval_macro_f1": 0.8572722694614372,
|
| 366 |
+
"eval_runtime": 1.4573,
|
| 367 |
+
"eval_samples_per_second": 509.151,
|
| 368 |
+
"eval_steps_per_second": 32.251,
|
| 369 |
+
"eval_weighted_f1": 0.8571712162304597,
|
| 370 |
+
"step": 1050
|
| 371 |
+
}
|
| 372 |
+
],
|
| 373 |
+
"logging_steps": 25,
|
| 374 |
+
"max_steps": 3500,
|
| 375 |
+
"num_input_tokens_seen": 0,
|
| 376 |
+
"num_train_epochs": 20,
|
| 377 |
+
"save_steps": 500,
|
| 378 |
+
"stateful_callbacks": {
|
| 379 |
+
"EarlyStoppingCallback": {
|
| 380 |
+
"args": {
|
| 381 |
+
"early_stopping_patience": 3,
|
| 382 |
+
"early_stopping_threshold": 0.0
|
| 383 |
+
},
|
| 384 |
+
"attributes": {
|
| 385 |
+
"early_stopping_patience_counter": 0
|
| 386 |
+
}
|
| 387 |
+
},
|
| 388 |
+
"TrainerControl": {
|
| 389 |
+
"args": {
|
| 390 |
+
"should_epoch_stop": false,
|
| 391 |
+
"should_evaluate": false,
|
| 392 |
+
"should_log": false,
|
| 393 |
+
"should_save": true,
|
| 394 |
+
"should_training_stop": false
|
| 395 |
+
},
|
| 396 |
+
"attributes": {}
|
| 397 |
+
}
|
| 398 |
+
},
|
| 399 |
+
"total_flos": 0.0,
|
| 400 |
+
"train_batch_size": 8,
|
| 401 |
+
"trial_name": null,
|
| 402 |
+
"trial_params": null
|
| 403 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/trainer_state.json
ADDED
|
@@ -0,0 +1,583 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 1050,
|
| 3 |
+
"best_metric": 0.8572722694614372,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1050",
|
| 5 |
+
"epoch": 9.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1575,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.1430615164520744,
|
| 14 |
+
"grad_norm": 7.234732627868652,
|
| 15 |
+
"learning_rate": 1.3714285714285717e-06,
|
| 16 |
+
"loss": 4.4028,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.2861230329041488,
|
| 21 |
+
"grad_norm": 6.111971855163574,
|
| 22 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 23 |
+
"loss": 4.4214,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.4291845493562232,
|
| 28 |
+
"grad_norm": 6.735830783843994,
|
| 29 |
+
"learning_rate": 4.228571428571429e-06,
|
| 30 |
+
"loss": 4.3905,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5722460658082976,
|
| 35 |
+
"grad_norm": 6.769274711608887,
|
| 36 |
+
"learning_rate": 5.6571428571428576e-06,
|
| 37 |
+
"loss": 4.4068,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.7153075822603719,
|
| 42 |
+
"grad_norm": 9.92778205871582,
|
| 43 |
+
"learning_rate": 7.085714285714286e-06,
|
| 44 |
+
"loss": 4.3451,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8583690987124464,
|
| 49 |
+
"grad_norm": 8.201742172241211,
|
| 50 |
+
"learning_rate": 8.514285714285715e-06,
|
| 51 |
+
"loss": 4.2443,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 1.0,
|
| 56 |
+
"grad_norm": 11.032033920288086,
|
| 57 |
+
"learning_rate": 9.942857142857145e-06,
|
| 58 |
+
"loss": 4.0712,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5876010781671159,
|
| 64 |
+
"eval_loss": 0.9521220922470093,
|
| 65 |
+
"eval_macro_f1": 0.5740345789141216,
|
| 66 |
+
"eval_runtime": 1.4608,
|
| 67 |
+
"eval_samples_per_second": 507.927,
|
| 68 |
+
"eval_steps_per_second": 32.173,
|
| 69 |
+
"eval_weighted_f1": 0.5774423182822452,
|
| 70 |
+
"step": 175
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.1430615164520743,
|
| 74 |
+
"grad_norm": 19.701786041259766,
|
| 75 |
+
"learning_rate": 1.1371428571428573e-05,
|
| 76 |
+
"loss": 3.7206,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.2861230329041489,
|
| 81 |
+
"grad_norm": 33.54481887817383,
|
| 82 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 83 |
+
"loss": 3.4389,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.4291845493562232,
|
| 88 |
+
"grad_norm": 25.37718963623047,
|
| 89 |
+
"learning_rate": 1.422857142857143e-05,
|
| 90 |
+
"loss": 3.0285,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.5722460658082977,
|
| 95 |
+
"grad_norm": 23.7049617767334,
|
| 96 |
+
"learning_rate": 1.5657142857142856e-05,
|
| 97 |
+
"loss": 2.5281,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.715307582260372,
|
| 102 |
+
"grad_norm": 34.06275177001953,
|
| 103 |
+
"learning_rate": 1.708571428571429e-05,
|
| 104 |
+
"loss": 2.4805,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.8583690987124464,
|
| 109 |
+
"grad_norm": 34.69179153442383,
|
| 110 |
+
"learning_rate": 1.8514285714285716e-05,
|
| 111 |
+
"loss": 2.0691,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.0,
|
| 116 |
+
"grad_norm": 12.415122032165527,
|
| 117 |
+
"learning_rate": 1.9942857142857142e-05,
|
| 118 |
+
"loss": 2.0887,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.0,
|
| 123 |
+
"eval_accuracy": 0.7991913746630728,
|
| 124 |
+
"eval_loss": 0.5084040760993958,
|
| 125 |
+
"eval_macro_f1": 0.7966981355469368,
|
| 126 |
+
"eval_runtime": 1.4012,
|
| 127 |
+
"eval_samples_per_second": 529.561,
|
| 128 |
+
"eval_steps_per_second": 33.544,
|
| 129 |
+
"eval_weighted_f1": 0.7971849691058734,
|
| 130 |
+
"step": 350
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.1430615164520743,
|
| 134 |
+
"grad_norm": 30.174409866333008,
|
| 135 |
+
"learning_rate": 1.984761904761905e-05,
|
| 136 |
+
"loss": 1.592,
|
| 137 |
+
"step": 375
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.2861230329041486,
|
| 141 |
+
"grad_norm": 38.79440689086914,
|
| 142 |
+
"learning_rate": 1.968888888888889e-05,
|
| 143 |
+
"loss": 1.5487,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.429184549356223,
|
| 148 |
+
"grad_norm": 52.02887725830078,
|
| 149 |
+
"learning_rate": 1.953015873015873e-05,
|
| 150 |
+
"loss": 1.8353,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.5722460658082977,
|
| 155 |
+
"grad_norm": 36.64469528198242,
|
| 156 |
+
"learning_rate": 1.937142857142857e-05,
|
| 157 |
+
"loss": 1.7984,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.715307582260372,
|
| 162 |
+
"grad_norm": 44.32551956176758,
|
| 163 |
+
"learning_rate": 1.9212698412698414e-05,
|
| 164 |
+
"loss": 1.4184,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.8583690987124464,
|
| 169 |
+
"grad_norm": 29.049049377441406,
|
| 170 |
+
"learning_rate": 1.9053968253968257e-05,
|
| 171 |
+
"loss": 1.5536,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 3.0,
|
| 176 |
+
"grad_norm": 23.547271728515625,
|
| 177 |
+
"learning_rate": 1.8895238095238096e-05,
|
| 178 |
+
"loss": 1.4959,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 3.0,
|
| 183 |
+
"eval_accuracy": 0.8261455525606469,
|
| 184 |
+
"eval_loss": 0.5115092992782593,
|
| 185 |
+
"eval_macro_f1": 0.8247734932614236,
|
| 186 |
+
"eval_runtime": 1.4826,
|
| 187 |
+
"eval_samples_per_second": 500.472,
|
| 188 |
+
"eval_steps_per_second": 31.701,
|
| 189 |
+
"eval_weighted_f1": 0.8253364622584316,
|
| 190 |
+
"step": 525
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 3.1430615164520743,
|
| 194 |
+
"grad_norm": 41.176551818847656,
|
| 195 |
+
"learning_rate": 1.8736507936507936e-05,
|
| 196 |
+
"loss": 1.1319,
|
| 197 |
+
"step": 550
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.2861230329041486,
|
| 201 |
+
"grad_norm": 69.8743667602539,
|
| 202 |
+
"learning_rate": 1.857777777777778e-05,
|
| 203 |
+
"loss": 1.1486,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.429184549356223,
|
| 208 |
+
"grad_norm": 95.68185424804688,
|
| 209 |
+
"learning_rate": 1.8419047619047622e-05,
|
| 210 |
+
"loss": 1.0582,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.5722460658082977,
|
| 215 |
+
"grad_norm": 26.72191047668457,
|
| 216 |
+
"learning_rate": 1.8260317460317462e-05,
|
| 217 |
+
"loss": 1.0875,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.715307582260372,
|
| 222 |
+
"grad_norm": 35.186214447021484,
|
| 223 |
+
"learning_rate": 1.81015873015873e-05,
|
| 224 |
+
"loss": 0.9114,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.8583690987124464,
|
| 229 |
+
"grad_norm": 21.028076171875,
|
| 230 |
+
"learning_rate": 1.7942857142857144e-05,
|
| 231 |
+
"loss": 1.0831,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 4.0,
|
| 236 |
+
"grad_norm": 74.53148651123047,
|
| 237 |
+
"learning_rate": 1.7784126984126987e-05,
|
| 238 |
+
"loss": 0.9946,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 4.0,
|
| 243 |
+
"eval_accuracy": 0.8126684636118598,
|
| 244 |
+
"eval_loss": 0.525178074836731,
|
| 245 |
+
"eval_macro_f1": 0.8126293393650315,
|
| 246 |
+
"eval_runtime": 1.6612,
|
| 247 |
+
"eval_samples_per_second": 446.661,
|
| 248 |
+
"eval_steps_per_second": 28.293,
|
| 249 |
+
"eval_weighted_f1": 0.8125059465433461,
|
| 250 |
+
"step": 700
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 4.143061516452074,
|
| 254 |
+
"grad_norm": 12.17982006072998,
|
| 255 |
+
"learning_rate": 1.7625396825396827e-05,
|
| 256 |
+
"loss": 0.535,
|
| 257 |
+
"step": 725
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.286123032904149,
|
| 261 |
+
"grad_norm": 29.905244827270508,
|
| 262 |
+
"learning_rate": 1.7466666666666667e-05,
|
| 263 |
+
"loss": 0.8032,
|
| 264 |
+
"step": 750
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.429184549356223,
|
| 268 |
+
"grad_norm": 63.210670471191406,
|
| 269 |
+
"learning_rate": 1.730793650793651e-05,
|
| 270 |
+
"loss": 0.756,
|
| 271 |
+
"step": 775
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.572246065808297,
|
| 275 |
+
"grad_norm": 47.123619079589844,
|
| 276 |
+
"learning_rate": 1.7149206349206353e-05,
|
| 277 |
+
"loss": 0.8204,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.715307582260372,
|
| 282 |
+
"grad_norm": 28.559167861938477,
|
| 283 |
+
"learning_rate": 1.6990476190476192e-05,
|
| 284 |
+
"loss": 0.8945,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.858369098712446,
|
| 289 |
+
"grad_norm": 55.45006561279297,
|
| 290 |
+
"learning_rate": 1.6831746031746032e-05,
|
| 291 |
+
"loss": 0.7219,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.0,
|
| 296 |
+
"grad_norm": 40.08330154418945,
|
| 297 |
+
"learning_rate": 1.6673015873015875e-05,
|
| 298 |
+
"loss": 0.7728,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 5.0,
|
| 303 |
+
"eval_accuracy": 0.8261455525606469,
|
| 304 |
+
"eval_loss": 0.5565875768661499,
|
| 305 |
+
"eval_macro_f1": 0.824687481830339,
|
| 306 |
+
"eval_runtime": 1.4901,
|
| 307 |
+
"eval_samples_per_second": 497.948,
|
| 308 |
+
"eval_steps_per_second": 31.541,
|
| 309 |
+
"eval_weighted_f1": 0.824909647320121,
|
| 310 |
+
"step": 875
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.143061516452074,
|
| 314 |
+
"grad_norm": 7.73151159286499,
|
| 315 |
+
"learning_rate": 1.6514285714285714e-05,
|
| 316 |
+
"loss": 0.72,
|
| 317 |
+
"step": 900
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 5.286123032904149,
|
| 321 |
+
"grad_norm": 7.84256649017334,
|
| 322 |
+
"learning_rate": 1.6355555555555557e-05,
|
| 323 |
+
"loss": 0.6782,
|
| 324 |
+
"step": 925
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 5.429184549356223,
|
| 328 |
+
"grad_norm": 21.195018768310547,
|
| 329 |
+
"learning_rate": 1.6196825396825397e-05,
|
| 330 |
+
"loss": 0.5648,
|
| 331 |
+
"step": 950
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.572246065808297,
|
| 335 |
+
"grad_norm": 2.495529890060425,
|
| 336 |
+
"learning_rate": 1.603809523809524e-05,
|
| 337 |
+
"loss": 0.4648,
|
| 338 |
+
"step": 975
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.715307582260372,
|
| 342 |
+
"grad_norm": 13.436622619628906,
|
| 343 |
+
"learning_rate": 1.587936507936508e-05,
|
| 344 |
+
"loss": 0.3205,
|
| 345 |
+
"step": 1000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.858369098712446,
|
| 349 |
+
"grad_norm": 8.621908187866211,
|
| 350 |
+
"learning_rate": 1.5720634920634923e-05,
|
| 351 |
+
"loss": 0.4704,
|
| 352 |
+
"step": 1025
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 6.0,
|
| 356 |
+
"grad_norm": 50.76936721801758,
|
| 357 |
+
"learning_rate": 1.5561904761904762e-05,
|
| 358 |
+
"loss": 0.5928,
|
| 359 |
+
"step": 1050
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 6.0,
|
| 363 |
+
"eval_accuracy": 0.8571428571428571,
|
| 364 |
+
"eval_loss": 0.6029855608940125,
|
| 365 |
+
"eval_macro_f1": 0.8572722694614372,
|
| 366 |
+
"eval_runtime": 1.4573,
|
| 367 |
+
"eval_samples_per_second": 509.151,
|
| 368 |
+
"eval_steps_per_second": 32.251,
|
| 369 |
+
"eval_weighted_f1": 0.8571712162304597,
|
| 370 |
+
"step": 1050
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 6.143061516452074,
|
| 374 |
+
"grad_norm": 21.049192428588867,
|
| 375 |
+
"learning_rate": 1.5403174603174605e-05,
|
| 376 |
+
"loss": 0.3161,
|
| 377 |
+
"step": 1075
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 6.286123032904149,
|
| 381 |
+
"grad_norm": 90.82843780517578,
|
| 382 |
+
"learning_rate": 1.5244444444444447e-05,
|
| 383 |
+
"loss": 0.4663,
|
| 384 |
+
"step": 1100
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 6.429184549356223,
|
| 388 |
+
"grad_norm": 0.8898224234580994,
|
| 389 |
+
"learning_rate": 1.5085714285714288e-05,
|
| 390 |
+
"loss": 0.3449,
|
| 391 |
+
"step": 1125
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 6.572246065808297,
|
| 395 |
+
"grad_norm": 23.016225814819336,
|
| 396 |
+
"learning_rate": 1.4926984126984128e-05,
|
| 397 |
+
"loss": 0.3313,
|
| 398 |
+
"step": 1150
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 6.715307582260372,
|
| 402 |
+
"grad_norm": 33.2369384765625,
|
| 403 |
+
"learning_rate": 1.4768253968253969e-05,
|
| 404 |
+
"loss": 0.4634,
|
| 405 |
+
"step": 1175
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 6.858369098712446,
|
| 409 |
+
"grad_norm": 77.57447814941406,
|
| 410 |
+
"learning_rate": 1.4609523809523812e-05,
|
| 411 |
+
"loss": 0.4808,
|
| 412 |
+
"step": 1200
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 7.0,
|
| 416 |
+
"grad_norm": 0.5025469660758972,
|
| 417 |
+
"learning_rate": 1.4450793650793653e-05,
|
| 418 |
+
"loss": 0.4547,
|
| 419 |
+
"step": 1225
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 7.0,
|
| 423 |
+
"eval_accuracy": 0.8477088948787062,
|
| 424 |
+
"eval_loss": 0.663605272769928,
|
| 425 |
+
"eval_macro_f1": 0.8476779339505424,
|
| 426 |
+
"eval_runtime": 1.4948,
|
| 427 |
+
"eval_samples_per_second": 496.39,
|
| 428 |
+
"eval_steps_per_second": 31.442,
|
| 429 |
+
"eval_weighted_f1": 0.8476903271952528,
|
| 430 |
+
"step": 1225
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 7.143061516452074,
|
| 434 |
+
"grad_norm": 93.99462890625,
|
| 435 |
+
"learning_rate": 1.4292063492063493e-05,
|
| 436 |
+
"loss": 0.3359,
|
| 437 |
+
"step": 1250
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 7.286123032904149,
|
| 441 |
+
"grad_norm": 17.987869262695312,
|
| 442 |
+
"learning_rate": 1.4133333333333334e-05,
|
| 443 |
+
"loss": 0.3637,
|
| 444 |
+
"step": 1275
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 7.429184549356223,
|
| 448 |
+
"grad_norm": 0.4351023733615875,
|
| 449 |
+
"learning_rate": 1.3974603174603177e-05,
|
| 450 |
+
"loss": 0.3333,
|
| 451 |
+
"step": 1300
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 7.572246065808297,
|
| 455 |
+
"grad_norm": 1.98640775680542,
|
| 456 |
+
"learning_rate": 1.3815873015873018e-05,
|
| 457 |
+
"loss": 0.257,
|
| 458 |
+
"step": 1325
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 7.715307582260372,
|
| 462 |
+
"grad_norm": 0.4878084659576416,
|
| 463 |
+
"learning_rate": 1.3657142857142858e-05,
|
| 464 |
+
"loss": 0.247,
|
| 465 |
+
"step": 1350
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 7.858369098712446,
|
| 469 |
+
"grad_norm": 207.392578125,
|
| 470 |
+
"learning_rate": 1.34984126984127e-05,
|
| 471 |
+
"loss": 0.3449,
|
| 472 |
+
"step": 1375
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 8.0,
|
| 476 |
+
"grad_norm": 86.43585205078125,
|
| 477 |
+
"learning_rate": 1.333968253968254e-05,
|
| 478 |
+
"loss": 0.3404,
|
| 479 |
+
"step": 1400
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 8.0,
|
| 483 |
+
"eval_accuracy": 0.8530997304582211,
|
| 484 |
+
"eval_loss": 0.7471036911010742,
|
| 485 |
+
"eval_macro_f1": 0.8532465471749783,
|
| 486 |
+
"eval_runtime": 1.4256,
|
| 487 |
+
"eval_samples_per_second": 520.491,
|
| 488 |
+
"eval_steps_per_second": 32.969,
|
| 489 |
+
"eval_weighted_f1": 0.8530844508546166,
|
| 490 |
+
"step": 1400
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 8.143061516452075,
|
| 494 |
+
"grad_norm": 1.6776268482208252,
|
| 495 |
+
"learning_rate": 1.318095238095238e-05,
|
| 496 |
+
"loss": 0.3292,
|
| 497 |
+
"step": 1425
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 8.286123032904149,
|
| 501 |
+
"grad_norm": 0.23851832747459412,
|
| 502 |
+
"learning_rate": 1.3022222222222223e-05,
|
| 503 |
+
"loss": 0.3591,
|
| 504 |
+
"step": 1450
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 8.429184549356224,
|
| 508 |
+
"grad_norm": 26.57066535949707,
|
| 509 |
+
"learning_rate": 1.2863492063492065e-05,
|
| 510 |
+
"loss": 0.1781,
|
| 511 |
+
"step": 1475
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 8.572246065808297,
|
| 515 |
+
"grad_norm": 99.14136505126953,
|
| 516 |
+
"learning_rate": 1.2704761904761906e-05,
|
| 517 |
+
"loss": 0.0926,
|
| 518 |
+
"step": 1500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 8.715307582260372,
|
| 522 |
+
"grad_norm": 0.21703723073005676,
|
| 523 |
+
"learning_rate": 1.2546031746031746e-05,
|
| 524 |
+
"loss": 0.2209,
|
| 525 |
+
"step": 1525
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 8.858369098712446,
|
| 529 |
+
"grad_norm": 7.657599925994873,
|
| 530 |
+
"learning_rate": 1.2387301587301589e-05,
|
| 531 |
+
"loss": 0.2332,
|
| 532 |
+
"step": 1550
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 9.0,
|
| 536 |
+
"grad_norm": 0.15074600279331207,
|
| 537 |
+
"learning_rate": 1.222857142857143e-05,
|
| 538 |
+
"loss": 0.2979,
|
| 539 |
+
"step": 1575
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 9.0,
|
| 543 |
+
"eval_accuracy": 0.839622641509434,
|
| 544 |
+
"eval_loss": 0.8274786472320557,
|
| 545 |
+
"eval_macro_f1": 0.8391995470451933,
|
| 546 |
+
"eval_runtime": 1.4463,
|
| 547 |
+
"eval_samples_per_second": 513.034,
|
| 548 |
+
"eval_steps_per_second": 32.497,
|
| 549 |
+
"eval_weighted_f1": 0.8392718263523362,
|
| 550 |
+
"step": 1575
|
| 551 |
+
}
|
| 552 |
+
],
|
| 553 |
+
"logging_steps": 25,
|
| 554 |
+
"max_steps": 3500,
|
| 555 |
+
"num_input_tokens_seen": 0,
|
| 556 |
+
"num_train_epochs": 20,
|
| 557 |
+
"save_steps": 500,
|
| 558 |
+
"stateful_callbacks": {
|
| 559 |
+
"EarlyStoppingCallback": {
|
| 560 |
+
"args": {
|
| 561 |
+
"early_stopping_patience": 3,
|
| 562 |
+
"early_stopping_threshold": 0.0
|
| 563 |
+
},
|
| 564 |
+
"attributes": {
|
| 565 |
+
"early_stopping_patience_counter": 3
|
| 566 |
+
}
|
| 567 |
+
},
|
| 568 |
+
"TrainerControl": {
|
| 569 |
+
"args": {
|
| 570 |
+
"should_epoch_stop": false,
|
| 571 |
+
"should_evaluate": false,
|
| 572 |
+
"should_log": false,
|
| 573 |
+
"should_save": true,
|
| 574 |
+
"should_training_stop": true
|
| 575 |
+
},
|
| 576 |
+
"attributes": {}
|
| 577 |
+
}
|
| 578 |
+
},
|
| 579 |
+
"total_flos": 0.0,
|
| 580 |
+
"train_batch_size": 8,
|
| 581 |
+
"trial_name": null,
|
| 582 |
+
"trial_params": null
|
| 583 |
+
}
|
ood-vifactcheck-phobert-topic-domain-05-s42-20260814T115412Z-00ca7096/trainer/checkpoint-1575/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/trainer_state.json
ADDED
|
@@ -0,0 +1,477 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 736,
|
| 3 |
+
"best_metric": 0.8546408139099864,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736",
|
| 5 |
+
"epoch": 7.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1288,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.1364256480218281,
|
| 14 |
+
"grad_norm": 7.697149276733398,
|
| 15 |
+
"learning_rate": 1.3043478260869566e-06,
|
| 16 |
+
"loss": 4.4145,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.2728512960436562,
|
| 21 |
+
"grad_norm": 7.441562652587891,
|
| 22 |
+
"learning_rate": 2.6630434782608698e-06,
|
| 23 |
+
"loss": 4.4112,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.4092769440654843,
|
| 28 |
+
"grad_norm": 9.718523025512695,
|
| 29 |
+
"learning_rate": 4.021739130434783e-06,
|
| 30 |
+
"loss": 4.3789,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5457025920873124,
|
| 35 |
+
"grad_norm": 9.041443824768066,
|
| 36 |
+
"learning_rate": 5.380434782608695e-06,
|
| 37 |
+
"loss": 4.3616,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6821282401091405,
|
| 42 |
+
"grad_norm": 10.699589729309082,
|
| 43 |
+
"learning_rate": 6.739130434782609e-06,
|
| 44 |
+
"loss": 4.3533,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8185538881309686,
|
| 49 |
+
"grad_norm": 6.946336269378662,
|
| 50 |
+
"learning_rate": 8.097826086956523e-06,
|
| 51 |
+
"loss": 4.2787,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.9549795361527967,
|
| 56 |
+
"grad_norm": 9.135581016540527,
|
| 57 |
+
"learning_rate": 9.456521739130436e-06,
|
| 58 |
+
"loss": 4.2136,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5748663101604278,
|
| 64 |
+
"eval_loss": 0.9590131044387817,
|
| 65 |
+
"eval_macro_f1": 0.5608447835536551,
|
| 66 |
+
"eval_runtime": 1.393,
|
| 67 |
+
"eval_samples_per_second": 536.973,
|
| 68 |
+
"eval_steps_per_second": 33.74,
|
| 69 |
+
"eval_weighted_f1": 0.563263122143691,
|
| 70 |
+
"step": 184
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.08731241473397,
|
| 74 |
+
"grad_norm": 32.14780807495117,
|
| 75 |
+
"learning_rate": 1.0815217391304349e-05,
|
| 76 |
+
"loss": 3.759,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.223738062755798,
|
| 81 |
+
"grad_norm": 57.07328414916992,
|
| 82 |
+
"learning_rate": 1.2173913043478263e-05,
|
| 83 |
+
"loss": 3.4549,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.360163710777626,
|
| 88 |
+
"grad_norm": 32.00222396850586,
|
| 89 |
+
"learning_rate": 1.3532608695652174e-05,
|
| 90 |
+
"loss": 3.1452,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.4965893587994543,
|
| 95 |
+
"grad_norm": 38.3240966796875,
|
| 96 |
+
"learning_rate": 1.4891304347826087e-05,
|
| 97 |
+
"loss": 2.6407,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.6330150068212825,
|
| 102 |
+
"grad_norm": 45.32682418823242,
|
| 103 |
+
"learning_rate": 1.6250000000000002e-05,
|
| 104 |
+
"loss": 2.5324,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.7694406548431105,
|
| 109 |
+
"grad_norm": 44.54869842529297,
|
| 110 |
+
"learning_rate": 1.7608695652173915e-05,
|
| 111 |
+
"loss": 2.3648,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.9058663028649385,
|
| 116 |
+
"grad_norm": 43.33744812011719,
|
| 117 |
+
"learning_rate": 1.8967391304347828e-05,
|
| 118 |
+
"loss": 2.2979,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.0,
|
| 123 |
+
"eval_accuracy": 0.820855614973262,
|
| 124 |
+
"eval_loss": 0.49087801575660706,
|
| 125 |
+
"eval_macro_f1": 0.8214316341406643,
|
| 126 |
+
"eval_runtime": 1.4774,
|
| 127 |
+
"eval_samples_per_second": 506.282,
|
| 128 |
+
"eval_steps_per_second": 31.812,
|
| 129 |
+
"eval_weighted_f1": 0.8215498843080979,
|
| 130 |
+
"step": 368
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.038199181446112,
|
| 134 |
+
"grad_norm": 59.5373649597168,
|
| 135 |
+
"learning_rate": 1.9963768115942032e-05,
|
| 136 |
+
"loss": 2.1284,
|
| 137 |
+
"step": 375
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.17462482946794,
|
| 141 |
+
"grad_norm": 25.12555694580078,
|
| 142 |
+
"learning_rate": 1.9812801932367153e-05,
|
| 143 |
+
"loss": 1.7497,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.311050477489768,
|
| 148 |
+
"grad_norm": 58.68057632446289,
|
| 149 |
+
"learning_rate": 1.966183574879227e-05,
|
| 150 |
+
"loss": 1.7005,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.447476125511596,
|
| 155 |
+
"grad_norm": 30.604564666748047,
|
| 156 |
+
"learning_rate": 1.9510869565217393e-05,
|
| 157 |
+
"loss": 1.6901,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.5839017735334244,
|
| 162 |
+
"grad_norm": 26.262935638427734,
|
| 163 |
+
"learning_rate": 1.9359903381642515e-05,
|
| 164 |
+
"loss": 1.6531,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.720327421555252,
|
| 169 |
+
"grad_norm": 64.99190521240234,
|
| 170 |
+
"learning_rate": 1.9208937198067633e-05,
|
| 171 |
+
"loss": 1.8155,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.8567530695770804,
|
| 176 |
+
"grad_norm": 25.000621795654297,
|
| 177 |
+
"learning_rate": 1.9057971014492754e-05,
|
| 178 |
+
"loss": 1.4477,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.9931787175989086,
|
| 183 |
+
"grad_norm": 36.30158615112305,
|
| 184 |
+
"learning_rate": 1.8907004830917876e-05,
|
| 185 |
+
"loss": 1.4264,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 3.0,
|
| 190 |
+
"eval_accuracy": 0.8248663101604278,
|
| 191 |
+
"eval_loss": 0.4865352213382721,
|
| 192 |
+
"eval_macro_f1": 0.8248397547752239,
|
| 193 |
+
"eval_runtime": 1.3883,
|
| 194 |
+
"eval_samples_per_second": 538.794,
|
| 195 |
+
"eval_steps_per_second": 33.855,
|
| 196 |
+
"eval_weighted_f1": 0.8254318128956593,
|
| 197 |
+
"step": 552
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.1255115961800817,
|
| 201 |
+
"grad_norm": 21.72858238220215,
|
| 202 |
+
"learning_rate": 1.8756038647342997e-05,
|
| 203 |
+
"loss": 1.142,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.26193724420191,
|
| 208 |
+
"grad_norm": 37.139976501464844,
|
| 209 |
+
"learning_rate": 1.8605072463768115e-05,
|
| 210 |
+
"loss": 1.0364,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.398362892223738,
|
| 215 |
+
"grad_norm": 624.9979248046875,
|
| 216 |
+
"learning_rate": 1.8454106280193237e-05,
|
| 217 |
+
"loss": 1.1093,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.5347885402455663,
|
| 222 |
+
"grad_norm": 19.94325828552246,
|
| 223 |
+
"learning_rate": 1.830314009661836e-05,
|
| 224 |
+
"loss": 1.1695,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.6712141882673945,
|
| 229 |
+
"grad_norm": 64.63702392578125,
|
| 230 |
+
"learning_rate": 1.815217391304348e-05,
|
| 231 |
+
"loss": 1.2231,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.8076398362892223,
|
| 236 |
+
"grad_norm": 66.94290924072266,
|
| 237 |
+
"learning_rate": 1.8001207729468598e-05,
|
| 238 |
+
"loss": 0.9801,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.9440654843110505,
|
| 243 |
+
"grad_norm": 70.0299072265625,
|
| 244 |
+
"learning_rate": 1.785024154589372e-05,
|
| 245 |
+
"loss": 1.1132,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.0,
|
| 250 |
+
"eval_accuracy": 0.8556149732620321,
|
| 251 |
+
"eval_loss": 0.4646115303039551,
|
| 252 |
+
"eval_macro_f1": 0.8546408139099864,
|
| 253 |
+
"eval_runtime": 1.3874,
|
| 254 |
+
"eval_samples_per_second": 539.137,
|
| 255 |
+
"eval_steps_per_second": 33.876,
|
| 256 |
+
"eval_weighted_f1": 0.8548216342577015,
|
| 257 |
+
"step": 736
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.076398362892224,
|
| 261 |
+
"grad_norm": 11.400529861450195,
|
| 262 |
+
"learning_rate": 1.769927536231884e-05,
|
| 263 |
+
"loss": 0.7875,
|
| 264 |
+
"step": 750
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.212824010914052,
|
| 268 |
+
"grad_norm": 43.03860092163086,
|
| 269 |
+
"learning_rate": 1.7548309178743963e-05,
|
| 270 |
+
"loss": 0.6275,
|
| 271 |
+
"step": 775
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.34924965893588,
|
| 275 |
+
"grad_norm": 33.661338806152344,
|
| 276 |
+
"learning_rate": 1.7397342995169084e-05,
|
| 277 |
+
"loss": 0.8072,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.485675306957708,
|
| 282 |
+
"grad_norm": 123.51994323730469,
|
| 283 |
+
"learning_rate": 1.7246376811594206e-05,
|
| 284 |
+
"loss": 0.7709,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.622100954979536,
|
| 289 |
+
"grad_norm": 23.630298614501953,
|
| 290 |
+
"learning_rate": 1.7095410628019324e-05,
|
| 291 |
+
"loss": 0.6849,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 4.758526603001364,
|
| 296 |
+
"grad_norm": 45.84729766845703,
|
| 297 |
+
"learning_rate": 1.6944444444444446e-05,
|
| 298 |
+
"loss": 0.8171,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.894952251023192,
|
| 303 |
+
"grad_norm": 11.474125862121582,
|
| 304 |
+
"learning_rate": 1.6793478260869567e-05,
|
| 305 |
+
"loss": 0.8442,
|
| 306 |
+
"step": 900
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 5.0,
|
| 310 |
+
"eval_accuracy": 0.8542780748663101,
|
| 311 |
+
"eval_loss": 0.4942897856235504,
|
| 312 |
+
"eval_macro_f1": 0.8542635658914728,
|
| 313 |
+
"eval_runtime": 1.4791,
|
| 314 |
+
"eval_samples_per_second": 505.727,
|
| 315 |
+
"eval_steps_per_second": 31.777,
|
| 316 |
+
"eval_weighted_f1": 0.8542707261647693,
|
| 317 |
+
"step": 920
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 5.0272851296043655,
|
| 321 |
+
"grad_norm": 38.750240325927734,
|
| 322 |
+
"learning_rate": 1.664251207729469e-05,
|
| 323 |
+
"loss": 0.6554,
|
| 324 |
+
"step": 925
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 5.163710777626194,
|
| 328 |
+
"grad_norm": 27.236188888549805,
|
| 329 |
+
"learning_rate": 1.649154589371981e-05,
|
| 330 |
+
"loss": 0.5835,
|
| 331 |
+
"step": 950
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.300136425648022,
|
| 335 |
+
"grad_norm": 33.5499153137207,
|
| 336 |
+
"learning_rate": 1.634057971014493e-05,
|
| 337 |
+
"loss": 0.5701,
|
| 338 |
+
"step": 975
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.43656207366985,
|
| 342 |
+
"grad_norm": 60.61030578613281,
|
| 343 |
+
"learning_rate": 1.618961352657005e-05,
|
| 344 |
+
"loss": 0.5773,
|
| 345 |
+
"step": 1000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.572987721691678,
|
| 349 |
+
"grad_norm": 11.466594696044922,
|
| 350 |
+
"learning_rate": 1.603864734299517e-05,
|
| 351 |
+
"loss": 0.4889,
|
| 352 |
+
"step": 1025
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 5.709413369713506,
|
| 356 |
+
"grad_norm": 14.258838653564453,
|
| 357 |
+
"learning_rate": 1.5887681159420293e-05,
|
| 358 |
+
"loss": 0.468,
|
| 359 |
+
"step": 1050
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.845839017735334,
|
| 363 |
+
"grad_norm": 41.26432800292969,
|
| 364 |
+
"learning_rate": 1.5736714975845415e-05,
|
| 365 |
+
"loss": 0.5505,
|
| 366 |
+
"step": 1075
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 5.982264665757162,
|
| 370 |
+
"grad_norm": 23.275590896606445,
|
| 371 |
+
"learning_rate": 1.5585748792270533e-05,
|
| 372 |
+
"loss": 0.6078,
|
| 373 |
+
"step": 1100
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 6.0,
|
| 377 |
+
"eval_accuracy": 0.8422459893048129,
|
| 378 |
+
"eval_loss": 0.6127049326896667,
|
| 379 |
+
"eval_macro_f1": 0.8417041933876886,
|
| 380 |
+
"eval_runtime": 1.4791,
|
| 381 |
+
"eval_samples_per_second": 505.701,
|
| 382 |
+
"eval_steps_per_second": 31.775,
|
| 383 |
+
"eval_weighted_f1": 0.8418951055552963,
|
| 384 |
+
"step": 1104
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 6.114597544338336,
|
| 388 |
+
"grad_norm": 1.2740169763565063,
|
| 389 |
+
"learning_rate": 1.5434782608695654e-05,
|
| 390 |
+
"loss": 0.3993,
|
| 391 |
+
"step": 1125
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 6.251023192360163,
|
| 395 |
+
"grad_norm": 2.6488542556762695,
|
| 396 |
+
"learning_rate": 1.5283816425120776e-05,
|
| 397 |
+
"loss": 0.4053,
|
| 398 |
+
"step": 1150
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 6.387448840381992,
|
| 402 |
+
"grad_norm": 142.55467224121094,
|
| 403 |
+
"learning_rate": 1.5132850241545896e-05,
|
| 404 |
+
"loss": 0.4601,
|
| 405 |
+
"step": 1175
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 6.52387448840382,
|
| 409 |
+
"grad_norm": 296.3175964355469,
|
| 410 |
+
"learning_rate": 1.4981884057971017e-05,
|
| 411 |
+
"loss": 0.2716,
|
| 412 |
+
"step": 1200
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 6.660300136425648,
|
| 416 |
+
"grad_norm": 4.013828754425049,
|
| 417 |
+
"learning_rate": 1.4830917874396135e-05,
|
| 418 |
+
"loss": 0.4055,
|
| 419 |
+
"step": 1225
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 6.796725784447476,
|
| 423 |
+
"grad_norm": 31.934812545776367,
|
| 424 |
+
"learning_rate": 1.4679951690821257e-05,
|
| 425 |
+
"loss": 0.3264,
|
| 426 |
+
"step": 1250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 6.933151432469304,
|
| 430 |
+
"grad_norm": 22.202177047729492,
|
| 431 |
+
"learning_rate": 1.4528985507246378e-05,
|
| 432 |
+
"loss": 0.3853,
|
| 433 |
+
"step": 1275
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 7.0,
|
| 437 |
+
"eval_accuracy": 0.8542780748663101,
|
| 438 |
+
"eval_loss": 0.6389632821083069,
|
| 439 |
+
"eval_macro_f1": 0.8536836413924723,
|
| 440 |
+
"eval_runtime": 1.4279,
|
| 441 |
+
"eval_samples_per_second": 523.852,
|
| 442 |
+
"eval_steps_per_second": 32.916,
|
| 443 |
+
"eval_weighted_f1": 0.8538266988941591,
|
| 444 |
+
"step": 1288
|
| 445 |
+
}
|
| 446 |
+
],
|
| 447 |
+
"logging_steps": 25,
|
| 448 |
+
"max_steps": 3680,
|
| 449 |
+
"num_input_tokens_seen": 0,
|
| 450 |
+
"num_train_epochs": 20,
|
| 451 |
+
"save_steps": 500,
|
| 452 |
+
"stateful_callbacks": {
|
| 453 |
+
"EarlyStoppingCallback": {
|
| 454 |
+
"args": {
|
| 455 |
+
"early_stopping_patience": 3,
|
| 456 |
+
"early_stopping_threshold": 0.0
|
| 457 |
+
},
|
| 458 |
+
"attributes": {
|
| 459 |
+
"early_stopping_patience_counter": 3
|
| 460 |
+
}
|
| 461 |
+
},
|
| 462 |
+
"TrainerControl": {
|
| 463 |
+
"args": {
|
| 464 |
+
"should_epoch_stop": false,
|
| 465 |
+
"should_evaluate": false,
|
| 466 |
+
"should_log": false,
|
| 467 |
+
"should_save": true,
|
| 468 |
+
"should_training_stop": true
|
| 469 |
+
},
|
| 470 |
+
"attributes": {}
|
| 471 |
+
}
|
| 472 |
+
},
|
| 473 |
+
"total_flos": 0.0,
|
| 474 |
+
"train_batch_size": 8,
|
| 475 |
+
"trial_name": null,
|
| 476 |
+
"trial_params": null
|
| 477 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-1288/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/trainer_state.json
ADDED
|
@@ -0,0 +1,290 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 736,
|
| 3 |
+
"best_metric": 0.8546408139099864,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736",
|
| 5 |
+
"epoch": 4.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 736,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.1364256480218281,
|
| 14 |
+
"grad_norm": 7.697149276733398,
|
| 15 |
+
"learning_rate": 1.3043478260869566e-06,
|
| 16 |
+
"loss": 4.4145,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.2728512960436562,
|
| 21 |
+
"grad_norm": 7.441562652587891,
|
| 22 |
+
"learning_rate": 2.6630434782608698e-06,
|
| 23 |
+
"loss": 4.4112,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.4092769440654843,
|
| 28 |
+
"grad_norm": 9.718523025512695,
|
| 29 |
+
"learning_rate": 4.021739130434783e-06,
|
| 30 |
+
"loss": 4.3789,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5457025920873124,
|
| 35 |
+
"grad_norm": 9.041443824768066,
|
| 36 |
+
"learning_rate": 5.380434782608695e-06,
|
| 37 |
+
"loss": 4.3616,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6821282401091405,
|
| 42 |
+
"grad_norm": 10.699589729309082,
|
| 43 |
+
"learning_rate": 6.739130434782609e-06,
|
| 44 |
+
"loss": 4.3533,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.8185538881309686,
|
| 49 |
+
"grad_norm": 6.946336269378662,
|
| 50 |
+
"learning_rate": 8.097826086956523e-06,
|
| 51 |
+
"loss": 4.2787,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.9549795361527967,
|
| 56 |
+
"grad_norm": 9.135581016540527,
|
| 57 |
+
"learning_rate": 9.456521739130436e-06,
|
| 58 |
+
"loss": 4.2136,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5748663101604278,
|
| 64 |
+
"eval_loss": 0.9590131044387817,
|
| 65 |
+
"eval_macro_f1": 0.5608447835536551,
|
| 66 |
+
"eval_runtime": 1.393,
|
| 67 |
+
"eval_samples_per_second": 536.973,
|
| 68 |
+
"eval_steps_per_second": 33.74,
|
| 69 |
+
"eval_weighted_f1": 0.563263122143691,
|
| 70 |
+
"step": 184
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.08731241473397,
|
| 74 |
+
"grad_norm": 32.14780807495117,
|
| 75 |
+
"learning_rate": 1.0815217391304349e-05,
|
| 76 |
+
"loss": 3.759,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.223738062755798,
|
| 81 |
+
"grad_norm": 57.07328414916992,
|
| 82 |
+
"learning_rate": 1.2173913043478263e-05,
|
| 83 |
+
"loss": 3.4549,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.360163710777626,
|
| 88 |
+
"grad_norm": 32.00222396850586,
|
| 89 |
+
"learning_rate": 1.3532608695652174e-05,
|
| 90 |
+
"loss": 3.1452,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.4965893587994543,
|
| 95 |
+
"grad_norm": 38.3240966796875,
|
| 96 |
+
"learning_rate": 1.4891304347826087e-05,
|
| 97 |
+
"loss": 2.6407,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.6330150068212825,
|
| 102 |
+
"grad_norm": 45.32682418823242,
|
| 103 |
+
"learning_rate": 1.6250000000000002e-05,
|
| 104 |
+
"loss": 2.5324,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.7694406548431105,
|
| 109 |
+
"grad_norm": 44.54869842529297,
|
| 110 |
+
"learning_rate": 1.7608695652173915e-05,
|
| 111 |
+
"loss": 2.3648,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.9058663028649385,
|
| 116 |
+
"grad_norm": 43.33744812011719,
|
| 117 |
+
"learning_rate": 1.8967391304347828e-05,
|
| 118 |
+
"loss": 2.2979,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.0,
|
| 123 |
+
"eval_accuracy": 0.820855614973262,
|
| 124 |
+
"eval_loss": 0.49087801575660706,
|
| 125 |
+
"eval_macro_f1": 0.8214316341406643,
|
| 126 |
+
"eval_runtime": 1.4774,
|
| 127 |
+
"eval_samples_per_second": 506.282,
|
| 128 |
+
"eval_steps_per_second": 31.812,
|
| 129 |
+
"eval_weighted_f1": 0.8215498843080979,
|
| 130 |
+
"step": 368
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.038199181446112,
|
| 134 |
+
"grad_norm": 59.5373649597168,
|
| 135 |
+
"learning_rate": 1.9963768115942032e-05,
|
| 136 |
+
"loss": 2.1284,
|
| 137 |
+
"step": 375
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.17462482946794,
|
| 141 |
+
"grad_norm": 25.12555694580078,
|
| 142 |
+
"learning_rate": 1.9812801932367153e-05,
|
| 143 |
+
"loss": 1.7497,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.311050477489768,
|
| 148 |
+
"grad_norm": 58.68057632446289,
|
| 149 |
+
"learning_rate": 1.966183574879227e-05,
|
| 150 |
+
"loss": 1.7005,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.447476125511596,
|
| 155 |
+
"grad_norm": 30.604564666748047,
|
| 156 |
+
"learning_rate": 1.9510869565217393e-05,
|
| 157 |
+
"loss": 1.6901,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.5839017735334244,
|
| 162 |
+
"grad_norm": 26.262935638427734,
|
| 163 |
+
"learning_rate": 1.9359903381642515e-05,
|
| 164 |
+
"loss": 1.6531,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.720327421555252,
|
| 169 |
+
"grad_norm": 64.99190521240234,
|
| 170 |
+
"learning_rate": 1.9208937198067633e-05,
|
| 171 |
+
"loss": 1.8155,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.8567530695770804,
|
| 176 |
+
"grad_norm": 25.000621795654297,
|
| 177 |
+
"learning_rate": 1.9057971014492754e-05,
|
| 178 |
+
"loss": 1.4477,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.9931787175989086,
|
| 183 |
+
"grad_norm": 36.30158615112305,
|
| 184 |
+
"learning_rate": 1.8907004830917876e-05,
|
| 185 |
+
"loss": 1.4264,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 3.0,
|
| 190 |
+
"eval_accuracy": 0.8248663101604278,
|
| 191 |
+
"eval_loss": 0.4865352213382721,
|
| 192 |
+
"eval_macro_f1": 0.8248397547752239,
|
| 193 |
+
"eval_runtime": 1.3883,
|
| 194 |
+
"eval_samples_per_second": 538.794,
|
| 195 |
+
"eval_steps_per_second": 33.855,
|
| 196 |
+
"eval_weighted_f1": 0.8254318128956593,
|
| 197 |
+
"step": 552
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 3.1255115961800817,
|
| 201 |
+
"grad_norm": 21.72858238220215,
|
| 202 |
+
"learning_rate": 1.8756038647342997e-05,
|
| 203 |
+
"loss": 1.142,
|
| 204 |
+
"step": 575
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.26193724420191,
|
| 208 |
+
"grad_norm": 37.139976501464844,
|
| 209 |
+
"learning_rate": 1.8605072463768115e-05,
|
| 210 |
+
"loss": 1.0364,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.398362892223738,
|
| 215 |
+
"grad_norm": 624.9979248046875,
|
| 216 |
+
"learning_rate": 1.8454106280193237e-05,
|
| 217 |
+
"loss": 1.1093,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.5347885402455663,
|
| 222 |
+
"grad_norm": 19.94325828552246,
|
| 223 |
+
"learning_rate": 1.830314009661836e-05,
|
| 224 |
+
"loss": 1.1695,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.6712141882673945,
|
| 229 |
+
"grad_norm": 64.63702392578125,
|
| 230 |
+
"learning_rate": 1.815217391304348e-05,
|
| 231 |
+
"loss": 1.2231,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.8076398362892223,
|
| 236 |
+
"grad_norm": 66.94290924072266,
|
| 237 |
+
"learning_rate": 1.8001207729468598e-05,
|
| 238 |
+
"loss": 0.9801,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.9440654843110505,
|
| 243 |
+
"grad_norm": 70.0299072265625,
|
| 244 |
+
"learning_rate": 1.785024154589372e-05,
|
| 245 |
+
"loss": 1.1132,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.0,
|
| 250 |
+
"eval_accuracy": 0.8556149732620321,
|
| 251 |
+
"eval_loss": 0.4646115303039551,
|
| 252 |
+
"eval_macro_f1": 0.8546408139099864,
|
| 253 |
+
"eval_runtime": 1.3874,
|
| 254 |
+
"eval_samples_per_second": 539.137,
|
| 255 |
+
"eval_steps_per_second": 33.876,
|
| 256 |
+
"eval_weighted_f1": 0.8548216342577015,
|
| 257 |
+
"step": 736
|
| 258 |
+
}
|
| 259 |
+
],
|
| 260 |
+
"logging_steps": 25,
|
| 261 |
+
"max_steps": 3680,
|
| 262 |
+
"num_input_tokens_seen": 0,
|
| 263 |
+
"num_train_epochs": 20,
|
| 264 |
+
"save_steps": 500,
|
| 265 |
+
"stateful_callbacks": {
|
| 266 |
+
"EarlyStoppingCallback": {
|
| 267 |
+
"args": {
|
| 268 |
+
"early_stopping_patience": 3,
|
| 269 |
+
"early_stopping_threshold": 0.0
|
| 270 |
+
},
|
| 271 |
+
"attributes": {
|
| 272 |
+
"early_stopping_patience_counter": 0
|
| 273 |
+
}
|
| 274 |
+
},
|
| 275 |
+
"TrainerControl": {
|
| 276 |
+
"args": {
|
| 277 |
+
"should_epoch_stop": false,
|
| 278 |
+
"should_evaluate": false,
|
| 279 |
+
"should_log": false,
|
| 280 |
+
"should_save": true,
|
| 281 |
+
"should_training_stop": false
|
| 282 |
+
},
|
| 283 |
+
"attributes": {}
|
| 284 |
+
}
|
| 285 |
+
},
|
| 286 |
+
"total_flos": 0.0,
|
| 287 |
+
"train_batch_size": 8,
|
| 288 |
+
"trial_name": null,
|
| 289 |
+
"trial_params": null
|
| 290 |
+
}
|
ood-vifactcheck-phobert-topic-domain-06-s42-20260814T120405Z-00ca7096/trainer/checkpoint-736/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/trainer_state.json
ADDED
|
@@ -0,0 +1,572 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 985,
|
| 3 |
+
"best_metric": 0.8603902977100812,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985",
|
| 5 |
+
"epoch": 8.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1576,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.12706480304955528,
|
| 14 |
+
"grad_norm": 8.6220121383667,
|
| 15 |
+
"learning_rate": 1.218274111675127e-06,
|
| 16 |
+
"loss": 4.4325,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.25412960609911056,
|
| 21 |
+
"grad_norm": 7.805527687072754,
|
| 22 |
+
"learning_rate": 2.487309644670051e-06,
|
| 23 |
+
"loss": 4.4162,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.3811944091486658,
|
| 28 |
+
"grad_norm": 10.720266342163086,
|
| 29 |
+
"learning_rate": 3.756345177664975e-06,
|
| 30 |
+
"loss": 4.3872,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5082592121982211,
|
| 35 |
+
"grad_norm": 9.079874992370605,
|
| 36 |
+
"learning_rate": 5.025380710659899e-06,
|
| 37 |
+
"loss": 4.3789,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6353240152477764,
|
| 42 |
+
"grad_norm": 7.274191856384277,
|
| 43 |
+
"learning_rate": 6.2944162436548225e-06,
|
| 44 |
+
"loss": 4.3244,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.7623888182973316,
|
| 49 |
+
"grad_norm": 8.526632308959961,
|
| 50 |
+
"learning_rate": 7.563451776649747e-06,
|
| 51 |
+
"loss": 4.2584,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.8894536213468869,
|
| 56 |
+
"grad_norm": 11.150956153869629,
|
| 57 |
+
"learning_rate": 8.832487309644671e-06,
|
| 58 |
+
"loss": 4.1435,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5994020926756353,
|
| 64 |
+
"eval_loss": 0.9035991430282593,
|
| 65 |
+
"eval_macro_f1": 0.580504110657113,
|
| 66 |
+
"eval_runtime": 1.2901,
|
| 67 |
+
"eval_samples_per_second": 518.553,
|
| 68 |
+
"eval_steps_per_second": 32.555,
|
| 69 |
+
"eval_weighted_f1": 0.5820152123678336,
|
| 70 |
+
"step": 197
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.0152477763659467,
|
| 74 |
+
"grad_norm": 17.665699005126953,
|
| 75 |
+
"learning_rate": 1.0101522842639596e-05,
|
| 76 |
+
"loss": 3.8449,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.1423125794155018,
|
| 81 |
+
"grad_norm": 18.85488510131836,
|
| 82 |
+
"learning_rate": 1.1370558375634519e-05,
|
| 83 |
+
"loss": 3.3981,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.2693773824650572,
|
| 88 |
+
"grad_norm": 27.287458419799805,
|
| 89 |
+
"learning_rate": 1.2639593908629442e-05,
|
| 90 |
+
"loss": 3.1573,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.3964421855146125,
|
| 95 |
+
"grad_norm": 36.977012634277344,
|
| 96 |
+
"learning_rate": 1.3908629441624366e-05,
|
| 97 |
+
"loss": 2.8441,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.5235069885641677,
|
| 102 |
+
"grad_norm": 43.64808654785156,
|
| 103 |
+
"learning_rate": 1.5177664974619291e-05,
|
| 104 |
+
"loss": 2.53,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.650571791613723,
|
| 109 |
+
"grad_norm": 44.9443244934082,
|
| 110 |
+
"learning_rate": 1.6446700507614214e-05,
|
| 111 |
+
"loss": 2.161,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.7776365946632784,
|
| 116 |
+
"grad_norm": 46.40912628173828,
|
| 117 |
+
"learning_rate": 1.771573604060914e-05,
|
| 118 |
+
"loss": 2.1218,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.9047013977128335,
|
| 123 |
+
"grad_norm": 42.61958312988281,
|
| 124 |
+
"learning_rate": 1.8984771573604063e-05,
|
| 125 |
+
"loss": 2.1179,
|
| 126 |
+
"step": 375
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 2.0,
|
| 130 |
+
"eval_accuracy": 0.8415545590433483,
|
| 131 |
+
"eval_loss": 0.4308476150035858,
|
| 132 |
+
"eval_macro_f1": 0.8397611856097527,
|
| 133 |
+
"eval_runtime": 1.3649,
|
| 134 |
+
"eval_samples_per_second": 490.158,
|
| 135 |
+
"eval_steps_per_second": 30.772,
|
| 136 |
+
"eval_weighted_f1": 0.8400435982981281,
|
| 137 |
+
"step": 394
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.0304955527318933,
|
| 141 |
+
"grad_norm": 45.20762634277344,
|
| 142 |
+
"learning_rate": 1.997179921037789e-05,
|
| 143 |
+
"loss": 1.8912,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.1575603557814484,
|
| 148 |
+
"grad_norm": 36.97223663330078,
|
| 149 |
+
"learning_rate": 1.9830795262267346e-05,
|
| 150 |
+
"loss": 1.4592,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.2846251588310036,
|
| 155 |
+
"grad_norm": 28.935998916625977,
|
| 156 |
+
"learning_rate": 1.96897913141568e-05,
|
| 157 |
+
"loss": 1.6084,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.411689961880559,
|
| 162 |
+
"grad_norm": 40.04739761352539,
|
| 163 |
+
"learning_rate": 1.954878736604625e-05,
|
| 164 |
+
"loss": 1.6174,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.5387547649301143,
|
| 169 |
+
"grad_norm": 23.60295295715332,
|
| 170 |
+
"learning_rate": 1.9407783417935703e-05,
|
| 171 |
+
"loss": 1.6742,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.6658195679796695,
|
| 176 |
+
"grad_norm": 46.64494705200195,
|
| 177 |
+
"learning_rate": 1.9266779469825156e-05,
|
| 178 |
+
"loss": 1.5438,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.792884371029225,
|
| 183 |
+
"grad_norm": 88.65325164794922,
|
| 184 |
+
"learning_rate": 1.912577552171461e-05,
|
| 185 |
+
"loss": 1.5668,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 2.91994917407878,
|
| 190 |
+
"grad_norm": 49.7031364440918,
|
| 191 |
+
"learning_rate": 1.8984771573604063e-05,
|
| 192 |
+
"loss": 1.6172,
|
| 193 |
+
"step": 575
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 3.0,
|
| 197 |
+
"eval_accuracy": 0.85949177877429,
|
| 198 |
+
"eval_loss": 0.3869464695453644,
|
| 199 |
+
"eval_macro_f1": 0.8591950425457222,
|
| 200 |
+
"eval_runtime": 1.407,
|
| 201 |
+
"eval_samples_per_second": 475.468,
|
| 202 |
+
"eval_steps_per_second": 29.85,
|
| 203 |
+
"eval_weighted_f1": 0.8593555068592202,
|
| 204 |
+
"step": 591
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.04574332909784,
|
| 208 |
+
"grad_norm": 24.9626522064209,
|
| 209 |
+
"learning_rate": 1.8843767625493517e-05,
|
| 210 |
+
"loss": 1.2955,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.172808132147395,
|
| 215 |
+
"grad_norm": 27.510896682739258,
|
| 216 |
+
"learning_rate": 1.870276367738297e-05,
|
| 217 |
+
"loss": 0.9673,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.2998729351969507,
|
| 222 |
+
"grad_norm": 50.77602767944336,
|
| 223 |
+
"learning_rate": 1.856175972927242e-05,
|
| 224 |
+
"loss": 1.0583,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.426937738246506,
|
| 229 |
+
"grad_norm": 34.61463165283203,
|
| 230 |
+
"learning_rate": 1.8420755781161873e-05,
|
| 231 |
+
"loss": 1.126,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.554002541296061,
|
| 236 |
+
"grad_norm": 28.8006649017334,
|
| 237 |
+
"learning_rate": 1.8279751833051327e-05,
|
| 238 |
+
"loss": 0.939,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.681067344345616,
|
| 243 |
+
"grad_norm": 38.91231155395508,
|
| 244 |
+
"learning_rate": 1.813874788494078e-05,
|
| 245 |
+
"loss": 1.0619,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.8081321473951717,
|
| 250 |
+
"grad_norm": 39.92739486694336,
|
| 251 |
+
"learning_rate": 1.7997743936830234e-05,
|
| 252 |
+
"loss": 0.9536,
|
| 253 |
+
"step": 750
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 3.935196950444727,
|
| 257 |
+
"grad_norm": 86.99784851074219,
|
| 258 |
+
"learning_rate": 1.7856739988719684e-05,
|
| 259 |
+
"loss": 1.0682,
|
| 260 |
+
"step": 775
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 4.0,
|
| 264 |
+
"eval_accuracy": 0.8535127055306427,
|
| 265 |
+
"eval_loss": 0.4100440442562103,
|
| 266 |
+
"eval_macro_f1": 0.8524424265366021,
|
| 267 |
+
"eval_runtime": 1.4137,
|
| 268 |
+
"eval_samples_per_second": 473.226,
|
| 269 |
+
"eval_steps_per_second": 29.709,
|
| 270 |
+
"eval_weighted_f1": 0.8527209277197818,
|
| 271 |
+
"step": 788
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.060991105463787,
|
| 275 |
+
"grad_norm": 37.58418273925781,
|
| 276 |
+
"learning_rate": 1.771573604060914e-05,
|
| 277 |
+
"loss": 0.8928,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.188055908513342,
|
| 282 |
+
"grad_norm": 5.276197910308838,
|
| 283 |
+
"learning_rate": 1.7574732092498594e-05,
|
| 284 |
+
"loss": 0.5611,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.315120711562897,
|
| 289 |
+
"grad_norm": 47.121063232421875,
|
| 290 |
+
"learning_rate": 1.7433728144388044e-05,
|
| 291 |
+
"loss": 0.7237,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 4.4421855146124525,
|
| 296 |
+
"grad_norm": 54.98345947265625,
|
| 297 |
+
"learning_rate": 1.7292724196277497e-05,
|
| 298 |
+
"loss": 0.6316,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.569250317662007,
|
| 303 |
+
"grad_norm": 87.73394012451172,
|
| 304 |
+
"learning_rate": 1.715172024816695e-05,
|
| 305 |
+
"loss": 0.8148,
|
| 306 |
+
"step": 900
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 4.696315120711563,
|
| 310 |
+
"grad_norm": 36.25434875488281,
|
| 311 |
+
"learning_rate": 1.7010716300056404e-05,
|
| 312 |
+
"loss": 0.6121,
|
| 313 |
+
"step": 925
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 4.823379923761118,
|
| 317 |
+
"grad_norm": 95.66836547851562,
|
| 318 |
+
"learning_rate": 1.6869712351945854e-05,
|
| 319 |
+
"loss": 0.6763,
|
| 320 |
+
"step": 950
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.950444726810673,
|
| 324 |
+
"grad_norm": 74.18147277832031,
|
| 325 |
+
"learning_rate": 1.672870840383531e-05,
|
| 326 |
+
"loss": 0.7392,
|
| 327 |
+
"step": 975
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 5.0,
|
| 331 |
+
"eval_accuracy": 0.8609865470852018,
|
| 332 |
+
"eval_loss": 0.4904754161834717,
|
| 333 |
+
"eval_macro_f1": 0.8603902977100812,
|
| 334 |
+
"eval_runtime": 1.3781,
|
| 335 |
+
"eval_samples_per_second": 485.434,
|
| 336 |
+
"eval_steps_per_second": 30.476,
|
| 337 |
+
"eval_weighted_f1": 0.8605260931834822,
|
| 338 |
+
"step": 985
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.076238881829733,
|
| 342 |
+
"grad_norm": 24.52606773376465,
|
| 343 |
+
"learning_rate": 1.6587704455724764e-05,
|
| 344 |
+
"loss": 0.5629,
|
| 345 |
+
"step": 1000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.203303684879288,
|
| 349 |
+
"grad_norm": 33.66092300415039,
|
| 350 |
+
"learning_rate": 1.6446700507614214e-05,
|
| 351 |
+
"loss": 0.4185,
|
| 352 |
+
"step": 1025
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 5.330368487928844,
|
| 356 |
+
"grad_norm": 24.667583465576172,
|
| 357 |
+
"learning_rate": 1.6305696559503667e-05,
|
| 358 |
+
"loss": 0.4883,
|
| 359 |
+
"step": 1050
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.457433290978399,
|
| 363 |
+
"grad_norm": 16.39928436279297,
|
| 364 |
+
"learning_rate": 1.616469261139312e-05,
|
| 365 |
+
"loss": 0.5369,
|
| 366 |
+
"step": 1075
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 5.584498094027954,
|
| 370 |
+
"grad_norm": 69.7935562133789,
|
| 371 |
+
"learning_rate": 1.6023688663282574e-05,
|
| 372 |
+
"loss": 0.5152,
|
| 373 |
+
"step": 1100
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 5.71156289707751,
|
| 377 |
+
"grad_norm": 6.037744522094727,
|
| 378 |
+
"learning_rate": 1.5882684715172024e-05,
|
| 379 |
+
"loss": 0.4274,
|
| 380 |
+
"step": 1125
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 5.838627700127065,
|
| 384 |
+
"grad_norm": 43.599754333496094,
|
| 385 |
+
"learning_rate": 1.5741680767061477e-05,
|
| 386 |
+
"loss": 0.5488,
|
| 387 |
+
"step": 1150
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 5.96569250317662,
|
| 391 |
+
"grad_norm": 108.26321411132812,
|
| 392 |
+
"learning_rate": 1.5600676818950934e-05,
|
| 393 |
+
"loss": 0.8107,
|
| 394 |
+
"step": 1175
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 6.0,
|
| 398 |
+
"eval_accuracy": 0.8400597907324364,
|
| 399 |
+
"eval_loss": 0.6087110042572021,
|
| 400 |
+
"eval_macro_f1": 0.8393996127241974,
|
| 401 |
+
"eval_runtime": 1.4002,
|
| 402 |
+
"eval_samples_per_second": 477.799,
|
| 403 |
+
"eval_steps_per_second": 29.996,
|
| 404 |
+
"eval_weighted_f1": 0.8395840488878226,
|
| 405 |
+
"step": 1182
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 6.09148665819568,
|
| 409 |
+
"grad_norm": 164.89027404785156,
|
| 410 |
+
"learning_rate": 1.5459672870840384e-05,
|
| 411 |
+
"loss": 0.4739,
|
| 412 |
+
"step": 1200
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 6.2185514612452355,
|
| 416 |
+
"grad_norm": 1.001871943473816,
|
| 417 |
+
"learning_rate": 1.5318668922729838e-05,
|
| 418 |
+
"loss": 0.3497,
|
| 419 |
+
"step": 1225
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 6.34561626429479,
|
| 423 |
+
"grad_norm": 0.4618174433708191,
|
| 424 |
+
"learning_rate": 1.5177664974619291e-05,
|
| 425 |
+
"loss": 0.4186,
|
| 426 |
+
"step": 1250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 6.472681067344346,
|
| 430 |
+
"grad_norm": 59.264102935791016,
|
| 431 |
+
"learning_rate": 1.5036661026508743e-05,
|
| 432 |
+
"loss": 0.4155,
|
| 433 |
+
"step": 1275
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 6.599745870393901,
|
| 437 |
+
"grad_norm": 200.21334838867188,
|
| 438 |
+
"learning_rate": 1.4895657078398196e-05,
|
| 439 |
+
"loss": 0.289,
|
| 440 |
+
"step": 1300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 6.726810673443456,
|
| 444 |
+
"grad_norm": 8.828402519226074,
|
| 445 |
+
"learning_rate": 1.4754653130287648e-05,
|
| 446 |
+
"loss": 0.3232,
|
| 447 |
+
"step": 1325
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.853875476493012,
|
| 451 |
+
"grad_norm": 104.2049331665039,
|
| 452 |
+
"learning_rate": 1.4613649182177103e-05,
|
| 453 |
+
"loss": 0.2852,
|
| 454 |
+
"step": 1350
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 6.980940279542566,
|
| 458 |
+
"grad_norm": 10.193021774291992,
|
| 459 |
+
"learning_rate": 1.4472645234066556e-05,
|
| 460 |
+
"loss": 0.5331,
|
| 461 |
+
"step": 1375
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 7.0,
|
| 465 |
+
"eval_accuracy": 0.8550074738415545,
|
| 466 |
+
"eval_loss": 0.5923581719398499,
|
| 467 |
+
"eval_macro_f1": 0.8547331575423476,
|
| 468 |
+
"eval_runtime": 1.3771,
|
| 469 |
+
"eval_samples_per_second": 485.804,
|
| 470 |
+
"eval_steps_per_second": 30.499,
|
| 471 |
+
"eval_weighted_f1": 0.8549270235548518,
|
| 472 |
+
"step": 1379
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 7.106734434561626,
|
| 476 |
+
"grad_norm": 16.848236083984375,
|
| 477 |
+
"learning_rate": 1.4331641285956008e-05,
|
| 478 |
+
"loss": 0.2243,
|
| 479 |
+
"step": 1400
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 7.233799237611182,
|
| 483 |
+
"grad_norm": 173.00979614257812,
|
| 484 |
+
"learning_rate": 1.4190637337845461e-05,
|
| 485 |
+
"loss": 0.2767,
|
| 486 |
+
"step": 1425
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 7.360864040660737,
|
| 490 |
+
"grad_norm": 117.08274841308594,
|
| 491 |
+
"learning_rate": 1.4049633389734913e-05,
|
| 492 |
+
"loss": 0.3777,
|
| 493 |
+
"step": 1450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 7.487928843710292,
|
| 497 |
+
"grad_norm": 390.5910949707031,
|
| 498 |
+
"learning_rate": 1.3908629441624366e-05,
|
| 499 |
+
"loss": 0.323,
|
| 500 |
+
"step": 1475
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 7.614993646759848,
|
| 504 |
+
"grad_norm": 87.0909423828125,
|
| 505 |
+
"learning_rate": 1.3767625493513818e-05,
|
| 506 |
+
"loss": 0.5457,
|
| 507 |
+
"step": 1500
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 7.742058449809403,
|
| 511 |
+
"grad_norm": 26.146286010742188,
|
| 512 |
+
"learning_rate": 1.3626621545403271e-05,
|
| 513 |
+
"loss": 0.2434,
|
| 514 |
+
"step": 1525
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 7.869123252858958,
|
| 518 |
+
"grad_norm": 0.44684383273124695,
|
| 519 |
+
"learning_rate": 1.3485617597292726e-05,
|
| 520 |
+
"loss": 0.3664,
|
| 521 |
+
"step": 1550
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 7.9961880559085134,
|
| 525 |
+
"grad_norm": 11.087890625,
|
| 526 |
+
"learning_rate": 1.3344613649182178e-05,
|
| 527 |
+
"loss": 0.2758,
|
| 528 |
+
"step": 1575
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 8.0,
|
| 532 |
+
"eval_accuracy": 0.8445440956651719,
|
| 533 |
+
"eval_loss": 0.7107126712799072,
|
| 534 |
+
"eval_macro_f1": 0.8438879563954013,
|
| 535 |
+
"eval_runtime": 1.4341,
|
| 536 |
+
"eval_samples_per_second": 466.503,
|
| 537 |
+
"eval_steps_per_second": 29.287,
|
| 538 |
+
"eval_weighted_f1": 0.8439698129479842,
|
| 539 |
+
"step": 1576
|
| 540 |
+
}
|
| 541 |
+
],
|
| 542 |
+
"logging_steps": 25,
|
| 543 |
+
"max_steps": 3940,
|
| 544 |
+
"num_input_tokens_seen": 0,
|
| 545 |
+
"num_train_epochs": 20,
|
| 546 |
+
"save_steps": 500,
|
| 547 |
+
"stateful_callbacks": {
|
| 548 |
+
"EarlyStoppingCallback": {
|
| 549 |
+
"args": {
|
| 550 |
+
"early_stopping_patience": 3,
|
| 551 |
+
"early_stopping_threshold": 0.0
|
| 552 |
+
},
|
| 553 |
+
"attributes": {
|
| 554 |
+
"early_stopping_patience_counter": 3
|
| 555 |
+
}
|
| 556 |
+
},
|
| 557 |
+
"TrainerControl": {
|
| 558 |
+
"args": {
|
| 559 |
+
"should_epoch_stop": false,
|
| 560 |
+
"should_evaluate": false,
|
| 561 |
+
"should_log": false,
|
| 562 |
+
"should_save": true,
|
| 563 |
+
"should_training_stop": true
|
| 564 |
+
},
|
| 565 |
+
"attributes": {}
|
| 566 |
+
}
|
| 567 |
+
},
|
| 568 |
+
"total_flos": 0.0,
|
| 569 |
+
"train_batch_size": 8,
|
| 570 |
+
"trial_name": null,
|
| 571 |
+
"trial_params": null
|
| 572 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-1576/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/trainer_state.json
ADDED
|
@@ -0,0 +1,371 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 985,
|
| 3 |
+
"best_metric": 0.8603902977100812,
|
| 4 |
+
"best_model_checkpoint": "/root/VNEncoder/results/checkpoints/ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985",
|
| 5 |
+
"epoch": 5.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 985,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.12706480304955528,
|
| 14 |
+
"grad_norm": 8.6220121383667,
|
| 15 |
+
"learning_rate": 1.218274111675127e-06,
|
| 16 |
+
"loss": 4.4325,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.25412960609911056,
|
| 21 |
+
"grad_norm": 7.805527687072754,
|
| 22 |
+
"learning_rate": 2.487309644670051e-06,
|
| 23 |
+
"loss": 4.4162,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.3811944091486658,
|
| 28 |
+
"grad_norm": 10.720266342163086,
|
| 29 |
+
"learning_rate": 3.756345177664975e-06,
|
| 30 |
+
"loss": 4.3872,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.5082592121982211,
|
| 35 |
+
"grad_norm": 9.079874992370605,
|
| 36 |
+
"learning_rate": 5.025380710659899e-06,
|
| 37 |
+
"loss": 4.3789,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.6353240152477764,
|
| 42 |
+
"grad_norm": 7.274191856384277,
|
| 43 |
+
"learning_rate": 6.2944162436548225e-06,
|
| 44 |
+
"loss": 4.3244,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.7623888182973316,
|
| 49 |
+
"grad_norm": 8.526632308959961,
|
| 50 |
+
"learning_rate": 7.563451776649747e-06,
|
| 51 |
+
"loss": 4.2584,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.8894536213468869,
|
| 56 |
+
"grad_norm": 11.150956153869629,
|
| 57 |
+
"learning_rate": 8.832487309644671e-06,
|
| 58 |
+
"loss": 4.1435,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.0,
|
| 63 |
+
"eval_accuracy": 0.5994020926756353,
|
| 64 |
+
"eval_loss": 0.9035991430282593,
|
| 65 |
+
"eval_macro_f1": 0.580504110657113,
|
| 66 |
+
"eval_runtime": 1.2901,
|
| 67 |
+
"eval_samples_per_second": 518.553,
|
| 68 |
+
"eval_steps_per_second": 32.555,
|
| 69 |
+
"eval_weighted_f1": 0.5820152123678336,
|
| 70 |
+
"step": 197
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 1.0152477763659467,
|
| 74 |
+
"grad_norm": 17.665699005126953,
|
| 75 |
+
"learning_rate": 1.0101522842639596e-05,
|
| 76 |
+
"loss": 3.8449,
|
| 77 |
+
"step": 200
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.1423125794155018,
|
| 81 |
+
"grad_norm": 18.85488510131836,
|
| 82 |
+
"learning_rate": 1.1370558375634519e-05,
|
| 83 |
+
"loss": 3.3981,
|
| 84 |
+
"step": 225
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.2693773824650572,
|
| 88 |
+
"grad_norm": 27.287458419799805,
|
| 89 |
+
"learning_rate": 1.2639593908629442e-05,
|
| 90 |
+
"loss": 3.1573,
|
| 91 |
+
"step": 250
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.3964421855146125,
|
| 95 |
+
"grad_norm": 36.977012634277344,
|
| 96 |
+
"learning_rate": 1.3908629441624366e-05,
|
| 97 |
+
"loss": 2.8441,
|
| 98 |
+
"step": 275
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.5235069885641677,
|
| 102 |
+
"grad_norm": 43.64808654785156,
|
| 103 |
+
"learning_rate": 1.5177664974619291e-05,
|
| 104 |
+
"loss": 2.53,
|
| 105 |
+
"step": 300
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.650571791613723,
|
| 109 |
+
"grad_norm": 44.9443244934082,
|
| 110 |
+
"learning_rate": 1.6446700507614214e-05,
|
| 111 |
+
"loss": 2.161,
|
| 112 |
+
"step": 325
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.7776365946632784,
|
| 116 |
+
"grad_norm": 46.40912628173828,
|
| 117 |
+
"learning_rate": 1.771573604060914e-05,
|
| 118 |
+
"loss": 2.1218,
|
| 119 |
+
"step": 350
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.9047013977128335,
|
| 123 |
+
"grad_norm": 42.61958312988281,
|
| 124 |
+
"learning_rate": 1.8984771573604063e-05,
|
| 125 |
+
"loss": 2.1179,
|
| 126 |
+
"step": 375
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 2.0,
|
| 130 |
+
"eval_accuracy": 0.8415545590433483,
|
| 131 |
+
"eval_loss": 0.4308476150035858,
|
| 132 |
+
"eval_macro_f1": 0.8397611856097527,
|
| 133 |
+
"eval_runtime": 1.3649,
|
| 134 |
+
"eval_samples_per_second": 490.158,
|
| 135 |
+
"eval_steps_per_second": 30.772,
|
| 136 |
+
"eval_weighted_f1": 0.8400435982981281,
|
| 137 |
+
"step": 394
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.0304955527318933,
|
| 141 |
+
"grad_norm": 45.20762634277344,
|
| 142 |
+
"learning_rate": 1.997179921037789e-05,
|
| 143 |
+
"loss": 1.8912,
|
| 144 |
+
"step": 400
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.1575603557814484,
|
| 148 |
+
"grad_norm": 36.97223663330078,
|
| 149 |
+
"learning_rate": 1.9830795262267346e-05,
|
| 150 |
+
"loss": 1.4592,
|
| 151 |
+
"step": 425
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.2846251588310036,
|
| 155 |
+
"grad_norm": 28.935998916625977,
|
| 156 |
+
"learning_rate": 1.96897913141568e-05,
|
| 157 |
+
"loss": 1.6084,
|
| 158 |
+
"step": 450
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.411689961880559,
|
| 162 |
+
"grad_norm": 40.04739761352539,
|
| 163 |
+
"learning_rate": 1.954878736604625e-05,
|
| 164 |
+
"loss": 1.6174,
|
| 165 |
+
"step": 475
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.5387547649301143,
|
| 169 |
+
"grad_norm": 23.60295295715332,
|
| 170 |
+
"learning_rate": 1.9407783417935703e-05,
|
| 171 |
+
"loss": 1.6742,
|
| 172 |
+
"step": 500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.6658195679796695,
|
| 176 |
+
"grad_norm": 46.64494705200195,
|
| 177 |
+
"learning_rate": 1.9266779469825156e-05,
|
| 178 |
+
"loss": 1.5438,
|
| 179 |
+
"step": 525
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.792884371029225,
|
| 183 |
+
"grad_norm": 88.65325164794922,
|
| 184 |
+
"learning_rate": 1.912577552171461e-05,
|
| 185 |
+
"loss": 1.5668,
|
| 186 |
+
"step": 550
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 2.91994917407878,
|
| 190 |
+
"grad_norm": 49.7031364440918,
|
| 191 |
+
"learning_rate": 1.8984771573604063e-05,
|
| 192 |
+
"loss": 1.6172,
|
| 193 |
+
"step": 575
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 3.0,
|
| 197 |
+
"eval_accuracy": 0.85949177877429,
|
| 198 |
+
"eval_loss": 0.3869464695453644,
|
| 199 |
+
"eval_macro_f1": 0.8591950425457222,
|
| 200 |
+
"eval_runtime": 1.407,
|
| 201 |
+
"eval_samples_per_second": 475.468,
|
| 202 |
+
"eval_steps_per_second": 29.85,
|
| 203 |
+
"eval_weighted_f1": 0.8593555068592202,
|
| 204 |
+
"step": 591
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.04574332909784,
|
| 208 |
+
"grad_norm": 24.9626522064209,
|
| 209 |
+
"learning_rate": 1.8843767625493517e-05,
|
| 210 |
+
"loss": 1.2955,
|
| 211 |
+
"step": 600
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.172808132147395,
|
| 215 |
+
"grad_norm": 27.510896682739258,
|
| 216 |
+
"learning_rate": 1.870276367738297e-05,
|
| 217 |
+
"loss": 0.9673,
|
| 218 |
+
"step": 625
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.2998729351969507,
|
| 222 |
+
"grad_norm": 50.77602767944336,
|
| 223 |
+
"learning_rate": 1.856175972927242e-05,
|
| 224 |
+
"loss": 1.0583,
|
| 225 |
+
"step": 650
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.426937738246506,
|
| 229 |
+
"grad_norm": 34.61463165283203,
|
| 230 |
+
"learning_rate": 1.8420755781161873e-05,
|
| 231 |
+
"loss": 1.126,
|
| 232 |
+
"step": 675
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.554002541296061,
|
| 236 |
+
"grad_norm": 28.8006649017334,
|
| 237 |
+
"learning_rate": 1.8279751833051327e-05,
|
| 238 |
+
"loss": 0.939,
|
| 239 |
+
"step": 700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.681067344345616,
|
| 243 |
+
"grad_norm": 38.91231155395508,
|
| 244 |
+
"learning_rate": 1.813874788494078e-05,
|
| 245 |
+
"loss": 1.0619,
|
| 246 |
+
"step": 725
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.8081321473951717,
|
| 250 |
+
"grad_norm": 39.92739486694336,
|
| 251 |
+
"learning_rate": 1.7997743936830234e-05,
|
| 252 |
+
"loss": 0.9536,
|
| 253 |
+
"step": 750
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 3.935196950444727,
|
| 257 |
+
"grad_norm": 86.99784851074219,
|
| 258 |
+
"learning_rate": 1.7856739988719684e-05,
|
| 259 |
+
"loss": 1.0682,
|
| 260 |
+
"step": 775
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 4.0,
|
| 264 |
+
"eval_accuracy": 0.8535127055306427,
|
| 265 |
+
"eval_loss": 0.4100440442562103,
|
| 266 |
+
"eval_macro_f1": 0.8524424265366021,
|
| 267 |
+
"eval_runtime": 1.4137,
|
| 268 |
+
"eval_samples_per_second": 473.226,
|
| 269 |
+
"eval_steps_per_second": 29.709,
|
| 270 |
+
"eval_weighted_f1": 0.8527209277197818,
|
| 271 |
+
"step": 788
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.060991105463787,
|
| 275 |
+
"grad_norm": 37.58418273925781,
|
| 276 |
+
"learning_rate": 1.771573604060914e-05,
|
| 277 |
+
"loss": 0.8928,
|
| 278 |
+
"step": 800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.188055908513342,
|
| 282 |
+
"grad_norm": 5.276197910308838,
|
| 283 |
+
"learning_rate": 1.7574732092498594e-05,
|
| 284 |
+
"loss": 0.5611,
|
| 285 |
+
"step": 825
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 4.315120711562897,
|
| 289 |
+
"grad_norm": 47.121063232421875,
|
| 290 |
+
"learning_rate": 1.7433728144388044e-05,
|
| 291 |
+
"loss": 0.7237,
|
| 292 |
+
"step": 850
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 4.4421855146124525,
|
| 296 |
+
"grad_norm": 54.98345947265625,
|
| 297 |
+
"learning_rate": 1.7292724196277497e-05,
|
| 298 |
+
"loss": 0.6316,
|
| 299 |
+
"step": 875
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.569250317662007,
|
| 303 |
+
"grad_norm": 87.73394012451172,
|
| 304 |
+
"learning_rate": 1.715172024816695e-05,
|
| 305 |
+
"loss": 0.8148,
|
| 306 |
+
"step": 900
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 4.696315120711563,
|
| 310 |
+
"grad_norm": 36.25434875488281,
|
| 311 |
+
"learning_rate": 1.7010716300056404e-05,
|
| 312 |
+
"loss": 0.6121,
|
| 313 |
+
"step": 925
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 4.823379923761118,
|
| 317 |
+
"grad_norm": 95.66836547851562,
|
| 318 |
+
"learning_rate": 1.6869712351945854e-05,
|
| 319 |
+
"loss": 0.6763,
|
| 320 |
+
"step": 950
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.950444726810673,
|
| 324 |
+
"grad_norm": 74.18147277832031,
|
| 325 |
+
"learning_rate": 1.672870840383531e-05,
|
| 326 |
+
"loss": 0.7392,
|
| 327 |
+
"step": 975
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 5.0,
|
| 331 |
+
"eval_accuracy": 0.8609865470852018,
|
| 332 |
+
"eval_loss": 0.4904754161834717,
|
| 333 |
+
"eval_macro_f1": 0.8603902977100812,
|
| 334 |
+
"eval_runtime": 1.3781,
|
| 335 |
+
"eval_samples_per_second": 485.434,
|
| 336 |
+
"eval_steps_per_second": 30.476,
|
| 337 |
+
"eval_weighted_f1": 0.8605260931834822,
|
| 338 |
+
"step": 985
|
| 339 |
+
}
|
| 340 |
+
],
|
| 341 |
+
"logging_steps": 25,
|
| 342 |
+
"max_steps": 3940,
|
| 343 |
+
"num_input_tokens_seen": 0,
|
| 344 |
+
"num_train_epochs": 20,
|
| 345 |
+
"save_steps": 500,
|
| 346 |
+
"stateful_callbacks": {
|
| 347 |
+
"EarlyStoppingCallback": {
|
| 348 |
+
"args": {
|
| 349 |
+
"early_stopping_patience": 3,
|
| 350 |
+
"early_stopping_threshold": 0.0
|
| 351 |
+
},
|
| 352 |
+
"attributes": {
|
| 353 |
+
"early_stopping_patience_counter": 0
|
| 354 |
+
}
|
| 355 |
+
},
|
| 356 |
+
"TrainerControl": {
|
| 357 |
+
"args": {
|
| 358 |
+
"should_epoch_stop": false,
|
| 359 |
+
"should_evaluate": false,
|
| 360 |
+
"should_log": false,
|
| 361 |
+
"should_save": true,
|
| 362 |
+
"should_training_stop": false
|
| 363 |
+
},
|
| 364 |
+
"attributes": {}
|
| 365 |
+
}
|
| 366 |
+
},
|
| 367 |
+
"total_flos": 0.0,
|
| 368 |
+
"train_batch_size": 8,
|
| 369 |
+
"trial_name": null,
|
| 370 |
+
"trial_params": null
|
| 371 |
+
}
|
ood-vifactcheck-phobert-topic-domain-07-s42-20260814T121241Z-00ca7096/trainer/checkpoint-985/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<mask>": 64000
|
| 3 |
+
}
|
ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/bpe.codes
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": "<mask>",
|
| 6 |
+
"pad_token": "<pad>",
|
| 7 |
+
"sep_token": "</s>",
|
| 8 |
+
"unk_token": "<unk>"
|
| 9 |
+
}
|
ood-vifactcheck-phobert-topic-domain-08-s42-20260814T122506Z-00ca7096/trainer/checkpoint-1700/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<pad>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "<unk>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"64000": {
|
| 36 |
+
"content": "<mask>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "<s>",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "<s>",
|
| 47 |
+
"eos_token": "</s>",
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "<mask>",
|
| 50 |
+
"model_max_length": 256,
|
| 51 |
+
"pad_token": "<pad>",
|
| 52 |
+
"sep_token": "</s>",
|
| 53 |
+
"tokenizer_class": "PhobertTokenizer",
|
| 54 |
+
"unk_token": "<unk>"
|
| 55 |
+
}
|