Commit
•
c35b1f3
1
Parent(s):
3657daf
uploaded omnitab model
Browse files- config.json +47 -0
- merges.txt +0 -0
- special_tokens_map.json +51 -0
- tokenizer_config.json +67 -0
- trainer_state.json +2701 -0
- training_args.bin +3 -0
- vocab.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "neulab/omnitab-large",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"BartForConditionalGeneration"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 0,
|
| 10 |
+
"classifier_dropout": 0.0,
|
| 11 |
+
"d_model": 1024,
|
| 12 |
+
"decoder_attention_heads": 16,
|
| 13 |
+
"decoder_ffn_dim": 4096,
|
| 14 |
+
"decoder_layerdrop": 0.0,
|
| 15 |
+
"decoder_layers": 12,
|
| 16 |
+
"decoder_start_token_id": 2,
|
| 17 |
+
"dropout": 0.1,
|
| 18 |
+
"encoder_attention_heads": 16,
|
| 19 |
+
"encoder_ffn_dim": 4096,
|
| 20 |
+
"encoder_layerdrop": 0.0,
|
| 21 |
+
"encoder_layers": 12,
|
| 22 |
+
"eos_token_id": 2,
|
| 23 |
+
"forced_bos_token_id": 0,
|
| 24 |
+
"forced_eos_token_id": 2,
|
| 25 |
+
"id2label": {
|
| 26 |
+
"0": "LABEL_0",
|
| 27 |
+
"1": "LABEL_1",
|
| 28 |
+
"2": "LABEL_2"
|
| 29 |
+
},
|
| 30 |
+
"init_std": 0.02,
|
| 31 |
+
"is_encoder_decoder": true,
|
| 32 |
+
"label2id": {
|
| 33 |
+
"LABEL_0": 0,
|
| 34 |
+
"LABEL_1": 1,
|
| 35 |
+
"LABEL_2": 2
|
| 36 |
+
},
|
| 37 |
+
"max_length": 1024,
|
| 38 |
+
"max_position_embeddings": 1024,
|
| 39 |
+
"model_type": "bart",
|
| 40 |
+
"num_hidden_layers": 12,
|
| 41 |
+
"pad_token_id": 1,
|
| 42 |
+
"scale_embedding": false,
|
| 43 |
+
"torch_dtype": "float32",
|
| 44 |
+
"transformers_version": "4.24.0",
|
| 45 |
+
"use_cache": true,
|
| 46 |
+
"vocab_size": 50265
|
| 47 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": true,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"cls_token": {
|
| 10 |
+
"content": "<s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": true,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"eos_token": {
|
| 17 |
+
"content": "</s>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": true,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"mask_token": {
|
| 24 |
+
"content": "<mask>",
|
| 25 |
+
"lstrip": true,
|
| 26 |
+
"normalized": true,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"pad_token": {
|
| 31 |
+
"content": "<pad>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": true,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
},
|
| 37 |
+
"sep_token": {
|
| 38 |
+
"content": "</s>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": true,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false
|
| 43 |
+
},
|
| 44 |
+
"unk_token": {
|
| 45 |
+
"content": "<unk>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": true,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false
|
| 50 |
+
}
|
| 51 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"bos_token": {
|
| 4 |
+
"__type": "AddedToken",
|
| 5 |
+
"content": "<s>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": true,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false
|
| 10 |
+
},
|
| 11 |
+
"cls_token": {
|
| 12 |
+
"__type": "AddedToken",
|
| 13 |
+
"content": "<s>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": true,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false
|
| 18 |
+
},
|
| 19 |
+
"do_lower_case": false,
|
| 20 |
+
"eos_token": {
|
| 21 |
+
"__type": "AddedToken",
|
| 22 |
+
"content": "</s>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": true,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false
|
| 27 |
+
},
|
| 28 |
+
"errors": "replace",
|
| 29 |
+
"mask_token": {
|
| 30 |
+
"__type": "AddedToken",
|
| 31 |
+
"content": "<mask>",
|
| 32 |
+
"lstrip": true,
|
| 33 |
+
"normalized": true,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
},
|
| 37 |
+
"max_cell_length": 15,
|
| 38 |
+
"model_max_length": 1024,
|
| 39 |
+
"name_or_path": "neulab/omnitab-large",
|
| 40 |
+
"pad_token": {
|
| 41 |
+
"__type": "AddedToken",
|
| 42 |
+
"content": "<pad>",
|
| 43 |
+
"lstrip": false,
|
| 44 |
+
"normalized": true,
|
| 45 |
+
"rstrip": false,
|
| 46 |
+
"single_word": false
|
| 47 |
+
},
|
| 48 |
+
"sep_token": {
|
| 49 |
+
"__type": "AddedToken",
|
| 50 |
+
"content": "</s>",
|
| 51 |
+
"lstrip": false,
|
| 52 |
+
"normalized": true,
|
| 53 |
+
"rstrip": false,
|
| 54 |
+
"single_word": false
|
| 55 |
+
},
|
| 56 |
+
"special_tokens_map_file": null,
|
| 57 |
+
"tokenizer_class": "TapexTokenizer",
|
| 58 |
+
"unk_token": {
|
| 59 |
+
"__type": "AddedToken",
|
| 60 |
+
"content": "<unk>",
|
| 61 |
+
"lstrip": false,
|
| 62 |
+
"normalized": true,
|
| 63 |
+
"rstrip": false,
|
| 64 |
+
"single_word": false
|
| 65 |
+
},
|
| 66 |
+
"use_fast": true
|
| 67 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2701 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 49.994350282485875,
|
| 5 |
+
"global_step": 4400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.11,
|
| 12 |
+
"learning_rate": 3.0000000000000004e-07,
|
| 13 |
+
"loss": 3.5742,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.23,
|
| 18 |
+
"learning_rate": 6.000000000000001e-07,
|
| 19 |
+
"loss": 3.4353,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.34,
|
| 24 |
+
"learning_rate": 9e-07,
|
| 25 |
+
"loss": 3.2438,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.45,
|
| 30 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 31 |
+
"loss": 3.0976,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.56,
|
| 36 |
+
"learning_rate": 1.5e-06,
|
| 37 |
+
"loss": 2.8575,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.68,
|
| 42 |
+
"learning_rate": 1.8e-06,
|
| 43 |
+
"loss": 2.6829,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.79,
|
| 48 |
+
"learning_rate": 2.1000000000000002e-06,
|
| 49 |
+
"loss": 2.4912,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.9,
|
| 54 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 55 |
+
"loss": 2.4056,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.02,
|
| 60 |
+
"learning_rate": 2.7e-06,
|
| 61 |
+
"loss": 2.4176,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.14,
|
| 66 |
+
"learning_rate": 3e-06,
|
| 67 |
+
"loss": 2.2921,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.25,
|
| 72 |
+
"learning_rate": 3.3e-06,
|
| 73 |
+
"loss": 2.2252,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 1.36,
|
| 78 |
+
"learning_rate": 3.6e-06,
|
| 79 |
+
"loss": 2.2169,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.47,
|
| 84 |
+
"learning_rate": 3.9e-06,
|
| 85 |
+
"loss": 2.156,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.59,
|
| 90 |
+
"learning_rate": 4.2000000000000004e-06,
|
| 91 |
+
"loss": 2.194,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.7,
|
| 96 |
+
"learning_rate": 4.5e-06,
|
| 97 |
+
"loss": 2.1136,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.81,
|
| 102 |
+
"learning_rate": 4.800000000000001e-06,
|
| 103 |
+
"loss": 2.1497,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.93,
|
| 108 |
+
"learning_rate": 5.1e-06,
|
| 109 |
+
"loss": 2.1208,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 2.05,
|
| 114 |
+
"learning_rate": 5.4e-06,
|
| 115 |
+
"loss": 2.2234,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 2.16,
|
| 120 |
+
"learning_rate": 5.7000000000000005e-06,
|
| 121 |
+
"loss": 2.0122,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 2.27,
|
| 126 |
+
"learning_rate": 6e-06,
|
| 127 |
+
"loss": 2.0509,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 2.38,
|
| 132 |
+
"learning_rate": 6.3e-06,
|
| 133 |
+
"loss": 2.0077,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 2.5,
|
| 138 |
+
"learning_rate": 6.6e-06,
|
| 139 |
+
"loss": 2.0297,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.61,
|
| 144 |
+
"learning_rate": 6.900000000000001e-06,
|
| 145 |
+
"loss": 2.0226,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 2.72,
|
| 150 |
+
"learning_rate": 7.2e-06,
|
| 151 |
+
"loss": 2.0274,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 2.84,
|
| 156 |
+
"learning_rate": 7.5e-06,
|
| 157 |
+
"loss": 2.0505,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.95,
|
| 162 |
+
"learning_rate": 7.8e-06,
|
| 163 |
+
"loss": 2.0163,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 3.07,
|
| 168 |
+
"learning_rate": 8.1e-06,
|
| 169 |
+
"loss": 2.0764,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.18,
|
| 174 |
+
"learning_rate": 8.400000000000001e-06,
|
| 175 |
+
"loss": 1.9411,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.29,
|
| 180 |
+
"learning_rate": 8.7e-06,
|
| 181 |
+
"loss": 1.9308,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.41,
|
| 186 |
+
"learning_rate": 9e-06,
|
| 187 |
+
"loss": 1.9507,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.52,
|
| 192 |
+
"learning_rate": 9.3e-06,
|
| 193 |
+
"loss": 1.9036,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.63,
|
| 198 |
+
"learning_rate": 9.600000000000001e-06,
|
| 199 |
+
"loss": 1.9728,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.75,
|
| 204 |
+
"learning_rate": 9.9e-06,
|
| 205 |
+
"loss": 1.9105,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.86,
|
| 210 |
+
"learning_rate": 1.02e-05,
|
| 211 |
+
"loss": 1.957,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.97,
|
| 216 |
+
"learning_rate": 1.05e-05,
|
| 217 |
+
"loss": 1.9615,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 4.09,
|
| 222 |
+
"learning_rate": 1.08e-05,
|
| 223 |
+
"loss": 1.9444,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 4.2,
|
| 228 |
+
"learning_rate": 1.11e-05,
|
| 229 |
+
"loss": 1.9014,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 4.32,
|
| 234 |
+
"learning_rate": 1.1400000000000001e-05,
|
| 235 |
+
"loss": 1.8752,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 4.43,
|
| 240 |
+
"learning_rate": 1.1700000000000001e-05,
|
| 241 |
+
"loss": 1.8787,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 4.54,
|
| 246 |
+
"learning_rate": 1.2e-05,
|
| 247 |
+
"loss": 1.8929,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 4.66,
|
| 252 |
+
"learning_rate": 1.2299999999999999e-05,
|
| 253 |
+
"loss": 1.9223,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 4.77,
|
| 258 |
+
"learning_rate": 1.26e-05,
|
| 259 |
+
"loss": 1.8559,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 4.88,
|
| 264 |
+
"learning_rate": 1.29e-05,
|
| 265 |
+
"loss": 1.8533,
|
| 266 |
+
"step": 430
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 4.99,
|
| 270 |
+
"learning_rate": 1.32e-05,
|
| 271 |
+
"loss": 1.8577,
|
| 272 |
+
"step": 440
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 5.11,
|
| 276 |
+
"learning_rate": 1.3500000000000001e-05,
|
| 277 |
+
"loss": 1.9091,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 5.23,
|
| 282 |
+
"learning_rate": 1.3800000000000002e-05,
|
| 283 |
+
"loss": 1.8096,
|
| 284 |
+
"step": 460
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 5.34,
|
| 288 |
+
"learning_rate": 1.4099999999999999e-05,
|
| 289 |
+
"loss": 1.8333,
|
| 290 |
+
"step": 470
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 5.45,
|
| 294 |
+
"learning_rate": 1.44e-05,
|
| 295 |
+
"loss": 1.8166,
|
| 296 |
+
"step": 480
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 5.56,
|
| 300 |
+
"learning_rate": 1.47e-05,
|
| 301 |
+
"loss": 1.8077,
|
| 302 |
+
"step": 490
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 5.68,
|
| 306 |
+
"learning_rate": 1.5e-05,
|
| 307 |
+
"loss": 1.8163,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 5.79,
|
| 312 |
+
"learning_rate": 1.53e-05,
|
| 313 |
+
"loss": 1.8269,
|
| 314 |
+
"step": 510
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 5.9,
|
| 318 |
+
"learning_rate": 1.56e-05,
|
| 319 |
+
"loss": 1.8201,
|
| 320 |
+
"step": 520
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 6.02,
|
| 324 |
+
"learning_rate": 1.59e-05,
|
| 325 |
+
"loss": 1.936,
|
| 326 |
+
"step": 530
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 6.14,
|
| 330 |
+
"learning_rate": 1.62e-05,
|
| 331 |
+
"loss": 1.7812,
|
| 332 |
+
"step": 540
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.25,
|
| 336 |
+
"learning_rate": 1.65e-05,
|
| 337 |
+
"loss": 1.7872,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.36,
|
| 342 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 343 |
+
"loss": 1.782,
|
| 344 |
+
"step": 560
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.47,
|
| 348 |
+
"learning_rate": 1.71e-05,
|
| 349 |
+
"loss": 1.783,
|
| 350 |
+
"step": 570
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.59,
|
| 354 |
+
"learning_rate": 1.74e-05,
|
| 355 |
+
"loss": 1.7638,
|
| 356 |
+
"step": 580
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.7,
|
| 360 |
+
"learning_rate": 1.77e-05,
|
| 361 |
+
"loss": 1.7805,
|
| 362 |
+
"step": 590
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.81,
|
| 366 |
+
"learning_rate": 1.8e-05,
|
| 367 |
+
"loss": 1.7768,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.93,
|
| 372 |
+
"learning_rate": 1.83e-05,
|
| 373 |
+
"loss": 1.7864,
|
| 374 |
+
"step": 610
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 7.05,
|
| 378 |
+
"learning_rate": 1.86e-05,
|
| 379 |
+
"loss": 1.8437,
|
| 380 |
+
"step": 620
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.16,
|
| 384 |
+
"learning_rate": 1.8900000000000002e-05,
|
| 385 |
+
"loss": 1.718,
|
| 386 |
+
"step": 630
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 7.27,
|
| 390 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 391 |
+
"loss": 1.7334,
|
| 392 |
+
"step": 640
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 7.38,
|
| 396 |
+
"learning_rate": 1.95e-05,
|
| 397 |
+
"loss": 1.7445,
|
| 398 |
+
"step": 650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 7.5,
|
| 402 |
+
"learning_rate": 1.98e-05,
|
| 403 |
+
"loss": 1.7553,
|
| 404 |
+
"step": 660
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 7.61,
|
| 408 |
+
"learning_rate": 2.01e-05,
|
| 409 |
+
"loss": 1.7567,
|
| 410 |
+
"step": 670
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 7.72,
|
| 414 |
+
"learning_rate": 2.04e-05,
|
| 415 |
+
"loss": 1.7498,
|
| 416 |
+
"step": 680
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 7.84,
|
| 420 |
+
"learning_rate": 2.07e-05,
|
| 421 |
+
"loss": 1.7273,
|
| 422 |
+
"step": 690
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 7.95,
|
| 426 |
+
"learning_rate": 2.1e-05,
|
| 427 |
+
"loss": 1.7588,
|
| 428 |
+
"step": 700
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 8.07,
|
| 432 |
+
"learning_rate": 2.13e-05,
|
| 433 |
+
"loss": 1.8182,
|
| 434 |
+
"step": 710
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 8.18,
|
| 438 |
+
"learning_rate": 2.16e-05,
|
| 439 |
+
"loss": 1.7256,
|
| 440 |
+
"step": 720
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 8.29,
|
| 444 |
+
"learning_rate": 2.19e-05,
|
| 445 |
+
"loss": 1.7198,
|
| 446 |
+
"step": 730
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 8.41,
|
| 450 |
+
"learning_rate": 2.22e-05,
|
| 451 |
+
"loss": 1.6916,
|
| 452 |
+
"step": 740
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 8.52,
|
| 456 |
+
"learning_rate": 2.25e-05,
|
| 457 |
+
"loss": 1.7189,
|
| 458 |
+
"step": 750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 8.63,
|
| 462 |
+
"learning_rate": 2.2800000000000002e-05,
|
| 463 |
+
"loss": 1.7172,
|
| 464 |
+
"step": 760
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 8.75,
|
| 468 |
+
"learning_rate": 2.3100000000000002e-05,
|
| 469 |
+
"loss": 1.7219,
|
| 470 |
+
"step": 770
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 8.86,
|
| 474 |
+
"learning_rate": 2.3400000000000003e-05,
|
| 475 |
+
"loss": 1.7425,
|
| 476 |
+
"step": 780
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 8.97,
|
| 480 |
+
"learning_rate": 2.37e-05,
|
| 481 |
+
"loss": 1.7384,
|
| 482 |
+
"step": 790
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 9.09,
|
| 486 |
+
"learning_rate": 2.4e-05,
|
| 487 |
+
"loss": 1.7836,
|
| 488 |
+
"step": 800
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 9.2,
|
| 492 |
+
"learning_rate": 2.43e-05,
|
| 493 |
+
"loss": 1.6577,
|
| 494 |
+
"step": 810
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 9.32,
|
| 498 |
+
"learning_rate": 2.4599999999999998e-05,
|
| 499 |
+
"loss": 1.6644,
|
| 500 |
+
"step": 820
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 9.43,
|
| 504 |
+
"learning_rate": 2.49e-05,
|
| 505 |
+
"loss": 1.6861,
|
| 506 |
+
"step": 830
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 9.54,
|
| 510 |
+
"learning_rate": 2.52e-05,
|
| 511 |
+
"loss": 1.6939,
|
| 512 |
+
"step": 840
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 9.66,
|
| 516 |
+
"learning_rate": 2.55e-05,
|
| 517 |
+
"loss": 1.7055,
|
| 518 |
+
"step": 850
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 9.77,
|
| 522 |
+
"learning_rate": 2.58e-05,
|
| 523 |
+
"loss": 1.7117,
|
| 524 |
+
"step": 860
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 9.88,
|
| 528 |
+
"learning_rate": 2.61e-05,
|
| 529 |
+
"loss": 1.6968,
|
| 530 |
+
"step": 870
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 9.99,
|
| 534 |
+
"learning_rate": 2.64e-05,
|
| 535 |
+
"loss": 1.7166,
|
| 536 |
+
"step": 880
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 10.11,
|
| 540 |
+
"learning_rate": 2.6700000000000002e-05,
|
| 541 |
+
"loss": 1.7732,
|
| 542 |
+
"step": 890
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 10.23,
|
| 546 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 547 |
+
"loss": 1.6664,
|
| 548 |
+
"step": 900
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 10.34,
|
| 552 |
+
"learning_rate": 2.7300000000000003e-05,
|
| 553 |
+
"loss": 1.6757,
|
| 554 |
+
"step": 910
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 10.45,
|
| 558 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 559 |
+
"loss": 1.6687,
|
| 560 |
+
"step": 920
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 10.56,
|
| 564 |
+
"learning_rate": 2.79e-05,
|
| 565 |
+
"loss": 1.6693,
|
| 566 |
+
"step": 930
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 10.68,
|
| 570 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 571 |
+
"loss": 1.6738,
|
| 572 |
+
"step": 940
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 10.79,
|
| 576 |
+
"learning_rate": 2.8499999999999998e-05,
|
| 577 |
+
"loss": 1.6807,
|
| 578 |
+
"step": 950
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 10.9,
|
| 582 |
+
"learning_rate": 2.88e-05,
|
| 583 |
+
"loss": 1.6875,
|
| 584 |
+
"step": 960
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 11.02,
|
| 588 |
+
"learning_rate": 2.91e-05,
|
| 589 |
+
"loss": 1.7664,
|
| 590 |
+
"step": 970
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 11.14,
|
| 594 |
+
"learning_rate": 2.94e-05,
|
| 595 |
+
"loss": 1.6674,
|
| 596 |
+
"step": 980
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 11.25,
|
| 600 |
+
"learning_rate": 2.97e-05,
|
| 601 |
+
"loss": 1.6369,
|
| 602 |
+
"step": 990
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 11.36,
|
| 606 |
+
"learning_rate": 3e-05,
|
| 607 |
+
"loss": 1.6675,
|
| 608 |
+
"step": 1000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 11.36,
|
| 612 |
+
"eval_denotation_accuracy": 0.18014835747085836,
|
| 613 |
+
"eval_loss": 2.312751293182373,
|
| 614 |
+
"eval_runtime": 1826.7928,
|
| 615 |
+
"eval_samples_per_second": 1.55,
|
| 616 |
+
"eval_steps_per_second": 0.097,
|
| 617 |
+
"step": 1000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 11.47,
|
| 621 |
+
"learning_rate": 2.9911764705882352e-05,
|
| 622 |
+
"loss": 1.6501,
|
| 623 |
+
"step": 1010
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 11.59,
|
| 627 |
+
"learning_rate": 2.9823529411764707e-05,
|
| 628 |
+
"loss": 1.6657,
|
| 629 |
+
"step": 1020
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 11.7,
|
| 633 |
+
"learning_rate": 2.973529411764706e-05,
|
| 634 |
+
"loss": 1.6592,
|
| 635 |
+
"step": 1030
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 11.81,
|
| 639 |
+
"learning_rate": 2.9647058823529414e-05,
|
| 640 |
+
"loss": 1.6575,
|
| 641 |
+
"step": 1040
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 11.93,
|
| 645 |
+
"learning_rate": 2.9558823529411766e-05,
|
| 646 |
+
"loss": 1.6638,
|
| 647 |
+
"step": 1050
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 12.05,
|
| 651 |
+
"learning_rate": 2.9470588235294117e-05,
|
| 652 |
+
"loss": 1.7401,
|
| 653 |
+
"step": 1060
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 12.16,
|
| 657 |
+
"learning_rate": 2.938235294117647e-05,
|
| 658 |
+
"loss": 1.6335,
|
| 659 |
+
"step": 1070
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 12.27,
|
| 663 |
+
"learning_rate": 2.9294117647058824e-05,
|
| 664 |
+
"loss": 1.6446,
|
| 665 |
+
"step": 1080
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 12.38,
|
| 669 |
+
"learning_rate": 2.9205882352941175e-05,
|
| 670 |
+
"loss": 1.6233,
|
| 671 |
+
"step": 1090
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 12.5,
|
| 675 |
+
"learning_rate": 2.911764705882353e-05,
|
| 676 |
+
"loss": 1.6528,
|
| 677 |
+
"step": 1100
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 12.61,
|
| 681 |
+
"learning_rate": 2.9029411764705882e-05,
|
| 682 |
+
"loss": 1.6488,
|
| 683 |
+
"step": 1110
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 12.72,
|
| 687 |
+
"learning_rate": 2.8941176470588237e-05,
|
| 688 |
+
"loss": 1.644,
|
| 689 |
+
"step": 1120
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 12.84,
|
| 693 |
+
"learning_rate": 2.885294117647059e-05,
|
| 694 |
+
"loss": 1.6455,
|
| 695 |
+
"step": 1130
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 12.95,
|
| 699 |
+
"learning_rate": 2.8764705882352943e-05,
|
| 700 |
+
"loss": 1.6705,
|
| 701 |
+
"step": 1140
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 13.07,
|
| 705 |
+
"learning_rate": 2.8676470588235295e-05,
|
| 706 |
+
"loss": 1.7129,
|
| 707 |
+
"step": 1150
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 13.18,
|
| 711 |
+
"learning_rate": 2.8588235294117647e-05,
|
| 712 |
+
"loss": 1.6363,
|
| 713 |
+
"step": 1160
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 13.29,
|
| 717 |
+
"learning_rate": 2.8499999999999998e-05,
|
| 718 |
+
"loss": 1.6237,
|
| 719 |
+
"step": 1170
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 13.41,
|
| 723 |
+
"learning_rate": 2.8411764705882353e-05,
|
| 724 |
+
"loss": 1.6011,
|
| 725 |
+
"step": 1180
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 13.52,
|
| 729 |
+
"learning_rate": 2.8323529411764705e-05,
|
| 730 |
+
"loss": 1.6302,
|
| 731 |
+
"step": 1190
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 13.63,
|
| 735 |
+
"learning_rate": 2.823529411764706e-05,
|
| 736 |
+
"loss": 1.6277,
|
| 737 |
+
"step": 1200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 13.75,
|
| 741 |
+
"learning_rate": 2.814705882352941e-05,
|
| 742 |
+
"loss": 1.6095,
|
| 743 |
+
"step": 1210
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 13.86,
|
| 747 |
+
"learning_rate": 2.8058823529411766e-05,
|
| 748 |
+
"loss": 1.6188,
|
| 749 |
+
"step": 1220
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 13.97,
|
| 753 |
+
"learning_rate": 2.7970588235294118e-05,
|
| 754 |
+
"loss": 1.6085,
|
| 755 |
+
"step": 1230
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 14.09,
|
| 759 |
+
"learning_rate": 2.7882352941176473e-05,
|
| 760 |
+
"loss": 1.6951,
|
| 761 |
+
"step": 1240
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 14.2,
|
| 765 |
+
"learning_rate": 2.7794117647058824e-05,
|
| 766 |
+
"loss": 1.6059,
|
| 767 |
+
"step": 1250
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 14.32,
|
| 771 |
+
"learning_rate": 2.770588235294118e-05,
|
| 772 |
+
"loss": 1.6191,
|
| 773 |
+
"step": 1260
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 14.43,
|
| 777 |
+
"learning_rate": 2.7617647058823528e-05,
|
| 778 |
+
"loss": 1.6178,
|
| 779 |
+
"step": 1270
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 14.54,
|
| 783 |
+
"learning_rate": 2.7529411764705883e-05,
|
| 784 |
+
"loss": 1.6226,
|
| 785 |
+
"step": 1280
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 14.66,
|
| 789 |
+
"learning_rate": 2.7441176470588234e-05,
|
| 790 |
+
"loss": 1.6099,
|
| 791 |
+
"step": 1290
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 14.77,
|
| 795 |
+
"learning_rate": 2.735294117647059e-05,
|
| 796 |
+
"loss": 1.5953,
|
| 797 |
+
"step": 1300
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 14.88,
|
| 801 |
+
"learning_rate": 2.726470588235294e-05,
|
| 802 |
+
"loss": 1.6067,
|
| 803 |
+
"step": 1310
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 14.99,
|
| 807 |
+
"learning_rate": 2.7176470588235296e-05,
|
| 808 |
+
"loss": 1.6062,
|
| 809 |
+
"step": 1320
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 15.11,
|
| 813 |
+
"learning_rate": 2.7088235294117647e-05,
|
| 814 |
+
"loss": 1.6788,
|
| 815 |
+
"step": 1330
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 15.23,
|
| 819 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 820 |
+
"loss": 1.5883,
|
| 821 |
+
"step": 1340
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 15.34,
|
| 825 |
+
"learning_rate": 2.6911764705882354e-05,
|
| 826 |
+
"loss": 1.582,
|
| 827 |
+
"step": 1350
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 15.45,
|
| 831 |
+
"learning_rate": 2.682352941176471e-05,
|
| 832 |
+
"loss": 1.5865,
|
| 833 |
+
"step": 1360
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 15.56,
|
| 837 |
+
"learning_rate": 2.673529411764706e-05,
|
| 838 |
+
"loss": 1.5797,
|
| 839 |
+
"step": 1370
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 15.68,
|
| 843 |
+
"learning_rate": 2.6647058823529412e-05,
|
| 844 |
+
"loss": 1.5991,
|
| 845 |
+
"step": 1380
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 15.79,
|
| 849 |
+
"learning_rate": 2.6558823529411764e-05,
|
| 850 |
+
"loss": 1.604,
|
| 851 |
+
"step": 1390
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 15.9,
|
| 855 |
+
"learning_rate": 2.647058823529412e-05,
|
| 856 |
+
"loss": 1.5845,
|
| 857 |
+
"step": 1400
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 16.02,
|
| 861 |
+
"learning_rate": 2.638235294117647e-05,
|
| 862 |
+
"loss": 1.661,
|
| 863 |
+
"step": 1410
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 16.14,
|
| 867 |
+
"learning_rate": 2.6294117647058825e-05,
|
| 868 |
+
"loss": 1.5842,
|
| 869 |
+
"step": 1420
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 16.25,
|
| 873 |
+
"learning_rate": 2.6205882352941177e-05,
|
| 874 |
+
"loss": 1.576,
|
| 875 |
+
"step": 1430
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 16.36,
|
| 879 |
+
"learning_rate": 2.6117647058823532e-05,
|
| 880 |
+
"loss": 1.5771,
|
| 881 |
+
"step": 1440
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 16.47,
|
| 885 |
+
"learning_rate": 2.6029411764705883e-05,
|
| 886 |
+
"loss": 1.5764,
|
| 887 |
+
"step": 1450
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 16.59,
|
| 891 |
+
"learning_rate": 2.594117647058824e-05,
|
| 892 |
+
"loss": 1.586,
|
| 893 |
+
"step": 1460
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 16.7,
|
| 897 |
+
"learning_rate": 2.585294117647059e-05,
|
| 898 |
+
"loss": 1.5773,
|
| 899 |
+
"step": 1470
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 16.81,
|
| 903 |
+
"learning_rate": 2.576470588235294e-05,
|
| 904 |
+
"loss": 1.5814,
|
| 905 |
+
"step": 1480
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 16.93,
|
| 909 |
+
"learning_rate": 2.5676470588235293e-05,
|
| 910 |
+
"loss": 1.5812,
|
| 911 |
+
"step": 1490
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 17.05,
|
| 915 |
+
"learning_rate": 2.5588235294117648e-05,
|
| 916 |
+
"loss": 1.6478,
|
| 917 |
+
"step": 1500
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 17.16,
|
| 921 |
+
"learning_rate": 2.55e-05,
|
| 922 |
+
"loss": 1.5619,
|
| 923 |
+
"step": 1510
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 17.27,
|
| 927 |
+
"learning_rate": 2.5411764705882355e-05,
|
| 928 |
+
"loss": 1.577,
|
| 929 |
+
"step": 1520
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 17.38,
|
| 933 |
+
"learning_rate": 2.5323529411764706e-05,
|
| 934 |
+
"loss": 1.5569,
|
| 935 |
+
"step": 1530
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 17.5,
|
| 939 |
+
"learning_rate": 2.523529411764706e-05,
|
| 940 |
+
"loss": 1.5831,
|
| 941 |
+
"step": 1540
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 17.61,
|
| 945 |
+
"learning_rate": 2.5147058823529413e-05,
|
| 946 |
+
"loss": 1.5613,
|
| 947 |
+
"step": 1550
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 17.72,
|
| 951 |
+
"learning_rate": 2.5058823529411768e-05,
|
| 952 |
+
"loss": 1.5594,
|
| 953 |
+
"step": 1560
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 17.84,
|
| 957 |
+
"learning_rate": 2.497058823529412e-05,
|
| 958 |
+
"loss": 1.5723,
|
| 959 |
+
"step": 1570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 17.95,
|
| 963 |
+
"learning_rate": 2.488235294117647e-05,
|
| 964 |
+
"loss": 1.576,
|
| 965 |
+
"step": 1580
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 18.07,
|
| 969 |
+
"learning_rate": 2.4794117647058823e-05,
|
| 970 |
+
"loss": 1.6366,
|
| 971 |
+
"step": 1590
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 18.18,
|
| 975 |
+
"learning_rate": 2.4705882352941174e-05,
|
| 976 |
+
"loss": 1.5548,
|
| 977 |
+
"step": 1600
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 18.29,
|
| 981 |
+
"learning_rate": 2.461764705882353e-05,
|
| 982 |
+
"loss": 1.5511,
|
| 983 |
+
"step": 1610
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 18.41,
|
| 987 |
+
"learning_rate": 2.452941176470588e-05,
|
| 988 |
+
"loss": 1.5625,
|
| 989 |
+
"step": 1620
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 18.52,
|
| 993 |
+
"learning_rate": 2.4441176470588236e-05,
|
| 994 |
+
"loss": 1.5676,
|
| 995 |
+
"step": 1630
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 18.63,
|
| 999 |
+
"learning_rate": 2.4352941176470587e-05,
|
| 1000 |
+
"loss": 1.5635,
|
| 1001 |
+
"step": 1640
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 18.75,
|
| 1005 |
+
"learning_rate": 2.4264705882352942e-05,
|
| 1006 |
+
"loss": 1.5552,
|
| 1007 |
+
"step": 1650
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 18.86,
|
| 1011 |
+
"learning_rate": 2.4176470588235294e-05,
|
| 1012 |
+
"loss": 1.5499,
|
| 1013 |
+
"step": 1660
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 18.97,
|
| 1017 |
+
"learning_rate": 2.408823529411765e-05,
|
| 1018 |
+
"loss": 1.5592,
|
| 1019 |
+
"step": 1670
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 19.09,
|
| 1023 |
+
"learning_rate": 2.4e-05,
|
| 1024 |
+
"loss": 1.6249,
|
| 1025 |
+
"step": 1680
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 19.2,
|
| 1029 |
+
"learning_rate": 2.3911764705882356e-05,
|
| 1030 |
+
"loss": 1.5408,
|
| 1031 |
+
"step": 1690
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 19.32,
|
| 1035 |
+
"learning_rate": 2.3823529411764704e-05,
|
| 1036 |
+
"loss": 1.5567,
|
| 1037 |
+
"step": 1700
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 19.43,
|
| 1041 |
+
"learning_rate": 2.373529411764706e-05,
|
| 1042 |
+
"loss": 1.5508,
|
| 1043 |
+
"step": 1710
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 19.54,
|
| 1047 |
+
"learning_rate": 2.364705882352941e-05,
|
| 1048 |
+
"loss": 1.5482,
|
| 1049 |
+
"step": 1720
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 19.66,
|
| 1053 |
+
"learning_rate": 2.3558823529411765e-05,
|
| 1054 |
+
"loss": 1.541,
|
| 1055 |
+
"step": 1730
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 19.77,
|
| 1059 |
+
"learning_rate": 2.3470588235294117e-05,
|
| 1060 |
+
"loss": 1.551,
|
| 1061 |
+
"step": 1740
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 19.88,
|
| 1065 |
+
"learning_rate": 2.3382352941176472e-05,
|
| 1066 |
+
"loss": 1.5503,
|
| 1067 |
+
"step": 1750
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 19.99,
|
| 1071 |
+
"learning_rate": 2.3294117647058824e-05,
|
| 1072 |
+
"loss": 1.5499,
|
| 1073 |
+
"step": 1760
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 20.11,
|
| 1077 |
+
"learning_rate": 2.320588235294118e-05,
|
| 1078 |
+
"loss": 1.6213,
|
| 1079 |
+
"step": 1770
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 20.23,
|
| 1083 |
+
"learning_rate": 2.311764705882353e-05,
|
| 1084 |
+
"loss": 1.5295,
|
| 1085 |
+
"step": 1780
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 20.34,
|
| 1089 |
+
"learning_rate": 2.3029411764705885e-05,
|
| 1090 |
+
"loss": 1.5382,
|
| 1091 |
+
"step": 1790
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 20.45,
|
| 1095 |
+
"learning_rate": 2.2941176470588233e-05,
|
| 1096 |
+
"loss": 1.5405,
|
| 1097 |
+
"step": 1800
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 20.56,
|
| 1101 |
+
"learning_rate": 2.2852941176470588e-05,
|
| 1102 |
+
"loss": 1.5342,
|
| 1103 |
+
"step": 1810
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 20.68,
|
| 1107 |
+
"learning_rate": 2.276470588235294e-05,
|
| 1108 |
+
"loss": 1.5444,
|
| 1109 |
+
"step": 1820
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 20.79,
|
| 1113 |
+
"learning_rate": 2.2676470588235295e-05,
|
| 1114 |
+
"loss": 1.5444,
|
| 1115 |
+
"step": 1830
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 20.9,
|
| 1119 |
+
"learning_rate": 2.2588235294117646e-05,
|
| 1120 |
+
"loss": 1.5363,
|
| 1121 |
+
"step": 1840
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 21.02,
|
| 1125 |
+
"learning_rate": 2.25e-05,
|
| 1126 |
+
"loss": 1.6267,
|
| 1127 |
+
"step": 1850
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 21.14,
|
| 1131 |
+
"learning_rate": 2.2411764705882353e-05,
|
| 1132 |
+
"loss": 1.5449,
|
| 1133 |
+
"step": 1860
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 21.25,
|
| 1137 |
+
"learning_rate": 2.2323529411764708e-05,
|
| 1138 |
+
"loss": 1.5241,
|
| 1139 |
+
"step": 1870
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 21.36,
|
| 1143 |
+
"learning_rate": 2.223529411764706e-05,
|
| 1144 |
+
"loss": 1.5301,
|
| 1145 |
+
"step": 1880
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 21.47,
|
| 1149 |
+
"learning_rate": 2.2147058823529415e-05,
|
| 1150 |
+
"loss": 1.5385,
|
| 1151 |
+
"step": 1890
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 21.59,
|
| 1155 |
+
"learning_rate": 2.2058823529411766e-05,
|
| 1156 |
+
"loss": 1.5206,
|
| 1157 |
+
"step": 1900
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 21.7,
|
| 1161 |
+
"learning_rate": 2.1970588235294118e-05,
|
| 1162 |
+
"loss": 1.5306,
|
| 1163 |
+
"step": 1910
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 21.81,
|
| 1167 |
+
"learning_rate": 2.188235294117647e-05,
|
| 1168 |
+
"loss": 1.5302,
|
| 1169 |
+
"step": 1920
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 21.93,
|
| 1173 |
+
"learning_rate": 2.1794117647058824e-05,
|
| 1174 |
+
"loss": 1.5401,
|
| 1175 |
+
"step": 1930
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 22.05,
|
| 1179 |
+
"learning_rate": 2.1705882352941176e-05,
|
| 1180 |
+
"loss": 1.6063,
|
| 1181 |
+
"step": 1940
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 22.16,
|
| 1185 |
+
"learning_rate": 2.161764705882353e-05,
|
| 1186 |
+
"loss": 1.5251,
|
| 1187 |
+
"step": 1950
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 22.27,
|
| 1191 |
+
"learning_rate": 2.1529411764705882e-05,
|
| 1192 |
+
"loss": 1.5197,
|
| 1193 |
+
"step": 1960
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 22.38,
|
| 1197 |
+
"learning_rate": 2.1441176470588237e-05,
|
| 1198 |
+
"loss": 1.524,
|
| 1199 |
+
"step": 1970
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 22.5,
|
| 1203 |
+
"learning_rate": 2.135294117647059e-05,
|
| 1204 |
+
"loss": 1.5233,
|
| 1205 |
+
"step": 1980
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 22.61,
|
| 1209 |
+
"learning_rate": 2.1264705882352944e-05,
|
| 1210 |
+
"loss": 1.5256,
|
| 1211 |
+
"step": 1990
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 22.72,
|
| 1215 |
+
"learning_rate": 2.1176470588235296e-05,
|
| 1216 |
+
"loss": 1.5264,
|
| 1217 |
+
"step": 2000
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 22.72,
|
| 1221 |
+
"eval_denotation_accuracy": 0.48675379724478984,
|
| 1222 |
+
"eval_loss": 2.4037857055664062,
|
| 1223 |
+
"eval_runtime": 1745.3919,
|
| 1224 |
+
"eval_samples_per_second": 1.622,
|
| 1225 |
+
"eval_steps_per_second": 0.101,
|
| 1226 |
+
"step": 2000
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 22.84,
|
| 1230 |
+
"learning_rate": 2.1088235294117647e-05,
|
| 1231 |
+
"loss": 1.527,
|
| 1232 |
+
"step": 2010
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 22.95,
|
| 1236 |
+
"learning_rate": 2.1e-05,
|
| 1237 |
+
"loss": 1.5229,
|
| 1238 |
+
"step": 2020
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 23.07,
|
| 1242 |
+
"learning_rate": 2.0911764705882354e-05,
|
| 1243 |
+
"loss": 1.6007,
|
| 1244 |
+
"step": 2030
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 23.18,
|
| 1248 |
+
"learning_rate": 2.0823529411764705e-05,
|
| 1249 |
+
"loss": 1.5107,
|
| 1250 |
+
"step": 2040
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 23.29,
|
| 1254 |
+
"learning_rate": 2.073529411764706e-05,
|
| 1255 |
+
"loss": 1.5148,
|
| 1256 |
+
"step": 2050
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 23.41,
|
| 1260 |
+
"learning_rate": 2.0647058823529412e-05,
|
| 1261 |
+
"loss": 1.5174,
|
| 1262 |
+
"step": 2060
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 23.52,
|
| 1266 |
+
"learning_rate": 2.0558823529411767e-05,
|
| 1267 |
+
"loss": 1.5141,
|
| 1268 |
+
"step": 2070
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 23.63,
|
| 1272 |
+
"learning_rate": 2.047058823529412e-05,
|
| 1273 |
+
"loss": 1.5166,
|
| 1274 |
+
"step": 2080
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 23.75,
|
| 1278 |
+
"learning_rate": 2.0382352941176474e-05,
|
| 1279 |
+
"loss": 1.5197,
|
| 1280 |
+
"step": 2090
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 23.86,
|
| 1284 |
+
"learning_rate": 2.0294117647058825e-05,
|
| 1285 |
+
"loss": 1.5154,
|
| 1286 |
+
"step": 2100
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 23.97,
|
| 1290 |
+
"learning_rate": 2.020588235294118e-05,
|
| 1291 |
+
"loss": 1.5155,
|
| 1292 |
+
"step": 2110
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 24.09,
|
| 1296 |
+
"learning_rate": 2.011764705882353e-05,
|
| 1297 |
+
"loss": 1.5843,
|
| 1298 |
+
"step": 2120
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 24.2,
|
| 1302 |
+
"learning_rate": 2.0029411764705883e-05,
|
| 1303 |
+
"loss": 1.5217,
|
| 1304 |
+
"step": 2130
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 24.32,
|
| 1308 |
+
"learning_rate": 1.9941176470588235e-05,
|
| 1309 |
+
"loss": 1.5099,
|
| 1310 |
+
"step": 2140
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 24.43,
|
| 1314 |
+
"learning_rate": 1.9852941176470586e-05,
|
| 1315 |
+
"loss": 1.5118,
|
| 1316 |
+
"step": 2150
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 24.54,
|
| 1320 |
+
"learning_rate": 1.976470588235294e-05,
|
| 1321 |
+
"loss": 1.5216,
|
| 1322 |
+
"step": 2160
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 24.66,
|
| 1326 |
+
"learning_rate": 1.9676470588235293e-05,
|
| 1327 |
+
"loss": 1.5088,
|
| 1328 |
+
"step": 2170
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 24.77,
|
| 1332 |
+
"learning_rate": 1.9588235294117648e-05,
|
| 1333 |
+
"loss": 1.5125,
|
| 1334 |
+
"step": 2180
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 24.88,
|
| 1338 |
+
"learning_rate": 1.95e-05,
|
| 1339 |
+
"loss": 1.517,
|
| 1340 |
+
"step": 2190
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 24.99,
|
| 1344 |
+
"learning_rate": 1.9411764705882355e-05,
|
| 1345 |
+
"loss": 1.5136,
|
| 1346 |
+
"step": 2200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 25.11,
|
| 1350 |
+
"learning_rate": 1.9323529411764706e-05,
|
| 1351 |
+
"loss": 1.5812,
|
| 1352 |
+
"step": 2210
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 25.23,
|
| 1356 |
+
"learning_rate": 1.923529411764706e-05,
|
| 1357 |
+
"loss": 1.513,
|
| 1358 |
+
"step": 2220
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 25.34,
|
| 1362 |
+
"learning_rate": 1.914705882352941e-05,
|
| 1363 |
+
"loss": 1.5135,
|
| 1364 |
+
"step": 2230
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 25.45,
|
| 1368 |
+
"learning_rate": 1.9058823529411764e-05,
|
| 1369 |
+
"loss": 1.5066,
|
| 1370 |
+
"step": 2240
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 25.56,
|
| 1374 |
+
"learning_rate": 1.8970588235294116e-05,
|
| 1375 |
+
"loss": 1.5087,
|
| 1376 |
+
"step": 2250
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 25.68,
|
| 1380 |
+
"learning_rate": 1.888235294117647e-05,
|
| 1381 |
+
"loss": 1.5013,
|
| 1382 |
+
"step": 2260
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 25.79,
|
| 1386 |
+
"learning_rate": 1.8794117647058823e-05,
|
| 1387 |
+
"loss": 1.5059,
|
| 1388 |
+
"step": 2270
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 25.9,
|
| 1392 |
+
"learning_rate": 1.8705882352941178e-05,
|
| 1393 |
+
"loss": 1.5057,
|
| 1394 |
+
"step": 2280
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 26.02,
|
| 1398 |
+
"learning_rate": 1.861764705882353e-05,
|
| 1399 |
+
"loss": 1.579,
|
| 1400 |
+
"step": 2290
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 26.14,
|
| 1404 |
+
"learning_rate": 1.8529411764705884e-05,
|
| 1405 |
+
"loss": 1.5009,
|
| 1406 |
+
"step": 2300
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 26.25,
|
| 1410 |
+
"learning_rate": 1.8441176470588236e-05,
|
| 1411 |
+
"loss": 1.4947,
|
| 1412 |
+
"step": 2310
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 26.36,
|
| 1416 |
+
"learning_rate": 1.835294117647059e-05,
|
| 1417 |
+
"loss": 1.502,
|
| 1418 |
+
"step": 2320
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 26.47,
|
| 1422 |
+
"learning_rate": 1.826470588235294e-05,
|
| 1423 |
+
"loss": 1.5006,
|
| 1424 |
+
"step": 2330
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 26.59,
|
| 1428 |
+
"learning_rate": 1.8176470588235294e-05,
|
| 1429 |
+
"loss": 1.4958,
|
| 1430 |
+
"step": 2340
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 26.7,
|
| 1434 |
+
"learning_rate": 1.8088235294117645e-05,
|
| 1435 |
+
"loss": 1.5004,
|
| 1436 |
+
"step": 2350
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 26.81,
|
| 1440 |
+
"learning_rate": 1.8e-05,
|
| 1441 |
+
"loss": 1.497,
|
| 1442 |
+
"step": 2360
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 26.93,
|
| 1446 |
+
"learning_rate": 1.7911764705882352e-05,
|
| 1447 |
+
"loss": 1.5019,
|
| 1448 |
+
"step": 2370
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 27.05,
|
| 1452 |
+
"learning_rate": 1.7823529411764707e-05,
|
| 1453 |
+
"loss": 1.5746,
|
| 1454 |
+
"step": 2380
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 27.16,
|
| 1458 |
+
"learning_rate": 1.773529411764706e-05,
|
| 1459 |
+
"loss": 1.4954,
|
| 1460 |
+
"step": 2390
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 27.27,
|
| 1464 |
+
"learning_rate": 1.7647058823529414e-05,
|
| 1465 |
+
"loss": 1.4968,
|
| 1466 |
+
"step": 2400
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 27.38,
|
| 1470 |
+
"learning_rate": 1.7558823529411765e-05,
|
| 1471 |
+
"loss": 1.4959,
|
| 1472 |
+
"step": 2410
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 27.5,
|
| 1476 |
+
"learning_rate": 1.747058823529412e-05,
|
| 1477 |
+
"loss": 1.5,
|
| 1478 |
+
"step": 2420
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 27.61,
|
| 1482 |
+
"learning_rate": 1.7382352941176472e-05,
|
| 1483 |
+
"loss": 1.4945,
|
| 1484 |
+
"step": 2430
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 27.72,
|
| 1488 |
+
"learning_rate": 1.7294117647058823e-05,
|
| 1489 |
+
"loss": 1.493,
|
| 1490 |
+
"step": 2440
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 27.84,
|
| 1494 |
+
"learning_rate": 1.7205882352941175e-05,
|
| 1495 |
+
"loss": 1.5007,
|
| 1496 |
+
"step": 2450
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 27.95,
|
| 1500 |
+
"learning_rate": 1.711764705882353e-05,
|
| 1501 |
+
"loss": 1.5059,
|
| 1502 |
+
"step": 2460
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 28.07,
|
| 1506 |
+
"learning_rate": 1.702941176470588e-05,
|
| 1507 |
+
"loss": 1.5709,
|
| 1508 |
+
"step": 2470
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 28.18,
|
| 1512 |
+
"learning_rate": 1.6941176470588237e-05,
|
| 1513 |
+
"loss": 1.4905,
|
| 1514 |
+
"step": 2480
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 28.29,
|
| 1518 |
+
"learning_rate": 1.6852941176470588e-05,
|
| 1519 |
+
"loss": 1.4894,
|
| 1520 |
+
"step": 2490
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 28.41,
|
| 1524 |
+
"learning_rate": 1.6764705882352943e-05,
|
| 1525 |
+
"loss": 1.5007,
|
| 1526 |
+
"step": 2500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 28.52,
|
| 1530 |
+
"learning_rate": 1.6676470588235295e-05,
|
| 1531 |
+
"loss": 1.4908,
|
| 1532 |
+
"step": 2510
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 28.63,
|
| 1536 |
+
"learning_rate": 1.658823529411765e-05,
|
| 1537 |
+
"loss": 1.4912,
|
| 1538 |
+
"step": 2520
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 28.75,
|
| 1542 |
+
"learning_rate": 1.65e-05,
|
| 1543 |
+
"loss": 1.493,
|
| 1544 |
+
"step": 2530
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 28.86,
|
| 1548 |
+
"learning_rate": 1.6411764705882356e-05,
|
| 1549 |
+
"loss": 1.4957,
|
| 1550 |
+
"step": 2540
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 28.97,
|
| 1554 |
+
"learning_rate": 1.6323529411764704e-05,
|
| 1555 |
+
"loss": 1.4919,
|
| 1556 |
+
"step": 2550
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 29.09,
|
| 1560 |
+
"learning_rate": 1.623529411764706e-05,
|
| 1561 |
+
"loss": 1.5667,
|
| 1562 |
+
"step": 2560
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 29.2,
|
| 1566 |
+
"learning_rate": 1.614705882352941e-05,
|
| 1567 |
+
"loss": 1.4908,
|
| 1568 |
+
"step": 2570
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 29.32,
|
| 1572 |
+
"learning_rate": 1.6058823529411766e-05,
|
| 1573 |
+
"loss": 1.4893,
|
| 1574 |
+
"step": 2580
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 29.43,
|
| 1578 |
+
"learning_rate": 1.5970588235294118e-05,
|
| 1579 |
+
"loss": 1.4898,
|
| 1580 |
+
"step": 2590
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 29.54,
|
| 1584 |
+
"learning_rate": 1.5882352941176473e-05,
|
| 1585 |
+
"loss": 1.4929,
|
| 1586 |
+
"step": 2600
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 29.66,
|
| 1590 |
+
"learning_rate": 1.5794117647058824e-05,
|
| 1591 |
+
"loss": 1.4858,
|
| 1592 |
+
"step": 2610
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 29.77,
|
| 1596 |
+
"learning_rate": 1.570588235294118e-05,
|
| 1597 |
+
"loss": 1.4946,
|
| 1598 |
+
"step": 2620
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 29.88,
|
| 1602 |
+
"learning_rate": 1.561764705882353e-05,
|
| 1603 |
+
"loss": 1.491,
|
| 1604 |
+
"step": 2630
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 29.99,
|
| 1608 |
+
"learning_rate": 1.5529411764705886e-05,
|
| 1609 |
+
"loss": 1.4873,
|
| 1610 |
+
"step": 2640
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 30.11,
|
| 1614 |
+
"learning_rate": 1.5441176470588234e-05,
|
| 1615 |
+
"loss": 1.5672,
|
| 1616 |
+
"step": 2650
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 30.23,
|
| 1620 |
+
"learning_rate": 1.535294117647059e-05,
|
| 1621 |
+
"loss": 1.4865,
|
| 1622 |
+
"step": 2660
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 30.34,
|
| 1626 |
+
"learning_rate": 1.526470588235294e-05,
|
| 1627 |
+
"loss": 1.4895,
|
| 1628 |
+
"step": 2670
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 30.45,
|
| 1632 |
+
"learning_rate": 1.5176470588235294e-05,
|
| 1633 |
+
"loss": 1.485,
|
| 1634 |
+
"step": 2680
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 30.56,
|
| 1638 |
+
"learning_rate": 1.5088235294117647e-05,
|
| 1639 |
+
"loss": 1.4892,
|
| 1640 |
+
"step": 2690
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 30.68,
|
| 1644 |
+
"learning_rate": 1.5e-05,
|
| 1645 |
+
"loss": 1.4854,
|
| 1646 |
+
"step": 2700
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 30.79,
|
| 1650 |
+
"learning_rate": 1.4911764705882354e-05,
|
| 1651 |
+
"loss": 1.4862,
|
| 1652 |
+
"step": 2710
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 30.9,
|
| 1656 |
+
"learning_rate": 1.4823529411764707e-05,
|
| 1657 |
+
"loss": 1.4886,
|
| 1658 |
+
"step": 2720
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 31.02,
|
| 1662 |
+
"learning_rate": 1.4735294117647059e-05,
|
| 1663 |
+
"loss": 1.5558,
|
| 1664 |
+
"step": 2730
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 31.14,
|
| 1668 |
+
"learning_rate": 1.4647058823529412e-05,
|
| 1669 |
+
"loss": 1.4916,
|
| 1670 |
+
"step": 2740
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 31.25,
|
| 1674 |
+
"learning_rate": 1.4558823529411765e-05,
|
| 1675 |
+
"loss": 1.4892,
|
| 1676 |
+
"step": 2750
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 31.36,
|
| 1680 |
+
"learning_rate": 1.4470588235294118e-05,
|
| 1681 |
+
"loss": 1.4861,
|
| 1682 |
+
"step": 2760
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 31.47,
|
| 1686 |
+
"learning_rate": 1.4382352941176472e-05,
|
| 1687 |
+
"loss": 1.4841,
|
| 1688 |
+
"step": 2770
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 31.59,
|
| 1692 |
+
"learning_rate": 1.4294117647058823e-05,
|
| 1693 |
+
"loss": 1.4834,
|
| 1694 |
+
"step": 2780
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 31.7,
|
| 1698 |
+
"learning_rate": 1.4205882352941177e-05,
|
| 1699 |
+
"loss": 1.4842,
|
| 1700 |
+
"step": 2790
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 31.81,
|
| 1704 |
+
"learning_rate": 1.411764705882353e-05,
|
| 1705 |
+
"loss": 1.4834,
|
| 1706 |
+
"step": 2800
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 31.93,
|
| 1710 |
+
"learning_rate": 1.4029411764705883e-05,
|
| 1711 |
+
"loss": 1.483,
|
| 1712 |
+
"step": 2810
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 32.05,
|
| 1716 |
+
"learning_rate": 1.3941176470588236e-05,
|
| 1717 |
+
"loss": 1.5567,
|
| 1718 |
+
"step": 2820
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 32.16,
|
| 1722 |
+
"learning_rate": 1.385294117647059e-05,
|
| 1723 |
+
"loss": 1.4841,
|
| 1724 |
+
"step": 2830
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 32.27,
|
| 1728 |
+
"learning_rate": 1.3764705882352941e-05,
|
| 1729 |
+
"loss": 1.4858,
|
| 1730 |
+
"step": 2840
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 32.38,
|
| 1734 |
+
"learning_rate": 1.3676470588235295e-05,
|
| 1735 |
+
"loss": 1.483,
|
| 1736 |
+
"step": 2850
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 32.5,
|
| 1740 |
+
"learning_rate": 1.3588235294117648e-05,
|
| 1741 |
+
"loss": 1.4781,
|
| 1742 |
+
"step": 2860
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 32.61,
|
| 1746 |
+
"learning_rate": 1.3500000000000001e-05,
|
| 1747 |
+
"loss": 1.4818,
|
| 1748 |
+
"step": 2870
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 32.72,
|
| 1752 |
+
"learning_rate": 1.3411764705882354e-05,
|
| 1753 |
+
"loss": 1.4802,
|
| 1754 |
+
"step": 2880
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 32.84,
|
| 1758 |
+
"learning_rate": 1.3323529411764706e-05,
|
| 1759 |
+
"loss": 1.4835,
|
| 1760 |
+
"step": 2890
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 32.95,
|
| 1764 |
+
"learning_rate": 1.323529411764706e-05,
|
| 1765 |
+
"loss": 1.486,
|
| 1766 |
+
"step": 2900
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 33.07,
|
| 1770 |
+
"learning_rate": 1.3147058823529413e-05,
|
| 1771 |
+
"loss": 1.5547,
|
| 1772 |
+
"step": 2910
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 33.18,
|
| 1776 |
+
"learning_rate": 1.3058823529411766e-05,
|
| 1777 |
+
"loss": 1.4813,
|
| 1778 |
+
"step": 2920
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 33.29,
|
| 1782 |
+
"learning_rate": 1.297058823529412e-05,
|
| 1783 |
+
"loss": 1.4857,
|
| 1784 |
+
"step": 2930
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 33.41,
|
| 1788 |
+
"learning_rate": 1.288235294117647e-05,
|
| 1789 |
+
"loss": 1.4853,
|
| 1790 |
+
"step": 2940
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 33.52,
|
| 1794 |
+
"learning_rate": 1.2794117647058824e-05,
|
| 1795 |
+
"loss": 1.4772,
|
| 1796 |
+
"step": 2950
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 33.63,
|
| 1800 |
+
"learning_rate": 1.2705882352941177e-05,
|
| 1801 |
+
"loss": 1.4826,
|
| 1802 |
+
"step": 2960
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 33.75,
|
| 1806 |
+
"learning_rate": 1.261764705882353e-05,
|
| 1807 |
+
"loss": 1.484,
|
| 1808 |
+
"step": 2970
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 33.86,
|
| 1812 |
+
"learning_rate": 1.2529411764705884e-05,
|
| 1813 |
+
"loss": 1.4864,
|
| 1814 |
+
"step": 2980
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 33.97,
|
| 1818 |
+
"learning_rate": 1.2441176470588236e-05,
|
| 1819 |
+
"loss": 1.4808,
|
| 1820 |
+
"step": 2990
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 34.09,
|
| 1824 |
+
"learning_rate": 1.2352941176470587e-05,
|
| 1825 |
+
"loss": 1.5574,
|
| 1826 |
+
"step": 3000
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 34.09,
|
| 1830 |
+
"eval_denotation_accuracy": 0.42352525609325326,
|
| 1831 |
+
"eval_loss": 2.4281280040740967,
|
| 1832 |
+
"eval_runtime": 1744.8445,
|
| 1833 |
+
"eval_samples_per_second": 1.622,
|
| 1834 |
+
"eval_steps_per_second": 0.101,
|
| 1835 |
+
"step": 3000
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 34.2,
|
| 1839 |
+
"learning_rate": 1.226470588235294e-05,
|
| 1840 |
+
"loss": 1.4832,
|
| 1841 |
+
"step": 3010
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 34.32,
|
| 1845 |
+
"learning_rate": 1.2176470588235294e-05,
|
| 1846 |
+
"loss": 1.4815,
|
| 1847 |
+
"step": 3020
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 34.43,
|
| 1851 |
+
"learning_rate": 1.2088235294117647e-05,
|
| 1852 |
+
"loss": 1.4813,
|
| 1853 |
+
"step": 3030
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 34.54,
|
| 1857 |
+
"learning_rate": 1.2e-05,
|
| 1858 |
+
"loss": 1.4826,
|
| 1859 |
+
"step": 3040
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 34.66,
|
| 1863 |
+
"learning_rate": 1.1911764705882352e-05,
|
| 1864 |
+
"loss": 1.4826,
|
| 1865 |
+
"step": 3050
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 34.77,
|
| 1869 |
+
"learning_rate": 1.1823529411764705e-05,
|
| 1870 |
+
"loss": 1.4812,
|
| 1871 |
+
"step": 3060
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 34.88,
|
| 1875 |
+
"learning_rate": 1.1735294117647058e-05,
|
| 1876 |
+
"loss": 1.48,
|
| 1877 |
+
"step": 3070
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 34.99,
|
| 1881 |
+
"learning_rate": 1.1647058823529412e-05,
|
| 1882 |
+
"loss": 1.4818,
|
| 1883 |
+
"step": 3080
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 35.11,
|
| 1887 |
+
"learning_rate": 1.1558823529411765e-05,
|
| 1888 |
+
"loss": 1.5577,
|
| 1889 |
+
"step": 3090
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 35.23,
|
| 1893 |
+
"learning_rate": 1.1470588235294117e-05,
|
| 1894 |
+
"loss": 1.4789,
|
| 1895 |
+
"step": 3100
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 35.34,
|
| 1899 |
+
"learning_rate": 1.138235294117647e-05,
|
| 1900 |
+
"loss": 1.477,
|
| 1901 |
+
"step": 3110
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 35.45,
|
| 1905 |
+
"learning_rate": 1.1294117647058823e-05,
|
| 1906 |
+
"loss": 1.4778,
|
| 1907 |
+
"step": 3120
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 35.56,
|
| 1911 |
+
"learning_rate": 1.1205882352941177e-05,
|
| 1912 |
+
"loss": 1.476,
|
| 1913 |
+
"step": 3130
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 35.68,
|
| 1917 |
+
"learning_rate": 1.111764705882353e-05,
|
| 1918 |
+
"loss": 1.4799,
|
| 1919 |
+
"step": 3140
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 35.79,
|
| 1923 |
+
"learning_rate": 1.1029411764705883e-05,
|
| 1924 |
+
"loss": 1.4747,
|
| 1925 |
+
"step": 3150
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 35.9,
|
| 1929 |
+
"learning_rate": 1.0941176470588235e-05,
|
| 1930 |
+
"loss": 1.4748,
|
| 1931 |
+
"step": 3160
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 36.02,
|
| 1935 |
+
"learning_rate": 1.0852941176470588e-05,
|
| 1936 |
+
"loss": 1.5561,
|
| 1937 |
+
"step": 3170
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 36.14,
|
| 1941 |
+
"learning_rate": 1.0764705882352941e-05,
|
| 1942 |
+
"loss": 1.4779,
|
| 1943 |
+
"step": 3180
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 36.25,
|
| 1947 |
+
"learning_rate": 1.0676470588235295e-05,
|
| 1948 |
+
"loss": 1.4774,
|
| 1949 |
+
"step": 3190
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 36.36,
|
| 1953 |
+
"learning_rate": 1.0588235294117648e-05,
|
| 1954 |
+
"loss": 1.4743,
|
| 1955 |
+
"step": 3200
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 36.47,
|
| 1959 |
+
"learning_rate": 1.05e-05,
|
| 1960 |
+
"loss": 1.4765,
|
| 1961 |
+
"step": 3210
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 36.59,
|
| 1965 |
+
"learning_rate": 1.0411764705882353e-05,
|
| 1966 |
+
"loss": 1.4755,
|
| 1967 |
+
"step": 3220
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 36.7,
|
| 1971 |
+
"learning_rate": 1.0323529411764706e-05,
|
| 1972 |
+
"loss": 1.4796,
|
| 1973 |
+
"step": 3230
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 36.81,
|
| 1977 |
+
"learning_rate": 1.023529411764706e-05,
|
| 1978 |
+
"loss": 1.4723,
|
| 1979 |
+
"step": 3240
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 36.93,
|
| 1983 |
+
"learning_rate": 1.0147058823529413e-05,
|
| 1984 |
+
"loss": 1.4781,
|
| 1985 |
+
"step": 3250
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 37.05,
|
| 1989 |
+
"learning_rate": 1.0058823529411764e-05,
|
| 1990 |
+
"loss": 1.5533,
|
| 1991 |
+
"step": 3260
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 37.16,
|
| 1995 |
+
"learning_rate": 9.970588235294117e-06,
|
| 1996 |
+
"loss": 1.4733,
|
| 1997 |
+
"step": 3270
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 37.27,
|
| 2001 |
+
"learning_rate": 9.88235294117647e-06,
|
| 2002 |
+
"loss": 1.4761,
|
| 2003 |
+
"step": 3280
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 37.38,
|
| 2007 |
+
"learning_rate": 9.794117647058824e-06,
|
| 2008 |
+
"loss": 1.4738,
|
| 2009 |
+
"step": 3290
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 37.5,
|
| 2013 |
+
"learning_rate": 9.705882352941177e-06,
|
| 2014 |
+
"loss": 1.4731,
|
| 2015 |
+
"step": 3300
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 37.61,
|
| 2019 |
+
"learning_rate": 9.61764705882353e-06,
|
| 2020 |
+
"loss": 1.4733,
|
| 2021 |
+
"step": 3310
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 37.72,
|
| 2025 |
+
"learning_rate": 9.529411764705882e-06,
|
| 2026 |
+
"loss": 1.4738,
|
| 2027 |
+
"step": 3320
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 37.84,
|
| 2031 |
+
"learning_rate": 9.441176470588235e-06,
|
| 2032 |
+
"loss": 1.4742,
|
| 2033 |
+
"step": 3330
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 37.95,
|
| 2037 |
+
"learning_rate": 9.352941176470589e-06,
|
| 2038 |
+
"loss": 1.4721,
|
| 2039 |
+
"step": 3340
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 38.07,
|
| 2043 |
+
"learning_rate": 9.264705882352942e-06,
|
| 2044 |
+
"loss": 1.5433,
|
| 2045 |
+
"step": 3350
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 38.18,
|
| 2049 |
+
"learning_rate": 9.176470588235295e-06,
|
| 2050 |
+
"loss": 1.4746,
|
| 2051 |
+
"step": 3360
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 38.29,
|
| 2055 |
+
"learning_rate": 9.088235294117647e-06,
|
| 2056 |
+
"loss": 1.4754,
|
| 2057 |
+
"step": 3370
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 38.41,
|
| 2061 |
+
"learning_rate": 9e-06,
|
| 2062 |
+
"loss": 1.4758,
|
| 2063 |
+
"step": 3380
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 38.52,
|
| 2067 |
+
"learning_rate": 8.911764705882354e-06,
|
| 2068 |
+
"loss": 1.4705,
|
| 2069 |
+
"step": 3390
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 38.63,
|
| 2073 |
+
"learning_rate": 8.823529411764707e-06,
|
| 2074 |
+
"loss": 1.4709,
|
| 2075 |
+
"step": 3400
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 38.75,
|
| 2079 |
+
"learning_rate": 8.73529411764706e-06,
|
| 2080 |
+
"loss": 1.4717,
|
| 2081 |
+
"step": 3410
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 38.86,
|
| 2085 |
+
"learning_rate": 8.647058823529412e-06,
|
| 2086 |
+
"loss": 1.4741,
|
| 2087 |
+
"step": 3420
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 38.97,
|
| 2091 |
+
"learning_rate": 8.558823529411765e-06,
|
| 2092 |
+
"loss": 1.47,
|
| 2093 |
+
"step": 3430
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 39.09,
|
| 2097 |
+
"learning_rate": 8.470588235294118e-06,
|
| 2098 |
+
"loss": 1.543,
|
| 2099 |
+
"step": 3440
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 39.2,
|
| 2103 |
+
"learning_rate": 8.382352941176472e-06,
|
| 2104 |
+
"loss": 1.4719,
|
| 2105 |
+
"step": 3450
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 39.32,
|
| 2109 |
+
"learning_rate": 8.294117647058825e-06,
|
| 2110 |
+
"loss": 1.4715,
|
| 2111 |
+
"step": 3460
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 39.43,
|
| 2115 |
+
"learning_rate": 8.205882352941178e-06,
|
| 2116 |
+
"loss": 1.4681,
|
| 2117 |
+
"step": 3470
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 39.54,
|
| 2121 |
+
"learning_rate": 8.11764705882353e-06,
|
| 2122 |
+
"loss": 1.4733,
|
| 2123 |
+
"step": 3480
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 39.66,
|
| 2127 |
+
"learning_rate": 8.029411764705883e-06,
|
| 2128 |
+
"loss": 1.4716,
|
| 2129 |
+
"step": 3490
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 39.77,
|
| 2133 |
+
"learning_rate": 7.941176470588236e-06,
|
| 2134 |
+
"loss": 1.4694,
|
| 2135 |
+
"step": 3500
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 39.88,
|
| 2139 |
+
"learning_rate": 7.85294117647059e-06,
|
| 2140 |
+
"loss": 1.4726,
|
| 2141 |
+
"step": 3510
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 39.99,
|
| 2145 |
+
"learning_rate": 7.764705882352943e-06,
|
| 2146 |
+
"loss": 1.4738,
|
| 2147 |
+
"step": 3520
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 40.11,
|
| 2151 |
+
"learning_rate": 7.676470588235294e-06,
|
| 2152 |
+
"loss": 1.5443,
|
| 2153 |
+
"step": 3530
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 40.23,
|
| 2157 |
+
"learning_rate": 7.588235294117647e-06,
|
| 2158 |
+
"loss": 1.4713,
|
| 2159 |
+
"step": 3540
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 40.34,
|
| 2163 |
+
"learning_rate": 7.5e-06,
|
| 2164 |
+
"loss": 1.4673,
|
| 2165 |
+
"step": 3550
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 40.45,
|
| 2169 |
+
"learning_rate": 7.4117647058823535e-06,
|
| 2170 |
+
"loss": 1.4705,
|
| 2171 |
+
"step": 3560
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 40.56,
|
| 2175 |
+
"learning_rate": 7.323529411764706e-06,
|
| 2176 |
+
"loss": 1.4695,
|
| 2177 |
+
"step": 3570
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 40.68,
|
| 2181 |
+
"learning_rate": 7.235294117647059e-06,
|
| 2182 |
+
"loss": 1.4678,
|
| 2183 |
+
"step": 3580
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 40.79,
|
| 2187 |
+
"learning_rate": 7.147058823529412e-06,
|
| 2188 |
+
"loss": 1.4679,
|
| 2189 |
+
"step": 3590
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 40.9,
|
| 2193 |
+
"learning_rate": 7.058823529411765e-06,
|
| 2194 |
+
"loss": 1.4693,
|
| 2195 |
+
"step": 3600
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 41.02,
|
| 2199 |
+
"learning_rate": 6.970588235294118e-06,
|
| 2200 |
+
"loss": 1.5428,
|
| 2201 |
+
"step": 3610
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 41.14,
|
| 2205 |
+
"learning_rate": 6.882352941176471e-06,
|
| 2206 |
+
"loss": 1.4741,
|
| 2207 |
+
"step": 3620
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 41.25,
|
| 2211 |
+
"learning_rate": 6.794117647058824e-06,
|
| 2212 |
+
"loss": 1.4677,
|
| 2213 |
+
"step": 3630
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 41.36,
|
| 2217 |
+
"learning_rate": 6.705882352941177e-06,
|
| 2218 |
+
"loss": 1.4695,
|
| 2219 |
+
"step": 3640
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 41.47,
|
| 2223 |
+
"learning_rate": 6.61764705882353e-06,
|
| 2224 |
+
"loss": 1.4659,
|
| 2225 |
+
"step": 3650
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 41.59,
|
| 2229 |
+
"learning_rate": 6.529411764705883e-06,
|
| 2230 |
+
"loss": 1.4693,
|
| 2231 |
+
"step": 3660
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 41.7,
|
| 2235 |
+
"learning_rate": 6.441176470588235e-06,
|
| 2236 |
+
"loss": 1.469,
|
| 2237 |
+
"step": 3670
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 41.81,
|
| 2241 |
+
"learning_rate": 6.352941176470589e-06,
|
| 2242 |
+
"loss": 1.466,
|
| 2243 |
+
"step": 3680
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 41.93,
|
| 2247 |
+
"learning_rate": 6.264705882352942e-06,
|
| 2248 |
+
"loss": 1.4688,
|
| 2249 |
+
"step": 3690
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 42.05,
|
| 2253 |
+
"learning_rate": 6.176470588235294e-06,
|
| 2254 |
+
"loss": 1.5432,
|
| 2255 |
+
"step": 3700
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 42.16,
|
| 2259 |
+
"learning_rate": 6.088235294117647e-06,
|
| 2260 |
+
"loss": 1.4697,
|
| 2261 |
+
"step": 3710
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 42.27,
|
| 2265 |
+
"learning_rate": 6e-06,
|
| 2266 |
+
"loss": 1.4696,
|
| 2267 |
+
"step": 3720
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 42.38,
|
| 2271 |
+
"learning_rate": 5.911764705882353e-06,
|
| 2272 |
+
"loss": 1.4661,
|
| 2273 |
+
"step": 3730
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 42.5,
|
| 2277 |
+
"learning_rate": 5.823529411764706e-06,
|
| 2278 |
+
"loss": 1.4701,
|
| 2279 |
+
"step": 3740
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 42.61,
|
| 2283 |
+
"learning_rate": 5.735294117647058e-06,
|
| 2284 |
+
"loss": 1.4664,
|
| 2285 |
+
"step": 3750
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 42.72,
|
| 2289 |
+
"learning_rate": 5.647058823529412e-06,
|
| 2290 |
+
"loss": 1.4661,
|
| 2291 |
+
"step": 3760
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 42.84,
|
| 2295 |
+
"learning_rate": 5.558823529411765e-06,
|
| 2296 |
+
"loss": 1.4668,
|
| 2297 |
+
"step": 3770
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 42.95,
|
| 2301 |
+
"learning_rate": 5.470588235294117e-06,
|
| 2302 |
+
"loss": 1.4688,
|
| 2303 |
+
"step": 3780
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 43.07,
|
| 2307 |
+
"learning_rate": 5.382352941176471e-06,
|
| 2308 |
+
"loss": 1.5398,
|
| 2309 |
+
"step": 3790
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 43.18,
|
| 2313 |
+
"learning_rate": 5.294117647058824e-06,
|
| 2314 |
+
"loss": 1.4654,
|
| 2315 |
+
"step": 3800
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 43.29,
|
| 2319 |
+
"learning_rate": 5.205882352941176e-06,
|
| 2320 |
+
"loss": 1.4637,
|
| 2321 |
+
"step": 3810
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 43.41,
|
| 2325 |
+
"learning_rate": 5.11764705882353e-06,
|
| 2326 |
+
"loss": 1.4706,
|
| 2327 |
+
"step": 3820
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 43.52,
|
| 2331 |
+
"learning_rate": 5.029411764705882e-06,
|
| 2332 |
+
"loss": 1.4638,
|
| 2333 |
+
"step": 3830
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 43.63,
|
| 2337 |
+
"learning_rate": 4.941176470588235e-06,
|
| 2338 |
+
"loss": 1.4671,
|
| 2339 |
+
"step": 3840
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 43.75,
|
| 2343 |
+
"learning_rate": 4.852941176470589e-06,
|
| 2344 |
+
"loss": 1.4681,
|
| 2345 |
+
"step": 3850
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 43.86,
|
| 2349 |
+
"learning_rate": 4.764705882352941e-06,
|
| 2350 |
+
"loss": 1.467,
|
| 2351 |
+
"step": 3860
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 43.97,
|
| 2355 |
+
"learning_rate": 4.676470588235294e-06,
|
| 2356 |
+
"loss": 1.4666,
|
| 2357 |
+
"step": 3870
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 44.09,
|
| 2361 |
+
"learning_rate": 4.588235294117648e-06,
|
| 2362 |
+
"loss": 1.5401,
|
| 2363 |
+
"step": 3880
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 44.2,
|
| 2367 |
+
"learning_rate": 4.5e-06,
|
| 2368 |
+
"loss": 1.4642,
|
| 2369 |
+
"step": 3890
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 44.32,
|
| 2373 |
+
"learning_rate": 4.411764705882353e-06,
|
| 2374 |
+
"loss": 1.466,
|
| 2375 |
+
"step": 3900
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 44.43,
|
| 2379 |
+
"learning_rate": 4.323529411764706e-06,
|
| 2380 |
+
"loss": 1.4647,
|
| 2381 |
+
"step": 3910
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 44.54,
|
| 2385 |
+
"learning_rate": 4.235294117647059e-06,
|
| 2386 |
+
"loss": 1.4689,
|
| 2387 |
+
"step": 3920
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 44.66,
|
| 2391 |
+
"learning_rate": 4.147058823529412e-06,
|
| 2392 |
+
"loss": 1.4642,
|
| 2393 |
+
"step": 3930
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 44.77,
|
| 2397 |
+
"learning_rate": 4.058823529411765e-06,
|
| 2398 |
+
"loss": 1.4696,
|
| 2399 |
+
"step": 3940
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 44.88,
|
| 2403 |
+
"learning_rate": 3.970588235294118e-06,
|
| 2404 |
+
"loss": 1.4666,
|
| 2405 |
+
"step": 3950
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 44.99,
|
| 2409 |
+
"learning_rate": 3.8823529411764714e-06,
|
| 2410 |
+
"loss": 1.4678,
|
| 2411 |
+
"step": 3960
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 45.11,
|
| 2415 |
+
"learning_rate": 3.7941176470588235e-06,
|
| 2416 |
+
"loss": 1.5383,
|
| 2417 |
+
"step": 3970
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 45.23,
|
| 2421 |
+
"learning_rate": 3.7058823529411767e-06,
|
| 2422 |
+
"loss": 1.4664,
|
| 2423 |
+
"step": 3980
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 45.34,
|
| 2427 |
+
"learning_rate": 3.6176470588235296e-06,
|
| 2428 |
+
"loss": 1.4645,
|
| 2429 |
+
"step": 3990
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 45.45,
|
| 2433 |
+
"learning_rate": 3.5294117647058825e-06,
|
| 2434 |
+
"loss": 1.4656,
|
| 2435 |
+
"step": 4000
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 45.45,
|
| 2439 |
+
"eval_denotation_accuracy": 0.4959378311550689,
|
| 2440 |
+
"eval_loss": 2.4232473373413086,
|
| 2441 |
+
"eval_runtime": 1762.2635,
|
| 2442 |
+
"eval_samples_per_second": 1.606,
|
| 2443 |
+
"eval_steps_per_second": 0.1,
|
| 2444 |
+
"step": 4000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 45.56,
|
| 2448 |
+
"learning_rate": 3.4411764705882353e-06,
|
| 2449 |
+
"loss": 1.4646,
|
| 2450 |
+
"step": 4010
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 45.68,
|
| 2454 |
+
"learning_rate": 3.3529411764705886e-06,
|
| 2455 |
+
"loss": 1.4656,
|
| 2456 |
+
"step": 4020
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 45.79,
|
| 2460 |
+
"learning_rate": 3.2647058823529415e-06,
|
| 2461 |
+
"loss": 1.4653,
|
| 2462 |
+
"step": 4030
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 45.9,
|
| 2466 |
+
"learning_rate": 3.1764705882352943e-06,
|
| 2467 |
+
"loss": 1.4659,
|
| 2468 |
+
"step": 4040
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 46.02,
|
| 2472 |
+
"learning_rate": 3.088235294117647e-06,
|
| 2473 |
+
"loss": 1.5391,
|
| 2474 |
+
"step": 4050
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 46.14,
|
| 2478 |
+
"learning_rate": 3e-06,
|
| 2479 |
+
"loss": 1.4667,
|
| 2480 |
+
"step": 4060
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 46.25,
|
| 2484 |
+
"learning_rate": 2.911764705882353e-06,
|
| 2485 |
+
"loss": 1.4653,
|
| 2486 |
+
"step": 4070
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 46.36,
|
| 2490 |
+
"learning_rate": 2.823529411764706e-06,
|
| 2491 |
+
"loss": 1.4646,
|
| 2492 |
+
"step": 4080
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 46.47,
|
| 2496 |
+
"learning_rate": 2.7352941176470587e-06,
|
| 2497 |
+
"loss": 1.4662,
|
| 2498 |
+
"step": 4090
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 46.59,
|
| 2502 |
+
"learning_rate": 2.647058823529412e-06,
|
| 2503 |
+
"loss": 1.4678,
|
| 2504 |
+
"step": 4100
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 46.7,
|
| 2508 |
+
"learning_rate": 2.558823529411765e-06,
|
| 2509 |
+
"loss": 1.4651,
|
| 2510 |
+
"step": 4110
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 46.81,
|
| 2514 |
+
"learning_rate": 2.4705882352941177e-06,
|
| 2515 |
+
"loss": 1.4628,
|
| 2516 |
+
"step": 4120
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 46.93,
|
| 2520 |
+
"learning_rate": 2.3823529411764705e-06,
|
| 2521 |
+
"loss": 1.4637,
|
| 2522 |
+
"step": 4130
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 47.05,
|
| 2526 |
+
"learning_rate": 2.294117647058824e-06,
|
| 2527 |
+
"loss": 1.5399,
|
| 2528 |
+
"step": 4140
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 47.16,
|
| 2532 |
+
"learning_rate": 2.2058823529411767e-06,
|
| 2533 |
+
"loss": 1.4615,
|
| 2534 |
+
"step": 4150
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 47.27,
|
| 2538 |
+
"learning_rate": 2.1176470588235296e-06,
|
| 2539 |
+
"loss": 1.4641,
|
| 2540 |
+
"step": 4160
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 47.38,
|
| 2544 |
+
"learning_rate": 2.0294117647058824e-06,
|
| 2545 |
+
"loss": 1.4653,
|
| 2546 |
+
"step": 4170
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 47.5,
|
| 2550 |
+
"learning_rate": 1.9411764705882357e-06,
|
| 2551 |
+
"loss": 1.4626,
|
| 2552 |
+
"step": 4180
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 47.61,
|
| 2556 |
+
"learning_rate": 1.8529411764705884e-06,
|
| 2557 |
+
"loss": 1.4606,
|
| 2558 |
+
"step": 4190
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 47.72,
|
| 2562 |
+
"learning_rate": 1.7647058823529412e-06,
|
| 2563 |
+
"loss": 1.4667,
|
| 2564 |
+
"step": 4200
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 47.84,
|
| 2568 |
+
"learning_rate": 1.6764705882352943e-06,
|
| 2569 |
+
"loss": 1.4606,
|
| 2570 |
+
"step": 4210
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 47.95,
|
| 2574 |
+
"learning_rate": 1.5882352941176472e-06,
|
| 2575 |
+
"loss": 1.4652,
|
| 2576 |
+
"step": 4220
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 48.07,
|
| 2580 |
+
"learning_rate": 1.5e-06,
|
| 2581 |
+
"loss": 1.5363,
|
| 2582 |
+
"step": 4230
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 48.18,
|
| 2586 |
+
"learning_rate": 1.411764705882353e-06,
|
| 2587 |
+
"loss": 1.4629,
|
| 2588 |
+
"step": 4240
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 48.29,
|
| 2592 |
+
"learning_rate": 1.323529411764706e-06,
|
| 2593 |
+
"loss": 1.4609,
|
| 2594 |
+
"step": 4250
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 48.41,
|
| 2598 |
+
"learning_rate": 1.2352941176470588e-06,
|
| 2599 |
+
"loss": 1.4615,
|
| 2600 |
+
"step": 4260
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 48.52,
|
| 2604 |
+
"learning_rate": 1.147058823529412e-06,
|
| 2605 |
+
"loss": 1.4631,
|
| 2606 |
+
"step": 4270
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 48.63,
|
| 2610 |
+
"learning_rate": 1.0588235294117648e-06,
|
| 2611 |
+
"loss": 1.4639,
|
| 2612 |
+
"step": 4280
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 48.75,
|
| 2616 |
+
"learning_rate": 9.705882352941179e-07,
|
| 2617 |
+
"loss": 1.4647,
|
| 2618 |
+
"step": 4290
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 48.86,
|
| 2622 |
+
"learning_rate": 8.823529411764706e-07,
|
| 2623 |
+
"loss": 1.4614,
|
| 2624 |
+
"step": 4300
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 48.97,
|
| 2628 |
+
"learning_rate": 7.941176470588236e-07,
|
| 2629 |
+
"loss": 1.4641,
|
| 2630 |
+
"step": 4310
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 49.09,
|
| 2634 |
+
"learning_rate": 7.058823529411765e-07,
|
| 2635 |
+
"loss": 1.537,
|
| 2636 |
+
"step": 4320
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 49.2,
|
| 2640 |
+
"learning_rate": 6.176470588235294e-07,
|
| 2641 |
+
"loss": 1.4627,
|
| 2642 |
+
"step": 4330
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 49.32,
|
| 2646 |
+
"learning_rate": 5.294117647058824e-07,
|
| 2647 |
+
"loss": 1.4662,
|
| 2648 |
+
"step": 4340
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 49.43,
|
| 2652 |
+
"learning_rate": 4.411764705882353e-07,
|
| 2653 |
+
"loss": 1.4648,
|
| 2654 |
+
"step": 4350
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 49.54,
|
| 2658 |
+
"learning_rate": 3.529411764705882e-07,
|
| 2659 |
+
"loss": 1.4633,
|
| 2660 |
+
"step": 4360
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 49.66,
|
| 2664 |
+
"learning_rate": 2.647058823529412e-07,
|
| 2665 |
+
"loss": 1.4631,
|
| 2666 |
+
"step": 4370
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 49.77,
|
| 2670 |
+
"learning_rate": 1.764705882352941e-07,
|
| 2671 |
+
"loss": 1.4604,
|
| 2672 |
+
"step": 4380
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 49.88,
|
| 2676 |
+
"learning_rate": 8.823529411764706e-08,
|
| 2677 |
+
"loss": 1.4621,
|
| 2678 |
+
"step": 4390
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 49.99,
|
| 2682 |
+
"learning_rate": 0.0,
|
| 2683 |
+
"loss": 1.4644,
|
| 2684 |
+
"step": 4400
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 49.99,
|
| 2688 |
+
"step": 4400,
|
| 2689 |
+
"total_flos": 1.218502107627307e+18,
|
| 2690 |
+
"train_loss": 1.6152976189960133,
|
| 2691 |
+
"train_runtime": 26687.1063,
|
| 2692 |
+
"train_samples_per_second": 21.211,
|
| 2693 |
+
"train_steps_per_second": 0.165
|
| 2694 |
+
}
|
| 2695 |
+
],
|
| 2696 |
+
"max_steps": 4400,
|
| 2697 |
+
"num_train_epochs": 50,
|
| 2698 |
+
"total_flos": 1.218502107627307e+18,
|
| 2699 |
+
"trial_name": null,
|
| 2700 |
+
"trial_params": null
|
| 2701 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:503668f2708455e35b05fe2b93a0fe12f7e5b099c660067bf1dbfd1cc219fb86
|
| 3 |
+
size 3503
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|