tyoyo
commited on
Commit
•
4d42c1b
1
Parent(s):
d60c14f
add train result manually
Browse files- .gitignore +2 -0
- config.json +33 -0
- pytorch_model.bin +3 -0
- runs/Nov26_01-50-41_fd647f080840/1637891461.117952/events.out.tfevents.1637891461.fd647f080840.26316.1 +3 -0
- runs/Nov26_01-50-41_fd647f080840/events.out.tfevents.1637891461.fd647f080840.26316.0 +3 -0
- special_tokens_map.json +1 -0
- spiece.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- trainer_state.json +2102 -0
- training_args.bin +3 -0
.gitignore
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
checkpoint-*
|
| 2 |
+
.ipynb_checkpoints
|
config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "sonoisa/t5-base-japanese",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"T5ForConditionalGeneration"
|
| 5 |
+
],
|
| 6 |
+
"bos_token_id": 0,
|
| 7 |
+
"d_ff": 3072,
|
| 8 |
+
"d_kv": 64,
|
| 9 |
+
"d_model": 768,
|
| 10 |
+
"decoder_start_token_id": 0,
|
| 11 |
+
"dropout_rate": 0.1,
|
| 12 |
+
"eos_token_id": 1,
|
| 13 |
+
"eos_token_ids": [
|
| 14 |
+
1
|
| 15 |
+
],
|
| 16 |
+
"feed_forward_proj": "relu",
|
| 17 |
+
"initializer_factor": 1.0,
|
| 18 |
+
"is_encoder_decoder": true,
|
| 19 |
+
"layer_norm_epsilon": 1e-06,
|
| 20 |
+
"max_length": 512,
|
| 21 |
+
"model_type": "t5",
|
| 22 |
+
"n_positions": 512,
|
| 23 |
+
"num_beams": 4,
|
| 24 |
+
"num_decoder_layers": 12,
|
| 25 |
+
"num_heads": 12,
|
| 26 |
+
"num_layers": 12,
|
| 27 |
+
"pad_token_id": 0,
|
| 28 |
+
"relative_attention_num_buckets": 32,
|
| 29 |
+
"torch_dtype": "float32",
|
| 30 |
+
"transformers_version": "4.12.5",
|
| 31 |
+
"use_cache": true,
|
| 32 |
+
"vocab_size": 32128
|
| 33 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:881490313bdf7dc72f28e72b60d4b4ace54ae3f316275058645217d0dec31ec6
|
| 3 |
+
size 891730879
|
runs/Nov26_01-50-41_fd647f080840/1637891461.117952/events.out.tfevents.1637891461.fd647f080840.26316.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9cb5f4383815ca3c1ddde5bb0d7fb40bc53e342078f8edaf98ac4329abbadef9
|
| 3 |
+
size 4861
|
runs/Nov26_01-50-41_fd647f080840/events.out.tfevents.1637891461.fd647f080840.26316.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:161149154594663816f0cd19890bc1018f37dea226fbcc0dea62bd6e3f8a62f9
|
| 3 |
+
size 69506
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "additional_special_tokens": ["<extra_id_0>", "<extra_id_1>", "<extra_id_2>", "<extra_id_3>", "<extra_id_4>", "<extra_id_5>", "<extra_id_6>", "<extra_id_7>", "<extra_id_8>", "<extra_id_9>", "<extra_id_10>", "<extra_id_11>", "<extra_id_12>", "<extra_id_13>", "<extra_id_14>", "<extra_id_15>", "<extra_id_16>", "<extra_id_17>", "<extra_id_18>", "<extra_id_19>", "<extra_id_20>", "<extra_id_21>", "<extra_id_22>", "<extra_id_23>", "<extra_id_24>", "<extra_id_25>", "<extra_id_26>", "<extra_id_27>", "<extra_id_28>", "<extra_id_29>", "<extra_id_30>", "<extra_id_31>", "<extra_id_32>", "<extra_id_33>", "<extra_id_34>", "<extra_id_35>", "<extra_id_36>", "<extra_id_37>", "<extra_id_38>", "<extra_id_39>", "<extra_id_40>", "<extra_id_41>", "<extra_id_42>", "<extra_id_43>", "<extra_id_44>", "<extra_id_45>", "<extra_id_46>", "<extra_id_47>", "<extra_id_48>", "<extra_id_49>", "<extra_id_50>", "<extra_id_51>", "<extra_id_52>", "<extra_id_53>", "<extra_id_54>", "<extra_id_55>", "<extra_id_56>", "<extra_id_57>", "<extra_id_58>", "<extra_id_59>", "<extra_id_60>", "<extra_id_61>", "<extra_id_62>", "<extra_id_63>", "<extra_id_64>", "<extra_id_65>", "<extra_id_66>", "<extra_id_67>", "<extra_id_68>", "<extra_id_69>", "<extra_id_70>", "<extra_id_71>", "<extra_id_72>", "<extra_id_73>", "<extra_id_74>", "<extra_id_75>", "<extra_id_76>", "<extra_id_77>", "<extra_id_78>", "<extra_id_79>", "<extra_id_80>", "<extra_id_81>", "<extra_id_82>", "<extra_id_83>", "<extra_id_84>", "<extra_id_85>", "<extra_id_86>", "<extra_id_87>", "<extra_id_88>", "<extra_id_89>", "<extra_id_90>", "<extra_id_91>", "<extra_id_92>", "<extra_id_93>", "<extra_id_94>", "<extra_id_95>", "<extra_id_96>", "<extra_id_97>", "<extra_id_98>", "<extra_id_99>"]}
|
spiece.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:575bba4012fcabb11c89be62a45afdcb2ef648e880f021703b96a8a173f6b3b7
|
| 3 |
+
size 804231
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 100, "additional_special_tokens": ["<extra_id_0>", "<extra_id_1>", "<extra_id_2>", "<extra_id_3>", "<extra_id_4>", "<extra_id_5>", "<extra_id_6>", "<extra_id_7>", "<extra_id_8>", "<extra_id_9>", "<extra_id_10>", "<extra_id_11>", "<extra_id_12>", "<extra_id_13>", "<extra_id_14>", "<extra_id_15>", "<extra_id_16>", "<extra_id_17>", "<extra_id_18>", "<extra_id_19>", "<extra_id_20>", "<extra_id_21>", "<extra_id_22>", "<extra_id_23>", "<extra_id_24>", "<extra_id_25>", "<extra_id_26>", "<extra_id_27>", "<extra_id_28>", "<extra_id_29>", "<extra_id_30>", "<extra_id_31>", "<extra_id_32>", "<extra_id_33>", "<extra_id_34>", "<extra_id_35>", "<extra_id_36>", "<extra_id_37>", "<extra_id_38>", "<extra_id_39>", "<extra_id_40>", "<extra_id_41>", "<extra_id_42>", "<extra_id_43>", "<extra_id_44>", "<extra_id_45>", "<extra_id_46>", "<extra_id_47>", "<extra_id_48>", "<extra_id_49>", "<extra_id_50>", "<extra_id_51>", "<extra_id_52>", "<extra_id_53>", "<extra_id_54>", "<extra_id_55>", "<extra_id_56>", "<extra_id_57>", "<extra_id_58>", "<extra_id_59>", "<extra_id_60>", "<extra_id_61>", "<extra_id_62>", "<extra_id_63>", "<extra_id_64>", "<extra_id_65>", "<extra_id_66>", "<extra_id_67>", "<extra_id_68>", "<extra_id_69>", "<extra_id_70>", "<extra_id_71>", "<extra_id_72>", "<extra_id_73>", "<extra_id_74>", "<extra_id_75>", "<extra_id_76>", "<extra_id_77>", "<extra_id_78>", "<extra_id_79>", "<extra_id_80>", "<extra_id_81>", "<extra_id_82>", "<extra_id_83>", "<extra_id_84>", "<extra_id_85>", "<extra_id_86>", "<extra_id_87>", "<extra_id_88>", "<extra_id_89>", "<extra_id_90>", "<extra_id_91>", "<extra_id_92>", "<extra_id_93>", "<extra_id_94>", "<extra_id_95>", "<extra_id_96>", "<extra_id_97>", "<extra_id_98>", "<extra_id_99>"], "is_fast": true, "special_tokens_map_file": null, "name_or_path": "sonoisa/t5-base-japanese", "sp_model_kwargs": {}, "tokenizer_class": "T5Tokenizer"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 8.71313672922252,
|
| 5 |
+
"global_step": 6500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 2.680965147453083e-06,
|
| 13 |
+
"loss": 35.9761,
|
| 14 |
+
"step": 20
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 5.361930294906166e-06,
|
| 19 |
+
"loss": 35.0955,
|
| 20 |
+
"step": 40
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.08,
|
| 24 |
+
"learning_rate": 8.04289544235925e-06,
|
| 25 |
+
"loss": 34.9335,
|
| 26 |
+
"step": 60
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.11,
|
| 30 |
+
"learning_rate": 1.0723860589812333e-05,
|
| 31 |
+
"loss": 34.6391,
|
| 32 |
+
"step": 80
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.13,
|
| 36 |
+
"learning_rate": 1.3404825737265417e-05,
|
| 37 |
+
"loss": 34.1356,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.16,
|
| 42 |
+
"learning_rate": 1.60857908847185e-05,
|
| 43 |
+
"loss": 33.2089,
|
| 44 |
+
"step": 120
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.19,
|
| 48 |
+
"learning_rate": 1.8766756032171583e-05,
|
| 49 |
+
"loss": 32.6544,
|
| 50 |
+
"step": 140
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.21,
|
| 54 |
+
"learning_rate": 2.1447721179624665e-05,
|
| 55 |
+
"loss": 32.3698,
|
| 56 |
+
"step": 160
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.24,
|
| 60 |
+
"learning_rate": 2.4128686327077747e-05,
|
| 61 |
+
"loss": 31.6695,
|
| 62 |
+
"step": 180
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.27,
|
| 66 |
+
"learning_rate": 2.6809651474530833e-05,
|
| 67 |
+
"loss": 32.0937,
|
| 68 |
+
"step": 200
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.29,
|
| 72 |
+
"learning_rate": 2.9490616621983912e-05,
|
| 73 |
+
"loss": 31.3979,
|
| 74 |
+
"step": 220
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.32,
|
| 78 |
+
"learning_rate": 3.2171581769437e-05,
|
| 79 |
+
"loss": 30.5171,
|
| 80 |
+
"step": 240
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.35,
|
| 84 |
+
"learning_rate": 3.485254691689008e-05,
|
| 85 |
+
"loss": 31.0177,
|
| 86 |
+
"step": 260
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.38,
|
| 90 |
+
"learning_rate": 3.7533512064343166e-05,
|
| 91 |
+
"loss": 30.2928,
|
| 92 |
+
"step": 280
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.4,
|
| 96 |
+
"learning_rate": 4.021447721179625e-05,
|
| 97 |
+
"loss": 29.3157,
|
| 98 |
+
"step": 300
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.43,
|
| 102 |
+
"learning_rate": 4.289544235924933e-05,
|
| 103 |
+
"loss": 27.5923,
|
| 104 |
+
"step": 320
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.46,
|
| 108 |
+
"learning_rate": 4.557640750670242e-05,
|
| 109 |
+
"loss": 25.9267,
|
| 110 |
+
"step": 340
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.48,
|
| 114 |
+
"learning_rate": 4.8257372654155495e-05,
|
| 115 |
+
"loss": 23.5854,
|
| 116 |
+
"step": 360
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.51,
|
| 120 |
+
"learning_rate": 5.0938337801608584e-05,
|
| 121 |
+
"loss": 20.8788,
|
| 122 |
+
"step": 380
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.54,
|
| 126 |
+
"learning_rate": 5.3619302949061666e-05,
|
| 127 |
+
"loss": 18.1499,
|
| 128 |
+
"step": 400
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.56,
|
| 132 |
+
"learning_rate": 5.6300268096514755e-05,
|
| 133 |
+
"loss": 14.4047,
|
| 134 |
+
"step": 420
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.59,
|
| 138 |
+
"learning_rate": 5.8981233243967824e-05,
|
| 139 |
+
"loss": 7.4829,
|
| 140 |
+
"step": 440
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.62,
|
| 144 |
+
"learning_rate": 6.16621983914209e-05,
|
| 145 |
+
"loss": 1.5146,
|
| 146 |
+
"step": 460
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.64,
|
| 150 |
+
"learning_rate": 6.4343163538874e-05,
|
| 151 |
+
"loss": 0.9193,
|
| 152 |
+
"step": 480
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.67,
|
| 156 |
+
"learning_rate": 6.702412868632708e-05,
|
| 157 |
+
"loss": 0.8398,
|
| 158 |
+
"step": 500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.7,
|
| 162 |
+
"learning_rate": 6.970509383378017e-05,
|
| 163 |
+
"loss": 0.7985,
|
| 164 |
+
"step": 520
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.72,
|
| 168 |
+
"learning_rate": 7.238605898123325e-05,
|
| 169 |
+
"loss": 0.724,
|
| 170 |
+
"step": 540
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.75,
|
| 174 |
+
"learning_rate": 7.506702412868633e-05,
|
| 175 |
+
"loss": 0.6716,
|
| 176 |
+
"step": 560
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.78,
|
| 180 |
+
"learning_rate": 7.774798927613941e-05,
|
| 181 |
+
"loss": 0.6661,
|
| 182 |
+
"step": 580
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.8,
|
| 186 |
+
"learning_rate": 8.04289544235925e-05,
|
| 187 |
+
"loss": 0.6585,
|
| 188 |
+
"step": 600
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.83,
|
| 192 |
+
"learning_rate": 8.310991957104558e-05,
|
| 193 |
+
"loss": 0.6667,
|
| 194 |
+
"step": 620
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.86,
|
| 198 |
+
"learning_rate": 8.579088471849866e-05,
|
| 199 |
+
"loss": 0.6471,
|
| 200 |
+
"step": 640
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.88,
|
| 204 |
+
"learning_rate": 8.847184986595176e-05,
|
| 205 |
+
"loss": 0.6125,
|
| 206 |
+
"step": 660
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.91,
|
| 210 |
+
"learning_rate": 9.115281501340484e-05,
|
| 211 |
+
"loss": 0.6344,
|
| 212 |
+
"step": 680
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.94,
|
| 216 |
+
"learning_rate": 9.383378016085791e-05,
|
| 217 |
+
"loss": 0.5928,
|
| 218 |
+
"step": 700
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.97,
|
| 222 |
+
"learning_rate": 9.651474530831099e-05,
|
| 223 |
+
"loss": 0.6335,
|
| 224 |
+
"step": 720
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.99,
|
| 228 |
+
"learning_rate": 9.919571045576409e-05,
|
| 229 |
+
"loss": 0.6122,
|
| 230 |
+
"step": 740
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.0,
|
| 234 |
+
"eval_cer": 0.22800093910360325,
|
| 235 |
+
"eval_deletions": 4828,
|
| 236 |
+
"eval_hits": 54224,
|
| 237 |
+
"eval_insertions": 3962,
|
| 238 |
+
"eval_loss": 0.4796110987663269,
|
| 239 |
+
"eval_mer": 0.23053781751099758,
|
| 240 |
+
"eval_runtime": 1011.4956,
|
| 241 |
+
"eval_samples_per_second": 6.204,
|
| 242 |
+
"eval_steps_per_second": 0.776,
|
| 243 |
+
"eval_substitutions": 7456,
|
| 244 |
+
"eval_wer": 0.24427136585072473,
|
| 245 |
+
"eval_wil": 0.3265158023477125,
|
| 246 |
+
"eval_wip": 0.6734841976522875,
|
| 247 |
+
"step": 746
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.02,
|
| 251 |
+
"learning_rate": 9.979148048853143e-05,
|
| 252 |
+
"loss": 0.5954,
|
| 253 |
+
"step": 760
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 1.05,
|
| 257 |
+
"learning_rate": 9.949359547214776e-05,
|
| 258 |
+
"loss": 0.5923,
|
| 259 |
+
"step": 780
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 1.07,
|
| 263 |
+
"learning_rate": 9.919571045576409e-05,
|
| 264 |
+
"loss": 0.5518,
|
| 265 |
+
"step": 800
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.1,
|
| 269 |
+
"learning_rate": 9.88978254393804e-05,
|
| 270 |
+
"loss": 0.5674,
|
| 271 |
+
"step": 820
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 1.13,
|
| 275 |
+
"learning_rate": 9.859994042299673e-05,
|
| 276 |
+
"loss": 0.5602,
|
| 277 |
+
"step": 840
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 1.15,
|
| 281 |
+
"learning_rate": 9.830205540661305e-05,
|
| 282 |
+
"loss": 0.5601,
|
| 283 |
+
"step": 860
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 1.18,
|
| 287 |
+
"learning_rate": 9.800417039022938e-05,
|
| 288 |
+
"loss": 0.5503,
|
| 289 |
+
"step": 880
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 1.21,
|
| 293 |
+
"learning_rate": 9.770628537384569e-05,
|
| 294 |
+
"loss": 0.5754,
|
| 295 |
+
"step": 900
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.23,
|
| 299 |
+
"learning_rate": 9.740840035746203e-05,
|
| 300 |
+
"loss": 0.5782,
|
| 301 |
+
"step": 920
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 1.26,
|
| 305 |
+
"learning_rate": 9.711051534107835e-05,
|
| 306 |
+
"loss": 0.5389,
|
| 307 |
+
"step": 940
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 1.29,
|
| 311 |
+
"learning_rate": 9.681263032469468e-05,
|
| 312 |
+
"loss": 0.571,
|
| 313 |
+
"step": 960
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 1.31,
|
| 317 |
+
"learning_rate": 9.651474530831099e-05,
|
| 318 |
+
"loss": 0.5699,
|
| 319 |
+
"step": 980
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 1.34,
|
| 323 |
+
"learning_rate": 9.621686029192732e-05,
|
| 324 |
+
"loss": 0.526,
|
| 325 |
+
"step": 1000
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.37,
|
| 329 |
+
"learning_rate": 9.591897527554365e-05,
|
| 330 |
+
"loss": 0.5505,
|
| 331 |
+
"step": 1020
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 1.39,
|
| 335 |
+
"learning_rate": 9.562109025915997e-05,
|
| 336 |
+
"loss": 0.5709,
|
| 337 |
+
"step": 1040
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 1.42,
|
| 341 |
+
"learning_rate": 9.532320524277629e-05,
|
| 342 |
+
"loss": 0.5532,
|
| 343 |
+
"step": 1060
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 1.45,
|
| 347 |
+
"learning_rate": 9.502532022639262e-05,
|
| 348 |
+
"loss": 0.5264,
|
| 349 |
+
"step": 1080
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 1.47,
|
| 353 |
+
"learning_rate": 9.472743521000894e-05,
|
| 354 |
+
"loss": 0.5521,
|
| 355 |
+
"step": 1100
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 1.5,
|
| 359 |
+
"learning_rate": 9.442955019362527e-05,
|
| 360 |
+
"loss": 0.5387,
|
| 361 |
+
"step": 1120
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.53,
|
| 365 |
+
"learning_rate": 9.413166517724159e-05,
|
| 366 |
+
"loss": 0.5352,
|
| 367 |
+
"step": 1140
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.55,
|
| 371 |
+
"learning_rate": 9.383378016085791e-05,
|
| 372 |
+
"loss": 0.5279,
|
| 373 |
+
"step": 1160
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 1.58,
|
| 377 |
+
"learning_rate": 9.353589514447424e-05,
|
| 378 |
+
"loss": 0.5198,
|
| 379 |
+
"step": 1180
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 1.61,
|
| 383 |
+
"learning_rate": 9.323801012809056e-05,
|
| 384 |
+
"loss": 0.5806,
|
| 385 |
+
"step": 1200
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 1.64,
|
| 389 |
+
"learning_rate": 9.294012511170688e-05,
|
| 390 |
+
"loss": 0.5378,
|
| 391 |
+
"step": 1220
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.66,
|
| 395 |
+
"learning_rate": 9.26422400953232e-05,
|
| 396 |
+
"loss": 0.5302,
|
| 397 |
+
"step": 1240
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.69,
|
| 401 |
+
"learning_rate": 9.234435507893954e-05,
|
| 402 |
+
"loss": 0.5466,
|
| 403 |
+
"step": 1260
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 1.72,
|
| 407 |
+
"learning_rate": 9.204647006255586e-05,
|
| 408 |
+
"loss": 0.5166,
|
| 409 |
+
"step": 1280
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.74,
|
| 413 |
+
"learning_rate": 9.174858504617218e-05,
|
| 414 |
+
"loss": 0.521,
|
| 415 |
+
"step": 1300
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.77,
|
| 419 |
+
"learning_rate": 9.14507000297885e-05,
|
| 420 |
+
"loss": 0.5241,
|
| 421 |
+
"step": 1320
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 1.8,
|
| 425 |
+
"learning_rate": 9.115281501340484e-05,
|
| 426 |
+
"loss": 0.5219,
|
| 427 |
+
"step": 1340
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.82,
|
| 431 |
+
"learning_rate": 9.085492999702116e-05,
|
| 432 |
+
"loss": 0.5093,
|
| 433 |
+
"step": 1360
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 1.85,
|
| 437 |
+
"learning_rate": 9.055704498063747e-05,
|
| 438 |
+
"loss": 0.5485,
|
| 439 |
+
"step": 1380
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 1.88,
|
| 443 |
+
"learning_rate": 9.02591599642538e-05,
|
| 444 |
+
"loss": 0.5173,
|
| 445 |
+
"step": 1400
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.9,
|
| 449 |
+
"learning_rate": 8.996127494787012e-05,
|
| 450 |
+
"loss": 0.5229,
|
| 451 |
+
"step": 1420
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.93,
|
| 455 |
+
"learning_rate": 8.966338993148646e-05,
|
| 456 |
+
"loss": 0.5474,
|
| 457 |
+
"step": 1440
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.96,
|
| 461 |
+
"learning_rate": 8.936550491510277e-05,
|
| 462 |
+
"loss": 0.5388,
|
| 463 |
+
"step": 1460
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.98,
|
| 467 |
+
"learning_rate": 8.90676198987191e-05,
|
| 468 |
+
"loss": 0.5345,
|
| 469 |
+
"step": 1480
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 2.0,
|
| 473 |
+
"eval_cer": 0.1926671660312811,
|
| 474 |
+
"eval_deletions": 4683,
|
| 475 |
+
"eval_hits": 54896,
|
| 476 |
+
"eval_insertions": 2609,
|
| 477 |
+
"eval_loss": 0.43797770142555237,
|
| 478 |
+
"eval_mer": 0.20575256449209312,
|
| 479 |
+
"eval_runtime": 945.3181,
|
| 480 |
+
"eval_samples_per_second": 6.638,
|
| 481 |
+
"eval_steps_per_second": 0.83,
|
| 482 |
+
"eval_substitutions": 6929,
|
| 483 |
+
"eval_wer": 0.21382390088410416,
|
| 484 |
+
"eval_wil": 0.296778046061947,
|
| 485 |
+
"eval_wip": 0.703221953938053,
|
| 486 |
+
"step": 1492
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 2.01,
|
| 490 |
+
"learning_rate": 8.876973488233542e-05,
|
| 491 |
+
"loss": 0.517,
|
| 492 |
+
"step": 1500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 2.04,
|
| 496 |
+
"learning_rate": 8.847184986595176e-05,
|
| 497 |
+
"loss": 0.4945,
|
| 498 |
+
"step": 1520
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 2.06,
|
| 502 |
+
"learning_rate": 8.817396484956806e-05,
|
| 503 |
+
"loss": 0.4924,
|
| 504 |
+
"step": 1540
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 2.09,
|
| 508 |
+
"learning_rate": 8.78760798331844e-05,
|
| 509 |
+
"loss": 0.4789,
|
| 510 |
+
"step": 1560
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 2.12,
|
| 514 |
+
"learning_rate": 8.757819481680072e-05,
|
| 515 |
+
"loss": 0.487,
|
| 516 |
+
"step": 1580
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 2.14,
|
| 520 |
+
"learning_rate": 8.728030980041705e-05,
|
| 521 |
+
"loss": 0.5086,
|
| 522 |
+
"step": 1600
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 2.17,
|
| 526 |
+
"learning_rate": 8.698242478403336e-05,
|
| 527 |
+
"loss": 0.4746,
|
| 528 |
+
"step": 1620
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 2.2,
|
| 532 |
+
"learning_rate": 8.668453976764968e-05,
|
| 533 |
+
"loss": 0.4631,
|
| 534 |
+
"step": 1640
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.23,
|
| 538 |
+
"learning_rate": 8.638665475126602e-05,
|
| 539 |
+
"loss": 0.509,
|
| 540 |
+
"step": 1660
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 2.25,
|
| 544 |
+
"learning_rate": 8.608876973488234e-05,
|
| 545 |
+
"loss": 0.4867,
|
| 546 |
+
"step": 1680
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 2.28,
|
| 550 |
+
"learning_rate": 8.579088471849866e-05,
|
| 551 |
+
"loss": 0.4788,
|
| 552 |
+
"step": 1700
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 2.31,
|
| 556 |
+
"learning_rate": 8.549299970211498e-05,
|
| 557 |
+
"loss": 0.4859,
|
| 558 |
+
"step": 1720
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 2.33,
|
| 562 |
+
"learning_rate": 8.519511468573132e-05,
|
| 563 |
+
"loss": 0.4856,
|
| 564 |
+
"step": 1740
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 2.36,
|
| 568 |
+
"learning_rate": 8.489722966934764e-05,
|
| 569 |
+
"loss": 0.4725,
|
| 570 |
+
"step": 1760
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 2.39,
|
| 574 |
+
"learning_rate": 8.459934465296396e-05,
|
| 575 |
+
"loss": 0.4691,
|
| 576 |
+
"step": 1780
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 2.41,
|
| 580 |
+
"learning_rate": 8.430145963658028e-05,
|
| 581 |
+
"loss": 0.4485,
|
| 582 |
+
"step": 1800
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 2.44,
|
| 586 |
+
"learning_rate": 8.400357462019661e-05,
|
| 587 |
+
"loss": 0.484,
|
| 588 |
+
"step": 1820
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 2.47,
|
| 592 |
+
"learning_rate": 8.370568960381294e-05,
|
| 593 |
+
"loss": 0.484,
|
| 594 |
+
"step": 1840
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 2.49,
|
| 598 |
+
"learning_rate": 8.340780458742926e-05,
|
| 599 |
+
"loss": 0.477,
|
| 600 |
+
"step": 1860
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 2.52,
|
| 604 |
+
"learning_rate": 8.310991957104558e-05,
|
| 605 |
+
"loss": 0.5029,
|
| 606 |
+
"step": 1880
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 2.55,
|
| 610 |
+
"learning_rate": 8.28120345546619e-05,
|
| 611 |
+
"loss": 0.4957,
|
| 612 |
+
"step": 1900
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 2.57,
|
| 616 |
+
"learning_rate": 8.251414953827823e-05,
|
| 617 |
+
"loss": 0.4917,
|
| 618 |
+
"step": 1920
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 2.6,
|
| 622 |
+
"learning_rate": 8.221626452189454e-05,
|
| 623 |
+
"loss": 0.4959,
|
| 624 |
+
"step": 1940
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 2.63,
|
| 628 |
+
"learning_rate": 8.191837950551088e-05,
|
| 629 |
+
"loss": 0.4756,
|
| 630 |
+
"step": 1960
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 2.65,
|
| 634 |
+
"learning_rate": 8.16204944891272e-05,
|
| 635 |
+
"loss": 0.4749,
|
| 636 |
+
"step": 1980
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 2.68,
|
| 640 |
+
"learning_rate": 8.132260947274353e-05,
|
| 641 |
+
"loss": 0.4821,
|
| 642 |
+
"step": 2000
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 2.71,
|
| 646 |
+
"learning_rate": 8.102472445635984e-05,
|
| 647 |
+
"loss": 0.4749,
|
| 648 |
+
"step": 2020
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 2.73,
|
| 652 |
+
"learning_rate": 8.072683943997617e-05,
|
| 653 |
+
"loss": 0.461,
|
| 654 |
+
"step": 2040
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 2.76,
|
| 658 |
+
"learning_rate": 8.04289544235925e-05,
|
| 659 |
+
"loss": 0.4985,
|
| 660 |
+
"step": 2060
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 2.79,
|
| 664 |
+
"learning_rate": 8.013106940720883e-05,
|
| 665 |
+
"loss": 0.4886,
|
| 666 |
+
"step": 2080
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 2.82,
|
| 670 |
+
"learning_rate": 7.983318439082514e-05,
|
| 671 |
+
"loss": 0.476,
|
| 672 |
+
"step": 2100
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 2.84,
|
| 676 |
+
"learning_rate": 7.953529937444147e-05,
|
| 677 |
+
"loss": 0.4872,
|
| 678 |
+
"step": 2120
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 2.87,
|
| 682 |
+
"learning_rate": 7.92374143580578e-05,
|
| 683 |
+
"loss": 0.4823,
|
| 684 |
+
"step": 2140
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 2.9,
|
| 688 |
+
"learning_rate": 7.893952934167412e-05,
|
| 689 |
+
"loss": 0.4901,
|
| 690 |
+
"step": 2160
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.92,
|
| 694 |
+
"learning_rate": 7.864164432529044e-05,
|
| 695 |
+
"loss": 0.4831,
|
| 696 |
+
"step": 2180
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 2.95,
|
| 700 |
+
"learning_rate": 7.834375930890676e-05,
|
| 701 |
+
"loss": 0.4747,
|
| 702 |
+
"step": 2200
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 2.98,
|
| 706 |
+
"learning_rate": 7.804587429252309e-05,
|
| 707 |
+
"loss": 0.4642,
|
| 708 |
+
"step": 2220
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.0,
|
| 712 |
+
"eval_cer": 0.18081657294263642,
|
| 713 |
+
"eval_deletions": 4438,
|
| 714 |
+
"eval_hits": 55151,
|
| 715 |
+
"eval_insertions": 2260,
|
| 716 |
+
"eval_loss": 0.4211605191230774,
|
| 717 |
+
"eval_mer": 0.19801361098185202,
|
| 718 |
+
"eval_runtime": 929.7123,
|
| 719 |
+
"eval_samples_per_second": 6.749,
|
| 720 |
+
"eval_steps_per_second": 0.844,
|
| 721 |
+
"eval_substitutions": 6919,
|
| 722 |
+
"eval_wer": 0.20474228664220845,
|
| 723 |
+
"eval_wil": 0.28908227331086855,
|
| 724 |
+
"eval_wip": 0.7109177266891314,
|
| 725 |
+
"step": 2238
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 3.0,
|
| 729 |
+
"learning_rate": 7.774798927613941e-05,
|
| 730 |
+
"loss": 0.4761,
|
| 731 |
+
"step": 2240
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 3.03,
|
| 735 |
+
"learning_rate": 7.745010425975573e-05,
|
| 736 |
+
"loss": 0.4689,
|
| 737 |
+
"step": 2260
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 3.06,
|
| 741 |
+
"learning_rate": 7.715221924337206e-05,
|
| 742 |
+
"loss": 0.4719,
|
| 743 |
+
"step": 2280
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 3.08,
|
| 747 |
+
"learning_rate": 7.685433422698839e-05,
|
| 748 |
+
"loss": 0.4701,
|
| 749 |
+
"step": 2300
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.11,
|
| 753 |
+
"learning_rate": 7.655644921060471e-05,
|
| 754 |
+
"loss": 0.4208,
|
| 755 |
+
"step": 2320
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.14,
|
| 759 |
+
"learning_rate": 7.625856419422103e-05,
|
| 760 |
+
"loss": 0.4471,
|
| 761 |
+
"step": 2340
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 3.16,
|
| 765 |
+
"learning_rate": 7.596067917783735e-05,
|
| 766 |
+
"loss": 0.4465,
|
| 767 |
+
"step": 2360
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 3.19,
|
| 771 |
+
"learning_rate": 7.566279416145369e-05,
|
| 772 |
+
"loss": 0.4379,
|
| 773 |
+
"step": 2380
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.22,
|
| 777 |
+
"learning_rate": 7.536490914507001e-05,
|
| 778 |
+
"loss": 0.4697,
|
| 779 |
+
"step": 2400
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.24,
|
| 783 |
+
"learning_rate": 7.506702412868633e-05,
|
| 784 |
+
"loss": 0.4335,
|
| 785 |
+
"step": 2420
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.27,
|
| 789 |
+
"learning_rate": 7.476913911230265e-05,
|
| 790 |
+
"loss": 0.454,
|
| 791 |
+
"step": 2440
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.3,
|
| 795 |
+
"learning_rate": 7.447125409591897e-05,
|
| 796 |
+
"loss": 0.4643,
|
| 797 |
+
"step": 2460
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.32,
|
| 801 |
+
"learning_rate": 7.417336907953531e-05,
|
| 802 |
+
"loss": 0.4512,
|
| 803 |
+
"step": 2480
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.35,
|
| 807 |
+
"learning_rate": 7.387548406315163e-05,
|
| 808 |
+
"loss": 0.4301,
|
| 809 |
+
"step": 2500
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.38,
|
| 813 |
+
"learning_rate": 7.357759904676795e-05,
|
| 814 |
+
"loss": 0.4546,
|
| 815 |
+
"step": 2520
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.4,
|
| 819 |
+
"learning_rate": 7.327971403038427e-05,
|
| 820 |
+
"loss": 0.4316,
|
| 821 |
+
"step": 2540
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.43,
|
| 825 |
+
"learning_rate": 7.29818290140006e-05,
|
| 826 |
+
"loss": 0.4463,
|
| 827 |
+
"step": 2560
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.46,
|
| 831 |
+
"learning_rate": 7.268394399761693e-05,
|
| 832 |
+
"loss": 0.4456,
|
| 833 |
+
"step": 2580
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 3.49,
|
| 837 |
+
"learning_rate": 7.238605898123325e-05,
|
| 838 |
+
"loss": 0.4487,
|
| 839 |
+
"step": 2600
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.51,
|
| 843 |
+
"learning_rate": 7.208817396484957e-05,
|
| 844 |
+
"loss": 0.4441,
|
| 845 |
+
"step": 2620
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 3.54,
|
| 849 |
+
"learning_rate": 7.17902889484659e-05,
|
| 850 |
+
"loss": 0.4364,
|
| 851 |
+
"step": 2640
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 3.57,
|
| 855 |
+
"learning_rate": 7.149240393208221e-05,
|
| 856 |
+
"loss": 0.4475,
|
| 857 |
+
"step": 2660
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 3.59,
|
| 861 |
+
"learning_rate": 7.119451891569855e-05,
|
| 862 |
+
"loss": 0.4462,
|
| 863 |
+
"step": 2680
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 3.62,
|
| 867 |
+
"learning_rate": 7.089663389931487e-05,
|
| 868 |
+
"loss": 0.4608,
|
| 869 |
+
"step": 2700
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 3.65,
|
| 873 |
+
"learning_rate": 7.059874888293119e-05,
|
| 874 |
+
"loss": 0.4439,
|
| 875 |
+
"step": 2720
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 3.67,
|
| 879 |
+
"learning_rate": 7.030086386654751e-05,
|
| 880 |
+
"loss": 0.4347,
|
| 881 |
+
"step": 2740
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.7,
|
| 885 |
+
"learning_rate": 7.000297885016383e-05,
|
| 886 |
+
"loss": 0.4608,
|
| 887 |
+
"step": 2760
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 3.73,
|
| 891 |
+
"learning_rate": 6.970509383378017e-05,
|
| 892 |
+
"loss": 0.4532,
|
| 893 |
+
"step": 2780
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.75,
|
| 897 |
+
"learning_rate": 6.940720881739649e-05,
|
| 898 |
+
"loss": 0.4375,
|
| 899 |
+
"step": 2800
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.78,
|
| 903 |
+
"learning_rate": 6.910932380101281e-05,
|
| 904 |
+
"loss": 0.4493,
|
| 905 |
+
"step": 2820
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.81,
|
| 909 |
+
"learning_rate": 6.881143878462913e-05,
|
| 910 |
+
"loss": 0.4443,
|
| 911 |
+
"step": 2840
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.83,
|
| 915 |
+
"learning_rate": 6.851355376824547e-05,
|
| 916 |
+
"loss": 0.4402,
|
| 917 |
+
"step": 2860
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.86,
|
| 921 |
+
"learning_rate": 6.821566875186179e-05,
|
| 922 |
+
"loss": 0.4413,
|
| 923 |
+
"step": 2880
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.89,
|
| 927 |
+
"learning_rate": 6.791778373547811e-05,
|
| 928 |
+
"loss": 0.4388,
|
| 929 |
+
"step": 2900
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.91,
|
| 933 |
+
"learning_rate": 6.761989871909443e-05,
|
| 934 |
+
"loss": 0.4439,
|
| 935 |
+
"step": 2920
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.94,
|
| 939 |
+
"learning_rate": 6.732201370271076e-05,
|
| 940 |
+
"loss": 0.448,
|
| 941 |
+
"step": 2940
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.97,
|
| 945 |
+
"learning_rate": 6.702412868632708e-05,
|
| 946 |
+
"loss": 0.4476,
|
| 947 |
+
"step": 2960
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 3.99,
|
| 951 |
+
"learning_rate": 6.67262436699434e-05,
|
| 952 |
+
"loss": 0.4481,
|
| 953 |
+
"step": 2980
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 4.0,
|
| 957 |
+
"eval_cer": 0.1899560633671336,
|
| 958 |
+
"eval_deletions": 4153,
|
| 959 |
+
"eval_hits": 55367,
|
| 960 |
+
"eval_insertions": 2944,
|
| 961 |
+
"eval_loss": 0.4186785817146301,
|
| 962 |
+
"eval_mer": 0.20280193514945574,
|
| 963 |
+
"eval_runtime": 923.6349,
|
| 964 |
+
"eval_samples_per_second": 6.794,
|
| 965 |
+
"eval_steps_per_second": 0.85,
|
| 966 |
+
"eval_substitutions": 6988,
|
| 967 |
+
"eval_wer": 0.21177903410116075,
|
| 968 |
+
"eval_wil": 0.2941351331957437,
|
| 969 |
+
"eval_wip": 0.7058648668042563,
|
| 970 |
+
"step": 2984
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 4.02,
|
| 974 |
+
"learning_rate": 6.642835865355973e-05,
|
| 975 |
+
"loss": 0.4268,
|
| 976 |
+
"step": 3000
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 4.05,
|
| 980 |
+
"learning_rate": 6.613047363717605e-05,
|
| 981 |
+
"loss": 0.4343,
|
| 982 |
+
"step": 3020
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 4.08,
|
| 986 |
+
"learning_rate": 6.583258862079238e-05,
|
| 987 |
+
"loss": 0.4045,
|
| 988 |
+
"step": 3040
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 4.1,
|
| 992 |
+
"learning_rate": 6.55347036044087e-05,
|
| 993 |
+
"loss": 0.4079,
|
| 994 |
+
"step": 3060
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 4.13,
|
| 998 |
+
"learning_rate": 6.523681858802503e-05,
|
| 999 |
+
"loss": 0.4329,
|
| 1000 |
+
"step": 3080
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 4.16,
|
| 1004 |
+
"learning_rate": 6.493893357164135e-05,
|
| 1005 |
+
"loss": 0.425,
|
| 1006 |
+
"step": 3100
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 4.18,
|
| 1010 |
+
"learning_rate": 6.464104855525768e-05,
|
| 1011 |
+
"loss": 0.4117,
|
| 1012 |
+
"step": 3120
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 4.21,
|
| 1016 |
+
"learning_rate": 6.4343163538874e-05,
|
| 1017 |
+
"loss": 0.429,
|
| 1018 |
+
"step": 3140
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 4.24,
|
| 1022 |
+
"learning_rate": 6.404527852249032e-05,
|
| 1023 |
+
"loss": 0.4186,
|
| 1024 |
+
"step": 3160
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 4.26,
|
| 1028 |
+
"learning_rate": 6.374739350610664e-05,
|
| 1029 |
+
"loss": 0.4209,
|
| 1030 |
+
"step": 3180
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 4.29,
|
| 1034 |
+
"learning_rate": 6.344950848972298e-05,
|
| 1035 |
+
"loss": 0.4236,
|
| 1036 |
+
"step": 3200
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 4.32,
|
| 1040 |
+
"learning_rate": 6.31516234733393e-05,
|
| 1041 |
+
"loss": 0.4057,
|
| 1042 |
+
"step": 3220
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 4.34,
|
| 1046 |
+
"learning_rate": 6.285373845695561e-05,
|
| 1047 |
+
"loss": 0.4042,
|
| 1048 |
+
"step": 3240
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 4.37,
|
| 1052 |
+
"learning_rate": 6.255585344057194e-05,
|
| 1053 |
+
"loss": 0.4413,
|
| 1054 |
+
"step": 3260
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 4.4,
|
| 1058 |
+
"learning_rate": 6.225796842418826e-05,
|
| 1059 |
+
"loss": 0.4316,
|
| 1060 |
+
"step": 3280
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 4.42,
|
| 1064 |
+
"learning_rate": 6.19600834078046e-05,
|
| 1065 |
+
"loss": 0.4273,
|
| 1066 |
+
"step": 3300
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 4.45,
|
| 1070 |
+
"learning_rate": 6.16621983914209e-05,
|
| 1071 |
+
"loss": 0.4361,
|
| 1072 |
+
"step": 3320
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 4.48,
|
| 1076 |
+
"learning_rate": 6.136431337503724e-05,
|
| 1077 |
+
"loss": 0.4485,
|
| 1078 |
+
"step": 3340
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 4.5,
|
| 1082 |
+
"learning_rate": 6.106642835865356e-05,
|
| 1083 |
+
"loss": 0.4034,
|
| 1084 |
+
"step": 3360
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 4.53,
|
| 1088 |
+
"learning_rate": 6.076854334226989e-05,
|
| 1089 |
+
"loss": 0.4119,
|
| 1090 |
+
"step": 3380
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 4.56,
|
| 1094 |
+
"learning_rate": 6.047065832588621e-05,
|
| 1095 |
+
"loss": 0.3945,
|
| 1096 |
+
"step": 3400
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 4.58,
|
| 1100 |
+
"learning_rate": 6.017277330950254e-05,
|
| 1101 |
+
"loss": 0.4258,
|
| 1102 |
+
"step": 3420
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 4.61,
|
| 1106 |
+
"learning_rate": 5.987488829311886e-05,
|
| 1107 |
+
"loss": 0.4311,
|
| 1108 |
+
"step": 3440
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 4.64,
|
| 1112 |
+
"learning_rate": 5.957700327673519e-05,
|
| 1113 |
+
"loss": 0.4134,
|
| 1114 |
+
"step": 3460
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 4.66,
|
| 1118 |
+
"learning_rate": 5.927911826035151e-05,
|
| 1119 |
+
"loss": 0.4262,
|
| 1120 |
+
"step": 3480
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 4.69,
|
| 1124 |
+
"learning_rate": 5.8981233243967824e-05,
|
| 1125 |
+
"loss": 0.4208,
|
| 1126 |
+
"step": 3500
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 4.72,
|
| 1130 |
+
"learning_rate": 5.868334822758416e-05,
|
| 1131 |
+
"loss": 0.4111,
|
| 1132 |
+
"step": 3520
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 4.75,
|
| 1136 |
+
"learning_rate": 5.838546321120047e-05,
|
| 1137 |
+
"loss": 0.4273,
|
| 1138 |
+
"step": 3540
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 4.77,
|
| 1142 |
+
"learning_rate": 5.808757819481681e-05,
|
| 1143 |
+
"loss": 0.3966,
|
| 1144 |
+
"step": 3560
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 4.8,
|
| 1148 |
+
"learning_rate": 5.778969317843312e-05,
|
| 1149 |
+
"loss": 0.4442,
|
| 1150 |
+
"step": 3580
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 4.83,
|
| 1154 |
+
"learning_rate": 5.749180816204946e-05,
|
| 1155 |
+
"loss": 0.4085,
|
| 1156 |
+
"step": 3600
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 4.85,
|
| 1160 |
+
"learning_rate": 5.719392314566577e-05,
|
| 1161 |
+
"loss": 0.4319,
|
| 1162 |
+
"step": 3620
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 4.88,
|
| 1166 |
+
"learning_rate": 5.6896038129282106e-05,
|
| 1167 |
+
"loss": 0.4201,
|
| 1168 |
+
"step": 3640
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 4.91,
|
| 1172 |
+
"learning_rate": 5.659815311289842e-05,
|
| 1173 |
+
"loss": 0.4209,
|
| 1174 |
+
"step": 3660
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 4.93,
|
| 1178 |
+
"learning_rate": 5.6300268096514755e-05,
|
| 1179 |
+
"loss": 0.4329,
|
| 1180 |
+
"step": 3680
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 4.96,
|
| 1184 |
+
"learning_rate": 5.600238308013107e-05,
|
| 1185 |
+
"loss": 0.4164,
|
| 1186 |
+
"step": 3700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 4.99,
|
| 1190 |
+
"learning_rate": 5.5704498063747404e-05,
|
| 1191 |
+
"loss": 0.4224,
|
| 1192 |
+
"step": 3720
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 5.0,
|
| 1196 |
+
"eval_cer": 0.180956320502644,
|
| 1197 |
+
"eval_deletions": 4305,
|
| 1198 |
+
"eval_hits": 55337,
|
| 1199 |
+
"eval_insertions": 2397,
|
| 1200 |
+
"eval_loss": 0.4172378480434418,
|
| 1201 |
+
"eval_mer": 0.19690878746099702,
|
| 1202 |
+
"eval_runtime": 915.2753,
|
| 1203 |
+
"eval_samples_per_second": 6.856,
|
| 1204 |
+
"eval_steps_per_second": 0.858,
|
| 1205 |
+
"eval_substitutions": 6866,
|
| 1206 |
+
"eval_wer": 0.20400553316894207,
|
| 1207 |
+
"eval_wil": 0.28727036701839537,
|
| 1208 |
+
"eval_wip": 0.7127296329816046,
|
| 1209 |
+
"step": 3730
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 5.01,
|
| 1213 |
+
"learning_rate": 5.540661304736372e-05,
|
| 1214 |
+
"loss": 0.3878,
|
| 1215 |
+
"step": 3740
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 5.04,
|
| 1219 |
+
"learning_rate": 5.510872803098004e-05,
|
| 1220 |
+
"loss": 0.4118,
|
| 1221 |
+
"step": 3760
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 5.07,
|
| 1225 |
+
"learning_rate": 5.481084301459637e-05,
|
| 1226 |
+
"loss": 0.3936,
|
| 1227 |
+
"step": 3780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 5.09,
|
| 1231 |
+
"learning_rate": 5.451295799821269e-05,
|
| 1232 |
+
"loss": 0.4022,
|
| 1233 |
+
"step": 3800
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 5.12,
|
| 1237 |
+
"learning_rate": 5.421507298182902e-05,
|
| 1238 |
+
"loss": 0.3973,
|
| 1239 |
+
"step": 3820
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.15,
|
| 1243 |
+
"learning_rate": 5.391718796544534e-05,
|
| 1244 |
+
"loss": 0.4062,
|
| 1245 |
+
"step": 3840
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 5.17,
|
| 1249 |
+
"learning_rate": 5.3619302949061666e-05,
|
| 1250 |
+
"loss": 0.4071,
|
| 1251 |
+
"step": 3860
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 5.2,
|
| 1255 |
+
"learning_rate": 5.332141793267799e-05,
|
| 1256 |
+
"loss": 0.4052,
|
| 1257 |
+
"step": 3880
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 5.23,
|
| 1261 |
+
"learning_rate": 5.3023532916294315e-05,
|
| 1262 |
+
"loss": 0.3924,
|
| 1263 |
+
"step": 3900
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 5.25,
|
| 1267 |
+
"learning_rate": 5.2725647899910637e-05,
|
| 1268 |
+
"loss": 0.3916,
|
| 1269 |
+
"step": 3920
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 5.28,
|
| 1273 |
+
"learning_rate": 5.2427762883526964e-05,
|
| 1274 |
+
"loss": 0.4163,
|
| 1275 |
+
"step": 3940
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 5.31,
|
| 1279 |
+
"learning_rate": 5.2129877867143286e-05,
|
| 1280 |
+
"loss": 0.4034,
|
| 1281 |
+
"step": 3960
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 5.34,
|
| 1285 |
+
"learning_rate": 5.1831992850759614e-05,
|
| 1286 |
+
"loss": 0.4115,
|
| 1287 |
+
"step": 3980
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 5.36,
|
| 1291 |
+
"learning_rate": 5.1534107834375935e-05,
|
| 1292 |
+
"loss": 0.4087,
|
| 1293 |
+
"step": 4000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 5.39,
|
| 1297 |
+
"learning_rate": 5.123622281799225e-05,
|
| 1298 |
+
"loss": 0.3804,
|
| 1299 |
+
"step": 4020
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 5.42,
|
| 1303 |
+
"learning_rate": 5.0938337801608584e-05,
|
| 1304 |
+
"loss": 0.4055,
|
| 1305 |
+
"step": 4040
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 5.44,
|
| 1309 |
+
"learning_rate": 5.06404527852249e-05,
|
| 1310 |
+
"loss": 0.3821,
|
| 1311 |
+
"step": 4060
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 5.47,
|
| 1315 |
+
"learning_rate": 5.034256776884123e-05,
|
| 1316 |
+
"loss": 0.3997,
|
| 1317 |
+
"step": 4080
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 5.5,
|
| 1321 |
+
"learning_rate": 5.004468275245755e-05,
|
| 1322 |
+
"loss": 0.3917,
|
| 1323 |
+
"step": 4100
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 5.52,
|
| 1327 |
+
"learning_rate": 4.974679773607388e-05,
|
| 1328 |
+
"loss": 0.4167,
|
| 1329 |
+
"step": 4120
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 5.55,
|
| 1333 |
+
"learning_rate": 4.94489127196902e-05,
|
| 1334 |
+
"loss": 0.376,
|
| 1335 |
+
"step": 4140
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 5.58,
|
| 1339 |
+
"learning_rate": 4.9151027703306525e-05,
|
| 1340 |
+
"loss": 0.4055,
|
| 1341 |
+
"step": 4160
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 5.6,
|
| 1345 |
+
"learning_rate": 4.8853142686922846e-05,
|
| 1346 |
+
"loss": 0.4081,
|
| 1347 |
+
"step": 4180
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 5.63,
|
| 1351 |
+
"learning_rate": 4.8555257670539174e-05,
|
| 1352 |
+
"loss": 0.4118,
|
| 1353 |
+
"step": 4200
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 5.66,
|
| 1357 |
+
"learning_rate": 4.8257372654155495e-05,
|
| 1358 |
+
"loss": 0.4007,
|
| 1359 |
+
"step": 4220
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 5.68,
|
| 1363 |
+
"learning_rate": 4.795948763777182e-05,
|
| 1364 |
+
"loss": 0.4243,
|
| 1365 |
+
"step": 4240
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 5.71,
|
| 1369 |
+
"learning_rate": 4.7661602621388144e-05,
|
| 1370 |
+
"loss": 0.4045,
|
| 1371 |
+
"step": 4260
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 5.74,
|
| 1375 |
+
"learning_rate": 4.736371760500447e-05,
|
| 1376 |
+
"loss": 0.3798,
|
| 1377 |
+
"step": 4280
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 5.76,
|
| 1381 |
+
"learning_rate": 4.706583258862079e-05,
|
| 1382 |
+
"loss": 0.3839,
|
| 1383 |
+
"step": 4300
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 5.79,
|
| 1387 |
+
"learning_rate": 4.676794757223712e-05,
|
| 1388 |
+
"loss": 0.3866,
|
| 1389 |
+
"step": 4320
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 5.82,
|
| 1393 |
+
"learning_rate": 4.647006255585344e-05,
|
| 1394 |
+
"loss": 0.3841,
|
| 1395 |
+
"step": 4340
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 5.84,
|
| 1399 |
+
"learning_rate": 4.617217753946977e-05,
|
| 1400 |
+
"loss": 0.4069,
|
| 1401 |
+
"step": 4360
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 5.87,
|
| 1405 |
+
"learning_rate": 4.587429252308609e-05,
|
| 1406 |
+
"loss": 0.4176,
|
| 1407 |
+
"step": 4380
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 5.9,
|
| 1411 |
+
"learning_rate": 4.557640750670242e-05,
|
| 1412 |
+
"loss": 0.4169,
|
| 1413 |
+
"step": 4400
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 5.92,
|
| 1417 |
+
"learning_rate": 4.5278522490318734e-05,
|
| 1418 |
+
"loss": 0.391,
|
| 1419 |
+
"step": 4420
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 5.95,
|
| 1423 |
+
"learning_rate": 4.498063747393506e-05,
|
| 1424 |
+
"loss": 0.3957,
|
| 1425 |
+
"step": 4440
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 5.98,
|
| 1429 |
+
"learning_rate": 4.468275245755138e-05,
|
| 1430 |
+
"loss": 0.3927,
|
| 1431 |
+
"step": 4460
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 6.0,
|
| 1435 |
+
"eval_cer": 0.18605990139412165,
|
| 1436 |
+
"eval_deletions": 3953,
|
| 1437 |
+
"eval_hits": 55575,
|
| 1438 |
+
"eval_insertions": 2665,
|
| 1439 |
+
"eval_loss": 0.4182780683040619,
|
| 1440 |
+
"eval_mer": 0.1965795903025747,
|
| 1441 |
+
"eval_runtime": 936.5757,
|
| 1442 |
+
"eval_samples_per_second": 6.7,
|
| 1443 |
+
"eval_steps_per_second": 0.838,
|
| 1444 |
+
"eval_substitutions": 6980,
|
| 1445 |
+
"eval_wer": 0.20445660672400312,
|
| 1446 |
+
"eval_wil": 0.28796021536777305,
|
| 1447 |
+
"eval_wip": 0.712039784632227,
|
| 1448 |
+
"step": 4476
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 6.01,
|
| 1452 |
+
"learning_rate": 4.438486744116771e-05,
|
| 1453 |
+
"loss": 0.3821,
|
| 1454 |
+
"step": 4480
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 6.03,
|
| 1458 |
+
"learning_rate": 4.408698242478403e-05,
|
| 1459 |
+
"loss": 0.3741,
|
| 1460 |
+
"step": 4500
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 6.06,
|
| 1464 |
+
"learning_rate": 4.378909740840036e-05,
|
| 1465 |
+
"loss": 0.3642,
|
| 1466 |
+
"step": 4520
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 6.09,
|
| 1470 |
+
"learning_rate": 4.349121239201668e-05,
|
| 1471 |
+
"loss": 0.3825,
|
| 1472 |
+
"step": 4540
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 6.11,
|
| 1476 |
+
"learning_rate": 4.319332737563301e-05,
|
| 1477 |
+
"loss": 0.3787,
|
| 1478 |
+
"step": 4560
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 6.14,
|
| 1482 |
+
"learning_rate": 4.289544235924933e-05,
|
| 1483 |
+
"loss": 0.3808,
|
| 1484 |
+
"step": 4580
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 6.17,
|
| 1488 |
+
"learning_rate": 4.259755734286566e-05,
|
| 1489 |
+
"loss": 0.3632,
|
| 1490 |
+
"step": 4600
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 6.19,
|
| 1494 |
+
"learning_rate": 4.229967232648198e-05,
|
| 1495 |
+
"loss": 0.3658,
|
| 1496 |
+
"step": 4620
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 6.22,
|
| 1500 |
+
"learning_rate": 4.200178731009831e-05,
|
| 1501 |
+
"loss": 0.3821,
|
| 1502 |
+
"step": 4640
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 6.25,
|
| 1506 |
+
"learning_rate": 4.170390229371463e-05,
|
| 1507 |
+
"loss": 0.3982,
|
| 1508 |
+
"step": 4660
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 6.27,
|
| 1512 |
+
"learning_rate": 4.140601727733095e-05,
|
| 1513 |
+
"loss": 0.3858,
|
| 1514 |
+
"step": 4680
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 6.3,
|
| 1518 |
+
"learning_rate": 4.110813226094727e-05,
|
| 1519 |
+
"loss": 0.3853,
|
| 1520 |
+
"step": 4700
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 6.33,
|
| 1524 |
+
"learning_rate": 4.08102472445636e-05,
|
| 1525 |
+
"loss": 0.3684,
|
| 1526 |
+
"step": 4720
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 6.35,
|
| 1530 |
+
"learning_rate": 4.051236222817992e-05,
|
| 1531 |
+
"loss": 0.3623,
|
| 1532 |
+
"step": 4740
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 6.38,
|
| 1536 |
+
"learning_rate": 4.021447721179625e-05,
|
| 1537 |
+
"loss": 0.38,
|
| 1538 |
+
"step": 4760
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 6.41,
|
| 1542 |
+
"learning_rate": 3.991659219541257e-05,
|
| 1543 |
+
"loss": 0.4107,
|
| 1544 |
+
"step": 4780
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 6.43,
|
| 1548 |
+
"learning_rate": 3.96187071790289e-05,
|
| 1549 |
+
"loss": 0.3882,
|
| 1550 |
+
"step": 4800
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 6.46,
|
| 1554 |
+
"learning_rate": 3.932082216264522e-05,
|
| 1555 |
+
"loss": 0.3907,
|
| 1556 |
+
"step": 4820
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 6.49,
|
| 1560 |
+
"learning_rate": 3.9022937146261546e-05,
|
| 1561 |
+
"loss": 0.3749,
|
| 1562 |
+
"step": 4840
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 6.51,
|
| 1566 |
+
"learning_rate": 3.872505212987787e-05,
|
| 1567 |
+
"loss": 0.3849,
|
| 1568 |
+
"step": 4860
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 6.54,
|
| 1572 |
+
"learning_rate": 3.8427167113494195e-05,
|
| 1573 |
+
"loss": 0.3726,
|
| 1574 |
+
"step": 4880
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 6.57,
|
| 1578 |
+
"learning_rate": 3.8129282097110517e-05,
|
| 1579 |
+
"loss": 0.3939,
|
| 1580 |
+
"step": 4900
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 6.6,
|
| 1584 |
+
"learning_rate": 3.7831397080726844e-05,
|
| 1585 |
+
"loss": 0.3641,
|
| 1586 |
+
"step": 4920
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 6.62,
|
| 1590 |
+
"learning_rate": 3.7533512064343166e-05,
|
| 1591 |
+
"loss": 0.3786,
|
| 1592 |
+
"step": 4940
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.65,
|
| 1596 |
+
"learning_rate": 3.723562704795949e-05,
|
| 1597 |
+
"loss": 0.3941,
|
| 1598 |
+
"step": 4960
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 6.68,
|
| 1602 |
+
"learning_rate": 3.6937742031575815e-05,
|
| 1603 |
+
"loss": 0.3876,
|
| 1604 |
+
"step": 4980
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 6.7,
|
| 1608 |
+
"learning_rate": 3.6639857015192136e-05,
|
| 1609 |
+
"loss": 0.3878,
|
| 1610 |
+
"step": 5000
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 6.73,
|
| 1614 |
+
"learning_rate": 3.6341971998808464e-05,
|
| 1615 |
+
"loss": 0.3897,
|
| 1616 |
+
"step": 5020
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 6.76,
|
| 1620 |
+
"learning_rate": 3.6044086982424785e-05,
|
| 1621 |
+
"loss": 0.3974,
|
| 1622 |
+
"step": 5040
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 6.78,
|
| 1626 |
+
"learning_rate": 3.5746201966041106e-05,
|
| 1627 |
+
"loss": 0.3781,
|
| 1628 |
+
"step": 5060
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 6.81,
|
| 1632 |
+
"learning_rate": 3.5448316949657434e-05,
|
| 1633 |
+
"loss": 0.39,
|
| 1634 |
+
"step": 5080
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.84,
|
| 1638 |
+
"learning_rate": 3.5150431933273755e-05,
|
| 1639 |
+
"loss": 0.4095,
|
| 1640 |
+
"step": 5100
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 6.86,
|
| 1644 |
+
"learning_rate": 3.485254691689008e-05,
|
| 1645 |
+
"loss": 0.3831,
|
| 1646 |
+
"step": 5120
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 6.89,
|
| 1650 |
+
"learning_rate": 3.4554661900506405e-05,
|
| 1651 |
+
"loss": 0.4011,
|
| 1652 |
+
"step": 5140
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 6.92,
|
| 1656 |
+
"learning_rate": 3.425677688412273e-05,
|
| 1657 |
+
"loss": 0.3948,
|
| 1658 |
+
"step": 5160
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 6.94,
|
| 1662 |
+
"learning_rate": 3.3958891867739054e-05,
|
| 1663 |
+
"loss": 0.3937,
|
| 1664 |
+
"step": 5180
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 6.97,
|
| 1668 |
+
"learning_rate": 3.366100685135538e-05,
|
| 1669 |
+
"loss": 0.3895,
|
| 1670 |
+
"step": 5200
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 7.0,
|
| 1674 |
+
"learning_rate": 3.33631218349717e-05,
|
| 1675 |
+
"loss": 0.3834,
|
| 1676 |
+
"step": 5220
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 7.0,
|
| 1680 |
+
"eval_cer": 0.18220286873791183,
|
| 1681 |
+
"eval_deletions": 4025,
|
| 1682 |
+
"eval_hits": 55571,
|
| 1683 |
+
"eval_insertions": 2507,
|
| 1684 |
+
"eval_loss": 0.41830888390541077,
|
| 1685 |
+
"eval_mer": 0.19479823226834747,
|
| 1686 |
+
"eval_runtime": 915.8341,
|
| 1687 |
+
"eval_samples_per_second": 6.852,
|
| 1688 |
+
"eval_steps_per_second": 0.857,
|
| 1689 |
+
"eval_substitutions": 6912,
|
| 1690 |
+
"eval_wer": 0.20214109580802309,
|
| 1691 |
+
"eval_wil": 0.28554315919063866,
|
| 1692 |
+
"eval_wip": 0.7144568408093613,
|
| 1693 |
+
"step": 5222
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 7.02,
|
| 1697 |
+
"learning_rate": 3.3065236818588024e-05,
|
| 1698 |
+
"loss": 0.3722,
|
| 1699 |
+
"step": 5240
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 7.05,
|
| 1703 |
+
"learning_rate": 3.276735180220435e-05,
|
| 1704 |
+
"loss": 0.3653,
|
| 1705 |
+
"step": 5260
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 7.08,
|
| 1709 |
+
"learning_rate": 3.246946678582067e-05,
|
| 1710 |
+
"loss": 0.3678,
|
| 1711 |
+
"step": 5280
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 7.1,
|
| 1715 |
+
"learning_rate": 3.2171581769437e-05,
|
| 1716 |
+
"loss": 0.3713,
|
| 1717 |
+
"step": 5300
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 7.13,
|
| 1721 |
+
"learning_rate": 3.187369675305332e-05,
|
| 1722 |
+
"loss": 0.3701,
|
| 1723 |
+
"step": 5320
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 7.16,
|
| 1727 |
+
"learning_rate": 3.157581173666965e-05,
|
| 1728 |
+
"loss": 0.3784,
|
| 1729 |
+
"step": 5340
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 7.18,
|
| 1733 |
+
"learning_rate": 3.127792672028597e-05,
|
| 1734 |
+
"loss": 0.3671,
|
| 1735 |
+
"step": 5360
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 7.21,
|
| 1739 |
+
"learning_rate": 3.09800417039023e-05,
|
| 1740 |
+
"loss": 0.3699,
|
| 1741 |
+
"step": 5380
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 7.24,
|
| 1745 |
+
"learning_rate": 3.068215668751862e-05,
|
| 1746 |
+
"loss": 0.3834,
|
| 1747 |
+
"step": 5400
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 7.27,
|
| 1751 |
+
"learning_rate": 3.0384271671134945e-05,
|
| 1752 |
+
"loss": 0.3745,
|
| 1753 |
+
"step": 5420
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 7.29,
|
| 1757 |
+
"learning_rate": 3.008638665475127e-05,
|
| 1758 |
+
"loss": 0.3831,
|
| 1759 |
+
"step": 5440
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 7.32,
|
| 1763 |
+
"learning_rate": 2.9788501638367594e-05,
|
| 1764 |
+
"loss": 0.3732,
|
| 1765 |
+
"step": 5460
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 7.35,
|
| 1769 |
+
"learning_rate": 2.9490616621983912e-05,
|
| 1770 |
+
"loss": 0.3685,
|
| 1771 |
+
"step": 5480
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 7.37,
|
| 1775 |
+
"learning_rate": 2.9192731605600237e-05,
|
| 1776 |
+
"loss": 0.3707,
|
| 1777 |
+
"step": 5500
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 7.4,
|
| 1781 |
+
"learning_rate": 2.889484658921656e-05,
|
| 1782 |
+
"loss": 0.3702,
|
| 1783 |
+
"step": 5520
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 7.43,
|
| 1787 |
+
"learning_rate": 2.8596961572832886e-05,
|
| 1788 |
+
"loss": 0.3707,
|
| 1789 |
+
"step": 5540
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 7.45,
|
| 1793 |
+
"learning_rate": 2.829907655644921e-05,
|
| 1794 |
+
"loss": 0.3767,
|
| 1795 |
+
"step": 5560
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 7.48,
|
| 1799 |
+
"learning_rate": 2.8001191540065535e-05,
|
| 1800 |
+
"loss": 0.345,
|
| 1801 |
+
"step": 5580
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 7.51,
|
| 1805 |
+
"learning_rate": 2.770330652368186e-05,
|
| 1806 |
+
"loss": 0.3677,
|
| 1807 |
+
"step": 5600
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 7.53,
|
| 1811 |
+
"learning_rate": 2.7405421507298184e-05,
|
| 1812 |
+
"loss": 0.371,
|
| 1813 |
+
"step": 5620
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 7.56,
|
| 1817 |
+
"learning_rate": 2.710753649091451e-05,
|
| 1818 |
+
"loss": 0.3642,
|
| 1819 |
+
"step": 5640
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 7.59,
|
| 1823 |
+
"learning_rate": 2.6809651474530833e-05,
|
| 1824 |
+
"loss": 0.3785,
|
| 1825 |
+
"step": 5660
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 7.61,
|
| 1829 |
+
"learning_rate": 2.6511766458147158e-05,
|
| 1830 |
+
"loss": 0.3696,
|
| 1831 |
+
"step": 5680
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 7.64,
|
| 1835 |
+
"learning_rate": 2.6213881441763482e-05,
|
| 1836 |
+
"loss": 0.3801,
|
| 1837 |
+
"step": 5700
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 7.67,
|
| 1841 |
+
"learning_rate": 2.5915996425379807e-05,
|
| 1842 |
+
"loss": 0.3793,
|
| 1843 |
+
"step": 5720
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 7.69,
|
| 1847 |
+
"learning_rate": 2.5618111408996125e-05,
|
| 1848 |
+
"loss": 0.3621,
|
| 1849 |
+
"step": 5740
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 7.72,
|
| 1853 |
+
"learning_rate": 2.532022639261245e-05,
|
| 1854 |
+
"loss": 0.3622,
|
| 1855 |
+
"step": 5760
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 7.75,
|
| 1859 |
+
"learning_rate": 2.5022341376228774e-05,
|
| 1860 |
+
"loss": 0.3758,
|
| 1861 |
+
"step": 5780
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 7.77,
|
| 1865 |
+
"learning_rate": 2.47244563598451e-05,
|
| 1866 |
+
"loss": 0.3677,
|
| 1867 |
+
"step": 5800
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 7.8,
|
| 1871 |
+
"learning_rate": 2.4426571343461423e-05,
|
| 1872 |
+
"loss": 0.3709,
|
| 1873 |
+
"step": 5820
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 7.83,
|
| 1877 |
+
"learning_rate": 2.4128686327077747e-05,
|
| 1878 |
+
"loss": 0.3702,
|
| 1879 |
+
"step": 5840
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 7.86,
|
| 1883 |
+
"learning_rate": 2.3830801310694072e-05,
|
| 1884 |
+
"loss": 0.3565,
|
| 1885 |
+
"step": 5860
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 7.88,
|
| 1889 |
+
"learning_rate": 2.3532916294310397e-05,
|
| 1890 |
+
"loss": 0.3621,
|
| 1891 |
+
"step": 5880
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 7.91,
|
| 1895 |
+
"learning_rate": 2.323503127792672e-05,
|
| 1896 |
+
"loss": 0.3875,
|
| 1897 |
+
"step": 5900
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 7.94,
|
| 1901 |
+
"learning_rate": 2.2937146261543046e-05,
|
| 1902 |
+
"loss": 0.3771,
|
| 1903 |
+
"step": 5920
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 7.96,
|
| 1907 |
+
"learning_rate": 2.2639261245159367e-05,
|
| 1908 |
+
"loss": 0.4039,
|
| 1909 |
+
"step": 5940
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 7.99,
|
| 1913 |
+
"learning_rate": 2.234137622877569e-05,
|
| 1914 |
+
"loss": 0.3484,
|
| 1915 |
+
"step": 5960
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 8.0,
|
| 1919 |
+
"eval_cer": 0.1825718022963319,
|
| 1920 |
+
"eval_deletions": 4028,
|
| 1921 |
+
"eval_hits": 55559,
|
| 1922 |
+
"eval_insertions": 2490,
|
| 1923 |
+
"eval_loss": 0.41909465193748474,
|
| 1924 |
+
"eval_mer": 0.19477376155830603,
|
| 1925 |
+
"eval_runtime": 894.6454,
|
| 1926 |
+
"eval_samples_per_second": 7.014,
|
| 1927 |
+
"eval_steps_per_second": 0.877,
|
| 1928 |
+
"eval_substitutions": 6921,
|
| 1929 |
+
"eval_wer": 0.2020659168821796,
|
| 1930 |
+
"eval_wil": 0.2856318460438537,
|
| 1931 |
+
"eval_wip": 0.7143681539561463,
|
| 1932 |
+
"step": 5968
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 8.02,
|
| 1936 |
+
"learning_rate": 2.2043491212392016e-05,
|
| 1937 |
+
"loss": 0.3722,
|
| 1938 |
+
"step": 5980
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 8.04,
|
| 1942 |
+
"learning_rate": 2.174560619600834e-05,
|
| 1943 |
+
"loss": 0.3512,
|
| 1944 |
+
"step": 6000
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 8.07,
|
| 1948 |
+
"learning_rate": 2.1447721179624665e-05,
|
| 1949 |
+
"loss": 0.377,
|
| 1950 |
+
"step": 6020
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 8.1,
|
| 1954 |
+
"learning_rate": 2.114983616324099e-05,
|
| 1955 |
+
"loss": 0.36,
|
| 1956 |
+
"step": 6040
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 8.12,
|
| 1960 |
+
"learning_rate": 2.0851951146857314e-05,
|
| 1961 |
+
"loss": 0.383,
|
| 1962 |
+
"step": 6060
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 8.15,
|
| 1966 |
+
"learning_rate": 2.0554066130473635e-05,
|
| 1967 |
+
"loss": 0.3542,
|
| 1968 |
+
"step": 6080
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 8.18,
|
| 1972 |
+
"learning_rate": 2.025618111408996e-05,
|
| 1973 |
+
"loss": 0.3594,
|
| 1974 |
+
"step": 6100
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 8.2,
|
| 1978 |
+
"learning_rate": 1.9958296097706285e-05,
|
| 1979 |
+
"loss": 0.3436,
|
| 1980 |
+
"step": 6120
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 8.23,
|
| 1984 |
+
"learning_rate": 1.966041108132261e-05,
|
| 1985 |
+
"loss": 0.3532,
|
| 1986 |
+
"step": 6140
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 8.26,
|
| 1990 |
+
"learning_rate": 1.9362526064938934e-05,
|
| 1991 |
+
"loss": 0.3565,
|
| 1992 |
+
"step": 6160
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 8.28,
|
| 1996 |
+
"learning_rate": 1.9064641048555258e-05,
|
| 1997 |
+
"loss": 0.3601,
|
| 1998 |
+
"step": 6180
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 8.31,
|
| 2002 |
+
"learning_rate": 1.8766756032171583e-05,
|
| 2003 |
+
"loss": 0.3498,
|
| 2004 |
+
"step": 6200
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 8.34,
|
| 2008 |
+
"learning_rate": 1.8468871015787907e-05,
|
| 2009 |
+
"loss": 0.3622,
|
| 2010 |
+
"step": 6220
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 8.36,
|
| 2014 |
+
"learning_rate": 1.8170985999404232e-05,
|
| 2015 |
+
"loss": 0.3594,
|
| 2016 |
+
"step": 6240
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 8.39,
|
| 2020 |
+
"learning_rate": 1.7873100983020553e-05,
|
| 2021 |
+
"loss": 0.3771,
|
| 2022 |
+
"step": 6260
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 8.42,
|
| 2026 |
+
"learning_rate": 1.7575215966636878e-05,
|
| 2027 |
+
"loss": 0.3673,
|
| 2028 |
+
"step": 6280
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 8.45,
|
| 2032 |
+
"learning_rate": 1.7277330950253202e-05,
|
| 2033 |
+
"loss": 0.36,
|
| 2034 |
+
"step": 6300
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 8.47,
|
| 2038 |
+
"learning_rate": 1.6979445933869527e-05,
|
| 2039 |
+
"loss": 0.3746,
|
| 2040 |
+
"step": 6320
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 8.5,
|
| 2044 |
+
"learning_rate": 1.668156091748585e-05,
|
| 2045 |
+
"loss": 0.3643,
|
| 2046 |
+
"step": 6340
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 8.53,
|
| 2050 |
+
"learning_rate": 1.6383675901102176e-05,
|
| 2051 |
+
"loss": 0.3572,
|
| 2052 |
+
"step": 6360
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 8.55,
|
| 2056 |
+
"learning_rate": 1.60857908847185e-05,
|
| 2057 |
+
"loss": 0.3704,
|
| 2058 |
+
"step": 6380
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 8.58,
|
| 2062 |
+
"learning_rate": 1.5787905868334825e-05,
|
| 2063 |
+
"loss": 0.3706,
|
| 2064 |
+
"step": 6400
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 8.61,
|
| 2068 |
+
"learning_rate": 1.549002085195115e-05,
|
| 2069 |
+
"loss": 0.367,
|
| 2070 |
+
"step": 6420
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 8.63,
|
| 2074 |
+
"learning_rate": 1.5192135835567473e-05,
|
| 2075 |
+
"loss": 0.3408,
|
| 2076 |
+
"step": 6440
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 8.66,
|
| 2080 |
+
"learning_rate": 1.4894250819183797e-05,
|
| 2081 |
+
"loss": 0.36,
|
| 2082 |
+
"step": 6460
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 8.69,
|
| 2086 |
+
"learning_rate": 1.4596365802800118e-05,
|
| 2087 |
+
"loss": 0.3711,
|
| 2088 |
+
"step": 6480
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 8.71,
|
| 2092 |
+
"learning_rate": 1.4298480786416443e-05,
|
| 2093 |
+
"loss": 0.3633,
|
| 2094 |
+
"step": 6500
|
| 2095 |
+
}
|
| 2096 |
+
],
|
| 2097 |
+
"max_steps": 7460,
|
| 2098 |
+
"num_train_epochs": 10,
|
| 2099 |
+
"total_flos": 3.15961607503872e+16,
|
| 2100 |
+
"trial_name": null,
|
| 2101 |
+
"trial_params": null
|
| 2102 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78b9781f5cd654242fcd55045aeec909e63b4d464e66623fd75e1b93e912eec1
|
| 3 |
+
size 3055
|