Hugging Face's logo

longcld
/
t5-small-e2e-qa

Text2Text Generation
Transformers
PyTorch
t5
text-generation-inference
Inference Endpoints
Model card Files Community
1
longcld commited on
Commit
14fe5bb
•
1 Parent(s): aba246b

loss 3.0

Browse files
Files changed (11) hide show
  1. added_tokens.json +1 -0
  2. config.json +58 -0
  3. optimizer.pt +3 -0
  4. pytorch_model.bin +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +1 -0
  8. spiece.model +3 -0
  9. tokenizer_config.json +1 -0
  10. trainer_state.json +3886 -0
  11. training_args.bin +3 -0
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<sep>": 250100, "<hl>": 250101}
config.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "t5-small",
3
+ "architectures": [
4
+ "T5ForConditionalGeneration"
5
+ ],
6
+ "d_ff": 2048,
7
+ "d_kv": 64,
8
+ "d_model": 512,
9
+ "decoder_start_token_id": 0,
10
+ "dropout_rate": 0.1,
11
+ "eos_token_id": 1,
12
+ "feed_forward_proj": "relu",
13
+ "gradient_checkpointing": false,
14
+ "initializer_factor": 1.0,
15
+ "is_encoder_decoder": true,
16
+ "layer_norm_epsilon": 1e-06,
17
+ "model_type": "t5",
18
+ "n_positions": 512,
19
+ "num_decoder_layers": 6,
20
+ "num_heads": 8,
21
+ "num_layers": 6,
22
+ "output_past": true,
23
+ "pad_token_id": 0,
24
+ "relative_attention_num_buckets": 32,
25
+ "task_specific_params": {
26
+ "summarization": {
27
+ "early_stopping": true,
28
+ "length_penalty": 2.0,
29
+ "max_length": 200,
30
+ "min_length": 30,
31
+ "no_repeat_ngram_size": 3,
32
+ "num_beams": 4,
33
+ "prefix": "summarize: "
34
+ },
35
+ "translation_en_to_de": {
36
+ "early_stopping": true,
37
+ "max_length": 300,
38
+ "num_beams": 4,
39
+ "prefix": "translate English to German: "
40
+ },
41
+ "translation_en_to_fr": {
42
+ "early_stopping": true,
43
+ "max_length": 300,
44
+ "num_beams": 4,
45
+ "prefix": "translate English to French: "
46
+ },
47
+ "translation_en_to_ro": {
48
+ "early_stopping": true,
49
+ "max_length": 300,
50
+ "num_beams": 4,
51
+ "prefix": "translate English to Romanian: "
52
+ }
53
+ },
54
+ "torch_dtype": "float32",
55
+ "transformers_version": "4.10.0",
56
+ "use_cache": true,
57
+ "vocab_size": 250102
58
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f1a1606eaccc38d548088e90223e9960163001a8bc8a545385918255fac95b
3
+ size 352532601
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9f02989a9ec6ef92d1f9d7a4083c69c59e574acbc98f6fc53963490f8bc7884
3
+ size 688496379
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ed4cc86e2aef33ae599c2753fdf26e392a4a2bee7004647e1e310f1bf0a4f42
3
+ size 14503
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4dfaa600da6606fbaba6875e4eec37cb5ac02d115d936093d0ee1e632d0783e
3
+ size 623
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef78f86560d809067d12bac6c09f19a462cb3af3f54d2b8acbba26e1433125d6
3
+ size 4309802
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0, "additional_special_tokens": null, "sp_model_kwargs": {}, "special_tokens_map_file": "C:\\Users\\hydra/.cache\\huggingface\\transformers\\bde22e2f08347211bf47b72fb4730b4b9a2e3323a6a1254932cf0fb969eb18cd.b02fba13e4584f40329b20be0bad88c2ef97b5fda7d9f4f69b1cb3a620568a86", "tokenizer_file": null, "name_or_path": "longcld/t5_small_qg_ae_hl", "tokenizer_class": "T5Tokenizer"}
trainer_state.json ADDED
@@ -0,0 +1,3886 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 9.690504807692308,
5
+ "global_step": 64500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 9.984975961538462e-05,
13
+ "loss": 9.5457,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.03,
18
+ "learning_rate": 9.969951923076923e-05,
19
+ "loss": 6.3223,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.05,
24
+ "learning_rate": 9.954927884615385e-05,
25
+ "loss": 5.9118,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 0.06,
30
+ "learning_rate": 9.939903846153847e-05,
31
+ "loss": 5.6779,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 0.08,
36
+ "learning_rate": 9.924879807692308e-05,
37
+ "loss": 5.4467,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 0.09,
42
+ "learning_rate": 9.909855769230769e-05,
43
+ "loss": 5.3733,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 0.11,
48
+ "learning_rate": 9.894831730769232e-05,
49
+ "loss": 5.3544,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 0.12,
54
+ "learning_rate": 9.879807692307693e-05,
55
+ "loss": 5.2784,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 0.14,
60
+ "learning_rate": 9.864783653846154e-05,
61
+ "loss": 5.1503,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 0.15,
66
+ "learning_rate": 9.849759615384617e-05,
67
+ "loss": 5.0107,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 0.17,
72
+ "learning_rate": 9.834735576923078e-05,
73
+ "loss": 5.1408,
74
+ "step": 1100
75
+ },
76
+ {
77
+ "epoch": 0.18,
78
+ "learning_rate": 9.819711538461539e-05,
79
+ "loss": 4.9742,
80
+ "step": 1200
81
+ },
82
+ {
83
+ "epoch": 0.2,
84
+ "learning_rate": 9.8046875e-05,
85
+ "loss": 4.8672,
86
+ "step": 1300
87
+ },
88
+ {
89
+ "epoch": 0.21,
90
+ "learning_rate": 9.789663461538462e-05,
91
+ "loss": 4.858,
92
+ "step": 1400
93
+ },
94
+ {
95
+ "epoch": 0.23,
96
+ "learning_rate": 9.774639423076923e-05,
97
+ "loss": 4.8912,
98
+ "step": 1500
99
+ },
100
+ {
101
+ "epoch": 0.24,
102
+ "learning_rate": 9.759615384615385e-05,
103
+ "loss": 4.8654,
104
+ "step": 1600
105
+ },
106
+ {
107
+ "epoch": 0.26,
108
+ "learning_rate": 9.744591346153847e-05,
109
+ "loss": 4.8192,
110
+ "step": 1700
111
+ },
112
+ {
113
+ "epoch": 0.27,
114
+ "learning_rate": 9.729567307692308e-05,
115
+ "loss": 4.7405,
116
+ "step": 1800
117
+ },
118
+ {
119
+ "epoch": 0.29,
120
+ "learning_rate": 9.714543269230769e-05,
121
+ "loss": 4.69,
122
+ "step": 1900
123
+ },
124
+ {
125
+ "epoch": 0.3,
126
+ "learning_rate": 9.699519230769232e-05,
127
+ "loss": 4.5945,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 0.32,
132
+ "learning_rate": 9.684495192307693e-05,
133
+ "loss": 4.621,
134
+ "step": 2100
135
+ },
136
+ {
137
+ "epoch": 0.33,
138
+ "learning_rate": 9.669471153846155e-05,
139
+ "loss": 4.68,
140
+ "step": 2200
141
+ },
142
+ {
143
+ "epoch": 0.35,
144
+ "learning_rate": 9.654447115384616e-05,
145
+ "loss": 4.5514,
146
+ "step": 2300
147
+ },
148
+ {
149
+ "epoch": 0.36,
150
+ "learning_rate": 9.639423076923078e-05,
151
+ "loss": 4.6192,
152
+ "step": 2400
153
+ },
154
+ {
155
+ "epoch": 0.38,
156
+ "learning_rate": 9.624399038461539e-05,
157
+ "loss": 4.4672,
158
+ "step": 2500
159
+ },
160
+ {
161
+ "epoch": 0.39,
162
+ "learning_rate": 9.609375e-05,
163
+ "loss": 4.5006,
164
+ "step": 2600
165
+ },
166
+ {
167
+ "epoch": 0.41,
168
+ "learning_rate": 9.594350961538461e-05,
169
+ "loss": 4.5668,
170
+ "step": 2700
171
+ },
172
+ {
173
+ "epoch": 0.42,
174
+ "learning_rate": 9.579326923076924e-05,
175
+ "loss": 4.5345,
176
+ "step": 2800
177
+ },
178
+ {
179
+ "epoch": 0.44,
180
+ "learning_rate": 9.564302884615385e-05,
181
+ "loss": 4.4806,
182
+ "step": 2900
183
+ },
184
+ {
185
+ "epoch": 0.45,
186
+ "learning_rate": 9.549278846153846e-05,
187
+ "loss": 4.3853,
188
+ "step": 3000
189
+ },
190
+ {
191
+ "epoch": 0.47,
192
+ "learning_rate": 9.534254807692308e-05,
193
+ "loss": 4.4482,
194
+ "step": 3100
195
+ },
196
+ {
197
+ "epoch": 0.48,
198
+ "learning_rate": 9.519230769230769e-05,
199
+ "loss": 4.4437,
200
+ "step": 3200
201
+ },
202
+ {
203
+ "epoch": 0.5,
204
+ "learning_rate": 9.504206730769231e-05,
205
+ "loss": 4.3163,
206
+ "step": 3300
207
+ },
208
+ {
209
+ "epoch": 0.51,
210
+ "learning_rate": 9.489182692307693e-05,
211
+ "loss": 4.3702,
212
+ "step": 3400
213
+ },
214
+ {
215
+ "epoch": 0.53,
216
+ "learning_rate": 9.474158653846155e-05,
217
+ "loss": 4.3657,
218
+ "step": 3500
219
+ },
220
+ {
221
+ "epoch": 0.54,
222
+ "learning_rate": 9.459134615384616e-05,
223
+ "loss": 4.4059,
224
+ "step": 3600
225
+ },
226
+ {
227
+ "epoch": 0.56,
228
+ "learning_rate": 9.444110576923078e-05,
229
+ "loss": 4.3584,
230
+ "step": 3700
231
+ },
232
+ {
233
+ "epoch": 0.57,
234
+ "learning_rate": 9.429086538461539e-05,
235
+ "loss": 4.2906,
236
+ "step": 3800
237
+ },
238
+ {
239
+ "epoch": 0.59,
240
+ "learning_rate": 9.4140625e-05,
241
+ "loss": 4.2862,
242
+ "step": 3900
243
+ },
244
+ {
245
+ "epoch": 0.6,
246
+ "learning_rate": 9.399038461538463e-05,
247
+ "loss": 4.3426,
248
+ "step": 4000
249
+ },
250
+ {
251
+ "epoch": 0.62,
252
+ "learning_rate": 9.384014423076924e-05,
253
+ "loss": 4.2767,
254
+ "step": 4100
255
+ },
256
+ {
257
+ "epoch": 0.63,
258
+ "learning_rate": 9.368990384615385e-05,
259
+ "loss": 4.3542,
260
+ "step": 4200
261
+ },
262
+ {
263
+ "epoch": 0.65,
264
+ "learning_rate": 9.353966346153846e-05,
265
+ "loss": 4.3341,
266
+ "step": 4300
267
+ },
268
+ {
269
+ "epoch": 0.66,
270
+ "learning_rate": 9.338942307692307e-05,
271
+ "loss": 4.2464,
272
+ "step": 4400
273
+ },
274
+ {
275
+ "epoch": 0.68,
276
+ "learning_rate": 9.323918269230769e-05,
277
+ "loss": 4.3169,
278
+ "step": 4500
279
+ },
280
+ {
281
+ "epoch": 0.69,
282
+ "learning_rate": 9.308894230769231e-05,
283
+ "loss": 4.1644,
284
+ "step": 4600
285
+ },
286
+ {
287
+ "epoch": 0.71,
288
+ "learning_rate": 9.293870192307694e-05,
289
+ "loss": 4.3213,
290
+ "step": 4700
291
+ },
292
+ {
293
+ "epoch": 0.72,
294
+ "learning_rate": 9.278846153846155e-05,
295
+ "loss": 4.2788,
296
+ "step": 4800
297
+ },
298
+ {
299
+ "epoch": 0.74,
300
+ "learning_rate": 9.263822115384616e-05,
301
+ "loss": 4.2807,
302
+ "step": 4900
303
+ },
304
+ {
305
+ "epoch": 0.75,
306
+ "learning_rate": 9.248798076923077e-05,
307
+ "loss": 4.1281,
308
+ "step": 5000
309
+ },
310
+ {
311
+ "epoch": 0.77,
312
+ "learning_rate": 9.233774038461539e-05,
313
+ "loss": 4.2481,
314
+ "step": 5100
315
+ },
316
+ {
317
+ "epoch": 0.78,
318
+ "learning_rate": 9.21875e-05,
319
+ "loss": 4.163,
320
+ "step": 5200
321
+ },
322
+ {
323
+ "epoch": 0.8,
324
+ "learning_rate": 9.203725961538462e-05,
325
+ "loss": 4.2567,
326
+ "step": 5300
327
+ },
328
+ {
329
+ "epoch": 0.81,
330
+ "learning_rate": 9.188701923076924e-05,
331
+ "loss": 4.2224,
332
+ "step": 5400
333
+ },
334
+ {
335
+ "epoch": 0.83,
336
+ "learning_rate": 9.173677884615385e-05,
337
+ "loss": 4.132,
338
+ "step": 5500
339
+ },
340
+ {
341
+ "epoch": 0.84,
342
+ "learning_rate": 9.158653846153846e-05,
343
+ "loss": 4.1601,
344
+ "step": 5600
345
+ },
346
+ {
347
+ "epoch": 0.86,
348
+ "learning_rate": 9.143629807692307e-05,
349
+ "loss": 4.0697,
350
+ "step": 5700
351
+ },
352
+ {
353
+ "epoch": 0.87,
354
+ "learning_rate": 9.12860576923077e-05,
355
+ "loss": 4.1043,
356
+ "step": 5800
357
+ },
358
+ {
359
+ "epoch": 0.89,
360
+ "learning_rate": 9.113581730769232e-05,
361
+ "loss": 4.1087,
362
+ "step": 5900
363
+ },
364
+ {
365
+ "epoch": 0.9,
366
+ "learning_rate": 9.098557692307694e-05,
367
+ "loss": 3.9814,
368
+ "step": 6000
369
+ },
370
+ {
371
+ "epoch": 0.92,
372
+ "learning_rate": 9.083533653846155e-05,
373
+ "loss": 4.0993,
374
+ "step": 6100
375
+ },
376
+ {
377
+ "epoch": 0.93,
378
+ "learning_rate": 9.068509615384616e-05,
379
+ "loss": 4.1064,
380
+ "step": 6200
381
+ },
382
+ {
383
+ "epoch": 0.95,
384
+ "learning_rate": 9.053485576923077e-05,
385
+ "loss": 4.065,
386
+ "step": 6300
387
+ },
388
+ {
389
+ "epoch": 0.96,
390
+ "learning_rate": 9.038461538461538e-05,
391
+ "loss": 4.0022,
392
+ "step": 6400
393
+ },
394
+ {
395
+ "epoch": 0.98,
396
+ "learning_rate": 9.023437500000001e-05,
397
+ "loss": 4.0968,
398
+ "step": 6500
399
+ },
400
+ {
401
+ "epoch": 0.99,
402
+ "learning_rate": 9.008413461538462e-05,
403
+ "loss": 4.1214,
404
+ "step": 6600
405
+ },
406
+ {
407
+ "epoch": 1.01,
408
+ "learning_rate": 8.993389423076923e-05,
409
+ "loss": 4.0841,
410
+ "step": 6700
411
+ },
412
+ {
413
+ "epoch": 1.02,
414
+ "learning_rate": 8.978365384615385e-05,
415
+ "loss": 4.0666,
416
+ "step": 6800
417
+ },
418
+ {
419
+ "epoch": 1.04,
420
+ "learning_rate": 8.963341346153846e-05,
421
+ "loss": 3.9632,
422
+ "step": 6900
423
+ },
424
+ {
425
+ "epoch": 1.05,
426
+ "learning_rate": 8.948317307692307e-05,
427
+ "loss": 3.9496,
428
+ "step": 7000
429
+ },
430
+ {
431
+ "epoch": 1.07,
432
+ "learning_rate": 8.93329326923077e-05,
433
+ "loss": 3.9757,
434
+ "step": 7100
435
+ },
436
+ {
437
+ "epoch": 1.08,
438
+ "learning_rate": 8.918269230769232e-05,
439
+ "loss": 3.9959,
440
+ "step": 7200
441
+ },
442
+ {
443
+ "epoch": 1.1,
444
+ "learning_rate": 8.903245192307693e-05,
445
+ "loss": 3.9594,
446
+ "step": 7300
447
+ },
448
+ {
449
+ "epoch": 1.11,
450
+ "learning_rate": 8.888221153846155e-05,
451
+ "loss": 3.9735,
452
+ "step": 7400
453
+ },
454
+ {
455
+ "epoch": 1.13,
456
+ "learning_rate": 8.873197115384616e-05,
457
+ "loss": 3.9647,
458
+ "step": 7500
459
+ },
460
+ {
461
+ "epoch": 1.14,
462
+ "learning_rate": 8.858173076923077e-05,
463
+ "loss": 3.8759,
464
+ "step": 7600
465
+ },
466
+ {
467
+ "epoch": 1.16,
468
+ "learning_rate": 8.84314903846154e-05,
469
+ "loss": 3.9956,
470
+ "step": 7700
471
+ },
472
+ {
473
+ "epoch": 1.17,
474
+ "learning_rate": 8.828125000000001e-05,
475
+ "loss": 3.9526,
476
+ "step": 7800
477
+ },
478
+ {
479
+ "epoch": 1.19,
480
+ "learning_rate": 8.813100961538462e-05,
481
+ "loss": 3.9409,
482
+ "step": 7900
483
+ },
484
+ {
485
+ "epoch": 1.2,
486
+ "learning_rate": 8.798076923076923e-05,
487
+ "loss": 3.9545,
488
+ "step": 8000
489
+ },
490
+ {
491
+ "epoch": 1.22,
492
+ "learning_rate": 8.783052884615384e-05,
493
+ "loss": 3.8668,
494
+ "step": 8100
495
+ },
496
+ {
497
+ "epoch": 1.23,
498
+ "learning_rate": 8.768028846153846e-05,
499
+ "loss": 3.8675,
500
+ "step": 8200
501
+ },
502
+ {
503
+ "epoch": 1.25,
504
+ "learning_rate": 8.753004807692308e-05,
505
+ "loss": 3.9234,
506
+ "step": 8300
507
+ },
508
+ {
509
+ "epoch": 1.26,
510
+ "learning_rate": 8.73798076923077e-05,
511
+ "loss": 3.9181,
512
+ "step": 8400
513
+ },
514
+ {
515
+ "epoch": 1.28,
516
+ "learning_rate": 8.722956730769232e-05,
517
+ "loss": 3.9288,
518
+ "step": 8500
519
+ },
520
+ {
521
+ "epoch": 1.29,
522
+ "learning_rate": 8.707932692307693e-05,
523
+ "loss": 3.9103,
524
+ "step": 8600
525
+ },
526
+ {
527
+ "epoch": 1.31,
528
+ "learning_rate": 8.692908653846154e-05,
529
+ "loss": 3.9452,
530
+ "step": 8700
531
+ },
532
+ {
533
+ "epoch": 1.32,
534
+ "learning_rate": 8.677884615384616e-05,
535
+ "loss": 3.8211,
536
+ "step": 8800
537
+ },
538
+ {
539
+ "epoch": 1.34,
540
+ "learning_rate": 8.662860576923077e-05,
541
+ "loss": 3.7997,
542
+ "step": 8900
543
+ },
544
+ {
545
+ "epoch": 1.35,
546
+ "learning_rate": 8.64783653846154e-05,
547
+ "loss": 3.8444,
548
+ "step": 9000
549
+ },
550
+ {
551
+ "epoch": 1.37,
552
+ "learning_rate": 8.6328125e-05,
553
+ "loss": 3.8875,
554
+ "step": 9100
555
+ },
556
+ {
557
+ "epoch": 1.38,
558
+ "learning_rate": 8.617788461538462e-05,
559
+ "loss": 3.8792,
560
+ "step": 9200
561
+ },
562
+ {
563
+ "epoch": 1.4,
564
+ "learning_rate": 8.602764423076923e-05,
565
+ "loss": 3.8994,
566
+ "step": 9300
567
+ },
568
+ {
569
+ "epoch": 1.41,
570
+ "learning_rate": 8.587740384615384e-05,
571
+ "loss": 3.8862,
572
+ "step": 9400
573
+ },
574
+ {
575
+ "epoch": 1.43,
576
+ "learning_rate": 8.572716346153847e-05,
577
+ "loss": 3.824,
578
+ "step": 9500
579
+ },
580
+ {
581
+ "epoch": 1.44,
582
+ "learning_rate": 8.557692307692308e-05,
583
+ "loss": 3.7989,
584
+ "step": 9600
585
+ },
586
+ {
587
+ "epoch": 1.46,
588
+ "learning_rate": 8.542668269230769e-05,
589
+ "loss": 3.8998,
590
+ "step": 9700
591
+ },
592
+ {
593
+ "epoch": 1.47,
594
+ "learning_rate": 8.527644230769232e-05,
595
+ "loss": 3.8514,
596
+ "step": 9800
597
+ },
598
+ {
599
+ "epoch": 1.49,
600
+ "learning_rate": 8.512620192307693e-05,
601
+ "loss": 3.8324,
602
+ "step": 9900
603
+ },
604
+ {
605
+ "epoch": 1.5,
606
+ "learning_rate": 8.497596153846154e-05,
607
+ "loss": 3.8852,
608
+ "step": 10000
609
+ },
610
+ {
611
+ "epoch": 1.52,
612
+ "learning_rate": 8.482572115384615e-05,
613
+ "loss": 3.7941,
614
+ "step": 10100
615
+ },
616
+ {
617
+ "epoch": 1.53,
618
+ "learning_rate": 8.467548076923078e-05,
619
+ "loss": 3.9133,
620
+ "step": 10200
621
+ },
622
+ {
623
+ "epoch": 1.55,
624
+ "learning_rate": 8.452524038461539e-05,
625
+ "loss": 3.7357,
626
+ "step": 10300
627
+ },
628
+ {
629
+ "epoch": 1.56,
630
+ "learning_rate": 8.4375e-05,
631
+ "loss": 3.7986,
632
+ "step": 10400
633
+ },
634
+ {
635
+ "epoch": 1.58,
636
+ "learning_rate": 8.422475961538462e-05,
637
+ "loss": 3.8262,
638
+ "step": 10500
639
+ },
640
+ {
641
+ "epoch": 1.59,
642
+ "learning_rate": 8.407451923076923e-05,
643
+ "loss": 3.888,
644
+ "step": 10600
645
+ },
646
+ {
647
+ "epoch": 1.61,
648
+ "learning_rate": 8.392427884615384e-05,
649
+ "loss": 3.7276,
650
+ "step": 10700
651
+ },
652
+ {
653
+ "epoch": 1.62,
654
+ "learning_rate": 8.377403846153847e-05,
655
+ "loss": 3.8952,
656
+ "step": 10800
657
+ },
658
+ {
659
+ "epoch": 1.64,
660
+ "learning_rate": 8.362379807692308e-05,
661
+ "loss": 3.89,
662
+ "step": 10900
663
+ },
664
+ {
665
+ "epoch": 1.65,
666
+ "learning_rate": 8.347355769230769e-05,
667
+ "loss": 3.8234,
668
+ "step": 11000
669
+ },
670
+ {
671
+ "epoch": 1.67,
672
+ "learning_rate": 8.332331730769232e-05,
673
+ "loss": 3.7485,
674
+ "step": 11100
675
+ },
676
+ {
677
+ "epoch": 1.68,
678
+ "learning_rate": 8.317307692307693e-05,
679
+ "loss": 3.7877,
680
+ "step": 11200
681
+ },
682
+ {
683
+ "epoch": 1.7,
684
+ "learning_rate": 8.302283653846154e-05,
685
+ "loss": 3.7776,
686
+ "step": 11300
687
+ },
688
+ {
689
+ "epoch": 1.71,
690
+ "learning_rate": 8.287259615384617e-05,
691
+ "loss": 3.7945,
692
+ "step": 11400
693
+ },
694
+ {
695
+ "epoch": 1.73,
696
+ "learning_rate": 8.272235576923078e-05,
697
+ "loss": 3.8574,
698
+ "step": 11500
699
+ },
700
+ {
701
+ "epoch": 1.74,
702
+ "learning_rate": 8.257211538461539e-05,
703
+ "loss": 3.7562,
704
+ "step": 11600
705
+ },
706
+ {
707
+ "epoch": 1.76,
708
+ "learning_rate": 8.2421875e-05,
709
+ "loss": 3.7713,
710
+ "step": 11700
711
+ },
712
+ {
713
+ "epoch": 1.77,
714
+ "learning_rate": 8.227163461538461e-05,
715
+ "loss": 3.7257,
716
+ "step": 11800
717
+ },
718
+ {
719
+ "epoch": 1.79,
720
+ "learning_rate": 8.212139423076923e-05,
721
+ "loss": 3.7542,
722
+ "step": 11900
723
+ },
724
+ {
725
+ "epoch": 1.8,
726
+ "learning_rate": 8.197115384615385e-05,
727
+ "loss": 3.6902,
728
+ "step": 12000
729
+ },
730
+ {
731
+ "epoch": 1.82,
732
+ "learning_rate": 8.182091346153846e-05,
733
+ "loss": 3.7354,
734
+ "step": 12100
735
+ },
736
+ {
737
+ "epoch": 1.83,
738
+ "learning_rate": 8.167067307692308e-05,
739
+ "loss": 3.745,
740
+ "step": 12200
741
+ },
742
+ {
743
+ "epoch": 1.85,
744
+ "learning_rate": 8.152043269230769e-05,
745
+ "loss": 3.7681,
746
+ "step": 12300
747
+ },
748
+ {
749
+ "epoch": 1.86,
750
+ "learning_rate": 8.137019230769231e-05,
751
+ "loss": 3.8431,
752
+ "step": 12400
753
+ },
754
+ {
755
+ "epoch": 1.88,
756
+ "learning_rate": 8.121995192307693e-05,
757
+ "loss": 3.7285,
758
+ "step": 12500
759
+ },
760
+ {
761
+ "epoch": 1.89,
762
+ "learning_rate": 8.106971153846155e-05,
763
+ "loss": 3.6801,
764
+ "step": 12600
765
+ },
766
+ {
767
+ "epoch": 1.91,
768
+ "learning_rate": 8.091947115384616e-05,
769
+ "loss": 3.6456,
770
+ "step": 12700
771
+ },
772
+ {
773
+ "epoch": 1.92,
774
+ "learning_rate": 8.076923076923078e-05,
775
+ "loss": 3.764,
776
+ "step": 12800
777
+ },
778
+ {
779
+ "epoch": 1.94,
780
+ "learning_rate": 8.061899038461539e-05,
781
+ "loss": 3.7537,
782
+ "step": 12900
783
+ },
784
+ {
785
+ "epoch": 1.95,
786
+ "learning_rate": 8.046875e-05,
787
+ "loss": 3.5876,
788
+ "step": 13000
789
+ },
790
+ {
791
+ "epoch": 1.97,
792
+ "learning_rate": 8.031850961538461e-05,
793
+ "loss": 3.8075,
794
+ "step": 13100
795
+ },
796
+ {
797
+ "epoch": 1.98,
798
+ "learning_rate": 8.016826923076924e-05,
799
+ "loss": 3.6953,
800
+ "step": 13200
801
+ },
802
+ {
803
+ "epoch": 2.0,
804
+ "learning_rate": 8.001802884615385e-05,
805
+ "loss": 3.6577,
806
+ "step": 13300
807
+ },
808
+ {
809
+ "epoch": 2.01,
810
+ "learning_rate": 7.986778846153846e-05,
811
+ "loss": 3.7064,
812
+ "step": 13400
813
+ },
814
+ {
815
+ "epoch": 2.03,
816
+ "learning_rate": 7.971754807692308e-05,
817
+ "loss": 3.6817,
818
+ "step": 13500
819
+ },
820
+ {
821
+ "epoch": 2.04,
822
+ "learning_rate": 7.956730769230769e-05,
823
+ "loss": 3.6729,
824
+ "step": 13600
825
+ },
826
+ {
827
+ "epoch": 2.06,
828
+ "learning_rate": 7.941706730769231e-05,
829
+ "loss": 3.6737,
830
+ "step": 13700
831
+ },
832
+ {
833
+ "epoch": 2.07,
834
+ "learning_rate": 7.926682692307693e-05,
835
+ "loss": 3.6222,
836
+ "step": 13800
837
+ },
838
+ {
839
+ "epoch": 2.09,
840
+ "learning_rate": 7.911658653846155e-05,
841
+ "loss": 3.6704,
842
+ "step": 13900
843
+ },
844
+ {
845
+ "epoch": 2.1,
846
+ "learning_rate": 7.896634615384616e-05,
847
+ "loss": 3.7445,
848
+ "step": 14000
849
+ },
850
+ {
851
+ "epoch": 2.12,
852
+ "learning_rate": 7.881610576923078e-05,
853
+ "loss": 3.6227,
854
+ "step": 14100
855
+ },
856
+ {
857
+ "epoch": 2.13,
858
+ "learning_rate": 7.866586538461539e-05,
859
+ "loss": 3.6822,
860
+ "step": 14200
861
+ },
862
+ {
863
+ "epoch": 2.15,
864
+ "learning_rate": 7.8515625e-05,
865
+ "loss": 3.6178,
866
+ "step": 14300
867
+ },
868
+ {
869
+ "epoch": 2.16,
870
+ "learning_rate": 7.836538461538462e-05,
871
+ "loss": 3.6329,
872
+ "step": 14400
873
+ },
874
+ {
875
+ "epoch": 2.18,
876
+ "learning_rate": 7.821514423076924e-05,
877
+ "loss": 3.627,
878
+ "step": 14500
879
+ },
880
+ {
881
+ "epoch": 2.19,
882
+ "learning_rate": 7.806490384615385e-05,
883
+ "loss": 3.7061,
884
+ "step": 14600
885
+ },
886
+ {
887
+ "epoch": 2.21,
888
+ "learning_rate": 7.791466346153846e-05,
889
+ "loss": 3.7037,
890
+ "step": 14700
891
+ },
892
+ {
893
+ "epoch": 2.22,
894
+ "learning_rate": 7.776442307692307e-05,
895
+ "loss": 3.5797,
896
+ "step": 14800
897
+ },
898
+ {
899
+ "epoch": 2.24,
900
+ "learning_rate": 7.761418269230769e-05,
901
+ "loss": 3.6865,
902
+ "step": 14900
903
+ },
904
+ {
905
+ "epoch": 2.25,
906
+ "learning_rate": 7.746394230769231e-05,
907
+ "loss": 3.7016,
908
+ "step": 15000
909
+ },
910
+ {
911
+ "epoch": 2.27,
912
+ "learning_rate": 7.731370192307694e-05,
913
+ "loss": 3.6149,
914
+ "step": 15100
915
+ },
916
+ {
917
+ "epoch": 2.28,
918
+ "learning_rate": 7.716346153846155e-05,
919
+ "loss": 3.614,
920
+ "step": 15200
921
+ },
922
+ {
923
+ "epoch": 2.3,
924
+ "learning_rate": 7.701322115384616e-05,
925
+ "loss": 3.5511,
926
+ "step": 15300
927
+ },
928
+ {
929
+ "epoch": 2.31,
930
+ "learning_rate": 7.686298076923077e-05,
931
+ "loss": 3.5887,
932
+ "step": 15400
933
+ },
934
+ {
935
+ "epoch": 2.33,
936
+ "learning_rate": 7.671274038461539e-05,
937
+ "loss": 3.6247,
938
+ "step": 15500
939
+ },
940
+ {
941
+ "epoch": 2.34,
942
+ "learning_rate": 7.65625e-05,
943
+ "loss": 3.5886,
944
+ "step": 15600
945
+ },
946
+ {
947
+ "epoch": 2.36,
948
+ "learning_rate": 7.641225961538462e-05,
949
+ "loss": 3.4512,
950
+ "step": 15700
951
+ },
952
+ {
953
+ "epoch": 2.37,
954
+ "learning_rate": 7.626201923076924e-05,
955
+ "loss": 3.4764,
956
+ "step": 15800
957
+ },
958
+ {
959
+ "epoch": 2.39,
960
+ "learning_rate": 7.611177884615385e-05,
961
+ "loss": 3.6115,
962
+ "step": 15900
963
+ },
964
+ {
965
+ "epoch": 2.4,
966
+ "learning_rate": 7.596153846153846e-05,
967
+ "loss": 3.5766,
968
+ "step": 16000
969
+ },
970
+ {
971
+ "epoch": 2.42,
972
+ "learning_rate": 7.581129807692307e-05,
973
+ "loss": 3.6717,
974
+ "step": 16100
975
+ },
976
+ {
977
+ "epoch": 2.43,
978
+ "learning_rate": 7.56610576923077e-05,
979
+ "loss": 3.4975,
980
+ "step": 16200
981
+ },
982
+ {
983
+ "epoch": 2.45,
984
+ "learning_rate": 7.551081730769232e-05,
985
+ "loss": 3.572,
986
+ "step": 16300
987
+ },
988
+ {
989
+ "epoch": 2.46,
990
+ "learning_rate": 7.536057692307694e-05,
991
+ "loss": 3.5285,
992
+ "step": 16400
993
+ },
994
+ {
995
+ "epoch": 2.48,
996
+ "learning_rate": 7.521033653846155e-05,
997
+ "loss": 3.634,
998
+ "step": 16500
999
+ },
1000
+ {
1001
+ "epoch": 2.49,
1002
+ "learning_rate": 7.506009615384616e-05,
1003
+ "loss": 3.5654,
1004
+ "step": 16600
1005
+ },
1006
+ {
1007
+ "epoch": 2.51,
1008
+ "learning_rate": 7.490985576923077e-05,
1009
+ "loss": 3.5617,
1010
+ "step": 16700
1011
+ },
1012
+ {
1013
+ "epoch": 2.52,
1014
+ "learning_rate": 7.475961538461538e-05,
1015
+ "loss": 3.6074,
1016
+ "step": 16800
1017
+ },
1018
+ {
1019
+ "epoch": 2.54,
1020
+ "learning_rate": 7.460937500000001e-05,
1021
+ "loss": 3.5804,
1022
+ "step": 16900
1023
+ },
1024
+ {
1025
+ "epoch": 2.55,
1026
+ "learning_rate": 7.445913461538462e-05,
1027
+ "loss": 3.5686,
1028
+ "step": 17000
1029
+ },
1030
+ {
1031
+ "epoch": 2.57,
1032
+ "learning_rate": 7.430889423076923e-05,
1033
+ "loss": 3.5895,
1034
+ "step": 17100
1035
+ },
1036
+ {
1037
+ "epoch": 2.58,
1038
+ "learning_rate": 7.415865384615385e-05,
1039
+ "loss": 3.5292,
1040
+ "step": 17200
1041
+ },
1042
+ {
1043
+ "epoch": 2.6,
1044
+ "learning_rate": 7.400841346153846e-05,
1045
+ "loss": 3.553,
1046
+ "step": 17300
1047
+ },
1048
+ {
1049
+ "epoch": 2.61,
1050
+ "learning_rate": 7.385817307692307e-05,
1051
+ "loss": 3.6004,
1052
+ "step": 17400
1053
+ },
1054
+ {
1055
+ "epoch": 2.63,
1056
+ "learning_rate": 7.37079326923077e-05,
1057
+ "loss": 3.5905,
1058
+ "step": 17500
1059
+ },
1060
+ {
1061
+ "epoch": 2.64,
1062
+ "learning_rate": 7.355769230769232e-05,
1063
+ "loss": 3.4882,
1064
+ "step": 17600
1065
+ },
1066
+ {
1067
+ "epoch": 2.66,
1068
+ "learning_rate": 7.340745192307693e-05,
1069
+ "loss": 3.7265,
1070
+ "step": 17700
1071
+ },
1072
+ {
1073
+ "epoch": 2.67,
1074
+ "learning_rate": 7.325721153846155e-05,
1075
+ "loss": 3.4666,
1076
+ "step": 17800
1077
+ },
1078
+ {
1079
+ "epoch": 2.69,
1080
+ "learning_rate": 7.310697115384616e-05,
1081
+ "loss": 3.5222,
1082
+ "step": 17900
1083
+ },
1084
+ {
1085
+ "epoch": 2.7,
1086
+ "learning_rate": 7.295673076923077e-05,
1087
+ "loss": 3.5472,
1088
+ "step": 18000
1089
+ },
1090
+ {
1091
+ "epoch": 2.72,
1092
+ "learning_rate": 7.28064903846154e-05,
1093
+ "loss": 3.4465,
1094
+ "step": 18100
1095
+ },
1096
+ {
1097
+ "epoch": 2.73,
1098
+ "learning_rate": 7.265625000000001e-05,
1099
+ "loss": 3.4855,
1100
+ "step": 18200
1101
+ },
1102
+ {
1103
+ "epoch": 2.75,
1104
+ "learning_rate": 7.250600961538462e-05,
1105
+ "loss": 3.4352,
1106
+ "step": 18300
1107
+ },
1108
+ {
1109
+ "epoch": 2.76,
1110
+ "learning_rate": 7.235576923076923e-05,
1111
+ "loss": 3.5702,
1112
+ "step": 18400
1113
+ },
1114
+ {
1115
+ "epoch": 2.78,
1116
+ "learning_rate": 7.220552884615384e-05,
1117
+ "loss": 3.5483,
1118
+ "step": 18500
1119
+ },
1120
+ {
1121
+ "epoch": 2.79,
1122
+ "learning_rate": 7.205528846153846e-05,
1123
+ "loss": 3.4681,
1124
+ "step": 18600
1125
+ },
1126
+ {
1127
+ "epoch": 2.81,
1128
+ "learning_rate": 7.190504807692308e-05,
1129
+ "loss": 3.4907,
1130
+ "step": 18700
1131
+ },
1132
+ {
1133
+ "epoch": 2.82,
1134
+ "learning_rate": 7.17548076923077e-05,
1135
+ "loss": 3.6045,
1136
+ "step": 18800
1137
+ },
1138
+ {
1139
+ "epoch": 2.84,
1140
+ "learning_rate": 7.160456730769232e-05,
1141
+ "loss": 3.4191,
1142
+ "step": 18900
1143
+ },
1144
+ {
1145
+ "epoch": 2.85,
1146
+ "learning_rate": 7.145432692307693e-05,
1147
+ "loss": 3.5984,
1148
+ "step": 19000
1149
+ },
1150
+ {
1151
+ "epoch": 2.87,
1152
+ "learning_rate": 7.130408653846154e-05,
1153
+ "loss": 3.4533,
1154
+ "step": 19100
1155
+ },
1156
+ {
1157
+ "epoch": 2.88,
1158
+ "learning_rate": 7.115384615384616e-05,
1159
+ "loss": 3.4745,
1160
+ "step": 19200
1161
+ },
1162
+ {
1163
+ "epoch": 2.9,
1164
+ "learning_rate": 7.100360576923077e-05,
1165
+ "loss": 3.4572,
1166
+ "step": 19300
1167
+ },
1168
+ {
1169
+ "epoch": 2.91,
1170
+ "learning_rate": 7.08533653846154e-05,
1171
+ "loss": 3.5035,
1172
+ "step": 19400
1173
+ },
1174
+ {
1175
+ "epoch": 2.93,
1176
+ "learning_rate": 7.0703125e-05,
1177
+ "loss": 3.5592,
1178
+ "step": 19500
1179
+ },
1180
+ {
1181
+ "epoch": 2.94,
1182
+ "learning_rate": 7.055288461538462e-05,
1183
+ "loss": 3.4138,
1184
+ "step": 19600
1185
+ },
1186
+ {
1187
+ "epoch": 2.96,
1188
+ "learning_rate": 7.040264423076923e-05,
1189
+ "loss": 3.4603,
1190
+ "step": 19700
1191
+ },
1192
+ {
1193
+ "epoch": 2.97,
1194
+ "learning_rate": 7.025240384615384e-05,
1195
+ "loss": 3.5511,
1196
+ "step": 19800
1197
+ },
1198
+ {
1199
+ "epoch": 2.99,
1200
+ "learning_rate": 7.010216346153847e-05,
1201
+ "loss": 3.4787,
1202
+ "step": 19900
1203
+ },
1204
+ {
1205
+ "epoch": 3.0,
1206
+ "learning_rate": 6.995192307692308e-05,
1207
+ "loss": 3.5999,
1208
+ "step": 20000
1209
+ },
1210
+ {
1211
+ "epoch": 3.02,
1212
+ "learning_rate": 6.980168269230769e-05,
1213
+ "loss": 3.4902,
1214
+ "step": 20100
1215
+ },
1216
+ {
1217
+ "epoch": 3.03,
1218
+ "learning_rate": 6.965144230769232e-05,
1219
+ "loss": 3.5083,
1220
+ "step": 20200
1221
+ },
1222
+ {
1223
+ "epoch": 3.05,
1224
+ "learning_rate": 6.950120192307693e-05,
1225
+ "loss": 3.5033,
1226
+ "step": 20300
1227
+ },
1228
+ {
1229
+ "epoch": 3.06,
1230
+ "learning_rate": 6.935096153846154e-05,
1231
+ "loss": 3.5604,
1232
+ "step": 20400
1233
+ },
1234
+ {
1235
+ "epoch": 3.08,
1236
+ "learning_rate": 6.920072115384615e-05,
1237
+ "loss": 3.5009,
1238
+ "step": 20500
1239
+ },
1240
+ {
1241
+ "epoch": 3.09,
1242
+ "learning_rate": 6.905048076923078e-05,
1243
+ "loss": 3.5076,
1244
+ "step": 20600
1245
+ },
1246
+ {
1247
+ "epoch": 3.11,
1248
+ "learning_rate": 6.890024038461539e-05,
1249
+ "loss": 3.4699,
1250
+ "step": 20700
1251
+ },
1252
+ {
1253
+ "epoch": 3.12,
1254
+ "learning_rate": 6.875e-05,
1255
+ "loss": 3.4902,
1256
+ "step": 20800
1257
+ },
1258
+ {
1259
+ "epoch": 3.14,
1260
+ "learning_rate": 6.859975961538462e-05,
1261
+ "loss": 3.4528,
1262
+ "step": 20900
1263
+ },
1264
+ {
1265
+ "epoch": 3.16,
1266
+ "learning_rate": 6.844951923076923e-05,
1267
+ "loss": 3.5253,
1268
+ "step": 21000
1269
+ },
1270
+ {
1271
+ "epoch": 3.17,
1272
+ "learning_rate": 6.829927884615384e-05,
1273
+ "loss": 3.48,
1274
+ "step": 21100
1275
+ },
1276
+ {
1277
+ "epoch": 3.19,
1278
+ "learning_rate": 6.814903846153847e-05,
1279
+ "loss": 3.3898,
1280
+ "step": 21200
1281
+ },
1282
+ {
1283
+ "epoch": 3.2,
1284
+ "learning_rate": 6.799879807692308e-05,
1285
+ "loss": 3.4181,
1286
+ "step": 21300
1287
+ },
1288
+ {
1289
+ "epoch": 3.22,
1290
+ "learning_rate": 6.784855769230769e-05,
1291
+ "loss": 3.4118,
1292
+ "step": 21400
1293
+ },
1294
+ {
1295
+ "epoch": 3.23,
1296
+ "learning_rate": 6.769831730769232e-05,
1297
+ "loss": 3.442,
1298
+ "step": 21500
1299
+ },
1300
+ {
1301
+ "epoch": 3.25,
1302
+ "learning_rate": 6.754807692307693e-05,
1303
+ "loss": 3.4115,
1304
+ "step": 21600
1305
+ },
1306
+ {
1307
+ "epoch": 3.26,
1308
+ "learning_rate": 6.739783653846154e-05,
1309
+ "loss": 3.4211,
1310
+ "step": 21700
1311
+ },
1312
+ {
1313
+ "epoch": 3.28,
1314
+ "learning_rate": 6.724759615384617e-05,
1315
+ "loss": 3.4422,
1316
+ "step": 21800
1317
+ },
1318
+ {
1319
+ "epoch": 3.29,
1320
+ "learning_rate": 6.709735576923078e-05,
1321
+ "loss": 3.5675,
1322
+ "step": 21900
1323
+ },
1324
+ {
1325
+ "epoch": 3.31,
1326
+ "learning_rate": 6.694711538461539e-05,
1327
+ "loss": 3.3725,
1328
+ "step": 22000
1329
+ },
1330
+ {
1331
+ "epoch": 3.32,
1332
+ "learning_rate": 6.6796875e-05,
1333
+ "loss": 3.4709,
1334
+ "step": 22100
1335
+ },
1336
+ {
1337
+ "epoch": 3.34,
1338
+ "learning_rate": 6.664663461538461e-05,
1339
+ "loss": 3.4247,
1340
+ "step": 22200
1341
+ },
1342
+ {
1343
+ "epoch": 3.35,
1344
+ "learning_rate": 6.649639423076923e-05,
1345
+ "loss": 3.411,
1346
+ "step": 22300
1347
+ },
1348
+ {
1349
+ "epoch": 3.37,
1350
+ "learning_rate": 6.634615384615385e-05,
1351
+ "loss": 3.3555,
1352
+ "step": 22400
1353
+ },
1354
+ {
1355
+ "epoch": 3.38,
1356
+ "learning_rate": 6.619591346153846e-05,
1357
+ "loss": 3.3727,
1358
+ "step": 22500
1359
+ },
1360
+ {
1361
+ "epoch": 3.4,
1362
+ "learning_rate": 6.604567307692308e-05,
1363
+ "loss": 3.4175,
1364
+ "step": 22600
1365
+ },
1366
+ {
1367
+ "epoch": 3.41,
1368
+ "learning_rate": 6.589543269230769e-05,
1369
+ "loss": 3.4083,
1370
+ "step": 22700
1371
+ },
1372
+ {
1373
+ "epoch": 3.43,
1374
+ "learning_rate": 6.574519230769231e-05,
1375
+ "loss": 3.4315,
1376
+ "step": 22800
1377
+ },
1378
+ {
1379
+ "epoch": 3.44,
1380
+ "learning_rate": 6.559495192307693e-05,
1381
+ "loss": 3.4095,
1382
+ "step": 22900
1383
+ },
1384
+ {
1385
+ "epoch": 3.46,
1386
+ "learning_rate": 6.544471153846154e-05,
1387
+ "loss": 3.4292,
1388
+ "step": 23000
1389
+ },
1390
+ {
1391
+ "epoch": 3.47,
1392
+ "learning_rate": 6.529447115384616e-05,
1393
+ "loss": 3.4097,
1394
+ "step": 23100
1395
+ },
1396
+ {
1397
+ "epoch": 3.49,
1398
+ "learning_rate": 6.514423076923078e-05,
1399
+ "loss": 3.5273,
1400
+ "step": 23200
1401
+ },
1402
+ {
1403
+ "epoch": 3.5,
1404
+ "learning_rate": 6.499399038461539e-05,
1405
+ "loss": 3.4549,
1406
+ "step": 23300
1407
+ },
1408
+ {
1409
+ "epoch": 3.52,
1410
+ "learning_rate": 6.484375e-05,
1411
+ "loss": 3.3474,
1412
+ "step": 23400
1413
+ },
1414
+ {
1415
+ "epoch": 3.53,
1416
+ "learning_rate": 6.469350961538461e-05,
1417
+ "loss": 3.3646,
1418
+ "step": 23500
1419
+ },
1420
+ {
1421
+ "epoch": 3.55,
1422
+ "learning_rate": 6.454326923076924e-05,
1423
+ "loss": 3.4457,
1424
+ "step": 23600
1425
+ },
1426
+ {
1427
+ "epoch": 3.56,
1428
+ "learning_rate": 6.439302884615385e-05,
1429
+ "loss": 3.4549,
1430
+ "step": 23700
1431
+ },
1432
+ {
1433
+ "epoch": 3.58,
1434
+ "learning_rate": 6.424278846153846e-05,
1435
+ "loss": 3.4032,
1436
+ "step": 23800
1437
+ },
1438
+ {
1439
+ "epoch": 3.59,
1440
+ "learning_rate": 6.409254807692307e-05,
1441
+ "loss": 3.3047,
1442
+ "step": 23900
1443
+ },
1444
+ {
1445
+ "epoch": 3.61,
1446
+ "learning_rate": 6.394230769230769e-05,
1447
+ "loss": 3.4121,
1448
+ "step": 24000
1449
+ },
1450
+ {
1451
+ "epoch": 3.62,
1452
+ "learning_rate": 6.379206730769231e-05,
1453
+ "loss": 3.4355,
1454
+ "step": 24100
1455
+ },
1456
+ {
1457
+ "epoch": 3.64,
1458
+ "learning_rate": 6.364182692307692e-05,
1459
+ "loss": 3.4153,
1460
+ "step": 24200
1461
+ },
1462
+ {
1463
+ "epoch": 3.65,
1464
+ "learning_rate": 6.349158653846155e-05,
1465
+ "loss": 3.3583,
1466
+ "step": 24300
1467
+ },
1468
+ {
1469
+ "epoch": 3.67,
1470
+ "learning_rate": 6.334134615384616e-05,
1471
+ "loss": 3.3435,
1472
+ "step": 24400
1473
+ },
1474
+ {
1475
+ "epoch": 3.68,
1476
+ "learning_rate": 6.319110576923077e-05,
1477
+ "loss": 3.3166,
1478
+ "step": 24500
1479
+ },
1480
+ {
1481
+ "epoch": 3.7,
1482
+ "learning_rate": 6.304086538461539e-05,
1483
+ "loss": 3.3751,
1484
+ "step": 24600
1485
+ },
1486
+ {
1487
+ "epoch": 3.71,
1488
+ "learning_rate": 6.2890625e-05,
1489
+ "loss": 3.4042,
1490
+ "step": 24700
1491
+ },
1492
+ {
1493
+ "epoch": 3.73,
1494
+ "learning_rate": 6.274038461538461e-05,
1495
+ "loss": 3.3774,
1496
+ "step": 24800
1497
+ },
1498
+ {
1499
+ "epoch": 3.74,
1500
+ "learning_rate": 6.259014423076924e-05,
1501
+ "loss": 3.4476,
1502
+ "step": 24900
1503
+ },
1504
+ {
1505
+ "epoch": 3.76,
1506
+ "learning_rate": 6.243990384615385e-05,
1507
+ "loss": 3.3689,
1508
+ "step": 25000
1509
+ },
1510
+ {
1511
+ "epoch": 3.77,
1512
+ "learning_rate": 6.228966346153846e-05,
1513
+ "loss": 3.4963,
1514
+ "step": 25100
1515
+ },
1516
+ {
1517
+ "epoch": 3.79,
1518
+ "learning_rate": 6.213942307692307e-05,
1519
+ "loss": 3.3515,
1520
+ "step": 25200
1521
+ },
1522
+ {
1523
+ "epoch": 3.8,
1524
+ "learning_rate": 6.198918269230769e-05,
1525
+ "loss": 3.329,
1526
+ "step": 25300
1527
+ },
1528
+ {
1529
+ "epoch": 3.82,
1530
+ "learning_rate": 6.183894230769231e-05,
1531
+ "loss": 3.4401,
1532
+ "step": 25400
1533
+ },
1534
+ {
1535
+ "epoch": 3.83,
1536
+ "learning_rate": 6.168870192307694e-05,
1537
+ "loss": 3.3613,
1538
+ "step": 25500
1539
+ },
1540
+ {
1541
+ "epoch": 3.85,
1542
+ "learning_rate": 6.153846153846155e-05,
1543
+ "loss": 3.3463,
1544
+ "step": 25600
1545
+ },
1546
+ {
1547
+ "epoch": 3.86,
1548
+ "learning_rate": 6.138822115384616e-05,
1549
+ "loss": 3.3742,
1550
+ "step": 25700
1551
+ },
1552
+ {
1553
+ "epoch": 3.88,
1554
+ "learning_rate": 6.123798076923077e-05,
1555
+ "loss": 3.3858,
1556
+ "step": 25800
1557
+ },
1558
+ {
1559
+ "epoch": 3.89,
1560
+ "learning_rate": 6.108774038461539e-05,
1561
+ "loss": 3.4393,
1562
+ "step": 25900
1563
+ },
1564
+ {
1565
+ "epoch": 3.91,
1566
+ "learning_rate": 6.0937500000000004e-05,
1567
+ "loss": 3.4649,
1568
+ "step": 26000
1569
+ },
1570
+ {
1571
+ "epoch": 3.92,
1572
+ "learning_rate": 6.0787259615384616e-05,
1573
+ "loss": 3.411,
1574
+ "step": 26100
1575
+ },
1576
+ {
1577
+ "epoch": 3.94,
1578
+ "learning_rate": 6.0637019230769235e-05,
1579
+ "loss": 3.4315,
1580
+ "step": 26200
1581
+ },
1582
+ {
1583
+ "epoch": 3.95,
1584
+ "learning_rate": 6.048677884615385e-05,
1585
+ "loss": 3.3696,
1586
+ "step": 26300
1587
+ },
1588
+ {
1589
+ "epoch": 3.97,
1590
+ "learning_rate": 6.033653846153846e-05,
1591
+ "loss": 3.4611,
1592
+ "step": 26400
1593
+ },
1594
+ {
1595
+ "epoch": 3.98,
1596
+ "learning_rate": 6.018629807692308e-05,
1597
+ "loss": 3.4147,
1598
+ "step": 26500
1599
+ },
1600
+ {
1601
+ "epoch": 4.0,
1602
+ "learning_rate": 6.003605769230769e-05,
1603
+ "loss": 3.3671,
1604
+ "step": 26600
1605
+ },
1606
+ {
1607
+ "epoch": 4.01,
1608
+ "learning_rate": 5.9885817307692316e-05,
1609
+ "loss": 3.2922,
1610
+ "step": 26700
1611
+ },
1612
+ {
1613
+ "epoch": 4.03,
1614
+ "learning_rate": 5.973557692307693e-05,
1615
+ "loss": 3.4404,
1616
+ "step": 26800
1617
+ },
1618
+ {
1619
+ "epoch": 4.04,
1620
+ "learning_rate": 5.958533653846155e-05,
1621
+ "loss": 3.2936,
1622
+ "step": 26900
1623
+ },
1624
+ {
1625
+ "epoch": 4.06,
1626
+ "learning_rate": 5.943509615384616e-05,
1627
+ "loss": 3.3042,
1628
+ "step": 27000
1629
+ },
1630
+ {
1631
+ "epoch": 4.07,
1632
+ "learning_rate": 5.928485576923077e-05,
1633
+ "loss": 3.3803,
1634
+ "step": 27100
1635
+ },
1636
+ {
1637
+ "epoch": 4.09,
1638
+ "learning_rate": 5.913461538461539e-05,
1639
+ "loss": 3.3067,
1640
+ "step": 27200
1641
+ },
1642
+ {
1643
+ "epoch": 4.1,
1644
+ "learning_rate": 5.8984375e-05,
1645
+ "loss": 3.2929,
1646
+ "step": 27300
1647
+ },
1648
+ {
1649
+ "epoch": 4.12,
1650
+ "learning_rate": 5.8834134615384614e-05,
1651
+ "loss": 3.3419,
1652
+ "step": 27400
1653
+ },
1654
+ {
1655
+ "epoch": 4.13,
1656
+ "learning_rate": 5.868389423076923e-05,
1657
+ "loss": 3.3165,
1658
+ "step": 27500
1659
+ },
1660
+ {
1661
+ "epoch": 4.15,
1662
+ "learning_rate": 5.8533653846153845e-05,
1663
+ "loss": 3.3961,
1664
+ "step": 27600
1665
+ },
1666
+ {
1667
+ "epoch": 4.16,
1668
+ "learning_rate": 5.8383413461538464e-05,
1669
+ "loss": 3.3305,
1670
+ "step": 27700
1671
+ },
1672
+ {
1673
+ "epoch": 4.18,
1674
+ "learning_rate": 5.8233173076923076e-05,
1675
+ "loss": 3.3619,
1676
+ "step": 27800
1677
+ },
1678
+ {
1679
+ "epoch": 4.19,
1680
+ "learning_rate": 5.808293269230769e-05,
1681
+ "loss": 3.3138,
1682
+ "step": 27900
1683
+ },
1684
+ {
1685
+ "epoch": 4.21,
1686
+ "learning_rate": 5.7932692307692314e-05,
1687
+ "loss": 3.3557,
1688
+ "step": 28000
1689
+ },
1690
+ {
1691
+ "epoch": 4.22,
1692
+ "learning_rate": 5.778245192307693e-05,
1693
+ "loss": 3.4086,
1694
+ "step": 28100
1695
+ },
1696
+ {
1697
+ "epoch": 4.24,
1698
+ "learning_rate": 5.7632211538461545e-05,
1699
+ "loss": 3.2462,
1700
+ "step": 28200
1701
+ },
1702
+ {
1703
+ "epoch": 4.25,
1704
+ "learning_rate": 5.748197115384616e-05,
1705
+ "loss": 3.3991,
1706
+ "step": 28300
1707
+ },
1708
+ {
1709
+ "epoch": 4.27,
1710
+ "learning_rate": 5.7331730769230776e-05,
1711
+ "loss": 3.3424,
1712
+ "step": 28400
1713
+ },
1714
+ {
1715
+ "epoch": 4.28,
1716
+ "learning_rate": 5.718149038461539e-05,
1717
+ "loss": 3.345,
1718
+ "step": 28500
1719
+ },
1720
+ {
1721
+ "epoch": 4.3,
1722
+ "learning_rate": 5.703125e-05,
1723
+ "loss": 3.3057,
1724
+ "step": 28600
1725
+ },
1726
+ {
1727
+ "epoch": 4.31,
1728
+ "learning_rate": 5.688100961538462e-05,
1729
+ "loss": 3.339,
1730
+ "step": 28700
1731
+ },
1732
+ {
1733
+ "epoch": 4.33,
1734
+ "learning_rate": 5.673076923076923e-05,
1735
+ "loss": 3.3355,
1736
+ "step": 28800
1737
+ },
1738
+ {
1739
+ "epoch": 4.34,
1740
+ "learning_rate": 5.6580528846153844e-05,
1741
+ "loss": 3.2835,
1742
+ "step": 28900
1743
+ },
1744
+ {
1745
+ "epoch": 4.36,
1746
+ "learning_rate": 5.643028846153846e-05,
1747
+ "loss": 3.1553,
1748
+ "step": 29000
1749
+ },
1750
+ {
1751
+ "epoch": 4.37,
1752
+ "learning_rate": 5.6280048076923075e-05,
1753
+ "loss": 3.3456,
1754
+ "step": 29100
1755
+ },
1756
+ {
1757
+ "epoch": 4.39,
1758
+ "learning_rate": 5.612980769230769e-05,
1759
+ "loss": 3.2839,
1760
+ "step": 29200
1761
+ },
1762
+ {
1763
+ "epoch": 4.4,
1764
+ "learning_rate": 5.597956730769231e-05,
1765
+ "loss": 3.3465,
1766
+ "step": 29300
1767
+ },
1768
+ {
1769
+ "epoch": 4.42,
1770
+ "learning_rate": 5.582932692307693e-05,
1771
+ "loss": 3.3,
1772
+ "step": 29400
1773
+ },
1774
+ {
1775
+ "epoch": 4.43,
1776
+ "learning_rate": 5.5679086538461544e-05,
1777
+ "loss": 3.395,
1778
+ "step": 29500
1779
+ },
1780
+ {
1781
+ "epoch": 4.45,
1782
+ "learning_rate": 5.552884615384616e-05,
1783
+ "loss": 3.3677,
1784
+ "step": 29600
1785
+ },
1786
+ {
1787
+ "epoch": 4.46,
1788
+ "learning_rate": 5.5378605769230775e-05,
1789
+ "loss": 3.2638,
1790
+ "step": 29700
1791
+ },
1792
+ {
1793
+ "epoch": 4.48,
1794
+ "learning_rate": 5.522836538461539e-05,
1795
+ "loss": 3.2915,
1796
+ "step": 29800
1797
+ },
1798
+ {
1799
+ "epoch": 4.49,
1800
+ "learning_rate": 5.5078125000000006e-05,
1801
+ "loss": 3.3927,
1802
+ "step": 29900
1803
+ },
1804
+ {
1805
+ "epoch": 4.51,
1806
+ "learning_rate": 5.492788461538462e-05,
1807
+ "loss": 3.2758,
1808
+ "step": 30000
1809
+ },
1810
+ {
1811
+ "epoch": 4.52,
1812
+ "learning_rate": 5.477764423076923e-05,
1813
+ "loss": 3.2846,
1814
+ "step": 30100
1815
+ },
1816
+ {
1817
+ "epoch": 4.54,
1818
+ "learning_rate": 5.462740384615385e-05,
1819
+ "loss": 3.2502,
1820
+ "step": 30200
1821
+ },
1822
+ {
1823
+ "epoch": 4.55,
1824
+ "learning_rate": 5.447716346153846e-05,
1825
+ "loss": 3.345,
1826
+ "step": 30300
1827
+ },
1828
+ {
1829
+ "epoch": 4.57,
1830
+ "learning_rate": 5.432692307692307e-05,
1831
+ "loss": 3.3753,
1832
+ "step": 30400
1833
+ },
1834
+ {
1835
+ "epoch": 4.58,
1836
+ "learning_rate": 5.417668269230769e-05,
1837
+ "loss": 3.2171,
1838
+ "step": 30500
1839
+ },
1840
+ {
1841
+ "epoch": 4.6,
1842
+ "learning_rate": 5.402644230769232e-05,
1843
+ "loss": 3.2815,
1844
+ "step": 30600
1845
+ },
1846
+ {
1847
+ "epoch": 4.61,
1848
+ "learning_rate": 5.387620192307693e-05,
1849
+ "loss": 3.3164,
1850
+ "step": 30700
1851
+ },
1852
+ {
1853
+ "epoch": 4.63,
1854
+ "learning_rate": 5.372596153846154e-05,
1855
+ "loss": 3.3076,
1856
+ "step": 30800
1857
+ },
1858
+ {
1859
+ "epoch": 4.64,
1860
+ "learning_rate": 5.357572115384616e-05,
1861
+ "loss": 3.3597,
1862
+ "step": 30900
1863
+ },
1864
+ {
1865
+ "epoch": 4.66,
1866
+ "learning_rate": 5.342548076923077e-05,
1867
+ "loss": 3.2528,
1868
+ "step": 31000
1869
+ },
1870
+ {
1871
+ "epoch": 4.67,
1872
+ "learning_rate": 5.3275240384615385e-05,
1873
+ "loss": 3.3317,
1874
+ "step": 31100
1875
+ },
1876
+ {
1877
+ "epoch": 4.69,
1878
+ "learning_rate": 5.3125000000000004e-05,
1879
+ "loss": 3.4087,
1880
+ "step": 31200
1881
+ },
1882
+ {
1883
+ "epoch": 4.7,
1884
+ "learning_rate": 5.2974759615384616e-05,
1885
+ "loss": 3.3444,
1886
+ "step": 31300
1887
+ },
1888
+ {
1889
+ "epoch": 4.72,
1890
+ "learning_rate": 5.2824519230769235e-05,
1891
+ "loss": 3.3282,
1892
+ "step": 31400
1893
+ },
1894
+ {
1895
+ "epoch": 4.73,
1896
+ "learning_rate": 5.267427884615385e-05,
1897
+ "loss": 3.2904,
1898
+ "step": 31500
1899
+ },
1900
+ {
1901
+ "epoch": 4.75,
1902
+ "learning_rate": 5.252403846153846e-05,
1903
+ "loss": 3.3006,
1904
+ "step": 31600
1905
+ },
1906
+ {
1907
+ "epoch": 4.76,
1908
+ "learning_rate": 5.237379807692308e-05,
1909
+ "loss": 3.3654,
1910
+ "step": 31700
1911
+ },
1912
+ {
1913
+ "epoch": 4.78,
1914
+ "learning_rate": 5.222355769230769e-05,
1915
+ "loss": 3.2973,
1916
+ "step": 31800
1917
+ },
1918
+ {
1919
+ "epoch": 4.79,
1920
+ "learning_rate": 5.2073317307692316e-05,
1921
+ "loss": 3.3668,
1922
+ "step": 31900
1923
+ },
1924
+ {
1925
+ "epoch": 4.81,
1926
+ "learning_rate": 5.192307692307693e-05,
1927
+ "loss": 3.3186,
1928
+ "step": 32000
1929
+ },
1930
+ {
1931
+ "epoch": 4.82,
1932
+ "learning_rate": 5.177283653846155e-05,
1933
+ "loss": 3.1989,
1934
+ "step": 32100
1935
+ },
1936
+ {
1937
+ "epoch": 4.84,
1938
+ "learning_rate": 5.162259615384616e-05,
1939
+ "loss": 3.3046,
1940
+ "step": 32200
1941
+ },
1942
+ {
1943
+ "epoch": 4.85,
1944
+ "learning_rate": 5.147235576923077e-05,
1945
+ "loss": 3.2191,
1946
+ "step": 32300
1947
+ },
1948
+ {
1949
+ "epoch": 4.87,
1950
+ "learning_rate": 5.132211538461539e-05,
1951
+ "loss": 3.2571,
1952
+ "step": 32400
1953
+ },
1954
+ {
1955
+ "epoch": 4.88,
1956
+ "learning_rate": 5.1171875e-05,
1957
+ "loss": 3.2072,
1958
+ "step": 32500
1959
+ },
1960
+ {
1961
+ "epoch": 4.9,
1962
+ "learning_rate": 5.1021634615384614e-05,
1963
+ "loss": 3.322,
1964
+ "step": 32600
1965
+ },
1966
+ {
1967
+ "epoch": 4.91,
1968
+ "learning_rate": 5.087139423076923e-05,
1969
+ "loss": 3.299,
1970
+ "step": 32700
1971
+ },
1972
+ {
1973
+ "epoch": 4.93,
1974
+ "learning_rate": 5.0721153846153845e-05,
1975
+ "loss": 3.2845,
1976
+ "step": 32800
1977
+ },
1978
+ {
1979
+ "epoch": 4.94,
1980
+ "learning_rate": 5.057091346153846e-05,
1981
+ "loss": 3.2909,
1982
+ "step": 32900
1983
+ },
1984
+ {
1985
+ "epoch": 4.96,
1986
+ "learning_rate": 5.0420673076923076e-05,
1987
+ "loss": 3.3085,
1988
+ "step": 33000
1989
+ },
1990
+ {
1991
+ "epoch": 4.97,
1992
+ "learning_rate": 5.027043269230769e-05,
1993
+ "loss": 3.3532,
1994
+ "step": 33100
1995
+ },
1996
+ {
1997
+ "epoch": 4.99,
1998
+ "learning_rate": 5.0120192307692314e-05,
1999
+ "loss": 3.2598,
2000
+ "step": 33200
2001
+ },
2002
+ {
2003
+ "epoch": 5.0,
2004
+ "learning_rate": 4.9969951923076926e-05,
2005
+ "loss": 3.259,
2006
+ "step": 33300
2007
+ },
2008
+ {
2009
+ "epoch": 5.02,
2010
+ "learning_rate": 4.981971153846154e-05,
2011
+ "loss": 3.194,
2012
+ "step": 33400
2013
+ },
2014
+ {
2015
+ "epoch": 5.03,
2016
+ "learning_rate": 4.966947115384616e-05,
2017
+ "loss": 3.202,
2018
+ "step": 33500
2019
+ },
2020
+ {
2021
+ "epoch": 5.05,
2022
+ "learning_rate": 4.9519230769230776e-05,
2023
+ "loss": 3.1681,
2024
+ "step": 33600
2025
+ },
2026
+ {
2027
+ "epoch": 5.06,
2028
+ "learning_rate": 4.936899038461539e-05,
2029
+ "loss": 3.256,
2030
+ "step": 33700
2031
+ },
2032
+ {
2033
+ "epoch": 5.08,
2034
+ "learning_rate": 4.921875e-05,
2035
+ "loss": 3.2368,
2036
+ "step": 33800
2037
+ },
2038
+ {
2039
+ "epoch": 5.09,
2040
+ "learning_rate": 4.906850961538462e-05,
2041
+ "loss": 3.3036,
2042
+ "step": 33900
2043
+ },
2044
+ {
2045
+ "epoch": 5.11,
2046
+ "learning_rate": 4.891826923076923e-05,
2047
+ "loss": 3.2212,
2048
+ "step": 34000
2049
+ },
2050
+ {
2051
+ "epoch": 5.12,
2052
+ "learning_rate": 4.8768028846153843e-05,
2053
+ "loss": 3.3221,
2054
+ "step": 34100
2055
+ },
2056
+ {
2057
+ "epoch": 5.14,
2058
+ "learning_rate": 4.861778846153847e-05,
2059
+ "loss": 3.2859,
2060
+ "step": 34200
2061
+ },
2062
+ {
2063
+ "epoch": 5.15,
2064
+ "learning_rate": 4.846754807692308e-05,
2065
+ "loss": 3.1449,
2066
+ "step": 34300
2067
+ },
2068
+ {
2069
+ "epoch": 5.17,
2070
+ "learning_rate": 4.8317307692307693e-05,
2071
+ "loss": 3.2694,
2072
+ "step": 34400
2073
+ },
2074
+ {
2075
+ "epoch": 5.18,
2076
+ "learning_rate": 4.816706730769231e-05,
2077
+ "loss": 3.2395,
2078
+ "step": 34500
2079
+ },
2080
+ {
2081
+ "epoch": 5.2,
2082
+ "learning_rate": 4.8016826923076924e-05,
2083
+ "loss": 3.2135,
2084
+ "step": 34600
2085
+ },
2086
+ {
2087
+ "epoch": 5.21,
2088
+ "learning_rate": 4.7866586538461537e-05,
2089
+ "loss": 3.2219,
2090
+ "step": 34700
2091
+ },
2092
+ {
2093
+ "epoch": 5.23,
2094
+ "learning_rate": 4.7716346153846155e-05,
2095
+ "loss": 3.2907,
2096
+ "step": 34800
2097
+ },
2098
+ {
2099
+ "epoch": 5.24,
2100
+ "learning_rate": 4.7566105769230774e-05,
2101
+ "loss": 3.3046,
2102
+ "step": 34900
2103
+ },
2104
+ {
2105
+ "epoch": 5.26,
2106
+ "learning_rate": 4.7415865384615386e-05,
2107
+ "loss": 3.2011,
2108
+ "step": 35000
2109
+ },
2110
+ {
2111
+ "epoch": 5.27,
2112
+ "learning_rate": 4.7265625000000005e-05,
2113
+ "loss": 3.2394,
2114
+ "step": 35100
2115
+ },
2116
+ {
2117
+ "epoch": 5.29,
2118
+ "learning_rate": 4.711538461538462e-05,
2119
+ "loss": 3.2004,
2120
+ "step": 35200
2121
+ },
2122
+ {
2123
+ "epoch": 5.3,
2124
+ "learning_rate": 4.696514423076923e-05,
2125
+ "loss": 3.2539,
2126
+ "step": 35300
2127
+ },
2128
+ {
2129
+ "epoch": 5.32,
2130
+ "learning_rate": 4.681490384615385e-05,
2131
+ "loss": 3.3304,
2132
+ "step": 35400
2133
+ },
2134
+ {
2135
+ "epoch": 5.33,
2136
+ "learning_rate": 4.666466346153847e-05,
2137
+ "loss": 3.2046,
2138
+ "step": 35500
2139
+ },
2140
+ {
2141
+ "epoch": 5.35,
2142
+ "learning_rate": 4.651442307692308e-05,
2143
+ "loss": 3.3181,
2144
+ "step": 35600
2145
+ },
2146
+ {
2147
+ "epoch": 5.36,
2148
+ "learning_rate": 4.636418269230769e-05,
2149
+ "loss": 3.1461,
2150
+ "step": 35700
2151
+ },
2152
+ {
2153
+ "epoch": 5.38,
2154
+ "learning_rate": 4.621394230769231e-05,
2155
+ "loss": 3.2141,
2156
+ "step": 35800
2157
+ },
2158
+ {
2159
+ "epoch": 5.39,
2160
+ "learning_rate": 4.606370192307692e-05,
2161
+ "loss": 3.2212,
2162
+ "step": 35900
2163
+ },
2164
+ {
2165
+ "epoch": 5.41,
2166
+ "learning_rate": 4.591346153846154e-05,
2167
+ "loss": 3.2862,
2168
+ "step": 36000
2169
+ },
2170
+ {
2171
+ "epoch": 5.42,
2172
+ "learning_rate": 4.576322115384616e-05,
2173
+ "loss": 3.1912,
2174
+ "step": 36100
2175
+ },
2176
+ {
2177
+ "epoch": 5.44,
2178
+ "learning_rate": 4.561298076923077e-05,
2179
+ "loss": 3.3004,
2180
+ "step": 36200
2181
+ },
2182
+ {
2183
+ "epoch": 5.45,
2184
+ "learning_rate": 4.5462740384615385e-05,
2185
+ "loss": 3.1604,
2186
+ "step": 36300
2187
+ },
2188
+ {
2189
+ "epoch": 5.47,
2190
+ "learning_rate": 4.5312500000000004e-05,
2191
+ "loss": 3.1868,
2192
+ "step": 36400
2193
+ },
2194
+ {
2195
+ "epoch": 5.48,
2196
+ "learning_rate": 4.5162259615384616e-05,
2197
+ "loss": 3.2882,
2198
+ "step": 36500
2199
+ },
2200
+ {
2201
+ "epoch": 5.5,
2202
+ "learning_rate": 4.501201923076923e-05,
2203
+ "loss": 3.2211,
2204
+ "step": 36600
2205
+ },
2206
+ {
2207
+ "epoch": 5.51,
2208
+ "learning_rate": 4.486177884615385e-05,
2209
+ "loss": 3.2298,
2210
+ "step": 36700
2211
+ },
2212
+ {
2213
+ "epoch": 5.53,
2214
+ "learning_rate": 4.4711538461538466e-05,
2215
+ "loss": 3.2108,
2216
+ "step": 36800
2217
+ },
2218
+ {
2219
+ "epoch": 5.54,
2220
+ "learning_rate": 4.456129807692308e-05,
2221
+ "loss": 3.2887,
2222
+ "step": 36900
2223
+ },
2224
+ {
2225
+ "epoch": 5.56,
2226
+ "learning_rate": 4.44110576923077e-05,
2227
+ "loss": 3.1887,
2228
+ "step": 37000
2229
+ },
2230
+ {
2231
+ "epoch": 5.57,
2232
+ "learning_rate": 4.426081730769231e-05,
2233
+ "loss": 3.3018,
2234
+ "step": 37100
2235
+ },
2236
+ {
2237
+ "epoch": 5.59,
2238
+ "learning_rate": 4.411057692307692e-05,
2239
+ "loss": 3.2316,
2240
+ "step": 37200
2241
+ },
2242
+ {
2243
+ "epoch": 5.6,
2244
+ "learning_rate": 4.396033653846154e-05,
2245
+ "loss": 3.3247,
2246
+ "step": 37300
2247
+ },
2248
+ {
2249
+ "epoch": 5.62,
2250
+ "learning_rate": 4.381009615384616e-05,
2251
+ "loss": 3.2524,
2252
+ "step": 37400
2253
+ },
2254
+ {
2255
+ "epoch": 5.63,
2256
+ "learning_rate": 4.365985576923077e-05,
2257
+ "loss": 3.4009,
2258
+ "step": 37500
2259
+ },
2260
+ {
2261
+ "epoch": 5.65,
2262
+ "learning_rate": 4.350961538461539e-05,
2263
+ "loss": 3.1606,
2264
+ "step": 37600
2265
+ },
2266
+ {
2267
+ "epoch": 5.66,
2268
+ "learning_rate": 4.3359375e-05,
2269
+ "loss": 3.3663,
2270
+ "step": 37700
2271
+ },
2272
+ {
2273
+ "epoch": 5.68,
2274
+ "learning_rate": 4.3209134615384614e-05,
2275
+ "loss": 3.2013,
2276
+ "step": 37800
2277
+ },
2278
+ {
2279
+ "epoch": 5.69,
2280
+ "learning_rate": 4.305889423076923e-05,
2281
+ "loss": 3.2427,
2282
+ "step": 37900
2283
+ },
2284
+ {
2285
+ "epoch": 5.71,
2286
+ "learning_rate": 4.2908653846153845e-05,
2287
+ "loss": 3.2398,
2288
+ "step": 38000
2289
+ },
2290
+ {
2291
+ "epoch": 5.72,
2292
+ "learning_rate": 4.2758413461538464e-05,
2293
+ "loss": 3.293,
2294
+ "step": 38100
2295
+ },
2296
+ {
2297
+ "epoch": 5.74,
2298
+ "learning_rate": 4.260817307692308e-05,
2299
+ "loss": 3.2233,
2300
+ "step": 38200
2301
+ },
2302
+ {
2303
+ "epoch": 5.75,
2304
+ "learning_rate": 4.2457932692307695e-05,
2305
+ "loss": 3.2299,
2306
+ "step": 38300
2307
+ },
2308
+ {
2309
+ "epoch": 5.77,
2310
+ "learning_rate": 4.230769230769231e-05,
2311
+ "loss": 3.1866,
2312
+ "step": 38400
2313
+ },
2314
+ {
2315
+ "epoch": 5.78,
2316
+ "learning_rate": 4.2157451923076926e-05,
2317
+ "loss": 3.2959,
2318
+ "step": 38500
2319
+ },
2320
+ {
2321
+ "epoch": 5.8,
2322
+ "learning_rate": 4.200721153846154e-05,
2323
+ "loss": 3.2326,
2324
+ "step": 38600
2325
+ },
2326
+ {
2327
+ "epoch": 5.81,
2328
+ "learning_rate": 4.185697115384616e-05,
2329
+ "loss": 3.2325,
2330
+ "step": 38700
2331
+ },
2332
+ {
2333
+ "epoch": 5.83,
2334
+ "learning_rate": 4.1706730769230776e-05,
2335
+ "loss": 3.312,
2336
+ "step": 38800
2337
+ },
2338
+ {
2339
+ "epoch": 5.84,
2340
+ "learning_rate": 4.155649038461539e-05,
2341
+ "loss": 3.2872,
2342
+ "step": 38900
2343
+ },
2344
+ {
2345
+ "epoch": 5.86,
2346
+ "learning_rate": 4.140625e-05,
2347
+ "loss": 3.3211,
2348
+ "step": 39000
2349
+ },
2350
+ {
2351
+ "epoch": 5.87,
2352
+ "learning_rate": 4.125600961538462e-05,
2353
+ "loss": 3.267,
2354
+ "step": 39100
2355
+ },
2356
+ {
2357
+ "epoch": 5.89,
2358
+ "learning_rate": 4.110576923076923e-05,
2359
+ "loss": 3.2381,
2360
+ "step": 39200
2361
+ },
2362
+ {
2363
+ "epoch": 5.9,
2364
+ "learning_rate": 4.095552884615384e-05,
2365
+ "loss": 3.1223,
2366
+ "step": 39300
2367
+ },
2368
+ {
2369
+ "epoch": 5.92,
2370
+ "learning_rate": 4.080528846153847e-05,
2371
+ "loss": 3.1428,
2372
+ "step": 39400
2373
+ },
2374
+ {
2375
+ "epoch": 5.93,
2376
+ "learning_rate": 4.065504807692308e-05,
2377
+ "loss": 3.3539,
2378
+ "step": 39500
2379
+ },
2380
+ {
2381
+ "epoch": 5.95,
2382
+ "learning_rate": 4.050480769230769e-05,
2383
+ "loss": 3.1743,
2384
+ "step": 39600
2385
+ },
2386
+ {
2387
+ "epoch": 5.96,
2388
+ "learning_rate": 4.035456730769231e-05,
2389
+ "loss": 3.1944,
2390
+ "step": 39700
2391
+ },
2392
+ {
2393
+ "epoch": 5.98,
2394
+ "learning_rate": 4.0204326923076924e-05,
2395
+ "loss": 3.2237,
2396
+ "step": 39800
2397
+ },
2398
+ {
2399
+ "epoch": 5.99,
2400
+ "learning_rate": 4.0054086538461536e-05,
2401
+ "loss": 3.3603,
2402
+ "step": 39900
2403
+ },
2404
+ {
2405
+ "epoch": 6.01,
2406
+ "learning_rate": 3.9903846153846155e-05,
2407
+ "loss": 3.1751,
2408
+ "step": 40000
2409
+ },
2410
+ {
2411
+ "epoch": 6.02,
2412
+ "learning_rate": 3.9753605769230774e-05,
2413
+ "loss": 3.1382,
2414
+ "step": 40100
2415
+ },
2416
+ {
2417
+ "epoch": 6.04,
2418
+ "learning_rate": 3.9603365384615386e-05,
2419
+ "loss": 3.2931,
2420
+ "step": 40200
2421
+ },
2422
+ {
2423
+ "epoch": 6.05,
2424
+ "learning_rate": 3.9453125000000005e-05,
2425
+ "loss": 3.1821,
2426
+ "step": 40300
2427
+ },
2428
+ {
2429
+ "epoch": 6.07,
2430
+ "learning_rate": 3.930288461538462e-05,
2431
+ "loss": 3.159,
2432
+ "step": 40400
2433
+ },
2434
+ {
2435
+ "epoch": 6.08,
2436
+ "learning_rate": 3.915264423076923e-05,
2437
+ "loss": 3.1653,
2438
+ "step": 40500
2439
+ },
2440
+ {
2441
+ "epoch": 6.1,
2442
+ "learning_rate": 3.900240384615385e-05,
2443
+ "loss": 3.2043,
2444
+ "step": 40600
2445
+ },
2446
+ {
2447
+ "epoch": 6.11,
2448
+ "learning_rate": 3.885216346153847e-05,
2449
+ "loss": 3.1382,
2450
+ "step": 40700
2451
+ },
2452
+ {
2453
+ "epoch": 6.13,
2454
+ "learning_rate": 3.870192307692308e-05,
2455
+ "loss": 3.1522,
2456
+ "step": 40800
2457
+ },
2458
+ {
2459
+ "epoch": 6.14,
2460
+ "learning_rate": 3.855168269230769e-05,
2461
+ "loss": 3.1957,
2462
+ "step": 40900
2463
+ },
2464
+ {
2465
+ "epoch": 6.16,
2466
+ "learning_rate": 3.840144230769231e-05,
2467
+ "loss": 3.1041,
2468
+ "step": 41000
2469
+ },
2470
+ {
2471
+ "epoch": 6.17,
2472
+ "learning_rate": 3.825120192307692e-05,
2473
+ "loss": 3.1838,
2474
+ "step": 41100
2475
+ },
2476
+ {
2477
+ "epoch": 6.19,
2478
+ "learning_rate": 3.810096153846154e-05,
2479
+ "loss": 3.1468,
2480
+ "step": 41200
2481
+ },
2482
+ {
2483
+ "epoch": 6.2,
2484
+ "learning_rate": 3.795072115384616e-05,
2485
+ "loss": 3.1075,
2486
+ "step": 41300
2487
+ },
2488
+ {
2489
+ "epoch": 6.22,
2490
+ "learning_rate": 3.780048076923077e-05,
2491
+ "loss": 3.1594,
2492
+ "step": 41400
2493
+ },
2494
+ {
2495
+ "epoch": 6.23,
2496
+ "learning_rate": 3.7650240384615385e-05,
2497
+ "loss": 3.1844,
2498
+ "step": 41500
2499
+ },
2500
+ {
2501
+ "epoch": 6.25,
2502
+ "learning_rate": 3.7500000000000003e-05,
2503
+ "loss": 3.3313,
2504
+ "step": 41600
2505
+ },
2506
+ {
2507
+ "epoch": 6.27,
2508
+ "learning_rate": 3.7349759615384616e-05,
2509
+ "loss": 3.1786,
2510
+ "step": 41700
2511
+ },
2512
+ {
2513
+ "epoch": 6.28,
2514
+ "learning_rate": 3.719951923076923e-05,
2515
+ "loss": 3.2375,
2516
+ "step": 41800
2517
+ },
2518
+ {
2519
+ "epoch": 6.3,
2520
+ "learning_rate": 3.704927884615385e-05,
2521
+ "loss": 3.1206,
2522
+ "step": 41900
2523
+ },
2524
+ {
2525
+ "epoch": 6.31,
2526
+ "learning_rate": 3.6899038461538466e-05,
2527
+ "loss": 3.1919,
2528
+ "step": 42000
2529
+ },
2530
+ {
2531
+ "epoch": 6.33,
2532
+ "learning_rate": 3.674879807692308e-05,
2533
+ "loss": 3.2956,
2534
+ "step": 42100
2535
+ },
2536
+ {
2537
+ "epoch": 6.34,
2538
+ "learning_rate": 3.6598557692307697e-05,
2539
+ "loss": 3.2287,
2540
+ "step": 42200
2541
+ },
2542
+ {
2543
+ "epoch": 6.36,
2544
+ "learning_rate": 3.644831730769231e-05,
2545
+ "loss": 3.1453,
2546
+ "step": 42300
2547
+ },
2548
+ {
2549
+ "epoch": 6.37,
2550
+ "learning_rate": 3.629807692307692e-05,
2551
+ "loss": 3.1535,
2552
+ "step": 42400
2553
+ },
2554
+ {
2555
+ "epoch": 6.39,
2556
+ "learning_rate": 3.614783653846154e-05,
2557
+ "loss": 3.165,
2558
+ "step": 42500
2559
+ },
2560
+ {
2561
+ "epoch": 6.4,
2562
+ "learning_rate": 3.599759615384616e-05,
2563
+ "loss": 3.1838,
2564
+ "step": 42600
2565
+ },
2566
+ {
2567
+ "epoch": 6.42,
2568
+ "learning_rate": 3.584735576923077e-05,
2569
+ "loss": 3.1737,
2570
+ "step": 42700
2571
+ },
2572
+ {
2573
+ "epoch": 6.43,
2574
+ "learning_rate": 3.569711538461539e-05,
2575
+ "loss": 3.1649,
2576
+ "step": 42800
2577
+ },
2578
+ {
2579
+ "epoch": 6.45,
2580
+ "learning_rate": 3.5546875e-05,
2581
+ "loss": 3.103,
2582
+ "step": 42900
2583
+ },
2584
+ {
2585
+ "epoch": 6.46,
2586
+ "learning_rate": 3.5396634615384614e-05,
2587
+ "loss": 3.1786,
2588
+ "step": 43000
2589
+ },
2590
+ {
2591
+ "epoch": 6.48,
2592
+ "learning_rate": 3.524639423076923e-05,
2593
+ "loss": 3.2883,
2594
+ "step": 43100
2595
+ },
2596
+ {
2597
+ "epoch": 6.49,
2598
+ "learning_rate": 3.5096153846153845e-05,
2599
+ "loss": 3.2017,
2600
+ "step": 43200
2601
+ },
2602
+ {
2603
+ "epoch": 6.51,
2604
+ "learning_rate": 3.4945913461538464e-05,
2605
+ "loss": 3.2765,
2606
+ "step": 43300
2607
+ },
2608
+ {
2609
+ "epoch": 6.52,
2610
+ "learning_rate": 3.479567307692308e-05,
2611
+ "loss": 3.1713,
2612
+ "step": 43400
2613
+ },
2614
+ {
2615
+ "epoch": 6.54,
2616
+ "learning_rate": 3.4645432692307695e-05,
2617
+ "loss": 3.1993,
2618
+ "step": 43500
2619
+ },
2620
+ {
2621
+ "epoch": 6.55,
2622
+ "learning_rate": 3.449519230769231e-05,
2623
+ "loss": 3.2256,
2624
+ "step": 43600
2625
+ },
2626
+ {
2627
+ "epoch": 6.57,
2628
+ "learning_rate": 3.4344951923076926e-05,
2629
+ "loss": 3.2242,
2630
+ "step": 43700
2631
+ },
2632
+ {
2633
+ "epoch": 6.58,
2634
+ "learning_rate": 3.419471153846154e-05,
2635
+ "loss": 3.256,
2636
+ "step": 43800
2637
+ },
2638
+ {
2639
+ "epoch": 6.6,
2640
+ "learning_rate": 3.404447115384616e-05,
2641
+ "loss": 3.1709,
2642
+ "step": 43900
2643
+ },
2644
+ {
2645
+ "epoch": 6.61,
2646
+ "learning_rate": 3.3894230769230776e-05,
2647
+ "loss": 3.1652,
2648
+ "step": 44000
2649
+ },
2650
+ {
2651
+ "epoch": 6.63,
2652
+ "learning_rate": 3.374399038461539e-05,
2653
+ "loss": 3.0949,
2654
+ "step": 44100
2655
+ },
2656
+ {
2657
+ "epoch": 6.64,
2658
+ "learning_rate": 3.359375e-05,
2659
+ "loss": 3.2357,
2660
+ "step": 44200
2661
+ },
2662
+ {
2663
+ "epoch": 6.66,
2664
+ "learning_rate": 3.344350961538462e-05,
2665
+ "loss": 3.1377,
2666
+ "step": 44300
2667
+ },
2668
+ {
2669
+ "epoch": 6.67,
2670
+ "learning_rate": 3.329326923076923e-05,
2671
+ "loss": 3.1905,
2672
+ "step": 44400
2673
+ },
2674
+ {
2675
+ "epoch": 6.69,
2676
+ "learning_rate": 3.314302884615384e-05,
2677
+ "loss": 3.1792,
2678
+ "step": 44500
2679
+ },
2680
+ {
2681
+ "epoch": 6.7,
2682
+ "learning_rate": 3.299278846153846e-05,
2683
+ "loss": 3.1817,
2684
+ "step": 44600
2685
+ },
2686
+ {
2687
+ "epoch": 6.72,
2688
+ "learning_rate": 3.284254807692308e-05,
2689
+ "loss": 3.1305,
2690
+ "step": 44700
2691
+ },
2692
+ {
2693
+ "epoch": 6.73,
2694
+ "learning_rate": 3.269230769230769e-05,
2695
+ "loss": 3.1406,
2696
+ "step": 44800
2697
+ },
2698
+ {
2699
+ "epoch": 6.75,
2700
+ "learning_rate": 3.254206730769231e-05,
2701
+ "loss": 3.2035,
2702
+ "step": 44900
2703
+ },
2704
+ {
2705
+ "epoch": 6.76,
2706
+ "learning_rate": 3.2391826923076924e-05,
2707
+ "loss": 3.115,
2708
+ "step": 45000
2709
+ },
2710
+ {
2711
+ "epoch": 6.78,
2712
+ "learning_rate": 3.2241586538461536e-05,
2713
+ "loss": 3.1376,
2714
+ "step": 45100
2715
+ },
2716
+ {
2717
+ "epoch": 6.79,
2718
+ "learning_rate": 3.2091346153846155e-05,
2719
+ "loss": 3.1234,
2720
+ "step": 45200
2721
+ },
2722
+ {
2723
+ "epoch": 6.81,
2724
+ "learning_rate": 3.1941105769230774e-05,
2725
+ "loss": 3.1036,
2726
+ "step": 45300
2727
+ },
2728
+ {
2729
+ "epoch": 6.82,
2730
+ "learning_rate": 3.1790865384615386e-05,
2731
+ "loss": 3.2273,
2732
+ "step": 45400
2733
+ },
2734
+ {
2735
+ "epoch": 6.84,
2736
+ "learning_rate": 3.1640625e-05,
2737
+ "loss": 3.1999,
2738
+ "step": 45500
2739
+ },
2740
+ {
2741
+ "epoch": 6.85,
2742
+ "learning_rate": 3.149038461538462e-05,
2743
+ "loss": 3.1843,
2744
+ "step": 45600
2745
+ },
2746
+ {
2747
+ "epoch": 6.87,
2748
+ "learning_rate": 3.134014423076923e-05,
2749
+ "loss": 3.1522,
2750
+ "step": 45700
2751
+ },
2752
+ {
2753
+ "epoch": 6.88,
2754
+ "learning_rate": 3.118990384615385e-05,
2755
+ "loss": 3.1722,
2756
+ "step": 45800
2757
+ },
2758
+ {
2759
+ "epoch": 6.9,
2760
+ "learning_rate": 3.103966346153847e-05,
2761
+ "loss": 3.2638,
2762
+ "step": 45900
2763
+ },
2764
+ {
2765
+ "epoch": 6.91,
2766
+ "learning_rate": 3.088942307692308e-05,
2767
+ "loss": 3.1483,
2768
+ "step": 46000
2769
+ },
2770
+ {
2771
+ "epoch": 6.93,
2772
+ "learning_rate": 3.073918269230769e-05,
2773
+ "loss": 3.2081,
2774
+ "step": 46100
2775
+ },
2776
+ {
2777
+ "epoch": 6.94,
2778
+ "learning_rate": 3.058894230769231e-05,
2779
+ "loss": 3.2939,
2780
+ "step": 46200
2781
+ },
2782
+ {
2783
+ "epoch": 6.96,
2784
+ "learning_rate": 3.0438701923076922e-05,
2785
+ "loss": 3.0819,
2786
+ "step": 46300
2787
+ },
2788
+ {
2789
+ "epoch": 6.97,
2790
+ "learning_rate": 3.0288461538461538e-05,
2791
+ "loss": 3.1972,
2792
+ "step": 46400
2793
+ },
2794
+ {
2795
+ "epoch": 6.99,
2796
+ "learning_rate": 3.0138221153846157e-05,
2797
+ "loss": 3.1313,
2798
+ "step": 46500
2799
+ },
2800
+ {
2801
+ "epoch": 7.0,
2802
+ "learning_rate": 2.9987980769230772e-05,
2803
+ "loss": 3.2346,
2804
+ "step": 46600
2805
+ },
2806
+ {
2807
+ "epoch": 7.02,
2808
+ "learning_rate": 2.9837740384615388e-05,
2809
+ "loss": 3.0856,
2810
+ "step": 46700
2811
+ },
2812
+ {
2813
+ "epoch": 7.03,
2814
+ "learning_rate": 2.96875e-05,
2815
+ "loss": 3.1688,
2816
+ "step": 46800
2817
+ },
2818
+ {
2819
+ "epoch": 7.05,
2820
+ "learning_rate": 2.9537259615384615e-05,
2821
+ "loss": 3.0855,
2822
+ "step": 46900
2823
+ },
2824
+ {
2825
+ "epoch": 7.06,
2826
+ "learning_rate": 2.938701923076923e-05,
2827
+ "loss": 3.107,
2828
+ "step": 47000
2829
+ },
2830
+ {
2831
+ "epoch": 7.08,
2832
+ "learning_rate": 2.9236778846153846e-05,
2833
+ "loss": 3.1544,
2834
+ "step": 47100
2835
+ },
2836
+ {
2837
+ "epoch": 7.09,
2838
+ "learning_rate": 2.9086538461538465e-05,
2839
+ "loss": 3.1732,
2840
+ "step": 47200
2841
+ },
2842
+ {
2843
+ "epoch": 7.11,
2844
+ "learning_rate": 2.893629807692308e-05,
2845
+ "loss": 3.1477,
2846
+ "step": 47300
2847
+ },
2848
+ {
2849
+ "epoch": 7.12,
2850
+ "learning_rate": 2.8786057692307693e-05,
2851
+ "loss": 3.137,
2852
+ "step": 47400
2853
+ },
2854
+ {
2855
+ "epoch": 7.14,
2856
+ "learning_rate": 2.863581730769231e-05,
2857
+ "loss": 3.1548,
2858
+ "step": 47500
2859
+ },
2860
+ {
2861
+ "epoch": 7.15,
2862
+ "learning_rate": 2.8485576923076924e-05,
2863
+ "loss": 3.2163,
2864
+ "step": 47600
2865
+ },
2866
+ {
2867
+ "epoch": 7.17,
2868
+ "learning_rate": 2.8335336538461536e-05,
2869
+ "loss": 3.1692,
2870
+ "step": 47700
2871
+ },
2872
+ {
2873
+ "epoch": 7.18,
2874
+ "learning_rate": 2.818509615384616e-05,
2875
+ "loss": 3.1894,
2876
+ "step": 47800
2877
+ },
2878
+ {
2879
+ "epoch": 7.2,
2880
+ "learning_rate": 2.8034855769230774e-05,
2881
+ "loss": 3.1581,
2882
+ "step": 47900
2883
+ },
2884
+ {
2885
+ "epoch": 7.21,
2886
+ "learning_rate": 2.7884615384615386e-05,
2887
+ "loss": 3.1314,
2888
+ "step": 48000
2889
+ },
2890
+ {
2891
+ "epoch": 7.23,
2892
+ "learning_rate": 2.7734375e-05,
2893
+ "loss": 3.1914,
2894
+ "step": 48100
2895
+ },
2896
+ {
2897
+ "epoch": 7.24,
2898
+ "learning_rate": 2.7584134615384617e-05,
2899
+ "loss": 3.193,
2900
+ "step": 48200
2901
+ },
2902
+ {
2903
+ "epoch": 7.26,
2904
+ "learning_rate": 2.743389423076923e-05,
2905
+ "loss": 3.1188,
2906
+ "step": 48300
2907
+ },
2908
+ {
2909
+ "epoch": 7.27,
2910
+ "learning_rate": 2.7283653846153845e-05,
2911
+ "loss": 3.1413,
2912
+ "step": 48400
2913
+ },
2914
+ {
2915
+ "epoch": 7.29,
2916
+ "learning_rate": 2.7133413461538464e-05,
2917
+ "loss": 3.1699,
2918
+ "step": 48500
2919
+ },
2920
+ {
2921
+ "epoch": 7.3,
2922
+ "learning_rate": 2.698317307692308e-05,
2923
+ "loss": 3.1479,
2924
+ "step": 48600
2925
+ },
2926
+ {
2927
+ "epoch": 7.32,
2928
+ "learning_rate": 2.6832932692307695e-05,
2929
+ "loss": 3.1354,
2930
+ "step": 48700
2931
+ },
2932
+ {
2933
+ "epoch": 7.33,
2934
+ "learning_rate": 2.668269230769231e-05,
2935
+ "loss": 3.1386,
2936
+ "step": 48800
2937
+ },
2938
+ {
2939
+ "epoch": 7.35,
2940
+ "learning_rate": 2.6532451923076922e-05,
2941
+ "loss": 3.1308,
2942
+ "step": 48900
2943
+ },
2944
+ {
2945
+ "epoch": 7.36,
2946
+ "learning_rate": 2.6382211538461538e-05,
2947
+ "loss": 3.1955,
2948
+ "step": 49000
2949
+ },
2950
+ {
2951
+ "epoch": 7.38,
2952
+ "learning_rate": 2.6231971153846157e-05,
2953
+ "loss": 3.0473,
2954
+ "step": 49100
2955
+ },
2956
+ {
2957
+ "epoch": 7.39,
2958
+ "learning_rate": 2.6081730769230772e-05,
2959
+ "loss": 3.1117,
2960
+ "step": 49200
2961
+ },
2962
+ {
2963
+ "epoch": 7.41,
2964
+ "learning_rate": 2.5931490384615388e-05,
2965
+ "loss": 3.127,
2966
+ "step": 49300
2967
+ },
2968
+ {
2969
+ "epoch": 7.42,
2970
+ "learning_rate": 2.578125e-05,
2971
+ "loss": 3.1794,
2972
+ "step": 49400
2973
+ },
2974
+ {
2975
+ "epoch": 7.44,
2976
+ "learning_rate": 2.5631009615384615e-05,
2977
+ "loss": 3.1297,
2978
+ "step": 49500
2979
+ },
2980
+ {
2981
+ "epoch": 7.45,
2982
+ "learning_rate": 2.548076923076923e-05,
2983
+ "loss": 3.204,
2984
+ "step": 49600
2985
+ },
2986
+ {
2987
+ "epoch": 7.47,
2988
+ "learning_rate": 2.5330528846153846e-05,
2989
+ "loss": 3.1469,
2990
+ "step": 49700
2991
+ },
2992
+ {
2993
+ "epoch": 7.48,
2994
+ "learning_rate": 2.5180288461538465e-05,
2995
+ "loss": 3.1524,
2996
+ "step": 49800
2997
+ },
2998
+ {
2999
+ "epoch": 7.5,
3000
+ "learning_rate": 2.503004807692308e-05,
3001
+ "loss": 3.1529,
3002
+ "step": 49900
3003
+ },
3004
+ {
3005
+ "epoch": 7.51,
3006
+ "learning_rate": 2.4879807692307693e-05,
3007
+ "loss": 3.209,
3008
+ "step": 50000
3009
+ },
3010
+ {
3011
+ "epoch": 7.53,
3012
+ "learning_rate": 2.472956730769231e-05,
3013
+ "loss": 3.1885,
3014
+ "step": 50100
3015
+ },
3016
+ {
3017
+ "epoch": 7.54,
3018
+ "learning_rate": 2.4579326923076924e-05,
3019
+ "loss": 3.1388,
3020
+ "step": 50200
3021
+ },
3022
+ {
3023
+ "epoch": 7.56,
3024
+ "learning_rate": 2.442908653846154e-05,
3025
+ "loss": 3.163,
3026
+ "step": 50300
3027
+ },
3028
+ {
3029
+ "epoch": 7.57,
3030
+ "learning_rate": 2.4278846153846155e-05,
3031
+ "loss": 3.1048,
3032
+ "step": 50400
3033
+ },
3034
+ {
3035
+ "epoch": 7.59,
3036
+ "learning_rate": 2.412860576923077e-05,
3037
+ "loss": 3.1563,
3038
+ "step": 50500
3039
+ },
3040
+ {
3041
+ "epoch": 7.6,
3042
+ "learning_rate": 2.3978365384615386e-05,
3043
+ "loss": 3.0561,
3044
+ "step": 50600
3045
+ },
3046
+ {
3047
+ "epoch": 7.62,
3048
+ "learning_rate": 2.3828125e-05,
3049
+ "loss": 3.1133,
3050
+ "step": 50700
3051
+ },
3052
+ {
3053
+ "epoch": 7.63,
3054
+ "learning_rate": 2.3677884615384617e-05,
3055
+ "loss": 3.0987,
3056
+ "step": 50800
3057
+ },
3058
+ {
3059
+ "epoch": 7.65,
3060
+ "learning_rate": 2.3527644230769233e-05,
3061
+ "loss": 3.1504,
3062
+ "step": 50900
3063
+ },
3064
+ {
3065
+ "epoch": 7.66,
3066
+ "learning_rate": 2.3377403846153848e-05,
3067
+ "loss": 3.057,
3068
+ "step": 51000
3069
+ },
3070
+ {
3071
+ "epoch": 7.68,
3072
+ "learning_rate": 2.3227163461538464e-05,
3073
+ "loss": 3.1533,
3074
+ "step": 51100
3075
+ },
3076
+ {
3077
+ "epoch": 7.69,
3078
+ "learning_rate": 2.307692307692308e-05,
3079
+ "loss": 3.1509,
3080
+ "step": 51200
3081
+ },
3082
+ {
3083
+ "epoch": 7.71,
3084
+ "learning_rate": 2.2926682692307695e-05,
3085
+ "loss": 3.1098,
3086
+ "step": 51300
3087
+ },
3088
+ {
3089
+ "epoch": 7.72,
3090
+ "learning_rate": 2.2776442307692307e-05,
3091
+ "loss": 3.1403,
3092
+ "step": 51400
3093
+ },
3094
+ {
3095
+ "epoch": 7.74,
3096
+ "learning_rate": 2.2626201923076922e-05,
3097
+ "loss": 3.175,
3098
+ "step": 51500
3099
+ },
3100
+ {
3101
+ "epoch": 7.75,
3102
+ "learning_rate": 2.247596153846154e-05,
3103
+ "loss": 3.082,
3104
+ "step": 51600
3105
+ },
3106
+ {
3107
+ "epoch": 7.77,
3108
+ "learning_rate": 2.2325721153846153e-05,
3109
+ "loss": 3.1393,
3110
+ "step": 51700
3111
+ },
3112
+ {
3113
+ "epoch": 7.78,
3114
+ "learning_rate": 2.217548076923077e-05,
3115
+ "loss": 3.0962,
3116
+ "step": 51800
3117
+ },
3118
+ {
3119
+ "epoch": 7.8,
3120
+ "learning_rate": 2.2025240384615388e-05,
3121
+ "loss": 3.1229,
3122
+ "step": 51900
3123
+ },
3124
+ {
3125
+ "epoch": 7.81,
3126
+ "learning_rate": 2.1875e-05,
3127
+ "loss": 3.1898,
3128
+ "step": 52000
3129
+ },
3130
+ {
3131
+ "epoch": 7.83,
3132
+ "learning_rate": 2.1724759615384615e-05,
3133
+ "loss": 3.1181,
3134
+ "step": 52100
3135
+ },
3136
+ {
3137
+ "epoch": 7.84,
3138
+ "learning_rate": 2.1574519230769234e-05,
3139
+ "loss": 3.1294,
3140
+ "step": 52200
3141
+ },
3142
+ {
3143
+ "epoch": 7.86,
3144
+ "learning_rate": 2.1424278846153846e-05,
3145
+ "loss": 3.1118,
3146
+ "step": 52300
3147
+ },
3148
+ {
3149
+ "epoch": 7.87,
3150
+ "learning_rate": 2.1274038461538462e-05,
3151
+ "loss": 3.2093,
3152
+ "step": 52400
3153
+ },
3154
+ {
3155
+ "epoch": 7.89,
3156
+ "learning_rate": 2.112379807692308e-05,
3157
+ "loss": 3.0968,
3158
+ "step": 52500
3159
+ },
3160
+ {
3161
+ "epoch": 7.9,
3162
+ "learning_rate": 2.0973557692307693e-05,
3163
+ "loss": 3.1043,
3164
+ "step": 52600
3165
+ },
3166
+ {
3167
+ "epoch": 7.92,
3168
+ "learning_rate": 2.082331730769231e-05,
3169
+ "loss": 3.1901,
3170
+ "step": 52700
3171
+ },
3172
+ {
3173
+ "epoch": 7.93,
3174
+ "learning_rate": 2.0673076923076924e-05,
3175
+ "loss": 3.1718,
3176
+ "step": 52800
3177
+ },
3178
+ {
3179
+ "epoch": 7.95,
3180
+ "learning_rate": 2.052283653846154e-05,
3181
+ "loss": 3.1754,
3182
+ "step": 52900
3183
+ },
3184
+ {
3185
+ "epoch": 7.96,
3186
+ "learning_rate": 2.0372596153846155e-05,
3187
+ "loss": 3.1013,
3188
+ "step": 53000
3189
+ },
3190
+ {
3191
+ "epoch": 7.98,
3192
+ "learning_rate": 2.022235576923077e-05,
3193
+ "loss": 3.1273,
3194
+ "step": 53100
3195
+ },
3196
+ {
3197
+ "epoch": 7.99,
3198
+ "learning_rate": 2.0072115384615386e-05,
3199
+ "loss": 3.0944,
3200
+ "step": 53200
3201
+ },
3202
+ {
3203
+ "epoch": 8.01,
3204
+ "learning_rate": 1.9921875e-05,
3205
+ "loss": 3.1341,
3206
+ "step": 53300
3207
+ },
3208
+ {
3209
+ "epoch": 8.02,
3210
+ "learning_rate": 1.9771634615384617e-05,
3211
+ "loss": 3.1024,
3212
+ "step": 53400
3213
+ },
3214
+ {
3215
+ "epoch": 8.04,
3216
+ "learning_rate": 1.9621394230769232e-05,
3217
+ "loss": 3.2147,
3218
+ "step": 53500
3219
+ },
3220
+ {
3221
+ "epoch": 8.05,
3222
+ "learning_rate": 1.9471153846153848e-05,
3223
+ "loss": 3.1102,
3224
+ "step": 53600
3225
+ },
3226
+ {
3227
+ "epoch": 8.07,
3228
+ "learning_rate": 1.9320913461538463e-05,
3229
+ "loss": 3.2247,
3230
+ "step": 53700
3231
+ },
3232
+ {
3233
+ "epoch": 8.08,
3234
+ "learning_rate": 1.917067307692308e-05,
3235
+ "loss": 3.1304,
3236
+ "step": 53800
3237
+ },
3238
+ {
3239
+ "epoch": 8.1,
3240
+ "learning_rate": 1.9020432692307695e-05,
3241
+ "loss": 3.1136,
3242
+ "step": 53900
3243
+ },
3244
+ {
3245
+ "epoch": 8.11,
3246
+ "learning_rate": 1.8870192307692307e-05,
3247
+ "loss": 3.1116,
3248
+ "step": 54000
3249
+ },
3250
+ {
3251
+ "epoch": 8.13,
3252
+ "learning_rate": 1.8719951923076922e-05,
3253
+ "loss": 3.0719,
3254
+ "step": 54100
3255
+ },
3256
+ {
3257
+ "epoch": 8.14,
3258
+ "learning_rate": 1.856971153846154e-05,
3259
+ "loss": 2.9952,
3260
+ "step": 54200
3261
+ },
3262
+ {
3263
+ "epoch": 8.16,
3264
+ "learning_rate": 1.8419471153846153e-05,
3265
+ "loss": 3.1379,
3266
+ "step": 54300
3267
+ },
3268
+ {
3269
+ "epoch": 8.17,
3270
+ "learning_rate": 1.826923076923077e-05,
3271
+ "loss": 3.0153,
3272
+ "step": 54400
3273
+ },
3274
+ {
3275
+ "epoch": 8.19,
3276
+ "learning_rate": 1.8118990384615388e-05,
3277
+ "loss": 3.0276,
3278
+ "step": 54500
3279
+ },
3280
+ {
3281
+ "epoch": 8.2,
3282
+ "learning_rate": 1.796875e-05,
3283
+ "loss": 3.1095,
3284
+ "step": 54600
3285
+ },
3286
+ {
3287
+ "epoch": 8.22,
3288
+ "learning_rate": 1.7818509615384615e-05,
3289
+ "loss": 3.173,
3290
+ "step": 54700
3291
+ },
3292
+ {
3293
+ "epoch": 8.23,
3294
+ "learning_rate": 1.7668269230769234e-05,
3295
+ "loss": 3.0593,
3296
+ "step": 54800
3297
+ },
3298
+ {
3299
+ "epoch": 8.25,
3300
+ "learning_rate": 1.7518028846153846e-05,
3301
+ "loss": 3.2008,
3302
+ "step": 54900
3303
+ },
3304
+ {
3305
+ "epoch": 8.26,
3306
+ "learning_rate": 1.7367788461538462e-05,
3307
+ "loss": 3.0976,
3308
+ "step": 55000
3309
+ },
3310
+ {
3311
+ "epoch": 8.28,
3312
+ "learning_rate": 1.721754807692308e-05,
3313
+ "loss": 3.0239,
3314
+ "step": 55100
3315
+ },
3316
+ {
3317
+ "epoch": 8.29,
3318
+ "learning_rate": 1.7067307692307693e-05,
3319
+ "loss": 2.9954,
3320
+ "step": 55200
3321
+ },
3322
+ {
3323
+ "epoch": 8.31,
3324
+ "learning_rate": 1.6917067307692308e-05,
3325
+ "loss": 3.0959,
3326
+ "step": 55300
3327
+ },
3328
+ {
3329
+ "epoch": 8.32,
3330
+ "learning_rate": 1.6766826923076924e-05,
3331
+ "loss": 3.1744,
3332
+ "step": 55400
3333
+ },
3334
+ {
3335
+ "epoch": 8.34,
3336
+ "learning_rate": 1.661658653846154e-05,
3337
+ "loss": 3.1568,
3338
+ "step": 55500
3339
+ },
3340
+ {
3341
+ "epoch": 8.35,
3342
+ "learning_rate": 1.6466346153846155e-05,
3343
+ "loss": 3.2058,
3344
+ "step": 55600
3345
+ },
3346
+ {
3347
+ "epoch": 8.37,
3348
+ "learning_rate": 1.631610576923077e-05,
3349
+ "loss": 3.2486,
3350
+ "step": 55700
3351
+ },
3352
+ {
3353
+ "epoch": 8.38,
3354
+ "learning_rate": 1.6165865384615386e-05,
3355
+ "loss": 3.0785,
3356
+ "step": 55800
3357
+ },
3358
+ {
3359
+ "epoch": 8.4,
3360
+ "learning_rate": 1.6015625e-05,
3361
+ "loss": 3.1703,
3362
+ "step": 55900
3363
+ },
3364
+ {
3365
+ "epoch": 8.41,
3366
+ "learning_rate": 1.5865384615384617e-05,
3367
+ "loss": 3.0091,
3368
+ "step": 56000
3369
+ },
3370
+ {
3371
+ "epoch": 8.43,
3372
+ "learning_rate": 1.5715144230769232e-05,
3373
+ "loss": 2.9814,
3374
+ "step": 56100
3375
+ },
3376
+ {
3377
+ "epoch": 8.44,
3378
+ "learning_rate": 1.5564903846153848e-05,
3379
+ "loss": 3.1474,
3380
+ "step": 56200
3381
+ },
3382
+ {
3383
+ "epoch": 8.46,
3384
+ "learning_rate": 1.541466346153846e-05,
3385
+ "loss": 3.0801,
3386
+ "step": 56300
3387
+ },
3388
+ {
3389
+ "epoch": 8.47,
3390
+ "learning_rate": 1.526442307692308e-05,
3391
+ "loss": 3.0667,
3392
+ "step": 56400
3393
+ },
3394
+ {
3395
+ "epoch": 8.49,
3396
+ "learning_rate": 1.5114182692307693e-05,
3397
+ "loss": 3.1068,
3398
+ "step": 56500
3399
+ },
3400
+ {
3401
+ "epoch": 8.5,
3402
+ "learning_rate": 1.4963942307692308e-05,
3403
+ "loss": 3.1035,
3404
+ "step": 56600
3405
+ },
3406
+ {
3407
+ "epoch": 8.52,
3408
+ "learning_rate": 1.4813701923076922e-05,
3409
+ "loss": 3.1616,
3410
+ "step": 56700
3411
+ },
3412
+ {
3413
+ "epoch": 8.53,
3414
+ "learning_rate": 1.466346153846154e-05,
3415
+ "loss": 3.2153,
3416
+ "step": 56800
3417
+ },
3418
+ {
3419
+ "epoch": 8.55,
3420
+ "learning_rate": 1.4513221153846155e-05,
3421
+ "loss": 3.1312,
3422
+ "step": 56900
3423
+ },
3424
+ {
3425
+ "epoch": 8.56,
3426
+ "learning_rate": 1.4362980769230769e-05,
3427
+ "loss": 3.1062,
3428
+ "step": 57000
3429
+ },
3430
+ {
3431
+ "epoch": 8.58,
3432
+ "learning_rate": 1.4212740384615386e-05,
3433
+ "loss": 3.092,
3434
+ "step": 57100
3435
+ },
3436
+ {
3437
+ "epoch": 8.59,
3438
+ "learning_rate": 1.4062500000000001e-05,
3439
+ "loss": 3.1349,
3440
+ "step": 57200
3441
+ },
3442
+ {
3443
+ "epoch": 8.61,
3444
+ "learning_rate": 1.3912259615384615e-05,
3445
+ "loss": 3.1018,
3446
+ "step": 57300
3447
+ },
3448
+ {
3449
+ "epoch": 8.62,
3450
+ "learning_rate": 1.3762019230769232e-05,
3451
+ "loss": 3.078,
3452
+ "step": 57400
3453
+ },
3454
+ {
3455
+ "epoch": 8.64,
3456
+ "learning_rate": 1.3611778846153848e-05,
3457
+ "loss": 3.0693,
3458
+ "step": 57500
3459
+ },
3460
+ {
3461
+ "epoch": 8.65,
3462
+ "learning_rate": 1.3461538461538462e-05,
3463
+ "loss": 3.0487,
3464
+ "step": 57600
3465
+ },
3466
+ {
3467
+ "epoch": 8.67,
3468
+ "learning_rate": 1.3311298076923079e-05,
3469
+ "loss": 3.0893,
3470
+ "step": 57700
3471
+ },
3472
+ {
3473
+ "epoch": 8.68,
3474
+ "learning_rate": 1.3161057692307693e-05,
3475
+ "loss": 3.0998,
3476
+ "step": 57800
3477
+ },
3478
+ {
3479
+ "epoch": 8.7,
3480
+ "learning_rate": 1.3010817307692308e-05,
3481
+ "loss": 3.0287,
3482
+ "step": 57900
3483
+ },
3484
+ {
3485
+ "epoch": 8.71,
3486
+ "learning_rate": 1.2860576923076922e-05,
3487
+ "loss": 3.0558,
3488
+ "step": 58000
3489
+ },
3490
+ {
3491
+ "epoch": 8.73,
3492
+ "learning_rate": 1.271033653846154e-05,
3493
+ "loss": 3.1865,
3494
+ "step": 58100
3495
+ },
3496
+ {
3497
+ "epoch": 8.74,
3498
+ "learning_rate": 1.2560096153846155e-05,
3499
+ "loss": 3.1941,
3500
+ "step": 58200
3501
+ },
3502
+ {
3503
+ "epoch": 8.76,
3504
+ "learning_rate": 1.240985576923077e-05,
3505
+ "loss": 3.1259,
3506
+ "step": 58300
3507
+ },
3508
+ {
3509
+ "epoch": 8.77,
3510
+ "learning_rate": 1.2259615384615384e-05,
3511
+ "loss": 3.0836,
3512
+ "step": 58400
3513
+ },
3514
+ {
3515
+ "epoch": 8.79,
3516
+ "learning_rate": 1.2109375000000001e-05,
3517
+ "loss": 3.2193,
3518
+ "step": 58500
3519
+ },
3520
+ {
3521
+ "epoch": 8.8,
3522
+ "learning_rate": 1.1959134615384617e-05,
3523
+ "loss": 3.091,
3524
+ "step": 58600
3525
+ },
3526
+ {
3527
+ "epoch": 8.82,
3528
+ "learning_rate": 1.180889423076923e-05,
3529
+ "loss": 3.1139,
3530
+ "step": 58700
3531
+ },
3532
+ {
3533
+ "epoch": 8.83,
3534
+ "learning_rate": 1.1658653846153846e-05,
3535
+ "loss": 3.1585,
3536
+ "step": 58800
3537
+ },
3538
+ {
3539
+ "epoch": 8.85,
3540
+ "learning_rate": 1.1508413461538462e-05,
3541
+ "loss": 3.1098,
3542
+ "step": 58900
3543
+ },
3544
+ {
3545
+ "epoch": 8.86,
3546
+ "learning_rate": 1.1358173076923077e-05,
3547
+ "loss": 3.0185,
3548
+ "step": 59000
3549
+ },
3550
+ {
3551
+ "epoch": 8.88,
3552
+ "learning_rate": 1.1207932692307693e-05,
3553
+ "loss": 3.1064,
3554
+ "step": 59100
3555
+ },
3556
+ {
3557
+ "epoch": 8.89,
3558
+ "learning_rate": 1.1057692307692308e-05,
3559
+ "loss": 3.1722,
3560
+ "step": 59200
3561
+ },
3562
+ {
3563
+ "epoch": 8.91,
3564
+ "learning_rate": 1.0907451923076924e-05,
3565
+ "loss": 3.107,
3566
+ "step": 59300
3567
+ },
3568
+ {
3569
+ "epoch": 8.92,
3570
+ "learning_rate": 1.075721153846154e-05,
3571
+ "loss": 3.0859,
3572
+ "step": 59400
3573
+ },
3574
+ {
3575
+ "epoch": 8.94,
3576
+ "learning_rate": 1.0606971153846155e-05,
3577
+ "loss": 3.1676,
3578
+ "step": 59500
3579
+ },
3580
+ {
3581
+ "epoch": 8.95,
3582
+ "learning_rate": 1.045673076923077e-05,
3583
+ "loss": 3.1134,
3584
+ "step": 59600
3585
+ },
3586
+ {
3587
+ "epoch": 8.97,
3588
+ "learning_rate": 1.0306490384615384e-05,
3589
+ "loss": 3.1622,
3590
+ "step": 59700
3591
+ },
3592
+ {
3593
+ "epoch": 8.98,
3594
+ "learning_rate": 1.0156250000000001e-05,
3595
+ "loss": 3.1357,
3596
+ "step": 59800
3597
+ },
3598
+ {
3599
+ "epoch": 9.0,
3600
+ "learning_rate": 1.0006009615384617e-05,
3601
+ "loss": 3.1807,
3602
+ "step": 59900
3603
+ },
3604
+ {
3605
+ "epoch": 9.01,
3606
+ "learning_rate": 9.85576923076923e-06,
3607
+ "loss": 3.1431,
3608
+ "step": 60000
3609
+ },
3610
+ {
3611
+ "epoch": 9.03,
3612
+ "learning_rate": 9.705528846153846e-06,
3613
+ "loss": 3.2043,
3614
+ "step": 60100
3615
+ },
3616
+ {
3617
+ "epoch": 9.04,
3618
+ "learning_rate": 9.555288461538462e-06,
3619
+ "loss": 3.0665,
3620
+ "step": 60200
3621
+ },
3622
+ {
3623
+ "epoch": 9.06,
3624
+ "learning_rate": 9.405048076923077e-06,
3625
+ "loss": 3.016,
3626
+ "step": 60300
3627
+ },
3628
+ {
3629
+ "epoch": 9.07,
3630
+ "learning_rate": 9.254807692307693e-06,
3631
+ "loss": 3.0571,
3632
+ "step": 60400
3633
+ },
3634
+ {
3635
+ "epoch": 9.09,
3636
+ "learning_rate": 9.104567307692308e-06,
3637
+ "loss": 3.1035,
3638
+ "step": 60500
3639
+ },
3640
+ {
3641
+ "epoch": 9.1,
3642
+ "learning_rate": 8.954326923076924e-06,
3643
+ "loss": 3.1543,
3644
+ "step": 60600
3645
+ },
3646
+ {
3647
+ "epoch": 9.12,
3648
+ "learning_rate": 8.804086538461539e-06,
3649
+ "loss": 3.1839,
3650
+ "step": 60700
3651
+ },
3652
+ {
3653
+ "epoch": 9.13,
3654
+ "learning_rate": 8.653846153846155e-06,
3655
+ "loss": 3.1786,
3656
+ "step": 60800
3657
+ },
3658
+ {
3659
+ "epoch": 9.15,
3660
+ "learning_rate": 8.50360576923077e-06,
3661
+ "loss": 3.0707,
3662
+ "step": 60900
3663
+ },
3664
+ {
3665
+ "epoch": 9.16,
3666
+ "learning_rate": 8.353365384615384e-06,
3667
+ "loss": 3.052,
3668
+ "step": 61000
3669
+ },
3670
+ {
3671
+ "epoch": 9.18,
3672
+ "learning_rate": 8.203125000000001e-06,
3673
+ "loss": 3.0002,
3674
+ "step": 61100
3675
+ },
3676
+ {
3677
+ "epoch": 9.19,
3678
+ "learning_rate": 8.052884615384617e-06,
3679
+ "loss": 3.0621,
3680
+ "step": 61200
3681
+ },
3682
+ {
3683
+ "epoch": 9.21,
3684
+ "learning_rate": 7.90264423076923e-06,
3685
+ "loss": 3.1472,
3686
+ "step": 61300
3687
+ },
3688
+ {
3689
+ "epoch": 9.22,
3690
+ "learning_rate": 7.752403846153846e-06,
3691
+ "loss": 3.2139,
3692
+ "step": 61400
3693
+ },
3694
+ {
3695
+ "epoch": 9.24,
3696
+ "learning_rate": 7.6021634615384615e-06,
3697
+ "loss": 3.1535,
3698
+ "step": 61500
3699
+ },
3700
+ {
3701
+ "epoch": 9.25,
3702
+ "learning_rate": 7.451923076923077e-06,
3703
+ "loss": 3.0481,
3704
+ "step": 61600
3705
+ },
3706
+ {
3707
+ "epoch": 9.27,
3708
+ "learning_rate": 7.301682692307693e-06,
3709
+ "loss": 3.1044,
3710
+ "step": 61700
3711
+ },
3712
+ {
3713
+ "epoch": 9.28,
3714
+ "learning_rate": 7.151442307692307e-06,
3715
+ "loss": 3.1502,
3716
+ "step": 61800
3717
+ },
3718
+ {
3719
+ "epoch": 9.3,
3720
+ "learning_rate": 7.001201923076924e-06,
3721
+ "loss": 3.16,
3722
+ "step": 61900
3723
+ },
3724
+ {
3725
+ "epoch": 9.31,
3726
+ "learning_rate": 6.850961538461539e-06,
3727
+ "loss": 3.1364,
3728
+ "step": 62000
3729
+ },
3730
+ {
3731
+ "epoch": 9.33,
3732
+ "learning_rate": 6.700721153846154e-06,
3733
+ "loss": 3.1023,
3734
+ "step": 62100
3735
+ },
3736
+ {
3737
+ "epoch": 9.34,
3738
+ "learning_rate": 6.55048076923077e-06,
3739
+ "loss": 3.0382,
3740
+ "step": 62200
3741
+ },
3742
+ {
3743
+ "epoch": 9.36,
3744
+ "learning_rate": 6.400240384615385e-06,
3745
+ "loss": 3.087,
3746
+ "step": 62300
3747
+ },
3748
+ {
3749
+ "epoch": 9.38,
3750
+ "learning_rate": 6.25e-06,
3751
+ "loss": 3.1023,
3752
+ "step": 62400
3753
+ },
3754
+ {
3755
+ "epoch": 9.39,
3756
+ "learning_rate": 6.099759615384616e-06,
3757
+ "loss": 3.1571,
3758
+ "step": 62500
3759
+ },
3760
+ {
3761
+ "epoch": 9.41,
3762
+ "learning_rate": 5.9495192307692305e-06,
3763
+ "loss": 3.0847,
3764
+ "step": 62600
3765
+ },
3766
+ {
3767
+ "epoch": 9.42,
3768
+ "learning_rate": 5.799278846153847e-06,
3769
+ "loss": 3.0513,
3770
+ "step": 62700
3771
+ },
3772
+ {
3773
+ "epoch": 9.44,
3774
+ "learning_rate": 5.6490384615384615e-06,
3775
+ "loss": 3.1568,
3776
+ "step": 62800
3777
+ },
3778
+ {
3779
+ "epoch": 9.45,
3780
+ "learning_rate": 5.498798076923077e-06,
3781
+ "loss": 3.1487,
3782
+ "step": 62900
3783
+ },
3784
+ {
3785
+ "epoch": 9.47,
3786
+ "learning_rate": 5.3485576923076925e-06,
3787
+ "loss": 3.1173,
3788
+ "step": 63000
3789
+ },
3790
+ {
3791
+ "epoch": 9.48,
3792
+ "learning_rate": 5.198317307692308e-06,
3793
+ "loss": 3.0394,
3794
+ "step": 63100
3795
+ },
3796
+ {
3797
+ "epoch": 9.5,
3798
+ "learning_rate": 5.0480769230769235e-06,
3799
+ "loss": 3.0607,
3800
+ "step": 63200
3801
+ },
3802
+ {
3803
+ "epoch": 9.51,
3804
+ "learning_rate": 4.897836538461539e-06,
3805
+ "loss": 3.0811,
3806
+ "step": 63300
3807
+ },
3808
+ {
3809
+ "epoch": 9.53,
3810
+ "learning_rate": 4.747596153846154e-06,
3811
+ "loss": 3.1086,
3812
+ "step": 63400
3813
+ },
3814
+ {
3815
+ "epoch": 9.54,
3816
+ "learning_rate": 4.59735576923077e-06,
3817
+ "loss": 3.0946,
3818
+ "step": 63500
3819
+ },
3820
+ {
3821
+ "epoch": 9.56,
3822
+ "learning_rate": 4.447115384615385e-06,
3823
+ "loss": 3.0247,
3824
+ "step": 63600
3825
+ },
3826
+ {
3827
+ "epoch": 9.57,
3828
+ "learning_rate": 4.296875e-06,
3829
+ "loss": 3.1085,
3830
+ "step": 63700
3831
+ },
3832
+ {
3833
+ "epoch": 9.59,
3834
+ "learning_rate": 4.146634615384616e-06,
3835
+ "loss": 3.1925,
3836
+ "step": 63800
3837
+ },
3838
+ {
3839
+ "epoch": 9.6,
3840
+ "learning_rate": 3.9963942307692304e-06,
3841
+ "loss": 3.1547,
3842
+ "step": 63900
3843
+ },
3844
+ {
3845
+ "epoch": 9.62,
3846
+ "learning_rate": 3.846153846153847e-06,
3847
+ "loss": 3.0974,
3848
+ "step": 64000
3849
+ },
3850
+ {
3851
+ "epoch": 9.63,
3852
+ "learning_rate": 3.695913461538462e-06,
3853
+ "loss": 3.0615,
3854
+ "step": 64100
3855
+ },
3856
+ {
3857
+ "epoch": 9.65,
3858
+ "learning_rate": 3.545673076923077e-06,
3859
+ "loss": 3.1745,
3860
+ "step": 64200
3861
+ },
3862
+ {
3863
+ "epoch": 9.66,
3864
+ "learning_rate": 3.395432692307692e-06,
3865
+ "loss": 3.0591,
3866
+ "step": 64300
3867
+ },
3868
+ {
3869
+ "epoch": 9.68,
3870
+ "learning_rate": 3.245192307692308e-06,
3871
+ "loss": 3.0274,
3872
+ "step": 64400
3873
+ },
3874
+ {
3875
+ "epoch": 9.69,
3876
+ "learning_rate": 3.0949519230769235e-06,
3877
+ "loss": 3.0092,
3878
+ "step": 64500
3879
+ }
3880
+ ],
3881
+ "max_steps": 66560,
3882
+ "num_train_epochs": 10,
3883
+ "total_flos": 1.3030638807239885e+17,
3884
+ "trial_name": null,
3885
+ "trial_params": null
3886
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8866efc220db599e3a3ed5db22928994cec12c97ba7ad4e862cf5bc62e056eb
3
+ size 2607