Hugging Face's logo

longcld
/
t5_small_squad_trans_old

Text2Text Generation
Transformers
PyTorch
t5
Inference Endpoints
text-generation-inference
Model card Files Community
longcld commited on
Commit
0b6e5f5
•
1 Parent(s): f6b53e1

upload

Browse files
Files changed (11) hide show
  1. added_tokens.json +1 -0
  2. config.json +57 -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 +3496 -0
  11. training_args.bin +3 -0
added_tokens.json ADDED
@@ -0,0 +1 @@
 
1
+ {"<sep>": 250100, "<hl>": 250101}
config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "longcld/t5_small_qg_ae_hl",
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
+ "transformers_version": "4.8.2",
55
+ "use_cache": true,
56
+ "vocab_size": 250102
57
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45dd816488d955a75f9225148850bc042e833c2594d860d0d13cec905cab0af4
3
+ size 352532601
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7bdbd8880ab8a569d6cabe92fa3af779f4701e1962a5cf450122b95ad09a2852
3
+ size 688496379
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55a92dfbd437791422bfd494f290b3e2151e8626d2a7be4267a6b4ef8ee02da8
3
+ size 14503
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46faa9d4b6cf6d00295e66819e68396a182980d08003bd4d22c668d4d358480e
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 +1,3496 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 7.239126365054602,
5
+ "global_step": 58000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 9.987518721917126e-06,
13
+ "loss": 2.674,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 9.97503744383425e-06,
19
+ "loss": 2.5884,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.04,
24
+ "learning_rate": 9.962556165751374e-06,
25
+ "loss": 2.614,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 0.05,
30
+ "learning_rate": 9.950074887668498e-06,
31
+ "loss": 2.6229,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 0.06,
36
+ "learning_rate": 9.937593609585623e-06,
37
+ "loss": 2.6261,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 0.07,
42
+ "learning_rate": 9.925112331502746e-06,
43
+ "loss": 2.6,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 0.09,
48
+ "learning_rate": 9.912631053419871e-06,
49
+ "loss": 2.6558,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 0.1,
54
+ "learning_rate": 9.900149775336996e-06,
55
+ "loss": 2.5839,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 0.11,
60
+ "learning_rate": 9.88766849725412e-06,
61
+ "loss": 2.6585,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 0.12,
66
+ "learning_rate": 9.875187219171243e-06,
67
+ "loss": 2.5986,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 0.14,
72
+ "learning_rate": 9.862705941088368e-06,
73
+ "loss": 2.6437,
74
+ "step": 1100
75
+ },
76
+ {
77
+ "epoch": 0.15,
78
+ "learning_rate": 9.850224663005492e-06,
79
+ "loss": 2.5858,
80
+ "step": 1200
81
+ },
82
+ {
83
+ "epoch": 0.16,
84
+ "learning_rate": 9.837743384922617e-06,
85
+ "loss": 2.6142,
86
+ "step": 1300
87
+ },
88
+ {
89
+ "epoch": 0.17,
90
+ "learning_rate": 9.825262106839742e-06,
91
+ "loss": 2.5436,
92
+ "step": 1400
93
+ },
94
+ {
95
+ "epoch": 0.19,
96
+ "learning_rate": 9.812780828756865e-06,
97
+ "loss": 2.5879,
98
+ "step": 1500
99
+ },
100
+ {
101
+ "epoch": 0.2,
102
+ "learning_rate": 9.80029955067399e-06,
103
+ "loss": 2.6459,
104
+ "step": 1600
105
+ },
106
+ {
107
+ "epoch": 0.21,
108
+ "learning_rate": 9.787818272591114e-06,
109
+ "loss": 2.6185,
110
+ "step": 1700
111
+ },
112
+ {
113
+ "epoch": 0.22,
114
+ "learning_rate": 9.775336994508239e-06,
115
+ "loss": 2.6672,
116
+ "step": 1800
117
+ },
118
+ {
119
+ "epoch": 0.24,
120
+ "learning_rate": 9.762855716425362e-06,
121
+ "loss": 2.6541,
122
+ "step": 1900
123
+ },
124
+ {
125
+ "epoch": 0.25,
126
+ "learning_rate": 9.750374438342487e-06,
127
+ "loss": 2.6669,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 0.26,
132
+ "learning_rate": 9.737893160259612e-06,
133
+ "loss": 2.6323,
134
+ "step": 2100
135
+ },
136
+ {
137
+ "epoch": 0.27,
138
+ "learning_rate": 9.725411882176736e-06,
139
+ "loss": 2.6764,
140
+ "step": 2200
141
+ },
142
+ {
143
+ "epoch": 0.29,
144
+ "learning_rate": 9.712930604093859e-06,
145
+ "loss": 2.5925,
146
+ "step": 2300
147
+ },
148
+ {
149
+ "epoch": 0.3,
150
+ "learning_rate": 9.700449326010984e-06,
151
+ "loss": 2.5392,
152
+ "step": 2400
153
+ },
154
+ {
155
+ "epoch": 0.31,
156
+ "learning_rate": 9.68796804792811e-06,
157
+ "loss": 2.5843,
158
+ "step": 2500
159
+ },
160
+ {
161
+ "epoch": 0.32,
162
+ "learning_rate": 9.675486769845233e-06,
163
+ "loss": 2.6209,
164
+ "step": 2600
165
+ },
166
+ {
167
+ "epoch": 0.34,
168
+ "learning_rate": 9.663005491762358e-06,
169
+ "loss": 2.6575,
170
+ "step": 2700
171
+ },
172
+ {
173
+ "epoch": 0.35,
174
+ "learning_rate": 9.650524213679481e-06,
175
+ "loss": 2.6499,
176
+ "step": 2800
177
+ },
178
+ {
179
+ "epoch": 0.36,
180
+ "learning_rate": 9.638042935596606e-06,
181
+ "loss": 2.5762,
182
+ "step": 2900
183
+ },
184
+ {
185
+ "epoch": 0.37,
186
+ "learning_rate": 9.62556165751373e-06,
187
+ "loss": 2.5637,
188
+ "step": 3000
189
+ },
190
+ {
191
+ "epoch": 0.39,
192
+ "learning_rate": 9.613080379430855e-06,
193
+ "loss": 2.5902,
194
+ "step": 3100
195
+ },
196
+ {
197
+ "epoch": 0.4,
198
+ "learning_rate": 9.60059910134798e-06,
199
+ "loss": 2.5784,
200
+ "step": 3200
201
+ },
202
+ {
203
+ "epoch": 0.41,
204
+ "learning_rate": 9.588117823265103e-06,
205
+ "loss": 2.6246,
206
+ "step": 3300
207
+ },
208
+ {
209
+ "epoch": 0.42,
210
+ "learning_rate": 9.575636545182228e-06,
211
+ "loss": 2.6543,
212
+ "step": 3400
213
+ },
214
+ {
215
+ "epoch": 0.44,
216
+ "learning_rate": 9.563155267099352e-06,
217
+ "loss": 2.581,
218
+ "step": 3500
219
+ },
220
+ {
221
+ "epoch": 0.45,
222
+ "learning_rate": 9.550673989016475e-06,
223
+ "loss": 2.6039,
224
+ "step": 3600
225
+ },
226
+ {
227
+ "epoch": 0.46,
228
+ "learning_rate": 9.5381927109336e-06,
229
+ "loss": 2.5848,
230
+ "step": 3700
231
+ },
232
+ {
233
+ "epoch": 0.47,
234
+ "learning_rate": 9.525711432850725e-06,
235
+ "loss": 2.6026,
236
+ "step": 3800
237
+ },
238
+ {
239
+ "epoch": 0.49,
240
+ "learning_rate": 9.513230154767849e-06,
241
+ "loss": 2.6178,
242
+ "step": 3900
243
+ },
244
+ {
245
+ "epoch": 0.5,
246
+ "learning_rate": 9.500748876684974e-06,
247
+ "loss": 2.5908,
248
+ "step": 4000
249
+ },
250
+ {
251
+ "epoch": 0.51,
252
+ "learning_rate": 9.488267598602097e-06,
253
+ "loss": 2.599,
254
+ "step": 4100
255
+ },
256
+ {
257
+ "epoch": 0.52,
258
+ "learning_rate": 9.475786320519222e-06,
259
+ "loss": 2.5937,
260
+ "step": 4200
261
+ },
262
+ {
263
+ "epoch": 0.54,
264
+ "learning_rate": 9.463305042436346e-06,
265
+ "loss": 2.6038,
266
+ "step": 4300
267
+ },
268
+ {
269
+ "epoch": 0.55,
270
+ "learning_rate": 9.45082376435347e-06,
271
+ "loss": 2.5825,
272
+ "step": 4400
273
+ },
274
+ {
275
+ "epoch": 0.56,
276
+ "learning_rate": 9.438342486270596e-06,
277
+ "loss": 2.5668,
278
+ "step": 4500
279
+ },
280
+ {
281
+ "epoch": 0.57,
282
+ "learning_rate": 9.42586120818772e-06,
283
+ "loss": 2.6231,
284
+ "step": 4600
285
+ },
286
+ {
287
+ "epoch": 0.59,
288
+ "learning_rate": 9.413379930104844e-06,
289
+ "loss": 2.5955,
290
+ "step": 4700
291
+ },
292
+ {
293
+ "epoch": 0.6,
294
+ "learning_rate": 9.400898652021968e-06,
295
+ "loss": 2.6169,
296
+ "step": 4800
297
+ },
298
+ {
299
+ "epoch": 0.61,
300
+ "learning_rate": 9.388417373939091e-06,
301
+ "loss": 2.5865,
302
+ "step": 4900
303
+ },
304
+ {
305
+ "epoch": 0.62,
306
+ "learning_rate": 9.375936095856216e-06,
307
+ "loss": 2.5837,
308
+ "step": 5000
309
+ },
310
+ {
311
+ "epoch": 0.64,
312
+ "learning_rate": 9.363454817773341e-06,
313
+ "loss": 2.5994,
314
+ "step": 5100
315
+ },
316
+ {
317
+ "epoch": 0.65,
318
+ "learning_rate": 9.350973539690466e-06,
319
+ "loss": 2.6152,
320
+ "step": 5200
321
+ },
322
+ {
323
+ "epoch": 0.66,
324
+ "learning_rate": 9.33849226160759e-06,
325
+ "loss": 2.5968,
326
+ "step": 5300
327
+ },
328
+ {
329
+ "epoch": 0.67,
330
+ "learning_rate": 9.326010983524713e-06,
331
+ "loss": 2.5876,
332
+ "step": 5400
333
+ },
334
+ {
335
+ "epoch": 0.69,
336
+ "learning_rate": 9.313529705441838e-06,
337
+ "loss": 2.5356,
338
+ "step": 5500
339
+ },
340
+ {
341
+ "epoch": 0.7,
342
+ "learning_rate": 9.301048427358962e-06,
343
+ "loss": 2.6401,
344
+ "step": 5600
345
+ },
346
+ {
347
+ "epoch": 0.71,
348
+ "learning_rate": 9.288567149276087e-06,
349
+ "loss": 2.6072,
350
+ "step": 5700
351
+ },
352
+ {
353
+ "epoch": 0.72,
354
+ "learning_rate": 9.276085871193212e-06,
355
+ "loss": 2.6391,
356
+ "step": 5800
357
+ },
358
+ {
359
+ "epoch": 0.74,
360
+ "learning_rate": 9.263604593110335e-06,
361
+ "loss": 2.6492,
362
+ "step": 5900
363
+ },
364
+ {
365
+ "epoch": 0.75,
366
+ "learning_rate": 9.251123315027458e-06,
367
+ "loss": 2.5952,
368
+ "step": 6000
369
+ },
370
+ {
371
+ "epoch": 0.76,
372
+ "learning_rate": 9.238642036944584e-06,
373
+ "loss": 2.606,
374
+ "step": 6100
375
+ },
376
+ {
377
+ "epoch": 0.77,
378
+ "learning_rate": 9.226160758861709e-06,
379
+ "loss": 2.6346,
380
+ "step": 6200
381
+ },
382
+ {
383
+ "epoch": 0.79,
384
+ "learning_rate": 9.213679480778832e-06,
385
+ "loss": 2.6238,
386
+ "step": 6300
387
+ },
388
+ {
389
+ "epoch": 0.8,
390
+ "learning_rate": 9.201198202695957e-06,
391
+ "loss": 2.5924,
392
+ "step": 6400
393
+ },
394
+ {
395
+ "epoch": 0.81,
396
+ "learning_rate": 9.188716924613082e-06,
397
+ "loss": 2.6511,
398
+ "step": 6500
399
+ },
400
+ {
401
+ "epoch": 0.82,
402
+ "learning_rate": 9.176235646530206e-06,
403
+ "loss": 2.5397,
404
+ "step": 6600
405
+ },
406
+ {
407
+ "epoch": 0.84,
408
+ "learning_rate": 9.163754368447329e-06,
409
+ "loss": 2.6166,
410
+ "step": 6700
411
+ },
412
+ {
413
+ "epoch": 0.85,
414
+ "learning_rate": 9.151273090364454e-06,
415
+ "loss": 2.5976,
416
+ "step": 6800
417
+ },
418
+ {
419
+ "epoch": 0.86,
420
+ "learning_rate": 9.13879181228158e-06,
421
+ "loss": 2.5556,
422
+ "step": 6900
423
+ },
424
+ {
425
+ "epoch": 0.87,
426
+ "learning_rate": 9.126310534198703e-06,
427
+ "loss": 2.5568,
428
+ "step": 7000
429
+ },
430
+ {
431
+ "epoch": 0.89,
432
+ "learning_rate": 9.113829256115828e-06,
433
+ "loss": 2.5994,
434
+ "step": 7100
435
+ },
436
+ {
437
+ "epoch": 0.9,
438
+ "learning_rate": 9.101347978032951e-06,
439
+ "loss": 2.5802,
440
+ "step": 7200
441
+ },
442
+ {
443
+ "epoch": 0.91,
444
+ "learning_rate": 9.088866699950074e-06,
445
+ "loss": 2.5724,
446
+ "step": 7300
447
+ },
448
+ {
449
+ "epoch": 0.92,
450
+ "learning_rate": 9.0763854218672e-06,
451
+ "loss": 2.5317,
452
+ "step": 7400
453
+ },
454
+ {
455
+ "epoch": 0.94,
456
+ "learning_rate": 9.063904143784325e-06,
457
+ "loss": 2.6202,
458
+ "step": 7500
459
+ },
460
+ {
461
+ "epoch": 0.95,
462
+ "learning_rate": 9.051422865701448e-06,
463
+ "loss": 2.5479,
464
+ "step": 7600
465
+ },
466
+ {
467
+ "epoch": 0.96,
468
+ "learning_rate": 9.038941587618573e-06,
469
+ "loss": 2.5918,
470
+ "step": 7700
471
+ },
472
+ {
473
+ "epoch": 0.97,
474
+ "learning_rate": 9.026460309535696e-06,
475
+ "loss": 2.5706,
476
+ "step": 7800
477
+ },
478
+ {
479
+ "epoch": 0.99,
480
+ "learning_rate": 9.013979031452822e-06,
481
+ "loss": 2.5578,
482
+ "step": 7900
483
+ },
484
+ {
485
+ "epoch": 1.0,
486
+ "learning_rate": 9.001497753369945e-06,
487
+ "loss": 2.6592,
488
+ "step": 8000
489
+ },
490
+ {
491
+ "epoch": 1.01,
492
+ "learning_rate": 8.98901647528707e-06,
493
+ "loss": 2.6137,
494
+ "step": 8100
495
+ },
496
+ {
497
+ "epoch": 1.02,
498
+ "learning_rate": 8.976535197204195e-06,
499
+ "loss": 2.5963,
500
+ "step": 8200
501
+ },
502
+ {
503
+ "epoch": 1.04,
504
+ "learning_rate": 8.964053919121319e-06,
505
+ "loss": 2.5891,
506
+ "step": 8300
507
+ },
508
+ {
509
+ "epoch": 1.05,
510
+ "learning_rate": 8.951572641038444e-06,
511
+ "loss": 2.5719,
512
+ "step": 8400
513
+ },
514
+ {
515
+ "epoch": 1.06,
516
+ "learning_rate": 8.939091362955567e-06,
517
+ "loss": 2.5967,
518
+ "step": 8500
519
+ },
520
+ {
521
+ "epoch": 1.07,
522
+ "learning_rate": 8.926610084872692e-06,
523
+ "loss": 2.5656,
524
+ "step": 8600
525
+ },
526
+ {
527
+ "epoch": 1.09,
528
+ "learning_rate": 8.914128806789815e-06,
529
+ "loss": 2.6024,
530
+ "step": 8700
531
+ },
532
+ {
533
+ "epoch": 1.1,
534
+ "learning_rate": 8.90164752870694e-06,
535
+ "loss": 2.566,
536
+ "step": 8800
537
+ },
538
+ {
539
+ "epoch": 1.11,
540
+ "learning_rate": 8.889166250624066e-06,
541
+ "loss": 2.5757,
542
+ "step": 8900
543
+ },
544
+ {
545
+ "epoch": 1.12,
546
+ "learning_rate": 8.876684972541189e-06,
547
+ "loss": 2.5595,
548
+ "step": 9000
549
+ },
550
+ {
551
+ "epoch": 1.14,
552
+ "learning_rate": 8.864203694458312e-06,
553
+ "loss": 2.605,
554
+ "step": 9100
555
+ },
556
+ {
557
+ "epoch": 1.15,
558
+ "learning_rate": 8.851722416375437e-06,
559
+ "loss": 2.594,
560
+ "step": 9200
561
+ },
562
+ {
563
+ "epoch": 1.16,
564
+ "learning_rate": 8.839241138292561e-06,
565
+ "loss": 2.6241,
566
+ "step": 9300
567
+ },
568
+ {
569
+ "epoch": 1.17,
570
+ "learning_rate": 8.826759860209686e-06,
571
+ "loss": 2.6328,
572
+ "step": 9400
573
+ },
574
+ {
575
+ "epoch": 1.19,
576
+ "learning_rate": 8.814278582126811e-06,
577
+ "loss": 2.5948,
578
+ "step": 9500
579
+ },
580
+ {
581
+ "epoch": 1.2,
582
+ "learning_rate": 8.801797304043934e-06,
583
+ "loss": 2.5232,
584
+ "step": 9600
585
+ },
586
+ {
587
+ "epoch": 1.21,
588
+ "learning_rate": 8.78931602596106e-06,
589
+ "loss": 2.614,
590
+ "step": 9700
591
+ },
592
+ {
593
+ "epoch": 1.22,
594
+ "learning_rate": 8.776834747878183e-06,
595
+ "loss": 2.561,
596
+ "step": 9800
597
+ },
598
+ {
599
+ "epoch": 1.24,
600
+ "learning_rate": 8.764353469795308e-06,
601
+ "loss": 2.6071,
602
+ "step": 9900
603
+ },
604
+ {
605
+ "epoch": 1.25,
606
+ "learning_rate": 8.751872191712431e-06,
607
+ "loss": 2.5902,
608
+ "step": 10000
609
+ },
610
+ {
611
+ "epoch": 1.26,
612
+ "learning_rate": 8.739390913629556e-06,
613
+ "loss": 2.5653,
614
+ "step": 10100
615
+ },
616
+ {
617
+ "epoch": 1.27,
618
+ "learning_rate": 8.726909635546682e-06,
619
+ "loss": 2.5746,
620
+ "step": 10200
621
+ },
622
+ {
623
+ "epoch": 1.29,
624
+ "learning_rate": 8.714428357463805e-06,
625
+ "loss": 2.525,
626
+ "step": 10300
627
+ },
628
+ {
629
+ "epoch": 1.3,
630
+ "learning_rate": 8.701947079380928e-06,
631
+ "loss": 2.6418,
632
+ "step": 10400
633
+ },
634
+ {
635
+ "epoch": 1.31,
636
+ "learning_rate": 8.689465801298053e-06,
637
+ "loss": 2.6579,
638
+ "step": 10500
639
+ },
640
+ {
641
+ "epoch": 1.32,
642
+ "learning_rate": 8.676984523215179e-06,
643
+ "loss": 2.5815,
644
+ "step": 10600
645
+ },
646
+ {
647
+ "epoch": 1.34,
648
+ "learning_rate": 8.664503245132302e-06,
649
+ "loss": 2.6014,
650
+ "step": 10700
651
+ },
652
+ {
653
+ "epoch": 1.35,
654
+ "learning_rate": 8.652021967049427e-06,
655
+ "loss": 2.6233,
656
+ "step": 10800
657
+ },
658
+ {
659
+ "epoch": 1.36,
660
+ "learning_rate": 8.63954068896655e-06,
661
+ "loss": 2.5808,
662
+ "step": 10900
663
+ },
664
+ {
665
+ "epoch": 1.37,
666
+ "learning_rate": 8.627059410883675e-06,
667
+ "loss": 2.6013,
668
+ "step": 11000
669
+ },
670
+ {
671
+ "epoch": 1.39,
672
+ "learning_rate": 8.614578132800799e-06,
673
+ "loss": 2.5305,
674
+ "step": 11100
675
+ },
676
+ {
677
+ "epoch": 1.4,
678
+ "learning_rate": 8.602096854717924e-06,
679
+ "loss": 2.5726,
680
+ "step": 11200
681
+ },
682
+ {
683
+ "epoch": 1.41,
684
+ "learning_rate": 8.589615576635049e-06,
685
+ "loss": 2.5862,
686
+ "step": 11300
687
+ },
688
+ {
689
+ "epoch": 1.42,
690
+ "learning_rate": 8.577134298552172e-06,
691
+ "loss": 2.5728,
692
+ "step": 11400
693
+ },
694
+ {
695
+ "epoch": 1.44,
696
+ "learning_rate": 8.564653020469298e-06,
697
+ "loss": 2.5863,
698
+ "step": 11500
699
+ },
700
+ {
701
+ "epoch": 1.45,
702
+ "learning_rate": 8.552171742386421e-06,
703
+ "loss": 2.5121,
704
+ "step": 11600
705
+ },
706
+ {
707
+ "epoch": 1.46,
708
+ "learning_rate": 8.539690464303544e-06,
709
+ "loss": 2.583,
710
+ "step": 11700
711
+ },
712
+ {
713
+ "epoch": 1.47,
714
+ "learning_rate": 8.52720918622067e-06,
715
+ "loss": 2.6128,
716
+ "step": 11800
717
+ },
718
+ {
719
+ "epoch": 1.49,
720
+ "learning_rate": 8.514727908137794e-06,
721
+ "loss": 2.5568,
722
+ "step": 11900
723
+ },
724
+ {
725
+ "epoch": 1.5,
726
+ "learning_rate": 8.502246630054918e-06,
727
+ "loss": 2.5913,
728
+ "step": 12000
729
+ },
730
+ {
731
+ "epoch": 1.51,
732
+ "learning_rate": 8.489765351972043e-06,
733
+ "loss": 2.6139,
734
+ "step": 12100
735
+ },
736
+ {
737
+ "epoch": 1.52,
738
+ "learning_rate": 8.477284073889166e-06,
739
+ "loss": 2.5659,
740
+ "step": 12200
741
+ },
742
+ {
743
+ "epoch": 1.54,
744
+ "learning_rate": 8.464802795806291e-06,
745
+ "loss": 2.5676,
746
+ "step": 12300
747
+ },
748
+ {
749
+ "epoch": 1.55,
750
+ "learning_rate": 8.452321517723415e-06,
751
+ "loss": 2.5329,
752
+ "step": 12400
753
+ },
754
+ {
755
+ "epoch": 1.56,
756
+ "learning_rate": 8.43984023964054e-06,
757
+ "loss": 2.5653,
758
+ "step": 12500
759
+ },
760
+ {
761
+ "epoch": 1.57,
762
+ "learning_rate": 8.427358961557665e-06,
763
+ "loss": 2.6023,
764
+ "step": 12600
765
+ },
766
+ {
767
+ "epoch": 1.59,
768
+ "learning_rate": 8.414877683474788e-06,
769
+ "loss": 2.5692,
770
+ "step": 12700
771
+ },
772
+ {
773
+ "epoch": 1.6,
774
+ "learning_rate": 8.402396405391913e-06,
775
+ "loss": 2.5947,
776
+ "step": 12800
777
+ },
778
+ {
779
+ "epoch": 1.61,
780
+ "learning_rate": 8.389915127309037e-06,
781
+ "loss": 2.6291,
782
+ "step": 12900
783
+ },
784
+ {
785
+ "epoch": 1.62,
786
+ "learning_rate": 8.377433849226162e-06,
787
+ "loss": 2.5965,
788
+ "step": 13000
789
+ },
790
+ {
791
+ "epoch": 1.64,
792
+ "learning_rate": 8.364952571143285e-06,
793
+ "loss": 2.5433,
794
+ "step": 13100
795
+ },
796
+ {
797
+ "epoch": 1.65,
798
+ "learning_rate": 8.35247129306041e-06,
799
+ "loss": 2.5804,
800
+ "step": 13200
801
+ },
802
+ {
803
+ "epoch": 1.66,
804
+ "learning_rate": 8.339990014977535e-06,
805
+ "loss": 2.5296,
806
+ "step": 13300
807
+ },
808
+ {
809
+ "epoch": 1.67,
810
+ "learning_rate": 8.327508736894659e-06,
811
+ "loss": 2.6349,
812
+ "step": 13400
813
+ },
814
+ {
815
+ "epoch": 1.68,
816
+ "learning_rate": 8.315027458811782e-06,
817
+ "loss": 2.5424,
818
+ "step": 13500
819
+ },
820
+ {
821
+ "epoch": 1.7,
822
+ "learning_rate": 8.302546180728907e-06,
823
+ "loss": 2.6139,
824
+ "step": 13600
825
+ },
826
+ {
827
+ "epoch": 1.71,
828
+ "learning_rate": 8.29006490264603e-06,
829
+ "loss": 2.585,
830
+ "step": 13700
831
+ },
832
+ {
833
+ "epoch": 1.72,
834
+ "learning_rate": 8.277583624563156e-06,
835
+ "loss": 2.5734,
836
+ "step": 13800
837
+ },
838
+ {
839
+ "epoch": 1.73,
840
+ "learning_rate": 8.265102346480281e-06,
841
+ "loss": 2.591,
842
+ "step": 13900
843
+ },
844
+ {
845
+ "epoch": 1.75,
846
+ "learning_rate": 8.252621068397404e-06,
847
+ "loss": 2.6089,
848
+ "step": 14000
849
+ },
850
+ {
851
+ "epoch": 1.76,
852
+ "learning_rate": 8.24013979031453e-06,
853
+ "loss": 2.6216,
854
+ "step": 14100
855
+ },
856
+ {
857
+ "epoch": 1.77,
858
+ "learning_rate": 8.227658512231653e-06,
859
+ "loss": 2.5636,
860
+ "step": 14200
861
+ },
862
+ {
863
+ "epoch": 1.78,
864
+ "learning_rate": 8.215177234148778e-06,
865
+ "loss": 2.5922,
866
+ "step": 14300
867
+ },
868
+ {
869
+ "epoch": 1.8,
870
+ "learning_rate": 8.202695956065901e-06,
871
+ "loss": 2.5404,
872
+ "step": 14400
873
+ },
874
+ {
875
+ "epoch": 1.81,
876
+ "learning_rate": 8.190214677983026e-06,
877
+ "loss": 2.6366,
878
+ "step": 14500
879
+ },
880
+ {
881
+ "epoch": 1.82,
882
+ "learning_rate": 8.177733399900151e-06,
883
+ "loss": 2.6201,
884
+ "step": 14600
885
+ },
886
+ {
887
+ "epoch": 1.83,
888
+ "learning_rate": 8.165252121817275e-06,
889
+ "loss": 2.5972,
890
+ "step": 14700
891
+ },
892
+ {
893
+ "epoch": 1.85,
894
+ "learning_rate": 8.152770843734398e-06,
895
+ "loss": 2.5356,
896
+ "step": 14800
897
+ },
898
+ {
899
+ "epoch": 1.86,
900
+ "learning_rate": 8.140289565651523e-06,
901
+ "loss": 2.6053,
902
+ "step": 14900
903
+ },
904
+ {
905
+ "epoch": 1.87,
906
+ "learning_rate": 8.127808287568648e-06,
907
+ "loss": 2.5312,
908
+ "step": 15000
909
+ },
910
+ {
911
+ "epoch": 1.88,
912
+ "learning_rate": 8.115327009485772e-06,
913
+ "loss": 2.5696,
914
+ "step": 15100
915
+ },
916
+ {
917
+ "epoch": 1.9,
918
+ "learning_rate": 8.102845731402897e-06,
919
+ "loss": 2.5485,
920
+ "step": 15200
921
+ },
922
+ {
923
+ "epoch": 1.91,
924
+ "learning_rate": 8.09036445332002e-06,
925
+ "loss": 2.5444,
926
+ "step": 15300
927
+ },
928
+ {
929
+ "epoch": 1.92,
930
+ "learning_rate": 8.077883175237144e-06,
931
+ "loss": 2.5953,
932
+ "step": 15400
933
+ },
934
+ {
935
+ "epoch": 1.93,
936
+ "learning_rate": 8.065401897154269e-06,
937
+ "loss": 2.5588,
938
+ "step": 15500
939
+ },
940
+ {
941
+ "epoch": 1.95,
942
+ "learning_rate": 8.052920619071394e-06,
943
+ "loss": 2.5685,
944
+ "step": 15600
945
+ },
946
+ {
947
+ "epoch": 1.96,
948
+ "learning_rate": 8.040439340988519e-06,
949
+ "loss": 2.6248,
950
+ "step": 15700
951
+ },
952
+ {
953
+ "epoch": 1.97,
954
+ "learning_rate": 8.027958062905642e-06,
955
+ "loss": 2.5816,
956
+ "step": 15800
957
+ },
958
+ {
959
+ "epoch": 1.98,
960
+ "learning_rate": 8.015476784822767e-06,
961
+ "loss": 2.5558,
962
+ "step": 15900
963
+ },
964
+ {
965
+ "epoch": 2.0,
966
+ "learning_rate": 8.00299550673989e-06,
967
+ "loss": 2.6264,
968
+ "step": 16000
969
+ },
970
+ {
971
+ "epoch": 2.01,
972
+ "learning_rate": 7.990514228657014e-06,
973
+ "loss": 2.5924,
974
+ "step": 16100
975
+ },
976
+ {
977
+ "epoch": 2.02,
978
+ "learning_rate": 7.97803295057414e-06,
979
+ "loss": 2.5352,
980
+ "step": 16200
981
+ },
982
+ {
983
+ "epoch": 2.03,
984
+ "learning_rate": 7.965551672491264e-06,
985
+ "loss": 2.5441,
986
+ "step": 16300
987
+ },
988
+ {
989
+ "epoch": 2.05,
990
+ "learning_rate": 7.953070394408388e-06,
991
+ "loss": 2.5431,
992
+ "step": 16400
993
+ },
994
+ {
995
+ "epoch": 2.06,
996
+ "learning_rate": 7.940589116325513e-06,
997
+ "loss": 2.5556,
998
+ "step": 16500
999
+ },
1000
+ {
1001
+ "epoch": 2.07,
1002
+ "learning_rate": 7.928107838242636e-06,
1003
+ "loss": 2.5459,
1004
+ "step": 16600
1005
+ },
1006
+ {
1007
+ "epoch": 2.08,
1008
+ "learning_rate": 7.915626560159761e-06,
1009
+ "loss": 2.6096,
1010
+ "step": 16700
1011
+ },
1012
+ {
1013
+ "epoch": 2.1,
1014
+ "learning_rate": 7.903145282076885e-06,
1015
+ "loss": 2.5705,
1016
+ "step": 16800
1017
+ },
1018
+ {
1019
+ "epoch": 2.11,
1020
+ "learning_rate": 7.89066400399401e-06,
1021
+ "loss": 2.6404,
1022
+ "step": 16900
1023
+ },
1024
+ {
1025
+ "epoch": 2.12,
1026
+ "learning_rate": 7.878182725911135e-06,
1027
+ "loss": 2.5122,
1028
+ "step": 17000
1029
+ },
1030
+ {
1031
+ "epoch": 2.13,
1032
+ "learning_rate": 7.865701447828258e-06,
1033
+ "loss": 2.5761,
1034
+ "step": 17100
1035
+ },
1036
+ {
1037
+ "epoch": 2.15,
1038
+ "learning_rate": 7.853220169745383e-06,
1039
+ "loss": 2.5586,
1040
+ "step": 17200
1041
+ },
1042
+ {
1043
+ "epoch": 2.16,
1044
+ "learning_rate": 7.840738891662507e-06,
1045
+ "loss": 2.5417,
1046
+ "step": 17300
1047
+ },
1048
+ {
1049
+ "epoch": 2.17,
1050
+ "learning_rate": 7.82825761357963e-06,
1051
+ "loss": 2.5312,
1052
+ "step": 17400
1053
+ },
1054
+ {
1055
+ "epoch": 2.18,
1056
+ "learning_rate": 7.815776335496755e-06,
1057
+ "loss": 2.6014,
1058
+ "step": 17500
1059
+ },
1060
+ {
1061
+ "epoch": 2.2,
1062
+ "learning_rate": 7.80329505741388e-06,
1063
+ "loss": 2.5719,
1064
+ "step": 17600
1065
+ },
1066
+ {
1067
+ "epoch": 2.21,
1068
+ "learning_rate": 7.790813779331005e-06,
1069
+ "loss": 2.6213,
1070
+ "step": 17700
1071
+ },
1072
+ {
1073
+ "epoch": 2.22,
1074
+ "learning_rate": 7.778332501248129e-06,
1075
+ "loss": 2.5363,
1076
+ "step": 17800
1077
+ },
1078
+ {
1079
+ "epoch": 2.23,
1080
+ "learning_rate": 7.765851223165252e-06,
1081
+ "loss": 2.5023,
1082
+ "step": 17900
1083
+ },
1084
+ {
1085
+ "epoch": 2.25,
1086
+ "learning_rate": 7.753369945082377e-06,
1087
+ "loss": 2.5229,
1088
+ "step": 18000
1089
+ },
1090
+ {
1091
+ "epoch": 2.26,
1092
+ "learning_rate": 7.7408886669995e-06,
1093
+ "loss": 2.5801,
1094
+ "step": 18100
1095
+ },
1096
+ {
1097
+ "epoch": 2.27,
1098
+ "learning_rate": 7.728407388916626e-06,
1099
+ "loss": 2.5993,
1100
+ "step": 18200
1101
+ },
1102
+ {
1103
+ "epoch": 2.28,
1104
+ "learning_rate": 7.71592611083375e-06,
1105
+ "loss": 2.5494,
1106
+ "step": 18300
1107
+ },
1108
+ {
1109
+ "epoch": 2.3,
1110
+ "learning_rate": 7.703444832750874e-06,
1111
+ "loss": 2.5426,
1112
+ "step": 18400
1113
+ },
1114
+ {
1115
+ "epoch": 2.31,
1116
+ "learning_rate": 7.690963554667998e-06,
1117
+ "loss": 2.5575,
1118
+ "step": 18500
1119
+ },
1120
+ {
1121
+ "epoch": 2.32,
1122
+ "learning_rate": 7.678482276585123e-06,
1123
+ "loss": 2.5636,
1124
+ "step": 18600
1125
+ },
1126
+ {
1127
+ "epoch": 2.33,
1128
+ "learning_rate": 7.666000998502248e-06,
1129
+ "loss": 2.6409,
1130
+ "step": 18700
1131
+ },
1132
+ {
1133
+ "epoch": 2.35,
1134
+ "learning_rate": 7.653519720419371e-06,
1135
+ "loss": 2.6014,
1136
+ "step": 18800
1137
+ },
1138
+ {
1139
+ "epoch": 2.36,
1140
+ "learning_rate": 7.641038442336496e-06,
1141
+ "loss": 2.5964,
1142
+ "step": 18900
1143
+ },
1144
+ {
1145
+ "epoch": 2.37,
1146
+ "learning_rate": 7.6285571642536205e-06,
1147
+ "loss": 2.5992,
1148
+ "step": 19000
1149
+ },
1150
+ {
1151
+ "epoch": 2.38,
1152
+ "learning_rate": 7.616075886170744e-06,
1153
+ "loss": 2.5852,
1154
+ "step": 19100
1155
+ },
1156
+ {
1157
+ "epoch": 2.4,
1158
+ "learning_rate": 7.603594608087869e-06,
1159
+ "loss": 2.6166,
1160
+ "step": 19200
1161
+ },
1162
+ {
1163
+ "epoch": 2.41,
1164
+ "learning_rate": 7.591113330004993e-06,
1165
+ "loss": 2.5835,
1166
+ "step": 19300
1167
+ },
1168
+ {
1169
+ "epoch": 2.42,
1170
+ "learning_rate": 7.578632051922118e-06,
1171
+ "loss": 2.5523,
1172
+ "step": 19400
1173
+ },
1174
+ {
1175
+ "epoch": 2.43,
1176
+ "learning_rate": 7.566150773839242e-06,
1177
+ "loss": 2.5704,
1178
+ "step": 19500
1179
+ },
1180
+ {
1181
+ "epoch": 2.45,
1182
+ "learning_rate": 7.553669495756366e-06,
1183
+ "loss": 2.6,
1184
+ "step": 19600
1185
+ },
1186
+ {
1187
+ "epoch": 2.46,
1188
+ "learning_rate": 7.541188217673491e-06,
1189
+ "loss": 2.5654,
1190
+ "step": 19700
1191
+ },
1192
+ {
1193
+ "epoch": 2.47,
1194
+ "learning_rate": 7.528706939590614e-06,
1195
+ "loss": 2.5707,
1196
+ "step": 19800
1197
+ },
1198
+ {
1199
+ "epoch": 2.48,
1200
+ "learning_rate": 7.516225661507739e-06,
1201
+ "loss": 2.5366,
1202
+ "step": 19900
1203
+ },
1204
+ {
1205
+ "epoch": 2.5,
1206
+ "learning_rate": 7.503744383424864e-06,
1207
+ "loss": 2.5468,
1208
+ "step": 20000
1209
+ },
1210
+ {
1211
+ "epoch": 2.51,
1212
+ "learning_rate": 7.491263105341987e-06,
1213
+ "loss": 2.5059,
1214
+ "step": 20100
1215
+ },
1216
+ {
1217
+ "epoch": 2.52,
1218
+ "learning_rate": 7.478781827259111e-06,
1219
+ "loss": 2.5545,
1220
+ "step": 20200
1221
+ },
1222
+ {
1223
+ "epoch": 2.53,
1224
+ "learning_rate": 7.466300549176236e-06,
1225
+ "loss": 2.6233,
1226
+ "step": 20300
1227
+ },
1228
+ {
1229
+ "epoch": 2.55,
1230
+ "learning_rate": 7.453819271093361e-06,
1231
+ "loss": 2.5522,
1232
+ "step": 20400
1233
+ },
1234
+ {
1235
+ "epoch": 2.56,
1236
+ "learning_rate": 7.441337993010485e-06,
1237
+ "loss": 2.5458,
1238
+ "step": 20500
1239
+ },
1240
+ {
1241
+ "epoch": 2.57,
1242
+ "learning_rate": 7.428856714927609e-06,
1243
+ "loss": 2.5973,
1244
+ "step": 20600
1245
+ },
1246
+ {
1247
+ "epoch": 2.58,
1248
+ "learning_rate": 7.416375436844734e-06,
1249
+ "loss": 2.5794,
1250
+ "step": 20700
1251
+ },
1252
+ {
1253
+ "epoch": 2.6,
1254
+ "learning_rate": 7.403894158761858e-06,
1255
+ "loss": 2.5503,
1256
+ "step": 20800
1257
+ },
1258
+ {
1259
+ "epoch": 2.61,
1260
+ "learning_rate": 7.391412880678982e-06,
1261
+ "loss": 2.5904,
1262
+ "step": 20900
1263
+ },
1264
+ {
1265
+ "epoch": 2.62,
1266
+ "learning_rate": 7.378931602596107e-06,
1267
+ "loss": 2.5643,
1268
+ "step": 21000
1269
+ },
1270
+ {
1271
+ "epoch": 2.63,
1272
+ "learning_rate": 7.366450324513231e-06,
1273
+ "loss": 2.6286,
1274
+ "step": 21100
1275
+ },
1276
+ {
1277
+ "epoch": 2.65,
1278
+ "learning_rate": 7.3539690464303545e-06,
1279
+ "loss": 2.4896,
1280
+ "step": 21200
1281
+ },
1282
+ {
1283
+ "epoch": 2.66,
1284
+ "learning_rate": 7.34148776834748e-06,
1285
+ "loss": 2.5652,
1286
+ "step": 21300
1287
+ },
1288
+ {
1289
+ "epoch": 2.67,
1290
+ "learning_rate": 7.329006490264604e-06,
1291
+ "loss": 2.5344,
1292
+ "step": 21400
1293
+ },
1294
+ {
1295
+ "epoch": 2.68,
1296
+ "learning_rate": 7.316525212181727e-06,
1297
+ "loss": 2.5794,
1298
+ "step": 21500
1299
+ },
1300
+ {
1301
+ "epoch": 2.7,
1302
+ "learning_rate": 7.304043934098852e-06,
1303
+ "loss": 2.5678,
1304
+ "step": 21600
1305
+ },
1306
+ {
1307
+ "epoch": 2.71,
1308
+ "learning_rate": 7.2915626560159766e-06,
1309
+ "loss": 2.5789,
1310
+ "step": 21700
1311
+ },
1312
+ {
1313
+ "epoch": 2.72,
1314
+ "learning_rate": 7.279081377933101e-06,
1315
+ "loss": 2.5663,
1316
+ "step": 21800
1317
+ },
1318
+ {
1319
+ "epoch": 2.73,
1320
+ "learning_rate": 7.266600099850225e-06,
1321
+ "loss": 2.5707,
1322
+ "step": 21900
1323
+ },
1324
+ {
1325
+ "epoch": 2.75,
1326
+ "learning_rate": 7.25411882176735e-06,
1327
+ "loss": 2.552,
1328
+ "step": 22000
1329
+ },
1330
+ {
1331
+ "epoch": 2.76,
1332
+ "learning_rate": 7.241637543684474e-06,
1333
+ "loss": 2.5694,
1334
+ "step": 22100
1335
+ },
1336
+ {
1337
+ "epoch": 2.77,
1338
+ "learning_rate": 7.229156265601598e-06,
1339
+ "loss": 2.5828,
1340
+ "step": 22200
1341
+ },
1342
+ {
1343
+ "epoch": 2.78,
1344
+ "learning_rate": 7.216674987518723e-06,
1345
+ "loss": 2.5308,
1346
+ "step": 22300
1347
+ },
1348
+ {
1349
+ "epoch": 2.8,
1350
+ "learning_rate": 7.204193709435847e-06,
1351
+ "loss": 2.4647,
1352
+ "step": 22400
1353
+ },
1354
+ {
1355
+ "epoch": 2.81,
1356
+ "learning_rate": 7.1917124313529705e-06,
1357
+ "loss": 2.5888,
1358
+ "step": 22500
1359
+ },
1360
+ {
1361
+ "epoch": 2.82,
1362
+ "learning_rate": 7.1792311532700956e-06,
1363
+ "loss": 2.5999,
1364
+ "step": 22600
1365
+ },
1366
+ {
1367
+ "epoch": 2.83,
1368
+ "learning_rate": 7.16674987518722e-06,
1369
+ "loss": 2.6143,
1370
+ "step": 22700
1371
+ },
1372
+ {
1373
+ "epoch": 2.85,
1374
+ "learning_rate": 7.154268597104343e-06,
1375
+ "loss": 2.6138,
1376
+ "step": 22800
1377
+ },
1378
+ {
1379
+ "epoch": 2.86,
1380
+ "learning_rate": 7.141787319021468e-06,
1381
+ "loss": 2.5534,
1382
+ "step": 22900
1383
+ },
1384
+ {
1385
+ "epoch": 2.87,
1386
+ "learning_rate": 7.1293060409385925e-06,
1387
+ "loss": 2.5856,
1388
+ "step": 23000
1389
+ },
1390
+ {
1391
+ "epoch": 2.88,
1392
+ "learning_rate": 7.116824762855718e-06,
1393
+ "loss": 2.5619,
1394
+ "step": 23100
1395
+ },
1396
+ {
1397
+ "epoch": 2.9,
1398
+ "learning_rate": 7.104343484772841e-06,
1399
+ "loss": 2.5742,
1400
+ "step": 23200
1401
+ },
1402
+ {
1403
+ "epoch": 2.91,
1404
+ "learning_rate": 7.091862206689965e-06,
1405
+ "loss": 2.5515,
1406
+ "step": 23300
1407
+ },
1408
+ {
1409
+ "epoch": 2.92,
1410
+ "learning_rate": 7.07938092860709e-06,
1411
+ "loss": 2.543,
1412
+ "step": 23400
1413
+ },
1414
+ {
1415
+ "epoch": 2.93,
1416
+ "learning_rate": 7.066899650524214e-06,
1417
+ "loss": 2.5166,
1418
+ "step": 23500
1419
+ },
1420
+ {
1421
+ "epoch": 2.95,
1422
+ "learning_rate": 7.054418372441339e-06,
1423
+ "loss": 2.5233,
1424
+ "step": 23600
1425
+ },
1426
+ {
1427
+ "epoch": 2.96,
1428
+ "learning_rate": 7.041937094358463e-06,
1429
+ "loss": 2.571,
1430
+ "step": 23700
1431
+ },
1432
+ {
1433
+ "epoch": 2.97,
1434
+ "learning_rate": 7.029455816275588e-06,
1435
+ "loss": 2.6225,
1436
+ "step": 23800
1437
+ },
1438
+ {
1439
+ "epoch": 2.98,
1440
+ "learning_rate": 7.0169745381927115e-06,
1441
+ "loss": 2.5945,
1442
+ "step": 23900
1443
+ },
1444
+ {
1445
+ "epoch": 3.0,
1446
+ "learning_rate": 7.004493260109836e-06,
1447
+ "loss": 2.5936,
1448
+ "step": 24000
1449
+ },
1450
+ {
1451
+ "epoch": 3.01,
1452
+ "learning_rate": 6.992011982026961e-06,
1453
+ "loss": 2.6008,
1454
+ "step": 24100
1455
+ },
1456
+ {
1457
+ "epoch": 3.02,
1458
+ "learning_rate": 6.979530703944084e-06,
1459
+ "loss": 2.5778,
1460
+ "step": 24200
1461
+ },
1462
+ {
1463
+ "epoch": 3.03,
1464
+ "learning_rate": 6.9670494258612085e-06,
1465
+ "loss": 2.5153,
1466
+ "step": 24300
1467
+ },
1468
+ {
1469
+ "epoch": 3.05,
1470
+ "learning_rate": 6.9545681477783335e-06,
1471
+ "loss": 2.5507,
1472
+ "step": 24400
1473
+ },
1474
+ {
1475
+ "epoch": 3.06,
1476
+ "learning_rate": 6.942086869695457e-06,
1477
+ "loss": 2.5331,
1478
+ "step": 24500
1479
+ },
1480
+ {
1481
+ "epoch": 3.07,
1482
+ "learning_rate": 6.929605591612581e-06,
1483
+ "loss": 2.4988,
1484
+ "step": 24600
1485
+ },
1486
+ {
1487
+ "epoch": 3.08,
1488
+ "learning_rate": 6.917124313529706e-06,
1489
+ "loss": 2.6262,
1490
+ "step": 24700
1491
+ },
1492
+ {
1493
+ "epoch": 3.1,
1494
+ "learning_rate": 6.9046430354468305e-06,
1495
+ "loss": 2.5813,
1496
+ "step": 24800
1497
+ },
1498
+ {
1499
+ "epoch": 3.11,
1500
+ "learning_rate": 6.892161757363955e-06,
1501
+ "loss": 2.475,
1502
+ "step": 24900
1503
+ },
1504
+ {
1505
+ "epoch": 3.12,
1506
+ "learning_rate": 6.879680479281079e-06,
1507
+ "loss": 2.5655,
1508
+ "step": 25000
1509
+ },
1510
+ {
1511
+ "epoch": 3.13,
1512
+ "learning_rate": 6.867199201198203e-06,
1513
+ "loss": 2.5762,
1514
+ "step": 25100
1515
+ },
1516
+ {
1517
+ "epoch": 3.15,
1518
+ "learning_rate": 6.8547179231153274e-06,
1519
+ "loss": 2.5742,
1520
+ "step": 25200
1521
+ },
1522
+ {
1523
+ "epoch": 3.16,
1524
+ "learning_rate": 6.842236645032452e-06,
1525
+ "loss": 2.6108,
1526
+ "step": 25300
1527
+ },
1528
+ {
1529
+ "epoch": 3.17,
1530
+ "learning_rate": 6.829755366949577e-06,
1531
+ "loss": 2.576,
1532
+ "step": 25400
1533
+ },
1534
+ {
1535
+ "epoch": 3.18,
1536
+ "learning_rate": 6.817274088866701e-06,
1537
+ "loss": 2.5012,
1538
+ "step": 25500
1539
+ },
1540
+ {
1541
+ "epoch": 3.2,
1542
+ "learning_rate": 6.804792810783824e-06,
1543
+ "loss": 2.528,
1544
+ "step": 25600
1545
+ },
1546
+ {
1547
+ "epoch": 3.21,
1548
+ "learning_rate": 6.7923115327009495e-06,
1549
+ "loss": 2.5362,
1550
+ "step": 25700
1551
+ },
1552
+ {
1553
+ "epoch": 3.22,
1554
+ "learning_rate": 6.779830254618074e-06,
1555
+ "loss": 2.5012,
1556
+ "step": 25800
1557
+ },
1558
+ {
1559
+ "epoch": 3.23,
1560
+ "learning_rate": 6.767348976535197e-06,
1561
+ "loss": 2.5318,
1562
+ "step": 25900
1563
+ },
1564
+ {
1565
+ "epoch": 3.25,
1566
+ "learning_rate": 6.754867698452322e-06,
1567
+ "loss": 2.5424,
1568
+ "step": 26000
1569
+ },
1570
+ {
1571
+ "epoch": 3.26,
1572
+ "learning_rate": 6.7423864203694464e-06,
1573
+ "loss": 2.4849,
1574
+ "step": 26100
1575
+ },
1576
+ {
1577
+ "epoch": 3.27,
1578
+ "learning_rate": 6.72990514228657e-06,
1579
+ "loss": 2.5104,
1580
+ "step": 26200
1581
+ },
1582
+ {
1583
+ "epoch": 3.28,
1584
+ "learning_rate": 6.717423864203695e-06,
1585
+ "loss": 2.6342,
1586
+ "step": 26300
1587
+ },
1588
+ {
1589
+ "epoch": 3.3,
1590
+ "learning_rate": 6.704942586120819e-06,
1591
+ "loss": 2.5654,
1592
+ "step": 26400
1593
+ },
1594
+ {
1595
+ "epoch": 3.31,
1596
+ "learning_rate": 6.692461308037944e-06,
1597
+ "loss": 2.5578,
1598
+ "step": 26500
1599
+ },
1600
+ {
1601
+ "epoch": 3.32,
1602
+ "learning_rate": 6.679980029955068e-06,
1603
+ "loss": 2.561,
1604
+ "step": 26600
1605
+ },
1606
+ {
1607
+ "epoch": 3.33,
1608
+ "learning_rate": 6.667498751872193e-06,
1609
+ "loss": 2.5678,
1610
+ "step": 26700
1611
+ },
1612
+ {
1613
+ "epoch": 3.34,
1614
+ "learning_rate": 6.655017473789317e-06,
1615
+ "loss": 2.5214,
1616
+ "step": 26800
1617
+ },
1618
+ {
1619
+ "epoch": 3.36,
1620
+ "learning_rate": 6.64253619570644e-06,
1621
+ "loss": 2.6021,
1622
+ "step": 26900
1623
+ },
1624
+ {
1625
+ "epoch": 3.37,
1626
+ "learning_rate": 6.6300549176235654e-06,
1627
+ "loss": 2.524,
1628
+ "step": 27000
1629
+ },
1630
+ {
1631
+ "epoch": 3.38,
1632
+ "learning_rate": 6.61757363954069e-06,
1633
+ "loss": 2.5641,
1634
+ "step": 27100
1635
+ },
1636
+ {
1637
+ "epoch": 3.39,
1638
+ "learning_rate": 6.605092361457813e-06,
1639
+ "loss": 2.59,
1640
+ "step": 27200
1641
+ },
1642
+ {
1643
+ "epoch": 3.41,
1644
+ "learning_rate": 6.592611083374938e-06,
1645
+ "loss": 2.5593,
1646
+ "step": 27300
1647
+ },
1648
+ {
1649
+ "epoch": 3.42,
1650
+ "learning_rate": 6.580129805292062e-06,
1651
+ "loss": 2.5457,
1652
+ "step": 27400
1653
+ },
1654
+ {
1655
+ "epoch": 3.43,
1656
+ "learning_rate": 6.5676485272091875e-06,
1657
+ "loss": 2.5725,
1658
+ "step": 27500
1659
+ },
1660
+ {
1661
+ "epoch": 3.44,
1662
+ "learning_rate": 6.555167249126311e-06,
1663
+ "loss": 2.5082,
1664
+ "step": 27600
1665
+ },
1666
+ {
1667
+ "epoch": 3.46,
1668
+ "learning_rate": 6.542685971043435e-06,
1669
+ "loss": 2.5807,
1670
+ "step": 27700
1671
+ },
1672
+ {
1673
+ "epoch": 3.47,
1674
+ "learning_rate": 6.53020469296056e-06,
1675
+ "loss": 2.6353,
1676
+ "step": 27800
1677
+ },
1678
+ {
1679
+ "epoch": 3.48,
1680
+ "learning_rate": 6.5177234148776836e-06,
1681
+ "loss": 2.6223,
1682
+ "step": 27900
1683
+ },
1684
+ {
1685
+ "epoch": 3.49,
1686
+ "learning_rate": 6.505242136794808e-06,
1687
+ "loss": 2.5538,
1688
+ "step": 28000
1689
+ },
1690
+ {
1691
+ "epoch": 3.51,
1692
+ "learning_rate": 6.492760858711933e-06,
1693
+ "loss": 2.5544,
1694
+ "step": 28100
1695
+ },
1696
+ {
1697
+ "epoch": 3.52,
1698
+ "learning_rate": 6.480279580629057e-06,
1699
+ "loss": 2.6314,
1700
+ "step": 28200
1701
+ },
1702
+ {
1703
+ "epoch": 3.53,
1704
+ "learning_rate": 6.467798302546181e-06,
1705
+ "loss": 2.5278,
1706
+ "step": 28300
1707
+ },
1708
+ {
1709
+ "epoch": 3.54,
1710
+ "learning_rate": 6.455317024463306e-06,
1711
+ "loss": 2.585,
1712
+ "step": 28400
1713
+ },
1714
+ {
1715
+ "epoch": 3.56,
1716
+ "learning_rate": 6.442835746380431e-06,
1717
+ "loss": 2.5468,
1718
+ "step": 28500
1719
+ },
1720
+ {
1721
+ "epoch": 3.57,
1722
+ "learning_rate": 6.430354468297554e-06,
1723
+ "loss": 2.5496,
1724
+ "step": 28600
1725
+ },
1726
+ {
1727
+ "epoch": 3.58,
1728
+ "learning_rate": 6.417873190214678e-06,
1729
+ "loss": 2.5182,
1730
+ "step": 28700
1731
+ },
1732
+ {
1733
+ "epoch": 3.59,
1734
+ "learning_rate": 6.405391912131803e-06,
1735
+ "loss": 2.6226,
1736
+ "step": 28800
1737
+ },
1738
+ {
1739
+ "epoch": 3.61,
1740
+ "learning_rate": 6.392910634048927e-06,
1741
+ "loss": 2.5062,
1742
+ "step": 28900
1743
+ },
1744
+ {
1745
+ "epoch": 3.62,
1746
+ "learning_rate": 6.380429355966051e-06,
1747
+ "loss": 2.5962,
1748
+ "step": 29000
1749
+ },
1750
+ {
1751
+ "epoch": 3.63,
1752
+ "learning_rate": 6.367948077883176e-06,
1753
+ "loss": 2.5113,
1754
+ "step": 29100
1755
+ },
1756
+ {
1757
+ "epoch": 3.64,
1758
+ "learning_rate": 6.3554667998003e-06,
1759
+ "loss": 2.5236,
1760
+ "step": 29200
1761
+ },
1762
+ {
1763
+ "epoch": 3.66,
1764
+ "learning_rate": 6.342985521717424e-06,
1765
+ "loss": 2.4738,
1766
+ "step": 29300
1767
+ },
1768
+ {
1769
+ "epoch": 3.67,
1770
+ "learning_rate": 6.330504243634549e-06,
1771
+ "loss": 2.5689,
1772
+ "step": 29400
1773
+ },
1774
+ {
1775
+ "epoch": 3.68,
1776
+ "learning_rate": 6.318022965551673e-06,
1777
+ "loss": 2.6064,
1778
+ "step": 29500
1779
+ },
1780
+ {
1781
+ "epoch": 3.69,
1782
+ "learning_rate": 6.305541687468797e-06,
1783
+ "loss": 2.5285,
1784
+ "step": 29600
1785
+ },
1786
+ {
1787
+ "epoch": 3.71,
1788
+ "learning_rate": 6.2930604093859216e-06,
1789
+ "loss": 2.6252,
1790
+ "step": 29700
1791
+ },
1792
+ {
1793
+ "epoch": 3.72,
1794
+ "learning_rate": 6.280579131303046e-06,
1795
+ "loss": 2.5797,
1796
+ "step": 29800
1797
+ },
1798
+ {
1799
+ "epoch": 3.73,
1800
+ "learning_rate": 6.26809785322017e-06,
1801
+ "loss": 2.526,
1802
+ "step": 29900
1803
+ },
1804
+ {
1805
+ "epoch": 3.74,
1806
+ "learning_rate": 6.255616575137294e-06,
1807
+ "loss": 2.5698,
1808
+ "step": 30000
1809
+ },
1810
+ {
1811
+ "epoch": 3.76,
1812
+ "learning_rate": 6.243135297054419e-06,
1813
+ "loss": 2.587,
1814
+ "step": 30100
1815
+ },
1816
+ {
1817
+ "epoch": 3.77,
1818
+ "learning_rate": 6.230654018971544e-06,
1819
+ "loss": 2.5236,
1820
+ "step": 30200
1821
+ },
1822
+ {
1823
+ "epoch": 3.78,
1824
+ "learning_rate": 6.218172740888667e-06,
1825
+ "loss": 2.539,
1826
+ "step": 30300
1827
+ },
1828
+ {
1829
+ "epoch": 3.79,
1830
+ "learning_rate": 6.205691462805792e-06,
1831
+ "loss": 2.5915,
1832
+ "step": 30400
1833
+ },
1834
+ {
1835
+ "epoch": 3.81,
1836
+ "learning_rate": 6.193210184722916e-06,
1837
+ "loss": 2.5222,
1838
+ "step": 30500
1839
+ },
1840
+ {
1841
+ "epoch": 3.82,
1842
+ "learning_rate": 6.18072890664004e-06,
1843
+ "loss": 2.5455,
1844
+ "step": 30600
1845
+ },
1846
+ {
1847
+ "epoch": 3.83,
1848
+ "learning_rate": 6.168247628557165e-06,
1849
+ "loss": 2.5418,
1850
+ "step": 30700
1851
+ },
1852
+ {
1853
+ "epoch": 3.84,
1854
+ "learning_rate": 6.155766350474289e-06,
1855
+ "loss": 2.5469,
1856
+ "step": 30800
1857
+ },
1858
+ {
1859
+ "epoch": 3.86,
1860
+ "learning_rate": 6.143285072391414e-06,
1861
+ "loss": 2.5557,
1862
+ "step": 30900
1863
+ },
1864
+ {
1865
+ "epoch": 3.87,
1866
+ "learning_rate": 6.1308037943085375e-06,
1867
+ "loss": 2.534,
1868
+ "step": 31000
1869
+ },
1870
+ {
1871
+ "epoch": 3.88,
1872
+ "learning_rate": 6.118322516225662e-06,
1873
+ "loss": 2.5967,
1874
+ "step": 31100
1875
+ },
1876
+ {
1877
+ "epoch": 3.89,
1878
+ "learning_rate": 6.105841238142787e-06,
1879
+ "loss": 2.493,
1880
+ "step": 31200
1881
+ },
1882
+ {
1883
+ "epoch": 3.91,
1884
+ "learning_rate": 6.09335996005991e-06,
1885
+ "loss": 2.4888,
1886
+ "step": 31300
1887
+ },
1888
+ {
1889
+ "epoch": 3.92,
1890
+ "learning_rate": 6.080878681977035e-06,
1891
+ "loss": 2.4598,
1892
+ "step": 31400
1893
+ },
1894
+ {
1895
+ "epoch": 3.93,
1896
+ "learning_rate": 6.0683974038941595e-06,
1897
+ "loss": 2.5489,
1898
+ "step": 31500
1899
+ },
1900
+ {
1901
+ "epoch": 3.94,
1902
+ "learning_rate": 6.055916125811283e-06,
1903
+ "loss": 2.6207,
1904
+ "step": 31600
1905
+ },
1906
+ {
1907
+ "epoch": 3.96,
1908
+ "learning_rate": 6.043434847728408e-06,
1909
+ "loss": 2.4955,
1910
+ "step": 31700
1911
+ },
1912
+ {
1913
+ "epoch": 3.97,
1914
+ "learning_rate": 6.030953569645532e-06,
1915
+ "loss": 2.5457,
1916
+ "step": 31800
1917
+ },
1918
+ {
1919
+ "epoch": 3.98,
1920
+ "learning_rate": 6.018472291562657e-06,
1921
+ "loss": 2.5991,
1922
+ "step": 31900
1923
+ },
1924
+ {
1925
+ "epoch": 3.99,
1926
+ "learning_rate": 6.005991013479781e-06,
1927
+ "loss": 2.6215,
1928
+ "step": 32000
1929
+ },
1930
+ {
1931
+ "epoch": 4.01,
1932
+ "learning_rate": 5.993509735396905e-06,
1933
+ "loss": 2.483,
1934
+ "step": 32100
1935
+ },
1936
+ {
1937
+ "epoch": 4.02,
1938
+ "learning_rate": 5.98102845731403e-06,
1939
+ "loss": 2.554,
1940
+ "step": 32200
1941
+ },
1942
+ {
1943
+ "epoch": 4.03,
1944
+ "learning_rate": 5.9685471792311534e-06,
1945
+ "loss": 2.5586,
1946
+ "step": 32300
1947
+ },
1948
+ {
1949
+ "epoch": 4.04,
1950
+ "learning_rate": 5.956065901148278e-06,
1951
+ "loss": 2.5568,
1952
+ "step": 32400
1953
+ },
1954
+ {
1955
+ "epoch": 4.06,
1956
+ "learning_rate": 5.943584623065403e-06,
1957
+ "loss": 2.5593,
1958
+ "step": 32500
1959
+ },
1960
+ {
1961
+ "epoch": 4.07,
1962
+ "learning_rate": 5.931103344982526e-06,
1963
+ "loss": 2.5854,
1964
+ "step": 32600
1965
+ },
1966
+ {
1967
+ "epoch": 4.08,
1968
+ "learning_rate": 5.91862206689965e-06,
1969
+ "loss": 2.5227,
1970
+ "step": 32700
1971
+ },
1972
+ {
1973
+ "epoch": 4.09,
1974
+ "learning_rate": 5.9061407888167755e-06,
1975
+ "loss": 2.5099,
1976
+ "step": 32800
1977
+ },
1978
+ {
1979
+ "epoch": 4.11,
1980
+ "learning_rate": 5.8936595107339e-06,
1981
+ "loss": 2.4718,
1982
+ "step": 32900
1983
+ },
1984
+ {
1985
+ "epoch": 4.12,
1986
+ "learning_rate": 5.881178232651024e-06,
1987
+ "loss": 2.5361,
1988
+ "step": 33000
1989
+ },
1990
+ {
1991
+ "epoch": 4.13,
1992
+ "learning_rate": 5.868696954568148e-06,
1993
+ "loss": 2.5505,
1994
+ "step": 33100
1995
+ },
1996
+ {
1997
+ "epoch": 4.14,
1998
+ "learning_rate": 5.856215676485273e-06,
1999
+ "loss": 2.5501,
2000
+ "step": 33200
2001
+ },
2002
+ {
2003
+ "epoch": 4.16,
2004
+ "learning_rate": 5.843734398402397e-06,
2005
+ "loss": 2.5911,
2006
+ "step": 33300
2007
+ },
2008
+ {
2009
+ "epoch": 4.17,
2010
+ "learning_rate": 5.831253120319521e-06,
2011
+ "loss": 2.5891,
2012
+ "step": 33400
2013
+ },
2014
+ {
2015
+ "epoch": 4.18,
2016
+ "learning_rate": 5.818771842236646e-06,
2017
+ "loss": 2.5229,
2018
+ "step": 33500
2019
+ },
2020
+ {
2021
+ "epoch": 4.19,
2022
+ "learning_rate": 5.80629056415377e-06,
2023
+ "loss": 2.518,
2024
+ "step": 33600
2025
+ },
2026
+ {
2027
+ "epoch": 4.21,
2028
+ "learning_rate": 5.793809286070894e-06,
2029
+ "loss": 2.5585,
2030
+ "step": 33700
2031
+ },
2032
+ {
2033
+ "epoch": 4.22,
2034
+ "learning_rate": 5.781328007988019e-06,
2035
+ "loss": 2.5518,
2036
+ "step": 33800
2037
+ },
2038
+ {
2039
+ "epoch": 4.23,
2040
+ "learning_rate": 5.768846729905143e-06,
2041
+ "loss": 2.5389,
2042
+ "step": 33900
2043
+ },
2044
+ {
2045
+ "epoch": 4.24,
2046
+ "learning_rate": 5.756365451822266e-06,
2047
+ "loss": 2.5181,
2048
+ "step": 34000
2049
+ },
2050
+ {
2051
+ "epoch": 4.26,
2052
+ "learning_rate": 5.743884173739391e-06,
2053
+ "loss": 2.4888,
2054
+ "step": 34100
2055
+ },
2056
+ {
2057
+ "epoch": 4.27,
2058
+ "learning_rate": 5.731402895656516e-06,
2059
+ "loss": 2.5526,
2060
+ "step": 34200
2061
+ },
2062
+ {
2063
+ "epoch": 4.28,
2064
+ "learning_rate": 5.71892161757364e-06,
2065
+ "loss": 2.5189,
2066
+ "step": 34300
2067
+ },
2068
+ {
2069
+ "epoch": 4.29,
2070
+ "learning_rate": 5.706440339490764e-06,
2071
+ "loss": 2.4954,
2072
+ "step": 34400
2073
+ },
2074
+ {
2075
+ "epoch": 4.31,
2076
+ "learning_rate": 5.693959061407888e-06,
2077
+ "loss": 2.5498,
2078
+ "step": 34500
2079
+ },
2080
+ {
2081
+ "epoch": 4.32,
2082
+ "learning_rate": 5.6814777833250135e-06,
2083
+ "loss": 2.5855,
2084
+ "step": 34600
2085
+ },
2086
+ {
2087
+ "epoch": 4.33,
2088
+ "learning_rate": 5.668996505242137e-06,
2089
+ "loss": 2.5233,
2090
+ "step": 34700
2091
+ },
2092
+ {
2093
+ "epoch": 4.34,
2094
+ "learning_rate": 5.656515227159262e-06,
2095
+ "loss": 2.4986,
2096
+ "step": 34800
2097
+ },
2098
+ {
2099
+ "epoch": 4.36,
2100
+ "learning_rate": 5.644033949076386e-06,
2101
+ "loss": 2.4563,
2102
+ "step": 34900
2103
+ },
2104
+ {
2105
+ "epoch": 4.37,
2106
+ "learning_rate": 5.6315526709935096e-06,
2107
+ "loss": 2.5183,
2108
+ "step": 35000
2109
+ },
2110
+ {
2111
+ "epoch": 4.38,
2112
+ "learning_rate": 5.619071392910635e-06,
2113
+ "loss": 2.5314,
2114
+ "step": 35100
2115
+ },
2116
+ {
2117
+ "epoch": 4.39,
2118
+ "learning_rate": 5.606590114827759e-06,
2119
+ "loss": 2.5084,
2120
+ "step": 35200
2121
+ },
2122
+ {
2123
+ "epoch": 4.41,
2124
+ "learning_rate": 5.594108836744882e-06,
2125
+ "loss": 2.5564,
2126
+ "step": 35300
2127
+ },
2128
+ {
2129
+ "epoch": 4.42,
2130
+ "learning_rate": 5.581627558662007e-06,
2131
+ "loss": 2.4687,
2132
+ "step": 35400
2133
+ },
2134
+ {
2135
+ "epoch": 4.43,
2136
+ "learning_rate": 5.569146280579132e-06,
2137
+ "loss": 2.5122,
2138
+ "step": 35500
2139
+ },
2140
+ {
2141
+ "epoch": 4.44,
2142
+ "learning_rate": 5.556665002496257e-06,
2143
+ "loss": 2.5395,
2144
+ "step": 35600
2145
+ },
2146
+ {
2147
+ "epoch": 4.46,
2148
+ "learning_rate": 5.54418372441338e-06,
2149
+ "loss": 2.6325,
2150
+ "step": 35700
2151
+ },
2152
+ {
2153
+ "epoch": 4.47,
2154
+ "learning_rate": 5.531702446330504e-06,
2155
+ "loss": 2.5592,
2156
+ "step": 35800
2157
+ },
2158
+ {
2159
+ "epoch": 4.48,
2160
+ "learning_rate": 5.519221168247629e-06,
2161
+ "loss": 2.5724,
2162
+ "step": 35900
2163
+ },
2164
+ {
2165
+ "epoch": 4.49,
2166
+ "learning_rate": 5.506739890164753e-06,
2167
+ "loss": 2.5779,
2168
+ "step": 36000
2169
+ },
2170
+ {
2171
+ "epoch": 4.51,
2172
+ "learning_rate": 5.494258612081878e-06,
2173
+ "loss": 2.556,
2174
+ "step": 36100
2175
+ },
2176
+ {
2177
+ "epoch": 4.52,
2178
+ "learning_rate": 5.481777333999002e-06,
2179
+ "loss": 2.4931,
2180
+ "step": 36200
2181
+ },
2182
+ {
2183
+ "epoch": 4.53,
2184
+ "learning_rate": 5.469296055916127e-06,
2185
+ "loss": 2.5409,
2186
+ "step": 36300
2187
+ },
2188
+ {
2189
+ "epoch": 4.54,
2190
+ "learning_rate": 5.456814777833251e-06,
2191
+ "loss": 2.5513,
2192
+ "step": 36400
2193
+ },
2194
+ {
2195
+ "epoch": 4.56,
2196
+ "learning_rate": 5.444333499750375e-06,
2197
+ "loss": 2.5558,
2198
+ "step": 36500
2199
+ },
2200
+ {
2201
+ "epoch": 4.57,
2202
+ "learning_rate": 5.4318522216675e-06,
2203
+ "loss": 2.5406,
2204
+ "step": 36600
2205
+ },
2206
+ {
2207
+ "epoch": 4.58,
2208
+ "learning_rate": 5.419370943584623e-06,
2209
+ "loss": 2.5845,
2210
+ "step": 36700
2211
+ },
2212
+ {
2213
+ "epoch": 4.59,
2214
+ "learning_rate": 5.4068896655017475e-06,
2215
+ "loss": 2.5388,
2216
+ "step": 36800
2217
+ },
2218
+ {
2219
+ "epoch": 4.61,
2220
+ "learning_rate": 5.394408387418873e-06,
2221
+ "loss": 2.5498,
2222
+ "step": 36900
2223
+ },
2224
+ {
2225
+ "epoch": 4.62,
2226
+ "learning_rate": 5.381927109335996e-06,
2227
+ "loss": 2.4883,
2228
+ "step": 37000
2229
+ },
2230
+ {
2231
+ "epoch": 4.63,
2232
+ "learning_rate": 5.36944583125312e-06,
2233
+ "loss": 2.5466,
2234
+ "step": 37100
2235
+ },
2236
+ {
2237
+ "epoch": 4.64,
2238
+ "learning_rate": 5.356964553170245e-06,
2239
+ "loss": 2.5449,
2240
+ "step": 37200
2241
+ },
2242
+ {
2243
+ "epoch": 4.66,
2244
+ "learning_rate": 5.3444832750873696e-06,
2245
+ "loss": 2.5573,
2246
+ "step": 37300
2247
+ },
2248
+ {
2249
+ "epoch": 4.67,
2250
+ "learning_rate": 5.332001997004494e-06,
2251
+ "loss": 2.5178,
2252
+ "step": 37400
2253
+ },
2254
+ {
2255
+ "epoch": 4.68,
2256
+ "learning_rate": 5.319520718921618e-06,
2257
+ "loss": 2.5307,
2258
+ "step": 37500
2259
+ },
2260
+ {
2261
+ "epoch": 4.69,
2262
+ "learning_rate": 5.307039440838742e-06,
2263
+ "loss": 2.5952,
2264
+ "step": 37600
2265
+ },
2266
+ {
2267
+ "epoch": 4.71,
2268
+ "learning_rate": 5.2945581627558665e-06,
2269
+ "loss": 2.6333,
2270
+ "step": 37700
2271
+ },
2272
+ {
2273
+ "epoch": 4.72,
2274
+ "learning_rate": 5.282076884672991e-06,
2275
+ "loss": 2.5498,
2276
+ "step": 37800
2277
+ },
2278
+ {
2279
+ "epoch": 4.73,
2280
+ "learning_rate": 5.269595606590116e-06,
2281
+ "loss": 2.5644,
2282
+ "step": 37900
2283
+ },
2284
+ {
2285
+ "epoch": 4.74,
2286
+ "learning_rate": 5.257114328507239e-06,
2287
+ "loss": 2.5154,
2288
+ "step": 38000
2289
+ },
2290
+ {
2291
+ "epoch": 4.76,
2292
+ "learning_rate": 5.2446330504243635e-06,
2293
+ "loss": 2.5258,
2294
+ "step": 38100
2295
+ },
2296
+ {
2297
+ "epoch": 4.77,
2298
+ "learning_rate": 5.2321517723414886e-06,
2299
+ "loss": 2.6026,
2300
+ "step": 38200
2301
+ },
2302
+ {
2303
+ "epoch": 4.78,
2304
+ "learning_rate": 5.219670494258613e-06,
2305
+ "loss": 2.5448,
2306
+ "step": 38300
2307
+ },
2308
+ {
2309
+ "epoch": 4.79,
2310
+ "learning_rate": 5.207189216175736e-06,
2311
+ "loss": 2.5479,
2312
+ "step": 38400
2313
+ },
2314
+ {
2315
+ "epoch": 4.81,
2316
+ "learning_rate": 5.194707938092861e-06,
2317
+ "loss": 2.6038,
2318
+ "step": 38500
2319
+ },
2320
+ {
2321
+ "epoch": 4.82,
2322
+ "learning_rate": 5.1822266600099855e-06,
2323
+ "loss": 2.5809,
2324
+ "step": 38600
2325
+ },
2326
+ {
2327
+ "epoch": 4.83,
2328
+ "learning_rate": 5.169745381927109e-06,
2329
+ "loss": 2.5219,
2330
+ "step": 38700
2331
+ },
2332
+ {
2333
+ "epoch": 4.84,
2334
+ "learning_rate": 5.157264103844234e-06,
2335
+ "loss": 2.5623,
2336
+ "step": 38800
2337
+ },
2338
+ {
2339
+ "epoch": 4.86,
2340
+ "learning_rate": 5.144782825761358e-06,
2341
+ "loss": 2.5276,
2342
+ "step": 38900
2343
+ },
2344
+ {
2345
+ "epoch": 4.87,
2346
+ "learning_rate": 5.132301547678483e-06,
2347
+ "loss": 2.5995,
2348
+ "step": 39000
2349
+ },
2350
+ {
2351
+ "epoch": 4.88,
2352
+ "learning_rate": 5.119820269595607e-06,
2353
+ "loss": 2.5641,
2354
+ "step": 39100
2355
+ },
2356
+ {
2357
+ "epoch": 4.89,
2358
+ "learning_rate": 5.107338991512732e-06,
2359
+ "loss": 2.5158,
2360
+ "step": 39200
2361
+ },
2362
+ {
2363
+ "epoch": 4.91,
2364
+ "learning_rate": 5.094857713429856e-06,
2365
+ "loss": 2.4968,
2366
+ "step": 39300
2367
+ },
2368
+ {
2369
+ "epoch": 4.92,
2370
+ "learning_rate": 5.082376435346979e-06,
2371
+ "loss": 2.5531,
2372
+ "step": 39400
2373
+ },
2374
+ {
2375
+ "epoch": 4.93,
2376
+ "learning_rate": 5.0698951572641045e-06,
2377
+ "loss": 2.5547,
2378
+ "step": 39500
2379
+ },
2380
+ {
2381
+ "epoch": 4.94,
2382
+ "learning_rate": 5.057413879181229e-06,
2383
+ "loss": 2.5746,
2384
+ "step": 39600
2385
+ },
2386
+ {
2387
+ "epoch": 4.96,
2388
+ "learning_rate": 5.044932601098352e-06,
2389
+ "loss": 2.461,
2390
+ "step": 39700
2391
+ },
2392
+ {
2393
+ "epoch": 4.97,
2394
+ "learning_rate": 5.032451323015477e-06,
2395
+ "loss": 2.5375,
2396
+ "step": 39800
2397
+ },
2398
+ {
2399
+ "epoch": 4.98,
2400
+ "learning_rate": 5.0199700449326015e-06,
2401
+ "loss": 2.5461,
2402
+ "step": 39900
2403
+ },
2404
+ {
2405
+ "epoch": 4.99,
2406
+ "learning_rate": 5.0074887668497265e-06,
2407
+ "loss": 2.5562,
2408
+ "step": 40000
2409
+ },
2410
+ {
2411
+ "epoch": 5.0,
2412
+ "learning_rate": 4.995007488766851e-06,
2413
+ "loss": 2.5117,
2414
+ "step": 40100
2415
+ },
2416
+ {
2417
+ "epoch": 5.02,
2418
+ "learning_rate": 4.982526210683974e-06,
2419
+ "loss": 2.5144,
2420
+ "step": 40200
2421
+ },
2422
+ {
2423
+ "epoch": 5.03,
2424
+ "learning_rate": 4.970044932601098e-06,
2425
+ "loss": 2.4917,
2426
+ "step": 40300
2427
+ },
2428
+ {
2429
+ "epoch": 5.04,
2430
+ "learning_rate": 4.9575636545182235e-06,
2431
+ "loss": 2.5242,
2432
+ "step": 40400
2433
+ },
2434
+ {
2435
+ "epoch": 5.05,
2436
+ "learning_rate": 4.945082376435347e-06,
2437
+ "loss": 2.5614,
2438
+ "step": 40500
2439
+ },
2440
+ {
2441
+ "epoch": 5.07,
2442
+ "learning_rate": 4.932601098352472e-06,
2443
+ "loss": 2.5885,
2444
+ "step": 40600
2445
+ },
2446
+ {
2447
+ "epoch": 5.08,
2448
+ "learning_rate": 4.920119820269596e-06,
2449
+ "loss": 2.5738,
2450
+ "step": 40700
2451
+ },
2452
+ {
2453
+ "epoch": 5.09,
2454
+ "learning_rate": 4.9076385421867204e-06,
2455
+ "loss": 2.5221,
2456
+ "step": 40800
2457
+ },
2458
+ {
2459
+ "epoch": 5.1,
2460
+ "learning_rate": 4.895157264103845e-06,
2461
+ "loss": 2.579,
2462
+ "step": 40900
2463
+ },
2464
+ {
2465
+ "epoch": 5.12,
2466
+ "learning_rate": 4.882675986020969e-06,
2467
+ "loss": 2.5234,
2468
+ "step": 41000
2469
+ },
2470
+ {
2471
+ "epoch": 5.13,
2472
+ "learning_rate": 4.870194707938093e-06,
2473
+ "loss": 2.5614,
2474
+ "step": 41100
2475
+ },
2476
+ {
2477
+ "epoch": 5.14,
2478
+ "learning_rate": 4.857713429855217e-06,
2479
+ "loss": 2.5128,
2480
+ "step": 41200
2481
+ },
2482
+ {
2483
+ "epoch": 5.15,
2484
+ "learning_rate": 4.845232151772342e-06,
2485
+ "loss": 2.4828,
2486
+ "step": 41300
2487
+ },
2488
+ {
2489
+ "epoch": 5.17,
2490
+ "learning_rate": 4.832750873689466e-06,
2491
+ "loss": 2.5557,
2492
+ "step": 41400
2493
+ },
2494
+ {
2495
+ "epoch": 5.18,
2496
+ "learning_rate": 4.82026959560659e-06,
2497
+ "loss": 2.556,
2498
+ "step": 41500
2499
+ },
2500
+ {
2501
+ "epoch": 5.19,
2502
+ "learning_rate": 4.807788317523715e-06,
2503
+ "loss": 2.5401,
2504
+ "step": 41600
2505
+ },
2506
+ {
2507
+ "epoch": 5.2,
2508
+ "learning_rate": 4.7953070394408394e-06,
2509
+ "loss": 2.501,
2510
+ "step": 41700
2511
+ },
2512
+ {
2513
+ "epoch": 5.22,
2514
+ "learning_rate": 4.782825761357963e-06,
2515
+ "loss": 2.4316,
2516
+ "step": 41800
2517
+ },
2518
+ {
2519
+ "epoch": 5.23,
2520
+ "learning_rate": 4.770344483275088e-06,
2521
+ "loss": 2.5133,
2522
+ "step": 41900
2523
+ },
2524
+ {
2525
+ "epoch": 5.24,
2526
+ "learning_rate": 4.757863205192212e-06,
2527
+ "loss": 2.574,
2528
+ "step": 42000
2529
+ },
2530
+ {
2531
+ "epoch": 5.25,
2532
+ "learning_rate": 4.745381927109336e-06,
2533
+ "loss": 2.6097,
2534
+ "step": 42100
2535
+ },
2536
+ {
2537
+ "epoch": 5.27,
2538
+ "learning_rate": 4.732900649026461e-06,
2539
+ "loss": 2.5314,
2540
+ "step": 42200
2541
+ },
2542
+ {
2543
+ "epoch": 5.28,
2544
+ "learning_rate": 4.720419370943585e-06,
2545
+ "loss": 2.5095,
2546
+ "step": 42300
2547
+ },
2548
+ {
2549
+ "epoch": 5.29,
2550
+ "learning_rate": 4.707938092860709e-06,
2551
+ "loss": 2.5508,
2552
+ "step": 42400
2553
+ },
2554
+ {
2555
+ "epoch": 5.3,
2556
+ "learning_rate": 4.695456814777833e-06,
2557
+ "loss": 2.4751,
2558
+ "step": 42500
2559
+ },
2560
+ {
2561
+ "epoch": 5.32,
2562
+ "learning_rate": 4.6829755366949584e-06,
2563
+ "loss": 2.5443,
2564
+ "step": 42600
2565
+ },
2566
+ {
2567
+ "epoch": 5.33,
2568
+ "learning_rate": 4.670494258612082e-06,
2569
+ "loss": 2.5164,
2570
+ "step": 42700
2571
+ },
2572
+ {
2573
+ "epoch": 5.34,
2574
+ "learning_rate": 4.658012980529207e-06,
2575
+ "loss": 2.5643,
2576
+ "step": 42800
2577
+ },
2578
+ {
2579
+ "epoch": 5.35,
2580
+ "learning_rate": 4.645531702446331e-06,
2581
+ "loss": 2.5964,
2582
+ "step": 42900
2583
+ },
2584
+ {
2585
+ "epoch": 5.37,
2586
+ "learning_rate": 4.633050424363455e-06,
2587
+ "loss": 2.5625,
2588
+ "step": 43000
2589
+ },
2590
+ {
2591
+ "epoch": 5.38,
2592
+ "learning_rate": 4.62056914628058e-06,
2593
+ "loss": 2.5365,
2594
+ "step": 43100
2595
+ },
2596
+ {
2597
+ "epoch": 5.39,
2598
+ "learning_rate": 4.608087868197704e-06,
2599
+ "loss": 2.5633,
2600
+ "step": 43200
2601
+ },
2602
+ {
2603
+ "epoch": 5.4,
2604
+ "learning_rate": 4.595606590114828e-06,
2605
+ "loss": 2.4372,
2606
+ "step": 43300
2607
+ },
2608
+ {
2609
+ "epoch": 5.42,
2610
+ "learning_rate": 4.583125312031952e-06,
2611
+ "loss": 2.5273,
2612
+ "step": 43400
2613
+ },
2614
+ {
2615
+ "epoch": 5.43,
2616
+ "learning_rate": 4.5706440339490766e-06,
2617
+ "loss": 2.4731,
2618
+ "step": 43500
2619
+ },
2620
+ {
2621
+ "epoch": 5.44,
2622
+ "learning_rate": 4.558162755866201e-06,
2623
+ "loss": 2.4437,
2624
+ "step": 43600
2625
+ },
2626
+ {
2627
+ "epoch": 5.45,
2628
+ "learning_rate": 4.545681477783325e-06,
2629
+ "loss": 2.5579,
2630
+ "step": 43700
2631
+ },
2632
+ {
2633
+ "epoch": 5.47,
2634
+ "learning_rate": 4.53320019970045e-06,
2635
+ "loss": 2.4914,
2636
+ "step": 43800
2637
+ },
2638
+ {
2639
+ "epoch": 5.48,
2640
+ "learning_rate": 4.520718921617574e-06,
2641
+ "loss": 2.532,
2642
+ "step": 43900
2643
+ },
2644
+ {
2645
+ "epoch": 5.49,
2646
+ "learning_rate": 4.508237643534698e-06,
2647
+ "loss": 2.5552,
2648
+ "step": 44000
2649
+ },
2650
+ {
2651
+ "epoch": 5.5,
2652
+ "learning_rate": 4.495756365451823e-06,
2653
+ "loss": 2.6329,
2654
+ "step": 44100
2655
+ },
2656
+ {
2657
+ "epoch": 5.52,
2658
+ "learning_rate": 4.483275087368947e-06,
2659
+ "loss": 2.5213,
2660
+ "step": 44200
2661
+ },
2662
+ {
2663
+ "epoch": 5.53,
2664
+ "learning_rate": 4.470793809286071e-06,
2665
+ "loss": 2.4853,
2666
+ "step": 44300
2667
+ },
2668
+ {
2669
+ "epoch": 5.54,
2670
+ "learning_rate": 4.4583125312031956e-06,
2671
+ "loss": 2.5358,
2672
+ "step": 44400
2673
+ },
2674
+ {
2675
+ "epoch": 5.55,
2676
+ "learning_rate": 4.44583125312032e-06,
2677
+ "loss": 2.4875,
2678
+ "step": 44500
2679
+ },
2680
+ {
2681
+ "epoch": 5.57,
2682
+ "learning_rate": 4.433349975037444e-06,
2683
+ "loss": 2.5111,
2684
+ "step": 44600
2685
+ },
2686
+ {
2687
+ "epoch": 5.58,
2688
+ "learning_rate": 4.420868696954568e-06,
2689
+ "loss": 2.5036,
2690
+ "step": 44700
2691
+ },
2692
+ {
2693
+ "epoch": 5.59,
2694
+ "learning_rate": 4.408387418871693e-06,
2695
+ "loss": 2.512,
2696
+ "step": 44800
2697
+ },
2698
+ {
2699
+ "epoch": 5.6,
2700
+ "learning_rate": 4.395906140788817e-06,
2701
+ "loss": 2.5135,
2702
+ "step": 44900
2703
+ },
2704
+ {
2705
+ "epoch": 5.62,
2706
+ "learning_rate": 4.383424862705942e-06,
2707
+ "loss": 2.5813,
2708
+ "step": 45000
2709
+ },
2710
+ {
2711
+ "epoch": 5.63,
2712
+ "learning_rate": 4.370943584623066e-06,
2713
+ "loss": 2.5975,
2714
+ "step": 45100
2715
+ },
2716
+ {
2717
+ "epoch": 5.64,
2718
+ "learning_rate": 4.3584623065401895e-06,
2719
+ "loss": 2.4869,
2720
+ "step": 45200
2721
+ },
2722
+ {
2723
+ "epoch": 5.65,
2724
+ "learning_rate": 4.3459810284573146e-06,
2725
+ "loss": 2.5271,
2726
+ "step": 45300
2727
+ },
2728
+ {
2729
+ "epoch": 5.67,
2730
+ "learning_rate": 4.333499750374439e-06,
2731
+ "loss": 2.5598,
2732
+ "step": 45400
2733
+ },
2734
+ {
2735
+ "epoch": 5.68,
2736
+ "learning_rate": 4.321018472291563e-06,
2737
+ "loss": 2.5388,
2738
+ "step": 45500
2739
+ },
2740
+ {
2741
+ "epoch": 5.69,
2742
+ "learning_rate": 4.308537194208687e-06,
2743
+ "loss": 2.5331,
2744
+ "step": 45600
2745
+ },
2746
+ {
2747
+ "epoch": 5.7,
2748
+ "learning_rate": 4.2960559161258115e-06,
2749
+ "loss": 2.5581,
2750
+ "step": 45700
2751
+ },
2752
+ {
2753
+ "epoch": 5.72,
2754
+ "learning_rate": 4.283574638042936e-06,
2755
+ "loss": 2.5059,
2756
+ "step": 45800
2757
+ },
2758
+ {
2759
+ "epoch": 5.73,
2760
+ "learning_rate": 4.27109335996006e-06,
2761
+ "loss": 2.5461,
2762
+ "step": 45900
2763
+ },
2764
+ {
2765
+ "epoch": 5.74,
2766
+ "learning_rate": 4.258612081877185e-06,
2767
+ "loss": 2.5158,
2768
+ "step": 46000
2769
+ },
2770
+ {
2771
+ "epoch": 5.75,
2772
+ "learning_rate": 4.2461308037943085e-06,
2773
+ "loss": 2.5132,
2774
+ "step": 46100
2775
+ },
2776
+ {
2777
+ "epoch": 5.77,
2778
+ "learning_rate": 4.233649525711433e-06,
2779
+ "loss": 2.5099,
2780
+ "step": 46200
2781
+ },
2782
+ {
2783
+ "epoch": 5.78,
2784
+ "learning_rate": 4.221168247628558e-06,
2785
+ "loss": 2.5432,
2786
+ "step": 46300
2787
+ },
2788
+ {
2789
+ "epoch": 5.79,
2790
+ "learning_rate": 4.208686969545682e-06,
2791
+ "loss": 2.5628,
2792
+ "step": 46400
2793
+ },
2794
+ {
2795
+ "epoch": 5.8,
2796
+ "learning_rate": 4.196205691462806e-06,
2797
+ "loss": 2.5235,
2798
+ "step": 46500
2799
+ },
2800
+ {
2801
+ "epoch": 5.82,
2802
+ "learning_rate": 4.1837244133799305e-06,
2803
+ "loss": 2.5383,
2804
+ "step": 46600
2805
+ },
2806
+ {
2807
+ "epoch": 5.83,
2808
+ "learning_rate": 4.171243135297055e-06,
2809
+ "loss": 2.4903,
2810
+ "step": 46700
2811
+ },
2812
+ {
2813
+ "epoch": 5.84,
2814
+ "learning_rate": 4.158761857214179e-06,
2815
+ "loss": 2.5872,
2816
+ "step": 46800
2817
+ },
2818
+ {
2819
+ "epoch": 5.85,
2820
+ "learning_rate": 4.146280579131303e-06,
2821
+ "loss": 2.5354,
2822
+ "step": 46900
2823
+ },
2824
+ {
2825
+ "epoch": 5.87,
2826
+ "learning_rate": 4.1337993010484274e-06,
2827
+ "loss": 2.5798,
2828
+ "step": 47000
2829
+ },
2830
+ {
2831
+ "epoch": 5.88,
2832
+ "learning_rate": 4.121318022965552e-06,
2833
+ "loss": 2.5454,
2834
+ "step": 47100
2835
+ },
2836
+ {
2837
+ "epoch": 5.89,
2838
+ "learning_rate": 4.108836744882677e-06,
2839
+ "loss": 2.5374,
2840
+ "step": 47200
2841
+ },
2842
+ {
2843
+ "epoch": 5.9,
2844
+ "learning_rate": 4.096355466799801e-06,
2845
+ "loss": 2.5371,
2846
+ "step": 47300
2847
+ },
2848
+ {
2849
+ "epoch": 5.92,
2850
+ "learning_rate": 4.083874188716924e-06,
2851
+ "loss": 2.4904,
2852
+ "step": 47400
2853
+ },
2854
+ {
2855
+ "epoch": 5.93,
2856
+ "learning_rate": 4.0713929106340495e-06,
2857
+ "loss": 2.5545,
2858
+ "step": 47500
2859
+ },
2860
+ {
2861
+ "epoch": 5.94,
2862
+ "learning_rate": 4.058911632551174e-06,
2863
+ "loss": 2.5668,
2864
+ "step": 47600
2865
+ },
2866
+ {
2867
+ "epoch": 5.95,
2868
+ "learning_rate": 4.046430354468298e-06,
2869
+ "loss": 2.5541,
2870
+ "step": 47700
2871
+ },
2872
+ {
2873
+ "epoch": 5.97,
2874
+ "learning_rate": 4.033949076385422e-06,
2875
+ "loss": 2.5027,
2876
+ "step": 47800
2877
+ },
2878
+ {
2879
+ "epoch": 5.98,
2880
+ "learning_rate": 4.0214677983025464e-06,
2881
+ "loss": 2.5735,
2882
+ "step": 47900
2883
+ },
2884
+ {
2885
+ "epoch": 5.99,
2886
+ "learning_rate": 4.008986520219671e-06,
2887
+ "loss": 2.5358,
2888
+ "step": 48000
2889
+ },
2890
+ {
2891
+ "epoch": 6.0,
2892
+ "learning_rate": 3.996505242136795e-06,
2893
+ "loss": 2.4827,
2894
+ "step": 48100
2895
+ },
2896
+ {
2897
+ "epoch": 6.02,
2898
+ "learning_rate": 3.98402396405392e-06,
2899
+ "loss": 2.4968,
2900
+ "step": 48200
2901
+ },
2902
+ {
2903
+ "epoch": 6.03,
2904
+ "learning_rate": 3.971542685971043e-06,
2905
+ "loss": 2.492,
2906
+ "step": 48300
2907
+ },
2908
+ {
2909
+ "epoch": 6.04,
2910
+ "learning_rate": 3.959061407888168e-06,
2911
+ "loss": 2.5652,
2912
+ "step": 48400
2913
+ },
2914
+ {
2915
+ "epoch": 6.05,
2916
+ "learning_rate": 3.946580129805293e-06,
2917
+ "loss": 2.5951,
2918
+ "step": 48500
2919
+ },
2920
+ {
2921
+ "epoch": 6.07,
2922
+ "learning_rate": 3.934098851722417e-06,
2923
+ "loss": 2.5054,
2924
+ "step": 48600
2925
+ },
2926
+ {
2927
+ "epoch": 6.08,
2928
+ "learning_rate": 3.921617573639541e-06,
2929
+ "loss": 2.5064,
2930
+ "step": 48700
2931
+ },
2932
+ {
2933
+ "epoch": 6.09,
2934
+ "learning_rate": 3.9091362955566654e-06,
2935
+ "loss": 2.517,
2936
+ "step": 48800
2937
+ },
2938
+ {
2939
+ "epoch": 6.1,
2940
+ "learning_rate": 3.89665501747379e-06,
2941
+ "loss": 2.4971,
2942
+ "step": 48900
2943
+ },
2944
+ {
2945
+ "epoch": 6.12,
2946
+ "learning_rate": 3.884173739390914e-06,
2947
+ "loss": 2.4405,
2948
+ "step": 49000
2949
+ },
2950
+ {
2951
+ "epoch": 6.13,
2952
+ "learning_rate": 3.871692461308038e-06,
2953
+ "loss": 2.4993,
2954
+ "step": 49100
2955
+ },
2956
+ {
2957
+ "epoch": 6.14,
2958
+ "learning_rate": 3.859211183225162e-06,
2959
+ "loss": 2.4815,
2960
+ "step": 49200
2961
+ },
2962
+ {
2963
+ "epoch": 6.15,
2964
+ "learning_rate": 3.846729905142287e-06,
2965
+ "loss": 2.4978,
2966
+ "step": 49300
2967
+ },
2968
+ {
2969
+ "epoch": 6.17,
2970
+ "learning_rate": 3.834248627059411e-06,
2971
+ "loss": 2.5125,
2972
+ "step": 49400
2973
+ },
2974
+ {
2975
+ "epoch": 6.18,
2976
+ "learning_rate": 3.821767348976536e-06,
2977
+ "loss": 2.5636,
2978
+ "step": 49500
2979
+ },
2980
+ {
2981
+ "epoch": 6.19,
2982
+ "learning_rate": 3.8092860708936598e-06,
2983
+ "loss": 2.5993,
2984
+ "step": 49600
2985
+ },
2986
+ {
2987
+ "epoch": 6.2,
2988
+ "learning_rate": 3.7968047928107844e-06,
2989
+ "loss": 2.4773,
2990
+ "step": 49700
2991
+ },
2992
+ {
2993
+ "epoch": 6.22,
2994
+ "learning_rate": 3.7843235147279082e-06,
2995
+ "loss": 2.46,
2996
+ "step": 49800
2997
+ },
2998
+ {
2999
+ "epoch": 6.23,
3000
+ "learning_rate": 3.771842236645033e-06,
3001
+ "loss": 2.538,
3002
+ "step": 49900
3003
+ },
3004
+ {
3005
+ "epoch": 6.24,
3006
+ "learning_rate": 3.759360958562157e-06,
3007
+ "loss": 2.5968,
3008
+ "step": 50000
3009
+ },
3010
+ {
3011
+ "epoch": 6.25,
3012
+ "learning_rate": 3.7468796804792814e-06,
3013
+ "loss": 2.5301,
3014
+ "step": 50100
3015
+ },
3016
+ {
3017
+ "epoch": 6.27,
3018
+ "learning_rate": 3.734398402396406e-06,
3019
+ "loss": 2.5282,
3020
+ "step": 50200
3021
+ },
3022
+ {
3023
+ "epoch": 6.28,
3024
+ "learning_rate": 3.72191712431353e-06,
3025
+ "loss": 2.5195,
3026
+ "step": 50300
3027
+ },
3028
+ {
3029
+ "epoch": 6.29,
3030
+ "learning_rate": 3.7094358462306545e-06,
3031
+ "loss": 2.5422,
3032
+ "step": 50400
3033
+ },
3034
+ {
3035
+ "epoch": 6.3,
3036
+ "learning_rate": 3.6969545681477787e-06,
3037
+ "loss": 2.5415,
3038
+ "step": 50500
3039
+ },
3040
+ {
3041
+ "epoch": 6.32,
3042
+ "learning_rate": 3.6844732900649026e-06,
3043
+ "loss": 2.5067,
3044
+ "step": 50600
3045
+ },
3046
+ {
3047
+ "epoch": 6.33,
3048
+ "learning_rate": 3.6719920119820272e-06,
3049
+ "loss": 2.514,
3050
+ "step": 50700
3051
+ },
3052
+ {
3053
+ "epoch": 6.34,
3054
+ "learning_rate": 3.6595107338991515e-06,
3055
+ "loss": 2.5947,
3056
+ "step": 50800
3057
+ },
3058
+ {
3059
+ "epoch": 6.35,
3060
+ "learning_rate": 3.647029455816276e-06,
3061
+ "loss": 2.4832,
3062
+ "step": 50900
3063
+ },
3064
+ {
3065
+ "epoch": 6.37,
3066
+ "learning_rate": 3.6345481777334004e-06,
3067
+ "loss": 2.5519,
3068
+ "step": 51000
3069
+ },
3070
+ {
3071
+ "epoch": 6.38,
3072
+ "learning_rate": 3.622066899650524e-06,
3073
+ "loss": 2.5448,
3074
+ "step": 51100
3075
+ },
3076
+ {
3077
+ "epoch": 6.39,
3078
+ "learning_rate": 3.609585621567649e-06,
3079
+ "loss": 2.4973,
3080
+ "step": 51200
3081
+ },
3082
+ {
3083
+ "epoch": 6.4,
3084
+ "learning_rate": 3.597104343484773e-06,
3085
+ "loss": 2.55,
3086
+ "step": 51300
3087
+ },
3088
+ {
3089
+ "epoch": 6.42,
3090
+ "learning_rate": 3.5846230654018977e-06,
3091
+ "loss": 2.4911,
3092
+ "step": 51400
3093
+ },
3094
+ {
3095
+ "epoch": 6.43,
3096
+ "learning_rate": 3.5721417873190215e-06,
3097
+ "loss": 2.4954,
3098
+ "step": 51500
3099
+ },
3100
+ {
3101
+ "epoch": 6.44,
3102
+ "learning_rate": 3.5596605092361458e-06,
3103
+ "loss": 2.5314,
3104
+ "step": 51600
3105
+ },
3106
+ {
3107
+ "epoch": 6.45,
3108
+ "learning_rate": 3.5471792311532704e-06,
3109
+ "loss": 2.474,
3110
+ "step": 51700
3111
+ },
3112
+ {
3113
+ "epoch": 6.47,
3114
+ "learning_rate": 3.5346979530703947e-06,
3115
+ "loss": 2.5239,
3116
+ "step": 51800
3117
+ },
3118
+ {
3119
+ "epoch": 6.48,
3120
+ "learning_rate": 3.5222166749875193e-06,
3121
+ "loss": 2.5104,
3122
+ "step": 51900
3123
+ },
3124
+ {
3125
+ "epoch": 6.49,
3126
+ "learning_rate": 3.509735396904643e-06,
3127
+ "loss": 2.5841,
3128
+ "step": 52000
3129
+ },
3130
+ {
3131
+ "epoch": 6.5,
3132
+ "learning_rate": 3.497254118821768e-06,
3133
+ "loss": 2.5434,
3134
+ "step": 52100
3135
+ },
3136
+ {
3137
+ "epoch": 6.52,
3138
+ "learning_rate": 3.484772840738892e-06,
3139
+ "loss": 2.5161,
3140
+ "step": 52200
3141
+ },
3142
+ {
3143
+ "epoch": 6.53,
3144
+ "learning_rate": 3.472291562656016e-06,
3145
+ "loss": 2.5273,
3146
+ "step": 52300
3147
+ },
3148
+ {
3149
+ "epoch": 6.54,
3150
+ "learning_rate": 3.4598102845731405e-06,
3151
+ "loss": 2.581,
3152
+ "step": 52400
3153
+ },
3154
+ {
3155
+ "epoch": 6.55,
3156
+ "learning_rate": 3.4473290064902648e-06,
3157
+ "loss": 2.5352,
3158
+ "step": 52500
3159
+ },
3160
+ {
3161
+ "epoch": 6.57,
3162
+ "learning_rate": 3.4348477284073894e-06,
3163
+ "loss": 2.5912,
3164
+ "step": 52600
3165
+ },
3166
+ {
3167
+ "epoch": 6.58,
3168
+ "learning_rate": 3.4223664503245137e-06,
3169
+ "loss": 2.5162,
3170
+ "step": 52700
3171
+ },
3172
+ {
3173
+ "epoch": 6.59,
3174
+ "learning_rate": 3.4098851722416375e-06,
3175
+ "loss": 2.5289,
3176
+ "step": 52800
3177
+ },
3178
+ {
3179
+ "epoch": 6.6,
3180
+ "learning_rate": 3.397403894158762e-06,
3181
+ "loss": 2.4994,
3182
+ "step": 52900
3183
+ },
3184
+ {
3185
+ "epoch": 6.62,
3186
+ "learning_rate": 3.3849226160758864e-06,
3187
+ "loss": 2.5098,
3188
+ "step": 53000
3189
+ },
3190
+ {
3191
+ "epoch": 6.63,
3192
+ "learning_rate": 3.372441337993011e-06,
3193
+ "loss": 2.5013,
3194
+ "step": 53100
3195
+ },
3196
+ {
3197
+ "epoch": 6.64,
3198
+ "learning_rate": 3.359960059910135e-06,
3199
+ "loss": 2.5134,
3200
+ "step": 53200
3201
+ },
3202
+ {
3203
+ "epoch": 6.65,
3204
+ "learning_rate": 3.347478781827259e-06,
3205
+ "loss": 2.5036,
3206
+ "step": 53300
3207
+ },
3208
+ {
3209
+ "epoch": 6.66,
3210
+ "learning_rate": 3.3349975037443838e-06,
3211
+ "loss": 2.5203,
3212
+ "step": 53400
3213
+ },
3214
+ {
3215
+ "epoch": 6.68,
3216
+ "learning_rate": 3.322516225661508e-06,
3217
+ "loss": 2.4901,
3218
+ "step": 53500
3219
+ },
3220
+ {
3221
+ "epoch": 6.69,
3222
+ "learning_rate": 3.3100349475786327e-06,
3223
+ "loss": 2.5451,
3224
+ "step": 53600
3225
+ },
3226
+ {
3227
+ "epoch": 6.7,
3228
+ "learning_rate": 3.2975536694957565e-06,
3229
+ "loss": 2.5166,
3230
+ "step": 53700
3231
+ },
3232
+ {
3233
+ "epoch": 6.71,
3234
+ "learning_rate": 3.2850723914128807e-06,
3235
+ "loss": 2.4924,
3236
+ "step": 53800
3237
+ },
3238
+ {
3239
+ "epoch": 6.73,
3240
+ "learning_rate": 3.2725911133300054e-06,
3241
+ "loss": 2.5529,
3242
+ "step": 53900
3243
+ },
3244
+ {
3245
+ "epoch": 6.74,
3246
+ "learning_rate": 3.2601098352471296e-06,
3247
+ "loss": 2.5539,
3248
+ "step": 54000
3249
+ },
3250
+ {
3251
+ "epoch": 6.75,
3252
+ "learning_rate": 3.2476285571642543e-06,
3253
+ "loss": 2.5476,
3254
+ "step": 54100
3255
+ },
3256
+ {
3257
+ "epoch": 6.76,
3258
+ "learning_rate": 3.235147279081378e-06,
3259
+ "loss": 2.5324,
3260
+ "step": 54200
3261
+ },
3262
+ {
3263
+ "epoch": 6.78,
3264
+ "learning_rate": 3.2226660009985023e-06,
3265
+ "loss": 2.5427,
3266
+ "step": 54300
3267
+ },
3268
+ {
3269
+ "epoch": 6.79,
3270
+ "learning_rate": 3.210184722915627e-06,
3271
+ "loss": 2.5474,
3272
+ "step": 54400
3273
+ },
3274
+ {
3275
+ "epoch": 6.8,
3276
+ "learning_rate": 3.197703444832751e-06,
3277
+ "loss": 2.5174,
3278
+ "step": 54500
3279
+ },
3280
+ {
3281
+ "epoch": 6.81,
3282
+ "learning_rate": 3.1852221667498755e-06,
3283
+ "loss": 2.5418,
3284
+ "step": 54600
3285
+ },
3286
+ {
3287
+ "epoch": 6.83,
3288
+ "learning_rate": 3.1727408886669997e-06,
3289
+ "loss": 2.4859,
3290
+ "step": 54700
3291
+ },
3292
+ {
3293
+ "epoch": 6.84,
3294
+ "learning_rate": 3.1602596105841244e-06,
3295
+ "loss": 2.471,
3296
+ "step": 54800
3297
+ },
3298
+ {
3299
+ "epoch": 6.85,
3300
+ "learning_rate": 3.1477783325012486e-06,
3301
+ "loss": 2.568,
3302
+ "step": 54900
3303
+ },
3304
+ {
3305
+ "epoch": 6.86,
3306
+ "learning_rate": 3.1352970544183724e-06,
3307
+ "loss": 2.537,
3308
+ "step": 55000
3309
+ },
3310
+ {
3311
+ "epoch": 6.88,
3312
+ "learning_rate": 3.122815776335497e-06,
3313
+ "loss": 2.5169,
3314
+ "step": 55100
3315
+ },
3316
+ {
3317
+ "epoch": 6.89,
3318
+ "learning_rate": 3.1103344982526213e-06,
3319
+ "loss": 2.5177,
3320
+ "step": 55200
3321
+ },
3322
+ {
3323
+ "epoch": 6.9,
3324
+ "learning_rate": 3.097853220169746e-06,
3325
+ "loss": 2.5503,
3326
+ "step": 55300
3327
+ },
3328
+ {
3329
+ "epoch": 6.91,
3330
+ "learning_rate": 3.08537194208687e-06,
3331
+ "loss": 2.4971,
3332
+ "step": 55400
3333
+ },
3334
+ {
3335
+ "epoch": 6.93,
3336
+ "learning_rate": 3.072890664003994e-06,
3337
+ "loss": 2.4498,
3338
+ "step": 55500
3339
+ },
3340
+ {
3341
+ "epoch": 6.94,
3342
+ "learning_rate": 3.0604093859211187e-06,
3343
+ "loss": 2.4854,
3344
+ "step": 55600
3345
+ },
3346
+ {
3347
+ "epoch": 6.95,
3348
+ "learning_rate": 3.047928107838243e-06,
3349
+ "loss": 2.5306,
3350
+ "step": 55700
3351
+ },
3352
+ {
3353
+ "epoch": 6.96,
3354
+ "learning_rate": 3.0354468297553676e-06,
3355
+ "loss": 2.4973,
3356
+ "step": 55800
3357
+ },
3358
+ {
3359
+ "epoch": 6.98,
3360
+ "learning_rate": 3.0229655516724914e-06,
3361
+ "loss": 2.474,
3362
+ "step": 55900
3363
+ },
3364
+ {
3365
+ "epoch": 6.99,
3366
+ "learning_rate": 3.0104842735896156e-06,
3367
+ "loss": 2.5066,
3368
+ "step": 56000
3369
+ },
3370
+ {
3371
+ "epoch": 7.0,
3372
+ "learning_rate": 2.9980029955067403e-06,
3373
+ "loss": 2.5626,
3374
+ "step": 56100
3375
+ },
3376
+ {
3377
+ "epoch": 7.01,
3378
+ "learning_rate": 2.985521717423864e-06,
3379
+ "loss": 2.5684,
3380
+ "step": 56200
3381
+ },
3382
+ {
3383
+ "epoch": 7.03,
3384
+ "learning_rate": 2.9730404393409888e-06,
3385
+ "loss": 2.5336,
3386
+ "step": 56300
3387
+ },
3388
+ {
3389
+ "epoch": 7.04,
3390
+ "learning_rate": 2.960559161258113e-06,
3391
+ "loss": 2.4971,
3392
+ "step": 56400
3393
+ },
3394
+ {
3395
+ "epoch": 7.05,
3396
+ "learning_rate": 2.9480778831752373e-06,
3397
+ "loss": 2.5267,
3398
+ "step": 56500
3399
+ },
3400
+ {
3401
+ "epoch": 7.06,
3402
+ "learning_rate": 2.935596605092362e-06,
3403
+ "loss": 2.472,
3404
+ "step": 56600
3405
+ },
3406
+ {
3407
+ "epoch": 7.08,
3408
+ "learning_rate": 2.9231153270094857e-06,
3409
+ "loss": 2.5235,
3410
+ "step": 56700
3411
+ },
3412
+ {
3413
+ "epoch": 7.09,
3414
+ "learning_rate": 2.9106340489266104e-06,
3415
+ "loss": 2.5225,
3416
+ "step": 56800
3417
+ },
3418
+ {
3419
+ "epoch": 7.1,
3420
+ "learning_rate": 2.8981527708437346e-06,
3421
+ "loss": 2.5145,
3422
+ "step": 56900
3423
+ },
3424
+ {
3425
+ "epoch": 7.11,
3426
+ "learning_rate": 2.885671492760859e-06,
3427
+ "loss": 2.4227,
3428
+ "step": 57000
3429
+ },
3430
+ {
3431
+ "epoch": 7.13,
3432
+ "learning_rate": 2.873190214677983e-06,
3433
+ "loss": 2.5316,
3434
+ "step": 57100
3435
+ },
3436
+ {
3437
+ "epoch": 7.14,
3438
+ "learning_rate": 2.8607089365951074e-06,
3439
+ "loss": 2.5347,
3440
+ "step": 57200
3441
+ },
3442
+ {
3443
+ "epoch": 7.15,
3444
+ "learning_rate": 2.848227658512232e-06,
3445
+ "loss": 2.5264,
3446
+ "step": 57300
3447
+ },
3448
+ {
3449
+ "epoch": 7.16,
3450
+ "learning_rate": 2.8357463804293563e-06,
3451
+ "loss": 2.5099,
3452
+ "step": 57400
3453
+ },
3454
+ {
3455
+ "epoch": 7.18,
3456
+ "learning_rate": 2.823265102346481e-06,
3457
+ "loss": 2.4761,
3458
+ "step": 57500
3459
+ },
3460
+ {
3461
+ "epoch": 7.19,
3462
+ "learning_rate": 2.8107838242636047e-06,
3463
+ "loss": 2.529,
3464
+ "step": 57600
3465
+ },
3466
+ {
3467
+ "epoch": 7.2,
3468
+ "learning_rate": 2.798302546180729e-06,
3469
+ "loss": 2.483,
3470
+ "step": 57700
3471
+ },
3472
+ {
3473
+ "epoch": 7.21,
3474
+ "learning_rate": 2.7858212680978536e-06,
3475
+ "loss": 2.4986,
3476
+ "step": 57800
3477
+ },
3478
+ {
3479
+ "epoch": 7.23,
3480
+ "learning_rate": 2.773339990014978e-06,
3481
+ "loss": 2.5386,
3482
+ "step": 57900
3483
+ },
3484
+ {
3485
+ "epoch": 7.24,
3486
+ "learning_rate": 2.760858711932102e-06,
3487
+ "loss": 2.4868,
3488
+ "step": 58000
3489
+ }
3490
+ ],
3491
+ "max_steps": 80120,
3492
+ "num_train_epochs": 10,
3493
+ "total_flos": 4.356850425234186e+17,
3494
+ "trial_name": null,
3495
+ "trial_params": null
3496
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:737e26502c6090960eb553ee6bb362b7c43cd9705f8a721737f7bf7282910dcd
3
+ size 2607