Hugging Face's logo

madhavsankar
/
qcpg-mscoco-bleurt-lr1e-4

Text2Text Generation
Transformers
PyTorch
t5
text-generation-inference
Model card Files Community
1
madhavsankar commited on
Commit
ad8f012
·
1 Parent(s): 8e687ac

Upload 12 files

Browse files
Files changed (12) hide show
  1. added_tokens.json +1 -0
  2. all_results.json +24 -0
  3. config.json +57 -0
  4. eval_results.json +13 -0
  5. pytorch_model.bin +3 -0
  6. special_tokens_map.json +1 -0
  7. spiece.model +3 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +1 -0
  10. train_results.json +14 -0
  11. trainer_state.json +2750 -0
  12. training_args.bin +3 -0
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"COND_SEMANTIC_SIM_10": 32159, "COND_SEMANTIC_SIM_55": 32169, "COND_SEMANTIC_SIM_40": 32165, "COND_MORPHOLOGICAL_DIV_75": 32135, "COND_SYNTACTIC_DIV_45": 32186, "COND_PHONOLOGICAL_DIV_80": 32156, "COND_LEXICAL_DIV_75": 32115, "COND_SYNTACTIC_DIV_65": 32191, "COND_SEMANTIC_SIM_35": 32164, "COND_SEMANTIC_SIM_75": 32173, "COND_SYNTACTIC_DIV_60": 32190, "COND_LEXICAL_DIV_70": 32114, "COND_LEXICAL_DIV_80": 32116, "COND_SYNTACTIC_DIV_70": 32192, "COND_LEXICAL_DIV_25": 32104, "COND_SEMANTIC_SIM_95": 32177, "COND_MORPHOLOGICAL_DIV_20": 32123, "COND_SEMANTIC_SIM_45": 32166, "COND_MORPHOLOGICAL_DIV_25": 32124, "COND_LEXICAL_DIV_0": 32100, "COND_PHONOLOGICAL_DIV_65": 32153, "COND_LEXICAL_DIV_15": 32102, "COND_MORPHOLOGICAL_DIV_0": 32119, "COND_MORPHOLOGICAL_DIV_35": 32126, "COND_MORPHOLOGICAL_DIV_15": 32122, "COND_MORPHOLOGICAL_DIV_50": 32130, "COND_MORPHOLOGICAL_DIV_90": 32138, "COND_PHONOLOGICAL_DIV_55": 32151, "COND_MORPHOLOGICAL_DIV_80": 32136, "COND_MORPHOLOGICAL_DIV_60": 32132, "COND_SEMANTIC_SIM_85": 32175, "COND_SYNTACTIC_DIV_40": 32185, "COND_SYNTACTIC_DIV_30": 32183, "COND_SEMANTIC_SIM_90": 32176, "COND_MORPHOLOGICAL_DIV_30": 32125, "COND_PHONOLOGICAL_DIV_50": 32150, "COND_SEMANTIC_SIM_50": 32168, "COND_SYNTACTIC_DIV_0": 32178, "COND_MORPHOLOGICAL_DIV_45": 32128, "COND_PHONOLOGICAL_DIV_60": 32152, "COND_PHONOLOGICAL_DIV_0": 32140, "COND_SEMANTIC_SIM_30": 32163, "COND_MORPHOLOGICAL_DIV_65": 32133, "COND_SYNTACTIC_DIV_25": 32182, "COND_PHONOLOGICAL_DIV_75": 32155, "COND_SYNTACTIC_DIV_55": 32189, "COND_SYNTACTIC_DIV_75": 32193, "COND_LEXICAL_DIV_10": 32101, "COND_LEXICAL_DIV_20": 32103, "COND_SEMANTIC_SIM_0": 32158, "COND_PHONOLOGICAL_DIV_10": 32141, "COND_SYNTACTIC_DIV_15": 32180, "COND_LEXICAL_DIV_50": 32110, "COND_MORPHOLOGICAL_DIV_5": 32129, "COND_SYNTACTIC_DIV_80": 32194, "COND_PHONOLOGICAL_DIV_45": 32148, "COND_LEXICAL_DIV_35": 32106, "COND_MORPHOLOGICAL_DIV_70": 32134, "COND_MORPHOLOGICAL_DIV_55": 32131, "COND_SYNTACTIC_DIV_10": 32179, "COND_MORPHOLOGICAL_DIV_85": 32137, "COND_LEXICAL_DIV_60": 32112, "COND_SEMANTIC_SIM_5": 32167, "COND_PHONOLOGICAL_DIV_20": 32143, "COND_SEMANTIC_SIM_65": 32171, "COND_MORPHOLOGICAL_DIV_10": 32120, "COND_MORPHOLOGICAL_DIV_100": 32121, "COND_MORPHOLOGICAL_DIV_40": 32127, "COND_SYNTACTIC_DIV_50": 32188, "COND_LEXICAL_DIV_55": 32111, "COND_PHONOLOGICAL_DIV_85": 32157, "COND_SEMANTIC_SIM_20": 32161, "COND_PHONOLOGICAL_DIV_35": 32146, "COND_SYNTACTIC_DIV_20": 32181, "COND_SEMANTIC_SIM_60": 32170, "COND_LEXICAL_DIV_85": 32117, "COND_LEXICAL_DIV_30": 32105, "COND_SYNTACTIC_DIV_35": 32184, "COND_SEMANTIC_SIM_70": 32172, "COND_LEXICAL_DIV_45": 32108, "COND_PHONOLOGICAL_DIV_30": 32145, "COND_LEXICAL_DIV_5": 32109, "COND_SYNTACTIC_DIV_5": 32187, "COND_PHONOLOGICAL_DIV_70": 32154, "COND_SEMANTIC_SIM_25": 32162, "COND_MORPHOLOGICAL_DIV_95": 32139, "COND_SEMANTIC_SIM_15": 32160, "COND_LEXICAL_DIV_90": 32118, "COND_PHONOLOGICAL_DIV_15": 32142, "COND_PHONOLOGICAL_DIV_40": 32147, "COND_PHONOLOGICAL_DIV_5": 32149, "COND_SEMANTIC_SIM_80": 32174, "COND_LEXICAL_DIV_65": 32113, "COND_LEXICAL_DIV_40": 32107, "COND_PHONOLOGICAL_DIV_25": 32144}
all_results.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_bleu": 11.7859,
4
+ "eval_gen_len": 15.7088,
5
+ "eval_loss": 1.7650222778320312,
6
+ "eval_mem_cpu_alloc_delta": 336457728,
7
+ "eval_mem_cpu_peaked_delta": 7503872,
8
+ "eval_mem_gpu_alloc_delta": 0,
9
+ "eval_mem_gpu_peaked_delta": 329739264,
10
+ "eval_runtime": 1333.5245,
11
+ "eval_samples": 14000,
12
+ "eval_samples_per_second": 10.498,
13
+ "init_mem_cpu_alloc_delta": 0,
14
+ "init_mem_cpu_peaked_delta": 0,
15
+ "init_mem_gpu_alloc_delta": 0,
16
+ "init_mem_gpu_peaked_delta": 0,
17
+ "train_mem_cpu_alloc_delta": -902914048,
18
+ "train_mem_cpu_peaked_delta": 902914048,
19
+ "train_mem_gpu_alloc_delta": 1783640576,
20
+ "train_mem_gpu_peaked_delta": 1783640064,
21
+ "train_runtime": 1.098,
22
+ "train_samples": 900000,
23
+ "train_samples_per_second": 204925.455
24
+ }
config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "t5-base",
3
+ "architectures": [
4
+ "T5ForConditionalGeneration"
5
+ ],
6
+ "d_ff": 3072,
7
+ "d_kv": 64,
8
+ "d_model": 768,
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": 12,
20
+ "num_heads": 12,
21
+ "num_layers": 12,
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": "4.6.1",
55
+ "use_cache": true,
56
+ "vocab_size": 32195
57
+ }
eval_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_bleu": 11.7859,
4
+ "eval_gen_len": 15.7088,
5
+ "eval_loss": 1.7650222778320312,
6
+ "eval_mem_cpu_alloc_delta": 336457728,
7
+ "eval_mem_cpu_peaked_delta": 7503872,
8
+ "eval_mem_gpu_alloc_delta": 0,
9
+ "eval_mem_gpu_peaked_delta": 329739264,
10
+ "eval_runtime": 1333.5245,
11
+ "eval_samples": 14000,
12
+ "eval_samples_per_second": 10.498
13
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa4c85ef9e4951f5e718cb6e873b06bb8bd066d2da9d8c4f0f0b8966f270fc3b
3
+ size 891938833
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "additional_special_tokens": ["COND_LEXICAL_DIV_0", "COND_LEXICAL_DIV_10", "COND_LEXICAL_DIV_15", "COND_LEXICAL_DIV_20", "COND_LEXICAL_DIV_25", "COND_LEXICAL_DIV_30", "COND_LEXICAL_DIV_35", "COND_LEXICAL_DIV_40", "COND_LEXICAL_DIV_45", "COND_LEXICAL_DIV_5", "COND_LEXICAL_DIV_50", "COND_LEXICAL_DIV_55", "COND_LEXICAL_DIV_60", "COND_LEXICAL_DIV_65", "COND_LEXICAL_DIV_70", "COND_LEXICAL_DIV_75", "COND_LEXICAL_DIV_80", "COND_LEXICAL_DIV_85", "COND_LEXICAL_DIV_90", "COND_MORPHOLOGICAL_DIV_0", "COND_MORPHOLOGICAL_DIV_10", "COND_MORPHOLOGICAL_DIV_100", "COND_MORPHOLOGICAL_DIV_15", "COND_MORPHOLOGICAL_DIV_20", "COND_MORPHOLOGICAL_DIV_25", "COND_MORPHOLOGICAL_DIV_30", "COND_MORPHOLOGICAL_DIV_35", "COND_MORPHOLOGICAL_DIV_40", "COND_MORPHOLOGICAL_DIV_45", "COND_MORPHOLOGICAL_DIV_5", "COND_MORPHOLOGICAL_DIV_50", "COND_MORPHOLOGICAL_DIV_55", "COND_MORPHOLOGICAL_DIV_60", "COND_MORPHOLOGICAL_DIV_65", "COND_MORPHOLOGICAL_DIV_70", "COND_MORPHOLOGICAL_DIV_75", "COND_MORPHOLOGICAL_DIV_80", "COND_MORPHOLOGICAL_DIV_85", "COND_MORPHOLOGICAL_DIV_90", "COND_MORPHOLOGICAL_DIV_95", "COND_PHONOLOGICAL_DIV_0", "COND_PHONOLOGICAL_DIV_10", "COND_PHONOLOGICAL_DIV_15", "COND_PHONOLOGICAL_DIV_20", "COND_PHONOLOGICAL_DIV_25", "COND_PHONOLOGICAL_DIV_30", "COND_PHONOLOGICAL_DIV_35", "COND_PHONOLOGICAL_DIV_40", "COND_PHONOLOGICAL_DIV_45", "COND_PHONOLOGICAL_DIV_5", "COND_PHONOLOGICAL_DIV_50", "COND_PHONOLOGICAL_DIV_55", "COND_PHONOLOGICAL_DIV_60", "COND_PHONOLOGICAL_DIV_65", "COND_PHONOLOGICAL_DIV_70", "COND_PHONOLOGICAL_DIV_75", "COND_PHONOLOGICAL_DIV_80", "COND_PHONOLOGICAL_DIV_85", "COND_SEMANTIC_SIM_0", "COND_SEMANTIC_SIM_10", "COND_SEMANTIC_SIM_15", "COND_SEMANTIC_SIM_20", "COND_SEMANTIC_SIM_25", "COND_SEMANTIC_SIM_30", "COND_SEMANTIC_SIM_35", "COND_SEMANTIC_SIM_40", "COND_SEMANTIC_SIM_45", "COND_SEMANTIC_SIM_5", "COND_SEMANTIC_SIM_50", "COND_SEMANTIC_SIM_55", "COND_SEMANTIC_SIM_60", "COND_SEMANTIC_SIM_65", "COND_SEMANTIC_SIM_70", "COND_SEMANTIC_SIM_75", "COND_SEMANTIC_SIM_80", "COND_SEMANTIC_SIM_85", "COND_SEMANTIC_SIM_90", "COND_SEMANTIC_SIM_95", "COND_SYNTACTIC_DIV_0", "COND_SYNTACTIC_DIV_10", "COND_SYNTACTIC_DIV_15", "COND_SYNTACTIC_DIV_20", "COND_SYNTACTIC_DIV_25", "COND_SYNTACTIC_DIV_30", "COND_SYNTACTIC_DIV_35", "COND_SYNTACTIC_DIV_40", "COND_SYNTACTIC_DIV_45", "COND_SYNTACTIC_DIV_5", "COND_SYNTACTIC_DIV_50", "COND_SYNTACTIC_DIV_55", "COND_SYNTACTIC_DIV_60", "COND_SYNTACTIC_DIV_65", "COND_SYNTACTIC_DIV_70", "COND_SYNTACTIC_DIV_75", "COND_SYNTACTIC_DIV_80"]}
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86
3
+ size 791656
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0, "additional_special_tokens": ["COND_LEXICAL_DIV_0", "COND_LEXICAL_DIV_10", "COND_LEXICAL_DIV_15", "COND_LEXICAL_DIV_20", "COND_LEXICAL_DIV_25", "COND_LEXICAL_DIV_30", "COND_LEXICAL_DIV_35", "COND_LEXICAL_DIV_40", "COND_LEXICAL_DIV_45", "COND_LEXICAL_DIV_5", "COND_LEXICAL_DIV_50", "COND_LEXICAL_DIV_55", "COND_LEXICAL_DIV_60", "COND_LEXICAL_DIV_65", "COND_LEXICAL_DIV_70", "COND_LEXICAL_DIV_75", "COND_LEXICAL_DIV_80", "COND_LEXICAL_DIV_85", "COND_LEXICAL_DIV_90", "COND_MORPHOLOGICAL_DIV_0", "COND_MORPHOLOGICAL_DIV_10", "COND_MORPHOLOGICAL_DIV_100", "COND_MORPHOLOGICAL_DIV_15", "COND_MORPHOLOGICAL_DIV_20", "COND_MORPHOLOGICAL_DIV_25", "COND_MORPHOLOGICAL_DIV_30", "COND_MORPHOLOGICAL_DIV_35", "COND_MORPHOLOGICAL_DIV_40", "COND_MORPHOLOGICAL_DIV_45", "COND_MORPHOLOGICAL_DIV_5", "COND_MORPHOLOGICAL_DIV_50", "COND_MORPHOLOGICAL_DIV_55", "COND_MORPHOLOGICAL_DIV_60", "COND_MORPHOLOGICAL_DIV_65", "COND_MORPHOLOGICAL_DIV_70", "COND_MORPHOLOGICAL_DIV_75", "COND_MORPHOLOGICAL_DIV_80", "COND_MORPHOLOGICAL_DIV_85", "COND_MORPHOLOGICAL_DIV_90", "COND_MORPHOLOGICAL_DIV_95", "COND_PHONOLOGICAL_DIV_0", "COND_PHONOLOGICAL_DIV_10", "COND_PHONOLOGICAL_DIV_15", "COND_PHONOLOGICAL_DIV_20", "COND_PHONOLOGICAL_DIV_25", "COND_PHONOLOGICAL_DIV_30", "COND_PHONOLOGICAL_DIV_35", "COND_PHONOLOGICAL_DIV_40", "COND_PHONOLOGICAL_DIV_45", "COND_PHONOLOGICAL_DIV_5", "COND_PHONOLOGICAL_DIV_50", "COND_PHONOLOGICAL_DIV_55", "COND_PHONOLOGICAL_DIV_60", "COND_PHONOLOGICAL_DIV_65", "COND_PHONOLOGICAL_DIV_70", "COND_PHONOLOGICAL_DIV_75", "COND_PHONOLOGICAL_DIV_80", "COND_PHONOLOGICAL_DIV_85", "COND_SEMANTIC_SIM_0", "COND_SEMANTIC_SIM_10", "COND_SEMANTIC_SIM_15", "COND_SEMANTIC_SIM_20", "COND_SEMANTIC_SIM_25", "COND_SEMANTIC_SIM_30", "COND_SEMANTIC_SIM_35", "COND_SEMANTIC_SIM_40", "COND_SEMANTIC_SIM_45", "COND_SEMANTIC_SIM_5", "COND_SEMANTIC_SIM_50", "COND_SEMANTIC_SIM_55", "COND_SEMANTIC_SIM_60", "COND_SEMANTIC_SIM_65", "COND_SEMANTIC_SIM_70", "COND_SEMANTIC_SIM_75", "COND_SEMANTIC_SIM_80", "COND_SEMANTIC_SIM_85", "COND_SEMANTIC_SIM_90", "COND_SEMANTIC_SIM_95", "COND_SYNTACTIC_DIV_0", "COND_SYNTACTIC_DIV_10", "COND_SYNTACTIC_DIV_15", "COND_SYNTACTIC_DIV_20", "COND_SYNTACTIC_DIV_25", "COND_SYNTACTIC_DIV_30", "COND_SYNTACTIC_DIV_35", "COND_SYNTACTIC_DIV_40", "COND_SYNTACTIC_DIV_45", "COND_SYNTACTIC_DIV_5", "COND_SYNTACTIC_DIV_50", "COND_SYNTACTIC_DIV_55", "COND_SYNTACTIC_DIV_60", "COND_SYNTACTIC_DIV_65", "COND_SYNTACTIC_DIV_70", "COND_SYNTACTIC_DIV_75", "COND_SYNTACTIC_DIV_80"], "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "t5-base"}
train_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "init_mem_cpu_alloc_delta": 0,
4
+ "init_mem_cpu_peaked_delta": 0,
5
+ "init_mem_gpu_alloc_delta": 0,
6
+ "init_mem_gpu_peaked_delta": 0,
7
+ "train_mem_cpu_alloc_delta": -902914048,
8
+ "train_mem_cpu_peaked_delta": 902914048,
9
+ "train_mem_gpu_alloc_delta": 1783640576,
10
+ "train_mem_gpu_peaked_delta": 1783640064,
11
+ "train_runtime": 1.098,
12
+ "train_samples": 900000,
13
+ "train_samples_per_second": 204925.455
14
+ }
trainer_state.json ADDED
@@ -0,0 +1,2750 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 4.0,
5
+ "global_step": 225000,
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": 0.0001,
13
+ "loss": 2.5359,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 0.0001,
19
+ "loss": 2.2048,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 0.0001,
25
+ "loss": 2.1844,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 0.0001,
31
+ "loss": 2.13,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.04,
36
+ "learning_rate": 0.0001,
37
+ "loss": 2.1054,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.05,
42
+ "learning_rate": 0.0001,
43
+ "loss": 2.0821,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.06,
48
+ "learning_rate": 0.0001,
49
+ "loss": 2.0846,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.07,
54
+ "learning_rate": 0.0001,
55
+ "loss": 2.0382,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.08,
60
+ "learning_rate": 0.0001,
61
+ "loss": 2.0451,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.09,
66
+ "learning_rate": 0.0001,
67
+ "loss": 2.0263,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.1,
72
+ "learning_rate": 0.0001,
73
+ "loss": 2.0045,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 0.0001,
79
+ "loss": 2.0001,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 0.0001,
85
+ "loss": 1.9757,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.12,
90
+ "learning_rate": 0.0001,
91
+ "loss": 1.9759,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.13,
96
+ "learning_rate": 0.0001,
97
+ "loss": 1.9791,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.14,
102
+ "learning_rate": 0.0001,
103
+ "loss": 1.9773,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.15,
108
+ "learning_rate": 0.0001,
109
+ "loss": 1.9644,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.16,
114
+ "learning_rate": 0.0001,
115
+ "loss": 1.9749,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.17,
120
+ "learning_rate": 0.0001,
121
+ "loss": 1.9496,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.18,
126
+ "learning_rate": 0.0001,
127
+ "loss": 1.9608,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.19,
132
+ "learning_rate": 0.0001,
133
+ "loss": 1.9397,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.2,
138
+ "learning_rate": 0.0001,
139
+ "loss": 1.9283,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.2,
144
+ "learning_rate": 0.0001,
145
+ "loss": 1.9276,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.21,
150
+ "learning_rate": 0.0001,
151
+ "loss": 1.9235,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.22,
156
+ "learning_rate": 0.0001,
157
+ "loss": 1.9289,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.23,
162
+ "learning_rate": 0.0001,
163
+ "loss": 1.9209,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.24,
168
+ "learning_rate": 0.0001,
169
+ "loss": 1.922,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 0.25,
174
+ "learning_rate": 0.0001,
175
+ "loss": 1.9066,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 0.26,
180
+ "learning_rate": 0.0001,
181
+ "loss": 1.9085,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 0.27,
186
+ "learning_rate": 0.0001,
187
+ "loss": 1.9046,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 0.28,
192
+ "learning_rate": 0.0001,
193
+ "loss": 1.8824,
194
+ "step": 15500
195
+ },
196
+ {
197
+ "epoch": 0.28,
198
+ "learning_rate": 0.0001,
199
+ "loss": 1.8896,
200
+ "step": 16000
201
+ },
202
+ {
203
+ "epoch": 0.29,
204
+ "learning_rate": 0.0001,
205
+ "loss": 1.895,
206
+ "step": 16500
207
+ },
208
+ {
209
+ "epoch": 0.3,
210
+ "learning_rate": 0.0001,
211
+ "loss": 1.8901,
212
+ "step": 17000
213
+ },
214
+ {
215
+ "epoch": 0.31,
216
+ "learning_rate": 0.0001,
217
+ "loss": 1.8835,
218
+ "step": 17500
219
+ },
220
+ {
221
+ "epoch": 0.32,
222
+ "learning_rate": 0.0001,
223
+ "loss": 1.8885,
224
+ "step": 18000
225
+ },
226
+ {
227
+ "epoch": 0.33,
228
+ "learning_rate": 0.0001,
229
+ "loss": 1.8658,
230
+ "step": 18500
231
+ },
232
+ {
233
+ "epoch": 0.34,
234
+ "learning_rate": 0.0001,
235
+ "loss": 1.8729,
236
+ "step": 19000
237
+ },
238
+ {
239
+ "epoch": 0.35,
240
+ "learning_rate": 0.0001,
241
+ "loss": 1.8631,
242
+ "step": 19500
243
+ },
244
+ {
245
+ "epoch": 0.36,
246
+ "learning_rate": 0.0001,
247
+ "loss": 1.8621,
248
+ "step": 20000
249
+ },
250
+ {
251
+ "epoch": 0.36,
252
+ "learning_rate": 0.0001,
253
+ "loss": 1.87,
254
+ "step": 20500
255
+ },
256
+ {
257
+ "epoch": 0.37,
258
+ "learning_rate": 0.0001,
259
+ "loss": 1.8591,
260
+ "step": 21000
261
+ },
262
+ {
263
+ "epoch": 0.38,
264
+ "learning_rate": 0.0001,
265
+ "loss": 1.8784,
266
+ "step": 21500
267
+ },
268
+ {
269
+ "epoch": 0.39,
270
+ "learning_rate": 0.0001,
271
+ "loss": 1.8624,
272
+ "step": 22000
273
+ },
274
+ {
275
+ "epoch": 0.4,
276
+ "learning_rate": 0.0001,
277
+ "loss": 1.8612,
278
+ "step": 22500
279
+ },
280
+ {
281
+ "epoch": 0.41,
282
+ "learning_rate": 0.0001,
283
+ "loss": 1.8378,
284
+ "step": 23000
285
+ },
286
+ {
287
+ "epoch": 0.42,
288
+ "learning_rate": 0.0001,
289
+ "loss": 1.8322,
290
+ "step": 23500
291
+ },
292
+ {
293
+ "epoch": 0.43,
294
+ "learning_rate": 0.0001,
295
+ "loss": 1.844,
296
+ "step": 24000
297
+ },
298
+ {
299
+ "epoch": 0.44,
300
+ "learning_rate": 0.0001,
301
+ "loss": 1.83,
302
+ "step": 24500
303
+ },
304
+ {
305
+ "epoch": 0.44,
306
+ "learning_rate": 0.0001,
307
+ "loss": 1.8332,
308
+ "step": 25000
309
+ },
310
+ {
311
+ "epoch": 0.45,
312
+ "learning_rate": 0.0001,
313
+ "loss": 1.8283,
314
+ "step": 25500
315
+ },
316
+ {
317
+ "epoch": 0.46,
318
+ "learning_rate": 0.0001,
319
+ "loss": 1.8475,
320
+ "step": 26000
321
+ },
322
+ {
323
+ "epoch": 0.47,
324
+ "learning_rate": 0.0001,
325
+ "loss": 1.8425,
326
+ "step": 26500
327
+ },
328
+ {
329
+ "epoch": 0.48,
330
+ "learning_rate": 0.0001,
331
+ "loss": 1.8237,
332
+ "step": 27000
333
+ },
334
+ {
335
+ "epoch": 0.49,
336
+ "learning_rate": 0.0001,
337
+ "loss": 1.8303,
338
+ "step": 27500
339
+ },
340
+ {
341
+ "epoch": 0.5,
342
+ "learning_rate": 0.0001,
343
+ "loss": 1.8212,
344
+ "step": 28000
345
+ },
346
+ {
347
+ "epoch": 0.51,
348
+ "learning_rate": 0.0001,
349
+ "loss": 1.824,
350
+ "step": 28500
351
+ },
352
+ {
353
+ "epoch": 0.52,
354
+ "learning_rate": 0.0001,
355
+ "loss": 1.8223,
356
+ "step": 29000
357
+ },
358
+ {
359
+ "epoch": 0.52,
360
+ "learning_rate": 0.0001,
361
+ "loss": 1.8139,
362
+ "step": 29500
363
+ },
364
+ {
365
+ "epoch": 0.53,
366
+ "learning_rate": 0.0001,
367
+ "loss": 1.7987,
368
+ "step": 30000
369
+ },
370
+ {
371
+ "epoch": 0.54,
372
+ "learning_rate": 0.0001,
373
+ "loss": 1.7972,
374
+ "step": 30500
375
+ },
376
+ {
377
+ "epoch": 0.55,
378
+ "learning_rate": 0.0001,
379
+ "loss": 1.8194,
380
+ "step": 31000
381
+ },
382
+ {
383
+ "epoch": 0.56,
384
+ "learning_rate": 0.0001,
385
+ "loss": 1.8134,
386
+ "step": 31500
387
+ },
388
+ {
389
+ "epoch": 0.57,
390
+ "learning_rate": 0.0001,
391
+ "loss": 1.8048,
392
+ "step": 32000
393
+ },
394
+ {
395
+ "epoch": 0.58,
396
+ "learning_rate": 0.0001,
397
+ "loss": 1.8054,
398
+ "step": 32500
399
+ },
400
+ {
401
+ "epoch": 0.59,
402
+ "learning_rate": 0.0001,
403
+ "loss": 1.804,
404
+ "step": 33000
405
+ },
406
+ {
407
+ "epoch": 0.6,
408
+ "learning_rate": 0.0001,
409
+ "loss": 1.7979,
410
+ "step": 33500
411
+ },
412
+ {
413
+ "epoch": 0.6,
414
+ "learning_rate": 0.0001,
415
+ "loss": 1.7976,
416
+ "step": 34000
417
+ },
418
+ {
419
+ "epoch": 0.61,
420
+ "learning_rate": 0.0001,
421
+ "loss": 1.7976,
422
+ "step": 34500
423
+ },
424
+ {
425
+ "epoch": 0.62,
426
+ "learning_rate": 0.0001,
427
+ "loss": 1.7868,
428
+ "step": 35000
429
+ },
430
+ {
431
+ "epoch": 0.63,
432
+ "learning_rate": 0.0001,
433
+ "loss": 1.798,
434
+ "step": 35500
435
+ },
436
+ {
437
+ "epoch": 0.64,
438
+ "learning_rate": 0.0001,
439
+ "loss": 1.7911,
440
+ "step": 36000
441
+ },
442
+ {
443
+ "epoch": 0.65,
444
+ "learning_rate": 0.0001,
445
+ "loss": 1.7787,
446
+ "step": 36500
447
+ },
448
+ {
449
+ "epoch": 0.66,
450
+ "learning_rate": 0.0001,
451
+ "loss": 1.7638,
452
+ "step": 37000
453
+ },
454
+ {
455
+ "epoch": 0.67,
456
+ "learning_rate": 0.0001,
457
+ "loss": 1.7904,
458
+ "step": 37500
459
+ },
460
+ {
461
+ "epoch": 0.68,
462
+ "learning_rate": 0.0001,
463
+ "loss": 1.7928,
464
+ "step": 38000
465
+ },
466
+ {
467
+ "epoch": 0.68,
468
+ "learning_rate": 0.0001,
469
+ "loss": 1.7665,
470
+ "step": 38500
471
+ },
472
+ {
473
+ "epoch": 0.69,
474
+ "learning_rate": 0.0001,
475
+ "loss": 1.7806,
476
+ "step": 39000
477
+ },
478
+ {
479
+ "epoch": 0.7,
480
+ "learning_rate": 0.0001,
481
+ "loss": 1.7774,
482
+ "step": 39500
483
+ },
484
+ {
485
+ "epoch": 0.71,
486
+ "learning_rate": 0.0001,
487
+ "loss": 1.7817,
488
+ "step": 40000
489
+ },
490
+ {
491
+ "epoch": 0.72,
492
+ "learning_rate": 0.0001,
493
+ "loss": 1.7737,
494
+ "step": 40500
495
+ },
496
+ {
497
+ "epoch": 0.73,
498
+ "learning_rate": 0.0001,
499
+ "loss": 1.7829,
500
+ "step": 41000
501
+ },
502
+ {
503
+ "epoch": 0.74,
504
+ "learning_rate": 0.0001,
505
+ "loss": 1.7916,
506
+ "step": 41500
507
+ },
508
+ {
509
+ "epoch": 0.75,
510
+ "learning_rate": 0.0001,
511
+ "loss": 1.7774,
512
+ "step": 42000
513
+ },
514
+ {
515
+ "epoch": 0.76,
516
+ "learning_rate": 0.0001,
517
+ "loss": 1.7595,
518
+ "step": 42500
519
+ },
520
+ {
521
+ "epoch": 0.76,
522
+ "learning_rate": 0.0001,
523
+ "loss": 1.7779,
524
+ "step": 43000
525
+ },
526
+ {
527
+ "epoch": 0.77,
528
+ "learning_rate": 0.0001,
529
+ "loss": 1.7643,
530
+ "step": 43500
531
+ },
532
+ {
533
+ "epoch": 0.78,
534
+ "learning_rate": 0.0001,
535
+ "loss": 1.7455,
536
+ "step": 44000
537
+ },
538
+ {
539
+ "epoch": 0.79,
540
+ "learning_rate": 0.0001,
541
+ "loss": 1.7623,
542
+ "step": 44500
543
+ },
544
+ {
545
+ "epoch": 0.8,
546
+ "learning_rate": 0.0001,
547
+ "loss": 1.7648,
548
+ "step": 45000
549
+ },
550
+ {
551
+ "epoch": 0.81,
552
+ "learning_rate": 0.0001,
553
+ "loss": 1.7844,
554
+ "step": 45500
555
+ },
556
+ {
557
+ "epoch": 0.82,
558
+ "learning_rate": 0.0001,
559
+ "loss": 1.7665,
560
+ "step": 46000
561
+ },
562
+ {
563
+ "epoch": 0.83,
564
+ "learning_rate": 0.0001,
565
+ "loss": 1.7516,
566
+ "step": 46500
567
+ },
568
+ {
569
+ "epoch": 0.84,
570
+ "learning_rate": 0.0001,
571
+ "loss": 1.759,
572
+ "step": 47000
573
+ },
574
+ {
575
+ "epoch": 0.84,
576
+ "learning_rate": 0.0001,
577
+ "loss": 1.755,
578
+ "step": 47500
579
+ },
580
+ {
581
+ "epoch": 0.85,
582
+ "learning_rate": 0.0001,
583
+ "loss": 1.7565,
584
+ "step": 48000
585
+ },
586
+ {
587
+ "epoch": 0.86,
588
+ "learning_rate": 0.0001,
589
+ "loss": 1.759,
590
+ "step": 48500
591
+ },
592
+ {
593
+ "epoch": 0.87,
594
+ "learning_rate": 0.0001,
595
+ "loss": 1.7388,
596
+ "step": 49000
597
+ },
598
+ {
599
+ "epoch": 0.88,
600
+ "learning_rate": 0.0001,
601
+ "loss": 1.7545,
602
+ "step": 49500
603
+ },
604
+ {
605
+ "epoch": 0.89,
606
+ "learning_rate": 0.0001,
607
+ "loss": 1.75,
608
+ "step": 50000
609
+ },
610
+ {
611
+ "epoch": 0.9,
612
+ "learning_rate": 0.0001,
613
+ "loss": 1.7473,
614
+ "step": 50500
615
+ },
616
+ {
617
+ "epoch": 0.91,
618
+ "learning_rate": 0.0001,
619
+ "loss": 1.7504,
620
+ "step": 51000
621
+ },
622
+ {
623
+ "epoch": 0.92,
624
+ "learning_rate": 0.0001,
625
+ "loss": 1.7397,
626
+ "step": 51500
627
+ },
628
+ {
629
+ "epoch": 0.92,
630
+ "learning_rate": 0.0001,
631
+ "loss": 1.7474,
632
+ "step": 52000
633
+ },
634
+ {
635
+ "epoch": 0.93,
636
+ "learning_rate": 0.0001,
637
+ "loss": 1.736,
638
+ "step": 52500
639
+ },
640
+ {
641
+ "epoch": 0.94,
642
+ "learning_rate": 0.0001,
643
+ "loss": 1.7259,
644
+ "step": 53000
645
+ },
646
+ {
647
+ "epoch": 0.95,
648
+ "learning_rate": 0.0001,
649
+ "loss": 1.7467,
650
+ "step": 53500
651
+ },
652
+ {
653
+ "epoch": 0.96,
654
+ "learning_rate": 0.0001,
655
+ "loss": 1.7328,
656
+ "step": 54000
657
+ },
658
+ {
659
+ "epoch": 0.97,
660
+ "learning_rate": 0.0001,
661
+ "loss": 1.7442,
662
+ "step": 54500
663
+ },
664
+ {
665
+ "epoch": 0.98,
666
+ "learning_rate": 0.0001,
667
+ "loss": 1.7526,
668
+ "step": 55000
669
+ },
670
+ {
671
+ "epoch": 0.99,
672
+ "learning_rate": 0.0001,
673
+ "loss": 1.7389,
674
+ "step": 55500
675
+ },
676
+ {
677
+ "epoch": 1.0,
678
+ "learning_rate": 0.0001,
679
+ "loss": 1.7493,
680
+ "step": 56000
681
+ },
682
+ {
683
+ "epoch": 1.0,
684
+ "eval_bleu": 11.4274,
685
+ "eval_gen_len": 15.0006,
686
+ "eval_loss": 1.7998178005218506,
687
+ "eval_runtime": 904.3577,
688
+ "eval_samples_per_second": 15.481,
689
+ "step": 56250
690
+ },
691
+ {
692
+ "epoch": 1.0,
693
+ "learning_rate": 0.0001,
694
+ "loss": 1.7023,
695
+ "step": 56500
696
+ },
697
+ {
698
+ "epoch": 1.01,
699
+ "learning_rate": 0.0001,
700
+ "loss": 1.675,
701
+ "step": 57000
702
+ },
703
+ {
704
+ "epoch": 1.02,
705
+ "learning_rate": 0.0001,
706
+ "loss": 1.6988,
707
+ "step": 57500
708
+ },
709
+ {
710
+ "epoch": 1.03,
711
+ "learning_rate": 0.0001,
712
+ "loss": 1.69,
713
+ "step": 58000
714
+ },
715
+ {
716
+ "epoch": 1.04,
717
+ "learning_rate": 0.0001,
718
+ "loss": 1.6773,
719
+ "step": 58500
720
+ },
721
+ {
722
+ "epoch": 1.05,
723
+ "learning_rate": 0.0001,
724
+ "loss": 1.6833,
725
+ "step": 59000
726
+ },
727
+ {
728
+ "epoch": 1.06,
729
+ "learning_rate": 0.0001,
730
+ "loss": 1.6796,
731
+ "step": 59500
732
+ },
733
+ {
734
+ "epoch": 1.07,
735
+ "learning_rate": 0.0001,
736
+ "loss": 1.6765,
737
+ "step": 60000
738
+ },
739
+ {
740
+ "epoch": 1.08,
741
+ "learning_rate": 0.0001,
742
+ "loss": 1.6798,
743
+ "step": 60500
744
+ },
745
+ {
746
+ "epoch": 1.08,
747
+ "learning_rate": 0.0001,
748
+ "loss": 1.6776,
749
+ "step": 61000
750
+ },
751
+ {
752
+ "epoch": 1.09,
753
+ "learning_rate": 0.0001,
754
+ "loss": 1.6885,
755
+ "step": 61500
756
+ },
757
+ {
758
+ "epoch": 1.1,
759
+ "learning_rate": 0.0001,
760
+ "loss": 1.6816,
761
+ "step": 62000
762
+ },
763
+ {
764
+ "epoch": 1.11,
765
+ "learning_rate": 0.0001,
766
+ "loss": 1.6709,
767
+ "step": 62500
768
+ },
769
+ {
770
+ "epoch": 1.12,
771
+ "learning_rate": 0.0001,
772
+ "loss": 1.6813,
773
+ "step": 63000
774
+ },
775
+ {
776
+ "epoch": 1.13,
777
+ "learning_rate": 0.0001,
778
+ "loss": 1.6864,
779
+ "step": 63500
780
+ },
781
+ {
782
+ "epoch": 1.14,
783
+ "learning_rate": 0.0001,
784
+ "loss": 1.6852,
785
+ "step": 64000
786
+ },
787
+ {
788
+ "epoch": 1.15,
789
+ "learning_rate": 0.0001,
790
+ "loss": 1.6781,
791
+ "step": 64500
792
+ },
793
+ {
794
+ "epoch": 1.16,
795
+ "learning_rate": 0.0001,
796
+ "loss": 1.6734,
797
+ "step": 65000
798
+ },
799
+ {
800
+ "epoch": 1.16,
801
+ "learning_rate": 0.0001,
802
+ "loss": 1.6812,
803
+ "step": 65500
804
+ },
805
+ {
806
+ "epoch": 1.17,
807
+ "learning_rate": 0.0001,
808
+ "loss": 1.6774,
809
+ "step": 66000
810
+ },
811
+ {
812
+ "epoch": 1.18,
813
+ "learning_rate": 0.0001,
814
+ "loss": 1.6805,
815
+ "step": 66500
816
+ },
817
+ {
818
+ "epoch": 1.19,
819
+ "learning_rate": 0.0001,
820
+ "loss": 1.6639,
821
+ "step": 67000
822
+ },
823
+ {
824
+ "epoch": 1.2,
825
+ "learning_rate": 0.0001,
826
+ "loss": 1.6631,
827
+ "step": 67500
828
+ },
829
+ {
830
+ "epoch": 1.21,
831
+ "learning_rate": 0.0001,
832
+ "loss": 1.6645,
833
+ "step": 68000
834
+ },
835
+ {
836
+ "epoch": 1.22,
837
+ "learning_rate": 0.0001,
838
+ "loss": 1.6752,
839
+ "step": 68500
840
+ },
841
+ {
842
+ "epoch": 1.23,
843
+ "learning_rate": 0.0001,
844
+ "loss": 1.6744,
845
+ "step": 69000
846
+ },
847
+ {
848
+ "epoch": 1.24,
849
+ "learning_rate": 0.0001,
850
+ "loss": 1.692,
851
+ "step": 69500
852
+ },
853
+ {
854
+ "epoch": 1.24,
855
+ "learning_rate": 0.0001,
856
+ "loss": 1.6632,
857
+ "step": 70000
858
+ },
859
+ {
860
+ "epoch": 1.25,
861
+ "learning_rate": 0.0001,
862
+ "loss": 1.6604,
863
+ "step": 70500
864
+ },
865
+ {
866
+ "epoch": 1.26,
867
+ "learning_rate": 0.0001,
868
+ "loss": 1.6638,
869
+ "step": 71000
870
+ },
871
+ {
872
+ "epoch": 1.27,
873
+ "learning_rate": 0.0001,
874
+ "loss": 1.6686,
875
+ "step": 71500
876
+ },
877
+ {
878
+ "epoch": 1.28,
879
+ "learning_rate": 0.0001,
880
+ "loss": 1.6699,
881
+ "step": 72000
882
+ },
883
+ {
884
+ "epoch": 1.29,
885
+ "learning_rate": 0.0001,
886
+ "loss": 1.6804,
887
+ "step": 72500
888
+ },
889
+ {
890
+ "epoch": 1.3,
891
+ "learning_rate": 0.0001,
892
+ "loss": 1.6674,
893
+ "step": 73000
894
+ },
895
+ {
896
+ "epoch": 1.31,
897
+ "learning_rate": 0.0001,
898
+ "loss": 1.6735,
899
+ "step": 73500
900
+ },
901
+ {
902
+ "epoch": 1.32,
903
+ "learning_rate": 0.0001,
904
+ "loss": 1.6632,
905
+ "step": 74000
906
+ },
907
+ {
908
+ "epoch": 1.32,
909
+ "learning_rate": 0.0001,
910
+ "loss": 1.655,
911
+ "step": 74500
912
+ },
913
+ {
914
+ "epoch": 1.33,
915
+ "learning_rate": 0.0001,
916
+ "loss": 1.6562,
917
+ "step": 75000
918
+ },
919
+ {
920
+ "epoch": 1.34,
921
+ "learning_rate": 0.0001,
922
+ "loss": 1.6486,
923
+ "step": 75500
924
+ },
925
+ {
926
+ "epoch": 1.35,
927
+ "learning_rate": 0.0001,
928
+ "loss": 1.6586,
929
+ "step": 76000
930
+ },
931
+ {
932
+ "epoch": 1.36,
933
+ "learning_rate": 0.0001,
934
+ "loss": 1.6505,
935
+ "step": 76500
936
+ },
937
+ {
938
+ "epoch": 1.37,
939
+ "learning_rate": 0.0001,
940
+ "loss": 1.6576,
941
+ "step": 77000
942
+ },
943
+ {
944
+ "epoch": 1.38,
945
+ "learning_rate": 0.0001,
946
+ "loss": 1.6541,
947
+ "step": 77500
948
+ },
949
+ {
950
+ "epoch": 1.39,
951
+ "learning_rate": 0.0001,
952
+ "loss": 1.6486,
953
+ "step": 78000
954
+ },
955
+ {
956
+ "epoch": 1.4,
957
+ "learning_rate": 0.0001,
958
+ "loss": 1.6578,
959
+ "step": 78500
960
+ },
961
+ {
962
+ "epoch": 1.4,
963
+ "learning_rate": 0.0001,
964
+ "loss": 1.6547,
965
+ "step": 79000
966
+ },
967
+ {
968
+ "epoch": 1.41,
969
+ "learning_rate": 0.0001,
970
+ "loss": 1.6497,
971
+ "step": 79500
972
+ },
973
+ {
974
+ "epoch": 1.42,
975
+ "learning_rate": 0.0001,
976
+ "loss": 1.6509,
977
+ "step": 80000
978
+ },
979
+ {
980
+ "epoch": 1.43,
981
+ "learning_rate": 0.0001,
982
+ "loss": 1.6484,
983
+ "step": 80500
984
+ },
985
+ {
986
+ "epoch": 1.44,
987
+ "learning_rate": 0.0001,
988
+ "loss": 1.6463,
989
+ "step": 81000
990
+ },
991
+ {
992
+ "epoch": 1.45,
993
+ "learning_rate": 0.0001,
994
+ "loss": 1.6513,
995
+ "step": 81500
996
+ },
997
+ {
998
+ "epoch": 1.46,
999
+ "learning_rate": 0.0001,
1000
+ "loss": 1.6377,
1001
+ "step": 82000
1002
+ },
1003
+ {
1004
+ "epoch": 1.47,
1005
+ "learning_rate": 0.0001,
1006
+ "loss": 1.6508,
1007
+ "step": 82500
1008
+ },
1009
+ {
1010
+ "epoch": 1.48,
1011
+ "learning_rate": 0.0001,
1012
+ "loss": 1.6432,
1013
+ "step": 83000
1014
+ },
1015
+ {
1016
+ "epoch": 1.48,
1017
+ "learning_rate": 0.0001,
1018
+ "loss": 1.635,
1019
+ "step": 83500
1020
+ },
1021
+ {
1022
+ "epoch": 1.49,
1023
+ "learning_rate": 0.0001,
1024
+ "loss": 1.6601,
1025
+ "step": 84000
1026
+ },
1027
+ {
1028
+ "epoch": 1.5,
1029
+ "learning_rate": 0.0001,
1030
+ "loss": 1.6419,
1031
+ "step": 84500
1032
+ },
1033
+ {
1034
+ "epoch": 1.51,
1035
+ "learning_rate": 0.0001,
1036
+ "loss": 1.6284,
1037
+ "step": 85000
1038
+ },
1039
+ {
1040
+ "epoch": 1.52,
1041
+ "learning_rate": 0.0001,
1042
+ "loss": 1.6547,
1043
+ "step": 85500
1044
+ },
1045
+ {
1046
+ "epoch": 1.53,
1047
+ "learning_rate": 0.0001,
1048
+ "loss": 1.6534,
1049
+ "step": 86000
1050
+ },
1051
+ {
1052
+ "epoch": 1.54,
1053
+ "learning_rate": 0.0001,
1054
+ "loss": 1.6409,
1055
+ "step": 86500
1056
+ },
1057
+ {
1058
+ "epoch": 1.55,
1059
+ "learning_rate": 0.0001,
1060
+ "loss": 1.6459,
1061
+ "step": 87000
1062
+ },
1063
+ {
1064
+ "epoch": 1.56,
1065
+ "learning_rate": 0.0001,
1066
+ "loss": 1.6374,
1067
+ "step": 87500
1068
+ },
1069
+ {
1070
+ "epoch": 1.56,
1071
+ "learning_rate": 0.0001,
1072
+ "loss": 1.6652,
1073
+ "step": 88000
1074
+ },
1075
+ {
1076
+ "epoch": 1.57,
1077
+ "learning_rate": 0.0001,
1078
+ "loss": 1.6435,
1079
+ "step": 88500
1080
+ },
1081
+ {
1082
+ "epoch": 1.58,
1083
+ "learning_rate": 0.0001,
1084
+ "loss": 1.6474,
1085
+ "step": 89000
1086
+ },
1087
+ {
1088
+ "epoch": 1.59,
1089
+ "learning_rate": 0.0001,
1090
+ "loss": 1.6413,
1091
+ "step": 89500
1092
+ },
1093
+ {
1094
+ "epoch": 1.6,
1095
+ "learning_rate": 0.0001,
1096
+ "loss": 1.6464,
1097
+ "step": 90000
1098
+ },
1099
+ {
1100
+ "epoch": 1.61,
1101
+ "learning_rate": 0.0001,
1102
+ "loss": 1.6251,
1103
+ "step": 90500
1104
+ },
1105
+ {
1106
+ "epoch": 1.62,
1107
+ "learning_rate": 0.0001,
1108
+ "loss": 1.6365,
1109
+ "step": 91000
1110
+ },
1111
+ {
1112
+ "epoch": 1.63,
1113
+ "learning_rate": 0.0001,
1114
+ "loss": 1.6322,
1115
+ "step": 91500
1116
+ },
1117
+ {
1118
+ "epoch": 1.64,
1119
+ "learning_rate": 0.0001,
1120
+ "loss": 1.6552,
1121
+ "step": 92000
1122
+ },
1123
+ {
1124
+ "epoch": 1.64,
1125
+ "learning_rate": 0.0001,
1126
+ "loss": 1.6345,
1127
+ "step": 92500
1128
+ },
1129
+ {
1130
+ "epoch": 1.65,
1131
+ "learning_rate": 0.0001,
1132
+ "loss": 1.6367,
1133
+ "step": 93000
1134
+ },
1135
+ {
1136
+ "epoch": 1.66,
1137
+ "learning_rate": 0.0001,
1138
+ "loss": 1.6475,
1139
+ "step": 93500
1140
+ },
1141
+ {
1142
+ "epoch": 1.67,
1143
+ "learning_rate": 0.0001,
1144
+ "loss": 1.6393,
1145
+ "step": 94000
1146
+ },
1147
+ {
1148
+ "epoch": 1.68,
1149
+ "learning_rate": 0.0001,
1150
+ "loss": 1.649,
1151
+ "step": 94500
1152
+ },
1153
+ {
1154
+ "epoch": 1.69,
1155
+ "learning_rate": 0.0001,
1156
+ "loss": 1.6516,
1157
+ "step": 95000
1158
+ },
1159
+ {
1160
+ "epoch": 1.7,
1161
+ "learning_rate": 0.0001,
1162
+ "loss": 1.6214,
1163
+ "step": 95500
1164
+ },
1165
+ {
1166
+ "epoch": 1.71,
1167
+ "learning_rate": 0.0001,
1168
+ "loss": 1.6312,
1169
+ "step": 96000
1170
+ },
1171
+ {
1172
+ "epoch": 1.72,
1173
+ "learning_rate": 0.0001,
1174
+ "loss": 1.6288,
1175
+ "step": 96500
1176
+ },
1177
+ {
1178
+ "epoch": 1.72,
1179
+ "learning_rate": 0.0001,
1180
+ "loss": 1.6271,
1181
+ "step": 97000
1182
+ },
1183
+ {
1184
+ "epoch": 1.73,
1185
+ "learning_rate": 0.0001,
1186
+ "loss": 1.6345,
1187
+ "step": 97500
1188
+ },
1189
+ {
1190
+ "epoch": 1.74,
1191
+ "learning_rate": 0.0001,
1192
+ "loss": 1.6379,
1193
+ "step": 98000
1194
+ },
1195
+ {
1196
+ "epoch": 1.75,
1197
+ "learning_rate": 0.0001,
1198
+ "loss": 1.6423,
1199
+ "step": 98500
1200
+ },
1201
+ {
1202
+ "epoch": 1.76,
1203
+ "learning_rate": 0.0001,
1204
+ "loss": 1.6311,
1205
+ "step": 99000
1206
+ },
1207
+ {
1208
+ "epoch": 1.77,
1209
+ "learning_rate": 0.0001,
1210
+ "loss": 1.6069,
1211
+ "step": 99500
1212
+ },
1213
+ {
1214
+ "epoch": 1.78,
1215
+ "learning_rate": 0.0001,
1216
+ "loss": 1.6339,
1217
+ "step": 100000
1218
+ },
1219
+ {
1220
+ "epoch": 1.79,
1221
+ "learning_rate": 0.0001,
1222
+ "loss": 1.6417,
1223
+ "step": 100500
1224
+ },
1225
+ {
1226
+ "epoch": 1.8,
1227
+ "learning_rate": 0.0001,
1228
+ "loss": 1.6317,
1229
+ "step": 101000
1230
+ },
1231
+ {
1232
+ "epoch": 1.8,
1233
+ "learning_rate": 0.0001,
1234
+ "loss": 1.6183,
1235
+ "step": 101500
1236
+ },
1237
+ {
1238
+ "epoch": 1.81,
1239
+ "learning_rate": 0.0001,
1240
+ "loss": 1.6274,
1241
+ "step": 102000
1242
+ },
1243
+ {
1244
+ "epoch": 1.82,
1245
+ "learning_rate": 0.0001,
1246
+ "loss": 1.6307,
1247
+ "step": 102500
1248
+ },
1249
+ {
1250
+ "epoch": 1.83,
1251
+ "learning_rate": 0.0001,
1252
+ "loss": 1.6261,
1253
+ "step": 103000
1254
+ },
1255
+ {
1256
+ "epoch": 1.84,
1257
+ "learning_rate": 0.0001,
1258
+ "loss": 1.6288,
1259
+ "step": 103500
1260
+ },
1261
+ {
1262
+ "epoch": 1.85,
1263
+ "learning_rate": 0.0001,
1264
+ "loss": 1.6218,
1265
+ "step": 104000
1266
+ },
1267
+ {
1268
+ "epoch": 1.86,
1269
+ "learning_rate": 0.0001,
1270
+ "loss": 1.625,
1271
+ "step": 104500
1272
+ },
1273
+ {
1274
+ "epoch": 1.87,
1275
+ "learning_rate": 0.0001,
1276
+ "loss": 1.616,
1277
+ "step": 105000
1278
+ },
1279
+ {
1280
+ "epoch": 1.88,
1281
+ "learning_rate": 0.0001,
1282
+ "loss": 1.6132,
1283
+ "step": 105500
1284
+ },
1285
+ {
1286
+ "epoch": 1.88,
1287
+ "learning_rate": 0.0001,
1288
+ "loss": 1.6102,
1289
+ "step": 106000
1290
+ },
1291
+ {
1292
+ "epoch": 1.89,
1293
+ "learning_rate": 0.0001,
1294
+ "loss": 1.6238,
1295
+ "step": 106500
1296
+ },
1297
+ {
1298
+ "epoch": 1.9,
1299
+ "learning_rate": 0.0001,
1300
+ "loss": 1.6207,
1301
+ "step": 107000
1302
+ },
1303
+ {
1304
+ "epoch": 1.91,
1305
+ "learning_rate": 0.0001,
1306
+ "loss": 1.6179,
1307
+ "step": 107500
1308
+ },
1309
+ {
1310
+ "epoch": 1.92,
1311
+ "learning_rate": 0.0001,
1312
+ "loss": 1.6041,
1313
+ "step": 108000
1314
+ },
1315
+ {
1316
+ "epoch": 1.93,
1317
+ "learning_rate": 0.0001,
1318
+ "loss": 1.6198,
1319
+ "step": 108500
1320
+ },
1321
+ {
1322
+ "epoch": 1.94,
1323
+ "learning_rate": 0.0001,
1324
+ "loss": 1.6224,
1325
+ "step": 109000
1326
+ },
1327
+ {
1328
+ "epoch": 1.95,
1329
+ "learning_rate": 0.0001,
1330
+ "loss": 1.6109,
1331
+ "step": 109500
1332
+ },
1333
+ {
1334
+ "epoch": 1.96,
1335
+ "learning_rate": 0.0001,
1336
+ "loss": 1.6159,
1337
+ "step": 110000
1338
+ },
1339
+ {
1340
+ "epoch": 1.96,
1341
+ "learning_rate": 0.0001,
1342
+ "loss": 1.5544,
1343
+ "step": 110500
1344
+ },
1345
+ {
1346
+ "epoch": 1.97,
1347
+ "learning_rate": 0.0001,
1348
+ "loss": 1.5557,
1349
+ "step": 111000
1350
+ },
1351
+ {
1352
+ "epoch": 1.98,
1353
+ "learning_rate": 0.0001,
1354
+ "loss": 1.5566,
1355
+ "step": 111500
1356
+ },
1357
+ {
1358
+ "epoch": 1.99,
1359
+ "learning_rate": 0.0001,
1360
+ "loss": 1.5543,
1361
+ "step": 112000
1362
+ },
1363
+ {
1364
+ "epoch": 2.0,
1365
+ "learning_rate": 0.0001,
1366
+ "loss": 1.5721,
1367
+ "step": 112500
1368
+ },
1369
+ {
1370
+ "epoch": 2.0,
1371
+ "eval_bleu": 11.8151,
1372
+ "eval_gen_len": 15.1413,
1373
+ "eval_loss": 1.7656184434890747,
1374
+ "eval_runtime": 908.9492,
1375
+ "eval_samples_per_second": 15.402,
1376
+ "step": 112500
1377
+ },
1378
+ {
1379
+ "epoch": 2.01,
1380
+ "learning_rate": 0.0001,
1381
+ "loss": 1.567,
1382
+ "step": 113000
1383
+ },
1384
+ {
1385
+ "epoch": 2.02,
1386
+ "learning_rate": 0.0001,
1387
+ "loss": 1.557,
1388
+ "step": 113500
1389
+ },
1390
+ {
1391
+ "epoch": 2.03,
1392
+ "learning_rate": 0.0001,
1393
+ "loss": 1.5747,
1394
+ "step": 114000
1395
+ },
1396
+ {
1397
+ "epoch": 2.04,
1398
+ "learning_rate": 0.0001,
1399
+ "loss": 1.5602,
1400
+ "step": 114500
1401
+ },
1402
+ {
1403
+ "epoch": 2.04,
1404
+ "learning_rate": 0.0001,
1405
+ "loss": 1.5574,
1406
+ "step": 115000
1407
+ },
1408
+ {
1409
+ "epoch": 2.05,
1410
+ "learning_rate": 0.0001,
1411
+ "loss": 1.5586,
1412
+ "step": 115500
1413
+ },
1414
+ {
1415
+ "epoch": 2.06,
1416
+ "learning_rate": 0.0001,
1417
+ "loss": 1.5582,
1418
+ "step": 116000
1419
+ },
1420
+ {
1421
+ "epoch": 2.07,
1422
+ "learning_rate": 0.0001,
1423
+ "loss": 1.5565,
1424
+ "step": 116500
1425
+ },
1426
+ {
1427
+ "epoch": 2.08,
1428
+ "learning_rate": 0.0001,
1429
+ "loss": 1.548,
1430
+ "step": 117000
1431
+ },
1432
+ {
1433
+ "epoch": 2.09,
1434
+ "learning_rate": 0.0001,
1435
+ "loss": 1.5543,
1436
+ "step": 117500
1437
+ },
1438
+ {
1439
+ "epoch": 2.1,
1440
+ "learning_rate": 0.0001,
1441
+ "loss": 1.5652,
1442
+ "step": 118000
1443
+ },
1444
+ {
1445
+ "epoch": 2.11,
1446
+ "learning_rate": 0.0001,
1447
+ "loss": 1.5506,
1448
+ "step": 118500
1449
+ },
1450
+ {
1451
+ "epoch": 2.12,
1452
+ "learning_rate": 0.0001,
1453
+ "loss": 1.5677,
1454
+ "step": 119000
1455
+ },
1456
+ {
1457
+ "epoch": 2.12,
1458
+ "learning_rate": 0.0001,
1459
+ "loss": 1.5566,
1460
+ "step": 119500
1461
+ },
1462
+ {
1463
+ "epoch": 2.13,
1464
+ "learning_rate": 0.0001,
1465
+ "loss": 1.5569,
1466
+ "step": 120000
1467
+ },
1468
+ {
1469
+ "epoch": 2.14,
1470
+ "learning_rate": 0.0001,
1471
+ "loss": 1.5489,
1472
+ "step": 120500
1473
+ },
1474
+ {
1475
+ "epoch": 2.15,
1476
+ "learning_rate": 0.0001,
1477
+ "loss": 1.5477,
1478
+ "step": 121000
1479
+ },
1480
+ {
1481
+ "epoch": 2.16,
1482
+ "learning_rate": 0.0001,
1483
+ "loss": 1.5648,
1484
+ "step": 121500
1485
+ },
1486
+ {
1487
+ "epoch": 2.17,
1488
+ "learning_rate": 0.0001,
1489
+ "loss": 1.5526,
1490
+ "step": 122000
1491
+ },
1492
+ {
1493
+ "epoch": 2.18,
1494
+ "learning_rate": 0.0001,
1495
+ "loss": 1.5571,
1496
+ "step": 122500
1497
+ },
1498
+ {
1499
+ "epoch": 2.19,
1500
+ "learning_rate": 0.0001,
1501
+ "loss": 1.5667,
1502
+ "step": 123000
1503
+ },
1504
+ {
1505
+ "epoch": 2.2,
1506
+ "learning_rate": 0.0001,
1507
+ "loss": 1.5603,
1508
+ "step": 123500
1509
+ },
1510
+ {
1511
+ "epoch": 2.2,
1512
+ "learning_rate": 0.0001,
1513
+ "loss": 1.5559,
1514
+ "step": 124000
1515
+ },
1516
+ {
1517
+ "epoch": 2.21,
1518
+ "learning_rate": 0.0001,
1519
+ "loss": 1.5616,
1520
+ "step": 124500
1521
+ },
1522
+ {
1523
+ "epoch": 2.22,
1524
+ "learning_rate": 0.0001,
1525
+ "loss": 1.5638,
1526
+ "step": 125000
1527
+ },
1528
+ {
1529
+ "epoch": 2.23,
1530
+ "learning_rate": 0.0001,
1531
+ "loss": 1.5465,
1532
+ "step": 125500
1533
+ },
1534
+ {
1535
+ "epoch": 2.24,
1536
+ "learning_rate": 0.0001,
1537
+ "loss": 1.5659,
1538
+ "step": 126000
1539
+ },
1540
+ {
1541
+ "epoch": 2.25,
1542
+ "learning_rate": 0.0001,
1543
+ "loss": 1.5706,
1544
+ "step": 126500
1545
+ },
1546
+ {
1547
+ "epoch": 2.26,
1548
+ "learning_rate": 0.0001,
1549
+ "loss": 1.5574,
1550
+ "step": 127000
1551
+ },
1552
+ {
1553
+ "epoch": 2.27,
1554
+ "learning_rate": 0.0001,
1555
+ "loss": 1.5539,
1556
+ "step": 127500
1557
+ },
1558
+ {
1559
+ "epoch": 2.28,
1560
+ "learning_rate": 0.0001,
1561
+ "loss": 1.5445,
1562
+ "step": 128000
1563
+ },
1564
+ {
1565
+ "epoch": 2.28,
1566
+ "learning_rate": 0.0001,
1567
+ "loss": 1.5449,
1568
+ "step": 128500
1569
+ },
1570
+ {
1571
+ "epoch": 2.29,
1572
+ "learning_rate": 0.0001,
1573
+ "loss": 1.5632,
1574
+ "step": 129000
1575
+ },
1576
+ {
1577
+ "epoch": 2.3,
1578
+ "learning_rate": 0.0001,
1579
+ "loss": 1.5586,
1580
+ "step": 129500
1581
+ },
1582
+ {
1583
+ "epoch": 2.31,
1584
+ "learning_rate": 0.0001,
1585
+ "loss": 1.5644,
1586
+ "step": 130000
1587
+ },
1588
+ {
1589
+ "epoch": 2.32,
1590
+ "learning_rate": 0.0001,
1591
+ "loss": 1.5549,
1592
+ "step": 130500
1593
+ },
1594
+ {
1595
+ "epoch": 2.33,
1596
+ "learning_rate": 0.0001,
1597
+ "loss": 1.557,
1598
+ "step": 131000
1599
+ },
1600
+ {
1601
+ "epoch": 2.34,
1602
+ "learning_rate": 0.0001,
1603
+ "loss": 1.555,
1604
+ "step": 131500
1605
+ },
1606
+ {
1607
+ "epoch": 2.35,
1608
+ "learning_rate": 0.0001,
1609
+ "loss": 1.5428,
1610
+ "step": 132000
1611
+ },
1612
+ {
1613
+ "epoch": 2.36,
1614
+ "learning_rate": 0.0001,
1615
+ "loss": 1.5554,
1616
+ "step": 132500
1617
+ },
1618
+ {
1619
+ "epoch": 2.36,
1620
+ "learning_rate": 0.0001,
1621
+ "loss": 1.5546,
1622
+ "step": 133000
1623
+ },
1624
+ {
1625
+ "epoch": 2.37,
1626
+ "learning_rate": 0.0001,
1627
+ "loss": 1.5553,
1628
+ "step": 133500
1629
+ },
1630
+ {
1631
+ "epoch": 2.38,
1632
+ "learning_rate": 0.0001,
1633
+ "loss": 1.5524,
1634
+ "step": 134000
1635
+ },
1636
+ {
1637
+ "epoch": 2.39,
1638
+ "learning_rate": 0.0001,
1639
+ "loss": 1.5456,
1640
+ "step": 134500
1641
+ },
1642
+ {
1643
+ "epoch": 2.4,
1644
+ "learning_rate": 0.0001,
1645
+ "loss": 1.5457,
1646
+ "step": 135000
1647
+ },
1648
+ {
1649
+ "epoch": 2.41,
1650
+ "learning_rate": 0.0001,
1651
+ "loss": 1.5386,
1652
+ "step": 135500
1653
+ },
1654
+ {
1655
+ "epoch": 2.42,
1656
+ "learning_rate": 0.0001,
1657
+ "loss": 1.5575,
1658
+ "step": 136000
1659
+ },
1660
+ {
1661
+ "epoch": 2.43,
1662
+ "learning_rate": 0.0001,
1663
+ "loss": 1.5474,
1664
+ "step": 136500
1665
+ },
1666
+ {
1667
+ "epoch": 2.44,
1668
+ "learning_rate": 0.0001,
1669
+ "loss": 1.5404,
1670
+ "step": 137000
1671
+ },
1672
+ {
1673
+ "epoch": 2.44,
1674
+ "learning_rate": 0.0001,
1675
+ "loss": 1.5357,
1676
+ "step": 137500
1677
+ },
1678
+ {
1679
+ "epoch": 2.45,
1680
+ "learning_rate": 0.0001,
1681
+ "loss": 1.5431,
1682
+ "step": 138000
1683
+ },
1684
+ {
1685
+ "epoch": 2.46,
1686
+ "learning_rate": 0.0001,
1687
+ "loss": 1.543,
1688
+ "step": 138500
1689
+ },
1690
+ {
1691
+ "epoch": 2.47,
1692
+ "learning_rate": 0.0001,
1693
+ "loss": 1.5417,
1694
+ "step": 139000
1695
+ },
1696
+ {
1697
+ "epoch": 2.48,
1698
+ "learning_rate": 0.0001,
1699
+ "loss": 1.5511,
1700
+ "step": 139500
1701
+ },
1702
+ {
1703
+ "epoch": 2.49,
1704
+ "learning_rate": 0.0001,
1705
+ "loss": 1.5515,
1706
+ "step": 140000
1707
+ },
1708
+ {
1709
+ "epoch": 2.5,
1710
+ "learning_rate": 0.0001,
1711
+ "loss": 1.5331,
1712
+ "step": 140500
1713
+ },
1714
+ {
1715
+ "epoch": 2.51,
1716
+ "learning_rate": 0.0001,
1717
+ "loss": 1.5541,
1718
+ "step": 141000
1719
+ },
1720
+ {
1721
+ "epoch": 2.52,
1722
+ "learning_rate": 0.0001,
1723
+ "loss": 1.5366,
1724
+ "step": 141500
1725
+ },
1726
+ {
1727
+ "epoch": 2.52,
1728
+ "learning_rate": 0.0001,
1729
+ "loss": 1.5504,
1730
+ "step": 142000
1731
+ },
1732
+ {
1733
+ "epoch": 2.53,
1734
+ "learning_rate": 0.0001,
1735
+ "loss": 1.546,
1736
+ "step": 142500
1737
+ },
1738
+ {
1739
+ "epoch": 2.54,
1740
+ "learning_rate": 0.0001,
1741
+ "loss": 1.5499,
1742
+ "step": 143000
1743
+ },
1744
+ {
1745
+ "epoch": 2.55,
1746
+ "learning_rate": 0.0001,
1747
+ "loss": 1.5367,
1748
+ "step": 143500
1749
+ },
1750
+ {
1751
+ "epoch": 2.56,
1752
+ "learning_rate": 0.0001,
1753
+ "loss": 1.5366,
1754
+ "step": 144000
1755
+ },
1756
+ {
1757
+ "epoch": 2.57,
1758
+ "learning_rate": 0.0001,
1759
+ "loss": 1.5543,
1760
+ "step": 144500
1761
+ },
1762
+ {
1763
+ "epoch": 2.58,
1764
+ "learning_rate": 0.0001,
1765
+ "loss": 1.5265,
1766
+ "step": 145000
1767
+ },
1768
+ {
1769
+ "epoch": 2.59,
1770
+ "learning_rate": 0.0001,
1771
+ "loss": 1.5382,
1772
+ "step": 145500
1773
+ },
1774
+ {
1775
+ "epoch": 2.6,
1776
+ "learning_rate": 0.0001,
1777
+ "loss": 1.5439,
1778
+ "step": 146000
1779
+ },
1780
+ {
1781
+ "epoch": 2.6,
1782
+ "learning_rate": 0.0001,
1783
+ "loss": 1.5358,
1784
+ "step": 146500
1785
+ },
1786
+ {
1787
+ "epoch": 2.61,
1788
+ "learning_rate": 0.0001,
1789
+ "loss": 1.5341,
1790
+ "step": 147000
1791
+ },
1792
+ {
1793
+ "epoch": 2.62,
1794
+ "learning_rate": 0.0001,
1795
+ "loss": 1.5402,
1796
+ "step": 147500
1797
+ },
1798
+ {
1799
+ "epoch": 2.63,
1800
+ "learning_rate": 0.0001,
1801
+ "loss": 1.5336,
1802
+ "step": 148000
1803
+ },
1804
+ {
1805
+ "epoch": 2.64,
1806
+ "learning_rate": 0.0001,
1807
+ "loss": 1.5453,
1808
+ "step": 148500
1809
+ },
1810
+ {
1811
+ "epoch": 2.65,
1812
+ "learning_rate": 0.0001,
1813
+ "loss": 1.5443,
1814
+ "step": 149000
1815
+ },
1816
+ {
1817
+ "epoch": 2.66,
1818
+ "learning_rate": 0.0001,
1819
+ "loss": 1.5387,
1820
+ "step": 149500
1821
+ },
1822
+ {
1823
+ "epoch": 2.67,
1824
+ "learning_rate": 0.0001,
1825
+ "loss": 1.541,
1826
+ "step": 150000
1827
+ },
1828
+ {
1829
+ "epoch": 2.68,
1830
+ "learning_rate": 0.0001,
1831
+ "loss": 1.5371,
1832
+ "step": 150500
1833
+ },
1834
+ {
1835
+ "epoch": 2.68,
1836
+ "learning_rate": 0.0001,
1837
+ "loss": 1.5307,
1838
+ "step": 151000
1839
+ },
1840
+ {
1841
+ "epoch": 2.69,
1842
+ "learning_rate": 0.0001,
1843
+ "loss": 1.5355,
1844
+ "step": 151500
1845
+ },
1846
+ {
1847
+ "epoch": 2.7,
1848
+ "learning_rate": 0.0001,
1849
+ "loss": 1.5379,
1850
+ "step": 152000
1851
+ },
1852
+ {
1853
+ "epoch": 2.71,
1854
+ "learning_rate": 0.0001,
1855
+ "loss": 1.5397,
1856
+ "step": 152500
1857
+ },
1858
+ {
1859
+ "epoch": 2.72,
1860
+ "learning_rate": 0.0001,
1861
+ "loss": 1.5379,
1862
+ "step": 153000
1863
+ },
1864
+ {
1865
+ "epoch": 2.73,
1866
+ "learning_rate": 0.0001,
1867
+ "loss": 1.5372,
1868
+ "step": 153500
1869
+ },
1870
+ {
1871
+ "epoch": 2.74,
1872
+ "learning_rate": 0.0001,
1873
+ "loss": 1.5358,
1874
+ "step": 154000
1875
+ },
1876
+ {
1877
+ "epoch": 2.75,
1878
+ "learning_rate": 0.0001,
1879
+ "loss": 1.5576,
1880
+ "step": 154500
1881
+ },
1882
+ {
1883
+ "epoch": 2.76,
1884
+ "learning_rate": 0.0001,
1885
+ "loss": 1.5357,
1886
+ "step": 155000
1887
+ },
1888
+ {
1889
+ "epoch": 2.76,
1890
+ "learning_rate": 0.0001,
1891
+ "loss": 1.5481,
1892
+ "step": 155500
1893
+ },
1894
+ {
1895
+ "epoch": 2.77,
1896
+ "learning_rate": 0.0001,
1897
+ "loss": 1.5354,
1898
+ "step": 156000
1899
+ },
1900
+ {
1901
+ "epoch": 2.78,
1902
+ "learning_rate": 0.0001,
1903
+ "loss": 1.5359,
1904
+ "step": 156500
1905
+ },
1906
+ {
1907
+ "epoch": 2.79,
1908
+ "learning_rate": 0.0001,
1909
+ "loss": 1.5362,
1910
+ "step": 157000
1911
+ },
1912
+ {
1913
+ "epoch": 2.8,
1914
+ "learning_rate": 0.0001,
1915
+ "loss": 1.5361,
1916
+ "step": 157500
1917
+ },
1918
+ {
1919
+ "epoch": 2.81,
1920
+ "learning_rate": 0.0001,
1921
+ "loss": 1.5222,
1922
+ "step": 158000
1923
+ },
1924
+ {
1925
+ "epoch": 2.82,
1926
+ "learning_rate": 0.0001,
1927
+ "loss": 1.523,
1928
+ "step": 158500
1929
+ },
1930
+ {
1931
+ "epoch": 2.83,
1932
+ "learning_rate": 0.0001,
1933
+ "loss": 1.5265,
1934
+ "step": 159000
1935
+ },
1936
+ {
1937
+ "epoch": 2.84,
1938
+ "learning_rate": 0.0001,
1939
+ "loss": 1.5283,
1940
+ "step": 159500
1941
+ },
1942
+ {
1943
+ "epoch": 2.84,
1944
+ "learning_rate": 0.0001,
1945
+ "loss": 1.5377,
1946
+ "step": 160000
1947
+ },
1948
+ {
1949
+ "epoch": 2.85,
1950
+ "learning_rate": 0.0001,
1951
+ "loss": 1.527,
1952
+ "step": 160500
1953
+ },
1954
+ {
1955
+ "epoch": 2.86,
1956
+ "learning_rate": 0.0001,
1957
+ "loss": 1.5429,
1958
+ "step": 161000
1959
+ },
1960
+ {
1961
+ "epoch": 2.87,
1962
+ "learning_rate": 0.0001,
1963
+ "loss": 1.526,
1964
+ "step": 161500
1965
+ },
1966
+ {
1967
+ "epoch": 2.88,
1968
+ "learning_rate": 0.0001,
1969
+ "loss": 1.5305,
1970
+ "step": 162000
1971
+ },
1972
+ {
1973
+ "epoch": 2.89,
1974
+ "learning_rate": 0.0001,
1975
+ "loss": 1.5251,
1976
+ "step": 162500
1977
+ },
1978
+ {
1979
+ "epoch": 2.9,
1980
+ "learning_rate": 0.0001,
1981
+ "loss": 1.5214,
1982
+ "step": 163000
1983
+ },
1984
+ {
1985
+ "epoch": 2.91,
1986
+ "learning_rate": 0.0001,
1987
+ "loss": 1.5256,
1988
+ "step": 163500
1989
+ },
1990
+ {
1991
+ "epoch": 2.92,
1992
+ "learning_rate": 0.0001,
1993
+ "loss": 1.5119,
1994
+ "step": 164000
1995
+ },
1996
+ {
1997
+ "epoch": 2.92,
1998
+ "learning_rate": 0.0001,
1999
+ "loss": 1.5314,
2000
+ "step": 164500
2001
+ },
2002
+ {
2003
+ "epoch": 2.93,
2004
+ "learning_rate": 0.0001,
2005
+ "loss": 1.5314,
2006
+ "step": 165000
2007
+ },
2008
+ {
2009
+ "epoch": 2.94,
2010
+ "learning_rate": 0.0001,
2011
+ "loss": 1.5328,
2012
+ "step": 165500
2013
+ },
2014
+ {
2015
+ "epoch": 2.95,
2016
+ "learning_rate": 0.0001,
2017
+ "loss": 1.5132,
2018
+ "step": 166000
2019
+ },
2020
+ {
2021
+ "epoch": 2.96,
2022
+ "learning_rate": 0.0001,
2023
+ "loss": 1.5353,
2024
+ "step": 166500
2025
+ },
2026
+ {
2027
+ "epoch": 2.97,
2028
+ "learning_rate": 0.0001,
2029
+ "loss": 1.5156,
2030
+ "step": 167000
2031
+ },
2032
+ {
2033
+ "epoch": 2.98,
2034
+ "learning_rate": 0.0001,
2035
+ "loss": 1.5236,
2036
+ "step": 167500
2037
+ },
2038
+ {
2039
+ "epoch": 2.99,
2040
+ "learning_rate": 0.0001,
2041
+ "loss": 1.5358,
2042
+ "step": 168000
2043
+ },
2044
+ {
2045
+ "epoch": 3.0,
2046
+ "learning_rate": 0.0001,
2047
+ "loss": 1.5124,
2048
+ "step": 168500
2049
+ },
2050
+ {
2051
+ "epoch": 3.0,
2052
+ "eval_bleu": 11.9547,
2053
+ "eval_gen_len": 15.0479,
2054
+ "eval_loss": 1.7495191097259521,
2055
+ "eval_runtime": 909.5834,
2056
+ "eval_samples_per_second": 15.392,
2057
+ "step": 168750
2058
+ },
2059
+ {
2060
+ "epoch": 3.0,
2061
+ "learning_rate": 0.0001,
2062
+ "loss": 1.4957,
2063
+ "step": 169000
2064
+ },
2065
+ {
2066
+ "epoch": 3.01,
2067
+ "learning_rate": 0.0001,
2068
+ "loss": 1.4527,
2069
+ "step": 169500
2070
+ },
2071
+ {
2072
+ "epoch": 3.02,
2073
+ "learning_rate": 0.0001,
2074
+ "loss": 1.4642,
2075
+ "step": 170000
2076
+ },
2077
+ {
2078
+ "epoch": 3.03,
2079
+ "learning_rate": 0.0001,
2080
+ "loss": 1.4671,
2081
+ "step": 170500
2082
+ },
2083
+ {
2084
+ "epoch": 3.04,
2085
+ "learning_rate": 0.0001,
2086
+ "loss": 1.4707,
2087
+ "step": 171000
2088
+ },
2089
+ {
2090
+ "epoch": 3.05,
2091
+ "learning_rate": 0.0001,
2092
+ "loss": 1.4747,
2093
+ "step": 171500
2094
+ },
2095
+ {
2096
+ "epoch": 3.06,
2097
+ "learning_rate": 0.0001,
2098
+ "loss": 1.4747,
2099
+ "step": 172000
2100
+ },
2101
+ {
2102
+ "epoch": 3.07,
2103
+ "learning_rate": 0.0001,
2104
+ "loss": 1.4657,
2105
+ "step": 172500
2106
+ },
2107
+ {
2108
+ "epoch": 3.08,
2109
+ "learning_rate": 0.0001,
2110
+ "loss": 1.4594,
2111
+ "step": 173000
2112
+ },
2113
+ {
2114
+ "epoch": 3.08,
2115
+ "learning_rate": 0.0001,
2116
+ "loss": 1.4777,
2117
+ "step": 173500
2118
+ },
2119
+ {
2120
+ "epoch": 3.09,
2121
+ "learning_rate": 0.0001,
2122
+ "loss": 1.4625,
2123
+ "step": 174000
2124
+ },
2125
+ {
2126
+ "epoch": 3.1,
2127
+ "learning_rate": 0.0001,
2128
+ "loss": 1.4824,
2129
+ "step": 174500
2130
+ },
2131
+ {
2132
+ "epoch": 3.11,
2133
+ "learning_rate": 0.0001,
2134
+ "loss": 1.4641,
2135
+ "step": 175000
2136
+ },
2137
+ {
2138
+ "epoch": 3.12,
2139
+ "learning_rate": 0.0001,
2140
+ "loss": 1.4659,
2141
+ "step": 175500
2142
+ },
2143
+ {
2144
+ "epoch": 3.13,
2145
+ "learning_rate": 0.0001,
2146
+ "loss": 1.4875,
2147
+ "step": 176000
2148
+ },
2149
+ {
2150
+ "epoch": 3.14,
2151
+ "learning_rate": 0.0001,
2152
+ "loss": 1.4798,
2153
+ "step": 176500
2154
+ },
2155
+ {
2156
+ "epoch": 3.15,
2157
+ "learning_rate": 0.0001,
2158
+ "loss": 1.4589,
2159
+ "step": 177000
2160
+ },
2161
+ {
2162
+ "epoch": 3.16,
2163
+ "learning_rate": 0.0001,
2164
+ "loss": 1.4544,
2165
+ "step": 177500
2166
+ },
2167
+ {
2168
+ "epoch": 3.16,
2169
+ "learning_rate": 0.0001,
2170
+ "loss": 1.4623,
2171
+ "step": 178000
2172
+ },
2173
+ {
2174
+ "epoch": 3.17,
2175
+ "learning_rate": 0.0001,
2176
+ "loss": 1.4754,
2177
+ "step": 178500
2178
+ },
2179
+ {
2180
+ "epoch": 3.18,
2181
+ "learning_rate": 0.0001,
2182
+ "loss": 1.4766,
2183
+ "step": 179000
2184
+ },
2185
+ {
2186
+ "epoch": 3.19,
2187
+ "learning_rate": 0.0001,
2188
+ "loss": 1.4635,
2189
+ "step": 179500
2190
+ },
2191
+ {
2192
+ "epoch": 3.2,
2193
+ "learning_rate": 0.0001,
2194
+ "loss": 1.4802,
2195
+ "step": 180000
2196
+ },
2197
+ {
2198
+ "epoch": 3.21,
2199
+ "learning_rate": 0.0001,
2200
+ "loss": 1.4808,
2201
+ "step": 180500
2202
+ },
2203
+ {
2204
+ "epoch": 3.22,
2205
+ "learning_rate": 0.0001,
2206
+ "loss": 1.4688,
2207
+ "step": 181000
2208
+ },
2209
+ {
2210
+ "epoch": 3.23,
2211
+ "learning_rate": 0.0001,
2212
+ "loss": 1.4749,
2213
+ "step": 181500
2214
+ },
2215
+ {
2216
+ "epoch": 3.24,
2217
+ "learning_rate": 0.0001,
2218
+ "loss": 1.4646,
2219
+ "step": 182000
2220
+ },
2221
+ {
2222
+ "epoch": 3.24,
2223
+ "learning_rate": 0.0001,
2224
+ "loss": 1.4751,
2225
+ "step": 182500
2226
+ },
2227
+ {
2228
+ "epoch": 3.25,
2229
+ "learning_rate": 0.0001,
2230
+ "loss": 1.4729,
2231
+ "step": 183000
2232
+ },
2233
+ {
2234
+ "epoch": 3.26,
2235
+ "learning_rate": 0.0001,
2236
+ "loss": 1.4595,
2237
+ "step": 183500
2238
+ },
2239
+ {
2240
+ "epoch": 3.27,
2241
+ "learning_rate": 0.0001,
2242
+ "loss": 1.4657,
2243
+ "step": 184000
2244
+ },
2245
+ {
2246
+ "epoch": 3.28,
2247
+ "learning_rate": 0.0001,
2248
+ "loss": 1.4852,
2249
+ "step": 184500
2250
+ },
2251
+ {
2252
+ "epoch": 3.29,
2253
+ "learning_rate": 0.0001,
2254
+ "loss": 1.4584,
2255
+ "step": 185000
2256
+ },
2257
+ {
2258
+ "epoch": 3.3,
2259
+ "learning_rate": 0.0001,
2260
+ "loss": 1.4669,
2261
+ "step": 185500
2262
+ },
2263
+ {
2264
+ "epoch": 3.31,
2265
+ "learning_rate": 0.0001,
2266
+ "loss": 1.4729,
2267
+ "step": 186000
2268
+ },
2269
+ {
2270
+ "epoch": 3.32,
2271
+ "learning_rate": 0.0001,
2272
+ "loss": 1.465,
2273
+ "step": 186500
2274
+ },
2275
+ {
2276
+ "epoch": 3.32,
2277
+ "learning_rate": 0.0001,
2278
+ "loss": 1.4668,
2279
+ "step": 187000
2280
+ },
2281
+ {
2282
+ "epoch": 3.33,
2283
+ "learning_rate": 0.0001,
2284
+ "loss": 1.4732,
2285
+ "step": 187500
2286
+ },
2287
+ {
2288
+ "epoch": 3.34,
2289
+ "learning_rate": 0.0001,
2290
+ "loss": 1.4719,
2291
+ "step": 188000
2292
+ },
2293
+ {
2294
+ "epoch": 3.35,
2295
+ "learning_rate": 0.0001,
2296
+ "loss": 1.4664,
2297
+ "step": 188500
2298
+ },
2299
+ {
2300
+ "epoch": 3.36,
2301
+ "learning_rate": 0.0001,
2302
+ "loss": 1.4694,
2303
+ "step": 189000
2304
+ },
2305
+ {
2306
+ "epoch": 3.37,
2307
+ "learning_rate": 0.0001,
2308
+ "loss": 1.4655,
2309
+ "step": 189500
2310
+ },
2311
+ {
2312
+ "epoch": 3.38,
2313
+ "learning_rate": 0.0001,
2314
+ "loss": 1.4708,
2315
+ "step": 190000
2316
+ },
2317
+ {
2318
+ "epoch": 3.39,
2319
+ "learning_rate": 0.0001,
2320
+ "loss": 1.4463,
2321
+ "step": 190500
2322
+ },
2323
+ {
2324
+ "epoch": 3.4,
2325
+ "learning_rate": 0.0001,
2326
+ "loss": 1.4351,
2327
+ "step": 191000
2328
+ },
2329
+ {
2330
+ "epoch": 3.4,
2331
+ "learning_rate": 0.0001,
2332
+ "loss": 1.4348,
2333
+ "step": 191500
2334
+ },
2335
+ {
2336
+ "epoch": 3.41,
2337
+ "learning_rate": 0.0001,
2338
+ "loss": 1.4408,
2339
+ "step": 192000
2340
+ },
2341
+ {
2342
+ "epoch": 3.42,
2343
+ "learning_rate": 0.0001,
2344
+ "loss": 1.4523,
2345
+ "step": 192500
2346
+ },
2347
+ {
2348
+ "epoch": 3.43,
2349
+ "learning_rate": 0.0001,
2350
+ "loss": 1.4364,
2351
+ "step": 193000
2352
+ },
2353
+ {
2354
+ "epoch": 3.44,
2355
+ "learning_rate": 0.0001,
2356
+ "loss": 1.4361,
2357
+ "step": 193500
2358
+ },
2359
+ {
2360
+ "epoch": 3.45,
2361
+ "learning_rate": 0.0001,
2362
+ "loss": 1.4307,
2363
+ "step": 194000
2364
+ },
2365
+ {
2366
+ "epoch": 3.46,
2367
+ "learning_rate": 0.0001,
2368
+ "loss": 1.4477,
2369
+ "step": 194500
2370
+ },
2371
+ {
2372
+ "epoch": 3.47,
2373
+ "learning_rate": 0.0001,
2374
+ "loss": 1.4355,
2375
+ "step": 195000
2376
+ },
2377
+ {
2378
+ "epoch": 3.48,
2379
+ "learning_rate": 0.0001,
2380
+ "loss": 1.4516,
2381
+ "step": 195500
2382
+ },
2383
+ {
2384
+ "epoch": 3.48,
2385
+ "learning_rate": 0.0001,
2386
+ "loss": 1.4461,
2387
+ "step": 196000
2388
+ },
2389
+ {
2390
+ "epoch": 3.49,
2391
+ "learning_rate": 0.0001,
2392
+ "loss": 1.432,
2393
+ "step": 196500
2394
+ },
2395
+ {
2396
+ "epoch": 3.5,
2397
+ "learning_rate": 0.0001,
2398
+ "loss": 1.4464,
2399
+ "step": 197000
2400
+ },
2401
+ {
2402
+ "epoch": 3.51,
2403
+ "learning_rate": 0.0001,
2404
+ "loss": 1.4444,
2405
+ "step": 197500
2406
+ },
2407
+ {
2408
+ "epoch": 3.52,
2409
+ "learning_rate": 0.0001,
2410
+ "loss": 1.4416,
2411
+ "step": 198000
2412
+ },
2413
+ {
2414
+ "epoch": 3.53,
2415
+ "learning_rate": 0.0001,
2416
+ "loss": 1.445,
2417
+ "step": 198500
2418
+ },
2419
+ {
2420
+ "epoch": 3.54,
2421
+ "learning_rate": 0.0001,
2422
+ "loss": 1.4472,
2423
+ "step": 199000
2424
+ },
2425
+ {
2426
+ "epoch": 3.55,
2427
+ "learning_rate": 0.0001,
2428
+ "loss": 1.4443,
2429
+ "step": 199500
2430
+ },
2431
+ {
2432
+ "epoch": 3.56,
2433
+ "learning_rate": 0.0001,
2434
+ "loss": 1.4274,
2435
+ "step": 200000
2436
+ },
2437
+ {
2438
+ "epoch": 3.56,
2439
+ "learning_rate": 0.0001,
2440
+ "loss": 1.4552,
2441
+ "step": 200500
2442
+ },
2443
+ {
2444
+ "epoch": 3.57,
2445
+ "learning_rate": 0.0001,
2446
+ "loss": 1.4347,
2447
+ "step": 201000
2448
+ },
2449
+ {
2450
+ "epoch": 3.58,
2451
+ "learning_rate": 0.0001,
2452
+ "loss": 1.4293,
2453
+ "step": 201500
2454
+ },
2455
+ {
2456
+ "epoch": 3.59,
2457
+ "learning_rate": 0.0001,
2458
+ "loss": 1.4355,
2459
+ "step": 202000
2460
+ },
2461
+ {
2462
+ "epoch": 3.6,
2463
+ "learning_rate": 0.0001,
2464
+ "loss": 1.4445,
2465
+ "step": 202500
2466
+ },
2467
+ {
2468
+ "epoch": 3.61,
2469
+ "learning_rate": 0.0001,
2470
+ "loss": 1.4313,
2471
+ "step": 203000
2472
+ },
2473
+ {
2474
+ "epoch": 3.62,
2475
+ "learning_rate": 0.0001,
2476
+ "loss": 1.4398,
2477
+ "step": 203500
2478
+ },
2479
+ {
2480
+ "epoch": 3.63,
2481
+ "learning_rate": 0.0001,
2482
+ "loss": 1.4468,
2483
+ "step": 204000
2484
+ },
2485
+ {
2486
+ "epoch": 3.64,
2487
+ "learning_rate": 0.0001,
2488
+ "loss": 1.4298,
2489
+ "step": 204500
2490
+ },
2491
+ {
2492
+ "epoch": 3.64,
2493
+ "learning_rate": 0.0001,
2494
+ "loss": 1.4497,
2495
+ "step": 205000
2496
+ },
2497
+ {
2498
+ "epoch": 3.65,
2499
+ "learning_rate": 0.0001,
2500
+ "loss": 1.4453,
2501
+ "step": 205500
2502
+ },
2503
+ {
2504
+ "epoch": 3.66,
2505
+ "learning_rate": 0.0001,
2506
+ "loss": 1.4357,
2507
+ "step": 206000
2508
+ },
2509
+ {
2510
+ "epoch": 3.67,
2511
+ "learning_rate": 0.0001,
2512
+ "loss": 1.4538,
2513
+ "step": 206500
2514
+ },
2515
+ {
2516
+ "epoch": 3.68,
2517
+ "learning_rate": 0.0001,
2518
+ "loss": 1.4302,
2519
+ "step": 207000
2520
+ },
2521
+ {
2522
+ "epoch": 3.69,
2523
+ "learning_rate": 0.0001,
2524
+ "loss": 1.4391,
2525
+ "step": 207500
2526
+ },
2527
+ {
2528
+ "epoch": 3.7,
2529
+ "learning_rate": 0.0001,
2530
+ "loss": 1.4429,
2531
+ "step": 208000
2532
+ },
2533
+ {
2534
+ "epoch": 3.71,
2535
+ "learning_rate": 0.0001,
2536
+ "loss": 1.4346,
2537
+ "step": 208500
2538
+ },
2539
+ {
2540
+ "epoch": 3.72,
2541
+ "learning_rate": 0.0001,
2542
+ "loss": 1.4406,
2543
+ "step": 209000
2544
+ },
2545
+ {
2546
+ "epoch": 3.72,
2547
+ "learning_rate": 0.0001,
2548
+ "loss": 1.4412,
2549
+ "step": 209500
2550
+ },
2551
+ {
2552
+ "epoch": 3.73,
2553
+ "learning_rate": 0.0001,
2554
+ "loss": 1.4398,
2555
+ "step": 210000
2556
+ },
2557
+ {
2558
+ "epoch": 3.74,
2559
+ "learning_rate": 0.0001,
2560
+ "loss": 1.4568,
2561
+ "step": 210500
2562
+ },
2563
+ {
2564
+ "epoch": 3.75,
2565
+ "learning_rate": 0.0001,
2566
+ "loss": 1.4383,
2567
+ "step": 211000
2568
+ },
2569
+ {
2570
+ "epoch": 3.76,
2571
+ "learning_rate": 0.0001,
2572
+ "loss": 1.4467,
2573
+ "step": 211500
2574
+ },
2575
+ {
2576
+ "epoch": 3.77,
2577
+ "learning_rate": 0.0001,
2578
+ "loss": 1.4461,
2579
+ "step": 212000
2580
+ },
2581
+ {
2582
+ "epoch": 3.78,
2583
+ "learning_rate": 0.0001,
2584
+ "loss": 1.4381,
2585
+ "step": 212500
2586
+ },
2587
+ {
2588
+ "epoch": 3.79,
2589
+ "learning_rate": 0.0001,
2590
+ "loss": 1.4449,
2591
+ "step": 213000
2592
+ },
2593
+ {
2594
+ "epoch": 3.8,
2595
+ "learning_rate": 0.0001,
2596
+ "loss": 1.4391,
2597
+ "step": 213500
2598
+ },
2599
+ {
2600
+ "epoch": 3.8,
2601
+ "learning_rate": 0.0001,
2602
+ "loss": 1.4392,
2603
+ "step": 214000
2604
+ },
2605
+ {
2606
+ "epoch": 3.81,
2607
+ "learning_rate": 0.0001,
2608
+ "loss": 1.4292,
2609
+ "step": 214500
2610
+ },
2611
+ {
2612
+ "epoch": 3.82,
2613
+ "learning_rate": 0.0001,
2614
+ "loss": 1.4242,
2615
+ "step": 215000
2616
+ },
2617
+ {
2618
+ "epoch": 3.83,
2619
+ "learning_rate": 0.0001,
2620
+ "loss": 1.4324,
2621
+ "step": 215500
2622
+ },
2623
+ {
2624
+ "epoch": 3.84,
2625
+ "learning_rate": 0.0001,
2626
+ "loss": 1.4413,
2627
+ "step": 216000
2628
+ },
2629
+ {
2630
+ "epoch": 3.85,
2631
+ "learning_rate": 0.0001,
2632
+ "loss": 1.4417,
2633
+ "step": 216500
2634
+ },
2635
+ {
2636
+ "epoch": 3.86,
2637
+ "learning_rate": 0.0001,
2638
+ "loss": 1.4337,
2639
+ "step": 217000
2640
+ },
2641
+ {
2642
+ "epoch": 3.87,
2643
+ "learning_rate": 0.0001,
2644
+ "loss": 1.4397,
2645
+ "step": 217500
2646
+ },
2647
+ {
2648
+ "epoch": 3.88,
2649
+ "learning_rate": 0.0001,
2650
+ "loss": 1.4337,
2651
+ "step": 218000
2652
+ },
2653
+ {
2654
+ "epoch": 3.88,
2655
+ "learning_rate": 0.0001,
2656
+ "loss": 1.4369,
2657
+ "step": 218500
2658
+ },
2659
+ {
2660
+ "epoch": 3.89,
2661
+ "learning_rate": 0.0001,
2662
+ "loss": 1.4321,
2663
+ "step": 219000
2664
+ },
2665
+ {
2666
+ "epoch": 3.9,
2667
+ "learning_rate": 0.0001,
2668
+ "loss": 1.4332,
2669
+ "step": 219500
2670
+ },
2671
+ {
2672
+ "epoch": 3.91,
2673
+ "learning_rate": 0.0001,
2674
+ "loss": 1.4186,
2675
+ "step": 220000
2676
+ },
2677
+ {
2678
+ "epoch": 3.92,
2679
+ "learning_rate": 0.0001,
2680
+ "loss": 1.4291,
2681
+ "step": 220500
2682
+ },
2683
+ {
2684
+ "epoch": 3.93,
2685
+ "learning_rate": 0.0001,
2686
+ "loss": 1.4397,
2687
+ "step": 221000
2688
+ },
2689
+ {
2690
+ "epoch": 3.94,
2691
+ "learning_rate": 0.0001,
2692
+ "loss": 1.4441,
2693
+ "step": 221500
2694
+ },
2695
+ {
2696
+ "epoch": 3.95,
2697
+ "learning_rate": 0.0001,
2698
+ "loss": 1.4286,
2699
+ "step": 222000
2700
+ },
2701
+ {
2702
+ "epoch": 3.96,
2703
+ "learning_rate": 0.0001,
2704
+ "loss": 1.4308,
2705
+ "step": 222500
2706
+ },
2707
+ {
2708
+ "epoch": 3.96,
2709
+ "learning_rate": 0.0001,
2710
+ "loss": 1.4285,
2711
+ "step": 223000
2712
+ },
2713
+ {
2714
+ "epoch": 3.97,
2715
+ "learning_rate": 0.0001,
2716
+ "loss": 1.4328,
2717
+ "step": 223500
2718
+ },
2719
+ {
2720
+ "epoch": 3.98,
2721
+ "learning_rate": 0.0001,
2722
+ "loss": 1.4368,
2723
+ "step": 224000
2724
+ },
2725
+ {
2726
+ "epoch": 3.99,
2727
+ "learning_rate": 0.0001,
2728
+ "loss": 1.4324,
2729
+ "step": 224500
2730
+ },
2731
+ {
2732
+ "epoch": 4.0,
2733
+ "learning_rate": 0.0001,
2734
+ "loss": 1.4309,
2735
+ "step": 225000
2736
+ },
2737
+ {
2738
+ "epoch": 4.0,
2739
+ "step": 225000,
2740
+ "total_flos": 0,
2741
+ "train_runtime": 1.098,
2742
+ "train_samples_per_second": 204925.455
2743
+ }
2744
+ ],
2745
+ "max_steps": 225000,
2746
+ "num_train_epochs": 4,
2747
+ "total_flos": 0,
2748
+ "trial_name": null,
2749
+ "trial_params": null
2750
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5cff30556273ccbc0392d9c7befc2da71b6809a94554830a029cd2fef673f6c
3
+ size 2683