Hugging Face's logo

PurCL
/
casp-moco-lmpa-c-only-sig

Transformers
Safetensors
assembly-source-dual-encoder
Model card Files Community
ziansu commited on
Commit
1308ff7
·
verified ·
1 Parent(s): 9cbd1c3

Training in progress, step 5000, checkpoint

Browse files
Files changed (8) hide show
  1. checkpoint-5000/config.json +184 -0
  2. checkpoint-5000/model.safetensors +3 -0
  3. checkpoint-5000/optimizer.pt +3 -0
  4. checkpoint-5000/rng_state_0.pth +3 -0
  5. checkpoint-5000/rng_state_1.pth +3 -0
  6. checkpoint-5000/scheduler.pt +3 -0
  7. checkpoint-5000/trainer_state.json +3601 -0
  8. checkpoint-5000/training_args.bin +3 -0
checkpoint-5000/config.json ADDED
@@ -0,0 +1,184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "K": 4096,
3
+ "architectures": [
4
+ "MomentumDualEncoderModel"
5
+ ],
6
+ "assembly_config": {
7
+ "_name_or_path": "",
8
+ "add_cross_attention": false,
9
+ "architectures": [
10
+ "LongelmForMaskedLMWithEdgePrediction"
11
+ ],
12
+ "attention_probs_dropout_prob": 0.1,
13
+ "bad_words_ids": null,
14
+ "begin_suppress_tokens": null,
15
+ "block_size": 8,
16
+ "bos_token_id": 1,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": null,
19
+ "cross_attention_hidden_size": null,
20
+ "decoder_start_token_id": null,
21
+ "diversity_penalty": 0.0,
22
+ "do_sample": false,
23
+ "early_stopping": false,
24
+ "encoder_no_repeat_ngram_size": 0,
25
+ "eos_token_id": 2,
26
+ "ep_add_linear_projection": true,
27
+ "exponential_decay_length_penalty": null,
28
+ "finetuning_task": null,
29
+ "forced_bos_token_id": null,
30
+ "forced_eos_token_id": null,
31
+ "global_memory_size": 1,
32
+ "hidden_act": "gelu",
33
+ "hidden_dropout_prob": 0.1,
34
+ "hidden_size": 768,
35
+ "id2label": {
36
+ "0": "LABEL_0",
37
+ "1": "LABEL_1"
38
+ },
39
+ "initializer_range": 0.02,
40
+ "intermediate_size": 3072,
41
+ "is_decoder": false,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layer_norm_eps": 1e-12,
48
+ "length_penalty": 1.0,
49
+ "max_blocks": 200,
50
+ "max_length": 20,
51
+ "max_position_embeddings": 1805,
52
+ "max_relative_position_embeddings": 8,
53
+ "min_length": 0,
54
+ "model_type": "longelm",
55
+ "no_repeat_ngram_size": 0,
56
+ "node_size": 1,
57
+ "num_attention_heads": 12,
58
+ "num_beam_groups": 1,
59
+ "num_beams": 1,
60
+ "num_hidden_layers": 12,
61
+ "num_return_sequences": 1,
62
+ "output_attentions": false,
63
+ "output_hidden_states": false,
64
+ "output_scores": false,
65
+ "pad_token_id": 3,
66
+ "position_embedding_type": "mixed",
67
+ "prefix": null,
68
+ "problem_type": null,
69
+ "pruned_heads": {},
70
+ "remove_invalid_values": false,
71
+ "repetition_penalty": 1.0,
72
+ "return_dict": true,
73
+ "return_dict_in_generate": false,
74
+ "sep_token_id": null,
75
+ "suppress_tokens": null,
76
+ "task_specific_params": null,
77
+ "temperature": 1.0,
78
+ "tf_legacy_loss": false,
79
+ "tie_encoder_decoder": false,
80
+ "tie_word_embeddings": true,
81
+ "tokenizer_class": null,
82
+ "top_k": 50,
83
+ "top_p": 1.0,
84
+ "torch_dtype": "float32",
85
+ "torchscript": false,
86
+ "type_vocab_size": 2,
87
+ "typical_p": 1.0,
88
+ "use_bfloat16": false,
89
+ "use_cache": true,
90
+ "vocab_size": 50265
91
+ },
92
+ "logit_scale_init_value": 2.6592,
93
+ "m": 0.999,
94
+ "model_type": "assembly-source-dual-encoder",
95
+ "projection_dim": 512,
96
+ "source_config": {
97
+ "_name_or_path": "Salesforce/codet5p-110m-embedding",
98
+ "add_cross_attention": false,
99
+ "architectures": [
100
+ "CodeT5p_Embedding"
101
+ ],
102
+ "auto_map": {
103
+ "AutoConfig": "Salesforce/codet5p-110m-embedding--configuration_codet5p_embedding.CodeT5pEmbeddingConfig",
104
+ "AutoModel": "Salesforce/codet5p-110m-embedding--modeling_codet5p_embedding.CodeT5pEmbeddingModel"
105
+ },
106
+ "bad_words_ids": null,
107
+ "begin_suppress_tokens": null,
108
+ "bos_token_id": 1,
109
+ "chunk_size_feed_forward": 0,
110
+ "cross_attention_hidden_size": null,
111
+ "d_ff": 3072,
112
+ "d_kv": 64,
113
+ "d_model": 768,
114
+ "decoder_start_token_id": 0,
115
+ "dense_act_fn": "relu",
116
+ "diversity_penalty": 0.0,
117
+ "do_sample": false,
118
+ "dropout_rate": 0.1,
119
+ "early_stopping": false,
120
+ "embed_dim": 256,
121
+ "encoder_no_repeat_ngram_size": 0,
122
+ "eos_token_id": 2,
123
+ "exponential_decay_length_penalty": null,
124
+ "feed_forward_proj": "relu",
125
+ "finetuning_task": null,
126
+ "forced_bos_token_id": null,
127
+ "forced_eos_token_id": null,
128
+ "gradient_checkpointing": false,
129
+ "id2label": {
130
+ "0": "LABEL_0"
131
+ },
132
+ "initializer_factor": 1.0,
133
+ "is_decoder": false,
134
+ "is_encoder_decoder": true,
135
+ "is_gated_act": false,
136
+ "label2id": {
137
+ "LABEL_0": 0
138
+ },
139
+ "layer_norm_epsilon": 1e-06,
140
+ "length_penalty": 1.0,
141
+ "max_length": 20,
142
+ "min_length": 0,
143
+ "model_type": "codet5p_embedding",
144
+ "n_positions": 512,
145
+ "no_repeat_ngram_size": 0,
146
+ "num_beam_groups": 1,
147
+ "num_beams": 1,
148
+ "num_heads": 12,
149
+ "num_layers": 12,
150
+ "num_return_sequences": 1,
151
+ "output_attentions": false,
152
+ "output_hidden_states": false,
153
+ "output_past": true,
154
+ "output_scores": false,
155
+ "pad_token_id": 0,
156
+ "prefix": null,
157
+ "problem_type": null,
158
+ "pruned_heads": {},
159
+ "relative_attention_max_distance": 128,
160
+ "relative_attention_num_buckets": 32,
161
+ "remove_invalid_values": false,
162
+ "repetition_penalty": 1.0,
163
+ "return_dict": true,
164
+ "return_dict_in_generate": false,
165
+ "sep_token_id": null,
166
+ "suppress_tokens": null,
167
+ "task_specific_params": null,
168
+ "temperature": 1.0,
169
+ "tf_legacy_loss": false,
170
+ "tie_encoder_decoder": false,
171
+ "tie_word_embeddings": true,
172
+ "tokenizer_class": null,
173
+ "top_k": 50,
174
+ "top_p": 1.0,
175
+ "torch_dtype": "float32",
176
+ "torchscript": false,
177
+ "typical_p": 1.0,
178
+ "use_bfloat16": false,
179
+ "use_cache": false,
180
+ "vocab_size": 32103
181
+ },
182
+ "torch_dtype": "float32",
183
+ "transformers_version": "4.39.3"
184
+ }
checkpoint-5000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c255a21eecff284af3ab43026149b5890339f1c8840b0d01ca9b7abb2373f640
3
+ size 1901991436
checkpoint-5000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43a8adaa1b12f3e2a71b7bd39693a0a71033c554867f8953536be57a1f671223
3
+ size 1888540293
checkpoint-5000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:200a9ee9ae70fc60c12d0bd32bdd5422fb503be83aaf4239efd1ef0cb48da5fa
3
+ size 15607
checkpoint-5000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ed6944ef76ff65f50ff24bbcc6828a2e21dde581ee9a7bd8873c286fb293fd71
3
+ size 15607
checkpoint-5000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d93799775e15c74cf6f6f99bbb3b9898341f84d0714ef8670acccaa3233abec
3
+ size 627
checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,3601 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.1984126984126984,
5
+ "eval_steps": 500,
6
+ "global_step": 5000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "grad_norm": 23.495065689086914,
14
+ "learning_rate": 4.0000000000000003e-07,
15
+ "loss": 8.6232,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.0,
20
+ "grad_norm": 24.87381935119629,
21
+ "learning_rate": 9e-07,
22
+ "loss": 8.4514,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.0,
27
+ "grad_norm": 22.63846778869629,
28
+ "learning_rate": 1.4000000000000001e-06,
29
+ "loss": 8.3152,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0,
34
+ "grad_norm": 20.65507698059082,
35
+ "learning_rate": 1.9e-06,
36
+ "loss": 7.9732,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.0,
41
+ "grad_norm": 16.147356033325195,
42
+ "learning_rate": 2.4000000000000003e-06,
43
+ "loss": 7.6879,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "grad_norm": 14.252687454223633,
49
+ "learning_rate": 2.9e-06,
50
+ "loss": 7.4152,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.0,
55
+ "grad_norm": 12.011382102966309,
56
+ "learning_rate": 3.4000000000000005e-06,
57
+ "loss": 7.2472,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0,
62
+ "grad_norm": 10.081536293029785,
63
+ "learning_rate": 3.9e-06,
64
+ "loss": 7.2433,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.0,
69
+ "grad_norm": 9.65556526184082,
70
+ "learning_rate": 4.4e-06,
71
+ "loss": 7.2298,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.0,
76
+ "grad_norm": 8.830799102783203,
77
+ "learning_rate": 4.9000000000000005e-06,
78
+ "loss": 7.19,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.0,
83
+ "grad_norm": 9.019427299499512,
84
+ "learning_rate": 5.4e-06,
85
+ "loss": 7.2362,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "grad_norm": 9.13768482208252,
91
+ "learning_rate": 5.9e-06,
92
+ "loss": 7.2804,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.01,
97
+ "grad_norm": 9.02840805053711,
98
+ "learning_rate": 6.4000000000000006e-06,
99
+ "loss": 7.2345,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.01,
104
+ "grad_norm": 8.178731918334961,
105
+ "learning_rate": 6.900000000000001e-06,
106
+ "loss": 7.3014,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.01,
111
+ "grad_norm": 8.83882999420166,
112
+ "learning_rate": 7.4e-06,
113
+ "loss": 7.3875,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.01,
118
+ "grad_norm": 8.3687162399292,
119
+ "learning_rate": 7.9e-06,
120
+ "loss": 7.4546,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.01,
125
+ "grad_norm": 8.09121036529541,
126
+ "learning_rate": 8.400000000000001e-06,
127
+ "loss": 7.407,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.01,
132
+ "grad_norm": 12.731450080871582,
133
+ "learning_rate": 8.9e-06,
134
+ "loss": 7.4321,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.01,
139
+ "grad_norm": 7.601162910461426,
140
+ "learning_rate": 9.4e-06,
141
+ "loss": 7.4155,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.01,
146
+ "grad_norm": 8.006213188171387,
147
+ "learning_rate": 9.900000000000002e-06,
148
+ "loss": 7.3705,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.01,
153
+ "grad_norm": 7.79794454574585,
154
+ "learning_rate": 1.04e-05,
155
+ "loss": 7.3759,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.01,
160
+ "grad_norm": 7.281346321105957,
161
+ "learning_rate": 1.09e-05,
162
+ "loss": 7.3798,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.01,
167
+ "grad_norm": 7.605266094207764,
168
+ "learning_rate": 1.1400000000000001e-05,
169
+ "loss": 7.3878,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.01,
174
+ "grad_norm": 7.446412563323975,
175
+ "learning_rate": 1.19e-05,
176
+ "loss": 7.3098,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.01,
181
+ "grad_norm": 7.6324334144592285,
182
+ "learning_rate": 1.24e-05,
183
+ "loss": 7.2995,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.01,
188
+ "grad_norm": 8.330737113952637,
189
+ "learning_rate": 1.29e-05,
190
+ "loss": 7.1887,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.01,
195
+ "grad_norm": 7.283708572387695,
196
+ "learning_rate": 1.3400000000000002e-05,
197
+ "loss": 7.1184,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.01,
202
+ "grad_norm": 6.950069904327393,
203
+ "learning_rate": 1.3900000000000002e-05,
204
+ "loss": 6.9876,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.01,
209
+ "grad_norm": 6.844958782196045,
210
+ "learning_rate": 1.44e-05,
211
+ "loss": 6.9319,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "grad_norm": 7.499362468719482,
217
+ "learning_rate": 1.49e-05,
218
+ "loss": 6.9253,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.01,
223
+ "grad_norm": 8.45572280883789,
224
+ "learning_rate": 1.54e-05,
225
+ "loss": 6.7999,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.01,
230
+ "grad_norm": 8.108625411987305,
231
+ "learning_rate": 1.59e-05,
232
+ "loss": 6.7667,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.01,
237
+ "grad_norm": 8.131596565246582,
238
+ "learning_rate": 1.6400000000000002e-05,
239
+ "loss": 6.7599,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.01,
244
+ "grad_norm": 7.2159247398376465,
245
+ "learning_rate": 1.69e-05,
246
+ "loss": 6.7379,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.01,
251
+ "grad_norm": 7.335048675537109,
252
+ "learning_rate": 1.74e-05,
253
+ "loss": 6.6553,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.01,
258
+ "grad_norm": 7.361081600189209,
259
+ "learning_rate": 1.79e-05,
260
+ "loss": 6.6905,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.01,
265
+ "grad_norm": 7.59442138671875,
266
+ "learning_rate": 1.84e-05,
267
+ "loss": 6.6356,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.02,
272
+ "grad_norm": 7.31695032119751,
273
+ "learning_rate": 1.8900000000000002e-05,
274
+ "loss": 6.6538,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.02,
279
+ "grad_norm": 7.541380882263184,
280
+ "learning_rate": 1.94e-05,
281
+ "loss": 6.6873,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.02,
286
+ "grad_norm": 8.056180000305176,
287
+ "learning_rate": 1.9900000000000003e-05,
288
+ "loss": 6.6332,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.02,
293
+ "grad_norm": 6.74859094619751,
294
+ "learning_rate": 2.04e-05,
295
+ "loss": 6.5488,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.02,
300
+ "grad_norm": 7.700525760650635,
301
+ "learning_rate": 2.09e-05,
302
+ "loss": 6.611,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.02,
307
+ "grad_norm": 7.347009181976318,
308
+ "learning_rate": 2.1400000000000002e-05,
309
+ "loss": 6.6621,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.02,
314
+ "grad_norm": 6.607337951660156,
315
+ "learning_rate": 2.19e-05,
316
+ "loss": 6.568,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.02,
321
+ "grad_norm": 6.816825866699219,
322
+ "learning_rate": 2.2400000000000002e-05,
323
+ "loss": 6.6293,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.02,
328
+ "grad_norm": 6.984443664550781,
329
+ "learning_rate": 2.29e-05,
330
+ "loss": 6.5667,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.02,
335
+ "grad_norm": 6.995021820068359,
336
+ "learning_rate": 2.3400000000000003e-05,
337
+ "loss": 6.5062,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.02,
342
+ "grad_norm": 6.876286029815674,
343
+ "learning_rate": 2.39e-05,
344
+ "loss": 6.5034,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.02,
349
+ "grad_norm": 6.953365802764893,
350
+ "learning_rate": 2.44e-05,
351
+ "loss": 6.459,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.02,
356
+ "grad_norm": 7.246520519256592,
357
+ "learning_rate": 2.4900000000000002e-05,
358
+ "loss": 6.4359,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.02,
363
+ "eval_loss": 6.133457660675049,
364
+ "eval_runtime": 422.8798,
365
+ "eval_samples_per_second": 50.182,
366
+ "eval_steps_per_second": 0.393,
367
+ "step": 500
368
+ },
369
+ {
370
+ "epoch": 0.02,
371
+ "grad_norm": 7.1657819747924805,
372
+ "learning_rate": 2.54e-05,
373
+ "loss": 6.4372,
374
+ "step": 510
375
+ },
376
+ {
377
+ "epoch": 0.02,
378
+ "grad_norm": 7.203813552856445,
379
+ "learning_rate": 2.5900000000000003e-05,
380
+ "loss": 6.3684,
381
+ "step": 520
382
+ },
383
+ {
384
+ "epoch": 0.02,
385
+ "grad_norm": 6.829228401184082,
386
+ "learning_rate": 2.64e-05,
387
+ "loss": 6.3275,
388
+ "step": 530
389
+ },
390
+ {
391
+ "epoch": 0.02,
392
+ "grad_norm": 6.641738414764404,
393
+ "learning_rate": 2.6900000000000003e-05,
394
+ "loss": 6.2923,
395
+ "step": 540
396
+ },
397
+ {
398
+ "epoch": 0.02,
399
+ "grad_norm": 7.824960708618164,
400
+ "learning_rate": 2.7400000000000002e-05,
401
+ "loss": 6.3685,
402
+ "step": 550
403
+ },
404
+ {
405
+ "epoch": 0.02,
406
+ "grad_norm": 6.7754364013671875,
407
+ "learning_rate": 2.7900000000000004e-05,
408
+ "loss": 6.3069,
409
+ "step": 560
410
+ },
411
+ {
412
+ "epoch": 0.02,
413
+ "grad_norm": 7.3608903884887695,
414
+ "learning_rate": 2.84e-05,
415
+ "loss": 6.2915,
416
+ "step": 570
417
+ },
418
+ {
419
+ "epoch": 0.02,
420
+ "grad_norm": 6.551064968109131,
421
+ "learning_rate": 2.8899999999999998e-05,
422
+ "loss": 6.1676,
423
+ "step": 580
424
+ },
425
+ {
426
+ "epoch": 0.02,
427
+ "grad_norm": 6.588657379150391,
428
+ "learning_rate": 2.94e-05,
429
+ "loss": 6.2311,
430
+ "step": 590
431
+ },
432
+ {
433
+ "epoch": 0.02,
434
+ "grad_norm": 6.631275177001953,
435
+ "learning_rate": 2.9900000000000002e-05,
436
+ "loss": 6.1824,
437
+ "step": 600
438
+ },
439
+ {
440
+ "epoch": 0.02,
441
+ "grad_norm": 6.2990288734436035,
442
+ "learning_rate": 3.04e-05,
443
+ "loss": 6.1641,
444
+ "step": 610
445
+ },
446
+ {
447
+ "epoch": 0.02,
448
+ "grad_norm": 6.6439714431762695,
449
+ "learning_rate": 3.09e-05,
450
+ "loss": 6.0125,
451
+ "step": 620
452
+ },
453
+ {
454
+ "epoch": 0.03,
455
+ "grad_norm": 6.402431964874268,
456
+ "learning_rate": 3.1400000000000004e-05,
457
+ "loss": 6.1314,
458
+ "step": 630
459
+ },
460
+ {
461
+ "epoch": 0.03,
462
+ "grad_norm": 5.883937358856201,
463
+ "learning_rate": 3.19e-05,
464
+ "loss": 5.9858,
465
+ "step": 640
466
+ },
467
+ {
468
+ "epoch": 0.03,
469
+ "grad_norm": 5.988973140716553,
470
+ "learning_rate": 3.24e-05,
471
+ "loss": 5.9886,
472
+ "step": 650
473
+ },
474
+ {
475
+ "epoch": 0.03,
476
+ "grad_norm": 6.186301231384277,
477
+ "learning_rate": 3.29e-05,
478
+ "loss": 5.9732,
479
+ "step": 660
480
+ },
481
+ {
482
+ "epoch": 0.03,
483
+ "grad_norm": 6.2982354164123535,
484
+ "learning_rate": 3.3400000000000005e-05,
485
+ "loss": 6.0391,
486
+ "step": 670
487
+ },
488
+ {
489
+ "epoch": 0.03,
490
+ "grad_norm": 6.037730693817139,
491
+ "learning_rate": 3.3900000000000004e-05,
492
+ "loss": 6.0547,
493
+ "step": 680
494
+ },
495
+ {
496
+ "epoch": 0.03,
497
+ "grad_norm": 6.280949592590332,
498
+ "learning_rate": 3.4399999999999996e-05,
499
+ "loss": 5.8734,
500
+ "step": 690
501
+ },
502
+ {
503
+ "epoch": 0.03,
504
+ "grad_norm": 6.861902236938477,
505
+ "learning_rate": 3.49e-05,
506
+ "loss": 5.8997,
507
+ "step": 700
508
+ },
509
+ {
510
+ "epoch": 0.03,
511
+ "grad_norm": 6.4033589363098145,
512
+ "learning_rate": 3.54e-05,
513
+ "loss": 5.8111,
514
+ "step": 710
515
+ },
516
+ {
517
+ "epoch": 0.03,
518
+ "grad_norm": 6.376819133758545,
519
+ "learning_rate": 3.59e-05,
520
+ "loss": 5.9027,
521
+ "step": 720
522
+ },
523
+ {
524
+ "epoch": 0.03,
525
+ "grad_norm": 5.777069568634033,
526
+ "learning_rate": 3.6400000000000004e-05,
527
+ "loss": 5.727,
528
+ "step": 730
529
+ },
530
+ {
531
+ "epoch": 0.03,
532
+ "grad_norm": 6.183162689208984,
533
+ "learning_rate": 3.69e-05,
534
+ "loss": 5.733,
535
+ "step": 740
536
+ },
537
+ {
538
+ "epoch": 0.03,
539
+ "grad_norm": 6.658604621887207,
540
+ "learning_rate": 3.74e-05,
541
+ "loss": 5.7018,
542
+ "step": 750
543
+ },
544
+ {
545
+ "epoch": 0.03,
546
+ "grad_norm": 6.263491153717041,
547
+ "learning_rate": 3.79e-05,
548
+ "loss": 5.7315,
549
+ "step": 760
550
+ },
551
+ {
552
+ "epoch": 0.03,
553
+ "grad_norm": 5.86756706237793,
554
+ "learning_rate": 3.8400000000000005e-05,
555
+ "loss": 5.6637,
556
+ "step": 770
557
+ },
558
+ {
559
+ "epoch": 0.03,
560
+ "grad_norm": 6.122657299041748,
561
+ "learning_rate": 3.8900000000000004e-05,
562
+ "loss": 5.6849,
563
+ "step": 780
564
+ },
565
+ {
566
+ "epoch": 0.03,
567
+ "grad_norm": 5.6024580001831055,
568
+ "learning_rate": 3.94e-05,
569
+ "loss": 5.5835,
570
+ "step": 790
571
+ },
572
+ {
573
+ "epoch": 0.03,
574
+ "grad_norm": 5.619604587554932,
575
+ "learning_rate": 3.99e-05,
576
+ "loss": 5.7,
577
+ "step": 800
578
+ },
579
+ {
580
+ "epoch": 0.03,
581
+ "grad_norm": 5.7342352867126465,
582
+ "learning_rate": 4.0400000000000006e-05,
583
+ "loss": 5.4825,
584
+ "step": 810
585
+ },
586
+ {
587
+ "epoch": 0.03,
588
+ "grad_norm": 6.796237468719482,
589
+ "learning_rate": 4.09e-05,
590
+ "loss": 5.6496,
591
+ "step": 820
592
+ },
593
+ {
594
+ "epoch": 0.03,
595
+ "grad_norm": 5.601135730743408,
596
+ "learning_rate": 4.14e-05,
597
+ "loss": 5.3526,
598
+ "step": 830
599
+ },
600
+ {
601
+ "epoch": 0.03,
602
+ "grad_norm": 5.576679229736328,
603
+ "learning_rate": 4.19e-05,
604
+ "loss": 5.3757,
605
+ "step": 840
606
+ },
607
+ {
608
+ "epoch": 0.03,
609
+ "grad_norm": 5.774380683898926,
610
+ "learning_rate": 4.24e-05,
611
+ "loss": 5.5164,
612
+ "step": 850
613
+ },
614
+ {
615
+ "epoch": 0.03,
616
+ "grad_norm": 6.140038967132568,
617
+ "learning_rate": 4.29e-05,
618
+ "loss": 5.3124,
619
+ "step": 860
620
+ },
621
+ {
622
+ "epoch": 0.03,
623
+ "grad_norm": 5.971868515014648,
624
+ "learning_rate": 4.3400000000000005e-05,
625
+ "loss": 5.3387,
626
+ "step": 870
627
+ },
628
+ {
629
+ "epoch": 0.03,
630
+ "grad_norm": 6.431824207305908,
631
+ "learning_rate": 4.39e-05,
632
+ "loss": 5.4363,
633
+ "step": 880
634
+ },
635
+ {
636
+ "epoch": 0.04,
637
+ "grad_norm": 6.174604892730713,
638
+ "learning_rate": 4.44e-05,
639
+ "loss": 5.282,
640
+ "step": 890
641
+ },
642
+ {
643
+ "epoch": 0.04,
644
+ "grad_norm": 6.313474178314209,
645
+ "learning_rate": 4.49e-05,
646
+ "loss": 5.3754,
647
+ "step": 900
648
+ },
649
+ {
650
+ "epoch": 0.04,
651
+ "grad_norm": 6.066281318664551,
652
+ "learning_rate": 4.5400000000000006e-05,
653
+ "loss": 5.3197,
654
+ "step": 910
655
+ },
656
+ {
657
+ "epoch": 0.04,
658
+ "grad_norm": 7.4433817863464355,
659
+ "learning_rate": 4.5900000000000004e-05,
660
+ "loss": 5.3277,
661
+ "step": 920
662
+ },
663
+ {
664
+ "epoch": 0.04,
665
+ "grad_norm": 5.884599208831787,
666
+ "learning_rate": 4.64e-05,
667
+ "loss": 5.0681,
668
+ "step": 930
669
+ },
670
+ {
671
+ "epoch": 0.04,
672
+ "grad_norm": 6.17655611038208,
673
+ "learning_rate": 4.69e-05,
674
+ "loss": 5.1159,
675
+ "step": 940
676
+ },
677
+ {
678
+ "epoch": 0.04,
679
+ "grad_norm": 7.090456485748291,
680
+ "learning_rate": 4.74e-05,
681
+ "loss": 5.2264,
682
+ "step": 950
683
+ },
684
+ {
685
+ "epoch": 0.04,
686
+ "grad_norm": 6.508240699768066,
687
+ "learning_rate": 4.79e-05,
688
+ "loss": 5.2113,
689
+ "step": 960
690
+ },
691
+ {
692
+ "epoch": 0.04,
693
+ "grad_norm": 6.5788445472717285,
694
+ "learning_rate": 4.8400000000000004e-05,
695
+ "loss": 4.9996,
696
+ "step": 970
697
+ },
698
+ {
699
+ "epoch": 0.04,
700
+ "grad_norm": 5.722221374511719,
701
+ "learning_rate": 4.89e-05,
702
+ "loss": 4.9688,
703
+ "step": 980
704
+ },
705
+ {
706
+ "epoch": 0.04,
707
+ "grad_norm": 5.813933849334717,
708
+ "learning_rate": 4.94e-05,
709
+ "loss": 4.8691,
710
+ "step": 990
711
+ },
712
+ {
713
+ "epoch": 0.04,
714
+ "grad_norm": 6.425748348236084,
715
+ "learning_rate": 4.99e-05,
716
+ "loss": 4.9883,
717
+ "step": 1000
718
+ },
719
+ {
720
+ "epoch": 0.04,
721
+ "eval_loss": 4.425854682922363,
722
+ "eval_runtime": 422.6592,
723
+ "eval_samples_per_second": 50.208,
724
+ "eval_steps_per_second": 0.393,
725
+ "step": 1000
726
+ },
727
+ {
728
+ "epoch": 0.04,
729
+ "grad_norm": 6.170848369598389,
730
+ "learning_rate": 4.9999996764541565e-05,
731
+ "loss": 4.8794,
732
+ "step": 1010
733
+ },
734
+ {
735
+ "epoch": 0.04,
736
+ "grad_norm": 6.247808933258057,
737
+ "learning_rate": 4.99999836204931e-05,
738
+ "loss": 4.9664,
739
+ "step": 1020
740
+ },
741
+ {
742
+ "epoch": 0.04,
743
+ "grad_norm": 6.27077054977417,
744
+ "learning_rate": 4.9999960365643774e-05,
745
+ "loss": 4.7753,
746
+ "step": 1030
747
+ },
748
+ {
749
+ "epoch": 0.04,
750
+ "grad_norm": 6.284353733062744,
751
+ "learning_rate": 4.999992700000298e-05,
752
+ "loss": 4.8909,
753
+ "step": 1040
754
+ },
755
+ {
756
+ "epoch": 0.04,
757
+ "grad_norm": 5.716318607330322,
758
+ "learning_rate": 4.999988352358421e-05,
759
+ "loss": 4.861,
760
+ "step": 1050
761
+ },
762
+ {
763
+ "epoch": 0.04,
764
+ "grad_norm": 6.448729991912842,
765
+ "learning_rate": 4.999982993640506e-05,
766
+ "loss": 4.8776,
767
+ "step": 1060
768
+ },
769
+ {
770
+ "epoch": 0.04,
771
+ "grad_norm": 5.401464939117432,
772
+ "learning_rate": 4.9999766238487204e-05,
773
+ "loss": 4.8507,
774
+ "step": 1070
775
+ },
776
+ {
777
+ "epoch": 0.04,
778
+ "grad_norm": 6.219631195068359,
779
+ "learning_rate": 4.999969242985639e-05,
780
+ "loss": 4.8234,
781
+ "step": 1080
782
+ },
783
+ {
784
+ "epoch": 0.04,
785
+ "grad_norm": 5.225848197937012,
786
+ "learning_rate": 4.9999608510542485e-05,
787
+ "loss": 4.6093,
788
+ "step": 1090
789
+ },
790
+ {
791
+ "epoch": 0.04,
792
+ "grad_norm": 6.117152690887451,
793
+ "learning_rate": 4.999951448057941e-05,
794
+ "loss": 4.786,
795
+ "step": 1100
796
+ },
797
+ {
798
+ "epoch": 0.04,
799
+ "grad_norm": 5.788968086242676,
800
+ "learning_rate": 4.999941034000521e-05,
801
+ "loss": 4.8389,
802
+ "step": 1110
803
+ },
804
+ {
805
+ "epoch": 0.04,
806
+ "grad_norm": 6.037156581878662,
807
+ "learning_rate": 4.999929608886198e-05,
808
+ "loss": 4.7674,
809
+ "step": 1120
810
+ },
811
+ {
812
+ "epoch": 0.04,
813
+ "grad_norm": 5.567563056945801,
814
+ "learning_rate": 4.999917172719596e-05,
815
+ "loss": 4.8247,
816
+ "step": 1130
817
+ },
818
+ {
819
+ "epoch": 0.05,
820
+ "grad_norm": 5.557987213134766,
821
+ "learning_rate": 4.999903725505742e-05,
822
+ "loss": 4.7517,
823
+ "step": 1140
824
+ },
825
+ {
826
+ "epoch": 0.05,
827
+ "grad_norm": 6.490599632263184,
828
+ "learning_rate": 4.999889267250075e-05,
829
+ "loss": 4.6037,
830
+ "step": 1150
831
+ },
832
+ {
833
+ "epoch": 0.05,
834
+ "grad_norm": 6.000855922698975,
835
+ "learning_rate": 4.999873797958443e-05,
836
+ "loss": 4.6077,
837
+ "step": 1160
838
+ },
839
+ {
840
+ "epoch": 0.05,
841
+ "grad_norm": 5.378462314605713,
842
+ "learning_rate": 4.999857317637102e-05,
843
+ "loss": 4.3662,
844
+ "step": 1170
845
+ },
846
+ {
847
+ "epoch": 0.05,
848
+ "grad_norm": 7.4290924072265625,
849
+ "learning_rate": 4.999839826292718e-05,
850
+ "loss": 4.5339,
851
+ "step": 1180
852
+ },
853
+ {
854
+ "epoch": 0.05,
855
+ "grad_norm": 5.727224826812744,
856
+ "learning_rate": 4.999821323932363e-05,
857
+ "loss": 4.5744,
858
+ "step": 1190
859
+ },
860
+ {
861
+ "epoch": 0.05,
862
+ "grad_norm": 5.893082618713379,
863
+ "learning_rate": 4.9998018105635214e-05,
864
+ "loss": 4.5247,
865
+ "step": 1200
866
+ },
867
+ {
868
+ "epoch": 0.05,
869
+ "grad_norm": 5.759618759155273,
870
+ "learning_rate": 4.999781286194085e-05,
871
+ "loss": 4.5744,
872
+ "step": 1210
873
+ },
874
+ {
875
+ "epoch": 0.05,
876
+ "grad_norm": 5.082376003265381,
877
+ "learning_rate": 4.999759750832355e-05,
878
+ "loss": 4.6011,
879
+ "step": 1220
880
+ },
881
+ {
882
+ "epoch": 0.05,
883
+ "grad_norm": 6.305749416351318,
884
+ "learning_rate": 4.999737204487039e-05,
885
+ "loss": 4.4587,
886
+ "step": 1230
887
+ },
888
+ {
889
+ "epoch": 0.05,
890
+ "grad_norm": 6.710574626922607,
891
+ "learning_rate": 4.999713647167258e-05,
892
+ "loss": 4.46,
893
+ "step": 1240
894
+ },
895
+ {
896
+ "epoch": 0.05,
897
+ "grad_norm": 6.753530502319336,
898
+ "learning_rate": 4.999689078882537e-05,
899
+ "loss": 4.5674,
900
+ "step": 1250
901
+ },
902
+ {
903
+ "epoch": 0.05,
904
+ "grad_norm": 5.949398040771484,
905
+ "learning_rate": 4.999663499642814e-05,
906
+ "loss": 4.3999,
907
+ "step": 1260
908
+ },
909
+ {
910
+ "epoch": 0.05,
911
+ "grad_norm": 6.530541896820068,
912
+ "learning_rate": 4.999636909458434e-05,
913
+ "loss": 4.533,
914
+ "step": 1270
915
+ },
916
+ {
917
+ "epoch": 0.05,
918
+ "grad_norm": 5.9646453857421875,
919
+ "learning_rate": 4.9996093083401495e-05,
920
+ "loss": 4.374,
921
+ "step": 1280
922
+ },
923
+ {
924
+ "epoch": 0.05,
925
+ "grad_norm": 6.346386432647705,
926
+ "learning_rate": 4.9995806962991246e-05,
927
+ "loss": 4.4758,
928
+ "step": 1290
929
+ },
930
+ {
931
+ "epoch": 0.05,
932
+ "grad_norm": 5.797152519226074,
933
+ "learning_rate": 4.9995510733469295e-05,
934
+ "loss": 4.6275,
935
+ "step": 1300
936
+ },
937
+ {
938
+ "epoch": 0.05,
939
+ "grad_norm": 6.426751136779785,
940
+ "learning_rate": 4.999520439495547e-05,
941
+ "loss": 4.3561,
942
+ "step": 1310
943
+ },
944
+ {
945
+ "epoch": 0.05,
946
+ "grad_norm": 5.3839030265808105,
947
+ "learning_rate": 4.9994887947573646e-05,
948
+ "loss": 4.3134,
949
+ "step": 1320
950
+ },
951
+ {
952
+ "epoch": 0.05,
953
+ "grad_norm": 6.01127290725708,
954
+ "learning_rate": 4.9994594501953555e-05,
955
+ "loss": 4.4336,
956
+ "step": 1330
957
+ },
958
+ {
959
+ "epoch": 0.05,
960
+ "grad_norm": 5.228607654571533,
961
+ "learning_rate": 4.999425884807851e-05,
962
+ "loss": 4.1543,
963
+ "step": 1340
964
+ },
965
+ {
966
+ "epoch": 0.05,
967
+ "grad_norm": 5.958486080169678,
968
+ "learning_rate": 4.999391308571788e-05,
969
+ "loss": 4.3806,
970
+ "step": 1350
971
+ },
972
+ {
973
+ "epoch": 0.05,
974
+ "grad_norm": 5.816386699676514,
975
+ "learning_rate": 4.99935572150115e-05,
976
+ "loss": 4.3759,
977
+ "step": 1360
978
+ },
979
+ {
980
+ "epoch": 0.05,
981
+ "grad_norm": 6.345638751983643,
982
+ "learning_rate": 4.9993191236103295e-05,
983
+ "loss": 4.5759,
984
+ "step": 1370
985
+ },
986
+ {
987
+ "epoch": 0.05,
988
+ "grad_norm": 5.738880634307861,
989
+ "learning_rate": 4.9992815149141294e-05,
990
+ "loss": 4.3274,
991
+ "step": 1380
992
+ },
993
+ {
994
+ "epoch": 0.06,
995
+ "grad_norm": 6.694160461425781,
996
+ "learning_rate": 4.9992428954277573e-05,
997
+ "loss": 4.376,
998
+ "step": 1390
999
+ },
1000
+ {
1001
+ "epoch": 0.06,
1002
+ "grad_norm": 5.050902843475342,
1003
+ "learning_rate": 4.999203265166834e-05,
1004
+ "loss": 4.1629,
1005
+ "step": 1400
1006
+ },
1007
+ {
1008
+ "epoch": 0.06,
1009
+ "grad_norm": 7.09380578994751,
1010
+ "learning_rate": 4.9991626241473876e-05,
1011
+ "loss": 4.3638,
1012
+ "step": 1410
1013
+ },
1014
+ {
1015
+ "epoch": 0.06,
1016
+ "grad_norm": 5.538166046142578,
1017
+ "learning_rate": 4.999120972385853e-05,
1018
+ "loss": 4.1318,
1019
+ "step": 1420
1020
+ },
1021
+ {
1022
+ "epoch": 0.06,
1023
+ "grad_norm": 6.052664279937744,
1024
+ "learning_rate": 4.9990783098990765e-05,
1025
+ "loss": 4.1856,
1026
+ "step": 1430
1027
+ },
1028
+ {
1029
+ "epoch": 0.06,
1030
+ "grad_norm": 6.707901477813721,
1031
+ "learning_rate": 4.9990346367043114e-05,
1032
+ "loss": 4.3361,
1033
+ "step": 1440
1034
+ },
1035
+ {
1036
+ "epoch": 0.06,
1037
+ "grad_norm": 5.3822150230407715,
1038
+ "learning_rate": 4.998989952819223e-05,
1039
+ "loss": 4.301,
1040
+ "step": 1450
1041
+ },
1042
+ {
1043
+ "epoch": 0.06,
1044
+ "grad_norm": 5.586170196533203,
1045
+ "learning_rate": 4.998944258261879e-05,
1046
+ "loss": 4.3131,
1047
+ "step": 1460
1048
+ },
1049
+ {
1050
+ "epoch": 0.06,
1051
+ "grad_norm": 5.62904691696167,
1052
+ "learning_rate": 4.998897553050763e-05,
1053
+ "loss": 4.3357,
1054
+ "step": 1470
1055
+ },
1056
+ {
1057
+ "epoch": 0.06,
1058
+ "grad_norm": 5.669075965881348,
1059
+ "learning_rate": 4.9988498372047633e-05,
1060
+ "loss": 4.2434,
1061
+ "step": 1480
1062
+ },
1063
+ {
1064
+ "epoch": 0.06,
1065
+ "grad_norm": 5.776549339294434,
1066
+ "learning_rate": 4.998801110743177e-05,
1067
+ "loss": 4.0441,
1068
+ "step": 1490
1069
+ },
1070
+ {
1071
+ "epoch": 0.06,
1072
+ "grad_norm": 5.442574977874756,
1073
+ "learning_rate": 4.9987513736857106e-05,
1074
+ "loss": 4.0931,
1075
+ "step": 1500
1076
+ },
1077
+ {
1078
+ "epoch": 0.06,
1079
+ "eval_loss": 3.676598310470581,
1080
+ "eval_runtime": 423.4536,
1081
+ "eval_samples_per_second": 50.114,
1082
+ "eval_steps_per_second": 0.392,
1083
+ "step": 1500
1084
+ },
1085
+ {
1086
+ "epoch": 0.06,
1087
+ "grad_norm": 5.529382705688477,
1088
+ "learning_rate": 4.998700626052481e-05,
1089
+ "loss": 4.1739,
1090
+ "step": 1510
1091
+ },
1092
+ {
1093
+ "epoch": 0.06,
1094
+ "grad_norm": 5.5605387687683105,
1095
+ "learning_rate": 4.99864886786401e-05,
1096
+ "loss": 4.1861,
1097
+ "step": 1520
1098
+ },
1099
+ {
1100
+ "epoch": 0.06,
1101
+ "grad_norm": 5.168206214904785,
1102
+ "learning_rate": 4.998596099141231e-05,
1103
+ "loss": 4.0245,
1104
+ "step": 1530
1105
+ },
1106
+ {
1107
+ "epoch": 0.06,
1108
+ "grad_norm": 5.634309768676758,
1109
+ "learning_rate": 4.998542319905486e-05,
1110
+ "loss": 4.2131,
1111
+ "step": 1540
1112
+ },
1113
+ {
1114
+ "epoch": 0.06,
1115
+ "grad_norm": 6.191194534301758,
1116
+ "learning_rate": 4.998487530178525e-05,
1117
+ "loss": 4.163,
1118
+ "step": 1550
1119
+ },
1120
+ {
1121
+ "epoch": 0.06,
1122
+ "grad_norm": 6.52882719039917,
1123
+ "learning_rate": 4.998431729982506e-05,
1124
+ "loss": 4.0668,
1125
+ "step": 1560
1126
+ },
1127
+ {
1128
+ "epoch": 0.06,
1129
+ "grad_norm": 5.614863395690918,
1130
+ "learning_rate": 4.998374919339998e-05,
1131
+ "loss": 4.2501,
1132
+ "step": 1570
1133
+ },
1134
+ {
1135
+ "epoch": 0.06,
1136
+ "grad_norm": 4.7852373123168945,
1137
+ "learning_rate": 4.9983170982739745e-05,
1138
+ "loss": 4.207,
1139
+ "step": 1580
1140
+ },
1141
+ {
1142
+ "epoch": 0.06,
1143
+ "grad_norm": 7.439071178436279,
1144
+ "learning_rate": 4.998258266807822e-05,
1145
+ "loss": 4.0462,
1146
+ "step": 1590
1147
+ },
1148
+ {
1149
+ "epoch": 0.06,
1150
+ "grad_norm": 5.510477542877197,
1151
+ "learning_rate": 4.998198424965333e-05,
1152
+ "loss": 4.277,
1153
+ "step": 1600
1154
+ },
1155
+ {
1156
+ "epoch": 0.06,
1157
+ "grad_norm": 5.421988487243652,
1158
+ "learning_rate": 4.9981375727707114e-05,
1159
+ "loss": 4.0289,
1160
+ "step": 1610
1161
+ },
1162
+ {
1163
+ "epoch": 0.06,
1164
+ "grad_norm": 5.632848739624023,
1165
+ "learning_rate": 4.998075710248565e-05,
1166
+ "loss": 4.2922,
1167
+ "step": 1620
1168
+ },
1169
+ {
1170
+ "epoch": 0.06,
1171
+ "grad_norm": 6.031602382659912,
1172
+ "learning_rate": 4.9980128374239156e-05,
1173
+ "loss": 3.9845,
1174
+ "step": 1630
1175
+ },
1176
+ {
1177
+ "epoch": 0.07,
1178
+ "grad_norm": 5.3867669105529785,
1179
+ "learning_rate": 4.997948954322189e-05,
1180
+ "loss": 4.1095,
1181
+ "step": 1640
1182
+ },
1183
+ {
1184
+ "epoch": 0.07,
1185
+ "grad_norm": 5.150663375854492,
1186
+ "learning_rate": 4.997884060969222e-05,
1187
+ "loss": 3.9247,
1188
+ "step": 1650
1189
+ },
1190
+ {
1191
+ "epoch": 0.07,
1192
+ "grad_norm": 5.717851638793945,
1193
+ "learning_rate": 4.99781815739126e-05,
1194
+ "loss": 4.1911,
1195
+ "step": 1660
1196
+ },
1197
+ {
1198
+ "epoch": 0.07,
1199
+ "grad_norm": 6.766101837158203,
1200
+ "learning_rate": 4.9977512436149563e-05,
1201
+ "loss": 3.9194,
1202
+ "step": 1670
1203
+ },
1204
+ {
1205
+ "epoch": 0.07,
1206
+ "grad_norm": 6.316621780395508,
1207
+ "learning_rate": 4.9976833196673735e-05,
1208
+ "loss": 4.101,
1209
+ "step": 1680
1210
+ },
1211
+ {
1212
+ "epoch": 0.07,
1213
+ "grad_norm": 5.107331275939941,
1214
+ "learning_rate": 4.9976143855759814e-05,
1215
+ "loss": 4.0706,
1216
+ "step": 1690
1217
+ },
1218
+ {
1219
+ "epoch": 0.07,
1220
+ "grad_norm": 5.533346652984619,
1221
+ "learning_rate": 4.99754444136866e-05,
1222
+ "loss": 4.0164,
1223
+ "step": 1700
1224
+ },
1225
+ {
1226
+ "epoch": 0.07,
1227
+ "grad_norm": 5.967229843139648,
1228
+ "learning_rate": 4.9974734870736975e-05,
1229
+ "loss": 4.1229,
1230
+ "step": 1710
1231
+ },
1232
+ {
1233
+ "epoch": 0.07,
1234
+ "grad_norm": 5.502392768859863,
1235
+ "learning_rate": 4.997401522719788e-05,
1236
+ "loss": 4.0456,
1237
+ "step": 1720
1238
+ },
1239
+ {
1240
+ "epoch": 0.07,
1241
+ "grad_norm": 5.6358962059021,
1242
+ "learning_rate": 4.997328548336038e-05,
1243
+ "loss": 3.9432,
1244
+ "step": 1730
1245
+ },
1246
+ {
1247
+ "epoch": 0.07,
1248
+ "grad_norm": 6.525423049926758,
1249
+ "learning_rate": 4.99725456395196e-05,
1250
+ "loss": 4.05,
1251
+ "step": 1740
1252
+ },
1253
+ {
1254
+ "epoch": 0.07,
1255
+ "grad_norm": 4.675175666809082,
1256
+ "learning_rate": 4.997179569597475e-05,
1257
+ "loss": 3.8145,
1258
+ "step": 1750
1259
+ },
1260
+ {
1261
+ "epoch": 0.07,
1262
+ "grad_norm": 5.409067630767822,
1263
+ "learning_rate": 4.9971035653029153e-05,
1264
+ "loss": 3.914,
1265
+ "step": 1760
1266
+ },
1267
+ {
1268
+ "epoch": 0.07,
1269
+ "grad_norm": 5.812373161315918,
1270
+ "learning_rate": 4.997026551099017e-05,
1271
+ "loss": 3.9817,
1272
+ "step": 1770
1273
+ },
1274
+ {
1275
+ "epoch": 0.07,
1276
+ "grad_norm": 5.165591239929199,
1277
+ "learning_rate": 4.996948527016929e-05,
1278
+ "loss": 3.9414,
1279
+ "step": 1780
1280
+ },
1281
+ {
1282
+ "epoch": 0.07,
1283
+ "grad_norm": 5.0605034828186035,
1284
+ "learning_rate": 4.996869493088205e-05,
1285
+ "loss": 3.8139,
1286
+ "step": 1790
1287
+ },
1288
+ {
1289
+ "epoch": 0.07,
1290
+ "grad_norm": 6.348527908325195,
1291
+ "learning_rate": 4.996789449344812e-05,
1292
+ "loss": 3.7905,
1293
+ "step": 1800
1294
+ },
1295
+ {
1296
+ "epoch": 0.07,
1297
+ "grad_norm": 5.135634422302246,
1298
+ "learning_rate": 4.9967083958191187e-05,
1299
+ "loss": 3.7834,
1300
+ "step": 1810
1301
+ },
1302
+ {
1303
+ "epoch": 0.07,
1304
+ "grad_norm": 8.48004150390625,
1305
+ "learning_rate": 4.996626332543907e-05,
1306
+ "loss": 3.7551,
1307
+ "step": 1820
1308
+ },
1309
+ {
1310
+ "epoch": 0.07,
1311
+ "grad_norm": 5.518482208251953,
1312
+ "learning_rate": 4.996543259552367e-05,
1313
+ "loss": 3.8619,
1314
+ "step": 1830
1315
+ },
1316
+ {
1317
+ "epoch": 0.07,
1318
+ "grad_norm": 5.7253737449646,
1319
+ "learning_rate": 4.996459176878095e-05,
1320
+ "loss": 3.7978,
1321
+ "step": 1840
1322
+ },
1323
+ {
1324
+ "epoch": 0.07,
1325
+ "grad_norm": 6.030267715454102,
1326
+ "learning_rate": 4.9963740845550965e-05,
1327
+ "loss": 3.9309,
1328
+ "step": 1850
1329
+ },
1330
+ {
1331
+ "epoch": 0.07,
1332
+ "grad_norm": 5.561445713043213,
1333
+ "learning_rate": 4.996287982617787e-05,
1334
+ "loss": 3.9761,
1335
+ "step": 1860
1336
+ },
1337
+ {
1338
+ "epoch": 0.07,
1339
+ "grad_norm": 5.555076599121094,
1340
+ "learning_rate": 4.996200871100988e-05,
1341
+ "loss": 3.8587,
1342
+ "step": 1870
1343
+ },
1344
+ {
1345
+ "epoch": 0.07,
1346
+ "grad_norm": 5.502940654754639,
1347
+ "learning_rate": 4.9961127500399295e-05,
1348
+ "loss": 3.847,
1349
+ "step": 1880
1350
+ },
1351
+ {
1352
+ "epoch": 0.07,
1353
+ "grad_norm": 5.557964324951172,
1354
+ "learning_rate": 4.996023619470252e-05,
1355
+ "loss": 3.6117,
1356
+ "step": 1890
1357
+ },
1358
+ {
1359
+ "epoch": 0.08,
1360
+ "grad_norm": 6.37682580947876,
1361
+ "learning_rate": 4.9959334794280014e-05,
1362
+ "loss": 3.7581,
1363
+ "step": 1900
1364
+ },
1365
+ {
1366
+ "epoch": 0.08,
1367
+ "grad_norm": 6.152722358703613,
1368
+ "learning_rate": 4.9958423299496344e-05,
1369
+ "loss": 3.8454,
1370
+ "step": 1910
1371
+ },
1372
+ {
1373
+ "epoch": 0.08,
1374
+ "grad_norm": 5.4067301750183105,
1375
+ "learning_rate": 4.995750171072014e-05,
1376
+ "loss": 3.9353,
1377
+ "step": 1920
1378
+ },
1379
+ {
1380
+ "epoch": 0.08,
1381
+ "grad_norm": 5.564638614654541,
1382
+ "learning_rate": 4.9956570028324124e-05,
1383
+ "loss": 3.738,
1384
+ "step": 1930
1385
+ },
1386
+ {
1387
+ "epoch": 0.08,
1388
+ "grad_norm": 5.320713520050049,
1389
+ "learning_rate": 4.99556282526851e-05,
1390
+ "loss": 3.5365,
1391
+ "step": 1940
1392
+ },
1393
+ {
1394
+ "epoch": 0.08,
1395
+ "grad_norm": 6.384575843811035,
1396
+ "learning_rate": 4.995467638418395e-05,
1397
+ "loss": 3.8153,
1398
+ "step": 1950
1399
+ },
1400
+ {
1401
+ "epoch": 0.08,
1402
+ "grad_norm": 6.424959659576416,
1403
+ "learning_rate": 4.995371442320565e-05,
1404
+ "loss": 3.779,
1405
+ "step": 1960
1406
+ },
1407
+ {
1408
+ "epoch": 0.08,
1409
+ "grad_norm": 5.928106784820557,
1410
+ "learning_rate": 4.9952742370139224e-05,
1411
+ "loss": 3.7395,
1412
+ "step": 1970
1413
+ },
1414
+ {
1415
+ "epoch": 0.08,
1416
+ "grad_norm": 6.09404182434082,
1417
+ "learning_rate": 4.995176022537784e-05,
1418
+ "loss": 3.757,
1419
+ "step": 1980
1420
+ },
1421
+ {
1422
+ "epoch": 0.08,
1423
+ "grad_norm": 8.332275390625,
1424
+ "learning_rate": 4.995076798931867e-05,
1425
+ "loss": 3.8866,
1426
+ "step": 1990
1427
+ },
1428
+ {
1429
+ "epoch": 0.08,
1430
+ "grad_norm": 6.44631814956665,
1431
+ "learning_rate": 4.994976566236303e-05,
1432
+ "loss": 3.8395,
1433
+ "step": 2000
1434
+ },
1435
+ {
1436
+ "epoch": 0.08,
1437
+ "eval_loss": 3.2655727863311768,
1438
+ "eval_runtime": 423.1622,
1439
+ "eval_samples_per_second": 50.149,
1440
+ "eval_steps_per_second": 0.392,
1441
+ "step": 2000
1442
+ },
1443
+ {
1444
+ "epoch": 0.08,
1445
+ "grad_norm": 5.428518295288086,
1446
+ "learning_rate": 4.994875324491629e-05,
1447
+ "loss": 3.7103,
1448
+ "step": 2010
1449
+ },
1450
+ {
1451
+ "epoch": 0.08,
1452
+ "grad_norm": 6.11751651763916,
1453
+ "learning_rate": 4.9947730737387885e-05,
1454
+ "loss": 3.8224,
1455
+ "step": 2020
1456
+ },
1457
+ {
1458
+ "epoch": 0.08,
1459
+ "grad_norm": 5.404431343078613,
1460
+ "learning_rate": 4.994669814019138e-05,
1461
+ "loss": 3.9208,
1462
+ "step": 2030
1463
+ },
1464
+ {
1465
+ "epoch": 0.08,
1466
+ "grad_norm": 5.487029075622559,
1467
+ "learning_rate": 4.994565545374438e-05,
1468
+ "loss": 3.7412,
1469
+ "step": 2040
1470
+ },
1471
+ {
1472
+ "epoch": 0.08,
1473
+ "grad_norm": 4.979548454284668,
1474
+ "learning_rate": 4.9944602678468575e-05,
1475
+ "loss": 3.6008,
1476
+ "step": 2050
1477
+ },
1478
+ {
1479
+ "epoch": 0.08,
1480
+ "grad_norm": 5.379942417144775,
1481
+ "learning_rate": 4.994353981478974e-05,
1482
+ "loss": 3.6077,
1483
+ "step": 2060
1484
+ },
1485
+ {
1486
+ "epoch": 0.08,
1487
+ "grad_norm": 5.666346549987793,
1488
+ "learning_rate": 4.994246686313774e-05,
1489
+ "loss": 3.8157,
1490
+ "step": 2070
1491
+ },
1492
+ {
1493
+ "epoch": 0.08,
1494
+ "grad_norm": 5.252011299133301,
1495
+ "learning_rate": 4.99413838239465e-05,
1496
+ "loss": 3.6929,
1497
+ "step": 2080
1498
+ },
1499
+ {
1500
+ "epoch": 0.08,
1501
+ "grad_norm": 6.579212188720703,
1502
+ "learning_rate": 4.9940290697654045e-05,
1503
+ "loss": 3.9075,
1504
+ "step": 2090
1505
+ },
1506
+ {
1507
+ "epoch": 0.08,
1508
+ "grad_norm": 5.747458457946777,
1509
+ "learning_rate": 4.993918748470246e-05,
1510
+ "loss": 3.5528,
1511
+ "step": 2100
1512
+ },
1513
+ {
1514
+ "epoch": 0.08,
1515
+ "grad_norm": 5.356452465057373,
1516
+ "learning_rate": 4.993818596932118e-05,
1517
+ "loss": 3.4987,
1518
+ "step": 2110
1519
+ },
1520
+ {
1521
+ "epoch": 0.08,
1522
+ "grad_norm": 5.761812686920166,
1523
+ "learning_rate": 4.993706359294986e-05,
1524
+ "loss": 3.7627,
1525
+ "step": 2120
1526
+ },
1527
+ {
1528
+ "epoch": 0.08,
1529
+ "grad_norm": 5.732122898101807,
1530
+ "learning_rate": 4.993593113122456e-05,
1531
+ "loss": 3.6106,
1532
+ "step": 2130
1533
+ },
1534
+ {
1535
+ "epoch": 0.08,
1536
+ "grad_norm": 11.450267791748047,
1537
+ "learning_rate": 4.993478858460329e-05,
1538
+ "loss": 3.4716,
1539
+ "step": 2140
1540
+ },
1541
+ {
1542
+ "epoch": 0.09,
1543
+ "grad_norm": 5.774083614349365,
1544
+ "learning_rate": 4.993363595354813e-05,
1545
+ "loss": 3.5741,
1546
+ "step": 2150
1547
+ },
1548
+ {
1549
+ "epoch": 0.09,
1550
+ "grad_norm": 5.652318000793457,
1551
+ "learning_rate": 4.9932473238525247e-05,
1552
+ "loss": 3.676,
1553
+ "step": 2160
1554
+ },
1555
+ {
1556
+ "epoch": 0.09,
1557
+ "grad_norm": 5.903106689453125,
1558
+ "learning_rate": 4.993130044000487e-05,
1559
+ "loss": 3.6334,
1560
+ "step": 2170
1561
+ },
1562
+ {
1563
+ "epoch": 0.09,
1564
+ "grad_norm": 5.358557224273682,
1565
+ "learning_rate": 4.993011755846132e-05,
1566
+ "loss": 3.5592,
1567
+ "step": 2180
1568
+ },
1569
+ {
1570
+ "epoch": 0.09,
1571
+ "grad_norm": 6.509594440460205,
1572
+ "learning_rate": 4.9928924594372995e-05,
1573
+ "loss": 3.6512,
1574
+ "step": 2190
1575
+ },
1576
+ {
1577
+ "epoch": 0.09,
1578
+ "grad_norm": 6.318192958831787,
1579
+ "learning_rate": 4.9927721548222374e-05,
1580
+ "loss": 3.3691,
1581
+ "step": 2200
1582
+ },
1583
+ {
1584
+ "epoch": 0.09,
1585
+ "grad_norm": 5.848221778869629,
1586
+ "learning_rate": 4.9926508420496e-05,
1587
+ "loss": 3.3721,
1588
+ "step": 2210
1589
+ },
1590
+ {
1591
+ "epoch": 0.09,
1592
+ "grad_norm": 5.9782490730285645,
1593
+ "learning_rate": 4.992528521168449e-05,
1594
+ "loss": 3.5859,
1595
+ "step": 2220
1596
+ },
1597
+ {
1598
+ "epoch": 0.09,
1599
+ "grad_norm": 6.406031608581543,
1600
+ "learning_rate": 4.992405192228258e-05,
1601
+ "loss": 3.6206,
1602
+ "step": 2230
1603
+ },
1604
+ {
1605
+ "epoch": 0.09,
1606
+ "grad_norm": 6.102032661437988,
1607
+ "learning_rate": 4.992280855278903e-05,
1608
+ "loss": 3.4615,
1609
+ "step": 2240
1610
+ },
1611
+ {
1612
+ "epoch": 0.09,
1613
+ "grad_norm": 5.5005035400390625,
1614
+ "learning_rate": 4.992155510370669e-05,
1615
+ "loss": 3.507,
1616
+ "step": 2250
1617
+ },
1618
+ {
1619
+ "epoch": 0.09,
1620
+ "grad_norm": 8.336762428283691,
1621
+ "learning_rate": 4.9920291575542515e-05,
1622
+ "loss": 3.3897,
1623
+ "step": 2260
1624
+ },
1625
+ {
1626
+ "epoch": 0.09,
1627
+ "grad_norm": 7.851747989654541,
1628
+ "learning_rate": 4.9919017968807514e-05,
1629
+ "loss": 3.5232,
1630
+ "step": 2270
1631
+ },
1632
+ {
1633
+ "epoch": 0.09,
1634
+ "grad_norm": 5.816250324249268,
1635
+ "learning_rate": 4.991773428401676e-05,
1636
+ "loss": 3.5638,
1637
+ "step": 2280
1638
+ },
1639
+ {
1640
+ "epoch": 0.09,
1641
+ "grad_norm": 4.746279716491699,
1642
+ "learning_rate": 4.9916440521689433e-05,
1643
+ "loss": 3.5446,
1644
+ "step": 2290
1645
+ },
1646
+ {
1647
+ "epoch": 0.09,
1648
+ "grad_norm": 5.8899922370910645,
1649
+ "learning_rate": 4.9915136682348774e-05,
1650
+ "loss": 3.3533,
1651
+ "step": 2300
1652
+ },
1653
+ {
1654
+ "epoch": 0.09,
1655
+ "grad_norm": 6.160922050476074,
1656
+ "learning_rate": 4.9913822766522076e-05,
1657
+ "loss": 3.4222,
1658
+ "step": 2310
1659
+ },
1660
+ {
1661
+ "epoch": 0.09,
1662
+ "grad_norm": 6.936460494995117,
1663
+ "learning_rate": 4.991249877474075e-05,
1664
+ "loss": 3.5763,
1665
+ "step": 2320
1666
+ },
1667
+ {
1668
+ "epoch": 0.09,
1669
+ "grad_norm": 16.517559051513672,
1670
+ "learning_rate": 4.991116470754025e-05,
1671
+ "loss": 3.5324,
1672
+ "step": 2330
1673
+ },
1674
+ {
1675
+ "epoch": 0.09,
1676
+ "grad_norm": 4.840503692626953,
1677
+ "learning_rate": 4.9909820565460116e-05,
1678
+ "loss": 3.5764,
1679
+ "step": 2340
1680
+ },
1681
+ {
1682
+ "epoch": 0.09,
1683
+ "grad_norm": 5.952908515930176,
1684
+ "learning_rate": 4.990846634904398e-05,
1685
+ "loss": 3.4369,
1686
+ "step": 2350
1687
+ },
1688
+ {
1689
+ "epoch": 0.09,
1690
+ "grad_norm": 5.108431339263916,
1691
+ "learning_rate": 4.9907102058839504e-05,
1692
+ "loss": 3.4518,
1693
+ "step": 2360
1694
+ },
1695
+ {
1696
+ "epoch": 0.09,
1697
+ "grad_norm": 5.068904399871826,
1698
+ "learning_rate": 4.990572769539848e-05,
1699
+ "loss": 3.5594,
1700
+ "step": 2370
1701
+ },
1702
+ {
1703
+ "epoch": 0.09,
1704
+ "grad_norm": 5.512117385864258,
1705
+ "learning_rate": 4.990434325927671e-05,
1706
+ "loss": 3.6366,
1707
+ "step": 2380
1708
+ },
1709
+ {
1710
+ "epoch": 0.09,
1711
+ "grad_norm": 5.916763782501221,
1712
+ "learning_rate": 4.990294875103413e-05,
1713
+ "loss": 3.4344,
1714
+ "step": 2390
1715
+ },
1716
+ {
1717
+ "epoch": 0.1,
1718
+ "grad_norm": 6.244757175445557,
1719
+ "learning_rate": 4.990154417123472e-05,
1720
+ "loss": 3.5858,
1721
+ "step": 2400
1722
+ },
1723
+ {
1724
+ "epoch": 0.1,
1725
+ "grad_norm": 6.071867942810059,
1726
+ "learning_rate": 4.990012952044654e-05,
1727
+ "loss": 3.5888,
1728
+ "step": 2410
1729
+ },
1730
+ {
1731
+ "epoch": 0.1,
1732
+ "grad_norm": 5.984373092651367,
1733
+ "learning_rate": 4.98987047992417e-05,
1734
+ "loss": 3.5084,
1735
+ "step": 2420
1736
+ },
1737
+ {
1738
+ "epoch": 0.1,
1739
+ "grad_norm": 7.062913417816162,
1740
+ "learning_rate": 4.989727000819644e-05,
1741
+ "loss": 3.4169,
1742
+ "step": 2430
1743
+ },
1744
+ {
1745
+ "epoch": 0.1,
1746
+ "grad_norm": 4.671230792999268,
1747
+ "learning_rate": 4.9895825147891e-05,
1748
+ "loss": 3.3354,
1749
+ "step": 2440
1750
+ },
1751
+ {
1752
+ "epoch": 0.1,
1753
+ "grad_norm": 5.235222339630127,
1754
+ "learning_rate": 4.989437021890975e-05,
1755
+ "loss": 3.4756,
1756
+ "step": 2450
1757
+ },
1758
+ {
1759
+ "epoch": 0.1,
1760
+ "grad_norm": 5.888786792755127,
1761
+ "learning_rate": 4.98929052218411e-05,
1762
+ "loss": 3.3409,
1763
+ "step": 2460
1764
+ },
1765
+ {
1766
+ "epoch": 0.1,
1767
+ "grad_norm": 6.014183044433594,
1768
+ "learning_rate": 4.989143015727754e-05,
1769
+ "loss": 3.4246,
1770
+ "step": 2470
1771
+ },
1772
+ {
1773
+ "epoch": 0.1,
1774
+ "grad_norm": 8.75373649597168,
1775
+ "learning_rate": 4.988994502581565e-05,
1776
+ "loss": 3.2431,
1777
+ "step": 2480
1778
+ },
1779
+ {
1780
+ "epoch": 0.1,
1781
+ "grad_norm": 5.800814628601074,
1782
+ "learning_rate": 4.988844982805605e-05,
1783
+ "loss": 3.4702,
1784
+ "step": 2490
1785
+ },
1786
+ {
1787
+ "epoch": 0.1,
1788
+ "grad_norm": 5.22312068939209,
1789
+ "learning_rate": 4.9886944564603444e-05,
1790
+ "loss": 3.2994,
1791
+ "step": 2500
1792
+ },
1793
+ {
1794
+ "epoch": 0.1,
1795
+ "eval_loss": 2.9097418785095215,
1796
+ "eval_runtime": 423.0547,
1797
+ "eval_samples_per_second": 50.161,
1798
+ "eval_steps_per_second": 0.392,
1799
+ "step": 2500
1800
+ },
1801
+ {
1802
+ "epoch": 0.1,
1803
+ "grad_norm": 5.41413688659668,
1804
+ "learning_rate": 4.988542923606663e-05,
1805
+ "loss": 3.5138,
1806
+ "step": 2510
1807
+ },
1808
+ {
1809
+ "epoch": 0.1,
1810
+ "grad_norm": 5.5836968421936035,
1811
+ "learning_rate": 4.988390384305843e-05,
1812
+ "loss": 3.4161,
1813
+ "step": 2520
1814
+ },
1815
+ {
1816
+ "epoch": 0.1,
1817
+ "grad_norm": 5.957060813903809,
1818
+ "learning_rate": 4.9882368386195776e-05,
1819
+ "loss": 3.4227,
1820
+ "step": 2530
1821
+ },
1822
+ {
1823
+ "epoch": 0.1,
1824
+ "grad_norm": 5.718446731567383,
1825
+ "learning_rate": 4.988082286609965e-05,
1826
+ "loss": 3.3632,
1827
+ "step": 2540
1828
+ },
1829
+ {
1830
+ "epoch": 0.1,
1831
+ "grad_norm": 6.273780822753906,
1832
+ "learning_rate": 4.9879267283395125e-05,
1833
+ "loss": 3.4018,
1834
+ "step": 2550
1835
+ },
1836
+ {
1837
+ "epoch": 0.1,
1838
+ "grad_norm": 5.000144004821777,
1839
+ "learning_rate": 4.9877701638711314e-05,
1840
+ "loss": 3.2865,
1841
+ "step": 2560
1842
+ },
1843
+ {
1844
+ "epoch": 0.1,
1845
+ "grad_norm": 6.369386672973633,
1846
+ "learning_rate": 4.987612593268142e-05,
1847
+ "loss": 3.2916,
1848
+ "step": 2570
1849
+ },
1850
+ {
1851
+ "epoch": 0.1,
1852
+ "grad_norm": 5.275691032409668,
1853
+ "learning_rate": 4.98745401659427e-05,
1854
+ "loss": 3.3678,
1855
+ "step": 2580
1856
+ },
1857
+ {
1858
+ "epoch": 0.1,
1859
+ "grad_norm": 6.77783727645874,
1860
+ "learning_rate": 4.9872944339136503e-05,
1861
+ "loss": 3.2567,
1862
+ "step": 2590
1863
+ },
1864
+ {
1865
+ "epoch": 0.1,
1866
+ "grad_norm": 6.3765435218811035,
1867
+ "learning_rate": 4.987133845290822e-05,
1868
+ "loss": 3.3649,
1869
+ "step": 2600
1870
+ },
1871
+ {
1872
+ "epoch": 0.1,
1873
+ "grad_norm": 6.956243991851807,
1874
+ "learning_rate": 4.9869722507907343e-05,
1875
+ "loss": 3.428,
1876
+ "step": 2610
1877
+ },
1878
+ {
1879
+ "epoch": 0.1,
1880
+ "grad_norm": 4.933529376983643,
1881
+ "learning_rate": 4.9868096504787395e-05,
1882
+ "loss": 3.0448,
1883
+ "step": 2620
1884
+ },
1885
+ {
1886
+ "epoch": 0.1,
1887
+ "grad_norm": 6.319052696228027,
1888
+ "learning_rate": 4.9866460444205983e-05,
1889
+ "loss": 3.3518,
1890
+ "step": 2630
1891
+ },
1892
+ {
1893
+ "epoch": 0.1,
1894
+ "grad_norm": 5.592201232910156,
1895
+ "learning_rate": 4.986481432682479e-05,
1896
+ "loss": 3.1236,
1897
+ "step": 2640
1898
+ },
1899
+ {
1900
+ "epoch": 0.11,
1901
+ "grad_norm": 5.243892192840576,
1902
+ "learning_rate": 4.986315815330956e-05,
1903
+ "loss": 3.2447,
1904
+ "step": 2650
1905
+ },
1906
+ {
1907
+ "epoch": 0.11,
1908
+ "grad_norm": 6.197099208831787,
1909
+ "learning_rate": 4.98614919243301e-05,
1910
+ "loss": 3.3031,
1911
+ "step": 2660
1912
+ },
1913
+ {
1914
+ "epoch": 0.11,
1915
+ "grad_norm": 6.438327312469482,
1916
+ "learning_rate": 4.98598156405603e-05,
1917
+ "loss": 3.5433,
1918
+ "step": 2670
1919
+ },
1920
+ {
1921
+ "epoch": 0.11,
1922
+ "grad_norm": 5.134957313537598,
1923
+ "learning_rate": 4.985812930267808e-05,
1924
+ "loss": 3.3124,
1925
+ "step": 2680
1926
+ },
1927
+ {
1928
+ "epoch": 0.11,
1929
+ "grad_norm": 5.727309703826904,
1930
+ "learning_rate": 4.985643291136546e-05,
1931
+ "loss": 3.2169,
1932
+ "step": 2690
1933
+ },
1934
+ {
1935
+ "epoch": 0.11,
1936
+ "grad_norm": 5.680230617523193,
1937
+ "learning_rate": 4.985472646730852e-05,
1938
+ "loss": 3.0445,
1939
+ "step": 2700
1940
+ },
1941
+ {
1942
+ "epoch": 0.11,
1943
+ "grad_norm": 5.9644598960876465,
1944
+ "learning_rate": 4.985300997119739e-05,
1945
+ "loss": 3.1588,
1946
+ "step": 2710
1947
+ },
1948
+ {
1949
+ "epoch": 0.11,
1950
+ "grad_norm": 5.430914402008057,
1951
+ "learning_rate": 4.985128342372629e-05,
1952
+ "loss": 3.2385,
1953
+ "step": 2720
1954
+ },
1955
+ {
1956
+ "epoch": 0.11,
1957
+ "grad_norm": 5.565753936767578,
1958
+ "learning_rate": 4.9849546825593476e-05,
1959
+ "loss": 3.2654,
1960
+ "step": 2730
1961
+ },
1962
+ {
1963
+ "epoch": 0.11,
1964
+ "grad_norm": 6.126991271972656,
1965
+ "learning_rate": 4.9847800177501305e-05,
1966
+ "loss": 3.3024,
1967
+ "step": 2740
1968
+ },
1969
+ {
1970
+ "epoch": 0.11,
1971
+ "grad_norm": 5.406001567840576,
1972
+ "learning_rate": 4.984604348015616e-05,
1973
+ "loss": 3.219,
1974
+ "step": 2750
1975
+ },
1976
+ {
1977
+ "epoch": 0.11,
1978
+ "grad_norm": 5.55892276763916,
1979
+ "learning_rate": 4.984427673426851e-05,
1980
+ "loss": 3.3882,
1981
+ "step": 2760
1982
+ },
1983
+ {
1984
+ "epoch": 0.11,
1985
+ "grad_norm": 5.547699451446533,
1986
+ "learning_rate": 4.9842499940552886e-05,
1987
+ "loss": 3.2334,
1988
+ "step": 2770
1989
+ },
1990
+ {
1991
+ "epoch": 0.11,
1992
+ "grad_norm": 6.514155864715576,
1993
+ "learning_rate": 4.9840713099727885e-05,
1994
+ "loss": 3.2041,
1995
+ "step": 2780
1996
+ },
1997
+ {
1998
+ "epoch": 0.11,
1999
+ "grad_norm": 5.4210991859436035,
2000
+ "learning_rate": 4.983891621251616e-05,
2001
+ "loss": 3.1746,
2002
+ "step": 2790
2003
+ },
2004
+ {
2005
+ "epoch": 0.11,
2006
+ "grad_norm": 5.854736328125,
2007
+ "learning_rate": 4.983710927964442e-05,
2008
+ "loss": 3.1961,
2009
+ "step": 2800
2010
+ },
2011
+ {
2012
+ "epoch": 0.11,
2013
+ "grad_norm": 5.277487754821777,
2014
+ "learning_rate": 4.9835292301843476e-05,
2015
+ "loss": 2.9688,
2016
+ "step": 2810
2017
+ },
2018
+ {
2019
+ "epoch": 0.11,
2020
+ "grad_norm": 5.925424098968506,
2021
+ "learning_rate": 4.983346527984813e-05,
2022
+ "loss": 3.2315,
2023
+ "step": 2820
2024
+ },
2025
+ {
2026
+ "epoch": 0.11,
2027
+ "grad_norm": 7.818734169006348,
2028
+ "learning_rate": 4.983162821439733e-05,
2029
+ "loss": 3.325,
2030
+ "step": 2830
2031
+ },
2032
+ {
2033
+ "epoch": 0.11,
2034
+ "grad_norm": 6.312294006347656,
2035
+ "learning_rate": 4.982978110623402e-05,
2036
+ "loss": 3.2141,
2037
+ "step": 2840
2038
+ },
2039
+ {
2040
+ "epoch": 0.11,
2041
+ "grad_norm": 6.804452419281006,
2042
+ "learning_rate": 4.982792395610524e-05,
2043
+ "loss": 3.1214,
2044
+ "step": 2850
2045
+ },
2046
+ {
2047
+ "epoch": 0.11,
2048
+ "grad_norm": 6.256209850311279,
2049
+ "learning_rate": 4.9826056764762074e-05,
2050
+ "loss": 3.2357,
2051
+ "step": 2860
2052
+ },
2053
+ {
2054
+ "epoch": 0.11,
2055
+ "grad_norm": 5.713292121887207,
2056
+ "learning_rate": 4.982417953295968e-05,
2057
+ "loss": 3.0448,
2058
+ "step": 2870
2059
+ },
2060
+ {
2061
+ "epoch": 0.11,
2062
+ "grad_norm": 5.762032985687256,
2063
+ "learning_rate": 4.982229226145727e-05,
2064
+ "loss": 3.1855,
2065
+ "step": 2880
2066
+ },
2067
+ {
2068
+ "epoch": 0.11,
2069
+ "grad_norm": 5.245543956756592,
2070
+ "learning_rate": 4.982039495101812e-05,
2071
+ "loss": 3.1024,
2072
+ "step": 2890
2073
+ },
2074
+ {
2075
+ "epoch": 0.12,
2076
+ "grad_norm": 5.456753730773926,
2077
+ "learning_rate": 4.9818487602409556e-05,
2078
+ "loss": 3.2703,
2079
+ "step": 2900
2080
+ },
2081
+ {
2082
+ "epoch": 0.12,
2083
+ "grad_norm": 5.831927299499512,
2084
+ "learning_rate": 4.981657021640298e-05,
2085
+ "loss": 3.2172,
2086
+ "step": 2910
2087
+ },
2088
+ {
2089
+ "epoch": 0.12,
2090
+ "grad_norm": 6.4401350021362305,
2091
+ "learning_rate": 4.981464279377384e-05,
2092
+ "loss": 3.2972,
2093
+ "step": 2920
2094
+ },
2095
+ {
2096
+ "epoch": 0.12,
2097
+ "grad_norm": 5.1732892990112305,
2098
+ "learning_rate": 4.981270533530164e-05,
2099
+ "loss": 3.1305,
2100
+ "step": 2930
2101
+ },
2102
+ {
2103
+ "epoch": 0.12,
2104
+ "grad_norm": 5.929070472717285,
2105
+ "learning_rate": 4.9810757841769976e-05,
2106
+ "loss": 3.0932,
2107
+ "step": 2940
2108
+ },
2109
+ {
2110
+ "epoch": 0.12,
2111
+ "grad_norm": 5.50985860824585,
2112
+ "learning_rate": 4.980880031396645e-05,
2113
+ "loss": 3.1417,
2114
+ "step": 2950
2115
+ },
2116
+ {
2117
+ "epoch": 0.12,
2118
+ "grad_norm": 6.071355819702148,
2119
+ "learning_rate": 4.980683275268277e-05,
2120
+ "loss": 3.1796,
2121
+ "step": 2960
2122
+ },
2123
+ {
2124
+ "epoch": 0.12,
2125
+ "grad_norm": 5.182791709899902,
2126
+ "learning_rate": 4.980485515871467e-05,
2127
+ "loss": 3.268,
2128
+ "step": 2970
2129
+ },
2130
+ {
2131
+ "epoch": 0.12,
2132
+ "grad_norm": 6.4410786628723145,
2133
+ "learning_rate": 4.980286753286195e-05,
2134
+ "loss": 3.2922,
2135
+ "step": 2980
2136
+ },
2137
+ {
2138
+ "epoch": 0.12,
2139
+ "grad_norm": 22.151193618774414,
2140
+ "learning_rate": 4.9800869875928475e-05,
2141
+ "loss": 3.0482,
2142
+ "step": 2990
2143
+ },
2144
+ {
2145
+ "epoch": 0.12,
2146
+ "grad_norm": 6.808634281158447,
2147
+ "learning_rate": 4.9798862188722164e-05,
2148
+ "loss": 3.1515,
2149
+ "step": 3000
2150
+ },
2151
+ {
2152
+ "epoch": 0.12,
2153
+ "eval_loss": 2.6660187244415283,
2154
+ "eval_runtime": 423.2518,
2155
+ "eval_samples_per_second": 50.138,
2156
+ "eval_steps_per_second": 0.392,
2157
+ "step": 3000
2158
+ },
2159
+ {
2160
+ "epoch": 0.12,
2161
+ "grad_norm": 9.806035995483398,
2162
+ "learning_rate": 4.979684447205499e-05,
2163
+ "loss": 3.0872,
2164
+ "step": 3010
2165
+ },
2166
+ {
2167
+ "epoch": 0.12,
2168
+ "grad_norm": 5.456057071685791,
2169
+ "learning_rate": 4.9794816726742976e-05,
2170
+ "loss": 2.882,
2171
+ "step": 3020
2172
+ },
2173
+ {
2174
+ "epoch": 0.12,
2175
+ "grad_norm": 6.348060607910156,
2176
+ "learning_rate": 4.979277895360622e-05,
2177
+ "loss": 3.1984,
2178
+ "step": 3030
2179
+ },
2180
+ {
2181
+ "epoch": 0.12,
2182
+ "grad_norm": 5.867720603942871,
2183
+ "learning_rate": 4.979073115346885e-05,
2184
+ "loss": 3.0328,
2185
+ "step": 3040
2186
+ },
2187
+ {
2188
+ "epoch": 0.12,
2189
+ "grad_norm": 5.720703601837158,
2190
+ "learning_rate": 4.9788673327159076e-05,
2191
+ "loss": 3.1852,
2192
+ "step": 3050
2193
+ },
2194
+ {
2195
+ "epoch": 0.12,
2196
+ "grad_norm": 6.027235984802246,
2197
+ "learning_rate": 4.978660547550914e-05,
2198
+ "loss": 3.1082,
2199
+ "step": 3060
2200
+ },
2201
+ {
2202
+ "epoch": 0.12,
2203
+ "grad_norm": 5.131529808044434,
2204
+ "learning_rate": 4.978452759935535e-05,
2205
+ "loss": 3.0318,
2206
+ "step": 3070
2207
+ },
2208
+ {
2209
+ "epoch": 0.12,
2210
+ "grad_norm": 5.773056983947754,
2211
+ "learning_rate": 4.978243969953806e-05,
2212
+ "loss": 3.3087,
2213
+ "step": 3080
2214
+ },
2215
+ {
2216
+ "epoch": 0.12,
2217
+ "grad_norm": 5.788400650024414,
2218
+ "learning_rate": 4.9780341776901696e-05,
2219
+ "loss": 2.9589,
2220
+ "step": 3090
2221
+ },
2222
+ {
2223
+ "epoch": 0.12,
2224
+ "grad_norm": 6.815254211425781,
2225
+ "learning_rate": 4.9778233832294724e-05,
2226
+ "loss": 3.1083,
2227
+ "step": 3100
2228
+ },
2229
+ {
2230
+ "epoch": 0.12,
2231
+ "grad_norm": 5.483798027038574,
2232
+ "learning_rate": 4.9776115866569654e-05,
2233
+ "loss": 3.1023,
2234
+ "step": 3110
2235
+ },
2236
+ {
2237
+ "epoch": 0.12,
2238
+ "grad_norm": 6.394237995147705,
2239
+ "learning_rate": 4.9773987880583075e-05,
2240
+ "loss": 3.0322,
2241
+ "step": 3120
2242
+ },
2243
+ {
2244
+ "epoch": 0.12,
2245
+ "grad_norm": 4.961243152618408,
2246
+ "learning_rate": 4.97718498751956e-05,
2247
+ "loss": 3.0204,
2248
+ "step": 3130
2249
+ },
2250
+ {
2251
+ "epoch": 0.12,
2252
+ "grad_norm": 5.827451229095459,
2253
+ "learning_rate": 4.976970185127191e-05,
2254
+ "loss": 3.1435,
2255
+ "step": 3140
2256
+ },
2257
+ {
2258
+ "epoch": 0.12,
2259
+ "grad_norm": 5.503066539764404,
2260
+ "learning_rate": 4.9767543809680744e-05,
2261
+ "loss": 3.0807,
2262
+ "step": 3150
2263
+ },
2264
+ {
2265
+ "epoch": 0.13,
2266
+ "grad_norm": 4.9387288093566895,
2267
+ "learning_rate": 4.9765375751294874e-05,
2268
+ "loss": 2.7987,
2269
+ "step": 3160
2270
+ },
2271
+ {
2272
+ "epoch": 0.13,
2273
+ "grad_norm": 6.066097259521484,
2274
+ "learning_rate": 4.9763197676991134e-05,
2275
+ "loss": 3.0099,
2276
+ "step": 3170
2277
+ },
2278
+ {
2279
+ "epoch": 0.13,
2280
+ "grad_norm": 7.227752685546875,
2281
+ "learning_rate": 4.976100958765042e-05,
2282
+ "loss": 2.9368,
2283
+ "step": 3180
2284
+ },
2285
+ {
2286
+ "epoch": 0.13,
2287
+ "grad_norm": 5.03649377822876,
2288
+ "learning_rate": 4.975881148415765e-05,
2289
+ "loss": 2.9237,
2290
+ "step": 3190
2291
+ },
2292
+ {
2293
+ "epoch": 0.13,
2294
+ "grad_norm": 5.1522979736328125,
2295
+ "learning_rate": 4.9756603367401814e-05,
2296
+ "loss": 2.96,
2297
+ "step": 3200
2298
+ },
2299
+ {
2300
+ "epoch": 0.13,
2301
+ "grad_norm": 6.068181037902832,
2302
+ "learning_rate": 4.975438523827594e-05,
2303
+ "loss": 2.9398,
2304
+ "step": 3210
2305
+ },
2306
+ {
2307
+ "epoch": 0.13,
2308
+ "grad_norm": 5.104649066925049,
2309
+ "learning_rate": 4.975215709767712e-05,
2310
+ "loss": 2.781,
2311
+ "step": 3220
2312
+ },
2313
+ {
2314
+ "epoch": 0.13,
2315
+ "grad_norm": 7.2377471923828125,
2316
+ "learning_rate": 4.974991894650649e-05,
2317
+ "loss": 3.2773,
2318
+ "step": 3230
2319
+ },
2320
+ {
2321
+ "epoch": 0.13,
2322
+ "grad_norm": 4.730634689331055,
2323
+ "learning_rate": 4.974767078566922e-05,
2324
+ "loss": 3.0184,
2325
+ "step": 3240
2326
+ },
2327
+ {
2328
+ "epoch": 0.13,
2329
+ "grad_norm": 5.900280952453613,
2330
+ "learning_rate": 4.974541261607454e-05,
2331
+ "loss": 2.9432,
2332
+ "step": 3250
2333
+ },
2334
+ {
2335
+ "epoch": 0.13,
2336
+ "grad_norm": 5.755638122558594,
2337
+ "learning_rate": 4.974314443863573e-05,
2338
+ "loss": 2.9988,
2339
+ "step": 3260
2340
+ },
2341
+ {
2342
+ "epoch": 0.13,
2343
+ "grad_norm": 6.039461612701416,
2344
+ "learning_rate": 4.974086625427011e-05,
2345
+ "loss": 3.0655,
2346
+ "step": 3270
2347
+ },
2348
+ {
2349
+ "epoch": 0.13,
2350
+ "grad_norm": 4.955098628997803,
2351
+ "learning_rate": 4.9738578063899066e-05,
2352
+ "loss": 2.822,
2353
+ "step": 3280
2354
+ },
2355
+ {
2356
+ "epoch": 0.13,
2357
+ "grad_norm": 5.892226696014404,
2358
+ "learning_rate": 4.9736279868447996e-05,
2359
+ "loss": 2.9688,
2360
+ "step": 3290
2361
+ },
2362
+ {
2363
+ "epoch": 0.13,
2364
+ "grad_norm": 5.9840497970581055,
2365
+ "learning_rate": 4.9733971668846376e-05,
2366
+ "loss": 2.9008,
2367
+ "step": 3300
2368
+ },
2369
+ {
2370
+ "epoch": 0.13,
2371
+ "grad_norm": 6.418862819671631,
2372
+ "learning_rate": 4.973165346602771e-05,
2373
+ "loss": 2.9117,
2374
+ "step": 3310
2375
+ },
2376
+ {
2377
+ "epoch": 0.13,
2378
+ "grad_norm": 5.666804790496826,
2379
+ "learning_rate": 4.972932526092956e-05,
2380
+ "loss": 3.0392,
2381
+ "step": 3320
2382
+ },
2383
+ {
2384
+ "epoch": 0.13,
2385
+ "grad_norm": 5.902047634124756,
2386
+ "learning_rate": 4.972698705449353e-05,
2387
+ "loss": 3.0601,
2388
+ "step": 3330
2389
+ },
2390
+ {
2391
+ "epoch": 0.13,
2392
+ "grad_norm": 6.236441612243652,
2393
+ "learning_rate": 4.972463884766525e-05,
2394
+ "loss": 2.961,
2395
+ "step": 3340
2396
+ },
2397
+ {
2398
+ "epoch": 0.13,
2399
+ "grad_norm": 6.548509120941162,
2400
+ "learning_rate": 4.972228064139444e-05,
2401
+ "loss": 2.8966,
2402
+ "step": 3350
2403
+ },
2404
+ {
2405
+ "epoch": 0.13,
2406
+ "grad_norm": 5.943577766418457,
2407
+ "learning_rate": 4.9719912436634796e-05,
2408
+ "loss": 3.0408,
2409
+ "step": 3360
2410
+ },
2411
+ {
2412
+ "epoch": 0.13,
2413
+ "grad_norm": 6.531796455383301,
2414
+ "learning_rate": 4.971753423434413e-05,
2415
+ "loss": 2.8598,
2416
+ "step": 3370
2417
+ },
2418
+ {
2419
+ "epoch": 0.13,
2420
+ "grad_norm": 6.425514221191406,
2421
+ "learning_rate": 4.971514603548425e-05,
2422
+ "loss": 3.0003,
2423
+ "step": 3380
2424
+ },
2425
+ {
2426
+ "epoch": 0.13,
2427
+ "grad_norm": 5.9626593589782715,
2428
+ "learning_rate": 4.9712747841021024e-05,
2429
+ "loss": 3.0137,
2430
+ "step": 3390
2431
+ },
2432
+ {
2433
+ "epoch": 0.13,
2434
+ "grad_norm": 5.935604572296143,
2435
+ "learning_rate": 4.971033965192435e-05,
2436
+ "loss": 2.9509,
2437
+ "step": 3400
2438
+ },
2439
+ {
2440
+ "epoch": 0.14,
2441
+ "grad_norm": 6.610740661621094,
2442
+ "learning_rate": 4.9708163737130644e-05,
2443
+ "loss": 3.0231,
2444
+ "step": 3410
2445
+ },
2446
+ {
2447
+ "epoch": 0.14,
2448
+ "grad_norm": 6.248142242431641,
2449
+ "learning_rate": 4.970573656091701e-05,
2450
+ "loss": 2.7345,
2451
+ "step": 3420
2452
+ },
2453
+ {
2454
+ "epoch": 0.14,
2455
+ "grad_norm": 5.92423677444458,
2456
+ "learning_rate": 4.9703299392905516e-05,
2457
+ "loss": 2.9733,
2458
+ "step": 3430
2459
+ },
2460
+ {
2461
+ "epoch": 0.14,
2462
+ "grad_norm": 5.833972454071045,
2463
+ "learning_rate": 4.9700852234081843e-05,
2464
+ "loss": 2.8882,
2465
+ "step": 3440
2466
+ },
2467
+ {
2468
+ "epoch": 0.14,
2469
+ "grad_norm": 5.712262153625488,
2470
+ "learning_rate": 4.96983950854357e-05,
2471
+ "loss": 2.9849,
2472
+ "step": 3450
2473
+ },
2474
+ {
2475
+ "epoch": 0.14,
2476
+ "grad_norm": 5.70693826675415,
2477
+ "learning_rate": 4.969592794796083e-05,
2478
+ "loss": 2.7034,
2479
+ "step": 3460
2480
+ },
2481
+ {
2482
+ "epoch": 0.14,
2483
+ "grad_norm": 6.603397369384766,
2484
+ "learning_rate": 4.969345082265503e-05,
2485
+ "loss": 2.9441,
2486
+ "step": 3470
2487
+ },
2488
+ {
2489
+ "epoch": 0.14,
2490
+ "grad_norm": 5.455355644226074,
2491
+ "learning_rate": 4.9690963710520125e-05,
2492
+ "loss": 3.0818,
2493
+ "step": 3480
2494
+ },
2495
+ {
2496
+ "epoch": 0.14,
2497
+ "grad_norm": 5.410728454589844,
2498
+ "learning_rate": 4.968846661256199e-05,
2499
+ "loss": 2.7837,
2500
+ "step": 3490
2501
+ },
2502
+ {
2503
+ "epoch": 0.14,
2504
+ "grad_norm": 5.591252326965332,
2505
+ "learning_rate": 4.9685959529790526e-05,
2506
+ "loss": 2.8103,
2507
+ "step": 3500
2508
+ },
2509
+ {
2510
+ "epoch": 0.14,
2511
+ "eval_loss": 2.493670701980591,
2512
+ "eval_runtime": 423.4603,
2513
+ "eval_samples_per_second": 50.113,
2514
+ "eval_steps_per_second": 0.392,
2515
+ "step": 3500
2516
+ },
2517
+ {
2518
+ "epoch": 0.14,
2519
+ "grad_norm": 6.1827216148376465,
2520
+ "learning_rate": 4.9683442463219674e-05,
2521
+ "loss": 2.9208,
2522
+ "step": 3510
2523
+ },
2524
+ {
2525
+ "epoch": 0.14,
2526
+ "grad_norm": 5.817008018493652,
2527
+ "learning_rate": 4.968091541386742e-05,
2528
+ "loss": 2.9904,
2529
+ "step": 3520
2530
+ },
2531
+ {
2532
+ "epoch": 0.14,
2533
+ "grad_norm": 5.470141887664795,
2534
+ "learning_rate": 4.96783783827558e-05,
2535
+ "loss": 2.7532,
2536
+ "step": 3530
2537
+ },
2538
+ {
2539
+ "epoch": 0.14,
2540
+ "grad_norm": 6.129313945770264,
2541
+ "learning_rate": 4.967583137091085e-05,
2542
+ "loss": 3.1164,
2543
+ "step": 3540
2544
+ },
2545
+ {
2546
+ "epoch": 0.14,
2547
+ "grad_norm": 5.951664924621582,
2548
+ "learning_rate": 4.967327437936268e-05,
2549
+ "loss": 3.0038,
2550
+ "step": 3550
2551
+ },
2552
+ {
2553
+ "epoch": 0.14,
2554
+ "grad_norm": 5.184429168701172,
2555
+ "learning_rate": 4.96707074091454e-05,
2556
+ "loss": 2.8789,
2557
+ "step": 3560
2558
+ },
2559
+ {
2560
+ "epoch": 0.14,
2561
+ "grad_norm": 5.133409023284912,
2562
+ "learning_rate": 4.9668130461297194e-05,
2563
+ "loss": 2.8137,
2564
+ "step": 3570
2565
+ },
2566
+ {
2567
+ "epoch": 0.14,
2568
+ "grad_norm": 5.30552339553833,
2569
+ "learning_rate": 4.9665543536860255e-05,
2570
+ "loss": 2.764,
2571
+ "step": 3580
2572
+ },
2573
+ {
2574
+ "epoch": 0.14,
2575
+ "grad_norm": 4.916990280151367,
2576
+ "learning_rate": 4.966294663688083e-05,
2577
+ "loss": 2.8359,
2578
+ "step": 3590
2579
+ },
2580
+ {
2581
+ "epoch": 0.14,
2582
+ "grad_norm": 5.573614597320557,
2583
+ "learning_rate": 4.966033976240916e-05,
2584
+ "loss": 2.9216,
2585
+ "step": 3600
2586
+ },
2587
+ {
2588
+ "epoch": 0.14,
2589
+ "grad_norm": 5.318258285522461,
2590
+ "learning_rate": 4.965772291449958e-05,
2591
+ "loss": 2.8818,
2592
+ "step": 3610
2593
+ },
2594
+ {
2595
+ "epoch": 0.14,
2596
+ "grad_norm": 6.087459564208984,
2597
+ "learning_rate": 4.965509609421042e-05,
2598
+ "loss": 2.6802,
2599
+ "step": 3620
2600
+ },
2601
+ {
2602
+ "epoch": 0.14,
2603
+ "grad_norm": 5.408172607421875,
2604
+ "learning_rate": 4.9652459302604046e-05,
2605
+ "loss": 2.7713,
2606
+ "step": 3630
2607
+ },
2608
+ {
2609
+ "epoch": 0.14,
2610
+ "grad_norm": 4.441077709197998,
2611
+ "learning_rate": 4.964981254074686e-05,
2612
+ "loss": 2.8727,
2613
+ "step": 3640
2614
+ },
2615
+ {
2616
+ "epoch": 0.14,
2617
+ "grad_norm": 6.799238681793213,
2618
+ "learning_rate": 4.9647155809709306e-05,
2619
+ "loss": 2.9911,
2620
+ "step": 3650
2621
+ },
2622
+ {
2623
+ "epoch": 0.15,
2624
+ "grad_norm": 5.362485885620117,
2625
+ "learning_rate": 4.9644489110565845e-05,
2626
+ "loss": 2.8088,
2627
+ "step": 3660
2628
+ },
2629
+ {
2630
+ "epoch": 0.15,
2631
+ "grad_norm": 6.419576168060303,
2632
+ "learning_rate": 4.964181244439498e-05,
2633
+ "loss": 2.8133,
2634
+ "step": 3670
2635
+ },
2636
+ {
2637
+ "epoch": 0.15,
2638
+ "grad_norm": 6.19252347946167,
2639
+ "learning_rate": 4.963912581227924e-05,
2640
+ "loss": 2.8429,
2641
+ "step": 3680
2642
+ },
2643
+ {
2644
+ "epoch": 0.15,
2645
+ "grad_norm": 6.425354480743408,
2646
+ "learning_rate": 4.963642921530518e-05,
2647
+ "loss": 2.8272,
2648
+ "step": 3690
2649
+ },
2650
+ {
2651
+ "epoch": 0.15,
2652
+ "grad_norm": 5.866875171661377,
2653
+ "learning_rate": 4.9633722654563405e-05,
2654
+ "loss": 2.816,
2655
+ "step": 3700
2656
+ },
2657
+ {
2658
+ "epoch": 0.15,
2659
+ "grad_norm": 5.425843238830566,
2660
+ "learning_rate": 4.963100613114852e-05,
2661
+ "loss": 2.8552,
2662
+ "step": 3710
2663
+ },
2664
+ {
2665
+ "epoch": 0.15,
2666
+ "grad_norm": 6.268444538116455,
2667
+ "learning_rate": 4.962827964615919e-05,
2668
+ "loss": 2.8642,
2669
+ "step": 3720
2670
+ },
2671
+ {
2672
+ "epoch": 0.15,
2673
+ "grad_norm": 4.511022090911865,
2674
+ "learning_rate": 4.962554320069808e-05,
2675
+ "loss": 2.8388,
2676
+ "step": 3730
2677
+ },
2678
+ {
2679
+ "epoch": 0.15,
2680
+ "grad_norm": 8.639039993286133,
2681
+ "learning_rate": 4.9622796795871904e-05,
2682
+ "loss": 2.809,
2683
+ "step": 3740
2684
+ },
2685
+ {
2686
+ "epoch": 0.15,
2687
+ "grad_norm": 6.79989767074585,
2688
+ "learning_rate": 4.9620040432791395e-05,
2689
+ "loss": 3.0316,
2690
+ "step": 3750
2691
+ },
2692
+ {
2693
+ "epoch": 0.15,
2694
+ "grad_norm": 5.568778991699219,
2695
+ "learning_rate": 4.961727411257131e-05,
2696
+ "loss": 2.7867,
2697
+ "step": 3760
2698
+ },
2699
+ {
2700
+ "epoch": 0.15,
2701
+ "grad_norm": 5.364297389984131,
2702
+ "learning_rate": 4.961449783633045e-05,
2703
+ "loss": 3.0213,
2704
+ "step": 3770
2705
+ },
2706
+ {
2707
+ "epoch": 0.15,
2708
+ "grad_norm": 5.523791790008545,
2709
+ "learning_rate": 4.9611711605191625e-05,
2710
+ "loss": 2.7875,
2711
+ "step": 3780
2712
+ },
2713
+ {
2714
+ "epoch": 0.15,
2715
+ "grad_norm": 5.39138650894165,
2716
+ "learning_rate": 4.9608915420281675e-05,
2717
+ "loss": 2.7144,
2718
+ "step": 3790
2719
+ },
2720
+ {
2721
+ "epoch": 0.15,
2722
+ "grad_norm": 5.800009250640869,
2723
+ "learning_rate": 4.9606109282731463e-05,
2724
+ "loss": 2.8091,
2725
+ "step": 3800
2726
+ },
2727
+ {
2728
+ "epoch": 0.15,
2729
+ "grad_norm": 5.470329284667969,
2730
+ "learning_rate": 4.96032931936759e-05,
2731
+ "loss": 2.8055,
2732
+ "step": 3810
2733
+ },
2734
+ {
2735
+ "epoch": 0.15,
2736
+ "grad_norm": 5.461717128753662,
2737
+ "learning_rate": 4.9600467154253885e-05,
2738
+ "loss": 2.8633,
2739
+ "step": 3820
2740
+ },
2741
+ {
2742
+ "epoch": 0.15,
2743
+ "grad_norm": 5.095701217651367,
2744
+ "learning_rate": 4.959763116560836e-05,
2745
+ "loss": 2.9004,
2746
+ "step": 3830
2747
+ },
2748
+ {
2749
+ "epoch": 0.15,
2750
+ "grad_norm": 5.429464340209961,
2751
+ "learning_rate": 4.95947852288863e-05,
2752
+ "loss": 2.6428,
2753
+ "step": 3840
2754
+ },
2755
+ {
2756
+ "epoch": 0.15,
2757
+ "grad_norm": 5.361551761627197,
2758
+ "learning_rate": 4.95919293452387e-05,
2759
+ "loss": 2.867,
2760
+ "step": 3850
2761
+ },
2762
+ {
2763
+ "epoch": 0.15,
2764
+ "grad_norm": 5.917137622833252,
2765
+ "learning_rate": 4.958906351582054e-05,
2766
+ "loss": 2.7266,
2767
+ "step": 3860
2768
+ },
2769
+ {
2770
+ "epoch": 0.15,
2771
+ "grad_norm": 5.074246406555176,
2772
+ "learning_rate": 4.958618774179089e-05,
2773
+ "loss": 2.8333,
2774
+ "step": 3870
2775
+ },
2776
+ {
2777
+ "epoch": 0.15,
2778
+ "grad_norm": 6.167778968811035,
2779
+ "learning_rate": 4.9583302024312796e-05,
2780
+ "loss": 2.7009,
2781
+ "step": 3880
2782
+ },
2783
+ {
2784
+ "epoch": 0.15,
2785
+ "grad_norm": 4.73693323135376,
2786
+ "learning_rate": 4.9580406364553325e-05,
2787
+ "loss": 2.5967,
2788
+ "step": 3890
2789
+ },
2790
+ {
2791
+ "epoch": 0.15,
2792
+ "grad_norm": 5.558803558349609,
2793
+ "learning_rate": 4.957750076368357e-05,
2794
+ "loss": 2.6763,
2795
+ "step": 3900
2796
+ },
2797
+ {
2798
+ "epoch": 0.16,
2799
+ "grad_norm": 7.1173481941223145,
2800
+ "learning_rate": 4.957458522287868e-05,
2801
+ "loss": 2.8593,
2802
+ "step": 3910
2803
+ },
2804
+ {
2805
+ "epoch": 0.16,
2806
+ "grad_norm": 7.378798484802246,
2807
+ "learning_rate": 4.9571659743317754e-05,
2808
+ "loss": 2.7876,
2809
+ "step": 3920
2810
+ },
2811
+ {
2812
+ "epoch": 0.16,
2813
+ "grad_norm": 6.375428199768066,
2814
+ "learning_rate": 4.956872432618399e-05,
2815
+ "loss": 2.7443,
2816
+ "step": 3930
2817
+ },
2818
+ {
2819
+ "epoch": 0.16,
2820
+ "grad_norm": 5.509357452392578,
2821
+ "learning_rate": 4.9565778972664534e-05,
2822
+ "loss": 2.8787,
2823
+ "step": 3940
2824
+ },
2825
+ {
2826
+ "epoch": 0.16,
2827
+ "grad_norm": 5.430800437927246,
2828
+ "learning_rate": 4.95628236839506e-05,
2829
+ "loss": 2.609,
2830
+ "step": 3950
2831
+ },
2832
+ {
2833
+ "epoch": 0.16,
2834
+ "grad_norm": 5.8563618659973145,
2835
+ "learning_rate": 4.95598584612374e-05,
2836
+ "loss": 2.7912,
2837
+ "step": 3960
2838
+ },
2839
+ {
2840
+ "epoch": 0.16,
2841
+ "grad_norm": 5.738502502441406,
2842
+ "learning_rate": 4.9556883305724155e-05,
2843
+ "loss": 2.8222,
2844
+ "step": 3970
2845
+ },
2846
+ {
2847
+ "epoch": 0.16,
2848
+ "grad_norm": 4.658450603485107,
2849
+ "learning_rate": 4.955389821861412e-05,
2850
+ "loss": 2.6731,
2851
+ "step": 3980
2852
+ },
2853
+ {
2854
+ "epoch": 0.16,
2855
+ "grad_norm": 6.250921726226807,
2856
+ "learning_rate": 4.955090320111457e-05,
2857
+ "loss": 2.6865,
2858
+ "step": 3990
2859
+ },
2860
+ {
2861
+ "epoch": 0.16,
2862
+ "grad_norm": 5.725744247436523,
2863
+ "learning_rate": 4.954789825443678e-05,
2864
+ "loss": 2.6249,
2865
+ "step": 4000
2866
+ },
2867
+ {
2868
+ "epoch": 0.16,
2869
+ "eval_loss": 2.342557668685913,
2870
+ "eval_runtime": 423.1443,
2871
+ "eval_samples_per_second": 50.151,
2872
+ "eval_steps_per_second": 0.392,
2873
+ "step": 4000
2874
+ },
2875
+ {
2876
+ "epoch": 0.16,
2877
+ "grad_norm": 6.008103847503662,
2878
+ "learning_rate": 4.954488337979604e-05,
2879
+ "loss": 2.6645,
2880
+ "step": 4010
2881
+ },
2882
+ {
2883
+ "epoch": 0.16,
2884
+ "grad_norm": 5.027274131774902,
2885
+ "learning_rate": 4.9541858578411674e-05,
2886
+ "loss": 2.8588,
2887
+ "step": 4020
2888
+ },
2889
+ {
2890
+ "epoch": 0.16,
2891
+ "grad_norm": 5.368979454040527,
2892
+ "learning_rate": 4.9538823851507006e-05,
2893
+ "loss": 2.7103,
2894
+ "step": 4030
2895
+ },
2896
+ {
2897
+ "epoch": 0.16,
2898
+ "grad_norm": 4.687747001647949,
2899
+ "learning_rate": 4.953577920030937e-05,
2900
+ "loss": 2.5958,
2901
+ "step": 4040
2902
+ },
2903
+ {
2904
+ "epoch": 0.16,
2905
+ "grad_norm": 5.669676780700684,
2906
+ "learning_rate": 4.953272462605013e-05,
2907
+ "loss": 2.7199,
2908
+ "step": 4050
2909
+ },
2910
+ {
2911
+ "epoch": 0.16,
2912
+ "grad_norm": 5.0332112312316895,
2913
+ "learning_rate": 4.952966012996466e-05,
2914
+ "loss": 2.62,
2915
+ "step": 4060
2916
+ },
2917
+ {
2918
+ "epoch": 0.16,
2919
+ "grad_norm": 5.005265235900879,
2920
+ "learning_rate": 4.952658571329233e-05,
2921
+ "loss": 2.5974,
2922
+ "step": 4070
2923
+ },
2924
+ {
2925
+ "epoch": 0.16,
2926
+ "grad_norm": 5.806674003601074,
2927
+ "learning_rate": 4.952350137727653e-05,
2928
+ "loss": 2.9317,
2929
+ "step": 4080
2930
+ },
2931
+ {
2932
+ "epoch": 0.16,
2933
+ "grad_norm": 5.117702484130859,
2934
+ "learning_rate": 4.9520407123164676e-05,
2935
+ "loss": 2.6551,
2936
+ "step": 4090
2937
+ },
2938
+ {
2939
+ "epoch": 0.16,
2940
+ "grad_norm": 6.5031023025512695,
2941
+ "learning_rate": 4.9517302952208174e-05,
2942
+ "loss": 2.7362,
2943
+ "step": 4100
2944
+ },
2945
+ {
2946
+ "epoch": 0.16,
2947
+ "grad_norm": 6.63228702545166,
2948
+ "learning_rate": 4.951418886566247e-05,
2949
+ "loss": 2.6743,
2950
+ "step": 4110
2951
+ },
2952
+ {
2953
+ "epoch": 0.16,
2954
+ "grad_norm": 5.421300888061523,
2955
+ "learning_rate": 4.951106486478698e-05,
2956
+ "loss": 2.6827,
2957
+ "step": 4120
2958
+ },
2959
+ {
2960
+ "epoch": 0.16,
2961
+ "grad_norm": 5.456374168395996,
2962
+ "learning_rate": 4.950793095084516e-05,
2963
+ "loss": 2.7266,
2964
+ "step": 4130
2965
+ },
2966
+ {
2967
+ "epoch": 0.16,
2968
+ "grad_norm": 6.39730978012085,
2969
+ "learning_rate": 4.950478712510446e-05,
2970
+ "loss": 2.8165,
2971
+ "step": 4140
2972
+ },
2973
+ {
2974
+ "epoch": 0.16,
2975
+ "grad_norm": 5.817816734313965,
2976
+ "learning_rate": 4.9501633388836354e-05,
2977
+ "loss": 2.6275,
2978
+ "step": 4150
2979
+ },
2980
+ {
2981
+ "epoch": 0.17,
2982
+ "grad_norm": 7.556663990020752,
2983
+ "learning_rate": 4.949846974331631e-05,
2984
+ "loss": 2.6368,
2985
+ "step": 4160
2986
+ },
2987
+ {
2988
+ "epoch": 0.17,
2989
+ "grad_norm": 6.171590805053711,
2990
+ "learning_rate": 4.949529618982381e-05,
2991
+ "loss": 2.8336,
2992
+ "step": 4170
2993
+ },
2994
+ {
2995
+ "epoch": 0.17,
2996
+ "grad_norm": 6.013977527618408,
2997
+ "learning_rate": 4.9492112729642346e-05,
2998
+ "loss": 2.6988,
2999
+ "step": 4180
3000
+ },
3001
+ {
3002
+ "epoch": 0.17,
3003
+ "grad_norm": 5.861149311065674,
3004
+ "learning_rate": 4.948891936405941e-05,
3005
+ "loss": 2.8883,
3006
+ "step": 4190
3007
+ },
3008
+ {
3009
+ "epoch": 0.17,
3010
+ "grad_norm": 5.240298271179199,
3011
+ "learning_rate": 4.948571609436649e-05,
3012
+ "loss": 2.8353,
3013
+ "step": 4200
3014
+ },
3015
+ {
3016
+ "epoch": 0.17,
3017
+ "grad_norm": 5.673245429992676,
3018
+ "learning_rate": 4.94825029218591e-05,
3019
+ "loss": 2.6835,
3020
+ "step": 4210
3021
+ },
3022
+ {
3023
+ "epoch": 0.17,
3024
+ "grad_norm": 5.680504322052002,
3025
+ "learning_rate": 4.947927984783674e-05,
3026
+ "loss": 2.594,
3027
+ "step": 4220
3028
+ },
3029
+ {
3030
+ "epoch": 0.17,
3031
+ "grad_norm": 4.967173099517822,
3032
+ "learning_rate": 4.947604687360295e-05,
3033
+ "loss": 2.5797,
3034
+ "step": 4230
3035
+ },
3036
+ {
3037
+ "epoch": 0.17,
3038
+ "grad_norm": 6.29856538772583,
3039
+ "learning_rate": 4.947280400046525e-05,
3040
+ "loss": 2.6197,
3041
+ "step": 4240
3042
+ },
3043
+ {
3044
+ "epoch": 0.17,
3045
+ "grad_norm": 5.518858432769775,
3046
+ "learning_rate": 4.9469551229735126e-05,
3047
+ "loss": 2.6727,
3048
+ "step": 4250
3049
+ },
3050
+ {
3051
+ "epoch": 0.17,
3052
+ "grad_norm": 5.493609428405762,
3053
+ "learning_rate": 4.946628856272814e-05,
3054
+ "loss": 2.521,
3055
+ "step": 4260
3056
+ },
3057
+ {
3058
+ "epoch": 0.17,
3059
+ "grad_norm": 5.581683158874512,
3060
+ "learning_rate": 4.9463016000763794e-05,
3061
+ "loss": 2.4878,
3062
+ "step": 4270
3063
+ },
3064
+ {
3065
+ "epoch": 0.17,
3066
+ "grad_norm": 5.769138336181641,
3067
+ "learning_rate": 4.945973354516564e-05,
3068
+ "loss": 2.6396,
3069
+ "step": 4280
3070
+ },
3071
+ {
3072
+ "epoch": 0.17,
3073
+ "grad_norm": 5.217662811279297,
3074
+ "learning_rate": 4.9456441197261195e-05,
3075
+ "loss": 2.3943,
3076
+ "step": 4290
3077
+ },
3078
+ {
3079
+ "epoch": 0.17,
3080
+ "grad_norm": 5.986959457397461,
3081
+ "learning_rate": 4.9453138958381994e-05,
3082
+ "loss": 2.9015,
3083
+ "step": 4300
3084
+ },
3085
+ {
3086
+ "epoch": 0.17,
3087
+ "grad_norm": 4.872792720794678,
3088
+ "learning_rate": 4.944982682986357e-05,
3089
+ "loss": 2.7326,
3090
+ "step": 4310
3091
+ },
3092
+ {
3093
+ "epoch": 0.17,
3094
+ "grad_norm": 5.723078727722168,
3095
+ "learning_rate": 4.944650481304545e-05,
3096
+ "loss": 2.7212,
3097
+ "step": 4320
3098
+ },
3099
+ {
3100
+ "epoch": 0.17,
3101
+ "grad_norm": 6.189296722412109,
3102
+ "learning_rate": 4.944317290927117e-05,
3103
+ "loss": 2.8079,
3104
+ "step": 4330
3105
+ },
3106
+ {
3107
+ "epoch": 0.17,
3108
+ "grad_norm": 5.358461856842041,
3109
+ "learning_rate": 4.943983111988827e-05,
3110
+ "loss": 2.6931,
3111
+ "step": 4340
3112
+ },
3113
+ {
3114
+ "epoch": 0.17,
3115
+ "grad_norm": 5.47238826751709,
3116
+ "learning_rate": 4.943647944624826e-05,
3117
+ "loss": 2.5772,
3118
+ "step": 4350
3119
+ },
3120
+ {
3121
+ "epoch": 0.17,
3122
+ "grad_norm": 5.701056957244873,
3123
+ "learning_rate": 4.943311788970667e-05,
3124
+ "loss": 2.5407,
3125
+ "step": 4360
3126
+ },
3127
+ {
3128
+ "epoch": 0.17,
3129
+ "grad_norm": 6.032464504241943,
3130
+ "learning_rate": 4.9429746451623024e-05,
3131
+ "loss": 2.6685,
3132
+ "step": 4370
3133
+ },
3134
+ {
3135
+ "epoch": 0.17,
3136
+ "grad_norm": 5.141815185546875,
3137
+ "learning_rate": 4.9426365133360834e-05,
3138
+ "loss": 2.7409,
3139
+ "step": 4380
3140
+ },
3141
+ {
3142
+ "epoch": 0.17,
3143
+ "grad_norm": 6.483497619628906,
3144
+ "learning_rate": 4.942297393628763e-05,
3145
+ "loss": 2.5305,
3146
+ "step": 4390
3147
+ },
3148
+ {
3149
+ "epoch": 0.17,
3150
+ "grad_norm": 6.223423957824707,
3151
+ "learning_rate": 4.941957286177491e-05,
3152
+ "loss": 2.8228,
3153
+ "step": 4400
3154
+ },
3155
+ {
3156
+ "epoch": 0.17,
3157
+ "grad_norm": 4.8881402015686035,
3158
+ "learning_rate": 4.9416161911198176e-05,
3159
+ "loss": 2.5842,
3160
+ "step": 4410
3161
+ },
3162
+ {
3163
+ "epoch": 0.18,
3164
+ "grad_norm": 5.5558929443359375,
3165
+ "learning_rate": 4.941274108593693e-05,
3166
+ "loss": 2.5049,
3167
+ "step": 4420
3168
+ },
3169
+ {
3170
+ "epoch": 0.18,
3171
+ "grad_norm": 7.3317694664001465,
3172
+ "learning_rate": 4.940931038737468e-05,
3173
+ "loss": 2.5361,
3174
+ "step": 4430
3175
+ },
3176
+ {
3177
+ "epoch": 0.18,
3178
+ "grad_norm": 6.913518905639648,
3179
+ "learning_rate": 4.940586981689888e-05,
3180
+ "loss": 2.7733,
3181
+ "step": 4440
3182
+ },
3183
+ {
3184
+ "epoch": 0.18,
3185
+ "grad_norm": 8.628571510314941,
3186
+ "learning_rate": 4.940241937590102e-05,
3187
+ "loss": 2.642,
3188
+ "step": 4450
3189
+ },
3190
+ {
3191
+ "epoch": 0.18,
3192
+ "grad_norm": 6.870291233062744,
3193
+ "learning_rate": 4.939895906577658e-05,
3194
+ "loss": 2.9244,
3195
+ "step": 4460
3196
+ },
3197
+ {
3198
+ "epoch": 0.18,
3199
+ "grad_norm": 5.8803253173828125,
3200
+ "learning_rate": 4.939548888792502e-05,
3201
+ "loss": 2.6324,
3202
+ "step": 4470
3203
+ },
3204
+ {
3205
+ "epoch": 0.18,
3206
+ "grad_norm": 6.114381790161133,
3207
+ "learning_rate": 4.939200884374979e-05,
3208
+ "loss": 2.674,
3209
+ "step": 4480
3210
+ },
3211
+ {
3212
+ "epoch": 0.18,
3213
+ "grad_norm": 7.145431041717529,
3214
+ "learning_rate": 4.938851893465832e-05,
3215
+ "loss": 2.6255,
3216
+ "step": 4490
3217
+ },
3218
+ {
3219
+ "epoch": 0.18,
3220
+ "grad_norm": 6.35526704788208,
3221
+ "learning_rate": 4.938501916206206e-05,
3222
+ "loss": 2.8467,
3223
+ "step": 4500
3224
+ },
3225
+ {
3226
+ "epoch": 0.18,
3227
+ "eval_loss": 2.259162664413452,
3228
+ "eval_runtime": 421.9557,
3229
+ "eval_samples_per_second": 50.292,
3230
+ "eval_steps_per_second": 0.393,
3231
+ "step": 4500
3232
+ },
3233
+ {
3234
+ "epoch": 0.18,
3235
+ "grad_norm": 6.096010208129883,
3236
+ "learning_rate": 4.938150952737642e-05,
3237
+ "loss": 2.6732,
3238
+ "step": 4510
3239
+ },
3240
+ {
3241
+ "epoch": 0.18,
3242
+ "grad_norm": 4.902897357940674,
3243
+ "learning_rate": 4.937799003202081e-05,
3244
+ "loss": 2.5127,
3245
+ "step": 4520
3246
+ },
3247
+ {
3248
+ "epoch": 0.18,
3249
+ "grad_norm": 5.106553077697754,
3250
+ "learning_rate": 4.9374460677418635e-05,
3251
+ "loss": 2.5394,
3252
+ "step": 4530
3253
+ },
3254
+ {
3255
+ "epoch": 0.18,
3256
+ "grad_norm": 5.985318183898926,
3257
+ "learning_rate": 4.937092146499727e-05,
3258
+ "loss": 2.6342,
3259
+ "step": 4540
3260
+ },
3261
+ {
3262
+ "epoch": 0.18,
3263
+ "grad_norm": 5.647487163543701,
3264
+ "learning_rate": 4.9367372396188095e-05,
3265
+ "loss": 2.7253,
3266
+ "step": 4550
3267
+ },
3268
+ {
3269
+ "epoch": 0.18,
3270
+ "grad_norm": 5.041297435760498,
3271
+ "learning_rate": 4.9363813472426465e-05,
3272
+ "loss": 2.7264,
3273
+ "step": 4560
3274
+ },
3275
+ {
3276
+ "epoch": 0.18,
3277
+ "grad_norm": 6.295587062835693,
3278
+ "learning_rate": 4.9360244695151716e-05,
3279
+ "loss": 2.6379,
3280
+ "step": 4570
3281
+ },
3282
+ {
3283
+ "epoch": 0.18,
3284
+ "grad_norm": 5.751054286956787,
3285
+ "learning_rate": 4.935666606580719e-05,
3286
+ "loss": 2.6497,
3287
+ "step": 4580
3288
+ },
3289
+ {
3290
+ "epoch": 0.18,
3291
+ "grad_norm": 5.367273330688477,
3292
+ "learning_rate": 4.935307758584019e-05,
3293
+ "loss": 2.3892,
3294
+ "step": 4590
3295
+ },
3296
+ {
3297
+ "epoch": 0.18,
3298
+ "grad_norm": 5.491433143615723,
3299
+ "learning_rate": 4.9349479256702023e-05,
3300
+ "loss": 2.4803,
3301
+ "step": 4600
3302
+ },
3303
+ {
3304
+ "epoch": 0.18,
3305
+ "grad_norm": 5.676270484924316,
3306
+ "learning_rate": 4.934587107984796e-05,
3307
+ "loss": 2.4972,
3308
+ "step": 4610
3309
+ },
3310
+ {
3311
+ "epoch": 0.18,
3312
+ "grad_norm": 5.060196876525879,
3313
+ "learning_rate": 4.934225305673728e-05,
3314
+ "loss": 2.4878,
3315
+ "step": 4620
3316
+ },
3317
+ {
3318
+ "epoch": 0.18,
3319
+ "grad_norm": 5.352468967437744,
3320
+ "learning_rate": 4.93386251888332e-05,
3321
+ "loss": 2.4713,
3322
+ "step": 4630
3323
+ },
3324
+ {
3325
+ "epoch": 0.18,
3326
+ "grad_norm": 8.105636596679688,
3327
+ "learning_rate": 4.9334987477602975e-05,
3328
+ "loss": 2.7637,
3329
+ "step": 4640
3330
+ },
3331
+ {
3332
+ "epoch": 0.18,
3333
+ "grad_norm": 5.352108478546143,
3334
+ "learning_rate": 4.933133992451779e-05,
3335
+ "loss": 2.5642,
3336
+ "step": 4650
3337
+ },
3338
+ {
3339
+ "epoch": 0.18,
3340
+ "grad_norm": 5.43104362487793,
3341
+ "learning_rate": 4.932768253105286e-05,
3342
+ "loss": 2.6106,
3343
+ "step": 4660
3344
+ },
3345
+ {
3346
+ "epoch": 0.19,
3347
+ "grad_norm": 5.641717433929443,
3348
+ "learning_rate": 4.932401529868733e-05,
3349
+ "loss": 2.6049,
3350
+ "step": 4670
3351
+ },
3352
+ {
3353
+ "epoch": 0.19,
3354
+ "grad_norm": 5.041056156158447,
3355
+ "learning_rate": 4.932033822890436e-05,
3356
+ "loss": 2.512,
3357
+ "step": 4680
3358
+ },
3359
+ {
3360
+ "epoch": 0.19,
3361
+ "grad_norm": 4.870728015899658,
3362
+ "learning_rate": 4.9316651323191067e-05,
3363
+ "loss": 2.4649,
3364
+ "step": 4690
3365
+ },
3366
+ {
3367
+ "epoch": 0.19,
3368
+ "grad_norm": 6.347024917602539,
3369
+ "learning_rate": 4.931295458303856e-05,
3370
+ "loss": 2.5996,
3371
+ "step": 4700
3372
+ },
3373
+ {
3374
+ "epoch": 0.19,
3375
+ "grad_norm": 5.286810398101807,
3376
+ "learning_rate": 4.9309248009941914e-05,
3377
+ "loss": 2.5899,
3378
+ "step": 4710
3379
+ },
3380
+ {
3381
+ "epoch": 0.19,
3382
+ "grad_norm": 6.839949131011963,
3383
+ "learning_rate": 4.93055316054002e-05,
3384
+ "loss": 2.5646,
3385
+ "step": 4720
3386
+ },
3387
+ {
3388
+ "epoch": 0.19,
3389
+ "grad_norm": 6.071991920471191,
3390
+ "learning_rate": 4.930180537091644e-05,
3391
+ "loss": 2.7168,
3392
+ "step": 4730
3393
+ },
3394
+ {
3395
+ "epoch": 0.19,
3396
+ "grad_norm": 5.042398452758789,
3397
+ "learning_rate": 4.929806930799765e-05,
3398
+ "loss": 2.7169,
3399
+ "step": 4740
3400
+ },
3401
+ {
3402
+ "epoch": 0.19,
3403
+ "grad_norm": 5.924988269805908,
3404
+ "learning_rate": 4.9294323418154805e-05,
3405
+ "loss": 2.4587,
3406
+ "step": 4750
3407
+ },
3408
+ {
3409
+ "epoch": 0.19,
3410
+ "grad_norm": 6.021846294403076,
3411
+ "learning_rate": 4.929056770290288e-05,
3412
+ "loss": 2.5327,
3413
+ "step": 4760
3414
+ },
3415
+ {
3416
+ "epoch": 0.19,
3417
+ "grad_norm": 5.043169975280762,
3418
+ "learning_rate": 4.928680216376079e-05,
3419
+ "loss": 2.4268,
3420
+ "step": 4770
3421
+ },
3422
+ {
3423
+ "epoch": 0.19,
3424
+ "grad_norm": 7.106345176696777,
3425
+ "learning_rate": 4.9283026802251454e-05,
3426
+ "loss": 2.5116,
3427
+ "step": 4780
3428
+ },
3429
+ {
3430
+ "epoch": 0.19,
3431
+ "grad_norm": 5.8474650382995605,
3432
+ "learning_rate": 4.927924161990175e-05,
3433
+ "loss": 2.5163,
3434
+ "step": 4790
3435
+ },
3436
+ {
3437
+ "epoch": 0.19,
3438
+ "grad_norm": 5.736364841461182,
3439
+ "learning_rate": 4.9275446618242516e-05,
3440
+ "loss": 2.4918,
3441
+ "step": 4800
3442
+ },
3443
+ {
3444
+ "epoch": 0.19,
3445
+ "grad_norm": 5.512024879455566,
3446
+ "learning_rate": 4.927164179880858e-05,
3447
+ "loss": 2.5756,
3448
+ "step": 4810
3449
+ },
3450
+ {
3451
+ "epoch": 0.19,
3452
+ "grad_norm": 4.911081790924072,
3453
+ "learning_rate": 4.926782716313874e-05,
3454
+ "loss": 2.4661,
3455
+ "step": 4820
3456
+ },
3457
+ {
3458
+ "epoch": 0.19,
3459
+ "grad_norm": 5.932026386260986,
3460
+ "learning_rate": 4.9264002712775745e-05,
3461
+ "loss": 2.5059,
3462
+ "step": 4830
3463
+ },
3464
+ {
3465
+ "epoch": 0.19,
3466
+ "grad_norm": 6.116228103637695,
3467
+ "learning_rate": 4.9260168449266335e-05,
3468
+ "loss": 2.4183,
3469
+ "step": 4840
3470
+ },
3471
+ {
3472
+ "epoch": 0.19,
3473
+ "grad_norm": 5.023027420043945,
3474
+ "learning_rate": 4.925632437416121e-05,
3475
+ "loss": 2.4144,
3476
+ "step": 4850
3477
+ },
3478
+ {
3479
+ "epoch": 0.19,
3480
+ "grad_norm": 5.4858832359313965,
3481
+ "learning_rate": 4.925247048901504e-05,
3482
+ "loss": 2.4708,
3483
+ "step": 4860
3484
+ },
3485
+ {
3486
+ "epoch": 0.19,
3487
+ "grad_norm": 6.690423965454102,
3488
+ "learning_rate": 4.924860679538645e-05,
3489
+ "loss": 2.7061,
3490
+ "step": 4870
3491
+ },
3492
+ {
3493
+ "epoch": 0.19,
3494
+ "grad_norm": 6.818970203399658,
3495
+ "learning_rate": 4.924473329483806e-05,
3496
+ "loss": 2.6152,
3497
+ "step": 4880
3498
+ },
3499
+ {
3500
+ "epoch": 0.19,
3501
+ "grad_norm": 5.238198280334473,
3502
+ "learning_rate": 4.924084998893642e-05,
3503
+ "loss": 2.3184,
3504
+ "step": 4890
3505
+ },
3506
+ {
3507
+ "epoch": 0.19,
3508
+ "grad_norm": 5.587952613830566,
3509
+ "learning_rate": 4.923695687925207e-05,
3510
+ "loss": 2.6381,
3511
+ "step": 4900
3512
+ },
3513
+ {
3514
+ "epoch": 0.19,
3515
+ "grad_norm": 5.2837066650390625,
3516
+ "learning_rate": 4.923305396735952e-05,
3517
+ "loss": 2.52,
3518
+ "step": 4910
3519
+ },
3520
+ {
3521
+ "epoch": 0.2,
3522
+ "grad_norm": 5.908483028411865,
3523
+ "learning_rate": 4.9229141254837226e-05,
3524
+ "loss": 2.675,
3525
+ "step": 4920
3526
+ },
3527
+ {
3528
+ "epoch": 0.2,
3529
+ "grad_norm": 5.803688049316406,
3530
+ "learning_rate": 4.922521874326761e-05,
3531
+ "loss": 2.6054,
3532
+ "step": 4930
3533
+ },
3534
+ {
3535
+ "epoch": 0.2,
3536
+ "grad_norm": 6.124865531921387,
3537
+ "learning_rate": 4.922128643423707e-05,
3538
+ "loss": 2.6076,
3539
+ "step": 4940
3540
+ },
3541
+ {
3542
+ "epoch": 0.2,
3543
+ "grad_norm": 6.534984111785889,
3544
+ "learning_rate": 4.921734432933596e-05,
3545
+ "loss": 2.4958,
3546
+ "step": 4950
3547
+ },
3548
+ {
3549
+ "epoch": 0.2,
3550
+ "grad_norm": 7.013035297393799,
3551
+ "learning_rate": 4.921339243015858e-05,
3552
+ "loss": 2.6099,
3553
+ "step": 4960
3554
+ },
3555
+ {
3556
+ "epoch": 0.2,
3557
+ "grad_norm": 6.598790645599365,
3558
+ "learning_rate": 4.920943073830322e-05,
3559
+ "loss": 2.493,
3560
+ "step": 4970
3561
+ },
3562
+ {
3563
+ "epoch": 0.2,
3564
+ "grad_norm": 5.958088397979736,
3565
+ "learning_rate": 4.920545925537212e-05,
3566
+ "loss": 2.5689,
3567
+ "step": 4980
3568
+ },
3569
+ {
3570
+ "epoch": 0.2,
3571
+ "grad_norm": 6.180896282196045,
3572
+ "learning_rate": 4.9201477982971466e-05,
3573
+ "loss": 2.6138,
3574
+ "step": 4990
3575
+ },
3576
+ {
3577
+ "epoch": 0.2,
3578
+ "grad_norm": 4.9859209060668945,
3579
+ "learning_rate": 4.919748692271141e-05,
3580
+ "loss": 2.5003,
3581
+ "step": 5000
3582
+ },
3583
+ {
3584
+ "epoch": 0.2,
3585
+ "eval_loss": 2.1625306606292725,
3586
+ "eval_runtime": 422.5397,
3587
+ "eval_samples_per_second": 50.223,
3588
+ "eval_steps_per_second": 0.393,
3589
+ "step": 5000
3590
+ }
3591
+ ],
3592
+ "logging_steps": 10,
3593
+ "max_steps": 50400,
3594
+ "num_input_tokens_seen": 0,
3595
+ "num_train_epochs": 2,
3596
+ "save_steps": 500,
3597
+ "total_flos": 0.0,
3598
+ "train_batch_size": 8,
3599
+ "trial_name": null,
3600
+ "trial_params": null
3601
+ }
checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37d08e731b46f2fc9ac1d4ac376bb688c70f2bf08a26eb4b0de280ef48858eb0
3
+ size 4539