Hugging Face's logo

qiufengqijun
/
mini_qwen_sft_checkpoint-7730

Text Generation
Safetensors
Chinese
English
qwen2
conversational
Model card Files Community
qiufengqijun commited on
Commit
5f6e31e
·
verified ·
1 Parent(s): 6642503

上传权重

Browse files
Files changed (13) hide show
  1. .gitattributes +1 -0
  2. added_tokens.json +24 -0
  3. config.json +29 -0
  4. generation_config.json +6 -0
  5. merges.txt +0 -0
  6. model.safetensors +3 -0
  7. special_tokens_map.json +31 -0
  8. tokenizer.json +3 -0
  9. tokenizer_config.json +207 -0
  10. trainer_state.json +2735 -0
  11. training_args.bin +3 -0
  12. vocab.json +0 -0
  13. zero_to_fp32.py +604 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "results/pt",
3
+ "architectures": [
4
+ "Qwen2ForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 1024,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "max_position_embeddings": 32768,
14
+ "max_window_layers": 21,
15
+ "model_type": "qwen2",
16
+ "num_attention_heads": 16,
17
+ "num_hidden_layers": 48,
18
+ "num_key_value_heads": 4,
19
+ "rms_norm_eps": 1e-06,
20
+ "rope_scaling": null,
21
+ "rope_theta": 1000000.0,
22
+ "sliding_window": null,
23
+ "tie_word_embeddings": true,
24
+ "torch_dtype": "bfloat16",
25
+ "transformers_version": "4.45.0",
26
+ "use_cache": true,
27
+ "use_sliding_window": false,
28
+ "vocab_size": 151936
29
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "4.45.0"
6
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e65c87db8ba6549db8bddcece93542499ab9c9eadcef176542b74aa194eb2e7f
3
+ size 2308852232
special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
3
+ size 11421896
tokenizer_config.json ADDED
@@ -0,0 +1,207 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ }
181
+ },
182
+ "additional_special_tokens": [
183
+ "<|im_start|>",
184
+ "<|im_end|>",
185
+ "<|object_ref_start|>",
186
+ "<|object_ref_end|>",
187
+ "<|box_start|>",
188
+ "<|box_end|>",
189
+ "<|quad_start|>",
190
+ "<|quad_end|>",
191
+ "<|vision_start|>",
192
+ "<|vision_end|>",
193
+ "<|vision_pad|>",
194
+ "<|image_pad|>",
195
+ "<|video_pad|>"
196
+ ],
197
+ "bos_token": null,
198
+ "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0]['role'] == 'system' %}\n {{- messages[0]['content'] }}\n {%- else %}\n {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n {%- endif %}\n {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n {%- if messages[0]['role'] == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n {%- else %}\n {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {{- '<|im_start|>' + message.role }}\n {%- if message.content %}\n {{- '\\n' + message.content }}\n {%- endif %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '\\n<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- message.content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
199
+ "clean_up_tokenization_spaces": false,
200
+ "eos_token": "<|im_end|>",
201
+ "errors": "replace",
202
+ "model_max_length": 131072,
203
+ "pad_token": "<|endoftext|>",
204
+ "split_special_tokens": false,
205
+ "tokenizer_class": "Qwen2Tokenizer",
206
+ "unk_token": null
207
+ }
trainer_state.json ADDED
@@ -0,0 +1,2735 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9998787339827803,
5
+ "eval_steps": 500,
6
+ "global_step": 7730,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.002587008367355188,
13
+ "grad_norm": 3.117795467376709,
14
+ "learning_rate": 8.624407072013799e-08,
15
+ "loss": 3.7293,
16
+ "step": 20
17
+ },
18
+ {
19
+ "epoch": 0.005174016734710376,
20
+ "grad_norm": 3.1848440170288086,
21
+ "learning_rate": 1.7248814144027598e-07,
22
+ "loss": 3.7345,
23
+ "step": 40
24
+ },
25
+ {
26
+ "epoch": 0.007761025102065565,
27
+ "grad_norm": 3.073965072631836,
28
+ "learning_rate": 2.5873221216041403e-07,
29
+ "loss": 3.7232,
30
+ "step": 60
31
+ },
32
+ {
33
+ "epoch": 0.010348033469420752,
34
+ "grad_norm": 2.366281032562256,
35
+ "learning_rate": 3.4497628288055197e-07,
36
+ "loss": 3.6733,
37
+ "step": 80
38
+ },
39
+ {
40
+ "epoch": 0.012935041836775941,
41
+ "grad_norm": 1.8071706295013428,
42
+ "learning_rate": 4.3122035360069e-07,
43
+ "loss": 3.5801,
44
+ "step": 100
45
+ },
46
+ {
47
+ "epoch": 0.01552205020413113,
48
+ "grad_norm": 1.3455595970153809,
49
+ "learning_rate": 5.174644243208281e-07,
50
+ "loss": 3.4895,
51
+ "step": 120
52
+ },
53
+ {
54
+ "epoch": 0.018109058571486317,
55
+ "grad_norm": 0.9497399926185608,
56
+ "learning_rate": 6.03708495040966e-07,
57
+ "loss": 3.3785,
58
+ "step": 140
59
+ },
60
+ {
61
+ "epoch": 0.020696066938841504,
62
+ "grad_norm": 0.9015834331512451,
63
+ "learning_rate": 6.899525657611039e-07,
64
+ "loss": 3.3095,
65
+ "step": 160
66
+ },
67
+ {
68
+ "epoch": 0.023283075306196693,
69
+ "grad_norm": 0.8537250757217407,
70
+ "learning_rate": 7.761966364812419e-07,
71
+ "loss": 3.1901,
72
+ "step": 180
73
+ },
74
+ {
75
+ "epoch": 0.025870083673551883,
76
+ "grad_norm": 0.8126162886619568,
77
+ "learning_rate": 8.6244070720138e-07,
78
+ "loss": 3.1323,
79
+ "step": 200
80
+ },
81
+ {
82
+ "epoch": 0.02845709204090707,
83
+ "grad_norm": 0.8042150139808655,
84
+ "learning_rate": 9.48684777921518e-07,
85
+ "loss": 3.0969,
86
+ "step": 220
87
+ },
88
+ {
89
+ "epoch": 0.03104410040826226,
90
+ "grad_norm": 0.7180504202842712,
91
+ "learning_rate": 1.0349288486416561e-06,
92
+ "loss": 3.0284,
93
+ "step": 240
94
+ },
95
+ {
96
+ "epoch": 0.033631108775617445,
97
+ "grad_norm": 0.7540543675422668,
98
+ "learning_rate": 1.1211729193617941e-06,
99
+ "loss": 3.043,
100
+ "step": 260
101
+ },
102
+ {
103
+ "epoch": 0.036218117142972635,
104
+ "grad_norm": 0.720626175403595,
105
+ "learning_rate": 1.207416990081932e-06,
106
+ "loss": 3.002,
107
+ "step": 280
108
+ },
109
+ {
110
+ "epoch": 0.038805125510327824,
111
+ "grad_norm": 0.761584997177124,
112
+ "learning_rate": 1.29366106080207e-06,
113
+ "loss": 3.0019,
114
+ "step": 300
115
+ },
116
+ {
117
+ "epoch": 0.04139213387768301,
118
+ "grad_norm": 0.7120524644851685,
119
+ "learning_rate": 1.3799051315222079e-06,
120
+ "loss": 2.9614,
121
+ "step": 320
122
+ },
123
+ {
124
+ "epoch": 0.0439791422450382,
125
+ "grad_norm": 0.6938987374305725,
126
+ "learning_rate": 1.4661492022423459e-06,
127
+ "loss": 2.9435,
128
+ "step": 340
129
+ },
130
+ {
131
+ "epoch": 0.04656615061239339,
132
+ "grad_norm": 0.7532190680503845,
133
+ "learning_rate": 1.5523932729624839e-06,
134
+ "loss": 2.9194,
135
+ "step": 360
136
+ },
137
+ {
138
+ "epoch": 0.049153158979748576,
139
+ "grad_norm": 0.7144018411636353,
140
+ "learning_rate": 1.6386373436826219e-06,
141
+ "loss": 2.909,
142
+ "step": 380
143
+ },
144
+ {
145
+ "epoch": 0.051740167347103766,
146
+ "grad_norm": 0.7019342184066772,
147
+ "learning_rate": 1.72488141440276e-06,
148
+ "loss": 2.8771,
149
+ "step": 400
150
+ },
151
+ {
152
+ "epoch": 0.05432717571445895,
153
+ "grad_norm": 0.7003944516181946,
154
+ "learning_rate": 1.8111254851228978e-06,
155
+ "loss": 2.892,
156
+ "step": 420
157
+ },
158
+ {
159
+ "epoch": 0.05691418408181414,
160
+ "grad_norm": 0.7309603095054626,
161
+ "learning_rate": 1.897369555843036e-06,
162
+ "loss": 2.8603,
163
+ "step": 440
164
+ },
165
+ {
166
+ "epoch": 0.05950119244916933,
167
+ "grad_norm": 0.6825790405273438,
168
+ "learning_rate": 1.983613626563174e-06,
169
+ "loss": 2.8596,
170
+ "step": 460
171
+ },
172
+ {
173
+ "epoch": 0.06208820081652452,
174
+ "grad_norm": 0.6874191164970398,
175
+ "learning_rate": 2.0698576972833122e-06,
176
+ "loss": 2.8482,
177
+ "step": 480
178
+ },
179
+ {
180
+ "epoch": 0.0646752091838797,
181
+ "grad_norm": 0.7430766224861145,
182
+ "learning_rate": 2.15610176800345e-06,
183
+ "loss": 2.8446,
184
+ "step": 500
185
+ },
186
+ {
187
+ "epoch": 0.06726221755123489,
188
+ "grad_norm": 0.7108047008514404,
189
+ "learning_rate": 2.2423458387235882e-06,
190
+ "loss": 2.828,
191
+ "step": 520
192
+ },
193
+ {
194
+ "epoch": 0.06984922591859008,
195
+ "grad_norm": 0.7121191620826721,
196
+ "learning_rate": 2.328589909443726e-06,
197
+ "loss": 2.8002,
198
+ "step": 540
199
+ },
200
+ {
201
+ "epoch": 0.07243623428594527,
202
+ "grad_norm": 0.7068054676055908,
203
+ "learning_rate": 2.414833980163864e-06,
204
+ "loss": 2.8119,
205
+ "step": 560
206
+ },
207
+ {
208
+ "epoch": 0.07502324265330046,
209
+ "grad_norm": 0.6924973726272583,
210
+ "learning_rate": 2.501078050884002e-06,
211
+ "loss": 2.7846,
212
+ "step": 580
213
+ },
214
+ {
215
+ "epoch": 0.07761025102065565,
216
+ "grad_norm": 0.7206343412399292,
217
+ "learning_rate": 2.58732212160414e-06,
218
+ "loss": 2.7896,
219
+ "step": 600
220
+ },
221
+ {
222
+ "epoch": 0.08019725938801084,
223
+ "grad_norm": 0.7203959822654724,
224
+ "learning_rate": 2.673566192324278e-06,
225
+ "loss": 2.7793,
226
+ "step": 620
227
+ },
228
+ {
229
+ "epoch": 0.08278426775536601,
230
+ "grad_norm": 0.7108203172683716,
231
+ "learning_rate": 2.7598102630444157e-06,
232
+ "loss": 2.7642,
233
+ "step": 640
234
+ },
235
+ {
236
+ "epoch": 0.0853712761227212,
237
+ "grad_norm": 0.7053371667861938,
238
+ "learning_rate": 2.846054333764554e-06,
239
+ "loss": 2.7588,
240
+ "step": 660
241
+ },
242
+ {
243
+ "epoch": 0.0879582844900764,
244
+ "grad_norm": 0.6986986398696899,
245
+ "learning_rate": 2.9322984044846917e-06,
246
+ "loss": 2.7453,
247
+ "step": 680
248
+ },
249
+ {
250
+ "epoch": 0.09054529285743158,
251
+ "grad_norm": 0.7130378484725952,
252
+ "learning_rate": 3.01854247520483e-06,
253
+ "loss": 2.7249,
254
+ "step": 700
255
+ },
256
+ {
257
+ "epoch": 0.09313230122478677,
258
+ "grad_norm": 0.7068546414375305,
259
+ "learning_rate": 3.1047865459249677e-06,
260
+ "loss": 2.7134,
261
+ "step": 720
262
+ },
263
+ {
264
+ "epoch": 0.09571930959214196,
265
+ "grad_norm": 0.7224396467208862,
266
+ "learning_rate": 3.191030616645106e-06,
267
+ "loss": 2.718,
268
+ "step": 740
269
+ },
270
+ {
271
+ "epoch": 0.09830631795949715,
272
+ "grad_norm": 0.7299512624740601,
273
+ "learning_rate": 3.2772746873652437e-06,
274
+ "loss": 2.7049,
275
+ "step": 760
276
+ },
277
+ {
278
+ "epoch": 0.10089332632685234,
279
+ "grad_norm": 0.7126713991165161,
280
+ "learning_rate": 3.363518758085382e-06,
281
+ "loss": 2.7199,
282
+ "step": 780
283
+ },
284
+ {
285
+ "epoch": 0.10348033469420753,
286
+ "grad_norm": 0.6644526720046997,
287
+ "learning_rate": 3.44976282880552e-06,
288
+ "loss": 2.7046,
289
+ "step": 800
290
+ },
291
+ {
292
+ "epoch": 0.10606734306156272,
293
+ "grad_norm": 0.7410451769828796,
294
+ "learning_rate": 3.536006899525658e-06,
295
+ "loss": 2.7012,
296
+ "step": 820
297
+ },
298
+ {
299
+ "epoch": 0.1086543514289179,
300
+ "grad_norm": 0.7218043208122253,
301
+ "learning_rate": 3.6222509702457957e-06,
302
+ "loss": 2.71,
303
+ "step": 840
304
+ },
305
+ {
306
+ "epoch": 0.11124135979627309,
307
+ "grad_norm": 0.6941894292831421,
308
+ "learning_rate": 3.708495040965934e-06,
309
+ "loss": 2.7035,
310
+ "step": 860
311
+ },
312
+ {
313
+ "epoch": 0.11382836816362828,
314
+ "grad_norm": 0.6824482679367065,
315
+ "learning_rate": 3.794739111686072e-06,
316
+ "loss": 2.6741,
317
+ "step": 880
318
+ },
319
+ {
320
+ "epoch": 0.11641537653098347,
321
+ "grad_norm": 0.7253878712654114,
322
+ "learning_rate": 3.88098318240621e-06,
323
+ "loss": 2.6831,
324
+ "step": 900
325
+ },
326
+ {
327
+ "epoch": 0.11900238489833866,
328
+ "grad_norm": 0.6949134469032288,
329
+ "learning_rate": 3.967227253126348e-06,
330
+ "loss": 2.6674,
331
+ "step": 920
332
+ },
333
+ {
334
+ "epoch": 0.12158939326569385,
335
+ "grad_norm": 0.7096832394599915,
336
+ "learning_rate": 4.053471323846485e-06,
337
+ "loss": 2.674,
338
+ "step": 940
339
+ },
340
+ {
341
+ "epoch": 0.12417640163304904,
342
+ "grad_norm": 0.6938119530677795,
343
+ "learning_rate": 4.1397153945666245e-06,
344
+ "loss": 2.6533,
345
+ "step": 960
346
+ },
347
+ {
348
+ "epoch": 0.12676341000040423,
349
+ "grad_norm": 0.6781166195869446,
350
+ "learning_rate": 4.225959465286762e-06,
351
+ "loss": 2.6426,
352
+ "step": 980
353
+ },
354
+ {
355
+ "epoch": 0.1293504183677594,
356
+ "grad_norm": 0.7600920796394348,
357
+ "learning_rate": 4.3122035360069e-06,
358
+ "loss": 2.6454,
359
+ "step": 1000
360
+ },
361
+ {
362
+ "epoch": 0.1319374267351146,
363
+ "grad_norm": 0.6884315609931946,
364
+ "learning_rate": 4.398447606727037e-06,
365
+ "loss": 2.6324,
366
+ "step": 1020
367
+ },
368
+ {
369
+ "epoch": 0.13452443510246978,
370
+ "grad_norm": 0.7092069387435913,
371
+ "learning_rate": 4.4846916774471764e-06,
372
+ "loss": 2.6181,
373
+ "step": 1040
374
+ },
375
+ {
376
+ "epoch": 0.13711144346982498,
377
+ "grad_norm": 0.729020357131958,
378
+ "learning_rate": 4.570935748167314e-06,
379
+ "loss": 2.6217,
380
+ "step": 1060
381
+ },
382
+ {
383
+ "epoch": 0.13969845183718016,
384
+ "grad_norm": 0.7104108333587646,
385
+ "learning_rate": 4.657179818887452e-06,
386
+ "loss": 2.6152,
387
+ "step": 1080
388
+ },
389
+ {
390
+ "epoch": 0.14228546020453534,
391
+ "grad_norm": 0.8092931509017944,
392
+ "learning_rate": 4.743423889607589e-06,
393
+ "loss": 2.621,
394
+ "step": 1100
395
+ },
396
+ {
397
+ "epoch": 0.14487246857189054,
398
+ "grad_norm": 0.6620950102806091,
399
+ "learning_rate": 4.829667960327728e-06,
400
+ "loss": 2.6166,
401
+ "step": 1120
402
+ },
403
+ {
404
+ "epoch": 0.14745947693924571,
405
+ "grad_norm": 0.6813467741012573,
406
+ "learning_rate": 4.915912031047866e-06,
407
+ "loss": 2.6171,
408
+ "step": 1140
409
+ },
410
+ {
411
+ "epoch": 0.15004648530660092,
412
+ "grad_norm": 0.7140293121337891,
413
+ "learning_rate": 5.002156101768004e-06,
414
+ "loss": 2.5911,
415
+ "step": 1160
416
+ },
417
+ {
418
+ "epoch": 0.1526334936739561,
419
+ "grad_norm": 0.7278040051460266,
420
+ "learning_rate": 5.088400172488141e-06,
421
+ "loss": 2.6055,
422
+ "step": 1180
423
+ },
424
+ {
425
+ "epoch": 0.1552205020413113,
426
+ "grad_norm": 0.7283148169517517,
427
+ "learning_rate": 5.17464424320828e-06,
428
+ "loss": 2.6054,
429
+ "step": 1200
430
+ },
431
+ {
432
+ "epoch": 0.15780751040866647,
433
+ "grad_norm": 0.7109535932540894,
434
+ "learning_rate": 5.260888313928419e-06,
435
+ "loss": 2.56,
436
+ "step": 1220
437
+ },
438
+ {
439
+ "epoch": 0.16039451877602168,
440
+ "grad_norm": 0.7203260064125061,
441
+ "learning_rate": 5.347132384648556e-06,
442
+ "loss": 2.578,
443
+ "step": 1240
444
+ },
445
+ {
446
+ "epoch": 0.16298152714337685,
447
+ "grad_norm": 0.7385180592536926,
448
+ "learning_rate": 5.433376455368694e-06,
449
+ "loss": 2.5829,
450
+ "step": 1260
451
+ },
452
+ {
453
+ "epoch": 0.16556853551073203,
454
+ "grad_norm": 0.7511777281761169,
455
+ "learning_rate": 5.5196205260888315e-06,
456
+ "loss": 2.5703,
457
+ "step": 1280
458
+ },
459
+ {
460
+ "epoch": 0.16815554387808723,
461
+ "grad_norm": 0.7461130619049072,
462
+ "learning_rate": 5.60586459680897e-06,
463
+ "loss": 2.5891,
464
+ "step": 1300
465
+ },
466
+ {
467
+ "epoch": 0.1707425522454424,
468
+ "grad_norm": 0.7192751169204712,
469
+ "learning_rate": 5.692108667529108e-06,
470
+ "loss": 2.552,
471
+ "step": 1320
472
+ },
473
+ {
474
+ "epoch": 0.1733295606127976,
475
+ "grad_norm": 0.7672246694564819,
476
+ "learning_rate": 5.778352738249245e-06,
477
+ "loss": 2.5451,
478
+ "step": 1340
479
+ },
480
+ {
481
+ "epoch": 0.1759165689801528,
482
+ "grad_norm": 0.8286859393119812,
483
+ "learning_rate": 5.8645968089693835e-06,
484
+ "loss": 2.5691,
485
+ "step": 1360
486
+ },
487
+ {
488
+ "epoch": 0.178503577347508,
489
+ "grad_norm": 0.8903458714485168,
490
+ "learning_rate": 5.9508408796895225e-06,
491
+ "loss": 2.5717,
492
+ "step": 1380
493
+ },
494
+ {
495
+ "epoch": 0.18109058571486317,
496
+ "grad_norm": 0.7192072868347168,
497
+ "learning_rate": 6.03708495040966e-06,
498
+ "loss": 2.547,
499
+ "step": 1400
500
+ },
501
+ {
502
+ "epoch": 0.18367759408221837,
503
+ "grad_norm": 0.7470182776451111,
504
+ "learning_rate": 6.123329021129798e-06,
505
+ "loss": 2.5309,
506
+ "step": 1420
507
+ },
508
+ {
509
+ "epoch": 0.18626460244957355,
510
+ "grad_norm": 0.7186440229415894,
511
+ "learning_rate": 6.2095730918499354e-06,
512
+ "loss": 2.5433,
513
+ "step": 1440
514
+ },
515
+ {
516
+ "epoch": 0.18885161081692872,
517
+ "grad_norm": 0.7359221577644348,
518
+ "learning_rate": 6.295817162570074e-06,
519
+ "loss": 2.5482,
520
+ "step": 1460
521
+ },
522
+ {
523
+ "epoch": 0.19143861918428393,
524
+ "grad_norm": 0.7859694957733154,
525
+ "learning_rate": 6.382061233290212e-06,
526
+ "loss": 2.5371,
527
+ "step": 1480
528
+ },
529
+ {
530
+ "epoch": 0.1940256275516391,
531
+ "grad_norm": 0.7339861392974854,
532
+ "learning_rate": 6.468305304010349e-06,
533
+ "loss": 2.5217,
534
+ "step": 1500
535
+ },
536
+ {
537
+ "epoch": 0.1966126359189943,
538
+ "grad_norm": 0.7527260780334473,
539
+ "learning_rate": 6.554549374730487e-06,
540
+ "loss": 2.5183,
541
+ "step": 1520
542
+ },
543
+ {
544
+ "epoch": 0.19919964428634948,
545
+ "grad_norm": 0.7345518469810486,
546
+ "learning_rate": 6.6407934454506265e-06,
547
+ "loss": 2.5249,
548
+ "step": 1540
549
+ },
550
+ {
551
+ "epoch": 0.20178665265370468,
552
+ "grad_norm": 0.736298680305481,
553
+ "learning_rate": 6.727037516170764e-06,
554
+ "loss": 2.5184,
555
+ "step": 1560
556
+ },
557
+ {
558
+ "epoch": 0.20437366102105986,
559
+ "grad_norm": 0.7677698135375977,
560
+ "learning_rate": 6.813281586890902e-06,
561
+ "loss": 2.5134,
562
+ "step": 1580
563
+ },
564
+ {
565
+ "epoch": 0.20696066938841506,
566
+ "grad_norm": 0.7627900838851929,
567
+ "learning_rate": 6.89952565761104e-06,
568
+ "loss": 2.5064,
569
+ "step": 1600
570
+ },
571
+ {
572
+ "epoch": 0.20954767775577024,
573
+ "grad_norm": 0.7290985584259033,
574
+ "learning_rate": 6.985769728331178e-06,
575
+ "loss": 2.5124,
576
+ "step": 1620
577
+ },
578
+ {
579
+ "epoch": 0.21213468612312544,
580
+ "grad_norm": 0.7347148060798645,
581
+ "learning_rate": 7.072013799051316e-06,
582
+ "loss": 2.4901,
583
+ "step": 1640
584
+ },
585
+ {
586
+ "epoch": 0.21472169449048062,
587
+ "grad_norm": 0.7257357239723206,
588
+ "learning_rate": 7.158257869771453e-06,
589
+ "loss": 2.4884,
590
+ "step": 1660
591
+ },
592
+ {
593
+ "epoch": 0.2173087028578358,
594
+ "grad_norm": 0.7803710699081421,
595
+ "learning_rate": 7.244501940491591e-06,
596
+ "loss": 2.4899,
597
+ "step": 1680
598
+ },
599
+ {
600
+ "epoch": 0.219895711225191,
601
+ "grad_norm": 0.6987377405166626,
602
+ "learning_rate": 7.33074601121173e-06,
603
+ "loss": 2.5001,
604
+ "step": 1700
605
+ },
606
+ {
607
+ "epoch": 0.22248271959254617,
608
+ "grad_norm": 0.6989637017250061,
609
+ "learning_rate": 7.416990081931868e-06,
610
+ "loss": 2.4941,
611
+ "step": 1720
612
+ },
613
+ {
614
+ "epoch": 0.22506972795990138,
615
+ "grad_norm": 0.7391577363014221,
616
+ "learning_rate": 7.503234152652006e-06,
617
+ "loss": 2.4811,
618
+ "step": 1740
619
+ },
620
+ {
621
+ "epoch": 0.22765673632725655,
622
+ "grad_norm": 0.7664337754249573,
623
+ "learning_rate": 7.589478223372144e-06,
624
+ "loss": 2.4806,
625
+ "step": 1760
626
+ },
627
+ {
628
+ "epoch": 0.23024374469461176,
629
+ "grad_norm": 0.7150381207466125,
630
+ "learning_rate": 7.675722294092282e-06,
631
+ "loss": 2.4674,
632
+ "step": 1780
633
+ },
634
+ {
635
+ "epoch": 0.23283075306196693,
636
+ "grad_norm": 0.7978541254997253,
637
+ "learning_rate": 7.76196636481242e-06,
638
+ "loss": 2.4633,
639
+ "step": 1800
640
+ },
641
+ {
642
+ "epoch": 0.23541776142932214,
643
+ "grad_norm": 0.7218653559684753,
644
+ "learning_rate": 7.848210435532557e-06,
645
+ "loss": 2.4589,
646
+ "step": 1820
647
+ },
648
+ {
649
+ "epoch": 0.2380047697966773,
650
+ "grad_norm": 0.723008930683136,
651
+ "learning_rate": 7.934454506252696e-06,
652
+ "loss": 2.467,
653
+ "step": 1840
654
+ },
655
+ {
656
+ "epoch": 0.2405917781640325,
657
+ "grad_norm": 0.7375757098197937,
658
+ "learning_rate": 8.020698576972833e-06,
659
+ "loss": 2.4812,
660
+ "step": 1860
661
+ },
662
+ {
663
+ "epoch": 0.2431787865313877,
664
+ "grad_norm": 0.7501986026763916,
665
+ "learning_rate": 8.10694264769297e-06,
666
+ "loss": 2.4587,
667
+ "step": 1880
668
+ },
669
+ {
670
+ "epoch": 0.24576579489874287,
671
+ "grad_norm": 0.7394606471061707,
672
+ "learning_rate": 8.19318671841311e-06,
673
+ "loss": 2.4514,
674
+ "step": 1900
675
+ },
676
+ {
677
+ "epoch": 0.24835280326609807,
678
+ "grad_norm": 0.7856109738349915,
679
+ "learning_rate": 8.279430789133249e-06,
680
+ "loss": 2.45,
681
+ "step": 1920
682
+ },
683
+ {
684
+ "epoch": 0.25093981163345325,
685
+ "grad_norm": 0.7603466510772705,
686
+ "learning_rate": 8.365674859853386e-06,
687
+ "loss": 2.4341,
688
+ "step": 1940
689
+ },
690
+ {
691
+ "epoch": 0.25352682000080845,
692
+ "grad_norm": 0.7223484516143799,
693
+ "learning_rate": 8.451918930573524e-06,
694
+ "loss": 2.4349,
695
+ "step": 1960
696
+ },
697
+ {
698
+ "epoch": 0.25611382836816365,
699
+ "grad_norm": 0.7488518357276917,
700
+ "learning_rate": 8.538163001293663e-06,
701
+ "loss": 2.4417,
702
+ "step": 1980
703
+ },
704
+ {
705
+ "epoch": 0.2587008367355188,
706
+ "grad_norm": 0.83389812707901,
707
+ "learning_rate": 8.6244070720138e-06,
708
+ "loss": 2.4266,
709
+ "step": 2000
710
+ },
711
+ {
712
+ "epoch": 0.261287845102874,
713
+ "grad_norm": 0.7574110627174377,
714
+ "learning_rate": 8.710651142733937e-06,
715
+ "loss": 2.4471,
716
+ "step": 2020
717
+ },
718
+ {
719
+ "epoch": 0.2638748534702292,
720
+ "grad_norm": 0.7482550740242004,
721
+ "learning_rate": 8.796895213454075e-06,
722
+ "loss": 2.4368,
723
+ "step": 2040
724
+ },
725
+ {
726
+ "epoch": 0.26646186183758436,
727
+ "grad_norm": 0.7341257333755493,
728
+ "learning_rate": 8.883139284174214e-06,
729
+ "loss": 2.4303,
730
+ "step": 2060
731
+ },
732
+ {
733
+ "epoch": 0.26904887020493956,
734
+ "grad_norm": 0.764855682849884,
735
+ "learning_rate": 8.969383354894353e-06,
736
+ "loss": 2.4176,
737
+ "step": 2080
738
+ },
739
+ {
740
+ "epoch": 0.27163587857229476,
741
+ "grad_norm": 0.7559799551963806,
742
+ "learning_rate": 9.055627425614489e-06,
743
+ "loss": 2.4088,
744
+ "step": 2100
745
+ },
746
+ {
747
+ "epoch": 0.27422288693964997,
748
+ "grad_norm": 0.7435436844825745,
749
+ "learning_rate": 9.141871496334628e-06,
750
+ "loss": 2.4241,
751
+ "step": 2120
752
+ },
753
+ {
754
+ "epoch": 0.2768098953070051,
755
+ "grad_norm": 0.7100109457969666,
756
+ "learning_rate": 9.228115567054767e-06,
757
+ "loss": 2.4122,
758
+ "step": 2140
759
+ },
760
+ {
761
+ "epoch": 0.2793969036743603,
762
+ "grad_norm": 0.7357873320579529,
763
+ "learning_rate": 9.314359637774904e-06,
764
+ "loss": 2.425,
765
+ "step": 2160
766
+ },
767
+ {
768
+ "epoch": 0.2819839120417155,
769
+ "grad_norm": 0.7281599640846252,
770
+ "learning_rate": 9.400603708495041e-06,
771
+ "loss": 2.4242,
772
+ "step": 2180
773
+ },
774
+ {
775
+ "epoch": 0.28457092040907067,
776
+ "grad_norm": 0.7816882729530334,
777
+ "learning_rate": 9.486847779215179e-06,
778
+ "loss": 2.4115,
779
+ "step": 2200
780
+ },
781
+ {
782
+ "epoch": 0.2871579287764259,
783
+ "grad_norm": 0.7402174472808838,
784
+ "learning_rate": 9.573091849935318e-06,
785
+ "loss": 2.4253,
786
+ "step": 2220
787
+ },
788
+ {
789
+ "epoch": 0.2897449371437811,
790
+ "grad_norm": 0.7378434538841248,
791
+ "learning_rate": 9.659335920655457e-06,
792
+ "loss": 2.4015,
793
+ "step": 2240
794
+ },
795
+ {
796
+ "epoch": 0.2923319455111363,
797
+ "grad_norm": 0.7792133688926697,
798
+ "learning_rate": 9.745579991375592e-06,
799
+ "loss": 2.3947,
800
+ "step": 2260
801
+ },
802
+ {
803
+ "epoch": 0.29491895387849143,
804
+ "grad_norm": 0.7180085778236389,
805
+ "learning_rate": 9.831824062095732e-06,
806
+ "loss": 2.3975,
807
+ "step": 2280
808
+ },
809
+ {
810
+ "epoch": 0.29750596224584663,
811
+ "grad_norm": 0.7536414861679077,
812
+ "learning_rate": 9.91806813281587e-06,
813
+ "loss": 2.4052,
814
+ "step": 2300
815
+ },
816
+ {
817
+ "epoch": 0.30009297061320184,
818
+ "grad_norm": 0.7593511939048767,
819
+ "learning_rate": 9.999999943356089e-06,
820
+ "loss": 2.3856,
821
+ "step": 2320
822
+ },
823
+ {
824
+ "epoch": 0.30267997898055704,
825
+ "grad_norm": 0.6879323720932007,
826
+ "learning_rate": 9.999975020055695e-06,
827
+ "loss": 2.4107,
828
+ "step": 2340
829
+ },
830
+ {
831
+ "epoch": 0.3052669873479122,
832
+ "grad_norm": 0.7193155884742737,
833
+ "learning_rate": 9.999904781886476e-06,
834
+ "loss": 2.3872,
835
+ "step": 2360
836
+ },
837
+ {
838
+ "epoch": 0.3078539957152674,
839
+ "grad_norm": 0.7621170282363892,
840
+ "learning_rate": 9.999789229485002e-06,
841
+ "loss": 2.3844,
842
+ "step": 2380
843
+ },
844
+ {
845
+ "epoch": 0.3104410040826226,
846
+ "grad_norm": 0.7357332110404968,
847
+ "learning_rate": 9.999628363898525e-06,
848
+ "loss": 2.3863,
849
+ "step": 2400
850
+ },
851
+ {
852
+ "epoch": 0.31302801244997774,
853
+ "grad_norm": 0.753011167049408,
854
+ "learning_rate": 9.999422186584978e-06,
855
+ "loss": 2.3807,
856
+ "step": 2420
857
+ },
858
+ {
859
+ "epoch": 0.31561502081733295,
860
+ "grad_norm": 0.7547248005867004,
861
+ "learning_rate": 9.999170699412942e-06,
862
+ "loss": 2.3691,
863
+ "step": 2440
864
+ },
865
+ {
866
+ "epoch": 0.31820202918468815,
867
+ "grad_norm": 0.7450150847434998,
868
+ "learning_rate": 9.998873904661655e-06,
869
+ "loss": 2.3495,
870
+ "step": 2460
871
+ },
872
+ {
873
+ "epoch": 0.32078903755204335,
874
+ "grad_norm": 0.7460102438926697,
875
+ "learning_rate": 9.998531805020974e-06,
876
+ "loss": 2.3582,
877
+ "step": 2480
878
+ },
879
+ {
880
+ "epoch": 0.3233760459193985,
881
+ "grad_norm": 0.7281093597412109,
882
+ "learning_rate": 9.998144403591352e-06,
883
+ "loss": 2.3559,
884
+ "step": 2500
885
+ },
886
+ {
887
+ "epoch": 0.3259630542867537,
888
+ "grad_norm": 0.7635049223899841,
889
+ "learning_rate": 9.99771170388382e-06,
890
+ "loss": 2.3668,
891
+ "step": 2520
892
+ },
893
+ {
894
+ "epoch": 0.3285500626541089,
895
+ "grad_norm": 0.7504300475120544,
896
+ "learning_rate": 9.997233709819935e-06,
897
+ "loss": 2.3642,
898
+ "step": 2540
899
+ },
900
+ {
901
+ "epoch": 0.33113707102146406,
902
+ "grad_norm": 0.73244309425354,
903
+ "learning_rate": 9.996710425731776e-06,
904
+ "loss": 2.3634,
905
+ "step": 2560
906
+ },
907
+ {
908
+ "epoch": 0.33372407938881926,
909
+ "grad_norm": 0.816100537776947,
910
+ "learning_rate": 9.996141856361871e-06,
911
+ "loss": 2.3405,
912
+ "step": 2580
913
+ },
914
+ {
915
+ "epoch": 0.33631108775617446,
916
+ "grad_norm": 0.7125614285469055,
917
+ "learning_rate": 9.995528006863175e-06,
918
+ "loss": 2.3594,
919
+ "step": 2600
920
+ },
921
+ {
922
+ "epoch": 0.33889809612352967,
923
+ "grad_norm": 0.7901700735092163,
924
+ "learning_rate": 9.994868882799022e-06,
925
+ "loss": 2.3636,
926
+ "step": 2620
927
+ },
928
+ {
929
+ "epoch": 0.3414851044908848,
930
+ "grad_norm": 0.7499126195907593,
931
+ "learning_rate": 9.994164490143062e-06,
932
+ "loss": 2.36,
933
+ "step": 2640
934
+ },
935
+ {
936
+ "epoch": 0.34407211285824,
937
+ "grad_norm": 0.78279048204422,
938
+ "learning_rate": 9.99341483527922e-06,
939
+ "loss": 2.3685,
940
+ "step": 2660
941
+ },
942
+ {
943
+ "epoch": 0.3466591212255952,
944
+ "grad_norm": 0.7683995366096497,
945
+ "learning_rate": 9.992619925001632e-06,
946
+ "loss": 2.3532,
947
+ "step": 2680
948
+ },
949
+ {
950
+ "epoch": 0.3492461295929504,
951
+ "grad_norm": 0.7231781482696533,
952
+ "learning_rate": 9.991779766514586e-06,
953
+ "loss": 2.377,
954
+ "step": 2700
955
+ },
956
+ {
957
+ "epoch": 0.3518331379603056,
958
+ "grad_norm": 0.7408224940299988,
959
+ "learning_rate": 9.990894367432453e-06,
960
+ "loss": 2.3406,
961
+ "step": 2720
962
+ },
963
+ {
964
+ "epoch": 0.3544201463276608,
965
+ "grad_norm": 0.7654145956039429,
966
+ "learning_rate": 9.989963735779623e-06,
967
+ "loss": 2.3627,
968
+ "step": 2740
969
+ },
970
+ {
971
+ "epoch": 0.357007154695016,
972
+ "grad_norm": 0.7166858315467834,
973
+ "learning_rate": 9.988987879990428e-06,
974
+ "loss": 2.3343,
975
+ "step": 2760
976
+ },
977
+ {
978
+ "epoch": 0.35959416306237113,
979
+ "grad_norm": 0.79310142993927,
980
+ "learning_rate": 9.987966808909069e-06,
981
+ "loss": 2.3521,
982
+ "step": 2780
983
+ },
984
+ {
985
+ "epoch": 0.36218117142972633,
986
+ "grad_norm": 0.7726064324378967,
987
+ "learning_rate": 9.98690053178953e-06,
988
+ "loss": 2.3408,
989
+ "step": 2800
990
+ },
991
+ {
992
+ "epoch": 0.36476817979708154,
993
+ "grad_norm": 0.7428059577941895,
994
+ "learning_rate": 9.985789058295501e-06,
995
+ "loss": 2.3238,
996
+ "step": 2820
997
+ },
998
+ {
999
+ "epoch": 0.36735518816443674,
1000
+ "grad_norm": 0.7808260321617126,
1001
+ "learning_rate": 9.984632398500289e-06,
1002
+ "loss": 2.3164,
1003
+ "step": 2840
1004
+ },
1005
+ {
1006
+ "epoch": 0.3699421965317919,
1007
+ "grad_norm": 0.7835705280303955,
1008
+ "learning_rate": 9.983430562886723e-06,
1009
+ "loss": 2.343,
1010
+ "step": 2860
1011
+ },
1012
+ {
1013
+ "epoch": 0.3725292048991471,
1014
+ "grad_norm": 0.7703275680541992,
1015
+ "learning_rate": 9.982183562347063e-06,
1016
+ "loss": 2.3354,
1017
+ "step": 2880
1018
+ },
1019
+ {
1020
+ "epoch": 0.3751162132665023,
1021
+ "grad_norm": 0.7307262420654297,
1022
+ "learning_rate": 9.980891408182897e-06,
1023
+ "loss": 2.3445,
1024
+ "step": 2900
1025
+ },
1026
+ {
1027
+ "epoch": 0.37770322163385744,
1028
+ "grad_norm": 0.6830443143844604,
1029
+ "learning_rate": 9.979554112105045e-06,
1030
+ "loss": 2.3437,
1031
+ "step": 2920
1032
+ },
1033
+ {
1034
+ "epoch": 0.38029023000121265,
1035
+ "grad_norm": 0.7646154761314392,
1036
+ "learning_rate": 9.978171686233445e-06,
1037
+ "loss": 2.3316,
1038
+ "step": 2940
1039
+ },
1040
+ {
1041
+ "epoch": 0.38287723836856785,
1042
+ "grad_norm": 0.7598294615745544,
1043
+ "learning_rate": 9.97674414309705e-06,
1044
+ "loss": 2.3135,
1045
+ "step": 2960
1046
+ },
1047
+ {
1048
+ "epoch": 0.38546424673592306,
1049
+ "grad_norm": 0.7368634343147278,
1050
+ "learning_rate": 9.975271495633709e-06,
1051
+ "loss": 2.3189,
1052
+ "step": 2980
1053
+ },
1054
+ {
1055
+ "epoch": 0.3880512551032782,
1056
+ "grad_norm": 0.7349269390106201,
1057
+ "learning_rate": 9.973753757190057e-06,
1058
+ "loss": 2.307,
1059
+ "step": 3000
1060
+ },
1061
+ {
1062
+ "epoch": 0.3906382634706334,
1063
+ "grad_norm": 0.7512723803520203,
1064
+ "learning_rate": 9.972190941521382e-06,
1065
+ "loss": 2.3195,
1066
+ "step": 3020
1067
+ },
1068
+ {
1069
+ "epoch": 0.3932252718379886,
1070
+ "grad_norm": 0.7430227398872375,
1071
+ "learning_rate": 9.970583062791517e-06,
1072
+ "loss": 2.3087,
1073
+ "step": 3040
1074
+ },
1075
+ {
1076
+ "epoch": 0.3958122802053438,
1077
+ "grad_norm": 0.7576317191123962,
1078
+ "learning_rate": 9.968930135572694e-06,
1079
+ "loss": 2.3,
1080
+ "step": 3060
1081
+ },
1082
+ {
1083
+ "epoch": 0.39839928857269896,
1084
+ "grad_norm": 0.7394402623176575,
1085
+ "learning_rate": 9.967232174845426e-06,
1086
+ "loss": 2.3164,
1087
+ "step": 3080
1088
+ },
1089
+ {
1090
+ "epoch": 0.40098629694005417,
1091
+ "grad_norm": 0.7061188220977783,
1092
+ "learning_rate": 9.965489195998363e-06,
1093
+ "loss": 2.3187,
1094
+ "step": 3100
1095
+ },
1096
+ {
1097
+ "epoch": 0.40357330530740937,
1098
+ "grad_norm": 0.7332024574279785,
1099
+ "learning_rate": 9.963701214828154e-06,
1100
+ "loss": 2.3085,
1101
+ "step": 3120
1102
+ },
1103
+ {
1104
+ "epoch": 0.4061603136747645,
1105
+ "grad_norm": 0.663925051689148,
1106
+ "learning_rate": 9.961868247539308e-06,
1107
+ "loss": 2.2978,
1108
+ "step": 3140
1109
+ },
1110
+ {
1111
+ "epoch": 0.4087473220421197,
1112
+ "grad_norm": 0.7256314158439636,
1113
+ "learning_rate": 9.959990310744042e-06,
1114
+ "loss": 2.3209,
1115
+ "step": 3160
1116
+ },
1117
+ {
1118
+ "epoch": 0.4113343304094749,
1119
+ "grad_norm": 0.711919903755188,
1120
+ "learning_rate": 9.958067421462133e-06,
1121
+ "loss": 2.3041,
1122
+ "step": 3180
1123
+ },
1124
+ {
1125
+ "epoch": 0.41392133877683013,
1126
+ "grad_norm": 0.714450478553772,
1127
+ "learning_rate": 9.956099597120762e-06,
1128
+ "loss": 2.2789,
1129
+ "step": 3200
1130
+ },
1131
+ {
1132
+ "epoch": 0.4165083471441853,
1133
+ "grad_norm": 0.7216628193855286,
1134
+ "learning_rate": 9.95408685555436e-06,
1135
+ "loss": 2.3056,
1136
+ "step": 3220
1137
+ },
1138
+ {
1139
+ "epoch": 0.4190953555115405,
1140
+ "grad_norm": 0.6629658341407776,
1141
+ "learning_rate": 9.952029215004441e-06,
1142
+ "loss": 2.3001,
1143
+ "step": 3240
1144
+ },
1145
+ {
1146
+ "epoch": 0.4216823638788957,
1147
+ "grad_norm": 0.8131959438323975,
1148
+ "learning_rate": 9.949926694119443e-06,
1149
+ "loss": 2.2881,
1150
+ "step": 3260
1151
+ },
1152
+ {
1153
+ "epoch": 0.4242693722462509,
1154
+ "grad_norm": 0.7133468389511108,
1155
+ "learning_rate": 9.94777931195455e-06,
1156
+ "loss": 2.2902,
1157
+ "step": 3280
1158
+ },
1159
+ {
1160
+ "epoch": 0.42685638061360603,
1161
+ "grad_norm": 0.7825130224227905,
1162
+ "learning_rate": 9.945587087971529e-06,
1163
+ "loss": 2.2829,
1164
+ "step": 3300
1165
+ },
1166
+ {
1167
+ "epoch": 0.42944338898096124,
1168
+ "grad_norm": 0.6878073811531067,
1169
+ "learning_rate": 9.943350042038545e-06,
1170
+ "loss": 2.2844,
1171
+ "step": 3320
1172
+ },
1173
+ {
1174
+ "epoch": 0.43203039734831644,
1175
+ "grad_norm": 0.7324455976486206,
1176
+ "learning_rate": 9.941068194429992e-06,
1177
+ "loss": 2.2932,
1178
+ "step": 3340
1179
+ },
1180
+ {
1181
+ "epoch": 0.4346174057156716,
1182
+ "grad_norm": 0.7330353856086731,
1183
+ "learning_rate": 9.938741565826295e-06,
1184
+ "loss": 2.2611,
1185
+ "step": 3360
1186
+ },
1187
+ {
1188
+ "epoch": 0.4372044140830268,
1189
+ "grad_norm": 0.8474377393722534,
1190
+ "learning_rate": 9.936370177313737e-06,
1191
+ "loss": 2.2938,
1192
+ "step": 3380
1193
+ },
1194
+ {
1195
+ "epoch": 0.439791422450382,
1196
+ "grad_norm": 0.776228666305542,
1197
+ "learning_rate": 9.933954050384253e-06,
1198
+ "loss": 2.292,
1199
+ "step": 3400
1200
+ },
1201
+ {
1202
+ "epoch": 0.4423784308177372,
1203
+ "grad_norm": 0.7257934212684631,
1204
+ "learning_rate": 9.931493206935249e-06,
1205
+ "loss": 2.2964,
1206
+ "step": 3420
1207
+ },
1208
+ {
1209
+ "epoch": 0.44496543918509235,
1210
+ "grad_norm": 0.7738561034202576,
1211
+ "learning_rate": 9.928987669269397e-06,
1212
+ "loss": 2.2585,
1213
+ "step": 3440
1214
+ },
1215
+ {
1216
+ "epoch": 0.44755244755244755,
1217
+ "grad_norm": 0.6888708472251892,
1218
+ "learning_rate": 9.926437460094431e-06,
1219
+ "loss": 2.281,
1220
+ "step": 3460
1221
+ },
1222
+ {
1223
+ "epoch": 0.45013945591980276,
1224
+ "grad_norm": 0.7120715975761414,
1225
+ "learning_rate": 9.923842602522948e-06,
1226
+ "loss": 2.2861,
1227
+ "step": 3480
1228
+ },
1229
+ {
1230
+ "epoch": 0.4527264642871579,
1231
+ "grad_norm": 0.770353376865387,
1232
+ "learning_rate": 9.92120312007219e-06,
1233
+ "loss": 2.271,
1234
+ "step": 3500
1235
+ },
1236
+ {
1237
+ "epoch": 0.4553134726545131,
1238
+ "grad_norm": 0.753852367401123,
1239
+ "learning_rate": 9.918519036663835e-06,
1240
+ "loss": 2.2592,
1241
+ "step": 3520
1242
+ },
1243
+ {
1244
+ "epoch": 0.4579004810218683,
1245
+ "grad_norm": 0.7479904294013977,
1246
+ "learning_rate": 9.915790376623785e-06,
1247
+ "loss": 2.2691,
1248
+ "step": 3540
1249
+ },
1250
+ {
1251
+ "epoch": 0.4604874893892235,
1252
+ "grad_norm": 0.7162041068077087,
1253
+ "learning_rate": 9.913017164681936e-06,
1254
+ "loss": 2.2637,
1255
+ "step": 3560
1256
+ },
1257
+ {
1258
+ "epoch": 0.46307449775657866,
1259
+ "grad_norm": 0.6859399080276489,
1260
+ "learning_rate": 9.91019942597196e-06,
1261
+ "loss": 2.2424,
1262
+ "step": 3580
1263
+ },
1264
+ {
1265
+ "epoch": 0.46566150612393387,
1266
+ "grad_norm": 0.7681860327720642,
1267
+ "learning_rate": 9.907337186031078e-06,
1268
+ "loss": 2.2758,
1269
+ "step": 3600
1270
+ },
1271
+ {
1272
+ "epoch": 0.46824851449128907,
1273
+ "grad_norm": 0.7416828274726868,
1274
+ "learning_rate": 9.904430470799826e-06,
1275
+ "loss": 2.2715,
1276
+ "step": 3620
1277
+ },
1278
+ {
1279
+ "epoch": 0.4708355228586443,
1280
+ "grad_norm": 0.7957201600074768,
1281
+ "learning_rate": 9.901479306621818e-06,
1282
+ "loss": 2.277,
1283
+ "step": 3640
1284
+ },
1285
+ {
1286
+ "epoch": 0.4734225312259994,
1287
+ "grad_norm": 0.71152663230896,
1288
+ "learning_rate": 9.89848372024351e-06,
1289
+ "loss": 2.2831,
1290
+ "step": 3660
1291
+ },
1292
+ {
1293
+ "epoch": 0.4760095395933546,
1294
+ "grad_norm": 0.6950436234474182,
1295
+ "learning_rate": 9.89544373881396e-06,
1296
+ "loss": 2.2678,
1297
+ "step": 3680
1298
+ },
1299
+ {
1300
+ "epoch": 0.47859654796070983,
1301
+ "grad_norm": 0.7226356267929077,
1302
+ "learning_rate": 9.89235938988458e-06,
1303
+ "loss": 2.2549,
1304
+ "step": 3700
1305
+ },
1306
+ {
1307
+ "epoch": 0.481183556328065,
1308
+ "grad_norm": 0.7032837867736816,
1309
+ "learning_rate": 9.88923070140888e-06,
1310
+ "loss": 2.2669,
1311
+ "step": 3720
1312
+ },
1313
+ {
1314
+ "epoch": 0.4837705646954202,
1315
+ "grad_norm": 0.749229371547699,
1316
+ "learning_rate": 9.886057701742222e-06,
1317
+ "loss": 2.2421,
1318
+ "step": 3740
1319
+ },
1320
+ {
1321
+ "epoch": 0.4863575730627754,
1322
+ "grad_norm": 0.7166919112205505,
1323
+ "learning_rate": 9.882840419641566e-06,
1324
+ "loss": 2.245,
1325
+ "step": 3760
1326
+ },
1327
+ {
1328
+ "epoch": 0.4889445814301306,
1329
+ "grad_norm": 0.7276502847671509,
1330
+ "learning_rate": 9.879578884265198e-06,
1331
+ "loss": 2.2542,
1332
+ "step": 3780
1333
+ },
1334
+ {
1335
+ "epoch": 0.49153158979748574,
1336
+ "grad_norm": 0.7321860194206238,
1337
+ "learning_rate": 9.876273125172476e-06,
1338
+ "loss": 2.2395,
1339
+ "step": 3800
1340
+ },
1341
+ {
1342
+ "epoch": 0.49411859816484094,
1343
+ "grad_norm": 0.6959764361381531,
1344
+ "learning_rate": 9.872923172323559e-06,
1345
+ "loss": 2.2588,
1346
+ "step": 3820
1347
+ },
1348
+ {
1349
+ "epoch": 0.49670560653219614,
1350
+ "grad_norm": 0.7237563729286194,
1351
+ "learning_rate": 9.869529056079133e-06,
1352
+ "loss": 2.2463,
1353
+ "step": 3840
1354
+ },
1355
+ {
1356
+ "epoch": 0.4992926148995513,
1357
+ "grad_norm": 0.7536144256591797,
1358
+ "learning_rate": 9.866090807200135e-06,
1359
+ "loss": 2.2394,
1360
+ "step": 3860
1361
+ },
1362
+ {
1363
+ "epoch": 0.5018796232669065,
1364
+ "grad_norm": 0.6835715770721436,
1365
+ "learning_rate": 9.862608456847484e-06,
1366
+ "loss": 2.2447,
1367
+ "step": 3880
1368
+ },
1369
+ {
1370
+ "epoch": 0.5044666316342616,
1371
+ "grad_norm": 0.7627236247062683,
1372
+ "learning_rate": 9.859082036581787e-06,
1373
+ "loss": 2.2727,
1374
+ "step": 3900
1375
+ },
1376
+ {
1377
+ "epoch": 0.5070536400016169,
1378
+ "grad_norm": 0.7260853052139282,
1379
+ "learning_rate": 9.855511578363057e-06,
1380
+ "loss": 2.2373,
1381
+ "step": 3920
1382
+ },
1383
+ {
1384
+ "epoch": 0.509640648368972,
1385
+ "grad_norm": 0.7366577386856079,
1386
+ "learning_rate": 9.851897114550423e-06,
1387
+ "loss": 2.2583,
1388
+ "step": 3940
1389
+ },
1390
+ {
1391
+ "epoch": 0.5122276567363273,
1392
+ "grad_norm": 0.718189001083374,
1393
+ "learning_rate": 9.848238677901844e-06,
1394
+ "loss": 2.2376,
1395
+ "step": 3960
1396
+ },
1397
+ {
1398
+ "epoch": 0.5148146651036825,
1399
+ "grad_norm": 0.6988586783409119,
1400
+ "learning_rate": 9.844536301573798e-06,
1401
+ "loss": 2.2413,
1402
+ "step": 3980
1403
+ },
1404
+ {
1405
+ "epoch": 0.5174016734710376,
1406
+ "grad_norm": 0.752648115158081,
1407
+ "learning_rate": 9.840790019120993e-06,
1408
+ "loss": 2.2346,
1409
+ "step": 4000
1410
+ },
1411
+ {
1412
+ "epoch": 0.5199886818383929,
1413
+ "grad_norm": 0.8000548481941223,
1414
+ "learning_rate": 9.836999864496058e-06,
1415
+ "loss": 2.2365,
1416
+ "step": 4020
1417
+ },
1418
+ {
1419
+ "epoch": 0.522575690205748,
1420
+ "grad_norm": 0.7324941158294678,
1421
+ "learning_rate": 9.833165872049235e-06,
1422
+ "loss": 2.2294,
1423
+ "step": 4040
1424
+ },
1425
+ {
1426
+ "epoch": 0.5251626985731032,
1427
+ "grad_norm": 0.765869140625,
1428
+ "learning_rate": 9.829288076528071e-06,
1429
+ "loss": 2.2418,
1430
+ "step": 4060
1431
+ },
1432
+ {
1433
+ "epoch": 0.5277497069404584,
1434
+ "grad_norm": 0.7222108244895935,
1435
+ "learning_rate": 9.825366513077104e-06,
1436
+ "loss": 2.2292,
1437
+ "step": 4080
1438
+ },
1439
+ {
1440
+ "epoch": 0.5303367153078136,
1441
+ "grad_norm": 0.7359380722045898,
1442
+ "learning_rate": 9.821401217237535e-06,
1443
+ "loss": 2.2491,
1444
+ "step": 4100
1445
+ },
1446
+ {
1447
+ "epoch": 0.5329237236751687,
1448
+ "grad_norm": 0.7288945317268372,
1449
+ "learning_rate": 9.817392224946916e-06,
1450
+ "loss": 2.2404,
1451
+ "step": 4120
1452
+ },
1453
+ {
1454
+ "epoch": 0.535510732042524,
1455
+ "grad_norm": 0.7518214583396912,
1456
+ "learning_rate": 9.813339572538822e-06,
1457
+ "loss": 2.2381,
1458
+ "step": 4140
1459
+ },
1460
+ {
1461
+ "epoch": 0.5380977404098791,
1462
+ "grad_norm": 0.753818154335022,
1463
+ "learning_rate": 9.809243296742516e-06,
1464
+ "loss": 2.2236,
1465
+ "step": 4160
1466
+ },
1467
+ {
1468
+ "epoch": 0.5406847487772344,
1469
+ "grad_norm": 0.7281918525695801,
1470
+ "learning_rate": 9.805103434682628e-06,
1471
+ "loss": 2.2142,
1472
+ "step": 4180
1473
+ },
1474
+ {
1475
+ "epoch": 0.5432717571445895,
1476
+ "grad_norm": 0.7605194449424744,
1477
+ "learning_rate": 9.800920023878803e-06,
1478
+ "loss": 2.2326,
1479
+ "step": 4200
1480
+ },
1481
+ {
1482
+ "epoch": 0.5458587655119447,
1483
+ "grad_norm": 0.7266237139701843,
1484
+ "learning_rate": 9.796693102245376e-06,
1485
+ "loss": 2.2144,
1486
+ "step": 4220
1487
+ },
1488
+ {
1489
+ "epoch": 0.5484457738792999,
1490
+ "grad_norm": 0.7575150728225708,
1491
+ "learning_rate": 9.792422708091014e-06,
1492
+ "loss": 2.2282,
1493
+ "step": 4240
1494
+ },
1495
+ {
1496
+ "epoch": 0.5510327822466551,
1497
+ "grad_norm": 0.7097237706184387,
1498
+ "learning_rate": 9.788108880118383e-06,
1499
+ "loss": 2.2139,
1500
+ "step": 4260
1501
+ },
1502
+ {
1503
+ "epoch": 0.5536197906140102,
1504
+ "grad_norm": 0.7074447870254517,
1505
+ "learning_rate": 9.783751657423787e-06,
1506
+ "loss": 2.2169,
1507
+ "step": 4280
1508
+ },
1509
+ {
1510
+ "epoch": 0.5562067989813655,
1511
+ "grad_norm": 0.7193735241889954,
1512
+ "learning_rate": 9.779351079496821e-06,
1513
+ "loss": 2.2435,
1514
+ "step": 4300
1515
+ },
1516
+ {
1517
+ "epoch": 0.5587938073487206,
1518
+ "grad_norm": 0.7104501128196716,
1519
+ "learning_rate": 9.774907186220005e-06,
1520
+ "loss": 2.198,
1521
+ "step": 4320
1522
+ },
1523
+ {
1524
+ "epoch": 0.5613808157160758,
1525
+ "grad_norm": 0.6711876392364502,
1526
+ "learning_rate": 9.770420017868426e-06,
1527
+ "loss": 2.1927,
1528
+ "step": 4340
1529
+ },
1530
+ {
1531
+ "epoch": 0.563967824083431,
1532
+ "grad_norm": 0.7272083163261414,
1533
+ "learning_rate": 9.765889615109379e-06,
1534
+ "loss": 2.2437,
1535
+ "step": 4360
1536
+ },
1537
+ {
1538
+ "epoch": 0.5665548324507862,
1539
+ "grad_norm": 0.7311263680458069,
1540
+ "learning_rate": 9.761316019001991e-06,
1541
+ "loss": 2.2106,
1542
+ "step": 4380
1543
+ },
1544
+ {
1545
+ "epoch": 0.5691418408181413,
1546
+ "grad_norm": 0.7337877154350281,
1547
+ "learning_rate": 9.756699270996848e-06,
1548
+ "loss": 2.2257,
1549
+ "step": 4400
1550
+ },
1551
+ {
1552
+ "epoch": 0.5717288491854966,
1553
+ "grad_norm": 0.7337206602096558,
1554
+ "learning_rate": 9.752039412935627e-06,
1555
+ "loss": 2.2066,
1556
+ "step": 4420
1557
+ },
1558
+ {
1559
+ "epoch": 0.5743158575528517,
1560
+ "grad_norm": 0.8056983947753906,
1561
+ "learning_rate": 9.74733648705071e-06,
1562
+ "loss": 2.2049,
1563
+ "step": 4440
1564
+ },
1565
+ {
1566
+ "epoch": 0.576902865920207,
1567
+ "grad_norm": 0.7203987240791321,
1568
+ "learning_rate": 9.742590535964805e-06,
1569
+ "loss": 2.2279,
1570
+ "step": 4460
1571
+ },
1572
+ {
1573
+ "epoch": 0.5794898742875622,
1574
+ "grad_norm": 0.7495117783546448,
1575
+ "learning_rate": 9.737801602690554e-06,
1576
+ "loss": 2.1986,
1577
+ "step": 4480
1578
+ },
1579
+ {
1580
+ "epoch": 0.5820768826549173,
1581
+ "grad_norm": 0.7343236207962036,
1582
+ "learning_rate": 9.732969730630153e-06,
1583
+ "loss": 2.2233,
1584
+ "step": 4500
1585
+ },
1586
+ {
1587
+ "epoch": 0.5846638910222726,
1588
+ "grad_norm": 0.7727257609367371,
1589
+ "learning_rate": 9.728094963574948e-06,
1590
+ "loss": 2.2042,
1591
+ "step": 4520
1592
+ },
1593
+ {
1594
+ "epoch": 0.5872508993896277,
1595
+ "grad_norm": 0.7064406275749207,
1596
+ "learning_rate": 9.723177345705048e-06,
1597
+ "loss": 2.2295,
1598
+ "step": 4540
1599
+ },
1600
+ {
1601
+ "epoch": 0.5898379077569829,
1602
+ "grad_norm": 0.7306511998176575,
1603
+ "learning_rate": 9.71821692158892e-06,
1604
+ "loss": 2.1982,
1605
+ "step": 4560
1606
+ },
1607
+ {
1608
+ "epoch": 0.5924249161243381,
1609
+ "grad_norm": 0.7717193365097046,
1610
+ "learning_rate": 9.713213736182976e-06,
1611
+ "loss": 2.206,
1612
+ "step": 4580
1613
+ },
1614
+ {
1615
+ "epoch": 0.5950119244916933,
1616
+ "grad_norm": 0.7180883288383484,
1617
+ "learning_rate": 9.708167834831183e-06,
1618
+ "loss": 2.2044,
1619
+ "step": 4600
1620
+ },
1621
+ {
1622
+ "epoch": 0.5975989328590484,
1623
+ "grad_norm": 0.7017503380775452,
1624
+ "learning_rate": 9.703079263264643e-06,
1625
+ "loss": 2.192,
1626
+ "step": 4620
1627
+ },
1628
+ {
1629
+ "epoch": 0.6001859412264037,
1630
+ "grad_norm": 0.7421537637710571,
1631
+ "learning_rate": 9.697948067601176e-06,
1632
+ "loss": 2.199,
1633
+ "step": 4640
1634
+ },
1635
+ {
1636
+ "epoch": 0.6027729495937588,
1637
+ "grad_norm": 0.7729679942131042,
1638
+ "learning_rate": 9.692774294344905e-06,
1639
+ "loss": 2.2073,
1640
+ "step": 4660
1641
+ },
1642
+ {
1643
+ "epoch": 0.6053599579611141,
1644
+ "grad_norm": 0.7517857551574707,
1645
+ "learning_rate": 9.687557990385836e-06,
1646
+ "loss": 2.1942,
1647
+ "step": 4680
1648
+ },
1649
+ {
1650
+ "epoch": 0.6079469663284692,
1651
+ "grad_norm": 0.7258435487747192,
1652
+ "learning_rate": 9.682299202999433e-06,
1653
+ "loss": 2.1916,
1654
+ "step": 4700
1655
+ },
1656
+ {
1657
+ "epoch": 0.6105339746958244,
1658
+ "grad_norm": 0.7256997227668762,
1659
+ "learning_rate": 9.676997979846183e-06,
1660
+ "loss": 2.1986,
1661
+ "step": 4720
1662
+ },
1663
+ {
1664
+ "epoch": 0.6131209830631796,
1665
+ "grad_norm": 0.7594742774963379,
1666
+ "learning_rate": 9.671654368971176e-06,
1667
+ "loss": 2.1867,
1668
+ "step": 4740
1669
+ },
1670
+ {
1671
+ "epoch": 0.6157079914305348,
1672
+ "grad_norm": 0.7546527981758118,
1673
+ "learning_rate": 9.666268418803655e-06,
1674
+ "loss": 2.1999,
1675
+ "step": 4760
1676
+ },
1677
+ {
1678
+ "epoch": 0.6182949997978899,
1679
+ "grad_norm": 0.7679339051246643,
1680
+ "learning_rate": 9.660840178156592e-06,
1681
+ "loss": 2.1964,
1682
+ "step": 4780
1683
+ },
1684
+ {
1685
+ "epoch": 0.6208820081652452,
1686
+ "grad_norm": 0.7719926834106445,
1687
+ "learning_rate": 9.655369696226235e-06,
1688
+ "loss": 2.1829,
1689
+ "step": 4800
1690
+ },
1691
+ {
1692
+ "epoch": 0.6234690165326003,
1693
+ "grad_norm": 0.7456310987472534,
1694
+ "learning_rate": 9.649857022591664e-06,
1695
+ "loss": 2.1725,
1696
+ "step": 4820
1697
+ },
1698
+ {
1699
+ "epoch": 0.6260560248999555,
1700
+ "grad_norm": 0.720140814781189,
1701
+ "learning_rate": 9.644302207214346e-06,
1702
+ "loss": 2.1806,
1703
+ "step": 4840
1704
+ },
1705
+ {
1706
+ "epoch": 0.6286430332673107,
1707
+ "grad_norm": 0.7672199010848999,
1708
+ "learning_rate": 9.63870530043768e-06,
1709
+ "loss": 2.1921,
1710
+ "step": 4860
1711
+ },
1712
+ {
1713
+ "epoch": 0.6312300416346659,
1714
+ "grad_norm": 0.7888000011444092,
1715
+ "learning_rate": 9.633066352986538e-06,
1716
+ "loss": 2.198,
1717
+ "step": 4880
1718
+ },
1719
+ {
1720
+ "epoch": 0.6338170500020212,
1721
+ "grad_norm": 0.7024748921394348,
1722
+ "learning_rate": 9.627385415966807e-06,
1723
+ "loss": 2.1579,
1724
+ "step": 4900
1725
+ },
1726
+ {
1727
+ "epoch": 0.6364040583693763,
1728
+ "grad_norm": 0.7126362919807434,
1729
+ "learning_rate": 9.62166254086493e-06,
1730
+ "loss": 2.1882,
1731
+ "step": 4920
1732
+ },
1733
+ {
1734
+ "epoch": 0.6389910667367315,
1735
+ "grad_norm": 0.6748985052108765,
1736
+ "learning_rate": 9.61589777954743e-06,
1737
+ "loss": 2.1931,
1738
+ "step": 4940
1739
+ },
1740
+ {
1741
+ "epoch": 0.6415780751040867,
1742
+ "grad_norm": 0.7744324207305908,
1743
+ "learning_rate": 9.61009118426045e-06,
1744
+ "loss": 2.1823,
1745
+ "step": 4960
1746
+ },
1747
+ {
1748
+ "epoch": 0.6441650834714419,
1749
+ "grad_norm": 0.6999046802520752,
1750
+ "learning_rate": 9.604242807629275e-06,
1751
+ "loss": 2.1598,
1752
+ "step": 4980
1753
+ },
1754
+ {
1755
+ "epoch": 0.646752091838797,
1756
+ "grad_norm": 0.7126173973083496,
1757
+ "learning_rate": 9.59835270265785e-06,
1758
+ "loss": 2.1661,
1759
+ "step": 5000
1760
+ },
1761
+ {
1762
+ "epoch": 0.6493391002061523,
1763
+ "grad_norm": 0.7425235509872437,
1764
+ "learning_rate": 9.592420922728312e-06,
1765
+ "loss": 2.1729,
1766
+ "step": 5020
1767
+ },
1768
+ {
1769
+ "epoch": 0.6519261085735074,
1770
+ "grad_norm": 0.735068142414093,
1771
+ "learning_rate": 9.586447521600496e-06,
1772
+ "loss": 2.1734,
1773
+ "step": 5040
1774
+ },
1775
+ {
1776
+ "epoch": 0.6545131169408626,
1777
+ "grad_norm": 0.772686779499054,
1778
+ "learning_rate": 9.580432553411446e-06,
1779
+ "loss": 2.1943,
1780
+ "step": 5060
1781
+ },
1782
+ {
1783
+ "epoch": 0.6571001253082178,
1784
+ "grad_norm": 0.8012945652008057,
1785
+ "learning_rate": 9.574376072674936e-06,
1786
+ "loss": 2.1678,
1787
+ "step": 5080
1788
+ },
1789
+ {
1790
+ "epoch": 0.659687133675573,
1791
+ "grad_norm": 0.7072063684463501,
1792
+ "learning_rate": 9.568278134280966e-06,
1793
+ "loss": 2.1749,
1794
+ "step": 5100
1795
+ },
1796
+ {
1797
+ "epoch": 0.6622741420429281,
1798
+ "grad_norm": 0.687140941619873,
1799
+ "learning_rate": 9.562138793495268e-06,
1800
+ "loss": 2.1844,
1801
+ "step": 5120
1802
+ },
1803
+ {
1804
+ "epoch": 0.6648611504102834,
1805
+ "grad_norm": 0.7809275984764099,
1806
+ "learning_rate": 9.555958105958805e-06,
1807
+ "loss": 2.1719,
1808
+ "step": 5140
1809
+ },
1810
+ {
1811
+ "epoch": 0.6674481587776385,
1812
+ "grad_norm": 0.7361642122268677,
1813
+ "learning_rate": 9.549736127687265e-06,
1814
+ "loss": 2.1706,
1815
+ "step": 5160
1816
+ },
1817
+ {
1818
+ "epoch": 0.6700351671449938,
1819
+ "grad_norm": 0.7150685787200928,
1820
+ "learning_rate": 9.543472915070555e-06,
1821
+ "loss": 2.1648,
1822
+ "step": 5180
1823
+ },
1824
+ {
1825
+ "epoch": 0.6726221755123489,
1826
+ "grad_norm": 0.7491324543952942,
1827
+ "learning_rate": 9.537168524872292e-06,
1828
+ "loss": 2.156,
1829
+ "step": 5200
1830
+ },
1831
+ {
1832
+ "epoch": 0.6752091838797041,
1833
+ "grad_norm": 0.7028157711029053,
1834
+ "learning_rate": 9.530823014229283e-06,
1835
+ "loss": 2.1588,
1836
+ "step": 5220
1837
+ },
1838
+ {
1839
+ "epoch": 0.6777961922470593,
1840
+ "grad_norm": 0.7949670553207397,
1841
+ "learning_rate": 9.52443644065101e-06,
1842
+ "loss": 2.1788,
1843
+ "step": 5240
1844
+ },
1845
+ {
1846
+ "epoch": 0.6803832006144145,
1847
+ "grad_norm": 0.7003277540206909,
1848
+ "learning_rate": 9.518008862019116e-06,
1849
+ "loss": 2.1917,
1850
+ "step": 5260
1851
+ },
1852
+ {
1853
+ "epoch": 0.6829702089817696,
1854
+ "grad_norm": 0.7252909541130066,
1855
+ "learning_rate": 9.511540336586864e-06,
1856
+ "loss": 2.1709,
1857
+ "step": 5280
1858
+ },
1859
+ {
1860
+ "epoch": 0.6855572173491249,
1861
+ "grad_norm": 0.7193975448608398,
1862
+ "learning_rate": 9.505030922978626e-06,
1863
+ "loss": 2.185,
1864
+ "step": 5300
1865
+ },
1866
+ {
1867
+ "epoch": 0.68814422571648,
1868
+ "grad_norm": 0.7502670288085938,
1869
+ "learning_rate": 9.49848068018934e-06,
1870
+ "loss": 2.1787,
1871
+ "step": 5320
1872
+ },
1873
+ {
1874
+ "epoch": 0.6907312340838352,
1875
+ "grad_norm": 0.7460989952087402,
1876
+ "learning_rate": 9.49188966758398e-06,
1877
+ "loss": 2.1557,
1878
+ "step": 5340
1879
+ },
1880
+ {
1881
+ "epoch": 0.6933182424511904,
1882
+ "grad_norm": 0.710442066192627,
1883
+ "learning_rate": 9.485257944897021e-06,
1884
+ "loss": 2.1547,
1885
+ "step": 5360
1886
+ },
1887
+ {
1888
+ "epoch": 0.6959052508185456,
1889
+ "grad_norm": 0.7363094091415405,
1890
+ "learning_rate": 9.478585572231891e-06,
1891
+ "loss": 2.1473,
1892
+ "step": 5380
1893
+ },
1894
+ {
1895
+ "epoch": 0.6984922591859009,
1896
+ "grad_norm": 0.7203119397163391,
1897
+ "learning_rate": 9.47187261006043e-06,
1898
+ "loss": 2.1743,
1899
+ "step": 5400
1900
+ },
1901
+ {
1902
+ "epoch": 0.701079267553256,
1903
+ "grad_norm": 0.715248167514801,
1904
+ "learning_rate": 9.465119119222346e-06,
1905
+ "loss": 2.1626,
1906
+ "step": 5420
1907
+ },
1908
+ {
1909
+ "epoch": 0.7036662759206112,
1910
+ "grad_norm": 0.7098533511161804,
1911
+ "learning_rate": 9.458325160924648e-06,
1912
+ "loss": 2.1807,
1913
+ "step": 5440
1914
+ },
1915
+ {
1916
+ "epoch": 0.7062532842879664,
1917
+ "grad_norm": 0.7714352607727051,
1918
+ "learning_rate": 9.451490796741117e-06,
1919
+ "loss": 2.1517,
1920
+ "step": 5460
1921
+ },
1922
+ {
1923
+ "epoch": 0.7088402926553216,
1924
+ "grad_norm": 0.720448911190033,
1925
+ "learning_rate": 9.444616088611718e-06,
1926
+ "loss": 2.1775,
1927
+ "step": 5480
1928
+ },
1929
+ {
1930
+ "epoch": 0.7114273010226767,
1931
+ "grad_norm": 0.7580015659332275,
1932
+ "learning_rate": 9.437701098842067e-06,
1933
+ "loss": 2.1529,
1934
+ "step": 5500
1935
+ },
1936
+ {
1937
+ "epoch": 0.714014309390032,
1938
+ "grad_norm": 0.727325439453125,
1939
+ "learning_rate": 9.430745890102849e-06,
1940
+ "loss": 2.1654,
1941
+ "step": 5520
1942
+ },
1943
+ {
1944
+ "epoch": 0.7166013177573871,
1945
+ "grad_norm": 0.7433022260665894,
1946
+ "learning_rate": 9.423750525429248e-06,
1947
+ "loss": 2.1565,
1948
+ "step": 5540
1949
+ },
1950
+ {
1951
+ "epoch": 0.7191883261247423,
1952
+ "grad_norm": 0.728881299495697,
1953
+ "learning_rate": 9.416715068220393e-06,
1954
+ "loss": 2.1734,
1955
+ "step": 5560
1956
+ },
1957
+ {
1958
+ "epoch": 0.7217753344920975,
1959
+ "grad_norm": 0.7467840909957886,
1960
+ "learning_rate": 9.409639582238761e-06,
1961
+ "loss": 2.1502,
1962
+ "step": 5580
1963
+ },
1964
+ {
1965
+ "epoch": 0.7243623428594527,
1966
+ "grad_norm": 0.7820594310760498,
1967
+ "learning_rate": 9.40252413160962e-06,
1968
+ "loss": 2.1725,
1969
+ "step": 5600
1970
+ },
1971
+ {
1972
+ "epoch": 0.7269493512268079,
1973
+ "grad_norm": 0.7077836394309998,
1974
+ "learning_rate": 9.395368780820433e-06,
1975
+ "loss": 2.1508,
1976
+ "step": 5620
1977
+ },
1978
+ {
1979
+ "epoch": 0.7295363595941631,
1980
+ "grad_norm": 0.7221850752830505,
1981
+ "learning_rate": 9.388173594720283e-06,
1982
+ "loss": 2.1649,
1983
+ "step": 5640
1984
+ },
1985
+ {
1986
+ "epoch": 0.7321233679615182,
1987
+ "grad_norm": 0.7874971628189087,
1988
+ "learning_rate": 9.380938638519274e-06,
1989
+ "loss": 2.158,
1990
+ "step": 5660
1991
+ },
1992
+ {
1993
+ "epoch": 0.7347103763288735,
1994
+ "grad_norm": 0.7347155213356018,
1995
+ "learning_rate": 9.373663977787956e-06,
1996
+ "loss": 2.1486,
1997
+ "step": 5680
1998
+ },
1999
+ {
2000
+ "epoch": 0.7372973846962286,
2001
+ "grad_norm": 0.7876786589622498,
2002
+ "learning_rate": 9.366349678456717e-06,
2003
+ "loss": 2.1501,
2004
+ "step": 5700
2005
+ },
2006
+ {
2007
+ "epoch": 0.7398843930635838,
2008
+ "grad_norm": 0.678352952003479,
2009
+ "learning_rate": 9.35899580681519e-06,
2010
+ "loss": 2.1644,
2011
+ "step": 5720
2012
+ },
2013
+ {
2014
+ "epoch": 0.742471401430939,
2015
+ "grad_norm": 0.7279735803604126,
2016
+ "learning_rate": 9.351602429511655e-06,
2017
+ "loss": 2.1508,
2018
+ "step": 5740
2019
+ },
2020
+ {
2021
+ "epoch": 0.7450584097982942,
2022
+ "grad_norm": 0.7041738629341125,
2023
+ "learning_rate": 9.344169613552428e-06,
2024
+ "loss": 2.1617,
2025
+ "step": 5760
2026
+ },
2027
+ {
2028
+ "epoch": 0.7476454181656493,
2029
+ "grad_norm": 0.750642716884613,
2030
+ "learning_rate": 9.336697426301267e-06,
2031
+ "loss": 2.1409,
2032
+ "step": 5780
2033
+ },
2034
+ {
2035
+ "epoch": 0.7502324265330046,
2036
+ "grad_norm": 0.7688580751419067,
2037
+ "learning_rate": 9.329185935478741e-06,
2038
+ "loss": 2.1459,
2039
+ "step": 5800
2040
+ },
2041
+ {
2042
+ "epoch": 0.7528194349003597,
2043
+ "grad_norm": 0.704578697681427,
2044
+ "learning_rate": 9.321635209161642e-06,
2045
+ "loss": 2.1417,
2046
+ "step": 5820
2047
+ },
2048
+ {
2049
+ "epoch": 0.7554064432677149,
2050
+ "grad_norm": 0.733323872089386,
2051
+ "learning_rate": 9.314045315782339e-06,
2052
+ "loss": 2.1516,
2053
+ "step": 5840
2054
+ },
2055
+ {
2056
+ "epoch": 0.7579934516350701,
2057
+ "grad_norm": 0.771930992603302,
2058
+ "learning_rate": 9.306416324128184e-06,
2059
+ "loss": 2.1256,
2060
+ "step": 5860
2061
+ },
2062
+ {
2063
+ "epoch": 0.7605804600024253,
2064
+ "grad_norm": 0.6910899877548218,
2065
+ "learning_rate": 9.298748303340871e-06,
2066
+ "loss": 2.1421,
2067
+ "step": 5880
2068
+ },
2069
+ {
2070
+ "epoch": 0.7631674683697806,
2071
+ "grad_norm": 0.7119818329811096,
2072
+ "learning_rate": 9.291041322915824e-06,
2073
+ "loss": 2.1631,
2074
+ "step": 5900
2075
+ },
2076
+ {
2077
+ "epoch": 0.7657544767371357,
2078
+ "grad_norm": 0.6817770600318909,
2079
+ "learning_rate": 9.283295452701549e-06,
2080
+ "loss": 2.143,
2081
+ "step": 5920
2082
+ },
2083
+ {
2084
+ "epoch": 0.7683414851044909,
2085
+ "grad_norm": 0.714541494846344,
2086
+ "learning_rate": 9.275510762899016e-06,
2087
+ "loss": 2.1546,
2088
+ "step": 5940
2089
+ },
2090
+ {
2091
+ "epoch": 0.7709284934718461,
2092
+ "grad_norm": 0.6969371438026428,
2093
+ "learning_rate": 9.267687324061016e-06,
2094
+ "loss": 2.1186,
2095
+ "step": 5960
2096
+ },
2097
+ {
2098
+ "epoch": 0.7735155018392013,
2099
+ "grad_norm": 0.7325617671012878,
2100
+ "learning_rate": 9.259825207091526e-06,
2101
+ "loss": 2.1363,
2102
+ "step": 5980
2103
+ },
2104
+ {
2105
+ "epoch": 0.7761025102065564,
2106
+ "grad_norm": 0.743165135383606,
2107
+ "learning_rate": 9.25192448324506e-06,
2108
+ "loss": 2.1269,
2109
+ "step": 6000
2110
+ },
2111
+ {
2112
+ "epoch": 0.7786895185739117,
2113
+ "grad_norm": 0.7538695931434631,
2114
+ "learning_rate": 9.243985224126031e-06,
2115
+ "loss": 2.1237,
2116
+ "step": 6020
2117
+ },
2118
+ {
2119
+ "epoch": 0.7812765269412668,
2120
+ "grad_norm": 0.7549970746040344,
2121
+ "learning_rate": 9.236007501688094e-06,
2122
+ "loss": 2.1343,
2123
+ "step": 6040
2124
+ },
2125
+ {
2126
+ "epoch": 0.783863535308622,
2127
+ "grad_norm": 0.732627272605896,
2128
+ "learning_rate": 9.2279913882335e-06,
2129
+ "loss": 2.1465,
2130
+ "step": 6060
2131
+ },
2132
+ {
2133
+ "epoch": 0.7864505436759772,
2134
+ "grad_norm": 0.7083766460418701,
2135
+ "learning_rate": 9.219936956412436e-06,
2136
+ "loss": 2.1394,
2137
+ "step": 6080
2138
+ },
2139
+ {
2140
+ "epoch": 0.7890375520433324,
2141
+ "grad_norm": 0.7239031195640564,
2142
+ "learning_rate": 9.211844279222376e-06,
2143
+ "loss": 2.1386,
2144
+ "step": 6100
2145
+ },
2146
+ {
2147
+ "epoch": 0.7916245604106876,
2148
+ "grad_norm": 0.7016878128051758,
2149
+ "learning_rate": 9.2037134300074e-06,
2150
+ "loss": 2.1313,
2151
+ "step": 6120
2152
+ },
2153
+ {
2154
+ "epoch": 0.7942115687780428,
2155
+ "grad_norm": 0.6846844553947449,
2156
+ "learning_rate": 9.195544482457555e-06,
2157
+ "loss": 2.1383,
2158
+ "step": 6140
2159
+ },
2160
+ {
2161
+ "epoch": 0.7967985771453979,
2162
+ "grad_norm": 0.7442426085472107,
2163
+ "learning_rate": 9.187337510608168e-06,
2164
+ "loss": 2.1177,
2165
+ "step": 6160
2166
+ },
2167
+ {
2168
+ "epoch": 0.7993855855127532,
2169
+ "grad_norm": 0.7443544268608093,
2170
+ "learning_rate": 9.179092588839178e-06,
2171
+ "loss": 2.1391,
2172
+ "step": 6180
2173
+ },
2174
+ {
2175
+ "epoch": 0.8019725938801083,
2176
+ "grad_norm": 0.7053080201148987,
2177
+ "learning_rate": 9.170809791874468e-06,
2178
+ "loss": 2.1221,
2179
+ "step": 6200
2180
+ },
2181
+ {
2182
+ "epoch": 0.8045596022474635,
2183
+ "grad_norm": 0.7665020227432251,
2184
+ "learning_rate": 9.16248919478119e-06,
2185
+ "loss": 2.1116,
2186
+ "step": 6220
2187
+ },
2188
+ {
2189
+ "epoch": 0.8071466106148187,
2190
+ "grad_norm": 0.7906709909439087,
2191
+ "learning_rate": 9.154130872969067e-06,
2192
+ "loss": 2.1363,
2193
+ "step": 6240
2194
+ },
2195
+ {
2196
+ "epoch": 0.8097336189821739,
2197
+ "grad_norm": 0.6853694915771484,
2198
+ "learning_rate": 9.145734902189733e-06,
2199
+ "loss": 2.149,
2200
+ "step": 6260
2201
+ },
2202
+ {
2203
+ "epoch": 0.812320627349529,
2204
+ "grad_norm": 0.7411865592002869,
2205
+ "learning_rate": 9.137301358536032e-06,
2206
+ "loss": 2.1355,
2207
+ "step": 6280
2208
+ },
2209
+ {
2210
+ "epoch": 0.8149076357168843,
2211
+ "grad_norm": 0.8013186454772949,
2212
+ "learning_rate": 9.128830318441327e-06,
2213
+ "loss": 2.1175,
2214
+ "step": 6300
2215
+ },
2216
+ {
2217
+ "epoch": 0.8174946440842394,
2218
+ "grad_norm": 0.7919278144836426,
2219
+ "learning_rate": 9.120321858678817e-06,
2220
+ "loss": 2.128,
2221
+ "step": 6320
2222
+ },
2223
+ {
2224
+ "epoch": 0.8200816524515947,
2225
+ "grad_norm": 0.7355955243110657,
2226
+ "learning_rate": 9.111776056360838e-06,
2227
+ "loss": 2.1253,
2228
+ "step": 6340
2229
+ },
2230
+ {
2231
+ "epoch": 0.8226686608189498,
2232
+ "grad_norm": 0.7750183343887329,
2233
+ "learning_rate": 9.103192988938155e-06,
2234
+ "loss": 2.1225,
2235
+ "step": 6360
2236
+ },
2237
+ {
2238
+ "epoch": 0.825255669186305,
2239
+ "grad_norm": 0.7022154927253723,
2240
+ "learning_rate": 9.094572734199271e-06,
2241
+ "loss": 2.1193,
2242
+ "step": 6380
2243
+ },
2244
+ {
2245
+ "epoch": 0.8278426775536603,
2246
+ "grad_norm": 0.7672535181045532,
2247
+ "learning_rate": 9.085915370269723e-06,
2248
+ "loss": 2.1188,
2249
+ "step": 6400
2250
+ },
2251
+ {
2252
+ "epoch": 0.8304296859210154,
2253
+ "grad_norm": 0.7107143998146057,
2254
+ "learning_rate": 9.077220975611363e-06,
2255
+ "loss": 2.1278,
2256
+ "step": 6420
2257
+ },
2258
+ {
2259
+ "epoch": 0.8330166942883706,
2260
+ "grad_norm": 0.7380732297897339,
2261
+ "learning_rate": 9.068489629021655e-06,
2262
+ "loss": 2.1374,
2263
+ "step": 6440
2264
+ },
2265
+ {
2266
+ "epoch": 0.8356037026557258,
2267
+ "grad_norm": 0.6959198117256165,
2268
+ "learning_rate": 9.05972140963296e-06,
2269
+ "loss": 2.129,
2270
+ "step": 6460
2271
+ },
2272
+ {
2273
+ "epoch": 0.838190711023081,
2274
+ "grad_norm": 0.742127001285553,
2275
+ "learning_rate": 9.050916396911818e-06,
2276
+ "loss": 2.1285,
2277
+ "step": 6480
2278
+ },
2279
+ {
2280
+ "epoch": 0.8407777193904361,
2281
+ "grad_norm": 0.6720155477523804,
2282
+ "learning_rate": 9.042074670658223e-06,
2283
+ "loss": 2.1172,
2284
+ "step": 6500
2285
+ },
2286
+ {
2287
+ "epoch": 0.8433647277577914,
2288
+ "grad_norm": 0.7020736932754517,
2289
+ "learning_rate": 9.033196311004915e-06,
2290
+ "loss": 2.1036,
2291
+ "step": 6520
2292
+ },
2293
+ {
2294
+ "epoch": 0.8459517361251465,
2295
+ "grad_norm": 0.746281087398529,
2296
+ "learning_rate": 9.024281398416632e-06,
2297
+ "loss": 2.1183,
2298
+ "step": 6540
2299
+ },
2300
+ {
2301
+ "epoch": 0.8485387444925018,
2302
+ "grad_norm": 0.7673128843307495,
2303
+ "learning_rate": 9.015330013689396e-06,
2304
+ "loss": 2.1435,
2305
+ "step": 6560
2306
+ },
2307
+ {
2308
+ "epoch": 0.8511257528598569,
2309
+ "grad_norm": 0.7635971903800964,
2310
+ "learning_rate": 9.006342237949782e-06,
2311
+ "loss": 2.1018,
2312
+ "step": 6580
2313
+ },
2314
+ {
2315
+ "epoch": 0.8537127612272121,
2316
+ "grad_norm": 0.7365761995315552,
2317
+ "learning_rate": 8.997318152654167e-06,
2318
+ "loss": 2.116,
2319
+ "step": 6600
2320
+ },
2321
+ {
2322
+ "epoch": 0.8562997695945673,
2323
+ "grad_norm": 0.7582866549491882,
2324
+ "learning_rate": 8.988257839588011e-06,
2325
+ "loss": 2.1146,
2326
+ "step": 6620
2327
+ },
2328
+ {
2329
+ "epoch": 0.8588867779619225,
2330
+ "grad_norm": 0.7569854259490967,
2331
+ "learning_rate": 8.979161380865104e-06,
2332
+ "loss": 2.1156,
2333
+ "step": 6640
2334
+ },
2335
+ {
2336
+ "epoch": 0.8614737863292776,
2337
+ "grad_norm": 0.6821621656417847,
2338
+ "learning_rate": 8.970028858926825e-06,
2339
+ "loss": 2.1134,
2340
+ "step": 6660
2341
+ },
2342
+ {
2343
+ "epoch": 0.8640607946966329,
2344
+ "grad_norm": 0.7307649850845337,
2345
+ "learning_rate": 8.96086035654139e-06,
2346
+ "loss": 2.124,
2347
+ "step": 6680
2348
+ },
2349
+ {
2350
+ "epoch": 0.866647803063988,
2351
+ "grad_norm": 0.7775338292121887,
2352
+ "learning_rate": 8.951655956803118e-06,
2353
+ "loss": 2.1052,
2354
+ "step": 6700
2355
+ },
2356
+ {
2357
+ "epoch": 0.8692348114313432,
2358
+ "grad_norm": 0.7425001859664917,
2359
+ "learning_rate": 8.942415743131651e-06,
2360
+ "loss": 2.106,
2361
+ "step": 6720
2362
+ },
2363
+ {
2364
+ "epoch": 0.8718218197986984,
2365
+ "grad_norm": 0.7379507422447205,
2366
+ "learning_rate": 8.933139799271229e-06,
2367
+ "loss": 2.0974,
2368
+ "step": 6740
2369
+ },
2370
+ {
2371
+ "epoch": 0.8744088281660536,
2372
+ "grad_norm": 0.7932276129722595,
2373
+ "learning_rate": 8.923828209289904e-06,
2374
+ "loss": 2.1257,
2375
+ "step": 6760
2376
+ },
2377
+ {
2378
+ "epoch": 0.8769958365334087,
2379
+ "grad_norm": 0.7237013578414917,
2380
+ "learning_rate": 8.914481057578791e-06,
2381
+ "loss": 2.1033,
2382
+ "step": 6780
2383
+ },
2384
+ {
2385
+ "epoch": 0.879582844900764,
2386
+ "grad_norm": 0.7310687899589539,
2387
+ "learning_rate": 8.905098428851309e-06,
2388
+ "loss": 2.1194,
2389
+ "step": 6800
2390
+ },
2391
+ {
2392
+ "epoch": 0.8821698532681191,
2393
+ "grad_norm": 0.6933685541152954,
2394
+ "learning_rate": 8.8956804081424e-06,
2395
+ "loss": 2.0891,
2396
+ "step": 6820
2397
+ },
2398
+ {
2399
+ "epoch": 0.8847568616354744,
2400
+ "grad_norm": 0.7352440357208252,
2401
+ "learning_rate": 8.886227080807762e-06,
2402
+ "loss": 2.1022,
2403
+ "step": 6840
2404
+ },
2405
+ {
2406
+ "epoch": 0.8873438700028295,
2407
+ "grad_norm": 0.7073638439178467,
2408
+ "learning_rate": 8.876738532523081e-06,
2409
+ "loss": 2.1065,
2410
+ "step": 6860
2411
+ },
2412
+ {
2413
+ "epoch": 0.8899308783701847,
2414
+ "grad_norm": 0.7440693974494934,
2415
+ "learning_rate": 8.867214849283252e-06,
2416
+ "loss": 2.1079,
2417
+ "step": 6880
2418
+ },
2419
+ {
2420
+ "epoch": 0.89251788673754,
2421
+ "grad_norm": 0.7651025056838989,
2422
+ "learning_rate": 8.8576561174016e-06,
2423
+ "loss": 2.121,
2424
+ "step": 6900
2425
+ },
2426
+ {
2427
+ "epoch": 0.8951048951048951,
2428
+ "grad_norm": 0.7150782346725464,
2429
+ "learning_rate": 8.84806242350909e-06,
2430
+ "loss": 2.1025,
2431
+ "step": 6920
2432
+ },
2433
+ {
2434
+ "epoch": 0.8976919034722503,
2435
+ "grad_norm": 0.7346411943435669,
2436
+ "learning_rate": 8.838433854553555e-06,
2437
+ "loss": 2.0959,
2438
+ "step": 6940
2439
+ },
2440
+ {
2441
+ "epoch": 0.9002789118396055,
2442
+ "grad_norm": 0.7531387209892273,
2443
+ "learning_rate": 8.828770497798897e-06,
2444
+ "loss": 2.0901,
2445
+ "step": 6960
2446
+ },
2447
+ {
2448
+ "epoch": 0.9028659202069607,
2449
+ "grad_norm": 0.7870126366615295,
2450
+ "learning_rate": 8.819072440824303e-06,
2451
+ "loss": 2.1034,
2452
+ "step": 6980
2453
+ },
2454
+ {
2455
+ "epoch": 0.9054529285743158,
2456
+ "grad_norm": 0.7796097993850708,
2457
+ "learning_rate": 8.80933977152345e-06,
2458
+ "loss": 2.0825,
2459
+ "step": 7000
2460
+ },
2461
+ {
2462
+ "epoch": 0.9080399369416711,
2463
+ "grad_norm": 0.7178963422775269,
2464
+ "learning_rate": 8.799572578103703e-06,
2465
+ "loss": 2.1252,
2466
+ "step": 7020
2467
+ },
2468
+ {
2469
+ "epoch": 0.9106269453090262,
2470
+ "grad_norm": 0.7850649952888489,
2471
+ "learning_rate": 8.789770949085321e-06,
2472
+ "loss": 2.0886,
2473
+ "step": 7040
2474
+ },
2475
+ {
2476
+ "epoch": 0.9132139536763815,
2477
+ "grad_norm": 0.7312331795692444,
2478
+ "learning_rate": 8.779934973300657e-06,
2479
+ "loss": 2.1015,
2480
+ "step": 7060
2481
+ },
2482
+ {
2483
+ "epoch": 0.9158009620437366,
2484
+ "grad_norm": 0.7429481148719788,
2485
+ "learning_rate": 8.770064739893346e-06,
2486
+ "loss": 2.0791,
2487
+ "step": 7080
2488
+ },
2489
+ {
2490
+ "epoch": 0.9183879704110918,
2491
+ "grad_norm": 0.70722895860672,
2492
+ "learning_rate": 8.7601603383175e-06,
2493
+ "loss": 2.0992,
2494
+ "step": 7100
2495
+ },
2496
+ {
2497
+ "epoch": 0.920974978778447,
2498
+ "grad_norm": 0.7041098475456238,
2499
+ "learning_rate": 8.750221858336902e-06,
2500
+ "loss": 2.1027,
2501
+ "step": 7120
2502
+ },
2503
+ {
2504
+ "epoch": 0.9235619871458022,
2505
+ "grad_norm": 0.7348500490188599,
2506
+ "learning_rate": 8.740249390024183e-06,
2507
+ "loss": 2.0899,
2508
+ "step": 7140
2509
+ },
2510
+ {
2511
+ "epoch": 0.9261489955131573,
2512
+ "grad_norm": 0.7186647653579712,
2513
+ "learning_rate": 8.730243023760012e-06,
2514
+ "loss": 2.1059,
2515
+ "step": 7160
2516
+ },
2517
+ {
2518
+ "epoch": 0.9287360038805126,
2519
+ "grad_norm": 0.7586842775344849,
2520
+ "learning_rate": 8.720202850232281e-06,
2521
+ "loss": 2.0929,
2522
+ "step": 7180
2523
+ },
2524
+ {
2525
+ "epoch": 0.9313230122478677,
2526
+ "grad_norm": 0.7751206159591675,
2527
+ "learning_rate": 8.710128960435271e-06,
2528
+ "loss": 2.0961,
2529
+ "step": 7200
2530
+ },
2531
+ {
2532
+ "epoch": 0.9339100206152229,
2533
+ "grad_norm": 0.7438105344772339,
2534
+ "learning_rate": 8.700021445668839e-06,
2535
+ "loss": 2.1013,
2536
+ "step": 7220
2537
+ },
2538
+ {
2539
+ "epoch": 0.9364970289825781,
2540
+ "grad_norm": 0.7343020439147949,
2541
+ "learning_rate": 8.68988039753758e-06,
2542
+ "loss": 2.0904,
2543
+ "step": 7240
2544
+ },
2545
+ {
2546
+ "epoch": 0.9390840373499333,
2547
+ "grad_norm": 0.7382264733314514,
2548
+ "learning_rate": 8.67970590795001e-06,
2549
+ "loss": 2.1016,
2550
+ "step": 7260
2551
+ },
2552
+ {
2553
+ "epoch": 0.9416710457172885,
2554
+ "grad_norm": 0.720942497253418,
2555
+ "learning_rate": 8.669498069117721e-06,
2556
+ "loss": 2.1167,
2557
+ "step": 7280
2558
+ },
2559
+ {
2560
+ "epoch": 0.9442580540846437,
2561
+ "grad_norm": 0.7112380266189575,
2562
+ "learning_rate": 8.65925697355455e-06,
2563
+ "loss": 2.0981,
2564
+ "step": 7300
2565
+ },
2566
+ {
2567
+ "epoch": 0.9468450624519988,
2568
+ "grad_norm": 0.7193992137908936,
2569
+ "learning_rate": 8.648982714075743e-06,
2570
+ "loss": 2.0853,
2571
+ "step": 7320
2572
+ },
2573
+ {
2574
+ "epoch": 0.9494320708193541,
2575
+ "grad_norm": 0.7250093221664429,
2576
+ "learning_rate": 8.638675383797106e-06,
2577
+ "loss": 2.0936,
2578
+ "step": 7340
2579
+ },
2580
+ {
2581
+ "epoch": 0.9520190791867092,
2582
+ "grad_norm": 0.727990448474884,
2583
+ "learning_rate": 8.628335076134173e-06,
2584
+ "loss": 2.0882,
2585
+ "step": 7360
2586
+ },
2587
+ {
2588
+ "epoch": 0.9546060875540644,
2589
+ "grad_norm": 0.7733245491981506,
2590
+ "learning_rate": 8.617961884801346e-06,
2591
+ "loss": 2.0799,
2592
+ "step": 7380
2593
+ },
2594
+ {
2595
+ "epoch": 0.9571930959214197,
2596
+ "grad_norm": 0.7614827752113342,
2597
+ "learning_rate": 8.60755590381106e-06,
2598
+ "loss": 2.0935,
2599
+ "step": 7400
2600
+ },
2601
+ {
2602
+ "epoch": 0.9597801042887748,
2603
+ "grad_norm": 0.7480348348617554,
2604
+ "learning_rate": 8.597117227472915e-06,
2605
+ "loss": 2.0889,
2606
+ "step": 7420
2607
+ },
2608
+ {
2609
+ "epoch": 0.96236711265613,
2610
+ "grad_norm": 0.7127954959869385,
2611
+ "learning_rate": 8.586645950392835e-06,
2612
+ "loss": 2.0981,
2613
+ "step": 7440
2614
+ },
2615
+ {
2616
+ "epoch": 0.9649541210234852,
2617
+ "grad_norm": 0.6996986865997314,
2618
+ "learning_rate": 8.576142167472204e-06,
2619
+ "loss": 2.0856,
2620
+ "step": 7460
2621
+ },
2622
+ {
2623
+ "epoch": 0.9675411293908404,
2624
+ "grad_norm": 0.757079005241394,
2625
+ "learning_rate": 8.565605973907006e-06,
2626
+ "loss": 2.0873,
2627
+ "step": 7480
2628
+ },
2629
+ {
2630
+ "epoch": 0.9701281377581955,
2631
+ "grad_norm": 0.7550200819969177,
2632
+ "learning_rate": 8.555037465186962e-06,
2633
+ "loss": 2.0817,
2634
+ "step": 7500
2635
+ },
2636
+ {
2637
+ "epoch": 0.9727151461255508,
2638
+ "grad_norm": 0.7152219414710999,
2639
+ "learning_rate": 8.544436737094672e-06,
2640
+ "loss": 2.0962,
2641
+ "step": 7520
2642
+ },
2643
+ {
2644
+ "epoch": 0.9753021544929059,
2645
+ "grad_norm": 0.774013876914978,
2646
+ "learning_rate": 8.533803885704732e-06,
2647
+ "loss": 2.0944,
2648
+ "step": 7540
2649
+ },
2650
+ {
2651
+ "epoch": 0.9778891628602612,
2652
+ "grad_norm": 0.721843421459198,
2653
+ "learning_rate": 8.523139007382881e-06,
2654
+ "loss": 2.0849,
2655
+ "step": 7560
2656
+ },
2657
+ {
2658
+ "epoch": 0.9804761712276163,
2659
+ "grad_norm": 0.7735098600387573,
2660
+ "learning_rate": 8.51244219878511e-06,
2661
+ "loss": 2.0906,
2662
+ "step": 7580
2663
+ },
2664
+ {
2665
+ "epoch": 0.9830631795949715,
2666
+ "grad_norm": 0.7077389359474182,
2667
+ "learning_rate": 8.501713556856803e-06,
2668
+ "loss": 2.0665,
2669
+ "step": 7600
2670
+ },
2671
+ {
2672
+ "epoch": 0.9856501879623267,
2673
+ "grad_norm": 0.7230038046836853,
2674
+ "learning_rate": 8.490953178831846e-06,
2675
+ "loss": 2.0766,
2676
+ "step": 7620
2677
+ },
2678
+ {
2679
+ "epoch": 0.9882371963296819,
2680
+ "grad_norm": 0.7460753321647644,
2681
+ "learning_rate": 8.480161162231747e-06,
2682
+ "loss": 2.0556,
2683
+ "step": 7640
2684
+ },
2685
+ {
2686
+ "epoch": 0.990824204697037,
2687
+ "grad_norm": 0.7653132081031799,
2688
+ "learning_rate": 8.469337604864759e-06,
2689
+ "loss": 2.0821,
2690
+ "step": 7660
2691
+ },
2692
+ {
2693
+ "epoch": 0.9934112130643923,
2694
+ "grad_norm": 0.7243251204490662,
2695
+ "learning_rate": 8.458482604824988e-06,
2696
+ "loss": 2.0926,
2697
+ "step": 7680
2698
+ },
2699
+ {
2700
+ "epoch": 0.9959982214317474,
2701
+ "grad_norm": 0.7776849865913391,
2702
+ "learning_rate": 8.447596260491508e-06,
2703
+ "loss": 2.0925,
2704
+ "step": 7700
2705
+ },
2706
+ {
2707
+ "epoch": 0.9985852297991026,
2708
+ "grad_norm": 0.7500579953193665,
2709
+ "learning_rate": 8.436678670527463e-06,
2710
+ "loss": 2.0736,
2711
+ "step": 7720
2712
+ }
2713
+ ],
2714
+ "logging_steps": 20,
2715
+ "max_steps": 23190,
2716
+ "num_input_tokens_seen": 0,
2717
+ "num_train_epochs": 3,
2718
+ "save_steps": 500,
2719
+ "stateful_callbacks": {
2720
+ "TrainerControl": {
2721
+ "args": {
2722
+ "should_epoch_stop": false,
2723
+ "should_evaluate": false,
2724
+ "should_log": false,
2725
+ "should_save": true,
2726
+ "should_training_stop": false
2727
+ },
2728
+ "attributes": {}
2729
+ }
2730
+ },
2731
+ "total_flos": 3.963409237367639e+19,
2732
+ "train_batch_size": 12,
2733
+ "trial_name": null,
2734
+ "trial_params": null
2735
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26ba0ad3aaaa7cbff9bc1003d64820cbd0ec396b0565bcc66973c21f42aab888
3
+ size 6712
vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
zero_to_fp32.py ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
215
+ exclude_frozen_parameters)
216
+ elif zero_stage == 3:
217
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
218
+ exclude_frozen_parameters)
219
+
220
+
221
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
222
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
223
+ return
224
+
225
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
226
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
227
+
228
+ if debug:
229
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
230
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
231
+
232
+ wanted_params = len(frozen_param_shapes)
233
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
234
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
235
+ print(f'Frozen params: Have {avail_numel} numels to process.')
236
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
237
+
238
+ total_params = 0
239
+ total_numel = 0
240
+ for name, shape in frozen_param_shapes.items():
241
+ total_params += 1
242
+ unpartitioned_numel = shape.numel()
243
+ total_numel += unpartitioned_numel
244
+
245
+ state_dict[name] = frozen_param_fragments[name]
246
+
247
+ if debug:
248
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
249
+
250
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
251
+
252
+
253
+ def _has_callable(obj, fn):
254
+ attr = getattr(obj, fn, None)
255
+ return callable(attr)
256
+
257
+
258
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
259
+ param_shapes = zero_model_states[0].param_shapes
260
+
261
+ # Reconstruction protocol:
262
+ #
263
+ # XXX: document this
264
+
265
+ if debug:
266
+ for i in range(world_size):
267
+ for j in range(len(fp32_flat_groups[0])):
268
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
269
+
270
+ # XXX: memory usage doubles here (zero2)
271
+ num_param_groups = len(fp32_flat_groups[0])
272
+ merged_single_partition_of_fp32_groups = []
273
+ for i in range(num_param_groups):
274
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
275
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
276
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
277
+ avail_numel = sum(
278
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
279
+
280
+ if debug:
281
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
282
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
283
+ # not asserting if there is a mismatch due to possible padding
284
+ print(f"Have {avail_numel} numels to process.")
285
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
286
+
287
+ # params
288
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
289
+ # out-of-core computing solution
290
+ total_numel = 0
291
+ total_params = 0
292
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
293
+ offset = 0
294
+ avail_numel = full_single_fp32_vector.numel()
295
+ for name, shape in shapes.items():
296
+
297
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
298
+ total_numel += unpartitioned_numel
299
+ total_params += 1
300
+
301
+ if debug:
302
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
303
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
304
+ offset += unpartitioned_numel
305
+
306
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
307
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
308
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
309
+ # live optimizer object, so we are checking that the numbers are within the right range
310
+ align_to = 2 * world_size
311
+
312
+ def zero2_align(x):
313
+ return align_to * math.ceil(x / align_to)
314
+
315
+ if debug:
316
+ print(f"original offset={offset}, avail_numel={avail_numel}")
317
+
318
+ offset = zero2_align(offset)
319
+ avail_numel = zero2_align(avail_numel)
320
+
321
+ if debug:
322
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
323
+
324
+ # Sanity check
325
+ if offset != avail_numel:
326
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
327
+
328
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
329
+
330
+
331
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
332
+ exclude_frozen_parameters):
333
+ state_dict = OrderedDict()
334
+
335
+ # buffers
336
+ buffers = zero_model_states[0].buffers
337
+ state_dict.update(buffers)
338
+ if debug:
339
+ print(f"added {len(buffers)} buffers")
340
+
341
+ if not exclude_frozen_parameters:
342
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
343
+
344
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
345
+
346
+ # recover shared parameters
347
+ for pair in zero_model_states[0].shared_params:
348
+ if pair[1] in state_dict:
349
+ state_dict[pair[0]] = state_dict[pair[1]]
350
+
351
+ return state_dict
352
+
353
+
354
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
355
+ remainder = unpartitioned_numel % world_size
356
+ padding_numel = (world_size - remainder) if remainder else 0
357
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
358
+ return partitioned_numel, padding_numel
359
+
360
+
361
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
362
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
363
+ return
364
+
365
+ if debug:
366
+ for i in range(world_size):
367
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
368
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
369
+
370
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
371
+ wanted_params = len(frozen_param_shapes)
372
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
373
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
374
+ print(f'Frozen params: Have {avail_numel} numels to process.')
375
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
376
+
377
+ total_params = 0
378
+ total_numel = 0
379
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
380
+ total_params += 1
381
+ unpartitioned_numel = shape.numel()
382
+ total_numel += unpartitioned_numel
383
+
384
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
385
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
386
+
387
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
388
+
389
+ if debug:
390
+ print(
391
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
392
+ )
393
+
394
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
395
+
396
+
397
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
398
+ param_shapes = zero_model_states[0].param_shapes
399
+ avail_numel = fp32_flat_groups[0].numel() * world_size
400
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
401
+ # param, re-consolidating each param, while dealing with padding if any
402
+
403
+ # merge list of dicts, preserving order
404
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
405
+
406
+ if debug:
407
+ for i in range(world_size):
408
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
409
+
410
+ wanted_params = len(param_shapes)
411
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
412
+ # not asserting if there is a mismatch due to possible padding
413
+ avail_numel = fp32_flat_groups[0].numel() * world_size
414
+ print(f"Trainable params: Have {avail_numel} numels to process.")
415
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
416
+
417
+ # params
418
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
419
+ # out-of-core computing solution
420
+ offset = 0
421
+ total_numel = 0
422
+ total_params = 0
423
+ for name, shape in param_shapes.items():
424
+
425
+ unpartitioned_numel = shape.numel()
426
+ total_numel += unpartitioned_numel
427
+ total_params += 1
428
+
429
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
430
+
431
+ if debug:
432
+ print(
433
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
434
+ )
435
+
436
+ # XXX: memory usage doubles here
437
+ state_dict[name] = torch.cat(
438
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
439
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
440
+ offset += partitioned_numel
441
+
442
+ offset *= world_size
443
+
444
+ # Sanity check
445
+ if offset != avail_numel:
446
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
447
+
448
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
449
+
450
+
451
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
452
+ exclude_frozen_parameters):
453
+ state_dict = OrderedDict()
454
+
455
+ # buffers
456
+ buffers = zero_model_states[0].buffers
457
+ state_dict.update(buffers)
458
+ if debug:
459
+ print(f"added {len(buffers)} buffers")
460
+
461
+ if not exclude_frozen_parameters:
462
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
463
+
464
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
465
+
466
+ # recover shared parameters
467
+ for pair in zero_model_states[0].shared_params:
468
+ if pair[1] in state_dict:
469
+ state_dict[pair[0]] = state_dict[pair[1]]
470
+
471
+ return state_dict
472
+
473
+
474
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
475
+ """
476
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
477
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
478
+ via a model hub.
479
+
480
+ Args:
481
+ - ``checkpoint_dir``: path to the desired checkpoint folder
482
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
483
+ - ``exclude_frozen_parameters``: exclude frozen parameters
484
+
485
+ Returns:
486
+ - pytorch ``state_dict``
487
+
488
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
489
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
490
+ the checkpoint.
491
+
492
+ A typical usage might be ::
493
+
494
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
495
+ # do the training and checkpoint saving
496
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
497
+ model = model.cpu() # move to cpu
498
+ model.load_state_dict(state_dict)
499
+ # submit to model hub or save the model to share with others
500
+
501
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
502
+ application. i.e. you will need to re-initialize the deepspeed engine, since
503
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
504
+
505
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
506
+
507
+ """
508
+ if tag is None:
509
+ latest_path = os.path.join(checkpoint_dir, 'latest')
510
+ if os.path.isfile(latest_path):
511
+ with open(latest_path, 'r') as fd:
512
+ tag = fd.read().strip()
513
+ else:
514
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
515
+
516
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
517
+
518
+ if not os.path.isdir(ds_checkpoint_dir):
519
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
520
+
521
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
522
+
523
+
524
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
525
+ """
526
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
527
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
528
+
529
+ Args:
530
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
531
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
532
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
533
+ - ``exclude_frozen_parameters``: exclude frozen parameters
534
+ """
535
+
536
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
537
+ print(f"Saving fp32 state dict to {output_file}")
538
+ torch.save(state_dict, output_file)
539
+
540
+
541
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
542
+ """
543
+ 1. Put the provided model to cpu
544
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
545
+ 3. Load it into the provided model
546
+
547
+ Args:
548
+ - ``model``: the model object to update
549
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
550
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
551
+
552
+ Returns:
553
+ - ``model`: modified model
554
+
555
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
556
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
557
+ conveniently placed for you in the checkpoint folder.
558
+
559
+ A typical usage might be ::
560
+
561
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
562
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
563
+ # submit to model hub or save the model to share with others
564
+
565
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
566
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
567
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
568
+
569
+ """
570
+ logger.info(f"Extracting fp32 weights")
571
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
572
+
573
+ logger.info(f"Overwriting model with fp32 weights")
574
+ model = model.cpu()
575
+ model.load_state_dict(state_dict, strict=False)
576
+
577
+ return model
578
+
579
+
580
+ if __name__ == "__main__":
581
+
582
+ parser = argparse.ArgumentParser()
583
+ parser.add_argument("checkpoint_dir",
584
+ type=str,
585
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
586
+ parser.add_argument(
587
+ "output_file",
588
+ type=str,
589
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
590
+ parser.add_argument("-t",
591
+ "--tag",
592
+ type=str,
593
+ default=None,
594
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
595
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
596
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
597
+ args = parser.parse_args()
598
+
599
+ debug = args.debug
600
+
601
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
602
+ args.output_file,
603
+ tag=args.tag,
604
+ exclude_frozen_parameters=args.exclude_frozen_parameters)