kaiyuanzh commited on
Commit
475ab5f
·
verified ·
1 Parent(s): 0995471

Training in progress, step 200, checkpoint

Browse files
checkpoint-200/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: GSAI-ML/LLaDA-8B-Instruct
3
+ library_name: peft
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+ ### Framework versions
201
+
202
+ - PEFT 0.15.1
checkpoint-200/adapter_config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "GSAI-ML/LLaDA-8B-Instruct",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 64,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.05,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 128,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "o_proj",
28
+ "up_proj",
29
+ "gate_proj",
30
+ "k_proj",
31
+ "down_proj",
32
+ "v_proj",
33
+ "q_proj"
34
+ ],
35
+ "task_type": "CAUSAL_LM",
36
+ "trainable_token_indices": null,
37
+ "use_dora": false,
38
+ "use_rslora": false
39
+ }
checkpoint-200/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3d4efb7d07b9fedcd31d422066a0e68ff1a3f77186cf4ba0beb698f6df5f14e
3
+ size 335580296
checkpoint-200/global_step200/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea0f6ab5f1e6588c75303ef3f81060f036b74857f54100825e8edc9c93d7a850
3
+ size 251665797
checkpoint-200/global_step200/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c7c982c4e70a3c80ca2e38665f83b5f44ad91ed9341b837551a4cf5da23667f
3
+ size 251665797
checkpoint-200/global_step200/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8081effe6322c14d3e70c7fd053bf1ffee317c8944368da273591065034e0ffb
3
+ size 251665797
checkpoint-200/global_step200/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d42d3edcb6dbaab6caf5e2ccd435864bb493adf5bb8b560bb34404b5d4e1e77a
3
+ size 251665861
checkpoint-200/global_step200/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6995615e51e540526b6e7d44a29e2338d61f2b93dc72fb2c6ce17d2a6f8f9830
3
+ size 251665861
checkpoint-200/global_step200/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2982bb2558f035307d65024ad3aee5b1226e832b082ec49d320e911f2045e711
3
+ size 251665861
checkpoint-200/global_step200/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:533a8d152bf66e0894ff7789941648154d7015012b9e01365258ed74e00d7890
3
+ size 251665861
checkpoint-200/global_step200/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:044e833d455d3e6657ba5f41471ee53326cf84b09dd7399149f917a1dd8516c8
3
+ size 251665861
checkpoint-200/global_step200/mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b2c39fb81601a9f23eafac0fde7ade80ef3ab3a0b6bb42bf843b3bf2750ba45
3
+ size 447220134
checkpoint-200/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step200
checkpoint-200/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99e5c549e0caef505cd965e796c7cf3b1c92466ab02e67d5c823380a866218db
3
+ size 16389
checkpoint-200/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e96568eb8163220619dfb7a6015a28aca6218cb0f0b7450f6e88761c4ec6782b
3
+ size 16389
checkpoint-200/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28866c94bd7980764735c025b28edd8ecdfd6c230918a0f328e2f4360c85ca20
3
+ size 16325
checkpoint-200/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2aa9b4f131fc779cdfb8cb7d8f007b18d5968136e2aaa090467d30116f73a3a
3
+ size 16325
checkpoint-200/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d70532ec46170f14c74589b96f47d5bc9e8df47e75bd431295676396648cac27
3
+ size 16389
checkpoint-200/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43ada95516598176819bb9d0e62ddd1b4e85cc12c1acf0b251b49c164c22b973
3
+ size 16325
checkpoint-200/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5dcc533ad9e488c961f8ea1e80dea393702d24beb0a9abd5de870e96615a20bc
3
+ size 16325
checkpoint-200/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43e7022b7c4417105bdae6eac032bb95de37ebd5a5ef567fdfecd9a8613e373b
3
+ size 16325
checkpoint-200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9aa91156c5c3363084222b9a829c114c05e65d59a7fcc7e609782c7351439aa5
3
+ size 1401
checkpoint-200/special_tokens_map.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<role>",
4
+ "</role>",
5
+ "<|arithmetic_start|>",
6
+ "<|arithmetic_end|>",
7
+ "<|number_start|>",
8
+ "<|number_end|>"
9
+ ],
10
+ "bos_token": {
11
+ "content": "<|startoftext|>",
12
+ "lstrip": false,
13
+ "normalized": false,
14
+ "rstrip": false,
15
+ "single_word": false
16
+ },
17
+ "cls_token": {
18
+ "content": "[CLS]",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "eos_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ },
31
+ "pad_token": {
32
+ "content": "<|endoftext|>",
33
+ "lstrip": false,
34
+ "normalized": false,
35
+ "rstrip": false,
36
+ "single_word": false
37
+ }
38
+ }
checkpoint-200/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-200/tokenizer_config.json ADDED
@@ -0,0 +1,2184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "126080": {
6
+ "content": "<|startoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "126081": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "126082": {
22
+ "content": "[CLS]",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "126083": {
30
+ "content": "[gMASK]",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "126084": {
38
+ "content": "<|reserved_token_0|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "126085": {
46
+ "content": "<|reserved_token_1|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "126086": {
54
+ "content": "<|reserved_token_2|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "126087": {
62
+ "content": "<|reserved_token_3|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "126088": {
70
+ "content": "<|reserved_token_4|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "126089": {
78
+ "content": "<|reserved_token_5|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "126090": {
86
+ "content": "<|reserved_token_6|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "126091": {
94
+ "content": "<|reserved_token_7|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "126092": {
102
+ "content": "<|reserved_token_8|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "126093": {
110
+ "content": "<|reserved_token_9|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "126094": {
118
+ "content": "<|reserved_token_10|>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": true
124
+ },
125
+ "126095": {
126
+ "content": "<|reserved_token_11|>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": true
132
+ },
133
+ "126096": {
134
+ "content": "<|reserved_token_12|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": true
140
+ },
141
+ "126097": {
142
+ "content": "<|reserved_token_13|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": true
148
+ },
149
+ "126098": {
150
+ "content": "<|reserved_token_14|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": true
156
+ },
157
+ "126099": {
158
+ "content": "<|reserved_token_15|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": true
164
+ },
165
+ "126100": {
166
+ "content": "<|reserved_token_16|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": true
172
+ },
173
+ "126101": {
174
+ "content": "<|reserved_token_17|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": true
180
+ },
181
+ "126102": {
182
+ "content": "<|reserved_token_18|>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": true
188
+ },
189
+ "126103": {
190
+ "content": "<|reserved_token_19|>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": true
196
+ },
197
+ "126104": {
198
+ "content": "<|reserved_token_20|>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": true
204
+ },
205
+ "126105": {
206
+ "content": "<|reserved_token_21|>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": true
212
+ },
213
+ "126106": {
214
+ "content": "<|reserved_token_22|>",
215
+ "lstrip": false,
216
+ "normalized": false,
217
+ "rstrip": false,
218
+ "single_word": false,
219
+ "special": true
220
+ },
221
+ "126107": {
222
+ "content": "<|reserved_token_23|>",
223
+ "lstrip": false,
224
+ "normalized": false,
225
+ "rstrip": false,
226
+ "single_word": false,
227
+ "special": true
228
+ },
229
+ "126108": {
230
+ "content": "<|reserved_token_24|>",
231
+ "lstrip": false,
232
+ "normalized": false,
233
+ "rstrip": false,
234
+ "single_word": false,
235
+ "special": true
236
+ },
237
+ "126109": {
238
+ "content": "<|reserved_token_25|>",
239
+ "lstrip": false,
240
+ "normalized": false,
241
+ "rstrip": false,
242
+ "single_word": false,
243
+ "special": true
244
+ },
245
+ "126110": {
246
+ "content": "<|reserved_token_26|>",
247
+ "lstrip": false,
248
+ "normalized": false,
249
+ "rstrip": false,
250
+ "single_word": false,
251
+ "special": true
252
+ },
253
+ "126111": {
254
+ "content": "<|reserved_token_27|>",
255
+ "lstrip": false,
256
+ "normalized": false,
257
+ "rstrip": false,
258
+ "single_word": false,
259
+ "special": true
260
+ },
261
+ "126112": {
262
+ "content": "<|reserved_token_28|>",
263
+ "lstrip": false,
264
+ "normalized": false,
265
+ "rstrip": false,
266
+ "single_word": false,
267
+ "special": true
268
+ },
269
+ "126113": {
270
+ "content": "<|reserved_token_29|>",
271
+ "lstrip": false,
272
+ "normalized": false,
273
+ "rstrip": false,
274
+ "single_word": false,
275
+ "special": true
276
+ },
277
+ "126114": {
278
+ "content": "<|reserved_token_30|>",
279
+ "lstrip": false,
280
+ "normalized": false,
281
+ "rstrip": false,
282
+ "single_word": false,
283
+ "special": true
284
+ },
285
+ "126115": {
286
+ "content": "<|reserved_token_31|>",
287
+ "lstrip": false,
288
+ "normalized": false,
289
+ "rstrip": false,
290
+ "single_word": false,
291
+ "special": true
292
+ },
293
+ "126116": {
294
+ "content": "<|reserved_token_32|>",
295
+ "lstrip": false,
296
+ "normalized": false,
297
+ "rstrip": false,
298
+ "single_word": false,
299
+ "special": true
300
+ },
301
+ "126117": {
302
+ "content": "<|reserved_token_33|>",
303
+ "lstrip": false,
304
+ "normalized": false,
305
+ "rstrip": false,
306
+ "single_word": false,
307
+ "special": true
308
+ },
309
+ "126118": {
310
+ "content": "<|reserved_token_34|>",
311
+ "lstrip": false,
312
+ "normalized": false,
313
+ "rstrip": false,
314
+ "single_word": false,
315
+ "special": true
316
+ },
317
+ "126119": {
318
+ "content": "<|reserved_token_35|>",
319
+ "lstrip": false,
320
+ "normalized": false,
321
+ "rstrip": false,
322
+ "single_word": false,
323
+ "special": true
324
+ },
325
+ "126120": {
326
+ "content": "<|reserved_token_36|>",
327
+ "lstrip": false,
328
+ "normalized": false,
329
+ "rstrip": false,
330
+ "single_word": false,
331
+ "special": true
332
+ },
333
+ "126121": {
334
+ "content": "<|reserved_token_37|>",
335
+ "lstrip": false,
336
+ "normalized": false,
337
+ "rstrip": false,
338
+ "single_word": false,
339
+ "special": true
340
+ },
341
+ "126122": {
342
+ "content": "<|reserved_token_38|>",
343
+ "lstrip": false,
344
+ "normalized": false,
345
+ "rstrip": false,
346
+ "single_word": false,
347
+ "special": true
348
+ },
349
+ "126123": {
350
+ "content": "<|reserved_token_39|>",
351
+ "lstrip": false,
352
+ "normalized": false,
353
+ "rstrip": false,
354
+ "single_word": false,
355
+ "special": true
356
+ },
357
+ "126124": {
358
+ "content": "<|reserved_token_40|>",
359
+ "lstrip": false,
360
+ "normalized": false,
361
+ "rstrip": false,
362
+ "single_word": false,
363
+ "special": true
364
+ },
365
+ "126125": {
366
+ "content": "<|reserved_token_41|>",
367
+ "lstrip": false,
368
+ "normalized": false,
369
+ "rstrip": false,
370
+ "single_word": false,
371
+ "special": true
372
+ },
373
+ "126126": {
374
+ "content": "<|reserved_token_42|>",
375
+ "lstrip": false,
376
+ "normalized": false,
377
+ "rstrip": false,
378
+ "single_word": false,
379
+ "special": true
380
+ },
381
+ "126127": {
382
+ "content": "<|reserved_token_43|>",
383
+ "lstrip": false,
384
+ "normalized": false,
385
+ "rstrip": false,
386
+ "single_word": false,
387
+ "special": true
388
+ },
389
+ "126128": {
390
+ "content": "<|reserved_token_44|>",
391
+ "lstrip": false,
392
+ "normalized": false,
393
+ "rstrip": false,
394
+ "single_word": false,
395
+ "special": true
396
+ },
397
+ "126129": {
398
+ "content": "<|reserved_token_45|>",
399
+ "lstrip": false,
400
+ "normalized": false,
401
+ "rstrip": false,
402
+ "single_word": false,
403
+ "special": true
404
+ },
405
+ "126130": {
406
+ "content": "<|reserved_token_46|>",
407
+ "lstrip": false,
408
+ "normalized": false,
409
+ "rstrip": false,
410
+ "single_word": false,
411
+ "special": true
412
+ },
413
+ "126131": {
414
+ "content": "<|reserved_token_47|>",
415
+ "lstrip": false,
416
+ "normalized": false,
417
+ "rstrip": false,
418
+ "single_word": false,
419
+ "special": true
420
+ },
421
+ "126132": {
422
+ "content": "<|reserved_token_48|>",
423
+ "lstrip": false,
424
+ "normalized": false,
425
+ "rstrip": false,
426
+ "single_word": false,
427
+ "special": true
428
+ },
429
+ "126133": {
430
+ "content": "<|reserved_token_49|>",
431
+ "lstrip": false,
432
+ "normalized": false,
433
+ "rstrip": false,
434
+ "single_word": false,
435
+ "special": true
436
+ },
437
+ "126134": {
438
+ "content": "<|reserved_token_50|>",
439
+ "lstrip": false,
440
+ "normalized": false,
441
+ "rstrip": false,
442
+ "single_word": false,
443
+ "special": true
444
+ },
445
+ "126135": {
446
+ "content": "<|reserved_token_51|>",
447
+ "lstrip": false,
448
+ "normalized": false,
449
+ "rstrip": false,
450
+ "single_word": false,
451
+ "special": true
452
+ },
453
+ "126136": {
454
+ "content": "<|reserved_token_52|>",
455
+ "lstrip": false,
456
+ "normalized": false,
457
+ "rstrip": false,
458
+ "single_word": false,
459
+ "special": true
460
+ },
461
+ "126137": {
462
+ "content": "<|reserved_token_53|>",
463
+ "lstrip": false,
464
+ "normalized": false,
465
+ "rstrip": false,
466
+ "single_word": false,
467
+ "special": true
468
+ },
469
+ "126138": {
470
+ "content": "<|reserved_token_54|>",
471
+ "lstrip": false,
472
+ "normalized": false,
473
+ "rstrip": false,
474
+ "single_word": false,
475
+ "special": true
476
+ },
477
+ "126139": {
478
+ "content": "<|reserved_token_55|>",
479
+ "lstrip": false,
480
+ "normalized": false,
481
+ "rstrip": false,
482
+ "single_word": false,
483
+ "special": true
484
+ },
485
+ "126140": {
486
+ "content": "<|reserved_token_56|>",
487
+ "lstrip": false,
488
+ "normalized": false,
489
+ "rstrip": false,
490
+ "single_word": false,
491
+ "special": true
492
+ },
493
+ "126141": {
494
+ "content": "<|reserved_token_57|>",
495
+ "lstrip": false,
496
+ "normalized": false,
497
+ "rstrip": false,
498
+ "single_word": false,
499
+ "special": true
500
+ },
501
+ "126142": {
502
+ "content": "<|reserved_token_58|>",
503
+ "lstrip": false,
504
+ "normalized": false,
505
+ "rstrip": false,
506
+ "single_word": false,
507
+ "special": true
508
+ },
509
+ "126143": {
510
+ "content": "<|reserved_token_59|>",
511
+ "lstrip": false,
512
+ "normalized": false,
513
+ "rstrip": false,
514
+ "single_word": false,
515
+ "special": true
516
+ },
517
+ "126144": {
518
+ "content": "<|reserved_token_60|>",
519
+ "lstrip": false,
520
+ "normalized": false,
521
+ "rstrip": false,
522
+ "single_word": false,
523
+ "special": true
524
+ },
525
+ "126145": {
526
+ "content": "<|reserved_token_61|>",
527
+ "lstrip": false,
528
+ "normalized": false,
529
+ "rstrip": false,
530
+ "single_word": false,
531
+ "special": true
532
+ },
533
+ "126146": {
534
+ "content": "<|reserved_token_62|>",
535
+ "lstrip": false,
536
+ "normalized": false,
537
+ "rstrip": false,
538
+ "single_word": false,
539
+ "special": true
540
+ },
541
+ "126147": {
542
+ "content": "<|reserved_token_63|>",
543
+ "lstrip": false,
544
+ "normalized": false,
545
+ "rstrip": false,
546
+ "single_word": false,
547
+ "special": true
548
+ },
549
+ "126148": {
550
+ "content": "<|reserved_token_64|>",
551
+ "lstrip": false,
552
+ "normalized": false,
553
+ "rstrip": false,
554
+ "single_word": false,
555
+ "special": true
556
+ },
557
+ "126149": {
558
+ "content": "<|reserved_token_65|>",
559
+ "lstrip": false,
560
+ "normalized": false,
561
+ "rstrip": false,
562
+ "single_word": false,
563
+ "special": true
564
+ },
565
+ "126150": {
566
+ "content": "<|reserved_token_66|>",
567
+ "lstrip": false,
568
+ "normalized": false,
569
+ "rstrip": false,
570
+ "single_word": false,
571
+ "special": true
572
+ },
573
+ "126151": {
574
+ "content": "<|reserved_token_67|>",
575
+ "lstrip": false,
576
+ "normalized": false,
577
+ "rstrip": false,
578
+ "single_word": false,
579
+ "special": true
580
+ },
581
+ "126152": {
582
+ "content": "<|reserved_token_68|>",
583
+ "lstrip": false,
584
+ "normalized": false,
585
+ "rstrip": false,
586
+ "single_word": false,
587
+ "special": true
588
+ },
589
+ "126153": {
590
+ "content": "<|reserved_token_69|>",
591
+ "lstrip": false,
592
+ "normalized": false,
593
+ "rstrip": false,
594
+ "single_word": false,
595
+ "special": true
596
+ },
597
+ "126154": {
598
+ "content": "<|reserved_token_70|>",
599
+ "lstrip": false,
600
+ "normalized": false,
601
+ "rstrip": false,
602
+ "single_word": false,
603
+ "special": true
604
+ },
605
+ "126155": {
606
+ "content": "<|reserved_token_71|>",
607
+ "lstrip": false,
608
+ "normalized": false,
609
+ "rstrip": false,
610
+ "single_word": false,
611
+ "special": true
612
+ },
613
+ "126156": {
614
+ "content": "<|reserved_token_72|>",
615
+ "lstrip": false,
616
+ "normalized": false,
617
+ "rstrip": false,
618
+ "single_word": false,
619
+ "special": true
620
+ },
621
+ "126157": {
622
+ "content": "<|reserved_token_73|>",
623
+ "lstrip": false,
624
+ "normalized": false,
625
+ "rstrip": false,
626
+ "single_word": false,
627
+ "special": true
628
+ },
629
+ "126158": {
630
+ "content": "<|reserved_token_74|>",
631
+ "lstrip": false,
632
+ "normalized": false,
633
+ "rstrip": false,
634
+ "single_word": false,
635
+ "special": true
636
+ },
637
+ "126159": {
638
+ "content": "<|reserved_token_75|>",
639
+ "lstrip": false,
640
+ "normalized": false,
641
+ "rstrip": false,
642
+ "single_word": false,
643
+ "special": true
644
+ },
645
+ "126160": {
646
+ "content": "<|reserved_token_76|>",
647
+ "lstrip": false,
648
+ "normalized": false,
649
+ "rstrip": false,
650
+ "single_word": false,
651
+ "special": true
652
+ },
653
+ "126161": {
654
+ "content": "<|reserved_token_77|>",
655
+ "lstrip": false,
656
+ "normalized": false,
657
+ "rstrip": false,
658
+ "single_word": false,
659
+ "special": true
660
+ },
661
+ "126162": {
662
+ "content": "<|reserved_token_78|>",
663
+ "lstrip": false,
664
+ "normalized": false,
665
+ "rstrip": false,
666
+ "single_word": false,
667
+ "special": true
668
+ },
669
+ "126163": {
670
+ "content": "<|reserved_token_79|>",
671
+ "lstrip": false,
672
+ "normalized": false,
673
+ "rstrip": false,
674
+ "single_word": false,
675
+ "special": true
676
+ },
677
+ "126164": {
678
+ "content": "<|reserved_token_80|>",
679
+ "lstrip": false,
680
+ "normalized": false,
681
+ "rstrip": false,
682
+ "single_word": false,
683
+ "special": true
684
+ },
685
+ "126165": {
686
+ "content": "<|reserved_token_81|>",
687
+ "lstrip": false,
688
+ "normalized": false,
689
+ "rstrip": false,
690
+ "single_word": false,
691
+ "special": true
692
+ },
693
+ "126166": {
694
+ "content": "<|reserved_token_82|>",
695
+ "lstrip": false,
696
+ "normalized": false,
697
+ "rstrip": false,
698
+ "single_word": false,
699
+ "special": true
700
+ },
701
+ "126167": {
702
+ "content": "<|reserved_token_83|>",
703
+ "lstrip": false,
704
+ "normalized": false,
705
+ "rstrip": false,
706
+ "single_word": false,
707
+ "special": true
708
+ },
709
+ "126168": {
710
+ "content": "<|reserved_token_84|>",
711
+ "lstrip": false,
712
+ "normalized": false,
713
+ "rstrip": false,
714
+ "single_word": false,
715
+ "special": true
716
+ },
717
+ "126169": {
718
+ "content": "<|reserved_token_85|>",
719
+ "lstrip": false,
720
+ "normalized": false,
721
+ "rstrip": false,
722
+ "single_word": false,
723
+ "special": true
724
+ },
725
+ "126170": {
726
+ "content": "<|reserved_token_86|>",
727
+ "lstrip": false,
728
+ "normalized": false,
729
+ "rstrip": false,
730
+ "single_word": false,
731
+ "special": true
732
+ },
733
+ "126171": {
734
+ "content": "<|reserved_token_87|>",
735
+ "lstrip": false,
736
+ "normalized": false,
737
+ "rstrip": false,
738
+ "single_word": false,
739
+ "special": true
740
+ },
741
+ "126172": {
742
+ "content": "<|reserved_token_88|>",
743
+ "lstrip": false,
744
+ "normalized": false,
745
+ "rstrip": false,
746
+ "single_word": false,
747
+ "special": true
748
+ },
749
+ "126173": {
750
+ "content": "<|reserved_token_89|>",
751
+ "lstrip": false,
752
+ "normalized": false,
753
+ "rstrip": false,
754
+ "single_word": false,
755
+ "special": true
756
+ },
757
+ "126174": {
758
+ "content": "<|reserved_token_90|>",
759
+ "lstrip": false,
760
+ "normalized": false,
761
+ "rstrip": false,
762
+ "single_word": false,
763
+ "special": true
764
+ },
765
+ "126175": {
766
+ "content": "<|reserved_token_91|>",
767
+ "lstrip": false,
768
+ "normalized": false,
769
+ "rstrip": false,
770
+ "single_word": false,
771
+ "special": true
772
+ },
773
+ "126176": {
774
+ "content": "<|reserved_token_92|>",
775
+ "lstrip": false,
776
+ "normalized": false,
777
+ "rstrip": false,
778
+ "single_word": false,
779
+ "special": true
780
+ },
781
+ "126177": {
782
+ "content": "<|reserved_token_93|>",
783
+ "lstrip": false,
784
+ "normalized": false,
785
+ "rstrip": false,
786
+ "single_word": false,
787
+ "special": true
788
+ },
789
+ "126178": {
790
+ "content": "<|reserved_token_94|>",
791
+ "lstrip": false,
792
+ "normalized": false,
793
+ "rstrip": false,
794
+ "single_word": false,
795
+ "special": true
796
+ },
797
+ "126179": {
798
+ "content": "<|reserved_token_95|>",
799
+ "lstrip": false,
800
+ "normalized": false,
801
+ "rstrip": false,
802
+ "single_word": false,
803
+ "special": true
804
+ },
805
+ "126180": {
806
+ "content": "<|reserved_token_96|>",
807
+ "lstrip": false,
808
+ "normalized": false,
809
+ "rstrip": false,
810
+ "single_word": false,
811
+ "special": true
812
+ },
813
+ "126181": {
814
+ "content": "<|reserved_token_97|>",
815
+ "lstrip": false,
816
+ "normalized": false,
817
+ "rstrip": false,
818
+ "single_word": false,
819
+ "special": true
820
+ },
821
+ "126182": {
822
+ "content": "<|reserved_token_98|>",
823
+ "lstrip": false,
824
+ "normalized": false,
825
+ "rstrip": false,
826
+ "single_word": false,
827
+ "special": true
828
+ },
829
+ "126183": {
830
+ "content": "<|reserved_token_99|>",
831
+ "lstrip": false,
832
+ "normalized": false,
833
+ "rstrip": false,
834
+ "single_word": false,
835
+ "special": true
836
+ },
837
+ "126184": {
838
+ "content": "<|reserved_token_100|>",
839
+ "lstrip": false,
840
+ "normalized": false,
841
+ "rstrip": false,
842
+ "single_word": false,
843
+ "special": true
844
+ },
845
+ "126185": {
846
+ "content": "<|reserved_token_101|>",
847
+ "lstrip": false,
848
+ "normalized": false,
849
+ "rstrip": false,
850
+ "single_word": false,
851
+ "special": true
852
+ },
853
+ "126186": {
854
+ "content": "<|reserved_token_102|>",
855
+ "lstrip": false,
856
+ "normalized": false,
857
+ "rstrip": false,
858
+ "single_word": false,
859
+ "special": true
860
+ },
861
+ "126187": {
862
+ "content": "<|reserved_token_103|>",
863
+ "lstrip": false,
864
+ "normalized": false,
865
+ "rstrip": false,
866
+ "single_word": false,
867
+ "special": true
868
+ },
869
+ "126188": {
870
+ "content": "<|reserved_token_104|>",
871
+ "lstrip": false,
872
+ "normalized": false,
873
+ "rstrip": false,
874
+ "single_word": false,
875
+ "special": true
876
+ },
877
+ "126189": {
878
+ "content": "<|reserved_token_105|>",
879
+ "lstrip": false,
880
+ "normalized": false,
881
+ "rstrip": false,
882
+ "single_word": false,
883
+ "special": true
884
+ },
885
+ "126190": {
886
+ "content": "<|reserved_token_106|>",
887
+ "lstrip": false,
888
+ "normalized": false,
889
+ "rstrip": false,
890
+ "single_word": false,
891
+ "special": true
892
+ },
893
+ "126191": {
894
+ "content": "<|reserved_token_107|>",
895
+ "lstrip": false,
896
+ "normalized": false,
897
+ "rstrip": false,
898
+ "single_word": false,
899
+ "special": true
900
+ },
901
+ "126192": {
902
+ "content": "<|reserved_token_108|>",
903
+ "lstrip": false,
904
+ "normalized": false,
905
+ "rstrip": false,
906
+ "single_word": false,
907
+ "special": true
908
+ },
909
+ "126193": {
910
+ "content": "<|reserved_token_109|>",
911
+ "lstrip": false,
912
+ "normalized": false,
913
+ "rstrip": false,
914
+ "single_word": false,
915
+ "special": true
916
+ },
917
+ "126194": {
918
+ "content": "<|reserved_token_110|>",
919
+ "lstrip": false,
920
+ "normalized": false,
921
+ "rstrip": false,
922
+ "single_word": false,
923
+ "special": true
924
+ },
925
+ "126195": {
926
+ "content": "<|reserved_token_111|>",
927
+ "lstrip": false,
928
+ "normalized": false,
929
+ "rstrip": false,
930
+ "single_word": false,
931
+ "special": true
932
+ },
933
+ "126196": {
934
+ "content": "<|reserved_token_112|>",
935
+ "lstrip": false,
936
+ "normalized": false,
937
+ "rstrip": false,
938
+ "single_word": false,
939
+ "special": true
940
+ },
941
+ "126197": {
942
+ "content": "<|reserved_token_113|>",
943
+ "lstrip": false,
944
+ "normalized": false,
945
+ "rstrip": false,
946
+ "single_word": false,
947
+ "special": true
948
+ },
949
+ "126198": {
950
+ "content": "<|reserved_token_114|>",
951
+ "lstrip": false,
952
+ "normalized": false,
953
+ "rstrip": false,
954
+ "single_word": false,
955
+ "special": true
956
+ },
957
+ "126199": {
958
+ "content": "<|reserved_token_115|>",
959
+ "lstrip": false,
960
+ "normalized": false,
961
+ "rstrip": false,
962
+ "single_word": false,
963
+ "special": true
964
+ },
965
+ "126200": {
966
+ "content": "<|reserved_token_116|>",
967
+ "lstrip": false,
968
+ "normalized": false,
969
+ "rstrip": false,
970
+ "single_word": false,
971
+ "special": true
972
+ },
973
+ "126201": {
974
+ "content": "<|reserved_token_117|>",
975
+ "lstrip": false,
976
+ "normalized": false,
977
+ "rstrip": false,
978
+ "single_word": false,
979
+ "special": true
980
+ },
981
+ "126202": {
982
+ "content": "<|reserved_token_118|>",
983
+ "lstrip": false,
984
+ "normalized": false,
985
+ "rstrip": false,
986
+ "single_word": false,
987
+ "special": true
988
+ },
989
+ "126203": {
990
+ "content": "<|reserved_token_119|>",
991
+ "lstrip": false,
992
+ "normalized": false,
993
+ "rstrip": false,
994
+ "single_word": false,
995
+ "special": true
996
+ },
997
+ "126204": {
998
+ "content": "<|reserved_token_120|>",
999
+ "lstrip": false,
1000
+ "normalized": false,
1001
+ "rstrip": false,
1002
+ "single_word": false,
1003
+ "special": true
1004
+ },
1005
+ "126205": {
1006
+ "content": "<|reserved_token_121|>",
1007
+ "lstrip": false,
1008
+ "normalized": false,
1009
+ "rstrip": false,
1010
+ "single_word": false,
1011
+ "special": true
1012
+ },
1013
+ "126206": {
1014
+ "content": "<|reserved_token_122|>",
1015
+ "lstrip": false,
1016
+ "normalized": false,
1017
+ "rstrip": false,
1018
+ "single_word": false,
1019
+ "special": true
1020
+ },
1021
+ "126207": {
1022
+ "content": "<|reserved_token_123|>",
1023
+ "lstrip": false,
1024
+ "normalized": false,
1025
+ "rstrip": false,
1026
+ "single_word": false,
1027
+ "special": true
1028
+ },
1029
+ "126208": {
1030
+ "content": "<|reserved_token_124|>",
1031
+ "lstrip": false,
1032
+ "normalized": false,
1033
+ "rstrip": false,
1034
+ "single_word": false,
1035
+ "special": true
1036
+ },
1037
+ "126209": {
1038
+ "content": "<|reserved_token_125|>",
1039
+ "lstrip": false,
1040
+ "normalized": false,
1041
+ "rstrip": false,
1042
+ "single_word": false,
1043
+ "special": true
1044
+ },
1045
+ "126210": {
1046
+ "content": "<|reserved_token_126|>",
1047
+ "lstrip": false,
1048
+ "normalized": false,
1049
+ "rstrip": false,
1050
+ "single_word": false,
1051
+ "special": true
1052
+ },
1053
+ "126211": {
1054
+ "content": "<|reserved_token_127|>",
1055
+ "lstrip": false,
1056
+ "normalized": false,
1057
+ "rstrip": false,
1058
+ "single_word": false,
1059
+ "special": true
1060
+ },
1061
+ "126212": {
1062
+ "content": "<|reserved_token_128|>",
1063
+ "lstrip": false,
1064
+ "normalized": false,
1065
+ "rstrip": false,
1066
+ "single_word": false,
1067
+ "special": true
1068
+ },
1069
+ "126213": {
1070
+ "content": "<|reserved_token_129|>",
1071
+ "lstrip": false,
1072
+ "normalized": false,
1073
+ "rstrip": false,
1074
+ "single_word": false,
1075
+ "special": true
1076
+ },
1077
+ "126214": {
1078
+ "content": "<|reserved_token_130|>",
1079
+ "lstrip": false,
1080
+ "normalized": false,
1081
+ "rstrip": false,
1082
+ "single_word": false,
1083
+ "special": true
1084
+ },
1085
+ "126215": {
1086
+ "content": "<|reserved_token_131|>",
1087
+ "lstrip": false,
1088
+ "normalized": false,
1089
+ "rstrip": false,
1090
+ "single_word": false,
1091
+ "special": true
1092
+ },
1093
+ "126216": {
1094
+ "content": "<|reserved_token_132|>",
1095
+ "lstrip": false,
1096
+ "normalized": false,
1097
+ "rstrip": false,
1098
+ "single_word": false,
1099
+ "special": true
1100
+ },
1101
+ "126217": {
1102
+ "content": "<|reserved_token_133|>",
1103
+ "lstrip": false,
1104
+ "normalized": false,
1105
+ "rstrip": false,
1106
+ "single_word": false,
1107
+ "special": true
1108
+ },
1109
+ "126218": {
1110
+ "content": "<|reserved_token_134|>",
1111
+ "lstrip": false,
1112
+ "normalized": false,
1113
+ "rstrip": false,
1114
+ "single_word": false,
1115
+ "special": true
1116
+ },
1117
+ "126219": {
1118
+ "content": "<|reserved_token_135|>",
1119
+ "lstrip": false,
1120
+ "normalized": false,
1121
+ "rstrip": false,
1122
+ "single_word": false,
1123
+ "special": true
1124
+ },
1125
+ "126220": {
1126
+ "content": "<|reserved_token_136|>",
1127
+ "lstrip": false,
1128
+ "normalized": false,
1129
+ "rstrip": false,
1130
+ "single_word": false,
1131
+ "special": true
1132
+ },
1133
+ "126221": {
1134
+ "content": "<|reserved_token_137|>",
1135
+ "lstrip": false,
1136
+ "normalized": false,
1137
+ "rstrip": false,
1138
+ "single_word": false,
1139
+ "special": true
1140
+ },
1141
+ "126222": {
1142
+ "content": "<|reserved_token_138|>",
1143
+ "lstrip": false,
1144
+ "normalized": false,
1145
+ "rstrip": false,
1146
+ "single_word": false,
1147
+ "special": true
1148
+ },
1149
+ "126223": {
1150
+ "content": "<|reserved_token_139|>",
1151
+ "lstrip": false,
1152
+ "normalized": false,
1153
+ "rstrip": false,
1154
+ "single_word": false,
1155
+ "special": true
1156
+ },
1157
+ "126224": {
1158
+ "content": "<|reserved_token_140|>",
1159
+ "lstrip": false,
1160
+ "normalized": false,
1161
+ "rstrip": false,
1162
+ "single_word": false,
1163
+ "special": true
1164
+ },
1165
+ "126225": {
1166
+ "content": "<|reserved_token_141|>",
1167
+ "lstrip": false,
1168
+ "normalized": false,
1169
+ "rstrip": false,
1170
+ "single_word": false,
1171
+ "special": true
1172
+ },
1173
+ "126226": {
1174
+ "content": "<|reserved_token_142|>",
1175
+ "lstrip": false,
1176
+ "normalized": false,
1177
+ "rstrip": false,
1178
+ "single_word": false,
1179
+ "special": true
1180
+ },
1181
+ "126227": {
1182
+ "content": "<|reserved_token_143|>",
1183
+ "lstrip": false,
1184
+ "normalized": false,
1185
+ "rstrip": false,
1186
+ "single_word": false,
1187
+ "special": true
1188
+ },
1189
+ "126228": {
1190
+ "content": "<|reserved_token_144|>",
1191
+ "lstrip": false,
1192
+ "normalized": false,
1193
+ "rstrip": false,
1194
+ "single_word": false,
1195
+ "special": true
1196
+ },
1197
+ "126229": {
1198
+ "content": "<|reserved_token_145|>",
1199
+ "lstrip": false,
1200
+ "normalized": false,
1201
+ "rstrip": false,
1202
+ "single_word": false,
1203
+ "special": true
1204
+ },
1205
+ "126230": {
1206
+ "content": "<|reserved_token_146|>",
1207
+ "lstrip": false,
1208
+ "normalized": false,
1209
+ "rstrip": false,
1210
+ "single_word": false,
1211
+ "special": true
1212
+ },
1213
+ "126231": {
1214
+ "content": "<|reserved_token_147|>",
1215
+ "lstrip": false,
1216
+ "normalized": false,
1217
+ "rstrip": false,
1218
+ "single_word": false,
1219
+ "special": true
1220
+ },
1221
+ "126232": {
1222
+ "content": "<|reserved_token_148|>",
1223
+ "lstrip": false,
1224
+ "normalized": false,
1225
+ "rstrip": false,
1226
+ "single_word": false,
1227
+ "special": true
1228
+ },
1229
+ "126233": {
1230
+ "content": "<|reserved_token_149|>",
1231
+ "lstrip": false,
1232
+ "normalized": false,
1233
+ "rstrip": false,
1234
+ "single_word": false,
1235
+ "special": true
1236
+ },
1237
+ "126234": {
1238
+ "content": "<|reserved_token_150|>",
1239
+ "lstrip": false,
1240
+ "normalized": false,
1241
+ "rstrip": false,
1242
+ "single_word": false,
1243
+ "special": true
1244
+ },
1245
+ "126235": {
1246
+ "content": "<|reserved_token_151|>",
1247
+ "lstrip": false,
1248
+ "normalized": false,
1249
+ "rstrip": false,
1250
+ "single_word": false,
1251
+ "special": true
1252
+ },
1253
+ "126236": {
1254
+ "content": "<|reserved_token_152|>",
1255
+ "lstrip": false,
1256
+ "normalized": false,
1257
+ "rstrip": false,
1258
+ "single_word": false,
1259
+ "special": true
1260
+ },
1261
+ "126237": {
1262
+ "content": "<|reserved_token_153|>",
1263
+ "lstrip": false,
1264
+ "normalized": false,
1265
+ "rstrip": false,
1266
+ "single_word": false,
1267
+ "special": true
1268
+ },
1269
+ "126238": {
1270
+ "content": "<|reserved_token_154|>",
1271
+ "lstrip": false,
1272
+ "normalized": false,
1273
+ "rstrip": false,
1274
+ "single_word": false,
1275
+ "special": true
1276
+ },
1277
+ "126239": {
1278
+ "content": "<|reserved_token_155|>",
1279
+ "lstrip": false,
1280
+ "normalized": false,
1281
+ "rstrip": false,
1282
+ "single_word": false,
1283
+ "special": true
1284
+ },
1285
+ "126240": {
1286
+ "content": "<|reserved_token_156|>",
1287
+ "lstrip": false,
1288
+ "normalized": false,
1289
+ "rstrip": false,
1290
+ "single_word": false,
1291
+ "special": true
1292
+ },
1293
+ "126241": {
1294
+ "content": "<|reserved_token_157|>",
1295
+ "lstrip": false,
1296
+ "normalized": false,
1297
+ "rstrip": false,
1298
+ "single_word": false,
1299
+ "special": true
1300
+ },
1301
+ "126242": {
1302
+ "content": "<|reserved_token_158|>",
1303
+ "lstrip": false,
1304
+ "normalized": false,
1305
+ "rstrip": false,
1306
+ "single_word": false,
1307
+ "special": true
1308
+ },
1309
+ "126243": {
1310
+ "content": "<|reserved_token_159|>",
1311
+ "lstrip": false,
1312
+ "normalized": false,
1313
+ "rstrip": false,
1314
+ "single_word": false,
1315
+ "special": true
1316
+ },
1317
+ "126244": {
1318
+ "content": "<|reserved_token_160|>",
1319
+ "lstrip": false,
1320
+ "normalized": false,
1321
+ "rstrip": false,
1322
+ "single_word": false,
1323
+ "special": true
1324
+ },
1325
+ "126245": {
1326
+ "content": "<|reserved_token_161|>",
1327
+ "lstrip": false,
1328
+ "normalized": false,
1329
+ "rstrip": false,
1330
+ "single_word": false,
1331
+ "special": true
1332
+ },
1333
+ "126246": {
1334
+ "content": "<|reserved_token_162|>",
1335
+ "lstrip": false,
1336
+ "normalized": false,
1337
+ "rstrip": false,
1338
+ "single_word": false,
1339
+ "special": true
1340
+ },
1341
+ "126247": {
1342
+ "content": "<|reserved_token_163|>",
1343
+ "lstrip": false,
1344
+ "normalized": false,
1345
+ "rstrip": false,
1346
+ "single_word": false,
1347
+ "special": true
1348
+ },
1349
+ "126248": {
1350
+ "content": "<|reserved_token_164|>",
1351
+ "lstrip": false,
1352
+ "normalized": false,
1353
+ "rstrip": false,
1354
+ "single_word": false,
1355
+ "special": true
1356
+ },
1357
+ "126249": {
1358
+ "content": "<|reserved_token_165|>",
1359
+ "lstrip": false,
1360
+ "normalized": false,
1361
+ "rstrip": false,
1362
+ "single_word": false,
1363
+ "special": true
1364
+ },
1365
+ "126250": {
1366
+ "content": "<|reserved_token_166|>",
1367
+ "lstrip": false,
1368
+ "normalized": false,
1369
+ "rstrip": false,
1370
+ "single_word": false,
1371
+ "special": true
1372
+ },
1373
+ "126251": {
1374
+ "content": "<|reserved_token_167|>",
1375
+ "lstrip": false,
1376
+ "normalized": false,
1377
+ "rstrip": false,
1378
+ "single_word": false,
1379
+ "special": true
1380
+ },
1381
+ "126252": {
1382
+ "content": "<|reserved_token_168|>",
1383
+ "lstrip": false,
1384
+ "normalized": false,
1385
+ "rstrip": false,
1386
+ "single_word": false,
1387
+ "special": true
1388
+ },
1389
+ "126253": {
1390
+ "content": "<|reserved_token_169|>",
1391
+ "lstrip": false,
1392
+ "normalized": false,
1393
+ "rstrip": false,
1394
+ "single_word": false,
1395
+ "special": true
1396
+ },
1397
+ "126254": {
1398
+ "content": "<|reserved_token_170|>",
1399
+ "lstrip": false,
1400
+ "normalized": false,
1401
+ "rstrip": false,
1402
+ "single_word": false,
1403
+ "special": true
1404
+ },
1405
+ "126255": {
1406
+ "content": "<|reserved_token_171|>",
1407
+ "lstrip": false,
1408
+ "normalized": false,
1409
+ "rstrip": false,
1410
+ "single_word": false,
1411
+ "special": true
1412
+ },
1413
+ "126256": {
1414
+ "content": "<|reserved_token_172|>",
1415
+ "lstrip": false,
1416
+ "normalized": false,
1417
+ "rstrip": false,
1418
+ "single_word": false,
1419
+ "special": true
1420
+ },
1421
+ "126257": {
1422
+ "content": "<|reserved_token_173|>",
1423
+ "lstrip": false,
1424
+ "normalized": false,
1425
+ "rstrip": false,
1426
+ "single_word": false,
1427
+ "special": true
1428
+ },
1429
+ "126258": {
1430
+ "content": "<|reserved_token_174|>",
1431
+ "lstrip": false,
1432
+ "normalized": false,
1433
+ "rstrip": false,
1434
+ "single_word": false,
1435
+ "special": true
1436
+ },
1437
+ "126259": {
1438
+ "content": "<|reserved_token_175|>",
1439
+ "lstrip": false,
1440
+ "normalized": false,
1441
+ "rstrip": false,
1442
+ "single_word": false,
1443
+ "special": true
1444
+ },
1445
+ "126260": {
1446
+ "content": "<|reserved_token_176|>",
1447
+ "lstrip": false,
1448
+ "normalized": false,
1449
+ "rstrip": false,
1450
+ "single_word": false,
1451
+ "special": true
1452
+ },
1453
+ "126261": {
1454
+ "content": "<|reserved_token_177|>",
1455
+ "lstrip": false,
1456
+ "normalized": false,
1457
+ "rstrip": false,
1458
+ "single_word": false,
1459
+ "special": true
1460
+ },
1461
+ "126262": {
1462
+ "content": "<|reserved_token_178|>",
1463
+ "lstrip": false,
1464
+ "normalized": false,
1465
+ "rstrip": false,
1466
+ "single_word": false,
1467
+ "special": true
1468
+ },
1469
+ "126263": {
1470
+ "content": "<|reserved_token_179|>",
1471
+ "lstrip": false,
1472
+ "normalized": false,
1473
+ "rstrip": false,
1474
+ "single_word": false,
1475
+ "special": true
1476
+ },
1477
+ "126264": {
1478
+ "content": "<|reserved_token_180|>",
1479
+ "lstrip": false,
1480
+ "normalized": false,
1481
+ "rstrip": false,
1482
+ "single_word": false,
1483
+ "special": true
1484
+ },
1485
+ "126265": {
1486
+ "content": "<|reserved_token_181|>",
1487
+ "lstrip": false,
1488
+ "normalized": false,
1489
+ "rstrip": false,
1490
+ "single_word": false,
1491
+ "special": true
1492
+ },
1493
+ "126266": {
1494
+ "content": "<|reserved_token_182|>",
1495
+ "lstrip": false,
1496
+ "normalized": false,
1497
+ "rstrip": false,
1498
+ "single_word": false,
1499
+ "special": true
1500
+ },
1501
+ "126267": {
1502
+ "content": "<|reserved_token_183|>",
1503
+ "lstrip": false,
1504
+ "normalized": false,
1505
+ "rstrip": false,
1506
+ "single_word": false,
1507
+ "special": true
1508
+ },
1509
+ "126268": {
1510
+ "content": "<|reserved_token_184|>",
1511
+ "lstrip": false,
1512
+ "normalized": false,
1513
+ "rstrip": false,
1514
+ "single_word": false,
1515
+ "special": true
1516
+ },
1517
+ "126269": {
1518
+ "content": "<|reserved_token_185|>",
1519
+ "lstrip": false,
1520
+ "normalized": false,
1521
+ "rstrip": false,
1522
+ "single_word": false,
1523
+ "special": true
1524
+ },
1525
+ "126270": {
1526
+ "content": "<|reserved_token_186|>",
1527
+ "lstrip": false,
1528
+ "normalized": false,
1529
+ "rstrip": false,
1530
+ "single_word": false,
1531
+ "special": true
1532
+ },
1533
+ "126271": {
1534
+ "content": "<|reserved_token_187|>",
1535
+ "lstrip": false,
1536
+ "normalized": false,
1537
+ "rstrip": false,
1538
+ "single_word": false,
1539
+ "special": true
1540
+ },
1541
+ "126272": {
1542
+ "content": "<|reserved_token_188|>",
1543
+ "lstrip": false,
1544
+ "normalized": false,
1545
+ "rstrip": false,
1546
+ "single_word": false,
1547
+ "special": true
1548
+ },
1549
+ "126273": {
1550
+ "content": "<|reserved_token_189|>",
1551
+ "lstrip": false,
1552
+ "normalized": false,
1553
+ "rstrip": false,
1554
+ "single_word": false,
1555
+ "special": true
1556
+ },
1557
+ "126274": {
1558
+ "content": "<|reserved_token_190|>",
1559
+ "lstrip": false,
1560
+ "normalized": false,
1561
+ "rstrip": false,
1562
+ "single_word": false,
1563
+ "special": true
1564
+ },
1565
+ "126275": {
1566
+ "content": "<|reserved_token_191|>",
1567
+ "lstrip": false,
1568
+ "normalized": false,
1569
+ "rstrip": false,
1570
+ "single_word": false,
1571
+ "special": true
1572
+ },
1573
+ "126276": {
1574
+ "content": "<|reserved_token_192|>",
1575
+ "lstrip": false,
1576
+ "normalized": false,
1577
+ "rstrip": false,
1578
+ "single_word": false,
1579
+ "special": true
1580
+ },
1581
+ "126277": {
1582
+ "content": "<|reserved_token_193|>",
1583
+ "lstrip": false,
1584
+ "normalized": false,
1585
+ "rstrip": false,
1586
+ "single_word": false,
1587
+ "special": true
1588
+ },
1589
+ "126278": {
1590
+ "content": "<|reserved_token_194|>",
1591
+ "lstrip": false,
1592
+ "normalized": false,
1593
+ "rstrip": false,
1594
+ "single_word": false,
1595
+ "special": true
1596
+ },
1597
+ "126279": {
1598
+ "content": "<|reserved_token_195|>",
1599
+ "lstrip": false,
1600
+ "normalized": false,
1601
+ "rstrip": false,
1602
+ "single_word": false,
1603
+ "special": true
1604
+ },
1605
+ "126280": {
1606
+ "content": "<|reserved_token_196|>",
1607
+ "lstrip": false,
1608
+ "normalized": false,
1609
+ "rstrip": false,
1610
+ "single_word": false,
1611
+ "special": true
1612
+ },
1613
+ "126281": {
1614
+ "content": "<|reserved_token_197|>",
1615
+ "lstrip": false,
1616
+ "normalized": false,
1617
+ "rstrip": false,
1618
+ "single_word": false,
1619
+ "special": true
1620
+ },
1621
+ "126282": {
1622
+ "content": "<|reserved_token_198|>",
1623
+ "lstrip": false,
1624
+ "normalized": false,
1625
+ "rstrip": false,
1626
+ "single_word": false,
1627
+ "special": true
1628
+ },
1629
+ "126283": {
1630
+ "content": "<|reserved_token_199|>",
1631
+ "lstrip": false,
1632
+ "normalized": false,
1633
+ "rstrip": false,
1634
+ "single_word": false,
1635
+ "special": true
1636
+ },
1637
+ "126284": {
1638
+ "content": "<|reserved_token_200|>",
1639
+ "lstrip": false,
1640
+ "normalized": false,
1641
+ "rstrip": false,
1642
+ "single_word": false,
1643
+ "special": true
1644
+ },
1645
+ "126285": {
1646
+ "content": "<|reserved_token_201|>",
1647
+ "lstrip": false,
1648
+ "normalized": false,
1649
+ "rstrip": false,
1650
+ "single_word": false,
1651
+ "special": true
1652
+ },
1653
+ "126286": {
1654
+ "content": "<|reserved_token_202|>",
1655
+ "lstrip": false,
1656
+ "normalized": false,
1657
+ "rstrip": false,
1658
+ "single_word": false,
1659
+ "special": true
1660
+ },
1661
+ "126287": {
1662
+ "content": "<|reserved_token_203|>",
1663
+ "lstrip": false,
1664
+ "normalized": false,
1665
+ "rstrip": false,
1666
+ "single_word": false,
1667
+ "special": true
1668
+ },
1669
+ "126288": {
1670
+ "content": "<|reserved_token_204|>",
1671
+ "lstrip": false,
1672
+ "normalized": false,
1673
+ "rstrip": false,
1674
+ "single_word": false,
1675
+ "special": true
1676
+ },
1677
+ "126289": {
1678
+ "content": "<|reserved_token_205|>",
1679
+ "lstrip": false,
1680
+ "normalized": false,
1681
+ "rstrip": false,
1682
+ "single_word": false,
1683
+ "special": true
1684
+ },
1685
+ "126290": {
1686
+ "content": "<|reserved_token_206|>",
1687
+ "lstrip": false,
1688
+ "normalized": false,
1689
+ "rstrip": false,
1690
+ "single_word": false,
1691
+ "special": true
1692
+ },
1693
+ "126291": {
1694
+ "content": "<|reserved_token_207|>",
1695
+ "lstrip": false,
1696
+ "normalized": false,
1697
+ "rstrip": false,
1698
+ "single_word": false,
1699
+ "special": true
1700
+ },
1701
+ "126292": {
1702
+ "content": "<|reserved_token_208|>",
1703
+ "lstrip": false,
1704
+ "normalized": false,
1705
+ "rstrip": false,
1706
+ "single_word": false,
1707
+ "special": true
1708
+ },
1709
+ "126293": {
1710
+ "content": "<|reserved_token_209|>",
1711
+ "lstrip": false,
1712
+ "normalized": false,
1713
+ "rstrip": false,
1714
+ "single_word": false,
1715
+ "special": true
1716
+ },
1717
+ "126294": {
1718
+ "content": "<|reserved_token_210|>",
1719
+ "lstrip": false,
1720
+ "normalized": false,
1721
+ "rstrip": false,
1722
+ "single_word": false,
1723
+ "special": true
1724
+ },
1725
+ "126295": {
1726
+ "content": "<|reserved_token_211|>",
1727
+ "lstrip": false,
1728
+ "normalized": false,
1729
+ "rstrip": false,
1730
+ "single_word": false,
1731
+ "special": true
1732
+ },
1733
+ "126296": {
1734
+ "content": "<|reserved_token_212|>",
1735
+ "lstrip": false,
1736
+ "normalized": false,
1737
+ "rstrip": false,
1738
+ "single_word": false,
1739
+ "special": true
1740
+ },
1741
+ "126297": {
1742
+ "content": "<|reserved_token_213|>",
1743
+ "lstrip": false,
1744
+ "normalized": false,
1745
+ "rstrip": false,
1746
+ "single_word": false,
1747
+ "special": true
1748
+ },
1749
+ "126298": {
1750
+ "content": "<|reserved_token_214|>",
1751
+ "lstrip": false,
1752
+ "normalized": false,
1753
+ "rstrip": false,
1754
+ "single_word": false,
1755
+ "special": true
1756
+ },
1757
+ "126299": {
1758
+ "content": "<|reserved_token_215|>",
1759
+ "lstrip": false,
1760
+ "normalized": false,
1761
+ "rstrip": false,
1762
+ "single_word": false,
1763
+ "special": true
1764
+ },
1765
+ "126300": {
1766
+ "content": "<|reserved_token_216|>",
1767
+ "lstrip": false,
1768
+ "normalized": false,
1769
+ "rstrip": false,
1770
+ "single_word": false,
1771
+ "special": true
1772
+ },
1773
+ "126301": {
1774
+ "content": "<|reserved_token_217|>",
1775
+ "lstrip": false,
1776
+ "normalized": false,
1777
+ "rstrip": false,
1778
+ "single_word": false,
1779
+ "special": true
1780
+ },
1781
+ "126302": {
1782
+ "content": "<|reserved_token_218|>",
1783
+ "lstrip": false,
1784
+ "normalized": false,
1785
+ "rstrip": false,
1786
+ "single_word": false,
1787
+ "special": true
1788
+ },
1789
+ "126303": {
1790
+ "content": "<|reserved_token_219|>",
1791
+ "lstrip": false,
1792
+ "normalized": false,
1793
+ "rstrip": false,
1794
+ "single_word": false,
1795
+ "special": true
1796
+ },
1797
+ "126304": {
1798
+ "content": "<|reserved_token_220|>",
1799
+ "lstrip": false,
1800
+ "normalized": false,
1801
+ "rstrip": false,
1802
+ "single_word": false,
1803
+ "special": true
1804
+ },
1805
+ "126305": {
1806
+ "content": "<|reserved_token_221|>",
1807
+ "lstrip": false,
1808
+ "normalized": false,
1809
+ "rstrip": false,
1810
+ "single_word": false,
1811
+ "special": true
1812
+ },
1813
+ "126306": {
1814
+ "content": "<|reserved_token_222|>",
1815
+ "lstrip": false,
1816
+ "normalized": false,
1817
+ "rstrip": false,
1818
+ "single_word": false,
1819
+ "special": true
1820
+ },
1821
+ "126307": {
1822
+ "content": "<|reserved_token_223|>",
1823
+ "lstrip": false,
1824
+ "normalized": false,
1825
+ "rstrip": false,
1826
+ "single_word": false,
1827
+ "special": true
1828
+ },
1829
+ "126308": {
1830
+ "content": "<|reserved_token_224|>",
1831
+ "lstrip": false,
1832
+ "normalized": false,
1833
+ "rstrip": false,
1834
+ "single_word": false,
1835
+ "special": true
1836
+ },
1837
+ "126309": {
1838
+ "content": "<|reserved_token_225|>",
1839
+ "lstrip": false,
1840
+ "normalized": false,
1841
+ "rstrip": false,
1842
+ "single_word": false,
1843
+ "special": true
1844
+ },
1845
+ "126310": {
1846
+ "content": "<|reserved_token_226|>",
1847
+ "lstrip": false,
1848
+ "normalized": false,
1849
+ "rstrip": false,
1850
+ "single_word": false,
1851
+ "special": true
1852
+ },
1853
+ "126311": {
1854
+ "content": "<|reserved_token_227|>",
1855
+ "lstrip": false,
1856
+ "normalized": false,
1857
+ "rstrip": false,
1858
+ "single_word": false,
1859
+ "special": true
1860
+ },
1861
+ "126312": {
1862
+ "content": "<|reserved_token_228|>",
1863
+ "lstrip": false,
1864
+ "normalized": false,
1865
+ "rstrip": false,
1866
+ "single_word": false,
1867
+ "special": true
1868
+ },
1869
+ "126313": {
1870
+ "content": "<|reserved_token_229|>",
1871
+ "lstrip": false,
1872
+ "normalized": false,
1873
+ "rstrip": false,
1874
+ "single_word": false,
1875
+ "special": true
1876
+ },
1877
+ "126314": {
1878
+ "content": "<|reserved_token_230|>",
1879
+ "lstrip": false,
1880
+ "normalized": false,
1881
+ "rstrip": false,
1882
+ "single_word": false,
1883
+ "special": true
1884
+ },
1885
+ "126315": {
1886
+ "content": "<|reserved_token_231|>",
1887
+ "lstrip": false,
1888
+ "normalized": false,
1889
+ "rstrip": false,
1890
+ "single_word": false,
1891
+ "special": true
1892
+ },
1893
+ "126316": {
1894
+ "content": "<|reserved_token_232|>",
1895
+ "lstrip": false,
1896
+ "normalized": false,
1897
+ "rstrip": false,
1898
+ "single_word": false,
1899
+ "special": true
1900
+ },
1901
+ "126317": {
1902
+ "content": "<|reserved_token_233|>",
1903
+ "lstrip": false,
1904
+ "normalized": false,
1905
+ "rstrip": false,
1906
+ "single_word": false,
1907
+ "special": true
1908
+ },
1909
+ "126318": {
1910
+ "content": "<|reserved_token_234|>",
1911
+ "lstrip": false,
1912
+ "normalized": false,
1913
+ "rstrip": false,
1914
+ "single_word": false,
1915
+ "special": true
1916
+ },
1917
+ "126319": {
1918
+ "content": "<|reserved_token_235|>",
1919
+ "lstrip": false,
1920
+ "normalized": false,
1921
+ "rstrip": false,
1922
+ "single_word": false,
1923
+ "special": true
1924
+ },
1925
+ "126320": {
1926
+ "content": "<|reserved_token_236|>",
1927
+ "lstrip": false,
1928
+ "normalized": false,
1929
+ "rstrip": false,
1930
+ "single_word": false,
1931
+ "special": true
1932
+ },
1933
+ "126321": {
1934
+ "content": "<|reserved_token_237|>",
1935
+ "lstrip": false,
1936
+ "normalized": false,
1937
+ "rstrip": false,
1938
+ "single_word": false,
1939
+ "special": true
1940
+ },
1941
+ "126322": {
1942
+ "content": "<|reserved_token_238|>",
1943
+ "lstrip": false,
1944
+ "normalized": false,
1945
+ "rstrip": false,
1946
+ "single_word": false,
1947
+ "special": true
1948
+ },
1949
+ "126323": {
1950
+ "content": "<|reserved_token_239|>",
1951
+ "lstrip": false,
1952
+ "normalized": false,
1953
+ "rstrip": false,
1954
+ "single_word": false,
1955
+ "special": true
1956
+ },
1957
+ "126324": {
1958
+ "content": "<|reserved_token_240|>",
1959
+ "lstrip": false,
1960
+ "normalized": false,
1961
+ "rstrip": false,
1962
+ "single_word": false,
1963
+ "special": true
1964
+ },
1965
+ "126325": {
1966
+ "content": "<|reserved_token_241|>",
1967
+ "lstrip": false,
1968
+ "normalized": false,
1969
+ "rstrip": false,
1970
+ "single_word": false,
1971
+ "special": true
1972
+ },
1973
+ "126326": {
1974
+ "content": "<|reserved_token_242|>",
1975
+ "lstrip": false,
1976
+ "normalized": false,
1977
+ "rstrip": false,
1978
+ "single_word": false,
1979
+ "special": true
1980
+ },
1981
+ "126327": {
1982
+ "content": "<|reserved_token_243|>",
1983
+ "lstrip": false,
1984
+ "normalized": false,
1985
+ "rstrip": false,
1986
+ "single_word": false,
1987
+ "special": true
1988
+ },
1989
+ "126328": {
1990
+ "content": "<|reserved_token_244|>",
1991
+ "lstrip": false,
1992
+ "normalized": false,
1993
+ "rstrip": false,
1994
+ "single_word": false,
1995
+ "special": true
1996
+ },
1997
+ "126329": {
1998
+ "content": "<|reserved_token_245|>",
1999
+ "lstrip": false,
2000
+ "normalized": false,
2001
+ "rstrip": false,
2002
+ "single_word": false,
2003
+ "special": true
2004
+ },
2005
+ "126330": {
2006
+ "content": "<|reserved_token_246|>",
2007
+ "lstrip": false,
2008
+ "normalized": false,
2009
+ "rstrip": false,
2010
+ "single_word": false,
2011
+ "special": true
2012
+ },
2013
+ "126331": {
2014
+ "content": "<|reserved_token_247|>",
2015
+ "lstrip": false,
2016
+ "normalized": false,
2017
+ "rstrip": false,
2018
+ "single_word": false,
2019
+ "special": true
2020
+ },
2021
+ "126332": {
2022
+ "content": "<|reserved_token_248|>",
2023
+ "lstrip": false,
2024
+ "normalized": false,
2025
+ "rstrip": false,
2026
+ "single_word": false,
2027
+ "special": true
2028
+ },
2029
+ "126333": {
2030
+ "content": "<|reserved_token_249|>",
2031
+ "lstrip": false,
2032
+ "normalized": false,
2033
+ "rstrip": false,
2034
+ "single_word": false,
2035
+ "special": true
2036
+ },
2037
+ "126334": {
2038
+ "content": "<|reserved_token_250|>",
2039
+ "lstrip": false,
2040
+ "normalized": false,
2041
+ "rstrip": false,
2042
+ "single_word": false,
2043
+ "special": true
2044
+ },
2045
+ "126335": {
2046
+ "content": "<|reserved_token_251|>",
2047
+ "lstrip": false,
2048
+ "normalized": false,
2049
+ "rstrip": false,
2050
+ "single_word": false,
2051
+ "special": true
2052
+ },
2053
+ "126336": {
2054
+ "content": "<|mdm_mask|>",
2055
+ "lstrip": false,
2056
+ "normalized": false,
2057
+ "rstrip": false,
2058
+ "single_word": false,
2059
+ "special": true
2060
+ },
2061
+ "126337": {
2062
+ "content": "<|reserved_token_253|>",
2063
+ "lstrip": false,
2064
+ "normalized": false,
2065
+ "rstrip": false,
2066
+ "single_word": false,
2067
+ "special": true
2068
+ },
2069
+ "126338": {
2070
+ "content": "<|reserved_token_254|>",
2071
+ "lstrip": false,
2072
+ "normalized": false,
2073
+ "rstrip": false,
2074
+ "single_word": false,
2075
+ "special": true
2076
+ },
2077
+ "126339": {
2078
+ "content": "<|reserved_token_255|>",
2079
+ "lstrip": false,
2080
+ "normalized": false,
2081
+ "rstrip": false,
2082
+ "single_word": false,
2083
+ "special": true
2084
+ },
2085
+ "126340": {
2086
+ "content": "<role>",
2087
+ "lstrip": false,
2088
+ "normalized": false,
2089
+ "rstrip": false,
2090
+ "single_word": false,
2091
+ "special": true
2092
+ },
2093
+ "126341": {
2094
+ "content": "</role>",
2095
+ "lstrip": false,
2096
+ "normalized": false,
2097
+ "rstrip": false,
2098
+ "single_word": false,
2099
+ "special": true
2100
+ },
2101
+ "126342": {
2102
+ "content": "<|arithmetic_start|>",
2103
+ "lstrip": false,
2104
+ "normalized": false,
2105
+ "rstrip": false,
2106
+ "single_word": false,
2107
+ "special": true
2108
+ },
2109
+ "126343": {
2110
+ "content": "<|arithmetic_end|>",
2111
+ "lstrip": false,
2112
+ "normalized": false,
2113
+ "rstrip": false,
2114
+ "single_word": false,
2115
+ "special": true
2116
+ },
2117
+ "126344": {
2118
+ "content": "<|number_start|>",
2119
+ "lstrip": false,
2120
+ "normalized": false,
2121
+ "rstrip": false,
2122
+ "single_word": false,
2123
+ "special": true
2124
+ },
2125
+ "126345": {
2126
+ "content": "<|number_end|>",
2127
+ "lstrip": false,
2128
+ "normalized": false,
2129
+ "rstrip": false,
2130
+ "single_word": false,
2131
+ "special": true
2132
+ },
2133
+ "126346": {
2134
+ "content": "<|start_header_id|>",
2135
+ "lstrip": false,
2136
+ "normalized": false,
2137
+ "rstrip": false,
2138
+ "single_word": false,
2139
+ "special": true
2140
+ },
2141
+ "126347": {
2142
+ "content": "<|end_header_id|>",
2143
+ "lstrip": false,
2144
+ "normalized": false,
2145
+ "rstrip": false,
2146
+ "single_word": false,
2147
+ "special": true
2148
+ },
2149
+ "126348": {
2150
+ "content": "<|eot_id|>",
2151
+ "lstrip": false,
2152
+ "normalized": false,
2153
+ "rstrip": false,
2154
+ "single_word": false,
2155
+ "special": true
2156
+ }
2157
+ },
2158
+ "additional_special_tokens": [
2159
+ "<role>",
2160
+ "</role>",
2161
+ "<|arithmetic_start|>",
2162
+ "<|arithmetic_end|>",
2163
+ "<|number_start|>",
2164
+ "<|number_end|>"
2165
+ ],
2166
+ "bos_token": "<|startoftext|>",
2167
+ "chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}",
2168
+ "clean_up_tokenization_spaces": false,
2169
+ "cls_token": "[CLS]",
2170
+ "eos_token": "<|endoftext|>",
2171
+ "extra_special_tokens": {},
2172
+ "fast_tokenizer": true,
2173
+ "gmask_token": "[gMASK]",
2174
+ "merges_file": null,
2175
+ "model_input_names": [
2176
+ "input_ids",
2177
+ "attention_mask"
2178
+ ],
2179
+ "model_max_length": 1000000000000000019884624838656,
2180
+ "pad_token": "<|endoftext|>",
2181
+ "padding_side": "left",
2182
+ "tokenizer_class": "PreTrainedTokenizer",
2183
+ "trust_remote_code": true
2184
+ }
checkpoint-200/trainer_state.json ADDED
@@ -0,0 +1,2033 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.21333333333333335,
5
+ "eval_steps": 500,
6
+ "global_step": 200,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "completion_length": 252.25,
13
+ "epoch": 0.0010666666666666667,
14
+ "grad_norm": 0.13917993009090424,
15
+ "kl": 0.0,
16
+ "learning_rate": 3e-06,
17
+ "loss": 0.1453,
18
+ "reward": 1.3802083730697632,
19
+ "rewards/boxed_and_answer_tags_format_reward": 0.6302083134651184,
20
+ "rewards/correctness_reward_func_math": 0.75,
21
+ "step": 1
22
+ },
23
+ {
24
+ "epoch": 0.0021333333333333334,
25
+ "grad_norm": 0.13937675952911377,
26
+ "kl": 0.0,
27
+ "learning_rate": 3e-06,
28
+ "loss": 0.1453,
29
+ "step": 2
30
+ },
31
+ {
32
+ "completion_length": 231.6979217529297,
33
+ "epoch": 0.0032,
34
+ "grad_norm": 0.07865716516971588,
35
+ "kl": 0.000684030877891928,
36
+ "learning_rate": 3e-06,
37
+ "loss": 0.0222,
38
+ "reward": 1.2968750298023224,
39
+ "rewards/boxed_and_answer_tags_format_reward": 0.6302083432674408,
40
+ "rewards/correctness_reward_func_math": 0.6666666865348816,
41
+ "step": 3
42
+ },
43
+ {
44
+ "epoch": 0.004266666666666667,
45
+ "grad_norm": 0.05887310951948166,
46
+ "kl": 0.000708846258930862,
47
+ "learning_rate": 3e-06,
48
+ "loss": 0.0223,
49
+ "step": 4
50
+ },
51
+ {
52
+ "completion_length": 248.09375762939453,
53
+ "epoch": 0.005333333333333333,
54
+ "grad_norm": 0.1470843404531479,
55
+ "kl": 0.0013292446383275092,
56
+ "learning_rate": 3e-06,
57
+ "loss": 0.0593,
58
+ "reward": 1.1302083432674408,
59
+ "rewards/boxed_and_answer_tags_format_reward": 0.5677083432674408,
60
+ "rewards/correctness_reward_func_math": 0.5625,
61
+ "step": 5
62
+ },
63
+ {
64
+ "epoch": 0.0064,
65
+ "grad_norm": 0.16160514950752258,
66
+ "kl": 0.0013866736553609371,
67
+ "learning_rate": 3e-06,
68
+ "loss": 0.0589,
69
+ "step": 6
70
+ },
71
+ {
72
+ "completion_length": 240.4375,
73
+ "epoch": 0.007466666666666667,
74
+ "grad_norm": 0.11771108210086823,
75
+ "kl": 0.0008424457046203315,
76
+ "learning_rate": 3e-06,
77
+ "loss": 0.1484,
78
+ "reward": 1.140625,
79
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916865348816,
80
+ "rewards/correctness_reward_func_math": 0.5208333283662796,
81
+ "step": 7
82
+ },
83
+ {
84
+ "epoch": 0.008533333333333334,
85
+ "grad_norm": 0.11521568894386292,
86
+ "kl": 0.0008330132404807955,
87
+ "learning_rate": 3e-06,
88
+ "loss": 0.1483,
89
+ "step": 8
90
+ },
91
+ {
92
+ "completion_length": 248.46875,
93
+ "epoch": 0.0096,
94
+ "grad_norm": 0.13317543268203735,
95
+ "kl": 0.0010271553765051067,
96
+ "learning_rate": 3e-06,
97
+ "loss": 0.1812,
98
+ "reward": 1.1822916865348816,
99
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916567325592,
100
+ "rewards/correctness_reward_func_math": 0.5625,
101
+ "step": 9
102
+ },
103
+ {
104
+ "epoch": 0.010666666666666666,
105
+ "grad_norm": 0.12070783227682114,
106
+ "kl": 0.001022065698634833,
107
+ "learning_rate": 3e-06,
108
+ "loss": 0.1816,
109
+ "step": 10
110
+ },
111
+ {
112
+ "completion_length": 243.93750762939453,
113
+ "epoch": 0.011733333333333333,
114
+ "grad_norm": 0.09472105652093887,
115
+ "kl": 0.0007579472730867565,
116
+ "learning_rate": 3e-06,
117
+ "loss": -0.0433,
118
+ "reward": 1.125,
119
+ "rewards/boxed_and_answer_tags_format_reward": 0.6041666567325592,
120
+ "rewards/correctness_reward_func_math": 0.5208333432674408,
121
+ "step": 11
122
+ },
123
+ {
124
+ "epoch": 0.0128,
125
+ "grad_norm": 0.08511383086442947,
126
+ "kl": 0.0008102746505755931,
127
+ "learning_rate": 3e-06,
128
+ "loss": -0.0435,
129
+ "step": 12
130
+ },
131
+ {
132
+ "completion_length": 246.86458587646484,
133
+ "epoch": 0.013866666666666666,
134
+ "grad_norm": 0.1304204910993576,
135
+ "kl": 0.0008793774177320302,
136
+ "learning_rate": 3e-06,
137
+ "loss": -0.0371,
138
+ "reward": 1.1614583730697632,
139
+ "rewards/boxed_and_answer_tags_format_reward": 0.5989583432674408,
140
+ "rewards/correctness_reward_func_math": 0.5625000074505806,
141
+ "step": 13
142
+ },
143
+ {
144
+ "epoch": 0.014933333333333333,
145
+ "grad_norm": 0.10566864907741547,
146
+ "kl": 0.0007963517564348876,
147
+ "learning_rate": 3e-06,
148
+ "loss": -0.037,
149
+ "step": 14
150
+ },
151
+ {
152
+ "completion_length": 244.59375762939453,
153
+ "epoch": 0.016,
154
+ "grad_norm": 0.09695824235677719,
155
+ "kl": 0.0009378301328979433,
156
+ "learning_rate": 3e-06,
157
+ "loss": 0.0214,
158
+ "reward": 1.0833333730697632,
159
+ "rewards/boxed_and_answer_tags_format_reward": 0.5833333134651184,
160
+ "rewards/correctness_reward_func_math": 0.5,
161
+ "step": 15
162
+ },
163
+ {
164
+ "epoch": 0.017066666666666667,
165
+ "grad_norm": 0.099588543176651,
166
+ "kl": 0.0009069846710190177,
167
+ "learning_rate": 3e-06,
168
+ "loss": 0.0217,
169
+ "step": 16
170
+ },
171
+ {
172
+ "completion_length": 239.80209350585938,
173
+ "epoch": 0.018133333333333335,
174
+ "grad_norm": 0.06958300620317459,
175
+ "kl": 0.001107214658986777,
176
+ "learning_rate": 3e-06,
177
+ "loss": 0.0169,
178
+ "reward": 0.916666716337204,
179
+ "rewards/boxed_and_answer_tags_format_reward": 0.5416666865348816,
180
+ "rewards/correctness_reward_func_math": 0.375,
181
+ "step": 17
182
+ },
183
+ {
184
+ "epoch": 0.0192,
185
+ "grad_norm": 0.1377073973417282,
186
+ "kl": 0.0008504675352014601,
187
+ "learning_rate": 3e-06,
188
+ "loss": 0.0173,
189
+ "step": 18
190
+ },
191
+ {
192
+ "completion_length": 246.2916717529297,
193
+ "epoch": 0.020266666666666665,
194
+ "grad_norm": 0.07782328873872757,
195
+ "kl": 0.0008457594667561352,
196
+ "learning_rate": 3e-06,
197
+ "loss": 0.0861,
198
+ "reward": 1.0989583730697632,
199
+ "rewards/boxed_and_answer_tags_format_reward": 0.6614583134651184,
200
+ "rewards/correctness_reward_func_math": 0.4375,
201
+ "step": 19
202
+ },
203
+ {
204
+ "epoch": 0.021333333333333333,
205
+ "grad_norm": 0.0780203640460968,
206
+ "kl": 0.000831472483696416,
207
+ "learning_rate": 3e-06,
208
+ "loss": 0.0857,
209
+ "step": 20
210
+ },
211
+ {
212
+ "completion_length": 238.67709350585938,
213
+ "epoch": 0.0224,
214
+ "grad_norm": 0.08472912758588791,
215
+ "kl": 0.0009613791771698743,
216
+ "learning_rate": 3e-06,
217
+ "loss": 0.1118,
218
+ "reward": 1.0052083432674408,
219
+ "rewards/boxed_and_answer_tags_format_reward": 0.5677083432674408,
220
+ "rewards/correctness_reward_func_math": 0.4375,
221
+ "step": 21
222
+ },
223
+ {
224
+ "epoch": 0.023466666666666667,
225
+ "grad_norm": 0.14066265523433685,
226
+ "kl": 0.0011330295819789171,
227
+ "learning_rate": 3e-06,
228
+ "loss": 0.1123,
229
+ "step": 22
230
+ },
231
+ {
232
+ "completion_length": 243.4791717529297,
233
+ "epoch": 0.024533333333333334,
234
+ "grad_norm": 0.14210180938243866,
235
+ "kl": 0.001092101214453578,
236
+ "learning_rate": 3e-06,
237
+ "loss": 0.0571,
238
+ "reward": 1.1145833730697632,
239
+ "rewards/boxed_and_answer_tags_format_reward": 0.65625,
240
+ "rewards/correctness_reward_func_math": 0.4583333283662796,
241
+ "step": 23
242
+ },
243
+ {
244
+ "epoch": 0.0256,
245
+ "grad_norm": 0.13714748620986938,
246
+ "kl": 0.0010952465236186981,
247
+ "learning_rate": 3e-06,
248
+ "loss": 0.0571,
249
+ "step": 24
250
+ },
251
+ {
252
+ "completion_length": 247.6666717529297,
253
+ "epoch": 0.02666666666666667,
254
+ "grad_norm": 0.12453050911426544,
255
+ "kl": 0.001361026312224567,
256
+ "learning_rate": 3e-06,
257
+ "loss": 0.0036,
258
+ "reward": 1.2447917461395264,
259
+ "rewards/boxed_and_answer_tags_format_reward": 0.640625,
260
+ "rewards/correctness_reward_func_math": 0.6041666716337204,
261
+ "step": 25
262
+ },
263
+ {
264
+ "epoch": 0.027733333333333332,
265
+ "grad_norm": 0.12724444270133972,
266
+ "kl": 0.0013856127625331283,
267
+ "learning_rate": 3e-06,
268
+ "loss": 0.0038,
269
+ "step": 26
270
+ },
271
+ {
272
+ "completion_length": 250.45833587646484,
273
+ "epoch": 0.0288,
274
+ "grad_norm": 0.09747665375471115,
275
+ "kl": 0.001093864208087325,
276
+ "learning_rate": 3e-06,
277
+ "loss": 0.0438,
278
+ "reward": 1.1718750596046448,
279
+ "rewards/boxed_and_answer_tags_format_reward": 0.5677083134651184,
280
+ "rewards/correctness_reward_func_math": 0.6041666716337204,
281
+ "step": 27
282
+ },
283
+ {
284
+ "epoch": 0.029866666666666666,
285
+ "grad_norm": 0.09332119673490524,
286
+ "kl": 0.001126218878198415,
287
+ "learning_rate": 3e-06,
288
+ "loss": 0.0436,
289
+ "step": 28
290
+ },
291
+ {
292
+ "completion_length": 239.3854217529297,
293
+ "epoch": 0.030933333333333334,
294
+ "grad_norm": 0.17529603838920593,
295
+ "kl": 0.00143107509939,
296
+ "learning_rate": 3e-06,
297
+ "loss": 0.2607,
298
+ "reward": 1.4583333730697632,
299
+ "rewards/boxed_and_answer_tags_format_reward": 0.6041666865348816,
300
+ "rewards/correctness_reward_func_math": 0.8541666567325592,
301
+ "step": 29
302
+ },
303
+ {
304
+ "epoch": 0.032,
305
+ "grad_norm": 0.14503324031829834,
306
+ "kl": 0.0013052276335656643,
307
+ "learning_rate": 3e-06,
308
+ "loss": 0.2604,
309
+ "step": 30
310
+ },
311
+ {
312
+ "completion_length": 247.8541717529297,
313
+ "epoch": 0.03306666666666667,
314
+ "grad_norm": 0.17090579867362976,
315
+ "kl": 0.0016485766973346472,
316
+ "learning_rate": 3e-06,
317
+ "loss": -0.0479,
318
+ "reward": 0.9114583432674408,
319
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916567325592,
320
+ "rewards/correctness_reward_func_math": 0.2916666567325592,
321
+ "step": 31
322
+ },
323
+ {
324
+ "epoch": 0.034133333333333335,
325
+ "grad_norm": 0.17482973635196686,
326
+ "kl": 0.0016103480011224747,
327
+ "learning_rate": 3e-06,
328
+ "loss": -0.0485,
329
+ "step": 32
330
+ },
331
+ {
332
+ "completion_length": 242.9479217529297,
333
+ "epoch": 0.0352,
334
+ "grad_norm": 0.08319579809904099,
335
+ "kl": 0.0013811211101710796,
336
+ "learning_rate": 3e-06,
337
+ "loss": -0.0021,
338
+ "reward": 1.1927083730697632,
339
+ "rewards/boxed_and_answer_tags_format_reward": 0.5885416567325592,
340
+ "rewards/correctness_reward_func_math": 0.6041666716337204,
341
+ "step": 33
342
+ },
343
+ {
344
+ "epoch": 0.03626666666666667,
345
+ "grad_norm": 0.08542275428771973,
346
+ "kl": 0.0013367344508878887,
347
+ "learning_rate": 3e-06,
348
+ "loss": -0.0023,
349
+ "step": 34
350
+ },
351
+ {
352
+ "completion_length": 248.25000762939453,
353
+ "epoch": 0.037333333333333336,
354
+ "grad_norm": 0.06514739990234375,
355
+ "kl": 0.0012877773842774332,
356
+ "learning_rate": 3e-06,
357
+ "loss": 0.0927,
358
+ "reward": 0.84375,
359
+ "rewards/boxed_and_answer_tags_format_reward": 0.6145833134651184,
360
+ "rewards/correctness_reward_func_math": 0.2291666641831398,
361
+ "step": 35
362
+ },
363
+ {
364
+ "epoch": 0.0384,
365
+ "grad_norm": 0.0697920098900795,
366
+ "kl": 0.0013648512540385127,
367
+ "learning_rate": 3e-06,
368
+ "loss": 0.0928,
369
+ "step": 36
370
+ },
371
+ {
372
+ "completion_length": 238.5104217529297,
373
+ "epoch": 0.039466666666666664,
374
+ "grad_norm": 0.09790534526109695,
375
+ "kl": 0.0015243100351653993,
376
+ "learning_rate": 3e-06,
377
+ "loss": 0.0506,
378
+ "reward": 1.1041667461395264,
379
+ "rewards/boxed_and_answer_tags_format_reward": 0.6041666567325592,
380
+ "rewards/correctness_reward_func_math": 0.5000000149011612,
381
+ "step": 37
382
+ },
383
+ {
384
+ "epoch": 0.04053333333333333,
385
+ "grad_norm": 0.09323987364768982,
386
+ "kl": 0.001548255153466016,
387
+ "learning_rate": 3e-06,
388
+ "loss": 0.05,
389
+ "step": 38
390
+ },
391
+ {
392
+ "completion_length": 244.18750762939453,
393
+ "epoch": 0.0416,
394
+ "grad_norm": 0.10544437915086746,
395
+ "kl": 0.0015407268656417727,
396
+ "learning_rate": 3e-06,
397
+ "loss": 0.08,
398
+ "reward": 1.1666666865348816,
399
+ "rewards/boxed_and_answer_tags_format_reward": 0.5833333432674408,
400
+ "rewards/correctness_reward_func_math": 0.5833333283662796,
401
+ "step": 39
402
+ },
403
+ {
404
+ "epoch": 0.042666666666666665,
405
+ "grad_norm": 0.12546436488628387,
406
+ "kl": 0.0015768679440952837,
407
+ "learning_rate": 3e-06,
408
+ "loss": 0.0804,
409
+ "step": 40
410
+ },
411
+ {
412
+ "completion_length": 237.65625762939453,
413
+ "epoch": 0.04373333333333333,
414
+ "grad_norm": 0.07941632717847824,
415
+ "kl": 0.0014703694032505155,
416
+ "learning_rate": 3e-06,
417
+ "loss": 0.0606,
418
+ "reward": 1.0156250596046448,
419
+ "rewards/boxed_and_answer_tags_format_reward": 0.5364583134651184,
420
+ "rewards/correctness_reward_func_math": 0.4791666716337204,
421
+ "step": 41
422
+ },
423
+ {
424
+ "epoch": 0.0448,
425
+ "grad_norm": 0.07871707528829575,
426
+ "kl": 0.0014033173210918903,
427
+ "learning_rate": 3e-06,
428
+ "loss": 0.0604,
429
+ "step": 42
430
+ },
431
+ {
432
+ "completion_length": 234.9791717529297,
433
+ "epoch": 0.04586666666666667,
434
+ "grad_norm": 0.07706955820322037,
435
+ "kl": 0.0015643976512365043,
436
+ "learning_rate": 3e-06,
437
+ "loss": 0.0771,
438
+ "reward": 1.2135416865348816,
439
+ "rewards/boxed_and_answer_tags_format_reward": 0.546875,
440
+ "rewards/correctness_reward_func_math": 0.6666666865348816,
441
+ "step": 43
442
+ },
443
+ {
444
+ "epoch": 0.046933333333333334,
445
+ "grad_norm": 0.06687943637371063,
446
+ "kl": 0.0016123607056215405,
447
+ "learning_rate": 3e-06,
448
+ "loss": 0.0768,
449
+ "step": 44
450
+ },
451
+ {
452
+ "completion_length": 245.6666717529297,
453
+ "epoch": 0.048,
454
+ "grad_norm": 0.0969594195485115,
455
+ "kl": 0.001440248393919319,
456
+ "learning_rate": 3e-06,
457
+ "loss": 0.0308,
458
+ "reward": 1.5572916865348816,
459
+ "rewards/boxed_and_answer_tags_format_reward": 0.640625,
460
+ "rewards/correctness_reward_func_math": 0.9166666567325592,
461
+ "step": 45
462
+ },
463
+ {
464
+ "epoch": 0.04906666666666667,
465
+ "grad_norm": 0.0994538962841034,
466
+ "kl": 0.0014502875274047256,
467
+ "learning_rate": 3e-06,
468
+ "loss": 0.0312,
469
+ "step": 46
470
+ },
471
+ {
472
+ "completion_length": 232.6354217529297,
473
+ "epoch": 0.050133333333333335,
474
+ "grad_norm": 0.0753404051065445,
475
+ "kl": 0.001390244986396283,
476
+ "learning_rate": 3e-06,
477
+ "loss": 0.1163,
478
+ "reward": 1.4895833730697632,
479
+ "rewards/boxed_and_answer_tags_format_reward": 0.59375,
480
+ "rewards/correctness_reward_func_math": 0.8958333432674408,
481
+ "step": 47
482
+ },
483
+ {
484
+ "epoch": 0.0512,
485
+ "grad_norm": 0.07540339231491089,
486
+ "kl": 0.0014414676115848124,
487
+ "learning_rate": 3e-06,
488
+ "loss": 0.1159,
489
+ "step": 48
490
+ },
491
+ {
492
+ "completion_length": 237.5416717529297,
493
+ "epoch": 0.05226666666666667,
494
+ "grad_norm": 0.11727641522884369,
495
+ "kl": 0.0016165076522156596,
496
+ "learning_rate": 3e-06,
497
+ "loss": 0.1507,
498
+ "reward": 1.1510416865348816,
499
+ "rewards/boxed_and_answer_tags_format_reward": 0.671875,
500
+ "rewards/correctness_reward_func_math": 0.4791666716337204,
501
+ "step": 49
502
+ },
503
+ {
504
+ "epoch": 0.05333333333333334,
505
+ "grad_norm": 0.12209934741258621,
506
+ "kl": 0.0017891249153763056,
507
+ "learning_rate": 3e-06,
508
+ "loss": 0.1503,
509
+ "step": 50
510
+ },
511
+ {
512
+ "completion_length": 241.14583587646484,
513
+ "epoch": 0.0544,
514
+ "grad_norm": 0.08470117300748825,
515
+ "kl": 0.001443372806534171,
516
+ "learning_rate": 3e-06,
517
+ "loss": -0.0777,
518
+ "reward": 1.2708333730697632,
519
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
520
+ "rewards/correctness_reward_func_math": 0.6458333283662796,
521
+ "step": 51
522
+ },
523
+ {
524
+ "epoch": 0.055466666666666664,
525
+ "grad_norm": 0.09066792577505112,
526
+ "kl": 0.0014335185405798256,
527
+ "learning_rate": 3e-06,
528
+ "loss": -0.0774,
529
+ "step": 52
530
+ },
531
+ {
532
+ "completion_length": 251.92708587646484,
533
+ "epoch": 0.05653333333333333,
534
+ "grad_norm": 0.08732093125581741,
535
+ "kl": 0.0016631721518933773,
536
+ "learning_rate": 3e-06,
537
+ "loss": -0.0009,
538
+ "reward": 1.2708333730697632,
539
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
540
+ "rewards/correctness_reward_func_math": 0.6458333544433117,
541
+ "step": 53
542
+ },
543
+ {
544
+ "epoch": 0.0576,
545
+ "grad_norm": 0.08641738444566727,
546
+ "kl": 0.0018310246523469687,
547
+ "learning_rate": 3e-06,
548
+ "loss": -0.0005,
549
+ "step": 54
550
+ },
551
+ {
552
+ "completion_length": 244.20833587646484,
553
+ "epoch": 0.058666666666666666,
554
+ "grad_norm": 0.10797394067049026,
555
+ "kl": 0.0015712621388956904,
556
+ "learning_rate": 3e-06,
557
+ "loss": 0.0333,
558
+ "reward": 1.109375,
559
+ "rewards/boxed_and_answer_tags_format_reward": 0.5260416567325592,
560
+ "rewards/correctness_reward_func_math": 0.5833333432674408,
561
+ "step": 55
562
+ },
563
+ {
564
+ "epoch": 0.05973333333333333,
565
+ "grad_norm": 0.11483337730169296,
566
+ "kl": 0.0016164358239620924,
567
+ "learning_rate": 3e-06,
568
+ "loss": 0.0336,
569
+ "step": 56
570
+ },
571
+ {
572
+ "completion_length": 244.52083587646484,
573
+ "epoch": 0.0608,
574
+ "grad_norm": 0.11415942013263702,
575
+ "kl": 0.002258998923934996,
576
+ "learning_rate": 3e-06,
577
+ "loss": 0.1416,
578
+ "reward": 1.1614583730697632,
579
+ "rewards/boxed_and_answer_tags_format_reward": 0.6614583432674408,
580
+ "rewards/correctness_reward_func_math": 0.4999999850988388,
581
+ "step": 57
582
+ },
583
+ {
584
+ "epoch": 0.06186666666666667,
585
+ "grad_norm": 0.11037321388721466,
586
+ "kl": 0.0020339792827144265,
587
+ "learning_rate": 3e-06,
588
+ "loss": 0.1416,
589
+ "step": 58
590
+ },
591
+ {
592
+ "completion_length": 252.5416717529297,
593
+ "epoch": 0.06293333333333333,
594
+ "grad_norm": 0.09595922380685806,
595
+ "kl": 0.0017390991561114788,
596
+ "learning_rate": 3e-06,
597
+ "loss": 0.0304,
598
+ "reward": 1.4270833730697632,
599
+ "rewards/boxed_and_answer_tags_format_reward": 0.6145833432674408,
600
+ "rewards/correctness_reward_func_math": 0.8125,
601
+ "step": 59
602
+ },
603
+ {
604
+ "epoch": 0.064,
605
+ "grad_norm": 0.10574597865343094,
606
+ "kl": 0.001767849549651146,
607
+ "learning_rate": 3e-06,
608
+ "loss": 0.0299,
609
+ "step": 60
610
+ },
611
+ {
612
+ "completion_length": 244.48958587646484,
613
+ "epoch": 0.06506666666666666,
614
+ "grad_norm": 0.09615544974803925,
615
+ "kl": 0.0017314389697276056,
616
+ "learning_rate": 3e-06,
617
+ "loss": 0.1035,
618
+ "reward": 1.1562500596046448,
619
+ "rewards/boxed_and_answer_tags_format_reward": 0.5729166567325592,
620
+ "rewards/correctness_reward_func_math": 0.5833333283662796,
621
+ "step": 61
622
+ },
623
+ {
624
+ "epoch": 0.06613333333333334,
625
+ "grad_norm": 0.0875302329659462,
626
+ "kl": 0.0015769051969982684,
627
+ "learning_rate": 3e-06,
628
+ "loss": 0.1032,
629
+ "step": 62
630
+ },
631
+ {
632
+ "completion_length": 248.92709350585938,
633
+ "epoch": 0.0672,
634
+ "grad_norm": 0.09455385059118271,
635
+ "kl": 0.00198291870765388,
636
+ "learning_rate": 3e-06,
637
+ "loss": 0.0419,
638
+ "reward": 1.4166667461395264,
639
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
640
+ "rewards/correctness_reward_func_math": 0.7916666567325592,
641
+ "step": 63
642
+ },
643
+ {
644
+ "epoch": 0.06826666666666667,
645
+ "grad_norm": 0.12381947785615921,
646
+ "kl": 0.002091848640702665,
647
+ "learning_rate": 3e-06,
648
+ "loss": 0.0418,
649
+ "step": 64
650
+ },
651
+ {
652
+ "completion_length": 244.25000762939453,
653
+ "epoch": 0.06933333333333333,
654
+ "grad_norm": 0.09528794884681702,
655
+ "kl": 0.0020358916372060776,
656
+ "learning_rate": 3e-06,
657
+ "loss": 0.0797,
658
+ "reward": 1.0833333432674408,
659
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
660
+ "rewards/correctness_reward_func_math": 0.4583333358168602,
661
+ "step": 65
662
+ },
663
+ {
664
+ "epoch": 0.0704,
665
+ "grad_norm": 0.09021297842264175,
666
+ "kl": 0.0021131213288754225,
667
+ "learning_rate": 3e-06,
668
+ "loss": 0.0796,
669
+ "step": 66
670
+ },
671
+ {
672
+ "completion_length": 242.2604217529297,
673
+ "epoch": 0.07146666666666666,
674
+ "grad_norm": 0.11209271103143692,
675
+ "kl": 0.0016552057350054383,
676
+ "learning_rate": 3e-06,
677
+ "loss": 0.0905,
678
+ "reward": 1.2291666865348816,
679
+ "rewards/boxed_and_answer_tags_format_reward": 0.5625,
680
+ "rewards/correctness_reward_func_math": 0.6666666865348816,
681
+ "step": 67
682
+ },
683
+ {
684
+ "epoch": 0.07253333333333334,
685
+ "grad_norm": 0.1326776146888733,
686
+ "kl": 0.0017586728790774941,
687
+ "learning_rate": 3e-06,
688
+ "loss": 0.0906,
689
+ "step": 68
690
+ },
691
+ {
692
+ "completion_length": 233.9166717529297,
693
+ "epoch": 0.0736,
694
+ "grad_norm": 0.09971653670072556,
695
+ "kl": 0.0016736852703616023,
696
+ "learning_rate": 3e-06,
697
+ "loss": 0.0235,
698
+ "reward": 1.1145833730697632,
699
+ "rewards/boxed_and_answer_tags_format_reward": 0.5729166567325592,
700
+ "rewards/correctness_reward_func_math": 0.5416666865348816,
701
+ "step": 69
702
+ },
703
+ {
704
+ "epoch": 0.07466666666666667,
705
+ "grad_norm": 0.09656917303800583,
706
+ "kl": 0.0016576185007579625,
707
+ "learning_rate": 3e-06,
708
+ "loss": 0.0236,
709
+ "step": 70
710
+ },
711
+ {
712
+ "completion_length": 245.3854217529297,
713
+ "epoch": 0.07573333333333333,
714
+ "grad_norm": 0.09661854803562164,
715
+ "kl": 0.0026833119336515665,
716
+ "learning_rate": 3e-06,
717
+ "loss": -0.0027,
718
+ "reward": 1.0520833730697632,
719
+ "rewards/boxed_and_answer_tags_format_reward": 0.5520833134651184,
720
+ "rewards/correctness_reward_func_math": 0.4999999850988388,
721
+ "step": 71
722
+ },
723
+ {
724
+ "epoch": 0.0768,
725
+ "grad_norm": 0.10463777184486389,
726
+ "kl": 0.0024712955346331,
727
+ "learning_rate": 3e-06,
728
+ "loss": -0.0035,
729
+ "step": 72
730
+ },
731
+ {
732
+ "completion_length": 251.48958587646484,
733
+ "epoch": 0.07786666666666667,
734
+ "grad_norm": 0.12469884753227234,
735
+ "kl": 0.002885082154534757,
736
+ "learning_rate": 3e-06,
737
+ "loss": 0.0742,
738
+ "reward": 1.0677083432674408,
739
+ "rewards/boxed_and_answer_tags_format_reward": 0.5885416567325592,
740
+ "rewards/correctness_reward_func_math": 0.4791666716337204,
741
+ "step": 73
742
+ },
743
+ {
744
+ "epoch": 0.07893333333333333,
745
+ "grad_norm": 0.12237768620252609,
746
+ "kl": 0.002940670936368406,
747
+ "learning_rate": 3e-06,
748
+ "loss": 0.0748,
749
+ "step": 74
750
+ },
751
+ {
752
+ "completion_length": 250.3125,
753
+ "epoch": 0.08,
754
+ "grad_norm": 0.1503533273935318,
755
+ "kl": 0.0019364022300578654,
756
+ "learning_rate": 3e-06,
757
+ "loss": 0.0117,
758
+ "reward": 1.078125,
759
+ "rewards/boxed_and_answer_tags_format_reward": 0.578125,
760
+ "rewards/correctness_reward_func_math": 0.5000000149011612,
761
+ "step": 75
762
+ },
763
+ {
764
+ "epoch": 0.08106666666666666,
765
+ "grad_norm": 0.12449654191732407,
766
+ "kl": 0.0019527390832081437,
767
+ "learning_rate": 3e-06,
768
+ "loss": 0.0122,
769
+ "step": 76
770
+ },
771
+ {
772
+ "completion_length": 241.89583587646484,
773
+ "epoch": 0.08213333333333334,
774
+ "grad_norm": 0.0918169841170311,
775
+ "kl": 0.002860512933693826,
776
+ "learning_rate": 3e-06,
777
+ "loss": 0.0417,
778
+ "reward": 1.2968750298023224,
779
+ "rewards/boxed_and_answer_tags_format_reward": 0.5468750149011612,
780
+ "rewards/correctness_reward_func_math": 0.7499999850988388,
781
+ "step": 77
782
+ },
783
+ {
784
+ "epoch": 0.0832,
785
+ "grad_norm": 0.09818709641695023,
786
+ "kl": 0.0027386642759665847,
787
+ "learning_rate": 3e-06,
788
+ "loss": 0.0418,
789
+ "step": 78
790
+ },
791
+ {
792
+ "completion_length": 250.05209350585938,
793
+ "epoch": 0.08426666666666667,
794
+ "grad_norm": 0.12854671478271484,
795
+ "kl": 0.002327249851077795,
796
+ "learning_rate": 3e-06,
797
+ "loss": -0.0002,
798
+ "reward": 1.0885416865348816,
799
+ "rewards/boxed_and_answer_tags_format_reward": 0.609375,
800
+ "rewards/correctness_reward_func_math": 0.4791666716337204,
801
+ "step": 79
802
+ },
803
+ {
804
+ "epoch": 0.08533333333333333,
805
+ "grad_norm": 0.13319538533687592,
806
+ "kl": 0.0023979160469025373,
807
+ "learning_rate": 3e-06,
808
+ "loss": -0.0004,
809
+ "step": 80
810
+ },
811
+ {
812
+ "completion_length": 244.50000762939453,
813
+ "epoch": 0.0864,
814
+ "grad_norm": 0.1296762228012085,
815
+ "kl": 0.0030258659971877933,
816
+ "learning_rate": 3e-06,
817
+ "loss": 0.123,
818
+ "reward": 1.4270833730697632,
819
+ "rewards/boxed_and_answer_tags_format_reward": 0.65625,
820
+ "rewards/correctness_reward_func_math": 0.7708333432674408,
821
+ "step": 81
822
+ },
823
+ {
824
+ "epoch": 0.08746666666666666,
825
+ "grad_norm": 0.11535439640283585,
826
+ "kl": 0.0028890471439808607,
827
+ "learning_rate": 3e-06,
828
+ "loss": 0.1225,
829
+ "step": 82
830
+ },
831
+ {
832
+ "completion_length": 246.8541717529297,
833
+ "epoch": 0.08853333333333334,
834
+ "grad_norm": 0.10547207295894623,
835
+ "kl": 0.003146483562886715,
836
+ "learning_rate": 3e-06,
837
+ "loss": 0.1346,
838
+ "reward": 1.4427083730697632,
839
+ "rewards/boxed_and_answer_tags_format_reward": 0.671875,
840
+ "rewards/correctness_reward_func_math": 0.7708333432674408,
841
+ "step": 83
842
+ },
843
+ {
844
+ "epoch": 0.0896,
845
+ "grad_norm": 0.1115085780620575,
846
+ "kl": 0.003225739812478423,
847
+ "learning_rate": 3e-06,
848
+ "loss": 0.1347,
849
+ "step": 84
850
+ },
851
+ {
852
+ "completion_length": 245.43750762939453,
853
+ "epoch": 0.09066666666666667,
854
+ "grad_norm": 0.10426100343465805,
855
+ "kl": 0.0026978254318237305,
856
+ "learning_rate": 3e-06,
857
+ "loss": -0.0577,
858
+ "reward": 1.0000000298023224,
859
+ "rewards/boxed_and_answer_tags_format_reward": 0.6041666567325592,
860
+ "rewards/correctness_reward_func_math": 0.3958333358168602,
861
+ "step": 85
862
+ },
863
+ {
864
+ "epoch": 0.09173333333333333,
865
+ "grad_norm": 0.10135605931282043,
866
+ "kl": 0.0027983213076367974,
867
+ "learning_rate": 3e-06,
868
+ "loss": -0.0582,
869
+ "step": 86
870
+ },
871
+ {
872
+ "completion_length": 244.25000762939453,
873
+ "epoch": 0.0928,
874
+ "grad_norm": 0.13423067331314087,
875
+ "kl": 0.003659309819340706,
876
+ "learning_rate": 3e-06,
877
+ "loss": 0.0473,
878
+ "reward": 1.3697916865348816,
879
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916567325592,
880
+ "rewards/correctness_reward_func_math": 0.75,
881
+ "step": 87
882
+ },
883
+ {
884
+ "epoch": 0.09386666666666667,
885
+ "grad_norm": 0.13260383903980255,
886
+ "kl": 0.0036486699245870113,
887
+ "learning_rate": 3e-06,
888
+ "loss": 0.0467,
889
+ "step": 88
890
+ },
891
+ {
892
+ "completion_length": 250.84375762939453,
893
+ "epoch": 0.09493333333333333,
894
+ "grad_norm": 0.08243498206138611,
895
+ "kl": 0.0021177935414016247,
896
+ "learning_rate": 3e-06,
897
+ "loss": -0.022,
898
+ "reward": 0.7708333730697632,
899
+ "rewards/boxed_and_answer_tags_format_reward": 0.4791666716337204,
900
+ "rewards/correctness_reward_func_math": 0.2916666641831398,
901
+ "step": 89
902
+ },
903
+ {
904
+ "epoch": 0.096,
905
+ "grad_norm": 0.08511421829462051,
906
+ "kl": 0.0021853672806173563,
907
+ "learning_rate": 3e-06,
908
+ "loss": -0.0225,
909
+ "step": 90
910
+ },
911
+ {
912
+ "completion_length": 246.9479217529297,
913
+ "epoch": 0.09706666666666666,
914
+ "grad_norm": 0.08923275023698807,
915
+ "kl": 0.0031975428573787212,
916
+ "learning_rate": 3e-06,
917
+ "loss": -0.0756,
918
+ "reward": 1.1145833730697632,
919
+ "rewards/boxed_and_answer_tags_format_reward": 0.59375,
920
+ "rewards/correctness_reward_func_math": 0.5208333283662796,
921
+ "step": 91
922
+ },
923
+ {
924
+ "epoch": 0.09813333333333334,
925
+ "grad_norm": 0.09071174263954163,
926
+ "kl": 0.003432211116887629,
927
+ "learning_rate": 3e-06,
928
+ "loss": -0.0761,
929
+ "step": 92
930
+ },
931
+ {
932
+ "completion_length": 236.875,
933
+ "epoch": 0.0992,
934
+ "grad_norm": 0.1429155021905899,
935
+ "kl": 0.004054961493238807,
936
+ "learning_rate": 3e-06,
937
+ "loss": -0.0223,
938
+ "reward": 1.4114583730697632,
939
+ "rewards/boxed_and_answer_tags_format_reward": 0.6822916567325592,
940
+ "rewards/correctness_reward_func_math": 0.729166641831398,
941
+ "step": 93
942
+ },
943
+ {
944
+ "epoch": 0.10026666666666667,
945
+ "grad_norm": 0.17755988240242004,
946
+ "kl": 0.004182609496638179,
947
+ "learning_rate": 3e-06,
948
+ "loss": -0.0215,
949
+ "step": 94
950
+ },
951
+ {
952
+ "completion_length": 240.81250762939453,
953
+ "epoch": 0.10133333333333333,
954
+ "grad_norm": 0.058556366711854935,
955
+ "kl": 0.0032166338060051203,
956
+ "learning_rate": 3e-06,
957
+ "loss": 0.0262,
958
+ "reward": 0.96875,
959
+ "rewards/boxed_and_answer_tags_format_reward": 0.5520833134651184,
960
+ "rewards/correctness_reward_func_math": 0.4166666716337204,
961
+ "step": 95
962
+ },
963
+ {
964
+ "epoch": 0.1024,
965
+ "grad_norm": 0.055329516530036926,
966
+ "kl": 0.0033893719082698226,
967
+ "learning_rate": 3e-06,
968
+ "loss": 0.0259,
969
+ "step": 96
970
+ },
971
+ {
972
+ "completion_length": 239.7916717529297,
973
+ "epoch": 0.10346666666666667,
974
+ "grad_norm": 0.12691481411457062,
975
+ "kl": 0.004895984660834074,
976
+ "learning_rate": 3e-06,
977
+ "loss": 0.0493,
978
+ "reward": 1.0572916865348816,
979
+ "rewards/boxed_and_answer_tags_format_reward": 0.640625,
980
+ "rewards/correctness_reward_func_math": 0.4166666567325592,
981
+ "step": 97
982
+ },
983
+ {
984
+ "epoch": 0.10453333333333334,
985
+ "grad_norm": 0.12851038575172424,
986
+ "kl": 0.004392273956909776,
987
+ "learning_rate": 3e-06,
988
+ "loss": 0.0489,
989
+ "step": 98
990
+ },
991
+ {
992
+ "completion_length": 243.52084350585938,
993
+ "epoch": 0.1056,
994
+ "grad_norm": 0.07937366515398026,
995
+ "kl": 0.0044018065091222525,
996
+ "learning_rate": 3e-06,
997
+ "loss": 0.0643,
998
+ "reward": 1.1562500298023224,
999
+ "rewards/boxed_and_answer_tags_format_reward": 0.5729166567325592,
1000
+ "rewards/correctness_reward_func_math": 0.5833333283662796,
1001
+ "step": 99
1002
+ },
1003
+ {
1004
+ "epoch": 0.10666666666666667,
1005
+ "grad_norm": 0.08331720530986786,
1006
+ "kl": 0.0046686476562172174,
1007
+ "learning_rate": 3e-06,
1008
+ "loss": 0.0648,
1009
+ "step": 100
1010
+ },
1011
+ {
1012
+ "completion_length": 250.5104217529297,
1013
+ "epoch": 0.10773333333333333,
1014
+ "grad_norm": 0.1329662948846817,
1015
+ "kl": 0.004440172808244824,
1016
+ "learning_rate": 3e-06,
1017
+ "loss": 0.1007,
1018
+ "reward": 1.2395833730697632,
1019
+ "rewards/boxed_and_answer_tags_format_reward": 0.6354166567325592,
1020
+ "rewards/correctness_reward_func_math": 0.6041666567325592,
1021
+ "step": 101
1022
+ },
1023
+ {
1024
+ "epoch": 0.1088,
1025
+ "grad_norm": 0.12988623976707458,
1026
+ "kl": 0.004622329259291291,
1027
+ "learning_rate": 3e-06,
1028
+ "loss": 0.1006,
1029
+ "step": 102
1030
+ },
1031
+ {
1032
+ "completion_length": 247.4479217529297,
1033
+ "epoch": 0.10986666666666667,
1034
+ "grad_norm": 0.10841701924800873,
1035
+ "kl": 0.003746375907212496,
1036
+ "learning_rate": 3e-06,
1037
+ "loss": 0.0357,
1038
+ "reward": 1.046875,
1039
+ "rewards/boxed_and_answer_tags_format_reward": 0.5885416567325592,
1040
+ "rewards/correctness_reward_func_math": 0.4583333283662796,
1041
+ "step": 103
1042
+ },
1043
+ {
1044
+ "epoch": 0.11093333333333333,
1045
+ "grad_norm": 0.10790833830833435,
1046
+ "kl": 0.0037733702920377254,
1047
+ "learning_rate": 3e-06,
1048
+ "loss": 0.0355,
1049
+ "step": 104
1050
+ },
1051
+ {
1052
+ "completion_length": 246.00000762939453,
1053
+ "epoch": 0.112,
1054
+ "grad_norm": 0.06712818145751953,
1055
+ "kl": 0.004377630655653775,
1056
+ "learning_rate": 3e-06,
1057
+ "loss": 0.0165,
1058
+ "reward": 1.359375,
1059
+ "rewards/boxed_and_answer_tags_format_reward": 0.6302083432674408,
1060
+ "rewards/correctness_reward_func_math": 0.7291666865348816,
1061
+ "step": 105
1062
+ },
1063
+ {
1064
+ "epoch": 0.11306666666666666,
1065
+ "grad_norm": 0.06803179532289505,
1066
+ "kl": 0.004485110635869205,
1067
+ "learning_rate": 3e-06,
1068
+ "loss": 0.0163,
1069
+ "step": 106
1070
+ },
1071
+ {
1072
+ "completion_length": 252.03125,
1073
+ "epoch": 0.11413333333333334,
1074
+ "grad_norm": 0.12938225269317627,
1075
+ "kl": 0.004964495077729225,
1076
+ "learning_rate": 3e-06,
1077
+ "loss": 0.0035,
1078
+ "reward": 1.6041666865348816,
1079
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1080
+ "rewards/correctness_reward_func_math": 0.9791666865348816,
1081
+ "step": 107
1082
+ },
1083
+ {
1084
+ "epoch": 0.1152,
1085
+ "grad_norm": 0.13050013780593872,
1086
+ "kl": 0.00515007833018899,
1087
+ "learning_rate": 3e-06,
1088
+ "loss": 0.0028,
1089
+ "step": 108
1090
+ },
1091
+ {
1092
+ "completion_length": 248.77084350585938,
1093
+ "epoch": 0.11626666666666667,
1094
+ "grad_norm": 0.0982803925871849,
1095
+ "kl": 0.004882118897512555,
1096
+ "learning_rate": 3e-06,
1097
+ "loss": 0.0344,
1098
+ "reward": 1.1458333730697632,
1099
+ "rewards/boxed_and_answer_tags_format_reward": 0.6458333432674408,
1100
+ "rewards/correctness_reward_func_math": 0.5000000149011612,
1101
+ "step": 109
1102
+ },
1103
+ {
1104
+ "epoch": 0.11733333333333333,
1105
+ "grad_norm": 0.10258838534355164,
1106
+ "kl": 0.004875096259638667,
1107
+ "learning_rate": 3e-06,
1108
+ "loss": 0.034,
1109
+ "step": 110
1110
+ },
1111
+ {
1112
+ "completion_length": 248.0416717529297,
1113
+ "epoch": 0.1184,
1114
+ "grad_norm": 0.10756522417068481,
1115
+ "kl": 0.0036248815013095737,
1116
+ "learning_rate": 3e-06,
1117
+ "loss": 0.0517,
1118
+ "reward": 0.8593750298023224,
1119
+ "rewards/boxed_and_answer_tags_format_reward": 0.5677083134651184,
1120
+ "rewards/correctness_reward_func_math": 0.2916666716337204,
1121
+ "step": 111
1122
+ },
1123
+ {
1124
+ "epoch": 0.11946666666666667,
1125
+ "grad_norm": 0.09450583159923553,
1126
+ "kl": 0.0038704528706148267,
1127
+ "learning_rate": 3e-06,
1128
+ "loss": 0.0517,
1129
+ "step": 112
1130
+ },
1131
+ {
1132
+ "completion_length": 230.9166717529297,
1133
+ "epoch": 0.12053333333333334,
1134
+ "grad_norm": 0.0950523242354393,
1135
+ "kl": 0.004498519003391266,
1136
+ "learning_rate": 3e-06,
1137
+ "loss": 0.0572,
1138
+ "reward": 1.4583333730697632,
1139
+ "rewards/boxed_and_answer_tags_format_reward": 0.6666666567325592,
1140
+ "rewards/correctness_reward_func_math": 0.791666641831398,
1141
+ "step": 113
1142
+ },
1143
+ {
1144
+ "epoch": 0.1216,
1145
+ "grad_norm": 0.09974662959575653,
1146
+ "kl": 0.00460103340446949,
1147
+ "learning_rate": 3e-06,
1148
+ "loss": 0.0566,
1149
+ "step": 114
1150
+ },
1151
+ {
1152
+ "completion_length": 234.34375762939453,
1153
+ "epoch": 0.12266666666666666,
1154
+ "grad_norm": 0.07447686791419983,
1155
+ "kl": 0.005343697499483824,
1156
+ "learning_rate": 3e-06,
1157
+ "loss": 0.0391,
1158
+ "reward": 0.9895833730697632,
1159
+ "rewards/boxed_and_answer_tags_format_reward": 0.6354166567325592,
1160
+ "rewards/correctness_reward_func_math": 0.3541666567325592,
1161
+ "step": 115
1162
+ },
1163
+ {
1164
+ "epoch": 0.12373333333333333,
1165
+ "grad_norm": 0.07214193046092987,
1166
+ "kl": 0.0053415135480463505,
1167
+ "learning_rate": 3e-06,
1168
+ "loss": 0.0391,
1169
+ "step": 116
1170
+ },
1171
+ {
1172
+ "completion_length": 237.39583587646484,
1173
+ "epoch": 0.1248,
1174
+ "grad_norm": 0.09537629038095474,
1175
+ "kl": 0.005689415615051985,
1176
+ "learning_rate": 3e-06,
1177
+ "loss": 0.0707,
1178
+ "reward": 1.2083333730697632,
1179
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1180
+ "rewards/correctness_reward_func_math": 0.5833333432674408,
1181
+ "step": 117
1182
+ },
1183
+ {
1184
+ "epoch": 0.12586666666666665,
1185
+ "grad_norm": 0.10276316851377487,
1186
+ "kl": 0.006041191751137376,
1187
+ "learning_rate": 3e-06,
1188
+ "loss": 0.0706,
1189
+ "step": 118
1190
+ },
1191
+ {
1192
+ "completion_length": 242.62500762939453,
1193
+ "epoch": 0.12693333333333334,
1194
+ "grad_norm": 0.09985591471195221,
1195
+ "kl": 0.006303998176008463,
1196
+ "learning_rate": 3e-06,
1197
+ "loss": 0.0517,
1198
+ "reward": 1.3125000596046448,
1199
+ "rewards/boxed_and_answer_tags_format_reward": 0.6666666567325592,
1200
+ "rewards/correctness_reward_func_math": 0.6458333432674408,
1201
+ "step": 119
1202
+ },
1203
+ {
1204
+ "epoch": 0.128,
1205
+ "grad_norm": 0.09995035827159882,
1206
+ "kl": 0.006645566551014781,
1207
+ "learning_rate": 3e-06,
1208
+ "loss": 0.0518,
1209
+ "step": 120
1210
+ },
1211
+ {
1212
+ "completion_length": 251.70833587646484,
1213
+ "epoch": 0.12906666666666666,
1214
+ "grad_norm": 0.08751357346773148,
1215
+ "kl": 0.005795367527753115,
1216
+ "learning_rate": 3e-06,
1217
+ "loss": 0.0718,
1218
+ "reward": 1.3854166865348816,
1219
+ "rewards/boxed_and_answer_tags_format_reward": 0.6354166865348816,
1220
+ "rewards/correctness_reward_func_math": 0.75,
1221
+ "step": 121
1222
+ },
1223
+ {
1224
+ "epoch": 0.13013333333333332,
1225
+ "grad_norm": 0.09613488614559174,
1226
+ "kl": 0.005708918673917651,
1227
+ "learning_rate": 3e-06,
1228
+ "loss": 0.0719,
1229
+ "step": 122
1230
+ },
1231
+ {
1232
+ "completion_length": 235.23958587646484,
1233
+ "epoch": 0.1312,
1234
+ "grad_norm": 0.15242788195610046,
1235
+ "kl": 0.0065653217025101185,
1236
+ "learning_rate": 3e-06,
1237
+ "loss": 0.1279,
1238
+ "reward": 1.3802083730697632,
1239
+ "rewards/boxed_and_answer_tags_format_reward": 0.6510416567325592,
1240
+ "rewards/correctness_reward_func_math": 0.729166641831398,
1241
+ "step": 123
1242
+ },
1243
+ {
1244
+ "epoch": 0.13226666666666667,
1245
+ "grad_norm": 0.1486300230026245,
1246
+ "kl": 0.006940900115296245,
1247
+ "learning_rate": 3e-06,
1248
+ "loss": 0.1283,
1249
+ "step": 124
1250
+ },
1251
+ {
1252
+ "completion_length": 244.3229217529297,
1253
+ "epoch": 0.13333333333333333,
1254
+ "grad_norm": 0.0940360352396965,
1255
+ "kl": 0.004931023344397545,
1256
+ "learning_rate": 3e-06,
1257
+ "loss": 0.0278,
1258
+ "reward": 0.9531250298023224,
1259
+ "rewards/boxed_and_answer_tags_format_reward": 0.4947916567325592,
1260
+ "rewards/correctness_reward_func_math": 0.4583333283662796,
1261
+ "step": 125
1262
+ },
1263
+ {
1264
+ "epoch": 0.1344,
1265
+ "grad_norm": 0.09261205792427063,
1266
+ "kl": 0.00510846171528101,
1267
+ "learning_rate": 3e-06,
1268
+ "loss": 0.0273,
1269
+ "step": 126
1270
+ },
1271
+ {
1272
+ "completion_length": 244.02083587646484,
1273
+ "epoch": 0.13546666666666668,
1274
+ "grad_norm": 0.15479284524917603,
1275
+ "kl": 0.008449542336165905,
1276
+ "learning_rate": 3e-06,
1277
+ "loss": -0.0406,
1278
+ "reward": 1.0989583432674408,
1279
+ "rewards/boxed_and_answer_tags_format_reward": 0.578125,
1280
+ "rewards/correctness_reward_func_math": 0.5208333432674408,
1281
+ "step": 127
1282
+ },
1283
+ {
1284
+ "epoch": 0.13653333333333334,
1285
+ "grad_norm": 0.15614548325538635,
1286
+ "kl": 0.008300963789224625,
1287
+ "learning_rate": 3e-06,
1288
+ "loss": -0.0413,
1289
+ "step": 128
1290
+ },
1291
+ {
1292
+ "completion_length": 241.4166717529297,
1293
+ "epoch": 0.1376,
1294
+ "grad_norm": 0.13294053077697754,
1295
+ "kl": 0.007204608526080847,
1296
+ "learning_rate": 3e-06,
1297
+ "loss": 0.0396,
1298
+ "reward": 1.1510416865348816,
1299
+ "rewards/boxed_and_answer_tags_format_reward": 0.609375,
1300
+ "rewards/correctness_reward_func_math": 0.5416666716337204,
1301
+ "step": 129
1302
+ },
1303
+ {
1304
+ "epoch": 0.13866666666666666,
1305
+ "grad_norm": 0.13154013454914093,
1306
+ "kl": 0.007467167917639017,
1307
+ "learning_rate": 3e-06,
1308
+ "loss": 0.0399,
1309
+ "step": 130
1310
+ },
1311
+ {
1312
+ "completion_length": 240.6041717529297,
1313
+ "epoch": 0.13973333333333332,
1314
+ "grad_norm": 0.09483897686004639,
1315
+ "kl": 0.007113585248589516,
1316
+ "learning_rate": 3e-06,
1317
+ "loss": 0.0289,
1318
+ "reward": 1.1666666865348816,
1319
+ "rewards/boxed_and_answer_tags_format_reward": 0.6041666567325592,
1320
+ "rewards/correctness_reward_func_math": 0.5625,
1321
+ "step": 131
1322
+ },
1323
+ {
1324
+ "epoch": 0.1408,
1325
+ "grad_norm": 0.09663183242082596,
1326
+ "kl": 0.007714268751442432,
1327
+ "learning_rate": 3e-06,
1328
+ "loss": 0.0287,
1329
+ "step": 132
1330
+ },
1331
+ {
1332
+ "completion_length": 240.0625,
1333
+ "epoch": 0.14186666666666667,
1334
+ "grad_norm": 0.11623935401439667,
1335
+ "kl": 0.009474250487983227,
1336
+ "learning_rate": 3e-06,
1337
+ "loss": -0.0188,
1338
+ "reward": 1.1614583730697632,
1339
+ "rewards/boxed_and_answer_tags_format_reward": 0.578125,
1340
+ "rewards/correctness_reward_func_math": 0.5833333283662796,
1341
+ "step": 133
1342
+ },
1343
+ {
1344
+ "epoch": 0.14293333333333333,
1345
+ "grad_norm": 0.11035683006048203,
1346
+ "kl": 0.0097966562025249,
1347
+ "learning_rate": 3e-06,
1348
+ "loss": -0.0192,
1349
+ "step": 134
1350
+ },
1351
+ {
1352
+ "completion_length": 245.53125,
1353
+ "epoch": 0.144,
1354
+ "grad_norm": 0.08813372999429703,
1355
+ "kl": 0.009550870396196842,
1356
+ "learning_rate": 3e-06,
1357
+ "loss": -0.0355,
1358
+ "reward": 1.4114583730697632,
1359
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916865348816,
1360
+ "rewards/correctness_reward_func_math": 0.7916666865348816,
1361
+ "step": 135
1362
+ },
1363
+ {
1364
+ "epoch": 0.14506666666666668,
1365
+ "grad_norm": 0.08638645708560944,
1366
+ "kl": 0.009892939124256372,
1367
+ "learning_rate": 3e-06,
1368
+ "loss": -0.0356,
1369
+ "step": 136
1370
+ },
1371
+ {
1372
+ "completion_length": 238.7604217529297,
1373
+ "epoch": 0.14613333333333334,
1374
+ "grad_norm": 0.09811553359031677,
1375
+ "kl": 0.00993044814094901,
1376
+ "learning_rate": 3e-06,
1377
+ "loss": -0.0298,
1378
+ "reward": 1.3125,
1379
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1380
+ "rewards/correctness_reward_func_math": 0.6875,
1381
+ "step": 137
1382
+ },
1383
+ {
1384
+ "epoch": 0.1472,
1385
+ "grad_norm": 0.10006941854953766,
1386
+ "kl": 0.010414571966975927,
1387
+ "learning_rate": 3e-06,
1388
+ "loss": -0.0301,
1389
+ "step": 138
1390
+ },
1391
+ {
1392
+ "completion_length": 241.1875,
1393
+ "epoch": 0.14826666666666666,
1394
+ "grad_norm": 0.09777302294969559,
1395
+ "kl": 0.008060592226684093,
1396
+ "learning_rate": 3e-06,
1397
+ "loss": 0.0531,
1398
+ "reward": 1.3437500596046448,
1399
+ "rewards/boxed_and_answer_tags_format_reward": 0.4687499850988388,
1400
+ "rewards/correctness_reward_func_math": 0.8749999701976776,
1401
+ "step": 139
1402
+ },
1403
+ {
1404
+ "epoch": 0.14933333333333335,
1405
+ "grad_norm": 0.1021089032292366,
1406
+ "kl": 0.0083432886749506,
1407
+ "learning_rate": 3e-06,
1408
+ "loss": 0.0534,
1409
+ "step": 140
1410
+ },
1411
+ {
1412
+ "completion_length": 249.84375762939453,
1413
+ "epoch": 0.1504,
1414
+ "grad_norm": 0.14115490019321442,
1415
+ "kl": 0.009720872156322002,
1416
+ "learning_rate": 3e-06,
1417
+ "loss": 0.0429,
1418
+ "reward": 0.9427083432674408,
1419
+ "rewards/boxed_and_answer_tags_format_reward": 0.671875,
1420
+ "rewards/correctness_reward_func_math": 0.2708333358168602,
1421
+ "step": 141
1422
+ },
1423
+ {
1424
+ "epoch": 0.15146666666666667,
1425
+ "grad_norm": 0.1028415858745575,
1426
+ "kl": 0.009948289953172207,
1427
+ "learning_rate": 3e-06,
1428
+ "loss": 0.0428,
1429
+ "step": 142
1430
+ },
1431
+ {
1432
+ "completion_length": 232.7916717529297,
1433
+ "epoch": 0.15253333333333333,
1434
+ "grad_norm": 0.09448906034231186,
1435
+ "kl": 0.010104351677000523,
1436
+ "learning_rate": 3e-06,
1437
+ "loss": 0.0293,
1438
+ "reward": 1.3020833730697632,
1439
+ "rewards/boxed_and_answer_tags_format_reward": 0.6770833432674408,
1440
+ "rewards/correctness_reward_func_math": 0.6250000149011612,
1441
+ "step": 143
1442
+ },
1443
+ {
1444
+ "epoch": 0.1536,
1445
+ "grad_norm": 0.10912659764289856,
1446
+ "kl": 0.010456445161253214,
1447
+ "learning_rate": 3e-06,
1448
+ "loss": 0.03,
1449
+ "step": 144
1450
+ },
1451
+ {
1452
+ "completion_length": 240.39583587646484,
1453
+ "epoch": 0.15466666666666667,
1454
+ "grad_norm": 0.09563589841127396,
1455
+ "kl": 0.013137076515704393,
1456
+ "learning_rate": 3e-06,
1457
+ "loss": 0.0798,
1458
+ "reward": 1.28125,
1459
+ "rewards/boxed_and_answer_tags_format_reward": 0.6979166865348816,
1460
+ "rewards/correctness_reward_func_math": 0.583333320915699,
1461
+ "step": 145
1462
+ },
1463
+ {
1464
+ "epoch": 0.15573333333333333,
1465
+ "grad_norm": 0.10613968223333359,
1466
+ "kl": 0.013441369403153658,
1467
+ "learning_rate": 3e-06,
1468
+ "loss": 0.0786,
1469
+ "step": 146
1470
+ },
1471
+ {
1472
+ "completion_length": 245.61458587646484,
1473
+ "epoch": 0.1568,
1474
+ "grad_norm": 0.1437290906906128,
1475
+ "kl": 0.012657003942877054,
1476
+ "learning_rate": 3e-06,
1477
+ "loss": 0.0246,
1478
+ "reward": 1.4270833730697632,
1479
+ "rewards/boxed_and_answer_tags_format_reward": 0.6145833432674408,
1480
+ "rewards/correctness_reward_func_math": 0.8125,
1481
+ "step": 147
1482
+ },
1483
+ {
1484
+ "epoch": 0.15786666666666666,
1485
+ "grad_norm": 0.14002259075641632,
1486
+ "kl": 0.012809256091713905,
1487
+ "learning_rate": 3e-06,
1488
+ "loss": 0.024,
1489
+ "step": 148
1490
+ },
1491
+ {
1492
+ "completion_length": 241.08334350585938,
1493
+ "epoch": 0.15893333333333334,
1494
+ "grad_norm": 0.08180505782365799,
1495
+ "kl": 0.01138412905856967,
1496
+ "learning_rate": 3e-06,
1497
+ "loss": 0.0151,
1498
+ "reward": 1.0572916865348816,
1499
+ "rewards/boxed_and_answer_tags_format_reward": 0.5989583134651184,
1500
+ "rewards/correctness_reward_func_math": 0.4583333432674408,
1501
+ "step": 149
1502
+ },
1503
+ {
1504
+ "epoch": 0.16,
1505
+ "grad_norm": 0.08484908193349838,
1506
+ "kl": 0.01177709735929966,
1507
+ "learning_rate": 3e-06,
1508
+ "loss": 0.0147,
1509
+ "step": 150
1510
+ },
1511
+ {
1512
+ "completion_length": 242.2916717529297,
1513
+ "epoch": 0.16106666666666666,
1514
+ "grad_norm": 0.1212899312376976,
1515
+ "kl": 0.013707389123737812,
1516
+ "learning_rate": 3e-06,
1517
+ "loss": 0.1331,
1518
+ "reward": 1.3177083730697632,
1519
+ "rewards/boxed_and_answer_tags_format_reward": 0.6510416567325592,
1520
+ "rewards/correctness_reward_func_math": 0.6666666567325592,
1521
+ "step": 151
1522
+ },
1523
+ {
1524
+ "epoch": 0.16213333333333332,
1525
+ "grad_norm": 0.11805413663387299,
1526
+ "kl": 0.014071326702833176,
1527
+ "learning_rate": 3e-06,
1528
+ "loss": 0.1326,
1529
+ "step": 152
1530
+ },
1531
+ {
1532
+ "completion_length": 244.6979217529297,
1533
+ "epoch": 0.1632,
1534
+ "grad_norm": 0.12482283264398575,
1535
+ "kl": 0.017437617294490337,
1536
+ "learning_rate": 3e-06,
1537
+ "loss": 0.0238,
1538
+ "reward": 1.2187500596046448,
1539
+ "rewards/boxed_and_answer_tags_format_reward": 0.6354166865348816,
1540
+ "rewards/correctness_reward_func_math": 0.5833333432674408,
1541
+ "step": 153
1542
+ },
1543
+ {
1544
+ "epoch": 0.16426666666666667,
1545
+ "grad_norm": 0.16741231083869934,
1546
+ "kl": 0.01778746023774147,
1547
+ "learning_rate": 3e-06,
1548
+ "loss": 0.024,
1549
+ "step": 154
1550
+ },
1551
+ {
1552
+ "completion_length": 237.86458587646484,
1553
+ "epoch": 0.16533333333333333,
1554
+ "grad_norm": 0.09586331993341446,
1555
+ "kl": 0.019033060409128666,
1556
+ "learning_rate": 3e-06,
1557
+ "loss": -0.0003,
1558
+ "reward": 1.3541666865348816,
1559
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1560
+ "rewards/correctness_reward_func_math": 0.7291666567325592,
1561
+ "step": 155
1562
+ },
1563
+ {
1564
+ "epoch": 0.1664,
1565
+ "grad_norm": 0.09274693578481674,
1566
+ "kl": 0.019481662660837173,
1567
+ "learning_rate": 3e-06,
1568
+ "loss": -0.0007,
1569
+ "step": 156
1570
+ },
1571
+ {
1572
+ "completion_length": 234.06250762939453,
1573
+ "epoch": 0.16746666666666668,
1574
+ "grad_norm": 0.1973016858100891,
1575
+ "kl": 0.02091437578201294,
1576
+ "learning_rate": 3e-06,
1577
+ "loss": 0.0658,
1578
+ "reward": 1.2864583730697632,
1579
+ "rewards/boxed_and_answer_tags_format_reward": 0.6614583432674408,
1580
+ "rewards/correctness_reward_func_math": 0.625,
1581
+ "step": 157
1582
+ },
1583
+ {
1584
+ "epoch": 0.16853333333333334,
1585
+ "grad_norm": 0.1657065898180008,
1586
+ "kl": 0.02124563418328762,
1587
+ "learning_rate": 3e-06,
1588
+ "loss": 0.0658,
1589
+ "step": 158
1590
+ },
1591
+ {
1592
+ "completion_length": 239.14583587646484,
1593
+ "epoch": 0.1696,
1594
+ "grad_norm": 0.12511895596981049,
1595
+ "kl": 0.018006207887083292,
1596
+ "learning_rate": 3e-06,
1597
+ "loss": 0.0626,
1598
+ "reward": 1.2916667461395264,
1599
+ "rewards/boxed_and_answer_tags_format_reward": 0.6458333432674408,
1600
+ "rewards/correctness_reward_func_math": 0.6458333432674408,
1601
+ "step": 159
1602
+ },
1603
+ {
1604
+ "epoch": 0.17066666666666666,
1605
+ "grad_norm": 0.17034728825092316,
1606
+ "kl": 0.01847991719841957,
1607
+ "learning_rate": 3e-06,
1608
+ "loss": 0.063,
1609
+ "step": 160
1610
+ },
1611
+ {
1612
+ "completion_length": 228.05208587646484,
1613
+ "epoch": 0.17173333333333332,
1614
+ "grad_norm": 0.1118399053812027,
1615
+ "kl": 0.0186180230230093,
1616
+ "learning_rate": 3e-06,
1617
+ "loss": 0.112,
1618
+ "reward": 1.1718750596046448,
1619
+ "rewards/boxed_and_answer_tags_format_reward": 0.6927083432674408,
1620
+ "rewards/correctness_reward_func_math": 0.4791666567325592,
1621
+ "step": 161
1622
+ },
1623
+ {
1624
+ "epoch": 0.1728,
1625
+ "grad_norm": 0.11686160415410995,
1626
+ "kl": 0.01933614257723093,
1627
+ "learning_rate": 3e-06,
1628
+ "loss": 0.1113,
1629
+ "step": 162
1630
+ },
1631
+ {
1632
+ "completion_length": 245.4166717529297,
1633
+ "epoch": 0.17386666666666667,
1634
+ "grad_norm": 0.13094253838062286,
1635
+ "kl": 0.02521906979382038,
1636
+ "learning_rate": 3e-06,
1637
+ "loss": 0.0199,
1638
+ "reward": 1.2760416865348816,
1639
+ "rewards/boxed_and_answer_tags_format_reward": 0.6927083134651184,
1640
+ "rewards/correctness_reward_func_math": 0.5833333283662796,
1641
+ "step": 163
1642
+ },
1643
+ {
1644
+ "epoch": 0.17493333333333333,
1645
+ "grad_norm": 0.14768710732460022,
1646
+ "kl": 0.025642136111855507,
1647
+ "learning_rate": 3e-06,
1648
+ "loss": 0.0201,
1649
+ "step": 164
1650
+ },
1651
+ {
1652
+ "completion_length": 235.80209350585938,
1653
+ "epoch": 0.176,
1654
+ "grad_norm": 0.12041536718606949,
1655
+ "kl": 0.022253177128732204,
1656
+ "learning_rate": 3e-06,
1657
+ "loss": 0.0575,
1658
+ "reward": 1.1614583730697632,
1659
+ "rewards/boxed_and_answer_tags_format_reward": 0.5989583134651184,
1660
+ "rewards/correctness_reward_func_math": 0.5625,
1661
+ "step": 165
1662
+ },
1663
+ {
1664
+ "epoch": 0.17706666666666668,
1665
+ "grad_norm": 0.1091667115688324,
1666
+ "kl": 0.02235335111618042,
1667
+ "learning_rate": 3e-06,
1668
+ "loss": 0.0579,
1669
+ "step": 166
1670
+ },
1671
+ {
1672
+ "completion_length": 242.7916717529297,
1673
+ "epoch": 0.17813333333333334,
1674
+ "grad_norm": 0.12823987007141113,
1675
+ "kl": 0.030961502343416214,
1676
+ "learning_rate": 3e-06,
1677
+ "loss": 0.1094,
1678
+ "reward": 1.296875,
1679
+ "rewards/boxed_and_answer_tags_format_reward": 0.6510416567325592,
1680
+ "rewards/correctness_reward_func_math": 0.6458333283662796,
1681
+ "step": 167
1682
+ },
1683
+ {
1684
+ "epoch": 0.1792,
1685
+ "grad_norm": 0.1254572570323944,
1686
+ "kl": 0.03183833882212639,
1687
+ "learning_rate": 3e-06,
1688
+ "loss": 0.1095,
1689
+ "step": 168
1690
+ },
1691
+ {
1692
+ "completion_length": 246.5416717529297,
1693
+ "epoch": 0.18026666666666666,
1694
+ "grad_norm": 0.11450909078121185,
1695
+ "kl": 0.030905899591743946,
1696
+ "learning_rate": 3e-06,
1697
+ "loss": -0.0066,
1698
+ "reward": 0.8958333432674408,
1699
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1700
+ "rewards/correctness_reward_func_math": 0.2708333283662796,
1701
+ "step": 169
1702
+ },
1703
+ {
1704
+ "epoch": 0.18133333333333335,
1705
+ "grad_norm": 0.09335505217313766,
1706
+ "kl": 0.031846022233366966,
1707
+ "learning_rate": 3e-06,
1708
+ "loss": -0.0061,
1709
+ "step": 170
1710
+ },
1711
+ {
1712
+ "completion_length": 229.46875762939453,
1713
+ "epoch": 0.1824,
1714
+ "grad_norm": 0.07376878708600998,
1715
+ "kl": 0.024884301237761974,
1716
+ "learning_rate": 3e-06,
1717
+ "loss": 0.0223,
1718
+ "reward": 1.1614583432674408,
1719
+ "rewards/boxed_and_answer_tags_format_reward": 0.5572916567325592,
1720
+ "rewards/correctness_reward_func_math": 0.6041666716337204,
1721
+ "step": 171
1722
+ },
1723
+ {
1724
+ "epoch": 0.18346666666666667,
1725
+ "grad_norm": 0.07343802601099014,
1726
+ "kl": 0.026218552142381668,
1727
+ "learning_rate": 3e-06,
1728
+ "loss": 0.022,
1729
+ "step": 172
1730
+ },
1731
+ {
1732
+ "completion_length": 246.3854217529297,
1733
+ "epoch": 0.18453333333333333,
1734
+ "grad_norm": 0.1438089907169342,
1735
+ "kl": 0.04622086510062218,
1736
+ "learning_rate": 3e-06,
1737
+ "loss": 0.0581,
1738
+ "reward": 1.2864583730697632,
1739
+ "rewards/boxed_and_answer_tags_format_reward": 0.6614583134651184,
1740
+ "rewards/correctness_reward_func_math": 0.625,
1741
+ "step": 173
1742
+ },
1743
+ {
1744
+ "epoch": 0.1856,
1745
+ "grad_norm": 0.2885880768299103,
1746
+ "kl": 0.048377882689237595,
1747
+ "learning_rate": 3e-06,
1748
+ "loss": 0.0578,
1749
+ "step": 174
1750
+ },
1751
+ {
1752
+ "completion_length": 248.05208587646484,
1753
+ "epoch": 0.18666666666666668,
1754
+ "grad_norm": 0.10133392363786697,
1755
+ "kl": 0.04044355824589729,
1756
+ "learning_rate": 3e-06,
1757
+ "loss": -0.0204,
1758
+ "reward": 1.2552083730697632,
1759
+ "rewards/boxed_and_answer_tags_format_reward": 0.609375,
1760
+ "rewards/correctness_reward_func_math": 0.6458333432674408,
1761
+ "step": 175
1762
+ },
1763
+ {
1764
+ "epoch": 0.18773333333333334,
1765
+ "grad_norm": 0.09640152007341385,
1766
+ "kl": 0.041533155366778374,
1767
+ "learning_rate": 3e-06,
1768
+ "loss": -0.0207,
1769
+ "step": 176
1770
+ },
1771
+ {
1772
+ "completion_length": 249.21875762939453,
1773
+ "epoch": 0.1888,
1774
+ "grad_norm": 0.10415184497833252,
1775
+ "kl": 0.040618300437927246,
1776
+ "learning_rate": 3e-06,
1777
+ "loss": 0.1303,
1778
+ "reward": 1.0364583432674408,
1779
+ "rewards/boxed_and_answer_tags_format_reward": 0.703125,
1780
+ "rewards/correctness_reward_func_math": 0.3333333283662796,
1781
+ "step": 177
1782
+ },
1783
+ {
1784
+ "epoch": 0.18986666666666666,
1785
+ "grad_norm": 0.09741508960723877,
1786
+ "kl": 0.042102932929992676,
1787
+ "learning_rate": 3e-06,
1788
+ "loss": 0.1304,
1789
+ "step": 178
1790
+ },
1791
+ {
1792
+ "completion_length": 227.33333587646484,
1793
+ "epoch": 0.19093333333333334,
1794
+ "grad_norm": 0.12159749865531921,
1795
+ "kl": 0.029760053381323814,
1796
+ "learning_rate": 3e-06,
1797
+ "loss": -0.0087,
1798
+ "reward": 1.2708333432674408,
1799
+ "rewards/boxed_and_answer_tags_format_reward": 0.5833333432674408,
1800
+ "rewards/correctness_reward_func_math": 0.6875000149011612,
1801
+ "step": 179
1802
+ },
1803
+ {
1804
+ "epoch": 0.192,
1805
+ "grad_norm": 0.12757156789302826,
1806
+ "kl": 0.03069307003170252,
1807
+ "learning_rate": 3e-06,
1808
+ "loss": -0.0091,
1809
+ "step": 180
1810
+ },
1811
+ {
1812
+ "completion_length": 244.65625,
1813
+ "epoch": 0.19306666666666666,
1814
+ "grad_norm": 0.14061614871025085,
1815
+ "kl": 0.05795930698513985,
1816
+ "learning_rate": 3e-06,
1817
+ "loss": -0.0672,
1818
+ "reward": 1.3958333730697632,
1819
+ "rewards/boxed_and_answer_tags_format_reward": 0.625,
1820
+ "rewards/correctness_reward_func_math": 0.7708333134651184,
1821
+ "step": 181
1822
+ },
1823
+ {
1824
+ "epoch": 0.19413333333333332,
1825
+ "grad_norm": 0.1459376960992813,
1826
+ "kl": 0.05885138362646103,
1827
+ "learning_rate": 3e-06,
1828
+ "loss": -0.0674,
1829
+ "step": 182
1830
+ },
1831
+ {
1832
+ "completion_length": 249.86458587646484,
1833
+ "epoch": 0.1952,
1834
+ "grad_norm": 0.09802168607711792,
1835
+ "kl": 0.0534309446811676,
1836
+ "learning_rate": 3e-06,
1837
+ "loss": 0.0978,
1838
+ "reward": 1.4739583730697632,
1839
+ "rewards/boxed_and_answer_tags_format_reward": 0.640625,
1840
+ "rewards/correctness_reward_func_math": 0.8333333432674408,
1841
+ "step": 183
1842
+ },
1843
+ {
1844
+ "epoch": 0.19626666666666667,
1845
+ "grad_norm": 0.10266973078250885,
1846
+ "kl": 0.054405322298407555,
1847
+ "learning_rate": 3e-06,
1848
+ "loss": 0.0974,
1849
+ "step": 184
1850
+ },
1851
+ {
1852
+ "completion_length": 239.48959350585938,
1853
+ "epoch": 0.19733333333333333,
1854
+ "grad_norm": 0.10332746058702469,
1855
+ "kl": 0.051794858649373055,
1856
+ "learning_rate": 3e-06,
1857
+ "loss": 0.1001,
1858
+ "reward": 1.4583333730697632,
1859
+ "rewards/boxed_and_answer_tags_format_reward": 0.6666666865348816,
1860
+ "rewards/correctness_reward_func_math": 0.7916666865348816,
1861
+ "step": 185
1862
+ },
1863
+ {
1864
+ "epoch": 0.1984,
1865
+ "grad_norm": 0.10028999298810959,
1866
+ "kl": 0.05242122523486614,
1867
+ "learning_rate": 3e-06,
1868
+ "loss": 0.0998,
1869
+ "step": 186
1870
+ },
1871
+ {
1872
+ "completion_length": 243.37500762939453,
1873
+ "epoch": 0.19946666666666665,
1874
+ "grad_norm": 0.09041033685207367,
1875
+ "kl": 0.06607511267066002,
1876
+ "learning_rate": 3e-06,
1877
+ "loss": 0.137,
1878
+ "reward": 1.0781250298023224,
1879
+ "rewards/boxed_and_answer_tags_format_reward": 0.703125,
1880
+ "rewards/correctness_reward_func_math": 0.375,
1881
+ "step": 187
1882
+ },
1883
+ {
1884
+ "epoch": 0.20053333333333334,
1885
+ "grad_norm": 0.08903069794178009,
1886
+ "kl": 0.0671824999153614,
1887
+ "learning_rate": 3e-06,
1888
+ "loss": 0.137,
1889
+ "step": 188
1890
+ },
1891
+ {
1892
+ "completion_length": 252.37500762939453,
1893
+ "epoch": 0.2016,
1894
+ "grad_norm": 0.11631181091070175,
1895
+ "kl": 0.06189817003905773,
1896
+ "learning_rate": 3e-06,
1897
+ "loss": 0.0803,
1898
+ "reward": 1.296875,
1899
+ "rewards/boxed_and_answer_tags_format_reward": 0.6510416567325592,
1900
+ "rewards/correctness_reward_func_math": 0.6458333432674408,
1901
+ "step": 189
1902
+ },
1903
+ {
1904
+ "epoch": 0.20266666666666666,
1905
+ "grad_norm": 0.1270066797733307,
1906
+ "kl": 0.06297731585800648,
1907
+ "learning_rate": 3e-06,
1908
+ "loss": 0.0799,
1909
+ "step": 190
1910
+ },
1911
+ {
1912
+ "completion_length": 243.14583587646484,
1913
+ "epoch": 0.20373333333333332,
1914
+ "grad_norm": 0.11611822247505188,
1915
+ "kl": 0.05705785006284714,
1916
+ "learning_rate": 3e-06,
1917
+ "loss": 0.0642,
1918
+ "reward": 1.1562500596046448,
1919
+ "rewards/boxed_and_answer_tags_format_reward": 0.6145833432674408,
1920
+ "rewards/correctness_reward_func_math": 0.5416666716337204,
1921
+ "step": 191
1922
+ },
1923
+ {
1924
+ "epoch": 0.2048,
1925
+ "grad_norm": 0.14279727637767792,
1926
+ "kl": 0.058118581771850586,
1927
+ "learning_rate": 3e-06,
1928
+ "loss": 0.0639,
1929
+ "step": 192
1930
+ },
1931
+ {
1932
+ "completion_length": 242.70833587646484,
1933
+ "epoch": 0.20586666666666667,
1934
+ "grad_norm": 0.13600584864616394,
1935
+ "kl": 0.05646519921720028,
1936
+ "learning_rate": 3e-06,
1937
+ "loss": -0.0209,
1938
+ "reward": 1.3177083730697632,
1939
+ "rewards/boxed_and_answer_tags_format_reward": 0.671875,
1940
+ "rewards/correctness_reward_func_math": 0.6458333283662796,
1941
+ "step": 193
1942
+ },
1943
+ {
1944
+ "epoch": 0.20693333333333333,
1945
+ "grad_norm": 0.13068649172782898,
1946
+ "kl": 0.056824104860424995,
1947
+ "learning_rate": 3e-06,
1948
+ "loss": -0.022,
1949
+ "step": 194
1950
+ },
1951
+ {
1952
+ "completion_length": 226.39583587646484,
1953
+ "epoch": 0.208,
1954
+ "grad_norm": 0.11672388762235641,
1955
+ "kl": 0.05406441166996956,
1956
+ "learning_rate": 3e-06,
1957
+ "loss": 0.073,
1958
+ "reward": 1.3489583730697632,
1959
+ "rewards/boxed_and_answer_tags_format_reward": 0.6197916865348816,
1960
+ "rewards/correctness_reward_func_math": 0.7291666865348816,
1961
+ "step": 195
1962
+ },
1963
+ {
1964
+ "epoch": 0.20906666666666668,
1965
+ "grad_norm": 0.12466276437044144,
1966
+ "kl": 0.054257895797491074,
1967
+ "learning_rate": 3e-06,
1968
+ "loss": 0.0732,
1969
+ "step": 196
1970
+ },
1971
+ {
1972
+ "completion_length": 239.45834350585938,
1973
+ "epoch": 0.21013333333333334,
1974
+ "grad_norm": 0.14174474775791168,
1975
+ "kl": 0.057876838371157646,
1976
+ "learning_rate": 3e-06,
1977
+ "loss": 0.0156,
1978
+ "reward": 1.4218750596046448,
1979
+ "rewards/boxed_and_answer_tags_format_reward": 0.6927083432674408,
1980
+ "rewards/correctness_reward_func_math": 0.7291666865348816,
1981
+ "step": 197
1982
+ },
1983
+ {
1984
+ "epoch": 0.2112,
1985
+ "grad_norm": 0.15798237919807434,
1986
+ "kl": 0.059085674583911896,
1987
+ "learning_rate": 3e-06,
1988
+ "loss": 0.0151,
1989
+ "step": 198
1990
+ },
1991
+ {
1992
+ "completion_length": 237.45834350585938,
1993
+ "epoch": 0.21226666666666666,
1994
+ "grad_norm": 0.10461069643497467,
1995
+ "kl": 0.05880102701485157,
1996
+ "learning_rate": 3e-06,
1997
+ "loss": 0.0335,
1998
+ "reward": 1.322916716337204,
1999
+ "rewards/boxed_and_answer_tags_format_reward": 0.6979166567325592,
2000
+ "rewards/correctness_reward_func_math": 0.6249999776482582,
2001
+ "step": 199
2002
+ },
2003
+ {
2004
+ "epoch": 0.21333333333333335,
2005
+ "grad_norm": 0.10501859337091446,
2006
+ "kl": 0.05938952974975109,
2007
+ "learning_rate": 3e-06,
2008
+ "loss": 0.0337,
2009
+ "step": 200
2010
+ }
2011
+ ],
2012
+ "logging_steps": 1,
2013
+ "max_steps": 9370,
2014
+ "num_input_tokens_seen": 0,
2015
+ "num_train_epochs": 10,
2016
+ "save_steps": 100,
2017
+ "stateful_callbacks": {
2018
+ "TrainerControl": {
2019
+ "args": {
2020
+ "should_epoch_stop": false,
2021
+ "should_evaluate": false,
2022
+ "should_log": false,
2023
+ "should_save": true,
2024
+ "should_training_stop": false
2025
+ },
2026
+ "attributes": {}
2027
+ }
2028
+ },
2029
+ "total_flos": 0.0,
2030
+ "train_batch_size": 6,
2031
+ "trial_name": null,
2032
+ "trial_params": null
2033
+ }
checkpoint-200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:596e3dfc57483ca62815b3a75154fd6577935d09e47ba1327dd06cce26795a1b
3
+ size 9873
checkpoint-200/zero_to_fp32.py ADDED
@@ -0,0 +1,760 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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:
14
+ # python zero_to_fp32.py . output_dir/
15
+ # or
16
+ # python zero_to_fp32.py . output_dir/ --safe_serialization
17
+
18
+ import argparse
19
+ import torch
20
+ import glob
21
+ import math
22
+ import os
23
+ import re
24
+ import gc
25
+ import json
26
+ import numpy as np
27
+ from tqdm import tqdm
28
+ from collections import OrderedDict
29
+ from dataclasses import dataclass
30
+
31
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
32
+ # DeepSpeed data structures it has to be available in the current python environment.
33
+ from deepspeed.utils import logger
34
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
35
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
36
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
37
+
38
+
39
+ @dataclass
40
+ class zero_model_state:
41
+ buffers: dict()
42
+ param_shapes: dict()
43
+ shared_params: list
44
+ ds_version: int
45
+ frozen_param_shapes: dict()
46
+ frozen_param_fragments: dict()
47
+
48
+
49
+ debug = 0
50
+
51
+ # load to cpu
52
+ device = torch.device('cpu')
53
+
54
+
55
+ def atoi(text):
56
+ return int(text) if text.isdigit() else text
57
+
58
+
59
+ def natural_keys(text):
60
+ '''
61
+ alist.sort(key=natural_keys) sorts in human order
62
+ http://nedbatchelder.com/blog/200712/human_sorting.html
63
+ (See Toothy's implementation in the comments)
64
+ '''
65
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
66
+
67
+
68
+ def get_model_state_file(checkpoint_dir, zero_stage):
69
+ if not os.path.isdir(checkpoint_dir):
70
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
71
+
72
+ # there should be only one file
73
+ if zero_stage <= 2:
74
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
75
+ elif zero_stage == 3:
76
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
77
+
78
+ if not os.path.exists(file):
79
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
80
+
81
+ return file
82
+
83
+
84
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
85
+ # XXX: need to test that this simple glob rule works for multi-node setup too
86
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
87
+
88
+ if len(ckpt_files) == 0:
89
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
90
+
91
+ return ckpt_files
92
+
93
+
94
+ def get_optim_files(checkpoint_dir):
95
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
96
+
97
+
98
+ def get_model_state_files(checkpoint_dir):
99
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
100
+
101
+
102
+ def parse_model_states(files):
103
+ zero_model_states = []
104
+ for file in files:
105
+ state_dict = torch.load(file, map_location=device, weights_only=False)
106
+
107
+ if BUFFER_NAMES not in state_dict:
108
+ raise ValueError(f"{file} is not a model state checkpoint")
109
+ buffer_names = state_dict[BUFFER_NAMES]
110
+ if debug:
111
+ print("Found buffers:", buffer_names)
112
+
113
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
114
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
115
+ param_shapes = state_dict[PARAM_SHAPES]
116
+
117
+ # collect parameters that are included in param_shapes
118
+ param_names = []
119
+ for s in param_shapes:
120
+ for name in s.keys():
121
+ param_names.append(name)
122
+
123
+ # update with frozen parameters
124
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
125
+ if frozen_param_shapes is not None:
126
+ if debug:
127
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
128
+ param_names += list(frozen_param_shapes.keys())
129
+
130
+ # handle shared params
131
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
132
+
133
+ ds_version = state_dict.get(DS_VERSION, None)
134
+
135
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
136
+
137
+ z_model_state = zero_model_state(buffers=buffers,
138
+ param_shapes=param_shapes,
139
+ shared_params=shared_params,
140
+ ds_version=ds_version,
141
+ frozen_param_shapes=frozen_param_shapes,
142
+ frozen_param_fragments=frozen_param_fragments)
143
+ zero_model_states.append(z_model_state)
144
+
145
+ return zero_model_states
146
+
147
+
148
+ def parse_optim_states(files, ds_checkpoint_dir):
149
+ total_files = len(files)
150
+ state_dicts = []
151
+ for f in tqdm(files, desc='Loading checkpoint shards'):
152
+ state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
153
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
154
+ # and also handle the case where it was already removed by another helper script
155
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
156
+ state_dicts.append(state_dict)
157
+
158
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
159
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
160
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
161
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
162
+
163
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
164
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
165
+ # use the max of the partition_count to get the dp world_size.
166
+
167
+ if type(world_size) is list:
168
+ world_size = max(world_size)
169
+
170
+ if world_size != total_files:
171
+ raise ValueError(
172
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
173
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
174
+ )
175
+
176
+ # the groups are named differently in each stage
177
+ if zero_stage <= 2:
178
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
179
+ elif zero_stage == 3:
180
+ fp32_groups_key = FP32_FLAT_GROUPS
181
+ else:
182
+ raise ValueError(f"unknown zero stage {zero_stage}")
183
+
184
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
185
+ return zero_stage, world_size, fp32_flat_groups
186
+
187
+
188
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
189
+ """
190
+ Returns fp32 state_dict reconstructed from ds checkpoint
191
+
192
+ Args:
193
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
194
+
195
+ """
196
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
197
+
198
+ optim_files = get_optim_files(ds_checkpoint_dir)
199
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
200
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
201
+
202
+ model_files = get_model_state_files(ds_checkpoint_dir)
203
+
204
+ zero_model_states = parse_model_states(model_files)
205
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
206
+
207
+ if zero_stage <= 2:
208
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
209
+ exclude_frozen_parameters)
210
+ elif zero_stage == 3:
211
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
212
+ exclude_frozen_parameters)
213
+
214
+
215
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
216
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
217
+ return
218
+
219
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
220
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
221
+
222
+ if debug:
223
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
224
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
225
+
226
+ wanted_params = len(frozen_param_shapes)
227
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
228
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
229
+ print(f'Frozen params: Have {avail_numel} numels to process.')
230
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
231
+
232
+ total_params = 0
233
+ total_numel = 0
234
+ for name, shape in frozen_param_shapes.items():
235
+ total_params += 1
236
+ unpartitioned_numel = shape.numel()
237
+ total_numel += unpartitioned_numel
238
+
239
+ state_dict[name] = frozen_param_fragments[name]
240
+
241
+ if debug:
242
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
243
+
244
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
245
+
246
+
247
+ def _has_callable(obj, fn):
248
+ attr = getattr(obj, fn, None)
249
+ return callable(attr)
250
+
251
+
252
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
253
+ param_shapes = zero_model_states[0].param_shapes
254
+
255
+ # Reconstruction protocol:
256
+ #
257
+ # XXX: document this
258
+
259
+ if debug:
260
+ for i in range(world_size):
261
+ for j in range(len(fp32_flat_groups[0])):
262
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
263
+
264
+ # XXX: memory usage doubles here (zero2)
265
+ num_param_groups = len(fp32_flat_groups[0])
266
+ merged_single_partition_of_fp32_groups = []
267
+ for i in range(num_param_groups):
268
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
269
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
270
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
271
+ avail_numel = sum(
272
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
273
+
274
+ if debug:
275
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
276
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
277
+ # not asserting if there is a mismatch due to possible padding
278
+ print(f"Have {avail_numel} numels to process.")
279
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
280
+
281
+ # params
282
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
283
+ # out-of-core computing solution
284
+ total_numel = 0
285
+ total_params = 0
286
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
287
+ offset = 0
288
+ avail_numel = full_single_fp32_vector.numel()
289
+ for name, shape in shapes.items():
290
+
291
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
292
+ total_numel += unpartitioned_numel
293
+ total_params += 1
294
+
295
+ if debug:
296
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
297
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
298
+ offset += unpartitioned_numel
299
+
300
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
301
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
302
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
303
+ # live optimizer object, so we are checking that the numbers are within the right range
304
+ align_to = 2 * world_size
305
+
306
+ def zero2_align(x):
307
+ return align_to * math.ceil(x / align_to)
308
+
309
+ if debug:
310
+ print(f"original offset={offset}, avail_numel={avail_numel}")
311
+
312
+ offset = zero2_align(offset)
313
+ avail_numel = zero2_align(avail_numel)
314
+
315
+ if debug:
316
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
317
+
318
+ # Sanity check
319
+ if offset != avail_numel:
320
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
321
+
322
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
323
+
324
+
325
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
326
+ exclude_frozen_parameters):
327
+ state_dict = OrderedDict()
328
+
329
+ # buffers
330
+ buffers = zero_model_states[0].buffers
331
+ state_dict.update(buffers)
332
+ if debug:
333
+ print(f"added {len(buffers)} buffers")
334
+
335
+ if not exclude_frozen_parameters:
336
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
337
+
338
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
339
+
340
+ # recover shared parameters
341
+ for pair in zero_model_states[0].shared_params:
342
+ if pair[1] in state_dict:
343
+ state_dict[pair[0]] = state_dict[pair[1]]
344
+
345
+ return state_dict
346
+
347
+
348
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
349
+ remainder = unpartitioned_numel % world_size
350
+ padding_numel = (world_size - remainder) if remainder else 0
351
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
352
+ return partitioned_numel, padding_numel
353
+
354
+
355
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
356
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
357
+ return
358
+
359
+ if debug:
360
+ for i in range(world_size):
361
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
362
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
363
+
364
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
365
+ wanted_params = len(frozen_param_shapes)
366
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
367
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
368
+ print(f'Frozen params: Have {avail_numel} numels to process.')
369
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
370
+
371
+ total_params = 0
372
+ total_numel = 0
373
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
374
+ total_params += 1
375
+ unpartitioned_numel = shape.numel()
376
+ total_numel += unpartitioned_numel
377
+
378
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
379
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
380
+
381
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
382
+
383
+ if debug:
384
+ print(
385
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
386
+ )
387
+
388
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
389
+
390
+
391
+ class GatheredTensor:
392
+ """
393
+ A pseudo tensor that collects partitioned weights.
394
+ It is more memory efficient when there are multiple groups.
395
+ """
396
+
397
+ def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
398
+ self.flat_groups = flat_groups
399
+ self.flat_groups_offset = flat_groups_offset
400
+ self.offset = offset
401
+ self.partitioned_numel = partitioned_numel
402
+ self.shape = shape
403
+ self.dtype = self.flat_groups[0][0].dtype
404
+
405
+ def contiguous(self):
406
+ """
407
+ Merge partitioned weights from flat_groups into a single tensor.
408
+ """
409
+ end_idx = self.offset + self.partitioned_numel
410
+ world_size = len(self.flat_groups)
411
+ pad_flat_param_chunks = []
412
+
413
+ for rank_i in range(world_size):
414
+ # for each rank, we need to collect weights from related group/groups
415
+ flat_groups_at_rank_i = self.flat_groups[rank_i]
416
+ start_group_id = None
417
+ end_group_id = None
418
+ for group_id in range(len(self.flat_groups_offset)):
419
+ if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
420
+ start_group_id = group_id
421
+ if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
422
+ end_group_id = group_id
423
+ break
424
+ # collect weights from related group/groups
425
+ for group_id in range(start_group_id, end_group_id + 1):
426
+ flat_tensor = flat_groups_at_rank_i[group_id]
427
+ start_offset = self.offset - self.flat_groups_offset[group_id]
428
+ end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
429
+ pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
430
+
431
+ # collect weights from all ranks
432
+ pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
433
+ param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
434
+ return param
435
+
436
+
437
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
438
+ param_shapes = zero_model_states[0].param_shapes
439
+ avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
440
+
441
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
442
+ # param, re-consolidating each param, while dealing with padding if any
443
+
444
+ # merge list of dicts, preserving order
445
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
446
+
447
+ if debug:
448
+ for i in range(world_size):
449
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
450
+
451
+ wanted_params = len(param_shapes)
452
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
453
+ # not asserting if there is a mismatch due to possible padding
454
+ avail_numel = fp32_flat_groups[0].numel() * world_size
455
+ print(f"Trainable params: Have {avail_numel} numels to process.")
456
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
457
+
458
+ # params
459
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
460
+ # out-of-core computing solution
461
+ offset = 0
462
+ total_numel = 0
463
+ total_params = 0
464
+ flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
465
+ for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
466
+ unpartitioned_numel = shape.numel()
467
+ total_numel += unpartitioned_numel
468
+ total_params += 1
469
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
470
+
471
+ if debug:
472
+ print(
473
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
474
+ )
475
+
476
+ # memory efficient tensor
477
+ tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
478
+ state_dict[name] = tensor
479
+ offset += partitioned_numel
480
+
481
+ offset *= world_size
482
+
483
+ # Sanity check
484
+ if offset != avail_numel:
485
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
486
+
487
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
488
+
489
+
490
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
491
+ exclude_frozen_parameters):
492
+ state_dict = OrderedDict()
493
+
494
+ # buffers
495
+ buffers = zero_model_states[0].buffers
496
+ state_dict.update(buffers)
497
+ if debug:
498
+ print(f"added {len(buffers)} buffers")
499
+
500
+ if not exclude_frozen_parameters:
501
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
502
+
503
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
504
+
505
+ # recover shared parameters
506
+ for pair in zero_model_states[0].shared_params:
507
+ if pair[1] in state_dict:
508
+ state_dict[pair[0]] = state_dict[pair[1]]
509
+
510
+ return state_dict
511
+
512
+
513
+ def to_torch_tensor(state_dict, return_empty_tensor=False):
514
+ """
515
+ Convert state_dict of GatheredTensor to torch tensor
516
+ """
517
+ torch_state_dict = {}
518
+ converted_tensors = {}
519
+ for name, tensor in state_dict.items():
520
+ tensor_id = id(tensor)
521
+ if tensor_id in converted_tensors: # shared tensors
522
+ shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
523
+ torch_state_dict[name] = shared_tensor
524
+ else:
525
+ converted_tensors[tensor_id] = name
526
+ if return_empty_tensor:
527
+ torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
528
+ else:
529
+ torch_state_dict[name] = tensor.contiguous()
530
+ return torch_state_dict
531
+
532
+
533
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
534
+ tag=None,
535
+ exclude_frozen_parameters=False,
536
+ lazy_mode=False):
537
+ """
538
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
539
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
540
+ via a model hub.
541
+
542
+ Args:
543
+ - ``checkpoint_dir``: path to the desired checkpoint folder
544
+ - ``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``
545
+ - ``exclude_frozen_parameters``: exclude frozen parameters
546
+ - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
547
+ Convert the pesduo tensor to torch tensor by ``.contiguous()``
548
+
549
+ Returns:
550
+ - pytorch ``state_dict``
551
+
552
+ A typical usage might be ::
553
+
554
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
555
+ # do the training and checkpoint saving
556
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
557
+ model = model.cpu() # move to cpu
558
+ model.load_state_dict(state_dict)
559
+ # submit to model hub or save the model to share with others
560
+
561
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
562
+ application. i.e. you will need to re-initialize the deepspeed engine, since
563
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
564
+
565
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
566
+
567
+ Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
568
+ You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
569
+ the checkpoint. Or you can load state_dict in lazy mode ::
570
+
571
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
572
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
573
+ for name, lazy_tensor in state_dict.item():
574
+ tensor = lazy_tensor.contiguous() # to cpu
575
+ print(name, tensor)
576
+ # del tensor to release memory if it no longer in use
577
+ """
578
+ if tag is None:
579
+ latest_path = os.path.join(checkpoint_dir, 'latest')
580
+ if os.path.isfile(latest_path):
581
+ with open(latest_path, 'r') as fd:
582
+ tag = fd.read().strip()
583
+ else:
584
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
585
+
586
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
587
+
588
+ if not os.path.isdir(ds_checkpoint_dir):
589
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
590
+
591
+ state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
592
+ if lazy_mode:
593
+ return state_dict
594
+ else:
595
+ return to_torch_tensor(state_dict)
596
+
597
+
598
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
599
+ output_dir,
600
+ max_shard_size="5GB",
601
+ safe_serialization=False,
602
+ tag=None,
603
+ exclude_frozen_parameters=False):
604
+ """
605
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
606
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
607
+
608
+ Args:
609
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
610
+ - ``output_dir``: directory to the pytorch fp32 state_dict output files
611
+ - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
612
+ - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
613
+ - ``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``
614
+ - ``exclude_frozen_parameters``: exclude frozen parameters
615
+ """
616
+
617
+ # Dependency pre-check
618
+ if safe_serialization:
619
+ try:
620
+ from safetensors.torch import save_file
621
+ except ImportError:
622
+ print('If you want to use `safe_serialization`, please `pip install safetensors`')
623
+ raise
624
+ if max_shard_size is not None:
625
+ try:
626
+ from huggingface_hub import split_torch_state_dict_into_shards
627
+ except ImportError:
628
+ print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
629
+ raise
630
+
631
+ # Convert zero checkpoint to state_dict
632
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
633
+ tag,
634
+ exclude_frozen_parameters,
635
+ lazy_mode=True)
636
+
637
+ # Shard the model if it is too big.
638
+ weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
639
+ if max_shard_size is not None:
640
+ filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
641
+ # an memory-efficient approach for sharding
642
+ empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
643
+ state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
644
+ filename_pattern=filename_pattern,
645
+ max_shard_size=max_shard_size)
646
+ else:
647
+ from collections import namedtuple
648
+ StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
649
+ state_dict_split = StateDictSplit(is_sharded=False,
650
+ filename_to_tensors={weights_name: list(state_dict.keys())})
651
+
652
+ # Save the model by shard
653
+ os.makedirs(output_dir, exist_ok=True)
654
+ filename_to_tensors = state_dict_split.filename_to_tensors.items()
655
+ for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
656
+ shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
657
+ shard_state_dict = to_torch_tensor(shard_state_dict)
658
+ output_path = os.path.join(output_dir, shard_file)
659
+ if safe_serialization:
660
+ save_file(shard_state_dict, output_path, metadata={"format": "pt"})
661
+ else:
662
+ torch.save(shard_state_dict, output_path)
663
+ # release the memory of current shard
664
+ for tensor_name in list(shard_state_dict.keys()):
665
+ del state_dict[tensor_name]
666
+ del shard_state_dict[tensor_name]
667
+ del shard_state_dict
668
+ gc.collect()
669
+
670
+ # Save index if sharded
671
+ if state_dict_split.is_sharded:
672
+ index = {
673
+ "metadata": state_dict_split.metadata,
674
+ "weight_map": state_dict_split.tensor_to_filename,
675
+ }
676
+ save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
677
+ save_index_file = os.path.join(output_dir, save_index_file)
678
+ with open(save_index_file, "w", encoding="utf-8") as f:
679
+ content = json.dumps(index, indent=2, sort_keys=True) + "\n"
680
+ f.write(content)
681
+
682
+
683
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
684
+ """
685
+ 1. Put the provided model to cpu
686
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
687
+ 3. Load it into the provided model
688
+
689
+ Args:
690
+ - ``model``: the model object to update
691
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
692
+ - ``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``
693
+
694
+ Returns:
695
+ - ``model`: modified model
696
+
697
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
698
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
699
+ conveniently placed for you in the checkpoint folder.
700
+
701
+ A typical usage might be ::
702
+
703
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
704
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
705
+ # submit to model hub or save the model to share with others
706
+
707
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
708
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
709
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
710
+
711
+ """
712
+ logger.info(f"Extracting fp32 weights")
713
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
714
+
715
+ logger.info(f"Overwriting model with fp32 weights")
716
+ model = model.cpu()
717
+ model.load_state_dict(state_dict, strict=False)
718
+
719
+ return model
720
+
721
+
722
+ if __name__ == "__main__":
723
+ parser = argparse.ArgumentParser()
724
+ parser.add_argument("checkpoint_dir",
725
+ type=str,
726
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
727
+ parser.add_argument("output_dir",
728
+ type=str,
729
+ help="directory to the pytorch fp32 state_dict output files"
730
+ "(e.g. path/checkpoint-12-output/)")
731
+ parser.add_argument(
732
+ "--max_shard_size",
733
+ type=str,
734
+ default="5GB",
735
+ help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
736
+ "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
737
+ "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
738
+ "without CPU OOM issues.")
739
+ parser.add_argument(
740
+ "--safe_serialization",
741
+ default=False,
742
+ action='store_true',
743
+ help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
744
+ parser.add_argument("-t",
745
+ "--tag",
746
+ type=str,
747
+ default=None,
748
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
749
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
750
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
751
+ args = parser.parse_args()
752
+
753
+ debug = args.debug
754
+
755
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
756
+ args.output_dir,
757
+ max_shard_size=args.max_shard_size,
758
+ safe_serialization=args.safe_serialization,
759
+ tag=args.tag,
760
+ exclude_frozen_parameters=args.exclude_frozen_parameters)