Hugging Face's logo

cjfcsjt
/
125k_refsom

PEFT
Safetensors
Model card Files Community
cjfcsjt commited on
Commit
1c2cd29
•
1 Parent(s): b29f7a2

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +202 -0
  2. adapter_config.json +380 -0
  3. adapter_model.safetensors +3 -0
  4. checkpoint-1600/README.md +202 -0
  5. checkpoint-1600/adapter_config.json +380 -0
  6. checkpoint-1600/adapter_model.safetensors +3 -0
  7. checkpoint-1600/latest +1 -0
  8. checkpoint-1600/qwen.tiktoken +0 -0
  9. checkpoint-1600/rng_state_0.pth +3 -0
  10. checkpoint-1600/rng_state_1.pth +3 -0
  11. checkpoint-1600/rng_state_2.pth +3 -0
  12. checkpoint-1600/rng_state_3.pth +3 -0
  13. checkpoint-1600/scheduler.pt +3 -0
  14. checkpoint-1600/special_tokens_map.json +3 -0
  15. checkpoint-1600/tokenizer_config.json +14 -0
  16. checkpoint-1600/trainer_state.json +1153 -0
  17. checkpoint-1600/training_args.bin +3 -0
  18. checkpoint-1600/zero_to_fp32.py +587 -0
  19. checkpoint-3200/README.md +202 -0
  20. checkpoint-3200/adapter_config.json +380 -0
  21. checkpoint-3200/adapter_model.safetensors +3 -0
  22. checkpoint-3200/latest +1 -0
  23. checkpoint-3200/qwen.tiktoken +0 -0
  24. checkpoint-3200/rng_state_0.pth +3 -0
  25. checkpoint-3200/rng_state_1.pth +3 -0
  26. checkpoint-3200/rng_state_2.pth +3 -0
  27. checkpoint-3200/rng_state_3.pth +3 -0
  28. checkpoint-3200/scheduler.pt +3 -0
  29. checkpoint-3200/special_tokens_map.json +3 -0
  30. checkpoint-3200/tokenizer_config.json +14 -0
  31. checkpoint-3200/trainer_state.json +2273 -0
  32. checkpoint-3200/training_args.bin +3 -0
  33. checkpoint-3200/zero_to_fp32.py +587 -0
  34. checkpoint-4800/README.md +202 -0
  35. checkpoint-4800/adapter_config.json +380 -0
  36. checkpoint-4800/adapter_model.safetensors +3 -0
  37. checkpoint-4800/latest +1 -0
  38. checkpoint-4800/qwen.tiktoken +0 -0
  39. checkpoint-4800/rng_state_0.pth +3 -0
  40. checkpoint-4800/rng_state_1.pth +3 -0
  41. checkpoint-4800/rng_state_2.pth +3 -0
  42. checkpoint-4800/rng_state_3.pth +3 -0
  43. checkpoint-4800/scheduler.pt +3 -0
  44. checkpoint-4800/special_tokens_map.json +3 -0
  45. checkpoint-4800/tokenizer_config.json +14 -0
  46. checkpoint-4800/trainer_state.json +3393 -0
  47. checkpoint-4800/training_args.bin +3 -0
  48. checkpoint-4800/zero_to_fp32.py +587 -0
  49. checkpoint-6400/README.md +202 -0
  50. checkpoint-6400/adapter_config.json +380 -0
README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: Qwen/Qwen-VL-Chat
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 0.10.0
adapter_config.json ADDED
@@ -0,0 +1,380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 64,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.25.mlp.w2",
24
+ "transformer.visual.transformer.resblocks.39.mlp.c_proj",
25
+ "transformer.visual.transformer.resblocks.21.attn.in_proj",
26
+ "transformer.visual.transformer.resblocks.44.mlp.c_proj",
27
+ "transformer.h.15.mlp.w2",
28
+ "transformer.visual.transformer.resblocks.7.mlp.c_proj",
29
+ "transformer.h.3.mlp.w2",
30
+ "transformer.visual.transformer.resblocks.23.mlp.c_fc",
31
+ "transformer.visual.transformer.resblocks.10.mlp.c_fc",
32
+ "transformer.h.20.attn.c_attn",
33
+ "transformer.h.23.mlp.c_proj",
34
+ "transformer.visual.transformer.resblocks.34.mlp.c_fc",
35
+ "transformer.h.0.attn.c_proj",
36
+ "transformer.h.9.mlp.w1",
37
+ "transformer.visual.transformer.resblocks.12.attn.out_proj",
38
+ "transformer.visual.transformer.resblocks.22.mlp.c_fc",
39
+ "transformer.visual.transformer.resblocks.23.attn.out_proj",
40
+ "transformer.h.27.mlp.c_proj",
41
+ "transformer.visual.transformer.resblocks.1.mlp.c_proj",
42
+ "transformer.h.22.attn.c_proj",
43
+ "transformer.h.25.mlp.w1",
44
+ "transformer.visual.transformer.resblocks.0.attn.in_proj",
45
+ "transformer.visual.transformer.resblocks.2.attn.in_proj",
46
+ "transformer.h.1.attn.c_proj",
47
+ "transformer.visual.transformer.resblocks.45.mlp.c_proj",
48
+ "transformer.h.29.attn.c_attn",
49
+ "transformer.visual.transformer.resblocks.43.attn.in_proj",
50
+ "transformer.visual.transformer.resblocks.46.mlp.c_proj",
51
+ "transformer.h.28.mlp.w2",
52
+ "transformer.h.20.attn.c_proj",
53
+ "transformer.h.12.attn.c_proj",
54
+ "transformer.visual.transformer.resblocks.27.attn.out_proj",
55
+ "transformer.visual.transformer.resblocks.17.mlp.c_proj",
56
+ "transformer.visual.transformer.resblocks.33.attn.out_proj",
57
+ "transformer.h.4.attn.c_proj",
58
+ "transformer.visual.transformer.resblocks.29.attn.in_proj",
59
+ "transformer.visual.transformer.resblocks.2.attn.out_proj",
60
+ "transformer.h.4.mlp.w2",
61
+ "transformer.h.30.mlp.w1",
62
+ "transformer.h.31.attn.c_attn",
63
+ "transformer.h.10.attn.c_proj",
64
+ "transformer.h.18.mlp.w2",
65
+ "transformer.visual.transformer.resblocks.12.attn.in_proj",
66
+ "transformer.h.20.mlp.c_proj",
67
+ "transformer.visual.transformer.resblocks.36.attn.out_proj",
68
+ "transformer.visual.transformer.resblocks.31.attn.out_proj",
69
+ "transformer.h.1.attn.c_attn",
70
+ "transformer.visual.transformer.resblocks.17.attn.out_proj",
71
+ "transformer.visual.transformer.resblocks.9.mlp.c_fc",
72
+ "transformer.h.22.attn.c_attn",
73
+ "transformer.h.16.attn.c_attn",
74
+ "transformer.visual.transformer.resblocks.20.mlp.c_fc",
75
+ "transformer.visual.transformer.resblocks.31.attn.in_proj",
76
+ "transformer.h.29.mlp.w1",
77
+ "transformer.h.27.attn.c_attn",
78
+ "transformer.visual.transformer.resblocks.18.mlp.c_proj",
79
+ "transformer.visual.transformer.resblocks.8.mlp.c_proj",
80
+ "transformer.visual.transformer.resblocks.44.attn.out_proj",
81
+ "transformer.h.5.attn.c_proj",
82
+ "transformer.h.20.mlp.w1",
83
+ "transformer.h.0.mlp.c_proj",
84
+ "transformer.visual.transformer.resblocks.9.attn.out_proj",
85
+ "transformer.h.2.mlp.w1",
86
+ "transformer.visual.transformer.resblocks.3.mlp.c_proj",
87
+ "transformer.visual.transformer.resblocks.30.mlp.c_fc",
88
+ "transformer.visual.transformer.resblocks.36.mlp.c_proj",
89
+ "transformer.h.17.mlp.w2",
90
+ "transformer.h.19.mlp.w2",
91
+ "transformer.visual.transformer.resblocks.30.attn.out_proj",
92
+ "transformer.visual.transformer.resblocks.40.mlp.c_proj",
93
+ "transformer.visual.transformer.resblocks.1.mlp.c_fc",
94
+ "transformer.h.10.mlp.w1",
95
+ "transformer.h.6.attn.c_attn",
96
+ "transformer.visual.transformer.resblocks.14.mlp.c_fc",
97
+ "transformer.h.24.mlp.w2",
98
+ "transformer.visual.transformer.resblocks.28.attn.out_proj",
99
+ "transformer.visual.transformer.resblocks.27.attn.in_proj",
100
+ "transformer.h.10.attn.c_attn",
101
+ "transformer.h.17.mlp.c_proj",
102
+ "transformer.visual.transformer.resblocks.0.mlp.c_fc",
103
+ "transformer.visual.transformer.resblocks.35.attn.out_proj",
104
+ "transformer.h.6.attn.c_proj",
105
+ "transformer.h.6.mlp.w2",
106
+ "transformer.visual.transformer.resblocks.21.mlp.c_fc",
107
+ "transformer.h.22.mlp.c_proj",
108
+ "transformer.h.4.mlp.w1",
109
+ "transformer.h.14.mlp.w1",
110
+ "transformer.h.2.mlp.c_proj",
111
+ "transformer.visual.transformer.resblocks.31.mlp.c_fc",
112
+ "transformer.h.28.attn.c_proj",
113
+ "transformer.visual.transformer.resblocks.19.attn.in_proj",
114
+ "transformer.h.26.attn.c_proj",
115
+ "transformer.visual.transformer.resblocks.47.mlp.c_fc",
116
+ "transformer.h.3.attn.c_proj",
117
+ "transformer.visual.transformer.resblocks.33.attn.in_proj",
118
+ "transformer.visual.transformer.resblocks.45.attn.out_proj",
119
+ "transformer.visual.transformer.resblocks.5.attn.out_proj",
120
+ "transformer.h.7.attn.c_attn",
121
+ "transformer.h.1.mlp.w1",
122
+ "transformer.h.19.mlp.c_proj",
123
+ "transformer.visual.transformer.resblocks.39.attn.in_proj",
124
+ "transformer.visual.transformer.resblocks.7.attn.out_proj",
125
+ "transformer.h.8.mlp.w1",
126
+ "transformer.h.22.mlp.w1",
127
+ "transformer.h.7.attn.c_proj",
128
+ "transformer.visual.transformer.resblocks.0.mlp.c_proj",
129
+ "transformer.h.7.mlp.w2",
130
+ "transformer.h.25.attn.c_proj",
131
+ "transformer.visual.transformer.resblocks.17.attn.in_proj",
132
+ "transformer.visual.transformer.resblocks.24.attn.out_proj",
133
+ "transformer.h.24.mlp.c_proj",
134
+ "transformer.visual.transformer.resblocks.19.mlp.c_fc",
135
+ "transformer.h.21.mlp.w2",
136
+ "transformer.h.9.mlp.w2",
137
+ "transformer.h.27.attn.c_proj",
138
+ "transformer.h.2.attn.c_proj",
139
+ "transformer.h.5.mlp.c_proj",
140
+ "transformer.visual.transformer.resblocks.18.attn.in_proj",
141
+ "transformer.h.18.attn.c_attn",
142
+ "transformer.visual.transformer.resblocks.37.attn.in_proj",
143
+ "transformer.visual.transformer.resblocks.15.attn.out_proj",
144
+ "transformer.h.9.mlp.c_proj",
145
+ "transformer.visual.transformer.resblocks.37.mlp.c_proj",
146
+ "transformer.h.17.attn.c_proj",
147
+ "transformer.visual.transformer.resblocks.4.attn.out_proj",
148
+ "transformer.h.11.mlp.w2",
149
+ "transformer.h.3.mlp.w1",
150
+ "transformer.visual.transformer.resblocks.32.attn.out_proj",
151
+ "transformer.h.30.attn.c_attn",
152
+ "transformer.h.7.mlp.c_proj",
153
+ "transformer.visual.transformer.resblocks.38.mlp.c_proj",
154
+ "transformer.visual.transformer.resblocks.22.mlp.c_proj",
155
+ "transformer.h.23.mlp.w2",
156
+ "transformer.visual.transformer.resblocks.8.mlp.c_fc",
157
+ "transformer.h.24.attn.c_attn",
158
+ "transformer.visual.transformer.resblocks.21.attn.out_proj",
159
+ "transformer.visual.transformer.resblocks.16.mlp.c_fc",
160
+ "transformer.visual.transformer.resblocks.30.mlp.c_proj",
161
+ "transformer.visual.transformer.resblocks.29.attn.out_proj",
162
+ "transformer.visual.transformer.resblocks.19.attn.out_proj",
163
+ "transformer.visual.transformer.resblocks.1.attn.in_proj",
164
+ "transformer.h.27.mlp.w1",
165
+ "transformer.h.1.mlp.w2",
166
+ "transformer.h.17.attn.c_attn",
167
+ "transformer.visual.transformer.resblocks.3.mlp.c_fc",
168
+ "transformer.visual.transformer.resblocks.24.attn.in_proj",
169
+ "transformer.visual.transformer.resblocks.44.attn.in_proj",
170
+ "transformer.h.6.mlp.c_proj",
171
+ "transformer.visual.transformer.resblocks.3.attn.in_proj",
172
+ "transformer.visual.transformer.resblocks.19.mlp.c_proj",
173
+ "transformer.visual.transformer.resblocks.38.mlp.c_fc",
174
+ "transformer.h.14.mlp.w2",
175
+ "transformer.h.8.mlp.w2",
176
+ "transformer.h.19.mlp.w1",
177
+ "transformer.visual.transformer.resblocks.15.mlp.c_fc",
178
+ "transformer.h.19.attn.c_attn",
179
+ "transformer.h.13.attn.c_proj",
180
+ "transformer.visual.transformer.resblocks.26.mlp.c_fc",
181
+ "transformer.visual.transformer.resblocks.35.mlp.c_proj",
182
+ "transformer.h.15.mlp.w1",
183
+ "transformer.h.23.attn.c_attn",
184
+ "transformer.visual.transformer.resblocks.6.attn.out_proj",
185
+ "transformer.visual.transformer.resblocks.24.mlp.c_fc",
186
+ "transformer.visual.transformer.resblocks.18.attn.out_proj",
187
+ "transformer.visual.transformer.resblocks.20.attn.in_proj",
188
+ "transformer.visual.transformer.resblocks.31.mlp.c_proj",
189
+ "transformer.visual.transformer.resblocks.43.attn.out_proj",
190
+ "transformer.h.5.attn.c_attn",
191
+ "transformer.visual.transformer.resblocks.40.attn.out_proj",
192
+ "transformer.visual.transformer.resblocks.29.mlp.c_fc",
193
+ "transformer.h.12.mlp.w1",
194
+ "transformer.visual.transformer.resblocks.13.mlp.c_proj",
195
+ "transformer.visual.transformer.resblocks.43.mlp.c_proj",
196
+ "transformer.visual.transformer.resblocks.11.mlp.c_proj",
197
+ "transformer.visual.transformer.resblocks.37.attn.out_proj",
198
+ "transformer.visual.transformer.resblocks.40.mlp.c_fc",
199
+ "transformer.h.16.mlp.w2",
200
+ "transformer.h.9.attn.c_proj",
201
+ "transformer.visual.transformer.resblocks.26.attn.out_proj",
202
+ "transformer.visual.transformer.resblocks.4.attn.in_proj",
203
+ "transformer.visual.transformer.resblocks.10.attn.out_proj",
204
+ "transformer.h.13.mlp.c_proj",
205
+ "transformer.visual.transformer.resblocks.32.attn.in_proj",
206
+ "transformer.h.25.mlp.c_proj",
207
+ "transformer.h.24.mlp.w1",
208
+ "transformer.h.4.mlp.c_proj",
209
+ "transformer.h.21.attn.c_attn",
210
+ "transformer.visual.transformer.resblocks.37.mlp.c_fc",
211
+ "transformer.h.11.attn.c_proj",
212
+ "transformer.h.30.mlp.c_proj",
213
+ "transformer.visual.transformer.resblocks.23.mlp.c_proj",
214
+ "transformer.visual.transformer.resblocks.32.mlp.c_proj",
215
+ "transformer.h.21.attn.c_proj",
216
+ "transformer.h.3.attn.c_attn",
217
+ "transformer.visual.transformer.resblocks.5.mlp.c_proj",
218
+ "transformer.visual.transformer.resblocks.12.mlp.c_fc",
219
+ "transformer.visual.transformer.resblocks.45.attn.in_proj",
220
+ "transformer.h.4.attn.c_attn",
221
+ "transformer.visual.transformer.resblocks.33.mlp.c_proj",
222
+ "transformer.h.19.attn.c_proj",
223
+ "transformer.h.5.mlp.w1",
224
+ "transformer.visual.transformer.resblocks.14.attn.in_proj",
225
+ "transformer.h.9.attn.c_attn",
226
+ "transformer.visual.transformer.resblocks.45.mlp.c_fc",
227
+ "transformer.h.22.mlp.w2",
228
+ "transformer.visual.transformer.resblocks.41.attn.in_proj",
229
+ "transformer.h.3.mlp.c_proj",
230
+ "transformer.visual.transformer.resblocks.5.mlp.c_fc",
231
+ "transformer.h.26.attn.c_attn",
232
+ "transformer.h.30.attn.c_proj",
233
+ "transformer.visual.transformer.resblocks.10.attn.in_proj",
234
+ "transformer.visual.transformer.resblocks.24.mlp.c_proj",
235
+ "transformer.visual.transformer.resblocks.4.mlp.c_proj",
236
+ "transformer.visual.transformer.resblocks.44.mlp.c_fc",
237
+ "transformer.visual.transformer.resblocks.38.attn.in_proj",
238
+ "transformer.visual.transformer.resblocks.13.mlp.c_fc",
239
+ "transformer.visual.transformer.resblocks.8.attn.out_proj",
240
+ "transformer.visual.transformer.resblocks.23.attn.in_proj",
241
+ "transformer.visual.transformer.resblocks.41.mlp.c_proj",
242
+ "transformer.visual.transformer.resblocks.42.mlp.c_fc",
243
+ "transformer.visual.transformer.resblocks.39.mlp.c_fc",
244
+ "transformer.h.17.mlp.w1",
245
+ "transformer.visual.transformer.resblocks.42.attn.out_proj",
246
+ "transformer.visual.transformer.resblocks.8.attn.in_proj",
247
+ "transformer.h.25.attn.c_attn",
248
+ "transformer.visual.transformer.resblocks.1.attn.out_proj",
249
+ "transformer.h.6.mlp.w1",
250
+ "transformer.visual.transformer.resblocks.46.attn.out_proj",
251
+ "transformer.h.8.attn.c_attn",
252
+ "transformer.h.13.mlp.w2",
253
+ "transformer.visual.transformer.resblocks.21.mlp.c_proj",
254
+ "transformer.h.2.mlp.w2",
255
+ "transformer.visual.transformer.resblocks.30.attn.in_proj",
256
+ "transformer.h.7.mlp.w1",
257
+ "transformer.visual.transformer.resblocks.25.mlp.c_fc",
258
+ "transformer.h.11.mlp.c_proj",
259
+ "transformer.visual.transformer.resblocks.36.mlp.c_fc",
260
+ "transformer.visual.transformer.resblocks.42.mlp.c_proj",
261
+ "transformer.h.11.attn.c_attn",
262
+ "transformer.h.29.mlp.c_proj",
263
+ "transformer.visual.transformer.resblocks.36.attn.in_proj",
264
+ "transformer.visual.transformer.resblocks.40.attn.in_proj",
265
+ "transformer.visual.transformer.resblocks.27.mlp.c_fc",
266
+ "transformer.h.11.mlp.w1",
267
+ "transformer.visual.transformer.resblocks.39.attn.out_proj",
268
+ "transformer.visual.transformer.resblocks.10.mlp.c_proj",
269
+ "transformer.visual.transformer.resblocks.46.mlp.c_fc",
270
+ "transformer.h.20.mlp.w2",
271
+ "transformer.h.18.attn.c_proj",
272
+ "transformer.visual.transformer.resblocks.4.mlp.c_fc",
273
+ "transformer.visual.transformer.resblocks.17.mlp.c_fc",
274
+ "transformer.visual.transformer.resblocks.2.mlp.c_proj",
275
+ "transformer.visual.transformer.resblocks.6.attn.in_proj",
276
+ "transformer.visual.transformer.resblocks.11.mlp.c_fc",
277
+ "transformer.h.12.attn.c_attn",
278
+ "transformer.visual.transformer.resblocks.16.attn.in_proj",
279
+ "transformer.h.14.attn.c_attn",
280
+ "transformer.h.31.mlp.c_proj",
281
+ "transformer.visual.transformer.resblocks.16.attn.out_proj",
282
+ "transformer.h.0.mlp.w1",
283
+ "transformer.visual.transformer.resblocks.43.mlp.c_fc",
284
+ "transformer.visual.transformer.resblocks.5.attn.in_proj",
285
+ "transformer.h.13.attn.c_attn",
286
+ "transformer.visual.transformer.resblocks.41.mlp.c_fc",
287
+ "transformer.h.23.attn.c_proj",
288
+ "transformer.visual.transformer.resblocks.35.mlp.c_fc",
289
+ "transformer.visual.transformer.resblocks.12.mlp.c_proj",
290
+ "transformer.h.8.attn.c_proj",
291
+ "transformer.h.31.attn.c_proj",
292
+ "transformer.visual.transformer.resblocks.26.mlp.c_proj",
293
+ "transformer.h.21.mlp.w1",
294
+ "transformer.visual.transformer.resblocks.16.mlp.c_proj",
295
+ "transformer.h.31.mlp.w1",
296
+ "transformer.visual.transformer.resblocks.13.attn.in_proj",
297
+ "transformer.h.1.mlp.c_proj",
298
+ "transformer.h.12.mlp.c_proj",
299
+ "transformer.h.30.mlp.w2",
300
+ "transformer.h.14.attn.c_proj",
301
+ "transformer.visual.transformer.resblocks.15.mlp.c_proj",
302
+ "transformer.visual.transformer.resblocks.7.attn.in_proj",
303
+ "transformer.visual.transformer.resblocks.22.attn.out_proj",
304
+ "transformer.visual.transformer.resblocks.0.attn.out_proj",
305
+ "transformer.h.31.mlp.w2",
306
+ "transformer.visual.transformer.resblocks.27.mlp.c_proj",
307
+ "transformer.visual.transformer.resblocks.25.attn.out_proj",
308
+ "transformer.visual.transformer.resblocks.22.attn.in_proj",
309
+ "transformer.visual.transformer.resblocks.33.mlp.c_fc",
310
+ "transformer.h.13.mlp.w1",
311
+ "transformer.visual.transformer.resblocks.6.mlp.c_fc",
312
+ "transformer.visual.transformer.resblocks.11.attn.in_proj",
313
+ "transformer.visual.transformer.resblocks.9.attn.in_proj",
314
+ "transformer.visual.transformer.resblocks.20.mlp.c_proj",
315
+ "transformer.h.16.mlp.c_proj",
316
+ "transformer.visual.transformer.resblocks.15.attn.in_proj",
317
+ "transformer.h.26.mlp.w2",
318
+ "transformer.h.28.attn.c_attn",
319
+ "transformer.visual.transformer.resblocks.29.mlp.c_proj",
320
+ "transformer.h.23.mlp.w1",
321
+ "transformer.h.14.mlp.c_proj",
322
+ "transformer.visual.transformer.resblocks.25.mlp.c_proj",
323
+ "transformer.visual.transformer.resblocks.9.mlp.c_proj",
324
+ "transformer.visual.transformer.resblocks.34.attn.in_proj",
325
+ "transformer.h.5.mlp.w2",
326
+ "transformer.h.26.mlp.c_proj",
327
+ "transformer.h.24.attn.c_proj",
328
+ "transformer.visual.transformer.resblocks.47.attn.in_proj",
329
+ "transformer.visual.transformer.resblocks.7.mlp.c_fc",
330
+ "transformer.h.2.attn.c_attn",
331
+ "transformer.visual.conv1",
332
+ "transformer.visual.transformer.resblocks.28.attn.in_proj",
333
+ "transformer.visual.transformer.resblocks.25.attn.in_proj",
334
+ "transformer.visual.transformer.resblocks.47.attn.out_proj",
335
+ "transformer.visual.transformer.resblocks.18.mlp.c_fc",
336
+ "transformer.h.28.mlp.w1",
337
+ "transformer.h.10.mlp.w2",
338
+ "transformer.h.15.attn.c_proj",
339
+ "transformer.h.21.mlp.c_proj",
340
+ "transformer.h.29.mlp.w2",
341
+ "transformer.visual.transformer.resblocks.32.mlp.c_fc",
342
+ "transformer.h.18.mlp.w1",
343
+ "transformer.visual.transformer.resblocks.41.attn.out_proj",
344
+ "transformer.h.8.mlp.c_proj",
345
+ "transformer.visual.transformer.resblocks.47.mlp.c_proj",
346
+ "transformer.visual.transformer.resblocks.28.mlp.c_fc",
347
+ "transformer.visual.transformer.resblocks.34.attn.out_proj",
348
+ "transformer.h.29.attn.c_proj",
349
+ "transformer.visual.transformer.resblocks.38.attn.out_proj",
350
+ "transformer.visual.transformer.resblocks.14.attn.out_proj",
351
+ "transformer.h.10.mlp.c_proj",
352
+ "transformer.visual.transformer.resblocks.6.mlp.c_proj",
353
+ "transformer.h.0.mlp.w2",
354
+ "transformer.visual.transformer.resblocks.26.attn.in_proj",
355
+ "transformer.visual.transformer.resblocks.42.attn.in_proj",
356
+ "transformer.h.15.attn.c_attn",
357
+ "transformer.h.16.mlp.w1",
358
+ "transformer.visual.transformer.resblocks.3.attn.out_proj",
359
+ "transformer.visual.transformer.resblocks.11.attn.out_proj",
360
+ "transformer.visual.transformer.resblocks.28.mlp.c_proj",
361
+ "transformer.visual.transformer.resblocks.14.mlp.c_proj",
362
+ "transformer.h.0.attn.c_attn",
363
+ "transformer.h.18.mlp.c_proj",
364
+ "transformer.h.12.mlp.w2",
365
+ "transformer.visual.transformer.resblocks.35.attn.in_proj",
366
+ "transformer.visual.transformer.resblocks.2.mlp.c_fc",
367
+ "transformer.visual.transformer.resblocks.34.mlp.c_proj",
368
+ "transformer.h.26.mlp.w1",
369
+ "transformer.visual.transformer.resblocks.13.attn.out_proj",
370
+ "transformer.visual.transformer.resblocks.20.attn.out_proj",
371
+ "transformer.visual.transformer.resblocks.46.attn.in_proj",
372
+ "transformer.h.16.attn.c_proj",
373
+ "transformer.h.15.mlp.c_proj",
374
+ "transformer.h.27.mlp.w2",
375
+ "transformer.h.28.mlp.c_proj"
376
+ ],
377
+ "task_type": "CAUSAL_LM",
378
+ "use_dora": false,
379
+ "use_rslora": false
380
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0904fa3a6459dee0fd0fcaa7627c81e48b6d55695f2ad8c9895405b496ccb50a
3
+ size 469105640
checkpoint-1600/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: Qwen/Qwen-VL-Chat
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 0.10.0
checkpoint-1600/adapter_config.json ADDED
@@ -0,0 +1,380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 64,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.25.mlp.w2",
24
+ "transformer.visual.transformer.resblocks.39.mlp.c_proj",
25
+ "transformer.visual.transformer.resblocks.21.attn.in_proj",
26
+ "transformer.visual.transformer.resblocks.44.mlp.c_proj",
27
+ "transformer.h.15.mlp.w2",
28
+ "transformer.visual.transformer.resblocks.7.mlp.c_proj",
29
+ "transformer.h.3.mlp.w2",
30
+ "transformer.visual.transformer.resblocks.23.mlp.c_fc",
31
+ "transformer.visual.transformer.resblocks.10.mlp.c_fc",
32
+ "transformer.h.20.attn.c_attn",
33
+ "transformer.h.23.mlp.c_proj",
34
+ "transformer.visual.transformer.resblocks.34.mlp.c_fc",
35
+ "transformer.h.0.attn.c_proj",
36
+ "transformer.h.9.mlp.w1",
37
+ "transformer.visual.transformer.resblocks.12.attn.out_proj",
38
+ "transformer.visual.transformer.resblocks.22.mlp.c_fc",
39
+ "transformer.visual.transformer.resblocks.23.attn.out_proj",
40
+ "transformer.h.27.mlp.c_proj",
41
+ "transformer.visual.transformer.resblocks.1.mlp.c_proj",
42
+ "transformer.h.22.attn.c_proj",
43
+ "transformer.h.25.mlp.w1",
44
+ "transformer.visual.transformer.resblocks.0.attn.in_proj",
45
+ "transformer.visual.transformer.resblocks.2.attn.in_proj",
46
+ "transformer.h.1.attn.c_proj",
47
+ "transformer.visual.transformer.resblocks.45.mlp.c_proj",
48
+ "transformer.h.29.attn.c_attn",
49
+ "transformer.visual.transformer.resblocks.43.attn.in_proj",
50
+ "transformer.visual.transformer.resblocks.46.mlp.c_proj",
51
+ "transformer.h.28.mlp.w2",
52
+ "transformer.h.20.attn.c_proj",
53
+ "transformer.h.12.attn.c_proj",
54
+ "transformer.visual.transformer.resblocks.27.attn.out_proj",
55
+ "transformer.visual.transformer.resblocks.17.mlp.c_proj",
56
+ "transformer.visual.transformer.resblocks.33.attn.out_proj",
57
+ "transformer.h.4.attn.c_proj",
58
+ "transformer.visual.transformer.resblocks.29.attn.in_proj",
59
+ "transformer.visual.transformer.resblocks.2.attn.out_proj",
60
+ "transformer.h.4.mlp.w2",
61
+ "transformer.h.30.mlp.w1",
62
+ "transformer.h.31.attn.c_attn",
63
+ "transformer.h.10.attn.c_proj",
64
+ "transformer.h.18.mlp.w2",
65
+ "transformer.visual.transformer.resblocks.12.attn.in_proj",
66
+ "transformer.h.20.mlp.c_proj",
67
+ "transformer.visual.transformer.resblocks.36.attn.out_proj",
68
+ "transformer.visual.transformer.resblocks.31.attn.out_proj",
69
+ "transformer.h.1.attn.c_attn",
70
+ "transformer.visual.transformer.resblocks.17.attn.out_proj",
71
+ "transformer.visual.transformer.resblocks.9.mlp.c_fc",
72
+ "transformer.h.22.attn.c_attn",
73
+ "transformer.h.16.attn.c_attn",
74
+ "transformer.visual.transformer.resblocks.20.mlp.c_fc",
75
+ "transformer.visual.transformer.resblocks.31.attn.in_proj",
76
+ "transformer.h.29.mlp.w1",
77
+ "transformer.h.27.attn.c_attn",
78
+ "transformer.visual.transformer.resblocks.18.mlp.c_proj",
79
+ "transformer.visual.transformer.resblocks.8.mlp.c_proj",
80
+ "transformer.visual.transformer.resblocks.44.attn.out_proj",
81
+ "transformer.h.5.attn.c_proj",
82
+ "transformer.h.20.mlp.w1",
83
+ "transformer.h.0.mlp.c_proj",
84
+ "transformer.visual.transformer.resblocks.9.attn.out_proj",
85
+ "transformer.h.2.mlp.w1",
86
+ "transformer.visual.transformer.resblocks.3.mlp.c_proj",
87
+ "transformer.visual.transformer.resblocks.30.mlp.c_fc",
88
+ "transformer.visual.transformer.resblocks.36.mlp.c_proj",
89
+ "transformer.h.17.mlp.w2",
90
+ "transformer.h.19.mlp.w2",
91
+ "transformer.visual.transformer.resblocks.30.attn.out_proj",
92
+ "transformer.visual.transformer.resblocks.40.mlp.c_proj",
93
+ "transformer.visual.transformer.resblocks.1.mlp.c_fc",
94
+ "transformer.h.10.mlp.w1",
95
+ "transformer.h.6.attn.c_attn",
96
+ "transformer.visual.transformer.resblocks.14.mlp.c_fc",
97
+ "transformer.h.24.mlp.w2",
98
+ "transformer.visual.transformer.resblocks.28.attn.out_proj",
99
+ "transformer.visual.transformer.resblocks.27.attn.in_proj",
100
+ "transformer.h.10.attn.c_attn",
101
+ "transformer.h.17.mlp.c_proj",
102
+ "transformer.visual.transformer.resblocks.0.mlp.c_fc",
103
+ "transformer.visual.transformer.resblocks.35.attn.out_proj",
104
+ "transformer.h.6.attn.c_proj",
105
+ "transformer.h.6.mlp.w2",
106
+ "transformer.visual.transformer.resblocks.21.mlp.c_fc",
107
+ "transformer.h.22.mlp.c_proj",
108
+ "transformer.h.4.mlp.w1",
109
+ "transformer.h.14.mlp.w1",
110
+ "transformer.h.2.mlp.c_proj",
111
+ "transformer.visual.transformer.resblocks.31.mlp.c_fc",
112
+ "transformer.h.28.attn.c_proj",
113
+ "transformer.visual.transformer.resblocks.19.attn.in_proj",
114
+ "transformer.h.26.attn.c_proj",
115
+ "transformer.visual.transformer.resblocks.47.mlp.c_fc",
116
+ "transformer.h.3.attn.c_proj",
117
+ "transformer.visual.transformer.resblocks.33.attn.in_proj",
118
+ "transformer.visual.transformer.resblocks.45.attn.out_proj",
119
+ "transformer.visual.transformer.resblocks.5.attn.out_proj",
120
+ "transformer.h.7.attn.c_attn",
121
+ "transformer.h.1.mlp.w1",
122
+ "transformer.h.19.mlp.c_proj",
123
+ "transformer.visual.transformer.resblocks.39.attn.in_proj",
124
+ "transformer.visual.transformer.resblocks.7.attn.out_proj",
125
+ "transformer.h.8.mlp.w1",
126
+ "transformer.h.22.mlp.w1",
127
+ "transformer.h.7.attn.c_proj",
128
+ "transformer.visual.transformer.resblocks.0.mlp.c_proj",
129
+ "transformer.h.7.mlp.w2",
130
+ "transformer.h.25.attn.c_proj",
131
+ "transformer.visual.transformer.resblocks.17.attn.in_proj",
132
+ "transformer.visual.transformer.resblocks.24.attn.out_proj",
133
+ "transformer.h.24.mlp.c_proj",
134
+ "transformer.visual.transformer.resblocks.19.mlp.c_fc",
135
+ "transformer.h.21.mlp.w2",
136
+ "transformer.h.9.mlp.w2",
137
+ "transformer.h.27.attn.c_proj",
138
+ "transformer.h.2.attn.c_proj",
139
+ "transformer.h.5.mlp.c_proj",
140
+ "transformer.visual.transformer.resblocks.18.attn.in_proj",
141
+ "transformer.h.18.attn.c_attn",
142
+ "transformer.visual.transformer.resblocks.37.attn.in_proj",
143
+ "transformer.visual.transformer.resblocks.15.attn.out_proj",
144
+ "transformer.h.9.mlp.c_proj",
145
+ "transformer.visual.transformer.resblocks.37.mlp.c_proj",
146
+ "transformer.h.17.attn.c_proj",
147
+ "transformer.visual.transformer.resblocks.4.attn.out_proj",
148
+ "transformer.h.11.mlp.w2",
149
+ "transformer.h.3.mlp.w1",
150
+ "transformer.visual.transformer.resblocks.32.attn.out_proj",
151
+ "transformer.h.30.attn.c_attn",
152
+ "transformer.h.7.mlp.c_proj",
153
+ "transformer.visual.transformer.resblocks.38.mlp.c_proj",
154
+ "transformer.visual.transformer.resblocks.22.mlp.c_proj",
155
+ "transformer.h.23.mlp.w2",
156
+ "transformer.visual.transformer.resblocks.8.mlp.c_fc",
157
+ "transformer.h.24.attn.c_attn",
158
+ "transformer.visual.transformer.resblocks.21.attn.out_proj",
159
+ "transformer.visual.transformer.resblocks.16.mlp.c_fc",
160
+ "transformer.visual.transformer.resblocks.30.mlp.c_proj",
161
+ "transformer.visual.transformer.resblocks.29.attn.out_proj",
162
+ "transformer.visual.transformer.resblocks.19.attn.out_proj",
163
+ "transformer.visual.transformer.resblocks.1.attn.in_proj",
164
+ "transformer.h.27.mlp.w1",
165
+ "transformer.h.1.mlp.w2",
166
+ "transformer.h.17.attn.c_attn",
167
+ "transformer.visual.transformer.resblocks.3.mlp.c_fc",
168
+ "transformer.visual.transformer.resblocks.24.attn.in_proj",
169
+ "transformer.visual.transformer.resblocks.44.attn.in_proj",
170
+ "transformer.h.6.mlp.c_proj",
171
+ "transformer.visual.transformer.resblocks.3.attn.in_proj",
172
+ "transformer.visual.transformer.resblocks.19.mlp.c_proj",
173
+ "transformer.visual.transformer.resblocks.38.mlp.c_fc",
174
+ "transformer.h.14.mlp.w2",
175
+ "transformer.h.8.mlp.w2",
176
+ "transformer.h.19.mlp.w1",
177
+ "transformer.visual.transformer.resblocks.15.mlp.c_fc",
178
+ "transformer.h.19.attn.c_attn",
179
+ "transformer.h.13.attn.c_proj",
180
+ "transformer.visual.transformer.resblocks.26.mlp.c_fc",
181
+ "transformer.visual.transformer.resblocks.35.mlp.c_proj",
182
+ "transformer.h.15.mlp.w1",
183
+ "transformer.h.23.attn.c_attn",
184
+ "transformer.visual.transformer.resblocks.6.attn.out_proj",
185
+ "transformer.visual.transformer.resblocks.24.mlp.c_fc",
186
+ "transformer.visual.transformer.resblocks.18.attn.out_proj",
187
+ "transformer.visual.transformer.resblocks.20.attn.in_proj",
188
+ "transformer.visual.transformer.resblocks.31.mlp.c_proj",
189
+ "transformer.visual.transformer.resblocks.43.attn.out_proj",
190
+ "transformer.h.5.attn.c_attn",
191
+ "transformer.visual.transformer.resblocks.40.attn.out_proj",
192
+ "transformer.visual.transformer.resblocks.29.mlp.c_fc",
193
+ "transformer.h.12.mlp.w1",
194
+ "transformer.visual.transformer.resblocks.13.mlp.c_proj",
195
+ "transformer.visual.transformer.resblocks.43.mlp.c_proj",
196
+ "transformer.visual.transformer.resblocks.11.mlp.c_proj",
197
+ "transformer.visual.transformer.resblocks.37.attn.out_proj",
198
+ "transformer.visual.transformer.resblocks.40.mlp.c_fc",
199
+ "transformer.h.16.mlp.w2",
200
+ "transformer.h.9.attn.c_proj",
201
+ "transformer.visual.transformer.resblocks.26.attn.out_proj",
202
+ "transformer.visual.transformer.resblocks.4.attn.in_proj",
203
+ "transformer.visual.transformer.resblocks.10.attn.out_proj",
204
+ "transformer.h.13.mlp.c_proj",
205
+ "transformer.visual.transformer.resblocks.32.attn.in_proj",
206
+ "transformer.h.25.mlp.c_proj",
207
+ "transformer.h.24.mlp.w1",
208
+ "transformer.h.4.mlp.c_proj",
209
+ "transformer.h.21.attn.c_attn",
210
+ "transformer.visual.transformer.resblocks.37.mlp.c_fc",
211
+ "transformer.h.11.attn.c_proj",
212
+ "transformer.h.30.mlp.c_proj",
213
+ "transformer.visual.transformer.resblocks.23.mlp.c_proj",
214
+ "transformer.visual.transformer.resblocks.32.mlp.c_proj",
215
+ "transformer.h.21.attn.c_proj",
216
+ "transformer.h.3.attn.c_attn",
217
+ "transformer.visual.transformer.resblocks.5.mlp.c_proj",
218
+ "transformer.visual.transformer.resblocks.12.mlp.c_fc",
219
+ "transformer.visual.transformer.resblocks.45.attn.in_proj",
220
+ "transformer.h.4.attn.c_attn",
221
+ "transformer.visual.transformer.resblocks.33.mlp.c_proj",
222
+ "transformer.h.19.attn.c_proj",
223
+ "transformer.h.5.mlp.w1",
224
+ "transformer.visual.transformer.resblocks.14.attn.in_proj",
225
+ "transformer.h.9.attn.c_attn",
226
+ "transformer.visual.transformer.resblocks.45.mlp.c_fc",
227
+ "transformer.h.22.mlp.w2",
228
+ "transformer.visual.transformer.resblocks.41.attn.in_proj",
229
+ "transformer.h.3.mlp.c_proj",
230
+ "transformer.visual.transformer.resblocks.5.mlp.c_fc",
231
+ "transformer.h.26.attn.c_attn",
232
+ "transformer.h.30.attn.c_proj",
233
+ "transformer.visual.transformer.resblocks.10.attn.in_proj",
234
+ "transformer.visual.transformer.resblocks.24.mlp.c_proj",
235
+ "transformer.visual.transformer.resblocks.4.mlp.c_proj",
236
+ "transformer.visual.transformer.resblocks.44.mlp.c_fc",
237
+ "transformer.visual.transformer.resblocks.38.attn.in_proj",
238
+ "transformer.visual.transformer.resblocks.13.mlp.c_fc",
239
+ "transformer.visual.transformer.resblocks.8.attn.out_proj",
240
+ "transformer.visual.transformer.resblocks.23.attn.in_proj",
241
+ "transformer.visual.transformer.resblocks.41.mlp.c_proj",
242
+ "transformer.visual.transformer.resblocks.42.mlp.c_fc",
243
+ "transformer.visual.transformer.resblocks.39.mlp.c_fc",
244
+ "transformer.h.17.mlp.w1",
245
+ "transformer.visual.transformer.resblocks.42.attn.out_proj",
246
+ "transformer.visual.transformer.resblocks.8.attn.in_proj",
247
+ "transformer.h.25.attn.c_attn",
248
+ "transformer.visual.transformer.resblocks.1.attn.out_proj",
249
+ "transformer.h.6.mlp.w1",
250
+ "transformer.visual.transformer.resblocks.46.attn.out_proj",
251
+ "transformer.h.8.attn.c_attn",
252
+ "transformer.h.13.mlp.w2",
253
+ "transformer.visual.transformer.resblocks.21.mlp.c_proj",
254
+ "transformer.h.2.mlp.w2",
255
+ "transformer.visual.transformer.resblocks.30.attn.in_proj",
256
+ "transformer.h.7.mlp.w1",
257
+ "transformer.visual.transformer.resblocks.25.mlp.c_fc",
258
+ "transformer.h.11.mlp.c_proj",
259
+ "transformer.visual.transformer.resblocks.36.mlp.c_fc",
260
+ "transformer.visual.transformer.resblocks.42.mlp.c_proj",
261
+ "transformer.h.11.attn.c_attn",
262
+ "transformer.h.29.mlp.c_proj",
263
+ "transformer.visual.transformer.resblocks.36.attn.in_proj",
264
+ "transformer.visual.transformer.resblocks.40.attn.in_proj",
265
+ "transformer.visual.transformer.resblocks.27.mlp.c_fc",
266
+ "transformer.h.11.mlp.w1",
267
+ "transformer.visual.transformer.resblocks.39.attn.out_proj",
268
+ "transformer.visual.transformer.resblocks.10.mlp.c_proj",
269
+ "transformer.visual.transformer.resblocks.46.mlp.c_fc",
270
+ "transformer.h.20.mlp.w2",
271
+ "transformer.h.18.attn.c_proj",
272
+ "transformer.visual.transformer.resblocks.4.mlp.c_fc",
273
+ "transformer.visual.transformer.resblocks.17.mlp.c_fc",
274
+ "transformer.visual.transformer.resblocks.2.mlp.c_proj",
275
+ "transformer.visual.transformer.resblocks.6.attn.in_proj",
276
+ "transformer.visual.transformer.resblocks.11.mlp.c_fc",
277
+ "transformer.h.12.attn.c_attn",
278
+ "transformer.visual.transformer.resblocks.16.attn.in_proj",
279
+ "transformer.h.14.attn.c_attn",
280
+ "transformer.h.31.mlp.c_proj",
281
+ "transformer.visual.transformer.resblocks.16.attn.out_proj",
282
+ "transformer.h.0.mlp.w1",
283
+ "transformer.visual.transformer.resblocks.43.mlp.c_fc",
284
+ "transformer.visual.transformer.resblocks.5.attn.in_proj",
285
+ "transformer.h.13.attn.c_attn",
286
+ "transformer.visual.transformer.resblocks.41.mlp.c_fc",
287
+ "transformer.h.23.attn.c_proj",
288
+ "transformer.visual.transformer.resblocks.35.mlp.c_fc",
289
+ "transformer.visual.transformer.resblocks.12.mlp.c_proj",
290
+ "transformer.h.8.attn.c_proj",
291
+ "transformer.h.31.attn.c_proj",
292
+ "transformer.visual.transformer.resblocks.26.mlp.c_proj",
293
+ "transformer.h.21.mlp.w1",
294
+ "transformer.visual.transformer.resblocks.16.mlp.c_proj",
295
+ "transformer.h.31.mlp.w1",
296
+ "transformer.visual.transformer.resblocks.13.attn.in_proj",
297
+ "transformer.h.1.mlp.c_proj",
298
+ "transformer.h.12.mlp.c_proj",
299
+ "transformer.h.30.mlp.w2",
300
+ "transformer.h.14.attn.c_proj",
301
+ "transformer.visual.transformer.resblocks.15.mlp.c_proj",
302
+ "transformer.visual.transformer.resblocks.7.attn.in_proj",
303
+ "transformer.visual.transformer.resblocks.22.attn.out_proj",
304
+ "transformer.visual.transformer.resblocks.0.attn.out_proj",
305
+ "transformer.h.31.mlp.w2",
306
+ "transformer.visual.transformer.resblocks.27.mlp.c_proj",
307
+ "transformer.visual.transformer.resblocks.25.attn.out_proj",
308
+ "transformer.visual.transformer.resblocks.22.attn.in_proj",
309
+ "transformer.visual.transformer.resblocks.33.mlp.c_fc",
310
+ "transformer.h.13.mlp.w1",
311
+ "transformer.visual.transformer.resblocks.6.mlp.c_fc",
312
+ "transformer.visual.transformer.resblocks.11.attn.in_proj",
313
+ "transformer.visual.transformer.resblocks.9.attn.in_proj",
314
+ "transformer.visual.transformer.resblocks.20.mlp.c_proj",
315
+ "transformer.h.16.mlp.c_proj",
316
+ "transformer.visual.transformer.resblocks.15.attn.in_proj",
317
+ "transformer.h.26.mlp.w2",
318
+ "transformer.h.28.attn.c_attn",
319
+ "transformer.visual.transformer.resblocks.29.mlp.c_proj",
320
+ "transformer.h.23.mlp.w1",
321
+ "transformer.h.14.mlp.c_proj",
322
+ "transformer.visual.transformer.resblocks.25.mlp.c_proj",
323
+ "transformer.visual.transformer.resblocks.9.mlp.c_proj",
324
+ "transformer.visual.transformer.resblocks.34.attn.in_proj",
325
+ "transformer.h.5.mlp.w2",
326
+ "transformer.h.26.mlp.c_proj",
327
+ "transformer.h.24.attn.c_proj",
328
+ "transformer.visual.transformer.resblocks.47.attn.in_proj",
329
+ "transformer.visual.transformer.resblocks.7.mlp.c_fc",
330
+ "transformer.h.2.attn.c_attn",
331
+ "transformer.visual.conv1",
332
+ "transformer.visual.transformer.resblocks.28.attn.in_proj",
333
+ "transformer.visual.transformer.resblocks.25.attn.in_proj",
334
+ "transformer.visual.transformer.resblocks.47.attn.out_proj",
335
+ "transformer.visual.transformer.resblocks.18.mlp.c_fc",
336
+ "transformer.h.28.mlp.w1",
337
+ "transformer.h.10.mlp.w2",
338
+ "transformer.h.15.attn.c_proj",
339
+ "transformer.h.21.mlp.c_proj",
340
+ "transformer.h.29.mlp.w2",
341
+ "transformer.visual.transformer.resblocks.32.mlp.c_fc",
342
+ "transformer.h.18.mlp.w1",
343
+ "transformer.visual.transformer.resblocks.41.attn.out_proj",
344
+ "transformer.h.8.mlp.c_proj",
345
+ "transformer.visual.transformer.resblocks.47.mlp.c_proj",
346
+ "transformer.visual.transformer.resblocks.28.mlp.c_fc",
347
+ "transformer.visual.transformer.resblocks.34.attn.out_proj",
348
+ "transformer.h.29.attn.c_proj",
349
+ "transformer.visual.transformer.resblocks.38.attn.out_proj",
350
+ "transformer.visual.transformer.resblocks.14.attn.out_proj",
351
+ "transformer.h.10.mlp.c_proj",
352
+ "transformer.visual.transformer.resblocks.6.mlp.c_proj",
353
+ "transformer.h.0.mlp.w2",
354
+ "transformer.visual.transformer.resblocks.26.attn.in_proj",
355
+ "transformer.visual.transformer.resblocks.42.attn.in_proj",
356
+ "transformer.h.15.attn.c_attn",
357
+ "transformer.h.16.mlp.w1",
358
+ "transformer.visual.transformer.resblocks.3.attn.out_proj",
359
+ "transformer.visual.transformer.resblocks.11.attn.out_proj",
360
+ "transformer.visual.transformer.resblocks.28.mlp.c_proj",
361
+ "transformer.visual.transformer.resblocks.14.mlp.c_proj",
362
+ "transformer.h.0.attn.c_attn",
363
+ "transformer.h.18.mlp.c_proj",
364
+ "transformer.h.12.mlp.w2",
365
+ "transformer.visual.transformer.resblocks.35.attn.in_proj",
366
+ "transformer.visual.transformer.resblocks.2.mlp.c_fc",
367
+ "transformer.visual.transformer.resblocks.34.mlp.c_proj",
368
+ "transformer.h.26.mlp.w1",
369
+ "transformer.visual.transformer.resblocks.13.attn.out_proj",
370
+ "transformer.visual.transformer.resblocks.20.attn.out_proj",
371
+ "transformer.visual.transformer.resblocks.46.attn.in_proj",
372
+ "transformer.h.16.attn.c_proj",
373
+ "transformer.h.15.mlp.c_proj",
374
+ "transformer.h.27.mlp.w2",
375
+ "transformer.h.28.mlp.c_proj"
376
+ ],
377
+ "task_type": "CAUSAL_LM",
378
+ "use_dora": false,
379
+ "use_rslora": false
380
+ }
checkpoint-1600/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6ea791c2343ae36a7e27b3df47e3002430442cdffb8c437b4bed2c2b8865e36
3
+ size 469105640
checkpoint-1600/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step1600
checkpoint-1600/qwen.tiktoken ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1600/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0943be1d73f8cd864a8d86cf602e01a0ee9483c4bbd1287ca1aa6a70a07a7d78
3
+ size 14960
checkpoint-1600/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1814e986cf45968069746e25fb44856688645b26635b32a478bed3330978b28
3
+ size 14960
checkpoint-1600/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bfdaf471dc63d05f3a4e9aff471d87f770511c8529541d9f1d14a82ae9e16fd9
3
+ size 14960
checkpoint-1600/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8557ae681bc439c82fd4086d0ae897ebef2a086d2acbb923505fe7e63067cbd2
3
+ size 14960
checkpoint-1600/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b92b7d41119c1f55a67eb2522899e6a677806ed9022c1df91778b1c215870a1
3
+ size 1064
checkpoint-1600/special_tokens_map.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "pad_token": "<|endoftext|>"
3
+ }
checkpoint-1600/tokenizer_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {},
3
+ "auto_map": {
4
+ "AutoTokenizer": [
5
+ "Qwen/Qwen-VL-Chat--tokenization_qwen.QWenTokenizer",
6
+ null
7
+ ]
8
+ },
9
+ "clean_up_tokenization_spaces": true,
10
+ "model_max_length": 1280,
11
+ "pad_token": "<|endoftext|>",
12
+ "padding_side": "right",
13
+ "tokenizer_class": "QWenTokenizer"
14
+ }
checkpoint-1600/trainer_state.json ADDED
@@ -0,0 +1,1153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.21341158424755743,
5
+ "eval_steps": 500,
6
+ "global_step": 1600,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001333822401547234,
13
+ "grad_norm": 72.53171450734499,
14
+ "learning_rate": 4e-06,
15
+ "loss": 1.0439,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.002667644803094468,
20
+ "grad_norm": 7.72170877312484,
21
+ "learning_rate": 8e-06,
22
+ "loss": 1.0385,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.004001467204641702,
27
+ "grad_norm": 28.037294890647996,
28
+ "learning_rate": 1.2e-05,
29
+ "loss": 1.0328,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.005335289606188936,
34
+ "grad_norm": 10.980963099753343,
35
+ "learning_rate": 1.6e-05,
36
+ "loss": 0.9608,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.00666911200773617,
41
+ "grad_norm": 9.853684997615185,
42
+ "learning_rate": 1.9999999999999998e-05,
43
+ "loss": 0.8941,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.008002934409283404,
48
+ "grad_norm": 33.396994669126855,
49
+ "learning_rate": 2.4e-05,
50
+ "loss": 0.9174,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.009336756810830639,
55
+ "grad_norm": 5.621190624979587,
56
+ "learning_rate": 2.8e-05,
57
+ "loss": 0.8146,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.010670579212377872,
62
+ "grad_norm": 2.4646688601781843,
63
+ "learning_rate": 2.9999966406213696e-05,
64
+ "loss": 0.7774,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.012004401613925107,
69
+ "grad_norm": 2.1272107013027215,
70
+ "learning_rate": 2.9999697656826056e-05,
71
+ "loss": 0.7426,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01333822401547234,
76
+ "grad_norm": 2.0001927521328575,
77
+ "learning_rate": 2.9999160162865885e-05,
78
+ "loss": 0.7251,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.014672046417019574,
83
+ "grad_norm": 4.278494734910313,
84
+ "learning_rate": 2.9998353933963273e-05,
85
+ "loss": 0.7283,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.016005868818566808,
90
+ "grad_norm": 84.74138622339441,
91
+ "learning_rate": 2.999727898456315e-05,
92
+ "loss": 0.721,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.017339691220114042,
97
+ "grad_norm": 2.5596285292704533,
98
+ "learning_rate": 2.999593533392503e-05,
99
+ "loss": 0.7476,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.018673513621661277,
104
+ "grad_norm": 3.167163073798878,
105
+ "learning_rate": 2.9994323006122654e-05,
106
+ "loss": 0.7701,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.02000733602320851,
111
+ "grad_norm": 12.240132620099228,
112
+ "learning_rate": 2.9992442030043557e-05,
113
+ "loss": 0.7042,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.021341158424755743,
118
+ "grad_norm": 2.266079140233148,
119
+ "learning_rate": 2.9990292439388565e-05,
120
+ "loss": 0.6945,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.022674980826302978,
125
+ "grad_norm": 4.493189214061043,
126
+ "learning_rate": 2.9987874272671168e-05,
127
+ "loss": 0.7811,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.024008803227850213,
132
+ "grad_norm": 3.8513915234854132,
133
+ "learning_rate": 2.9985187573216855e-05,
134
+ "loss": 0.7229,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.025342625629397444,
139
+ "grad_norm": 2.1061250870336896,
140
+ "learning_rate": 2.998223238916232e-05,
141
+ "loss": 0.7253,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.02667644803094468,
146
+ "grad_norm": 2.8326879963181373,
147
+ "learning_rate": 2.9979008773454618e-05,
148
+ "loss": 0.7576,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.028010270432491914,
153
+ "grad_norm": 3.0294324007099402,
154
+ "learning_rate": 2.997551678385019e-05,
155
+ "loss": 0.7362,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02934409283403915,
160
+ "grad_norm": 1.5691302199679913,
161
+ "learning_rate": 2.997175648291384e-05,
162
+ "loss": 0.7212,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.03067791523558638,
167
+ "grad_norm": 4.379709379404857,
168
+ "learning_rate": 2.996772793801763e-05,
169
+ "loss": 0.7382,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.032011737637133615,
174
+ "grad_norm": 1.899348727085658,
175
+ "learning_rate": 2.996343122133965e-05,
176
+ "loss": 0.7056,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.033345560038680847,
181
+ "grad_norm": 21.481584335801955,
182
+ "learning_rate": 2.9958866409862745e-05,
183
+ "loss": 0.697,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.034679382440228085,
188
+ "grad_norm": 3.022034785435926,
189
+ "learning_rate": 2.9954033585373108e-05,
190
+ "loss": 0.69,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.036013204841775316,
195
+ "grad_norm": 2.869274575136443,
196
+ "learning_rate": 2.994893283445885e-05,
197
+ "loss": 0.7026,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.037347027243322554,
202
+ "grad_norm": 7.8785627268474,
203
+ "learning_rate": 2.9943564248508415e-05,
204
+ "loss": 0.7021,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.038680849644869786,
209
+ "grad_norm": 4.450751706186503,
210
+ "learning_rate": 2.9937927923708966e-05,
211
+ "loss": 0.6814,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.04001467204641702,
216
+ "grad_norm": 11.613541400971123,
217
+ "learning_rate": 2.993202396104465e-05,
218
+ "loss": 0.7043,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.041348494447964255,
223
+ "grad_norm": 2.0602161594051536,
224
+ "learning_rate": 2.9925852466294795e-05,
225
+ "loss": 0.7073,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.04268231684951149,
230
+ "grad_norm": 19.383133688390824,
231
+ "learning_rate": 2.9919413550032014e-05,
232
+ "loss": 0.6965,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.04401613925105872,
237
+ "grad_norm": 7.9603948136759906,
238
+ "learning_rate": 2.991270732762022e-05,
239
+ "loss": 0.7029,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.045349961652605957,
244
+ "grad_norm": 2.810319267000959,
245
+ "learning_rate": 2.990573391921255e-05,
246
+ "loss": 0.7412,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.04668378405415319,
251
+ "grad_norm": 6.082487526538632,
252
+ "learning_rate": 2.989849344974924e-05,
253
+ "loss": 0.6913,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.048017606455700426,
258
+ "grad_norm": 1.3651265332328906,
259
+ "learning_rate": 2.9890986048955368e-05,
260
+ "loss": 0.6802,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.04935142885724766,
265
+ "grad_norm": 2.7766095141894023,
266
+ "learning_rate": 2.9883211851338516e-05,
267
+ "loss": 0.7254,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.05068525125879489,
272
+ "grad_norm": 2.2284551774435033,
273
+ "learning_rate": 2.9875170996186392e-05,
274
+ "loss": 0.7026,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.05201907366034213,
279
+ "grad_norm": 2.1609680967998957,
280
+ "learning_rate": 2.986686362756431e-05,
281
+ "loss": 0.7415,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.05335289606188936,
286
+ "grad_norm": 2.071743222471935,
287
+ "learning_rate": 2.9858289894312617e-05,
288
+ "loss": 0.7081,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.0546867184634366,
293
+ "grad_norm": 7.349693198515479,
294
+ "learning_rate": 2.9849449950044036e-05,
295
+ "loss": 0.7212,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.05602054086498383,
300
+ "grad_norm": 12.990346405367537,
301
+ "learning_rate": 2.984034395314088e-05,
302
+ "loss": 0.7187,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.05735436326653106,
307
+ "grad_norm": 2.4106321189642723,
308
+ "learning_rate": 2.983097206675227e-05,
309
+ "loss": 0.6881,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0586881856680783,
314
+ "grad_norm": 3.1647778114563163,
315
+ "learning_rate": 2.9821334458791156e-05,
316
+ "loss": 0.7529,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.06002200806962553,
321
+ "grad_norm": 2.7769936556882713,
322
+ "learning_rate": 2.9811431301931344e-05,
323
+ "loss": 0.7073,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.06135583047117276,
328
+ "grad_norm": 2.834916196253906,
329
+ "learning_rate": 2.9801262773604377e-05,
330
+ "loss": 0.6649,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.06268965287271999,
335
+ "grad_norm": 2.431783869557187,
336
+ "learning_rate": 2.9790829055996398e-05,
337
+ "loss": 0.6889,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.06402347527426723,
342
+ "grad_norm": 3.7325728788625825,
343
+ "learning_rate": 2.978013033604483e-05,
344
+ "loss": 0.6771,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.06535729767581447,
349
+ "grad_norm": 10.194734403509758,
350
+ "learning_rate": 2.976916680543506e-05,
351
+ "loss": 0.6917,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.06669112007736169,
356
+ "grad_norm": 6.4699962355353655,
357
+ "learning_rate": 2.975793866059701e-05,
358
+ "loss": 0.6656,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.06802494247890893,
363
+ "grad_norm": 7.664735661456941,
364
+ "learning_rate": 2.9746446102701606e-05,
365
+ "loss": 0.7268,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.06935876488045617,
370
+ "grad_norm": 2.926294421836929,
371
+ "learning_rate": 2.9734689337657157e-05,
372
+ "loss": 0.7498,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.07069258728200341,
377
+ "grad_norm": 8.95556138792813,
378
+ "learning_rate": 2.9722668576105703e-05,
379
+ "loss": 0.7237,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.07202640968355063,
384
+ "grad_norm": 2.725769364825004,
385
+ "learning_rate": 2.971038403341921e-05,
386
+ "loss": 0.701,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.07336023208509787,
391
+ "grad_norm": 2.458130413236654,
392
+ "learning_rate": 2.9697835929695727e-05,
393
+ "loss": 0.6775,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.07469405448664511,
398
+ "grad_norm": 1.908510150241819,
399
+ "learning_rate": 2.968502448975544e-05,
400
+ "loss": 0.7042,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.07602787688819233,
405
+ "grad_norm": 26.340984665432142,
406
+ "learning_rate": 2.967194994313663e-05,
407
+ "loss": 0.7234,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.07736169928973957,
412
+ "grad_norm": 5.677162249188698,
413
+ "learning_rate": 2.9658612524091594e-05,
414
+ "loss": 0.7159,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.07869552169128681,
419
+ "grad_norm": 1.7975075976512287,
420
+ "learning_rate": 2.9645012471582406e-05,
421
+ "loss": 0.7268,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.08002934409283403,
426
+ "grad_norm": 3.031655038731278,
427
+ "learning_rate": 2.9631150029276662e-05,
428
+ "loss": 0.6904,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.08136316649438127,
433
+ "grad_norm": 13.019769162242698,
434
+ "learning_rate": 2.9617025445543114e-05,
435
+ "loss": 0.683,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.08269698889592851,
440
+ "grad_norm": 5.982097939523946,
441
+ "learning_rate": 2.9602638973447218e-05,
442
+ "loss": 0.7511,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.08403081129747574,
447
+ "grad_norm": 2.7343884277014054,
448
+ "learning_rate": 2.9587990870746574e-05,
449
+ "loss": 0.7114,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.08536463369902297,
454
+ "grad_norm": 5.586409371459215,
455
+ "learning_rate": 2.9573081399886356e-05,
456
+ "loss": 0.6907,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.08669845610057021,
461
+ "grad_norm": 9.87255558712134,
462
+ "learning_rate": 2.9557910827994568e-05,
463
+ "loss": 0.7319,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.08803227850211744,
468
+ "grad_norm": 3.9867937712251664,
469
+ "learning_rate": 2.9542479426877283e-05,
470
+ "loss": 0.6759,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.08936610090366467,
475
+ "grad_norm": 3.111751395898768,
476
+ "learning_rate": 2.9526787473013753e-05,
477
+ "loss": 0.7086,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.09069992330521191,
482
+ "grad_norm": 3.0578210081080535,
483
+ "learning_rate": 2.9510835247551485e-05,
484
+ "loss": 0.7209,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.09203374570675915,
489
+ "grad_norm": 16.575220829800145,
490
+ "learning_rate": 2.949462303630116e-05,
491
+ "loss": 0.7125,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.09336756810830638,
496
+ "grad_norm": 1.8236956939037314,
497
+ "learning_rate": 2.9478151129731567e-05,
498
+ "loss": 0.6591,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.09470139050985361,
503
+ "grad_norm": 1.9746052093914703,
504
+ "learning_rate": 2.9461419822964348e-05,
505
+ "loss": 0.7048,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.09603521291140085,
510
+ "grad_norm": 5.9710918845264,
511
+ "learning_rate": 2.9444429415768726e-05,
512
+ "loss": 0.725,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.09736903531294808,
517
+ "grad_norm": 9.961068856364284,
518
+ "learning_rate": 2.942718021255617e-05,
519
+ "loss": 0.7155,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.09870285771449532,
524
+ "grad_norm": 9.195914431100427,
525
+ "learning_rate": 2.940967252237488e-05,
526
+ "loss": 0.7171,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.10003668011604255,
531
+ "grad_norm": 5.972259927477824,
532
+ "learning_rate": 2.9391906658904296e-05,
533
+ "loss": 0.7088,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.10137050251758978,
538
+ "grad_norm": 3.0581670197906226,
539
+ "learning_rate": 2.937388294044946e-05,
540
+ "loss": 0.6645,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.10270432491913702,
545
+ "grad_norm": 3.2614276687575945,
546
+ "learning_rate": 2.9355601689935315e-05,
547
+ "loss": 0.6997,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.10403814732068425,
552
+ "grad_norm": 3.212661762840432,
553
+ "learning_rate": 2.933706323490092e-05,
554
+ "loss": 0.7005,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.10537196972223148,
559
+ "grad_norm": 10.732038885056795,
560
+ "learning_rate": 2.9318267907493583e-05,
561
+ "loss": 0.6588,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.10670579212377872,
566
+ "grad_norm": 2.495380329507228,
567
+ "learning_rate": 2.9299216044462903e-05,
568
+ "loss": 0.6965,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.10803961452532596,
573
+ "grad_norm": 4.60295956813534,
574
+ "learning_rate": 2.927990798715475e-05,
575
+ "loss": 0.6959,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.1093734369268732,
580
+ "grad_norm": 5.95010483593304,
581
+ "learning_rate": 2.926034408150513e-05,
582
+ "loss": 0.7083,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.11070725932842042,
587
+ "grad_norm": 10.48663715000287,
588
+ "learning_rate": 2.9240524678034016e-05,
589
+ "loss": 0.6977,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.11204108172996766,
594
+ "grad_norm": 3.159424497856185,
595
+ "learning_rate": 2.9220450131839037e-05,
596
+ "loss": 0.7129,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.1133749041315149,
601
+ "grad_norm": 3.984530217756339,
602
+ "learning_rate": 2.920012080258912e-05,
603
+ "loss": 0.6618,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.11470872653306212,
608
+ "grad_norm": 4.212436954015185,
609
+ "learning_rate": 2.9179537054518085e-05,
610
+ "loss": 0.6958,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.11604254893460936,
615
+ "grad_norm": 2.5397356754848657,
616
+ "learning_rate": 2.9158699256418056e-05,
617
+ "loss": 0.6593,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.1173763713361566,
622
+ "grad_norm": 7.8192219616904115,
623
+ "learning_rate": 2.9137607781632913e-05,
624
+ "loss": 0.6932,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.11871019373770382,
629
+ "grad_norm": 3.890974772100691,
630
+ "learning_rate": 2.911626300805155e-05,
631
+ "loss": 0.674,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.12004401613925106,
636
+ "grad_norm": 2.241784461612369,
637
+ "learning_rate": 2.9094665318101155e-05,
638
+ "loss": 0.6808,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.1213778385407983,
643
+ "grad_norm": 2.8295407624221456,
644
+ "learning_rate": 2.9072815098740326e-05,
645
+ "loss": 0.7074,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.12271166094234552,
650
+ "grad_norm": 2.8191065019379455,
651
+ "learning_rate": 2.9050712741452136e-05,
652
+ "loss": 0.7218,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.12404548334389276,
657
+ "grad_norm": 2.1348105213857362,
658
+ "learning_rate": 2.902835864223715e-05,
659
+ "loss": 0.6901,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.12537930574543998,
664
+ "grad_norm": 2.547712785034466,
665
+ "learning_rate": 2.9005753201606287e-05,
666
+ "loss": 0.7024,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.12671312814698724,
671
+ "grad_norm": 6.0714243848770195,
672
+ "learning_rate": 2.8982896824573678e-05,
673
+ "loss": 0.665,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.12804695054853446,
678
+ "grad_norm": 2.928297684665459,
679
+ "learning_rate": 2.8959789920649394e-05,
680
+ "loss": 0.7178,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.12938077295008168,
685
+ "grad_norm": 9.456759373673547,
686
+ "learning_rate": 2.893643290383212e-05,
687
+ "loss": 0.6759,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.13071459535162894,
692
+ "grad_norm": 3.9700721280634648,
693
+ "learning_rate": 2.891282619260172e-05,
694
+ "loss": 0.6701,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.13204841775317616,
699
+ "grad_norm": 1.9216628539066036,
700
+ "learning_rate": 2.8888970209911754e-05,
701
+ "loss": 0.6725,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.13338224015472339,
706
+ "grad_norm": 6.51104150647371,
707
+ "learning_rate": 2.8864865383181893e-05,
708
+ "loss": 0.6811,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.13471606255627064,
713
+ "grad_norm": 7.1765693144596705,
714
+ "learning_rate": 2.8840512144290273e-05,
715
+ "loss": 0.7085,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.13604988495781786,
720
+ "grad_norm": 4.136786758396937,
721
+ "learning_rate": 2.8815910929565734e-05,
722
+ "loss": 0.7128,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.1373837073593651,
727
+ "grad_norm": 3.407893493053672,
728
+ "learning_rate": 2.879106217978002e-05,
729
+ "loss": 0.6782,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.13871752976091234,
734
+ "grad_norm": 2.5759617833329544,
735
+ "learning_rate": 2.8765966340139892e-05,
736
+ "loss": 0.7387,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.14005135216245956,
741
+ "grad_norm": 8.496713788438253,
742
+ "learning_rate": 2.8740623860279116e-05,
743
+ "loss": 0.6988,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.14138517456400682,
748
+ "grad_norm": 1.438197837288145,
749
+ "learning_rate": 2.871503519425044e-05,
750
+ "loss": 0.6797,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.14271899696555404,
755
+ "grad_norm": 3.305947931404651,
756
+ "learning_rate": 2.8689200800517448e-05,
757
+ "loss": 0.727,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.14405281936710126,
762
+ "grad_norm": 4.384984204254061,
763
+ "learning_rate": 2.866312114194634e-05,
764
+ "loss": 0.7073,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.14538664176864852,
769
+ "grad_norm": 3.0424755710445455,
770
+ "learning_rate": 2.8636796685797657e-05,
771
+ "loss": 0.6712,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.14672046417019574,
776
+ "grad_norm": 3.7562374375017415,
777
+ "learning_rate": 2.8610227903717876e-05,
778
+ "loss": 0.6811,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.14805428657174297,
783
+ "grad_norm": 18.148320631618972,
784
+ "learning_rate": 2.8583415271730994e-05,
785
+ "loss": 0.669,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.14938810897329022,
790
+ "grad_norm": 10.095266883651389,
791
+ "learning_rate": 2.855635927022998e-05,
792
+ "loss": 0.7034,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.15072193137483744,
797
+ "grad_norm": 6.298167661594497,
798
+ "learning_rate": 2.8529060383968175e-05,
799
+ "loss": 0.6736,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.15205575377638467,
804
+ "grad_norm": 1.7112633381805116,
805
+ "learning_rate": 2.850151910205061e-05,
806
+ "loss": 0.7072,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.15338957617793192,
811
+ "grad_norm": 4.478017946548926,
812
+ "learning_rate": 2.847373591792523e-05,
813
+ "loss": 0.679,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.15472339857947914,
818
+ "grad_norm": 6.208223721690807,
819
+ "learning_rate": 2.844571132937407e-05,
820
+ "loss": 0.688,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.15605722098102637,
825
+ "grad_norm": 3.5734211349612974,
826
+ "learning_rate": 2.841744583850431e-05,
827
+ "loss": 0.7054,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.15739104338257362,
832
+ "grad_norm": 2.9418941665585874,
833
+ "learning_rate": 2.838893995173932e-05,
834
+ "loss": 0.7051,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.15872486578412084,
839
+ "grad_norm": 2.8796543096657055,
840
+ "learning_rate": 2.836019417980955e-05,
841
+ "loss": 0.6856,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.16005868818566807,
846
+ "grad_norm": 1.3218998593276705,
847
+ "learning_rate": 2.8331209037743387e-05,
848
+ "loss": 0.69,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.16139251058721532,
853
+ "grad_norm": 1.566163155963456,
854
+ "learning_rate": 2.8301985044857947e-05,
855
+ "loss": 0.6384,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.16272633298876255,
860
+ "grad_norm": 6.731577281026319,
861
+ "learning_rate": 2.8272522724749743e-05,
862
+ "loss": 0.7241,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.16406015539030977,
867
+ "grad_norm": 3.3521245738531813,
868
+ "learning_rate": 2.8242822605285323e-05,
869
+ "loss": 0.6931,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.16539397779185702,
874
+ "grad_norm": 11.826633616347152,
875
+ "learning_rate": 2.8212885218591812e-05,
876
+ "loss": 0.702,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.16672780019340425,
881
+ "grad_norm": 1.826627133990473,
882
+ "learning_rate": 2.8182711101047362e-05,
883
+ "loss": 0.7061,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.16806162259495147,
888
+ "grad_norm": 2.573810458184169,
889
+ "learning_rate": 2.815230079327156e-05,
890
+ "loss": 0.7369,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.16939544499649872,
895
+ "grad_norm": 3.2839708945300665,
896
+ "learning_rate": 2.8121654840115734e-05,
897
+ "loss": 0.6966,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.17072926739804595,
902
+ "grad_norm": 2.3799212816187136,
903
+ "learning_rate": 2.809077379065319e-05,
904
+ "loss": 0.6586,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.17206308979959317,
909
+ "grad_norm": 8.449506341181594,
910
+ "learning_rate": 2.805965819816937e-05,
911
+ "loss": 0.6759,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.17339691220114042,
916
+ "grad_norm": 19.59186920922371,
917
+ "learning_rate": 2.802830862015196e-05,
918
+ "loss": 0.6537,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.17473073460268765,
923
+ "grad_norm": 1.9822479021376653,
924
+ "learning_rate": 2.799672561828087e-05,
925
+ "loss": 0.6854,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.17606455700423487,
930
+ "grad_norm": 34.5695029919346,
931
+ "learning_rate": 2.79649097584182e-05,
932
+ "loss": 0.696,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.17739837940578213,
937
+ "grad_norm": 1.886275024045834,
938
+ "learning_rate": 2.7932861610598077e-05,
939
+ "loss": 0.7308,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.17873220180732935,
944
+ "grad_norm": 12.128310547971918,
945
+ "learning_rate": 2.7900581749016466e-05,
946
+ "loss": 0.6775,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.1800660242088766,
951
+ "grad_norm": 5.164562576250181,
952
+ "learning_rate": 2.7868070752020865e-05,
953
+ "loss": 0.7096,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.18139984661042383,
958
+ "grad_norm": 10.603161437547913,
959
+ "learning_rate": 2.7835329202099944e-05,
960
+ "loss": 0.7128,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.18273366901197105,
965
+ "grad_norm": 4.184026615605208,
966
+ "learning_rate": 2.7802357685873117e-05,
967
+ "loss": 0.6958,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.1840674914135183,
972
+ "grad_norm": 4.135408851494439,
973
+ "learning_rate": 2.7769156794080033e-05,
974
+ "loss": 0.6933,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.18540131381506553,
979
+ "grad_norm": 1.7535535971850658,
980
+ "learning_rate": 2.7735727121569967e-05,
981
+ "loss": 0.7088,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.18673513621661275,
986
+ "grad_norm": 5.354882317397714,
987
+ "learning_rate": 2.770206926729121e-05,
988
+ "loss": 0.6777,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.18806895861816,
993
+ "grad_norm": 3.2593978573810523,
994
+ "learning_rate": 2.7668183834280284e-05,
995
+ "loss": 0.6535,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.18940278101970723,
1000
+ "grad_norm": 2.48929554913285,
1001
+ "learning_rate": 2.763407142965117e-05,
1002
+ "loss": 0.6853,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.19073660342125445,
1007
+ "grad_norm": 7.550246976134938,
1008
+ "learning_rate": 2.759973266458444e-05,
1009
+ "loss": 0.679,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1920704258228017,
1014
+ "grad_norm": 5.033109901639584,
1015
+ "learning_rate": 2.756516815431627e-05,
1016
+ "loss": 0.6875,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.19340424822434893,
1021
+ "grad_norm": 6.1146492204468545,
1022
+ "learning_rate": 2.7530378518127445e-05,
1023
+ "loss": 0.6762,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.19473807062589615,
1028
+ "grad_norm": 72.48946246157955,
1029
+ "learning_rate": 2.7495364379332256e-05,
1030
+ "loss": 0.6766,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.1960718930274434,
1035
+ "grad_norm": 2.9676182245937426,
1036
+ "learning_rate": 2.7460126365267335e-05,
1037
+ "loss": 0.6731,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.19740571542899063,
1042
+ "grad_norm": 2.31300235137622,
1043
+ "learning_rate": 2.7424665107280402e-05,
1044
+ "loss": 0.6936,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.19873953783053785,
1049
+ "grad_norm": 2.594823775876976,
1050
+ "learning_rate": 2.738898124071898e-05,
1051
+ "loss": 0.6703,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.2000733602320851,
1056
+ "grad_norm": 2.8354629666044575,
1057
+ "learning_rate": 2.735307540491898e-05,
1058
+ "loss": 0.7022,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.20140718263363233,
1063
+ "grad_norm": 4.553033073507926,
1064
+ "learning_rate": 2.7316948243193273e-05,
1065
+ "loss": 0.6948,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.20274100503517956,
1070
+ "grad_norm": 10.052746707722667,
1071
+ "learning_rate": 2.7280600402820146e-05,
1072
+ "loss": 0.7216,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.2040748274367268,
1077
+ "grad_norm": 2.3783106567256977,
1078
+ "learning_rate": 2.724403253503171e-05,
1079
+ "loss": 0.6955,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.20540864983827403,
1084
+ "grad_norm": 2.26171864458522,
1085
+ "learning_rate": 2.7207245295002242e-05,
1086
+ "loss": 0.7062,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.20674247223982126,
1091
+ "grad_norm": 3.223919319241339,
1092
+ "learning_rate": 2.7170239341836436e-05,
1093
+ "loss": 0.6659,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.2080762946413685,
1098
+ "grad_norm": 1.6810934171118763,
1099
+ "learning_rate": 2.7133015338557585e-05,
1100
+ "loss": 0.7069,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.20941011704291573,
1105
+ "grad_norm": 2.1432477935562515,
1106
+ "learning_rate": 2.7095573952095727e-05,
1107
+ "loss": 0.703,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.21074393944446296,
1112
+ "grad_norm": 1.5654642597457342,
1113
+ "learning_rate": 2.705791585327568e-05,
1114
+ "loss": 0.7037,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.2120777618460102,
1119
+ "grad_norm": 2.5797486712147992,
1120
+ "learning_rate": 2.7020041716805014e-05,
1121
+ "loss": 0.6945,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.21341158424755743,
1126
+ "grad_norm": 6.993511462468879,
1127
+ "learning_rate": 2.6981952221261986e-05,
1128
+ "loss": 0.6965,
1129
+ "step": 1600
1130
+ }
1131
+ ],
1132
+ "logging_steps": 10,
1133
+ "max_steps": 7497,
1134
+ "num_input_tokens_seen": 0,
1135
+ "num_train_epochs": 1,
1136
+ "save_steps": 400,
1137
+ "stateful_callbacks": {
1138
+ "TrainerControl": {
1139
+ "args": {
1140
+ "should_epoch_stop": false,
1141
+ "should_evaluate": false,
1142
+ "should_log": false,
1143
+ "should_save": true,
1144
+ "should_training_stop": false
1145
+ },
1146
+ "attributes": {}
1147
+ }
1148
+ },
1149
+ "total_flos": 7.289521573986304e+18,
1150
+ "train_batch_size": 4,
1151
+ "trial_name": null,
1152
+ "trial_params": null
1153
+ }
checkpoint-1600/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3917773d9035785be322ba4c24f526b38823b179d4e738034f481213b6bfc50c
3
+ size 6520
checkpoint-1600/zero_to_fp32.py ADDED
@@ -0,0 +1,587 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
252
+ param_shapes = zero_model_states[0].param_shapes
253
+
254
+ # Reconstruction protocol:
255
+ #
256
+ # XXX: document this
257
+
258
+ if debug:
259
+ for i in range(world_size):
260
+ for j in range(len(fp32_flat_groups[0])):
261
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
262
+
263
+ # XXX: memory usage doubles here (zero2)
264
+ num_param_groups = len(fp32_flat_groups[0])
265
+ merged_single_partition_of_fp32_groups = []
266
+ for i in range(num_param_groups):
267
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
268
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
269
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
270
+ avail_numel = sum(
271
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
272
+
273
+ if debug:
274
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
275
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
276
+ # not asserting if there is a mismatch due to possible padding
277
+ print(f"Have {avail_numel} numels to process.")
278
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
279
+
280
+ # params
281
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
282
+ # out-of-core computing solution
283
+ total_numel = 0
284
+ total_params = 0
285
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
286
+ offset = 0
287
+ avail_numel = full_single_fp32_vector.numel()
288
+ for name, shape in shapes.items():
289
+
290
+ unpartitioned_numel = shape.numel()
291
+ total_numel += unpartitioned_numel
292
+ total_params += 1
293
+
294
+ if debug:
295
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
296
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
297
+ offset += unpartitioned_numel
298
+
299
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
300
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
301
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
302
+ # live optimizer object, so we are checking that the numbers are within the right range
303
+ align_to = 2 * world_size
304
+
305
+ def zero2_align(x):
306
+ return align_to * math.ceil(x / align_to)
307
+
308
+ if debug:
309
+ print(f"original offset={offset}, avail_numel={avail_numel}")
310
+
311
+ offset = zero2_align(offset)
312
+ avail_numel = zero2_align(avail_numel)
313
+
314
+ if debug:
315
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
316
+
317
+ # Sanity check
318
+ if offset != avail_numel:
319
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
320
+
321
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
322
+
323
+
324
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
325
+ state_dict = OrderedDict()
326
+
327
+ # buffers
328
+ buffers = zero_model_states[0].buffers
329
+ state_dict.update(buffers)
330
+ if debug:
331
+ print(f"added {len(buffers)} buffers")
332
+
333
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
334
+
335
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
336
+
337
+ # recover shared parameters
338
+ for pair in zero_model_states[0].shared_params:
339
+ if pair[1] in state_dict:
340
+ state_dict[pair[0]] = state_dict[pair[1]]
341
+
342
+ return state_dict
343
+
344
+
345
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
346
+ remainder = unpartitioned_numel % world_size
347
+ padding_numel = (world_size - remainder) if remainder else 0
348
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
349
+ return partitioned_numel, padding_numel
350
+
351
+
352
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
353
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
354
+ return
355
+
356
+ if debug:
357
+ for i in range(world_size):
358
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
359
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
360
+
361
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
362
+ wanted_params = len(frozen_param_shapes)
363
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
364
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
365
+ print(f'Frozen params: Have {avail_numel} numels to process.')
366
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
367
+
368
+ total_params = 0
369
+ total_numel = 0
370
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
371
+ total_params += 1
372
+ unpartitioned_numel = shape.numel()
373
+ total_numel += unpartitioned_numel
374
+
375
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
376
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
377
+
378
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
379
+
380
+ if debug:
381
+ print(
382
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
383
+ )
384
+
385
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
386
+
387
+
388
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
389
+ param_shapes = zero_model_states[0].param_shapes
390
+ avail_numel = fp32_flat_groups[0].numel() * world_size
391
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
392
+ # param, re-consolidating each param, while dealing with padding if any
393
+
394
+ # merge list of dicts, preserving order
395
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
396
+
397
+ if debug:
398
+ for i in range(world_size):
399
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
400
+
401
+ wanted_params = len(param_shapes)
402
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
403
+ # not asserting if there is a mismatch due to possible padding
404
+ avail_numel = fp32_flat_groups[0].numel() * world_size
405
+ print(f"Trainable params: Have {avail_numel} numels to process.")
406
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
407
+
408
+ # params
409
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
410
+ # out-of-core computing solution
411
+ offset = 0
412
+ total_numel = 0
413
+ total_params = 0
414
+ for name, shape in param_shapes.items():
415
+
416
+ unpartitioned_numel = shape.numel()
417
+ total_numel += unpartitioned_numel
418
+ total_params += 1
419
+
420
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
421
+
422
+ if debug:
423
+ print(
424
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
425
+ )
426
+
427
+ # XXX: memory usage doubles here
428
+ state_dict[name] = torch.cat(
429
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
430
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
431
+ offset += partitioned_numel
432
+
433
+ offset *= world_size
434
+
435
+ # Sanity check
436
+ if offset != avail_numel:
437
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
438
+
439
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
440
+
441
+
442
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
443
+ state_dict = OrderedDict()
444
+
445
+ # buffers
446
+ buffers = zero_model_states[0].buffers
447
+ state_dict.update(buffers)
448
+ if debug:
449
+ print(f"added {len(buffers)} buffers")
450
+
451
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
452
+
453
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
454
+
455
+ # recover shared parameters
456
+ for pair in zero_model_states[0].shared_params:
457
+ if pair[1] in state_dict:
458
+ state_dict[pair[0]] = state_dict[pair[1]]
459
+
460
+ return state_dict
461
+
462
+
463
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
464
+ """
465
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
466
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
467
+ via a model hub.
468
+
469
+ Args:
470
+ - ``checkpoint_dir``: path to the desired checkpoint folder
471
+ - ``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``
472
+
473
+ Returns:
474
+ - pytorch ``state_dict``
475
+
476
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
477
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
478
+ the checkpoint.
479
+
480
+ A typical usage might be ::
481
+
482
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
483
+ # do the training and checkpoint saving
484
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
485
+ model = model.cpu() # move to cpu
486
+ model.load_state_dict(state_dict)
487
+ # submit to model hub or save the model to share with others
488
+
489
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
490
+ application. i.e. you will need to re-initialize the deepspeed engine, since
491
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
492
+
493
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
494
+
495
+ """
496
+ if tag is None:
497
+ latest_path = os.path.join(checkpoint_dir, 'latest')
498
+ if os.path.isfile(latest_path):
499
+ with open(latest_path, 'r') as fd:
500
+ tag = fd.read().strip()
501
+ else:
502
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
503
+
504
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
505
+
506
+ if not os.path.isdir(ds_checkpoint_dir):
507
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
508
+
509
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
510
+
511
+
512
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
513
+ """
514
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
515
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
516
+
517
+ Args:
518
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
519
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
520
+ - ``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``
521
+ """
522
+
523
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
524
+ print(f"Saving fp32 state dict to {output_file}")
525
+ torch.save(state_dict, output_file)
526
+
527
+
528
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
529
+ """
530
+ 1. Put the provided model to cpu
531
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
532
+ 3. Load it into the provided model
533
+
534
+ Args:
535
+ - ``model``: the model object to update
536
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
537
+ - ``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``
538
+
539
+ Returns:
540
+ - ``model`: modified model
541
+
542
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
543
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
544
+ conveniently placed for you in the checkpoint folder.
545
+
546
+ A typical usage might be ::
547
+
548
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
549
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
550
+ # submit to model hub or save the model to share with others
551
+
552
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
553
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
554
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
555
+
556
+ """
557
+ logger.info(f"Extracting fp32 weights")
558
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
559
+
560
+ logger.info(f"Overwriting model with fp32 weights")
561
+ model = model.cpu()
562
+ model.load_state_dict(state_dict, strict=False)
563
+
564
+ return model
565
+
566
+
567
+ if __name__ == "__main__":
568
+
569
+ parser = argparse.ArgumentParser()
570
+ parser.add_argument("checkpoint_dir",
571
+ type=str,
572
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
573
+ parser.add_argument(
574
+ "output_file",
575
+ type=str,
576
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
577
+ parser.add_argument("-t",
578
+ "--tag",
579
+ type=str,
580
+ default=None,
581
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
582
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
583
+ args = parser.parse_args()
584
+
585
+ debug = args.debug
586
+
587
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
checkpoint-3200/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: Qwen/Qwen-VL-Chat
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 0.10.0
checkpoint-3200/adapter_config.json ADDED
@@ -0,0 +1,380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 64,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.25.mlp.w2",
24
+ "transformer.visual.transformer.resblocks.39.mlp.c_proj",
25
+ "transformer.visual.transformer.resblocks.21.attn.in_proj",
26
+ "transformer.visual.transformer.resblocks.44.mlp.c_proj",
27
+ "transformer.h.15.mlp.w2",
28
+ "transformer.visual.transformer.resblocks.7.mlp.c_proj",
29
+ "transformer.h.3.mlp.w2",
30
+ "transformer.visual.transformer.resblocks.23.mlp.c_fc",
31
+ "transformer.visual.transformer.resblocks.10.mlp.c_fc",
32
+ "transformer.h.20.attn.c_attn",
33
+ "transformer.h.23.mlp.c_proj",
34
+ "transformer.visual.transformer.resblocks.34.mlp.c_fc",
35
+ "transformer.h.0.attn.c_proj",
36
+ "transformer.h.9.mlp.w1",
37
+ "transformer.visual.transformer.resblocks.12.attn.out_proj",
38
+ "transformer.visual.transformer.resblocks.22.mlp.c_fc",
39
+ "transformer.visual.transformer.resblocks.23.attn.out_proj",
40
+ "transformer.h.27.mlp.c_proj",
41
+ "transformer.visual.transformer.resblocks.1.mlp.c_proj",
42
+ "transformer.h.22.attn.c_proj",
43
+ "transformer.h.25.mlp.w1",
44
+ "transformer.visual.transformer.resblocks.0.attn.in_proj",
45
+ "transformer.visual.transformer.resblocks.2.attn.in_proj",
46
+ "transformer.h.1.attn.c_proj",
47
+ "transformer.visual.transformer.resblocks.45.mlp.c_proj",
48
+ "transformer.h.29.attn.c_attn",
49
+ "transformer.visual.transformer.resblocks.43.attn.in_proj",
50
+ "transformer.visual.transformer.resblocks.46.mlp.c_proj",
51
+ "transformer.h.28.mlp.w2",
52
+ "transformer.h.20.attn.c_proj",
53
+ "transformer.h.12.attn.c_proj",
54
+ "transformer.visual.transformer.resblocks.27.attn.out_proj",
55
+ "transformer.visual.transformer.resblocks.17.mlp.c_proj",
56
+ "transformer.visual.transformer.resblocks.33.attn.out_proj",
57
+ "transformer.h.4.attn.c_proj",
58
+ "transformer.visual.transformer.resblocks.29.attn.in_proj",
59
+ "transformer.visual.transformer.resblocks.2.attn.out_proj",
60
+ "transformer.h.4.mlp.w2",
61
+ "transformer.h.30.mlp.w1",
62
+ "transformer.h.31.attn.c_attn",
63
+ "transformer.h.10.attn.c_proj",
64
+ "transformer.h.18.mlp.w2",
65
+ "transformer.visual.transformer.resblocks.12.attn.in_proj",
66
+ "transformer.h.20.mlp.c_proj",
67
+ "transformer.visual.transformer.resblocks.36.attn.out_proj",
68
+ "transformer.visual.transformer.resblocks.31.attn.out_proj",
69
+ "transformer.h.1.attn.c_attn",
70
+ "transformer.visual.transformer.resblocks.17.attn.out_proj",
71
+ "transformer.visual.transformer.resblocks.9.mlp.c_fc",
72
+ "transformer.h.22.attn.c_attn",
73
+ "transformer.h.16.attn.c_attn",
74
+ "transformer.visual.transformer.resblocks.20.mlp.c_fc",
75
+ "transformer.visual.transformer.resblocks.31.attn.in_proj",
76
+ "transformer.h.29.mlp.w1",
77
+ "transformer.h.27.attn.c_attn",
78
+ "transformer.visual.transformer.resblocks.18.mlp.c_proj",
79
+ "transformer.visual.transformer.resblocks.8.mlp.c_proj",
80
+ "transformer.visual.transformer.resblocks.44.attn.out_proj",
81
+ "transformer.h.5.attn.c_proj",
82
+ "transformer.h.20.mlp.w1",
83
+ "transformer.h.0.mlp.c_proj",
84
+ "transformer.visual.transformer.resblocks.9.attn.out_proj",
85
+ "transformer.h.2.mlp.w1",
86
+ "transformer.visual.transformer.resblocks.3.mlp.c_proj",
87
+ "transformer.visual.transformer.resblocks.30.mlp.c_fc",
88
+ "transformer.visual.transformer.resblocks.36.mlp.c_proj",
89
+ "transformer.h.17.mlp.w2",
90
+ "transformer.h.19.mlp.w2",
91
+ "transformer.visual.transformer.resblocks.30.attn.out_proj",
92
+ "transformer.visual.transformer.resblocks.40.mlp.c_proj",
93
+ "transformer.visual.transformer.resblocks.1.mlp.c_fc",
94
+ "transformer.h.10.mlp.w1",
95
+ "transformer.h.6.attn.c_attn",
96
+ "transformer.visual.transformer.resblocks.14.mlp.c_fc",
97
+ "transformer.h.24.mlp.w2",
98
+ "transformer.visual.transformer.resblocks.28.attn.out_proj",
99
+ "transformer.visual.transformer.resblocks.27.attn.in_proj",
100
+ "transformer.h.10.attn.c_attn",
101
+ "transformer.h.17.mlp.c_proj",
102
+ "transformer.visual.transformer.resblocks.0.mlp.c_fc",
103
+ "transformer.visual.transformer.resblocks.35.attn.out_proj",
104
+ "transformer.h.6.attn.c_proj",
105
+ "transformer.h.6.mlp.w2",
106
+ "transformer.visual.transformer.resblocks.21.mlp.c_fc",
107
+ "transformer.h.22.mlp.c_proj",
108
+ "transformer.h.4.mlp.w1",
109
+ "transformer.h.14.mlp.w1",
110
+ "transformer.h.2.mlp.c_proj",
111
+ "transformer.visual.transformer.resblocks.31.mlp.c_fc",
112
+ "transformer.h.28.attn.c_proj",
113
+ "transformer.visual.transformer.resblocks.19.attn.in_proj",
114
+ "transformer.h.26.attn.c_proj",
115
+ "transformer.visual.transformer.resblocks.47.mlp.c_fc",
116
+ "transformer.h.3.attn.c_proj",
117
+ "transformer.visual.transformer.resblocks.33.attn.in_proj",
118
+ "transformer.visual.transformer.resblocks.45.attn.out_proj",
119
+ "transformer.visual.transformer.resblocks.5.attn.out_proj",
120
+ "transformer.h.7.attn.c_attn",
121
+ "transformer.h.1.mlp.w1",
122
+ "transformer.h.19.mlp.c_proj",
123
+ "transformer.visual.transformer.resblocks.39.attn.in_proj",
124
+ "transformer.visual.transformer.resblocks.7.attn.out_proj",
125
+ "transformer.h.8.mlp.w1",
126
+ "transformer.h.22.mlp.w1",
127
+ "transformer.h.7.attn.c_proj",
128
+ "transformer.visual.transformer.resblocks.0.mlp.c_proj",
129
+ "transformer.h.7.mlp.w2",
130
+ "transformer.h.25.attn.c_proj",
131
+ "transformer.visual.transformer.resblocks.17.attn.in_proj",
132
+ "transformer.visual.transformer.resblocks.24.attn.out_proj",
133
+ "transformer.h.24.mlp.c_proj",
134
+ "transformer.visual.transformer.resblocks.19.mlp.c_fc",
135
+ "transformer.h.21.mlp.w2",
136
+ "transformer.h.9.mlp.w2",
137
+ "transformer.h.27.attn.c_proj",
138
+ "transformer.h.2.attn.c_proj",
139
+ "transformer.h.5.mlp.c_proj",
140
+ "transformer.visual.transformer.resblocks.18.attn.in_proj",
141
+ "transformer.h.18.attn.c_attn",
142
+ "transformer.visual.transformer.resblocks.37.attn.in_proj",
143
+ "transformer.visual.transformer.resblocks.15.attn.out_proj",
144
+ "transformer.h.9.mlp.c_proj",
145
+ "transformer.visual.transformer.resblocks.37.mlp.c_proj",
146
+ "transformer.h.17.attn.c_proj",
147
+ "transformer.visual.transformer.resblocks.4.attn.out_proj",
148
+ "transformer.h.11.mlp.w2",
149
+ "transformer.h.3.mlp.w1",
150
+ "transformer.visual.transformer.resblocks.32.attn.out_proj",
151
+ "transformer.h.30.attn.c_attn",
152
+ "transformer.h.7.mlp.c_proj",
153
+ "transformer.visual.transformer.resblocks.38.mlp.c_proj",
154
+ "transformer.visual.transformer.resblocks.22.mlp.c_proj",
155
+ "transformer.h.23.mlp.w2",
156
+ "transformer.visual.transformer.resblocks.8.mlp.c_fc",
157
+ "transformer.h.24.attn.c_attn",
158
+ "transformer.visual.transformer.resblocks.21.attn.out_proj",
159
+ "transformer.visual.transformer.resblocks.16.mlp.c_fc",
160
+ "transformer.visual.transformer.resblocks.30.mlp.c_proj",
161
+ "transformer.visual.transformer.resblocks.29.attn.out_proj",
162
+ "transformer.visual.transformer.resblocks.19.attn.out_proj",
163
+ "transformer.visual.transformer.resblocks.1.attn.in_proj",
164
+ "transformer.h.27.mlp.w1",
165
+ "transformer.h.1.mlp.w2",
166
+ "transformer.h.17.attn.c_attn",
167
+ "transformer.visual.transformer.resblocks.3.mlp.c_fc",
168
+ "transformer.visual.transformer.resblocks.24.attn.in_proj",
169
+ "transformer.visual.transformer.resblocks.44.attn.in_proj",
170
+ "transformer.h.6.mlp.c_proj",
171
+ "transformer.visual.transformer.resblocks.3.attn.in_proj",
172
+ "transformer.visual.transformer.resblocks.19.mlp.c_proj",
173
+ "transformer.visual.transformer.resblocks.38.mlp.c_fc",
174
+ "transformer.h.14.mlp.w2",
175
+ "transformer.h.8.mlp.w2",
176
+ "transformer.h.19.mlp.w1",
177
+ "transformer.visual.transformer.resblocks.15.mlp.c_fc",
178
+ "transformer.h.19.attn.c_attn",
179
+ "transformer.h.13.attn.c_proj",
180
+ "transformer.visual.transformer.resblocks.26.mlp.c_fc",
181
+ "transformer.visual.transformer.resblocks.35.mlp.c_proj",
182
+ "transformer.h.15.mlp.w1",
183
+ "transformer.h.23.attn.c_attn",
184
+ "transformer.visual.transformer.resblocks.6.attn.out_proj",
185
+ "transformer.visual.transformer.resblocks.24.mlp.c_fc",
186
+ "transformer.visual.transformer.resblocks.18.attn.out_proj",
187
+ "transformer.visual.transformer.resblocks.20.attn.in_proj",
188
+ "transformer.visual.transformer.resblocks.31.mlp.c_proj",
189
+ "transformer.visual.transformer.resblocks.43.attn.out_proj",
190
+ "transformer.h.5.attn.c_attn",
191
+ "transformer.visual.transformer.resblocks.40.attn.out_proj",
192
+ "transformer.visual.transformer.resblocks.29.mlp.c_fc",
193
+ "transformer.h.12.mlp.w1",
194
+ "transformer.visual.transformer.resblocks.13.mlp.c_proj",
195
+ "transformer.visual.transformer.resblocks.43.mlp.c_proj",
196
+ "transformer.visual.transformer.resblocks.11.mlp.c_proj",
197
+ "transformer.visual.transformer.resblocks.37.attn.out_proj",
198
+ "transformer.visual.transformer.resblocks.40.mlp.c_fc",
199
+ "transformer.h.16.mlp.w2",
200
+ "transformer.h.9.attn.c_proj",
201
+ "transformer.visual.transformer.resblocks.26.attn.out_proj",
202
+ "transformer.visual.transformer.resblocks.4.attn.in_proj",
203
+ "transformer.visual.transformer.resblocks.10.attn.out_proj",
204
+ "transformer.h.13.mlp.c_proj",
205
+ "transformer.visual.transformer.resblocks.32.attn.in_proj",
206
+ "transformer.h.25.mlp.c_proj",
207
+ "transformer.h.24.mlp.w1",
208
+ "transformer.h.4.mlp.c_proj",
209
+ "transformer.h.21.attn.c_attn",
210
+ "transformer.visual.transformer.resblocks.37.mlp.c_fc",
211
+ "transformer.h.11.attn.c_proj",
212
+ "transformer.h.30.mlp.c_proj",
213
+ "transformer.visual.transformer.resblocks.23.mlp.c_proj",
214
+ "transformer.visual.transformer.resblocks.32.mlp.c_proj",
215
+ "transformer.h.21.attn.c_proj",
216
+ "transformer.h.3.attn.c_attn",
217
+ "transformer.visual.transformer.resblocks.5.mlp.c_proj",
218
+ "transformer.visual.transformer.resblocks.12.mlp.c_fc",
219
+ "transformer.visual.transformer.resblocks.45.attn.in_proj",
220
+ "transformer.h.4.attn.c_attn",
221
+ "transformer.visual.transformer.resblocks.33.mlp.c_proj",
222
+ "transformer.h.19.attn.c_proj",
223
+ "transformer.h.5.mlp.w1",
224
+ "transformer.visual.transformer.resblocks.14.attn.in_proj",
225
+ "transformer.h.9.attn.c_attn",
226
+ "transformer.visual.transformer.resblocks.45.mlp.c_fc",
227
+ "transformer.h.22.mlp.w2",
228
+ "transformer.visual.transformer.resblocks.41.attn.in_proj",
229
+ "transformer.h.3.mlp.c_proj",
230
+ "transformer.visual.transformer.resblocks.5.mlp.c_fc",
231
+ "transformer.h.26.attn.c_attn",
232
+ "transformer.h.30.attn.c_proj",
233
+ "transformer.visual.transformer.resblocks.10.attn.in_proj",
234
+ "transformer.visual.transformer.resblocks.24.mlp.c_proj",
235
+ "transformer.visual.transformer.resblocks.4.mlp.c_proj",
236
+ "transformer.visual.transformer.resblocks.44.mlp.c_fc",
237
+ "transformer.visual.transformer.resblocks.38.attn.in_proj",
238
+ "transformer.visual.transformer.resblocks.13.mlp.c_fc",
239
+ "transformer.visual.transformer.resblocks.8.attn.out_proj",
240
+ "transformer.visual.transformer.resblocks.23.attn.in_proj",
241
+ "transformer.visual.transformer.resblocks.41.mlp.c_proj",
242
+ "transformer.visual.transformer.resblocks.42.mlp.c_fc",
243
+ "transformer.visual.transformer.resblocks.39.mlp.c_fc",
244
+ "transformer.h.17.mlp.w1",
245
+ "transformer.visual.transformer.resblocks.42.attn.out_proj",
246
+ "transformer.visual.transformer.resblocks.8.attn.in_proj",
247
+ "transformer.h.25.attn.c_attn",
248
+ "transformer.visual.transformer.resblocks.1.attn.out_proj",
249
+ "transformer.h.6.mlp.w1",
250
+ "transformer.visual.transformer.resblocks.46.attn.out_proj",
251
+ "transformer.h.8.attn.c_attn",
252
+ "transformer.h.13.mlp.w2",
253
+ "transformer.visual.transformer.resblocks.21.mlp.c_proj",
254
+ "transformer.h.2.mlp.w2",
255
+ "transformer.visual.transformer.resblocks.30.attn.in_proj",
256
+ "transformer.h.7.mlp.w1",
257
+ "transformer.visual.transformer.resblocks.25.mlp.c_fc",
258
+ "transformer.h.11.mlp.c_proj",
259
+ "transformer.visual.transformer.resblocks.36.mlp.c_fc",
260
+ "transformer.visual.transformer.resblocks.42.mlp.c_proj",
261
+ "transformer.h.11.attn.c_attn",
262
+ "transformer.h.29.mlp.c_proj",
263
+ "transformer.visual.transformer.resblocks.36.attn.in_proj",
264
+ "transformer.visual.transformer.resblocks.40.attn.in_proj",
265
+ "transformer.visual.transformer.resblocks.27.mlp.c_fc",
266
+ "transformer.h.11.mlp.w1",
267
+ "transformer.visual.transformer.resblocks.39.attn.out_proj",
268
+ "transformer.visual.transformer.resblocks.10.mlp.c_proj",
269
+ "transformer.visual.transformer.resblocks.46.mlp.c_fc",
270
+ "transformer.h.20.mlp.w2",
271
+ "transformer.h.18.attn.c_proj",
272
+ "transformer.visual.transformer.resblocks.4.mlp.c_fc",
273
+ "transformer.visual.transformer.resblocks.17.mlp.c_fc",
274
+ "transformer.visual.transformer.resblocks.2.mlp.c_proj",
275
+ "transformer.visual.transformer.resblocks.6.attn.in_proj",
276
+ "transformer.visual.transformer.resblocks.11.mlp.c_fc",
277
+ "transformer.h.12.attn.c_attn",
278
+ "transformer.visual.transformer.resblocks.16.attn.in_proj",
279
+ "transformer.h.14.attn.c_attn",
280
+ "transformer.h.31.mlp.c_proj",
281
+ "transformer.visual.transformer.resblocks.16.attn.out_proj",
282
+ "transformer.h.0.mlp.w1",
283
+ "transformer.visual.transformer.resblocks.43.mlp.c_fc",
284
+ "transformer.visual.transformer.resblocks.5.attn.in_proj",
285
+ "transformer.h.13.attn.c_attn",
286
+ "transformer.visual.transformer.resblocks.41.mlp.c_fc",
287
+ "transformer.h.23.attn.c_proj",
288
+ "transformer.visual.transformer.resblocks.35.mlp.c_fc",
289
+ "transformer.visual.transformer.resblocks.12.mlp.c_proj",
290
+ "transformer.h.8.attn.c_proj",
291
+ "transformer.h.31.attn.c_proj",
292
+ "transformer.visual.transformer.resblocks.26.mlp.c_proj",
293
+ "transformer.h.21.mlp.w1",
294
+ "transformer.visual.transformer.resblocks.16.mlp.c_proj",
295
+ "transformer.h.31.mlp.w1",
296
+ "transformer.visual.transformer.resblocks.13.attn.in_proj",
297
+ "transformer.h.1.mlp.c_proj",
298
+ "transformer.h.12.mlp.c_proj",
299
+ "transformer.h.30.mlp.w2",
300
+ "transformer.h.14.attn.c_proj",
301
+ "transformer.visual.transformer.resblocks.15.mlp.c_proj",
302
+ "transformer.visual.transformer.resblocks.7.attn.in_proj",
303
+ "transformer.visual.transformer.resblocks.22.attn.out_proj",
304
+ "transformer.visual.transformer.resblocks.0.attn.out_proj",
305
+ "transformer.h.31.mlp.w2",
306
+ "transformer.visual.transformer.resblocks.27.mlp.c_proj",
307
+ "transformer.visual.transformer.resblocks.25.attn.out_proj",
308
+ "transformer.visual.transformer.resblocks.22.attn.in_proj",
309
+ "transformer.visual.transformer.resblocks.33.mlp.c_fc",
310
+ "transformer.h.13.mlp.w1",
311
+ "transformer.visual.transformer.resblocks.6.mlp.c_fc",
312
+ "transformer.visual.transformer.resblocks.11.attn.in_proj",
313
+ "transformer.visual.transformer.resblocks.9.attn.in_proj",
314
+ "transformer.visual.transformer.resblocks.20.mlp.c_proj",
315
+ "transformer.h.16.mlp.c_proj",
316
+ "transformer.visual.transformer.resblocks.15.attn.in_proj",
317
+ "transformer.h.26.mlp.w2",
318
+ "transformer.h.28.attn.c_attn",
319
+ "transformer.visual.transformer.resblocks.29.mlp.c_proj",
320
+ "transformer.h.23.mlp.w1",
321
+ "transformer.h.14.mlp.c_proj",
322
+ "transformer.visual.transformer.resblocks.25.mlp.c_proj",
323
+ "transformer.visual.transformer.resblocks.9.mlp.c_proj",
324
+ "transformer.visual.transformer.resblocks.34.attn.in_proj",
325
+ "transformer.h.5.mlp.w2",
326
+ "transformer.h.26.mlp.c_proj",
327
+ "transformer.h.24.attn.c_proj",
328
+ "transformer.visual.transformer.resblocks.47.attn.in_proj",
329
+ "transformer.visual.transformer.resblocks.7.mlp.c_fc",
330
+ "transformer.h.2.attn.c_attn",
331
+ "transformer.visual.conv1",
332
+ "transformer.visual.transformer.resblocks.28.attn.in_proj",
333
+ "transformer.visual.transformer.resblocks.25.attn.in_proj",
334
+ "transformer.visual.transformer.resblocks.47.attn.out_proj",
335
+ "transformer.visual.transformer.resblocks.18.mlp.c_fc",
336
+ "transformer.h.28.mlp.w1",
337
+ "transformer.h.10.mlp.w2",
338
+ "transformer.h.15.attn.c_proj",
339
+ "transformer.h.21.mlp.c_proj",
340
+ "transformer.h.29.mlp.w2",
341
+ "transformer.visual.transformer.resblocks.32.mlp.c_fc",
342
+ "transformer.h.18.mlp.w1",
343
+ "transformer.visual.transformer.resblocks.41.attn.out_proj",
344
+ "transformer.h.8.mlp.c_proj",
345
+ "transformer.visual.transformer.resblocks.47.mlp.c_proj",
346
+ "transformer.visual.transformer.resblocks.28.mlp.c_fc",
347
+ "transformer.visual.transformer.resblocks.34.attn.out_proj",
348
+ "transformer.h.29.attn.c_proj",
349
+ "transformer.visual.transformer.resblocks.38.attn.out_proj",
350
+ "transformer.visual.transformer.resblocks.14.attn.out_proj",
351
+ "transformer.h.10.mlp.c_proj",
352
+ "transformer.visual.transformer.resblocks.6.mlp.c_proj",
353
+ "transformer.h.0.mlp.w2",
354
+ "transformer.visual.transformer.resblocks.26.attn.in_proj",
355
+ "transformer.visual.transformer.resblocks.42.attn.in_proj",
356
+ "transformer.h.15.attn.c_attn",
357
+ "transformer.h.16.mlp.w1",
358
+ "transformer.visual.transformer.resblocks.3.attn.out_proj",
359
+ "transformer.visual.transformer.resblocks.11.attn.out_proj",
360
+ "transformer.visual.transformer.resblocks.28.mlp.c_proj",
361
+ "transformer.visual.transformer.resblocks.14.mlp.c_proj",
362
+ "transformer.h.0.attn.c_attn",
363
+ "transformer.h.18.mlp.c_proj",
364
+ "transformer.h.12.mlp.w2",
365
+ "transformer.visual.transformer.resblocks.35.attn.in_proj",
366
+ "transformer.visual.transformer.resblocks.2.mlp.c_fc",
367
+ "transformer.visual.transformer.resblocks.34.mlp.c_proj",
368
+ "transformer.h.26.mlp.w1",
369
+ "transformer.visual.transformer.resblocks.13.attn.out_proj",
370
+ "transformer.visual.transformer.resblocks.20.attn.out_proj",
371
+ "transformer.visual.transformer.resblocks.46.attn.in_proj",
372
+ "transformer.h.16.attn.c_proj",
373
+ "transformer.h.15.mlp.c_proj",
374
+ "transformer.h.27.mlp.w2",
375
+ "transformer.h.28.mlp.c_proj"
376
+ ],
377
+ "task_type": "CAUSAL_LM",
378
+ "use_dora": false,
379
+ "use_rslora": false
380
+ }
checkpoint-3200/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52fa03b5d3f60fe9a4c51cb0b16aafc218c8609e4f138129e9f8c1714f435083
3
+ size 469105640
checkpoint-3200/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step3200
checkpoint-3200/qwen.tiktoken ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-3200/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3639e6e74e4e1cfeae5cdc2ce9cd5b94f31fff478e7e917a5c449e4cae9026b
3
+ size 14960
checkpoint-3200/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c52f1e9778cd17f1a3644cac97323d0a2e7c0b349b3b00dd2e95c6a4224af088
3
+ size 14960
checkpoint-3200/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87e692a08bc8bbf1f61da918a3909e038839c02f5b2e391d3ec7aab121a15d14
3
+ size 14960
checkpoint-3200/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e71530355bd30c84720be857f8caa99718c58da763e9dd9ce35d28577582ea2d
3
+ size 14960
checkpoint-3200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:578ab9959f00bcfb2c7fa12abda00e812b706036a543db6f8556dc91c2db5694
3
+ size 1064
checkpoint-3200/special_tokens_map.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "pad_token": "<|endoftext|>"
3
+ }
checkpoint-3200/tokenizer_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {},
3
+ "auto_map": {
4
+ "AutoTokenizer": [
5
+ "Qwen/Qwen-VL-Chat--tokenization_qwen.QWenTokenizer",
6
+ null
7
+ ]
8
+ },
9
+ "clean_up_tokenization_spaces": true,
10
+ "model_max_length": 1280,
11
+ "pad_token": "<|endoftext|>",
12
+ "padding_side": "right",
13
+ "tokenizer_class": "QWenTokenizer"
14
+ }
checkpoint-3200/trainer_state.json ADDED
@@ -0,0 +1,2273 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.42682316849511487,
5
+ "eval_steps": 500,
6
+ "global_step": 3200,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001333822401547234,
13
+ "grad_norm": 72.53171450734499,
14
+ "learning_rate": 4e-06,
15
+ "loss": 1.0439,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.002667644803094468,
20
+ "grad_norm": 7.72170877312484,
21
+ "learning_rate": 8e-06,
22
+ "loss": 1.0385,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.004001467204641702,
27
+ "grad_norm": 28.037294890647996,
28
+ "learning_rate": 1.2e-05,
29
+ "loss": 1.0328,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.005335289606188936,
34
+ "grad_norm": 10.980963099753343,
35
+ "learning_rate": 1.6e-05,
36
+ "loss": 0.9608,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.00666911200773617,
41
+ "grad_norm": 9.853684997615185,
42
+ "learning_rate": 1.9999999999999998e-05,
43
+ "loss": 0.8941,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.008002934409283404,
48
+ "grad_norm": 33.396994669126855,
49
+ "learning_rate": 2.4e-05,
50
+ "loss": 0.9174,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.009336756810830639,
55
+ "grad_norm": 5.621190624979587,
56
+ "learning_rate": 2.8e-05,
57
+ "loss": 0.8146,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.010670579212377872,
62
+ "grad_norm": 2.4646688601781843,
63
+ "learning_rate": 2.9999966406213696e-05,
64
+ "loss": 0.7774,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.012004401613925107,
69
+ "grad_norm": 2.1272107013027215,
70
+ "learning_rate": 2.9999697656826056e-05,
71
+ "loss": 0.7426,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01333822401547234,
76
+ "grad_norm": 2.0001927521328575,
77
+ "learning_rate": 2.9999160162865885e-05,
78
+ "loss": 0.7251,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.014672046417019574,
83
+ "grad_norm": 4.278494734910313,
84
+ "learning_rate": 2.9998353933963273e-05,
85
+ "loss": 0.7283,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.016005868818566808,
90
+ "grad_norm": 84.74138622339441,
91
+ "learning_rate": 2.999727898456315e-05,
92
+ "loss": 0.721,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.017339691220114042,
97
+ "grad_norm": 2.5596285292704533,
98
+ "learning_rate": 2.999593533392503e-05,
99
+ "loss": 0.7476,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.018673513621661277,
104
+ "grad_norm": 3.167163073798878,
105
+ "learning_rate": 2.9994323006122654e-05,
106
+ "loss": 0.7701,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.02000733602320851,
111
+ "grad_norm": 12.240132620099228,
112
+ "learning_rate": 2.9992442030043557e-05,
113
+ "loss": 0.7042,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.021341158424755743,
118
+ "grad_norm": 2.266079140233148,
119
+ "learning_rate": 2.9990292439388565e-05,
120
+ "loss": 0.6945,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.022674980826302978,
125
+ "grad_norm": 4.493189214061043,
126
+ "learning_rate": 2.9987874272671168e-05,
127
+ "loss": 0.7811,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.024008803227850213,
132
+ "grad_norm": 3.8513915234854132,
133
+ "learning_rate": 2.9985187573216855e-05,
134
+ "loss": 0.7229,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.025342625629397444,
139
+ "grad_norm": 2.1061250870336896,
140
+ "learning_rate": 2.998223238916232e-05,
141
+ "loss": 0.7253,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.02667644803094468,
146
+ "grad_norm": 2.8326879963181373,
147
+ "learning_rate": 2.9979008773454618e-05,
148
+ "loss": 0.7576,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.028010270432491914,
153
+ "grad_norm": 3.0294324007099402,
154
+ "learning_rate": 2.997551678385019e-05,
155
+ "loss": 0.7362,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02934409283403915,
160
+ "grad_norm": 1.5691302199679913,
161
+ "learning_rate": 2.997175648291384e-05,
162
+ "loss": 0.7212,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.03067791523558638,
167
+ "grad_norm": 4.379709379404857,
168
+ "learning_rate": 2.996772793801763e-05,
169
+ "loss": 0.7382,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.032011737637133615,
174
+ "grad_norm": 1.899348727085658,
175
+ "learning_rate": 2.996343122133965e-05,
176
+ "loss": 0.7056,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.033345560038680847,
181
+ "grad_norm": 21.481584335801955,
182
+ "learning_rate": 2.9958866409862745e-05,
183
+ "loss": 0.697,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.034679382440228085,
188
+ "grad_norm": 3.022034785435926,
189
+ "learning_rate": 2.9954033585373108e-05,
190
+ "loss": 0.69,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.036013204841775316,
195
+ "grad_norm": 2.869274575136443,
196
+ "learning_rate": 2.994893283445885e-05,
197
+ "loss": 0.7026,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.037347027243322554,
202
+ "grad_norm": 7.8785627268474,
203
+ "learning_rate": 2.9943564248508415e-05,
204
+ "loss": 0.7021,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.038680849644869786,
209
+ "grad_norm": 4.450751706186503,
210
+ "learning_rate": 2.9937927923708966e-05,
211
+ "loss": 0.6814,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.04001467204641702,
216
+ "grad_norm": 11.613541400971123,
217
+ "learning_rate": 2.993202396104465e-05,
218
+ "loss": 0.7043,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.041348494447964255,
223
+ "grad_norm": 2.0602161594051536,
224
+ "learning_rate": 2.9925852466294795e-05,
225
+ "loss": 0.7073,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.04268231684951149,
230
+ "grad_norm": 19.383133688390824,
231
+ "learning_rate": 2.9919413550032014e-05,
232
+ "loss": 0.6965,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.04401613925105872,
237
+ "grad_norm": 7.9603948136759906,
238
+ "learning_rate": 2.991270732762022e-05,
239
+ "loss": 0.7029,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.045349961652605957,
244
+ "grad_norm": 2.810319267000959,
245
+ "learning_rate": 2.990573391921255e-05,
246
+ "loss": 0.7412,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.04668378405415319,
251
+ "grad_norm": 6.082487526538632,
252
+ "learning_rate": 2.989849344974924e-05,
253
+ "loss": 0.6913,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.048017606455700426,
258
+ "grad_norm": 1.3651265332328906,
259
+ "learning_rate": 2.9890986048955368e-05,
260
+ "loss": 0.6802,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.04935142885724766,
265
+ "grad_norm": 2.7766095141894023,
266
+ "learning_rate": 2.9883211851338516e-05,
267
+ "loss": 0.7254,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.05068525125879489,
272
+ "grad_norm": 2.2284551774435033,
273
+ "learning_rate": 2.9875170996186392e-05,
274
+ "loss": 0.7026,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.05201907366034213,
279
+ "grad_norm": 2.1609680967998957,
280
+ "learning_rate": 2.986686362756431e-05,
281
+ "loss": 0.7415,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.05335289606188936,
286
+ "grad_norm": 2.071743222471935,
287
+ "learning_rate": 2.9858289894312617e-05,
288
+ "loss": 0.7081,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.0546867184634366,
293
+ "grad_norm": 7.349693198515479,
294
+ "learning_rate": 2.9849449950044036e-05,
295
+ "loss": 0.7212,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.05602054086498383,
300
+ "grad_norm": 12.990346405367537,
301
+ "learning_rate": 2.984034395314088e-05,
302
+ "loss": 0.7187,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.05735436326653106,
307
+ "grad_norm": 2.4106321189642723,
308
+ "learning_rate": 2.983097206675227e-05,
309
+ "loss": 0.6881,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0586881856680783,
314
+ "grad_norm": 3.1647778114563163,
315
+ "learning_rate": 2.9821334458791156e-05,
316
+ "loss": 0.7529,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.06002200806962553,
321
+ "grad_norm": 2.7769936556882713,
322
+ "learning_rate": 2.9811431301931344e-05,
323
+ "loss": 0.7073,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.06135583047117276,
328
+ "grad_norm": 2.834916196253906,
329
+ "learning_rate": 2.9801262773604377e-05,
330
+ "loss": 0.6649,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.06268965287271999,
335
+ "grad_norm": 2.431783869557187,
336
+ "learning_rate": 2.9790829055996398e-05,
337
+ "loss": 0.6889,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.06402347527426723,
342
+ "grad_norm": 3.7325728788625825,
343
+ "learning_rate": 2.978013033604483e-05,
344
+ "loss": 0.6771,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.06535729767581447,
349
+ "grad_norm": 10.194734403509758,
350
+ "learning_rate": 2.976916680543506e-05,
351
+ "loss": 0.6917,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.06669112007736169,
356
+ "grad_norm": 6.4699962355353655,
357
+ "learning_rate": 2.975793866059701e-05,
358
+ "loss": 0.6656,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.06802494247890893,
363
+ "grad_norm": 7.664735661456941,
364
+ "learning_rate": 2.9746446102701606e-05,
365
+ "loss": 0.7268,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.06935876488045617,
370
+ "grad_norm": 2.926294421836929,
371
+ "learning_rate": 2.9734689337657157e-05,
372
+ "loss": 0.7498,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.07069258728200341,
377
+ "grad_norm": 8.95556138792813,
378
+ "learning_rate": 2.9722668576105703e-05,
379
+ "loss": 0.7237,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.07202640968355063,
384
+ "grad_norm": 2.725769364825004,
385
+ "learning_rate": 2.971038403341921e-05,
386
+ "loss": 0.701,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.07336023208509787,
391
+ "grad_norm": 2.458130413236654,
392
+ "learning_rate": 2.9697835929695727e-05,
393
+ "loss": 0.6775,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.07469405448664511,
398
+ "grad_norm": 1.908510150241819,
399
+ "learning_rate": 2.968502448975544e-05,
400
+ "loss": 0.7042,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.07602787688819233,
405
+ "grad_norm": 26.340984665432142,
406
+ "learning_rate": 2.967194994313663e-05,
407
+ "loss": 0.7234,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.07736169928973957,
412
+ "grad_norm": 5.677162249188698,
413
+ "learning_rate": 2.9658612524091594e-05,
414
+ "loss": 0.7159,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.07869552169128681,
419
+ "grad_norm": 1.7975075976512287,
420
+ "learning_rate": 2.9645012471582406e-05,
421
+ "loss": 0.7268,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.08002934409283403,
426
+ "grad_norm": 3.031655038731278,
427
+ "learning_rate": 2.9631150029276662e-05,
428
+ "loss": 0.6904,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.08136316649438127,
433
+ "grad_norm": 13.019769162242698,
434
+ "learning_rate": 2.9617025445543114e-05,
435
+ "loss": 0.683,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.08269698889592851,
440
+ "grad_norm": 5.982097939523946,
441
+ "learning_rate": 2.9602638973447218e-05,
442
+ "loss": 0.7511,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.08403081129747574,
447
+ "grad_norm": 2.7343884277014054,
448
+ "learning_rate": 2.9587990870746574e-05,
449
+ "loss": 0.7114,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.08536463369902297,
454
+ "grad_norm": 5.586409371459215,
455
+ "learning_rate": 2.9573081399886356e-05,
456
+ "loss": 0.6907,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.08669845610057021,
461
+ "grad_norm": 9.87255558712134,
462
+ "learning_rate": 2.9557910827994568e-05,
463
+ "loss": 0.7319,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.08803227850211744,
468
+ "grad_norm": 3.9867937712251664,
469
+ "learning_rate": 2.9542479426877283e-05,
470
+ "loss": 0.6759,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.08936610090366467,
475
+ "grad_norm": 3.111751395898768,
476
+ "learning_rate": 2.9526787473013753e-05,
477
+ "loss": 0.7086,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.09069992330521191,
482
+ "grad_norm": 3.0578210081080535,
483
+ "learning_rate": 2.9510835247551485e-05,
484
+ "loss": 0.7209,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.09203374570675915,
489
+ "grad_norm": 16.575220829800145,
490
+ "learning_rate": 2.949462303630116e-05,
491
+ "loss": 0.7125,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.09336756810830638,
496
+ "grad_norm": 1.8236956939037314,
497
+ "learning_rate": 2.9478151129731567e-05,
498
+ "loss": 0.6591,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.09470139050985361,
503
+ "grad_norm": 1.9746052093914703,
504
+ "learning_rate": 2.9461419822964348e-05,
505
+ "loss": 0.7048,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.09603521291140085,
510
+ "grad_norm": 5.9710918845264,
511
+ "learning_rate": 2.9444429415768726e-05,
512
+ "loss": 0.725,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.09736903531294808,
517
+ "grad_norm": 9.961068856364284,
518
+ "learning_rate": 2.942718021255617e-05,
519
+ "loss": 0.7155,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.09870285771449532,
524
+ "grad_norm": 9.195914431100427,
525
+ "learning_rate": 2.940967252237488e-05,
526
+ "loss": 0.7171,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.10003668011604255,
531
+ "grad_norm": 5.972259927477824,
532
+ "learning_rate": 2.9391906658904296e-05,
533
+ "loss": 0.7088,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.10137050251758978,
538
+ "grad_norm": 3.0581670197906226,
539
+ "learning_rate": 2.937388294044946e-05,
540
+ "loss": 0.6645,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.10270432491913702,
545
+ "grad_norm": 3.2614276687575945,
546
+ "learning_rate": 2.9355601689935315e-05,
547
+ "loss": 0.6997,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.10403814732068425,
552
+ "grad_norm": 3.212661762840432,
553
+ "learning_rate": 2.933706323490092e-05,
554
+ "loss": 0.7005,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.10537196972223148,
559
+ "grad_norm": 10.732038885056795,
560
+ "learning_rate": 2.9318267907493583e-05,
561
+ "loss": 0.6588,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.10670579212377872,
566
+ "grad_norm": 2.495380329507228,
567
+ "learning_rate": 2.9299216044462903e-05,
568
+ "loss": 0.6965,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.10803961452532596,
573
+ "grad_norm": 4.60295956813534,
574
+ "learning_rate": 2.927990798715475e-05,
575
+ "loss": 0.6959,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.1093734369268732,
580
+ "grad_norm": 5.95010483593304,
581
+ "learning_rate": 2.926034408150513e-05,
582
+ "loss": 0.7083,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.11070725932842042,
587
+ "grad_norm": 10.48663715000287,
588
+ "learning_rate": 2.9240524678034016e-05,
589
+ "loss": 0.6977,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.11204108172996766,
594
+ "grad_norm": 3.159424497856185,
595
+ "learning_rate": 2.9220450131839037e-05,
596
+ "loss": 0.7129,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.1133749041315149,
601
+ "grad_norm": 3.984530217756339,
602
+ "learning_rate": 2.920012080258912e-05,
603
+ "loss": 0.6618,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.11470872653306212,
608
+ "grad_norm": 4.212436954015185,
609
+ "learning_rate": 2.9179537054518085e-05,
610
+ "loss": 0.6958,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.11604254893460936,
615
+ "grad_norm": 2.5397356754848657,
616
+ "learning_rate": 2.9158699256418056e-05,
617
+ "loss": 0.6593,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.1173763713361566,
622
+ "grad_norm": 7.8192219616904115,
623
+ "learning_rate": 2.9137607781632913e-05,
624
+ "loss": 0.6932,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.11871019373770382,
629
+ "grad_norm": 3.890974772100691,
630
+ "learning_rate": 2.911626300805155e-05,
631
+ "loss": 0.674,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.12004401613925106,
636
+ "grad_norm": 2.241784461612369,
637
+ "learning_rate": 2.9094665318101155e-05,
638
+ "loss": 0.6808,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.1213778385407983,
643
+ "grad_norm": 2.8295407624221456,
644
+ "learning_rate": 2.9072815098740326e-05,
645
+ "loss": 0.7074,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.12271166094234552,
650
+ "grad_norm": 2.8191065019379455,
651
+ "learning_rate": 2.9050712741452136e-05,
652
+ "loss": 0.7218,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.12404548334389276,
657
+ "grad_norm": 2.1348105213857362,
658
+ "learning_rate": 2.902835864223715e-05,
659
+ "loss": 0.6901,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.12537930574543998,
664
+ "grad_norm": 2.547712785034466,
665
+ "learning_rate": 2.9005753201606287e-05,
666
+ "loss": 0.7024,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.12671312814698724,
671
+ "grad_norm": 6.0714243848770195,
672
+ "learning_rate": 2.8982896824573678e-05,
673
+ "loss": 0.665,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.12804695054853446,
678
+ "grad_norm": 2.928297684665459,
679
+ "learning_rate": 2.8959789920649394e-05,
680
+ "loss": 0.7178,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.12938077295008168,
685
+ "grad_norm": 9.456759373673547,
686
+ "learning_rate": 2.893643290383212e-05,
687
+ "loss": 0.6759,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.13071459535162894,
692
+ "grad_norm": 3.9700721280634648,
693
+ "learning_rate": 2.891282619260172e-05,
694
+ "loss": 0.6701,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.13204841775317616,
699
+ "grad_norm": 1.9216628539066036,
700
+ "learning_rate": 2.8888970209911754e-05,
701
+ "loss": 0.6725,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.13338224015472339,
706
+ "grad_norm": 6.51104150647371,
707
+ "learning_rate": 2.8864865383181893e-05,
708
+ "loss": 0.6811,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.13471606255627064,
713
+ "grad_norm": 7.1765693144596705,
714
+ "learning_rate": 2.8840512144290273e-05,
715
+ "loss": 0.7085,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.13604988495781786,
720
+ "grad_norm": 4.136786758396937,
721
+ "learning_rate": 2.8815910929565734e-05,
722
+ "loss": 0.7128,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.1373837073593651,
727
+ "grad_norm": 3.407893493053672,
728
+ "learning_rate": 2.879106217978002e-05,
729
+ "loss": 0.6782,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.13871752976091234,
734
+ "grad_norm": 2.5759617833329544,
735
+ "learning_rate": 2.8765966340139892e-05,
736
+ "loss": 0.7387,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.14005135216245956,
741
+ "grad_norm": 8.496713788438253,
742
+ "learning_rate": 2.8740623860279116e-05,
743
+ "loss": 0.6988,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.14138517456400682,
748
+ "grad_norm": 1.438197837288145,
749
+ "learning_rate": 2.871503519425044e-05,
750
+ "loss": 0.6797,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.14271899696555404,
755
+ "grad_norm": 3.305947931404651,
756
+ "learning_rate": 2.8689200800517448e-05,
757
+ "loss": 0.727,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.14405281936710126,
762
+ "grad_norm": 4.384984204254061,
763
+ "learning_rate": 2.866312114194634e-05,
764
+ "loss": 0.7073,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.14538664176864852,
769
+ "grad_norm": 3.0424755710445455,
770
+ "learning_rate": 2.8636796685797657e-05,
771
+ "loss": 0.6712,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.14672046417019574,
776
+ "grad_norm": 3.7562374375017415,
777
+ "learning_rate": 2.8610227903717876e-05,
778
+ "loss": 0.6811,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.14805428657174297,
783
+ "grad_norm": 18.148320631618972,
784
+ "learning_rate": 2.8583415271730994e-05,
785
+ "loss": 0.669,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.14938810897329022,
790
+ "grad_norm": 10.095266883651389,
791
+ "learning_rate": 2.855635927022998e-05,
792
+ "loss": 0.7034,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.15072193137483744,
797
+ "grad_norm": 6.298167661594497,
798
+ "learning_rate": 2.8529060383968175e-05,
799
+ "loss": 0.6736,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.15205575377638467,
804
+ "grad_norm": 1.7112633381805116,
805
+ "learning_rate": 2.850151910205061e-05,
806
+ "loss": 0.7072,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.15338957617793192,
811
+ "grad_norm": 4.478017946548926,
812
+ "learning_rate": 2.847373591792523e-05,
813
+ "loss": 0.679,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.15472339857947914,
818
+ "grad_norm": 6.208223721690807,
819
+ "learning_rate": 2.844571132937407e-05,
820
+ "loss": 0.688,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.15605722098102637,
825
+ "grad_norm": 3.5734211349612974,
826
+ "learning_rate": 2.841744583850431e-05,
827
+ "loss": 0.7054,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.15739104338257362,
832
+ "grad_norm": 2.9418941665585874,
833
+ "learning_rate": 2.838893995173932e-05,
834
+ "loss": 0.7051,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.15872486578412084,
839
+ "grad_norm": 2.8796543096657055,
840
+ "learning_rate": 2.836019417980955e-05,
841
+ "loss": 0.6856,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.16005868818566807,
846
+ "grad_norm": 1.3218998593276705,
847
+ "learning_rate": 2.8331209037743387e-05,
848
+ "loss": 0.69,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.16139251058721532,
853
+ "grad_norm": 1.566163155963456,
854
+ "learning_rate": 2.8301985044857947e-05,
855
+ "loss": 0.6384,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.16272633298876255,
860
+ "grad_norm": 6.731577281026319,
861
+ "learning_rate": 2.8272522724749743e-05,
862
+ "loss": 0.7241,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.16406015539030977,
867
+ "grad_norm": 3.3521245738531813,
868
+ "learning_rate": 2.8242822605285323e-05,
869
+ "loss": 0.6931,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.16539397779185702,
874
+ "grad_norm": 11.826633616347152,
875
+ "learning_rate": 2.8212885218591812e-05,
876
+ "loss": 0.702,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.16672780019340425,
881
+ "grad_norm": 1.826627133990473,
882
+ "learning_rate": 2.8182711101047362e-05,
883
+ "loss": 0.7061,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.16806162259495147,
888
+ "grad_norm": 2.573810458184169,
889
+ "learning_rate": 2.815230079327156e-05,
890
+ "loss": 0.7369,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.16939544499649872,
895
+ "grad_norm": 3.2839708945300665,
896
+ "learning_rate": 2.8121654840115734e-05,
897
+ "loss": 0.6966,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.17072926739804595,
902
+ "grad_norm": 2.3799212816187136,
903
+ "learning_rate": 2.809077379065319e-05,
904
+ "loss": 0.6586,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.17206308979959317,
909
+ "grad_norm": 8.449506341181594,
910
+ "learning_rate": 2.805965819816937e-05,
911
+ "loss": 0.6759,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.17339691220114042,
916
+ "grad_norm": 19.59186920922371,
917
+ "learning_rate": 2.802830862015196e-05,
918
+ "loss": 0.6537,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.17473073460268765,
923
+ "grad_norm": 1.9822479021376653,
924
+ "learning_rate": 2.799672561828087e-05,
925
+ "loss": 0.6854,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.17606455700423487,
930
+ "grad_norm": 34.5695029919346,
931
+ "learning_rate": 2.79649097584182e-05,
932
+ "loss": 0.696,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.17739837940578213,
937
+ "grad_norm": 1.886275024045834,
938
+ "learning_rate": 2.7932861610598077e-05,
939
+ "loss": 0.7308,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.17873220180732935,
944
+ "grad_norm": 12.128310547971918,
945
+ "learning_rate": 2.7900581749016466e-05,
946
+ "loss": 0.6775,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.1800660242088766,
951
+ "grad_norm": 5.164562576250181,
952
+ "learning_rate": 2.7868070752020865e-05,
953
+ "loss": 0.7096,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.18139984661042383,
958
+ "grad_norm": 10.603161437547913,
959
+ "learning_rate": 2.7835329202099944e-05,
960
+ "loss": 0.7128,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.18273366901197105,
965
+ "grad_norm": 4.184026615605208,
966
+ "learning_rate": 2.7802357685873117e-05,
967
+ "loss": 0.6958,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.1840674914135183,
972
+ "grad_norm": 4.135408851494439,
973
+ "learning_rate": 2.7769156794080033e-05,
974
+ "loss": 0.6933,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.18540131381506553,
979
+ "grad_norm": 1.7535535971850658,
980
+ "learning_rate": 2.7735727121569967e-05,
981
+ "loss": 0.7088,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.18673513621661275,
986
+ "grad_norm": 5.354882317397714,
987
+ "learning_rate": 2.770206926729121e-05,
988
+ "loss": 0.6777,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.18806895861816,
993
+ "grad_norm": 3.2593978573810523,
994
+ "learning_rate": 2.7668183834280284e-05,
995
+ "loss": 0.6535,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.18940278101970723,
1000
+ "grad_norm": 2.48929554913285,
1001
+ "learning_rate": 2.763407142965117e-05,
1002
+ "loss": 0.6853,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.19073660342125445,
1007
+ "grad_norm": 7.550246976134938,
1008
+ "learning_rate": 2.759973266458444e-05,
1009
+ "loss": 0.679,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1920704258228017,
1014
+ "grad_norm": 5.033109901639584,
1015
+ "learning_rate": 2.756516815431627e-05,
1016
+ "loss": 0.6875,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.19340424822434893,
1021
+ "grad_norm": 6.1146492204468545,
1022
+ "learning_rate": 2.7530378518127445e-05,
1023
+ "loss": 0.6762,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.19473807062589615,
1028
+ "grad_norm": 72.48946246157955,
1029
+ "learning_rate": 2.7495364379332256e-05,
1030
+ "loss": 0.6766,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.1960718930274434,
1035
+ "grad_norm": 2.9676182245937426,
1036
+ "learning_rate": 2.7460126365267335e-05,
1037
+ "loss": 0.6731,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.19740571542899063,
1042
+ "grad_norm": 2.31300235137622,
1043
+ "learning_rate": 2.7424665107280402e-05,
1044
+ "loss": 0.6936,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.19873953783053785,
1049
+ "grad_norm": 2.594823775876976,
1050
+ "learning_rate": 2.738898124071898e-05,
1051
+ "loss": 0.6703,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.2000733602320851,
1056
+ "grad_norm": 2.8354629666044575,
1057
+ "learning_rate": 2.735307540491898e-05,
1058
+ "loss": 0.7022,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.20140718263363233,
1063
+ "grad_norm": 4.553033073507926,
1064
+ "learning_rate": 2.7316948243193273e-05,
1065
+ "loss": 0.6948,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.20274100503517956,
1070
+ "grad_norm": 10.052746707722667,
1071
+ "learning_rate": 2.7280600402820146e-05,
1072
+ "loss": 0.7216,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.2040748274367268,
1077
+ "grad_norm": 2.3783106567256977,
1078
+ "learning_rate": 2.724403253503171e-05,
1079
+ "loss": 0.6955,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.20540864983827403,
1084
+ "grad_norm": 2.26171864458522,
1085
+ "learning_rate": 2.7207245295002242e-05,
1086
+ "loss": 0.7062,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.20674247223982126,
1091
+ "grad_norm": 3.223919319241339,
1092
+ "learning_rate": 2.7170239341836436e-05,
1093
+ "loss": 0.6659,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.2080762946413685,
1098
+ "grad_norm": 1.6810934171118763,
1099
+ "learning_rate": 2.7133015338557585e-05,
1100
+ "loss": 0.7069,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.20941011704291573,
1105
+ "grad_norm": 2.1432477935562515,
1106
+ "learning_rate": 2.7095573952095727e-05,
1107
+ "loss": 0.703,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.21074393944446296,
1112
+ "grad_norm": 1.5654642597457342,
1113
+ "learning_rate": 2.705791585327568e-05,
1114
+ "loss": 0.7037,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.2120777618460102,
1119
+ "grad_norm": 2.5797486712147992,
1120
+ "learning_rate": 2.7020041716805014e-05,
1121
+ "loss": 0.6945,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.21341158424755743,
1126
+ "grad_norm": 6.993511462468879,
1127
+ "learning_rate": 2.6981952221261986e-05,
1128
+ "loss": 0.6965,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.21474540664910466,
1133
+ "grad_norm": 1.9963649259483505,
1134
+ "learning_rate": 2.6943648049083366e-05,
1135
+ "loss": 0.7052,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.2160792290506519,
1140
+ "grad_norm": 13.563069274486924,
1141
+ "learning_rate": 2.6905129886552208e-05,
1142
+ "loss": 0.6925,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.21741305145219914,
1147
+ "grad_norm": 1.8811719402961335,
1148
+ "learning_rate": 2.6866398423785568e-05,
1149
+ "loss": 0.6739,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.2187468738537464,
1154
+ "grad_norm": 3.400454288874876,
1155
+ "learning_rate": 2.682745435472212e-05,
1156
+ "loss": 0.7115,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.2200806962552936,
1161
+ "grad_norm": 1.7511437629316988,
1162
+ "learning_rate": 2.6788298377109748e-05,
1163
+ "loss": 0.6939,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.22141451865684084,
1168
+ "grad_norm": 7.03338780434775,
1169
+ "learning_rate": 2.6748931192493017e-05,
1170
+ "loss": 0.7081,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.2227483410583881,
1175
+ "grad_norm": 7.485978307196749,
1176
+ "learning_rate": 2.670935350620063e-05,
1177
+ "loss": 0.7483,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.2240821634599353,
1182
+ "grad_norm": 11.472423495255576,
1183
+ "learning_rate": 2.6669566027332767e-05,
1184
+ "loss": 0.6653,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.22541598586148254,
1189
+ "grad_norm": 2.8967622506628388,
1190
+ "learning_rate": 2.6629569468748404e-05,
1191
+ "loss": 0.6994,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.2267498082630298,
1196
+ "grad_norm": 3.689173173355099,
1197
+ "learning_rate": 2.658936454705251e-05,
1198
+ "loss": 0.7121,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.22808363066457701,
1203
+ "grad_norm": 2.5714663434660414,
1204
+ "learning_rate": 2.6548951982583246e-05,
1205
+ "loss": 0.679,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.22941745306612424,
1210
+ "grad_norm": 2.6005673266384854,
1211
+ "learning_rate": 2.650833249939903e-05,
1212
+ "loss": 0.6826,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.2307512754676715,
1217
+ "grad_norm": 25.941043620884358,
1218
+ "learning_rate": 2.6467506825265573e-05,
1219
+ "loss": 0.7202,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.23208509786921871,
1224
+ "grad_norm": 2.9541333611408396,
1225
+ "learning_rate": 2.642647569164284e-05,
1226
+ "loss": 0.6855,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.23341892027076594,
1231
+ "grad_norm": 1.6808077602808311,
1232
+ "learning_rate": 2.638523983367194e-05,
1233
+ "loss": 0.6901,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.2347527426723132,
1238
+ "grad_norm": 8.145841190370119,
1239
+ "learning_rate": 2.634379999016198e-05,
1240
+ "loss": 0.7,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.23608656507386042,
1245
+ "grad_norm": 17.930965708795497,
1246
+ "learning_rate": 2.6302156903576784e-05,
1247
+ "loss": 0.6315,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.23742038747540764,
1252
+ "grad_norm": 3.8449192904768186,
1253
+ "learning_rate": 2.6260311320021628e-05,
1254
+ "loss": 0.6369,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.2387542098769549,
1259
+ "grad_norm": 1.7681330868700245,
1260
+ "learning_rate": 2.6218263989229855e-05,
1261
+ "loss": 0.674,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.24008803227850212,
1266
+ "grad_norm": 2.1491823327138317,
1267
+ "learning_rate": 2.617601566454944e-05,
1268
+ "loss": 0.6979,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.24142185468004934,
1273
+ "grad_norm": 4.667708666590814,
1274
+ "learning_rate": 2.613356710292951e-05,
1275
+ "loss": 0.6932,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.2427556770815966,
1280
+ "grad_norm": 2.70251113889306,
1281
+ "learning_rate": 2.6090919064906766e-05,
1282
+ "loss": 0.6842,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.24408949948314382,
1287
+ "grad_norm": 4.187423249865094,
1288
+ "learning_rate": 2.6048072314591854e-05,
1289
+ "loss": 0.6642,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.24542332188469104,
1294
+ "grad_norm": 1.6832068899399832,
1295
+ "learning_rate": 2.600502761965569e-05,
1296
+ "loss": 0.6984,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.2467571442862383,
1301
+ "grad_norm": 5.37697139339205,
1302
+ "learning_rate": 2.59617857513157e-05,
1303
+ "loss": 0.6489,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.24809096668778552,
1308
+ "grad_norm": 2.6302546632419186,
1309
+ "learning_rate": 2.591834748432198e-05,
1310
+ "loss": 0.6923,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.24942478908933274,
1315
+ "grad_norm": 3.8259011796471927,
1316
+ "learning_rate": 2.5874713596943465e-05,
1317
+ "loss": 0.7163,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.25075861149087997,
1322
+ "grad_norm": 5.322548910178761,
1323
+ "learning_rate": 2.5830884870953933e-05,
1324
+ "loss": 0.7188,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.25209243389242725,
1329
+ "grad_norm": 1.5202627897221725,
1330
+ "learning_rate": 2.578686209161803e-05,
1331
+ "loss": 0.6975,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.2534262562939745,
1336
+ "grad_norm": 2.7405063442524966,
1337
+ "learning_rate": 2.5742646047677186e-05,
1338
+ "loss": 0.7257,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.2547600786955217,
1343
+ "grad_norm": 3.7267206506339225,
1344
+ "learning_rate": 2.5698237531335493e-05,
1345
+ "loss": 0.7078,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.2560939010970689,
1350
+ "grad_norm": 1.7323421214775496,
1351
+ "learning_rate": 2.56536373382455e-05,
1352
+ "loss": 0.6739,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.25742772349861615,
1357
+ "grad_norm": 5.812235836466133,
1358
+ "learning_rate": 2.5608846267493974e-05,
1359
+ "loss": 0.668,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.25876154590016337,
1364
+ "grad_norm": 5.482474929884452,
1365
+ "learning_rate": 2.5563865121587563e-05,
1366
+ "loss": 0.6633,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.26009536830171065,
1371
+ "grad_norm": 10.990988768370446,
1372
+ "learning_rate": 2.5518694706438445e-05,
1373
+ "loss": 0.7351,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.2614291907032579,
1378
+ "grad_norm": 5.066143279063377,
1379
+ "learning_rate": 2.5473335831349842e-05,
1380
+ "loss": 0.6479,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.2627630131048051,
1385
+ "grad_norm": 1.8538054067880405,
1386
+ "learning_rate": 2.5427789309001577e-05,
1387
+ "loss": 0.691,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.2640968355063523,
1392
+ "grad_norm": 3.642801968239499,
1393
+ "learning_rate": 2.538205595543548e-05,
1394
+ "loss": 0.6981,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.26543065790789955,
1399
+ "grad_norm": 4.027276498286695,
1400
+ "learning_rate": 2.5336136590040767e-05,
1401
+ "loss": 0.6471,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.26676448030944677,
1406
+ "grad_norm": 5.225146913060103,
1407
+ "learning_rate": 2.529003203553937e-05,
1408
+ "loss": 0.6768,
1409
+ "step": 2000
1410
+ },
1411
+ {
1412
+ "epoch": 0.26809830271099405,
1413
+ "grad_norm": 3.253526388391669,
1414
+ "learning_rate": 2.5243743117971186e-05,
1415
+ "loss": 0.7292,
1416
+ "step": 2010
1417
+ },
1418
+ {
1419
+ "epoch": 0.2694321251125413,
1420
+ "grad_norm": 3.595165338865758,
1421
+ "learning_rate": 2.5197270666679295e-05,
1422
+ "loss": 0.7153,
1423
+ "step": 2020
1424
+ },
1425
+ {
1426
+ "epoch": 0.2707659475140885,
1427
+ "grad_norm": 3.041241411415701,
1428
+ "learning_rate": 2.515061551429509e-05,
1429
+ "loss": 0.6911,
1430
+ "step": 2030
1431
+ },
1432
+ {
1433
+ "epoch": 0.2720997699156357,
1434
+ "grad_norm": 1.8998842731166448,
1435
+ "learning_rate": 2.5103778496723334e-05,
1436
+ "loss": 0.7129,
1437
+ "step": 2040
1438
+ },
1439
+ {
1440
+ "epoch": 0.27343359231718295,
1441
+ "grad_norm": 2.421232470526233,
1442
+ "learning_rate": 2.5056760453127242e-05,
1443
+ "loss": 0.7009,
1444
+ "step": 2050
1445
+ },
1446
+ {
1447
+ "epoch": 0.2747674147187302,
1448
+ "grad_norm": 1.6780177844017854,
1449
+ "learning_rate": 2.5009562225913385e-05,
1450
+ "loss": 0.6772,
1451
+ "step": 2060
1452
+ },
1453
+ {
1454
+ "epoch": 0.27610123712027745,
1455
+ "grad_norm": 4.807962830195143,
1456
+ "learning_rate": 2.4962184660716645e-05,
1457
+ "loss": 0.7037,
1458
+ "step": 2070
1459
+ },
1460
+ {
1461
+ "epoch": 0.2774350595218247,
1462
+ "grad_norm": 2.9514583345109253,
1463
+ "learning_rate": 2.4914628606385022e-05,
1464
+ "loss": 0.7142,
1465
+ "step": 2080
1466
+ },
1467
+ {
1468
+ "epoch": 0.2787688819233719,
1469
+ "grad_norm": 7.40404674094443,
1470
+ "learning_rate": 2.4866894914964462e-05,
1471
+ "loss": 0.6811,
1472
+ "step": 2090
1473
+ },
1474
+ {
1475
+ "epoch": 0.2801027043249191,
1476
+ "grad_norm": 2.925771960255195,
1477
+ "learning_rate": 2.481898444168357e-05,
1478
+ "loss": 0.6545,
1479
+ "step": 2100
1480
+ },
1481
+ {
1482
+ "epoch": 0.28143652672646635,
1483
+ "grad_norm": 5.444798449384416,
1484
+ "learning_rate": 2.4770898044938284e-05,
1485
+ "loss": 0.7043,
1486
+ "step": 2110
1487
+ },
1488
+ {
1489
+ "epoch": 0.28277034912801363,
1490
+ "grad_norm": 9.858110881841988,
1491
+ "learning_rate": 2.4722636586276522e-05,
1492
+ "loss": 0.665,
1493
+ "step": 2120
1494
+ },
1495
+ {
1496
+ "epoch": 0.28410417152956086,
1497
+ "grad_norm": 5.298660202543338,
1498
+ "learning_rate": 2.4674200930382712e-05,
1499
+ "loss": 0.6881,
1500
+ "step": 2130
1501
+ },
1502
+ {
1503
+ "epoch": 0.2854379939311081,
1504
+ "grad_norm": 3.705317238676521,
1505
+ "learning_rate": 2.4625591945062326e-05,
1506
+ "loss": 0.7149,
1507
+ "step": 2140
1508
+ },
1509
+ {
1510
+ "epoch": 0.2867718163326553,
1511
+ "grad_norm": 3.4572860020453855,
1512
+ "learning_rate": 2.4576810501226318e-05,
1513
+ "loss": 0.6714,
1514
+ "step": 2150
1515
+ },
1516
+ {
1517
+ "epoch": 0.28810563873420253,
1518
+ "grad_norm": 2.603696480584208,
1519
+ "learning_rate": 2.4527857472875515e-05,
1520
+ "loss": 0.6706,
1521
+ "step": 2160
1522
+ },
1523
+ {
1524
+ "epoch": 0.28943946113574975,
1525
+ "grad_norm": 3.6276512973605524,
1526
+ "learning_rate": 2.447873373708498e-05,
1527
+ "loss": 0.6915,
1528
+ "step": 2170
1529
+ },
1530
+ {
1531
+ "epoch": 0.29077328353729703,
1532
+ "grad_norm": 3.2247665616500414,
1533
+ "learning_rate": 2.4429440173988275e-05,
1534
+ "loss": 0.698,
1535
+ "step": 2180
1536
+ },
1537
+ {
1538
+ "epoch": 0.29210710593884426,
1539
+ "grad_norm": 2.902123587923174,
1540
+ "learning_rate": 2.43799776667617e-05,
1541
+ "loss": 0.6566,
1542
+ "step": 2190
1543
+ },
1544
+ {
1545
+ "epoch": 0.2934409283403915,
1546
+ "grad_norm": 2.1614031926364214,
1547
+ "learning_rate": 2.4330347101608492e-05,
1548
+ "loss": 0.7077,
1549
+ "step": 2200
1550
+ },
1551
+ {
1552
+ "epoch": 0.2947747507419387,
1553
+ "grad_norm": 9.153437367973524,
1554
+ "learning_rate": 2.428054936774289e-05,
1555
+ "loss": 0.6163,
1556
+ "step": 2210
1557
+ },
1558
+ {
1559
+ "epoch": 0.29610857314348593,
1560
+ "grad_norm": 1.9839793734516538,
1561
+ "learning_rate": 2.423058535737427e-05,
1562
+ "loss": 0.6525,
1563
+ "step": 2220
1564
+ },
1565
+ {
1566
+ "epoch": 0.29744239554503316,
1567
+ "grad_norm": 7.536682448957617,
1568
+ "learning_rate": 2.418045596569111e-05,
1569
+ "loss": 0.6892,
1570
+ "step": 2230
1571
+ },
1572
+ {
1573
+ "epoch": 0.29877621794658044,
1574
+ "grad_norm": 7.678647037780194,
1575
+ "learning_rate": 2.4130162090844976e-05,
1576
+ "loss": 0.6787,
1577
+ "step": 2240
1578
+ },
1579
+ {
1580
+ "epoch": 0.30011004034812766,
1581
+ "grad_norm": 4.791025776780419,
1582
+ "learning_rate": 2.4079704633934427e-05,
1583
+ "loss": 0.6458,
1584
+ "step": 2250
1585
+ },
1586
+ {
1587
+ "epoch": 0.3014438627496749,
1588
+ "grad_norm": 2.274268244339417,
1589
+ "learning_rate": 2.4029084498988864e-05,
1590
+ "loss": 0.6962,
1591
+ "step": 2260
1592
+ },
1593
+ {
1594
+ "epoch": 0.3027776851512221,
1595
+ "grad_norm": 2.156854296054033,
1596
+ "learning_rate": 2.3978302592952332e-05,
1597
+ "loss": 0.6852,
1598
+ "step": 2270
1599
+ },
1600
+ {
1601
+ "epoch": 0.30411150755276933,
1602
+ "grad_norm": 3.294748786762216,
1603
+ "learning_rate": 2.392735982566728e-05,
1604
+ "loss": 0.7508,
1605
+ "step": 2280
1606
+ },
1607
+ {
1608
+ "epoch": 0.30544532995431656,
1609
+ "grad_norm": 1.6409137290114766,
1610
+ "learning_rate": 2.387625710985826e-05,
1611
+ "loss": 0.7167,
1612
+ "step": 2290
1613
+ },
1614
+ {
1615
+ "epoch": 0.30677915235586384,
1616
+ "grad_norm": 3.5000955023678713,
1617
+ "learning_rate": 2.3824995361115552e-05,
1618
+ "loss": 0.6903,
1619
+ "step": 2300
1620
+ },
1621
+ {
1622
+ "epoch": 0.30811297475741106,
1623
+ "grad_norm": 1.085755861309128,
1624
+ "learning_rate": 2.3773575497878784e-05,
1625
+ "loss": 0.7182,
1626
+ "step": 2310
1627
+ },
1628
+ {
1629
+ "epoch": 0.3094467971589583,
1630
+ "grad_norm": 2.6834548847634445,
1631
+ "learning_rate": 2.372199844142048e-05,
1632
+ "loss": 0.6984,
1633
+ "step": 2320
1634
+ },
1635
+ {
1636
+ "epoch": 0.3107806195605055,
1637
+ "grad_norm": 2.493536604926798,
1638
+ "learning_rate": 2.3670265115829523e-05,
1639
+ "loss": 0.6536,
1640
+ "step": 2330
1641
+ },
1642
+ {
1643
+ "epoch": 0.31211444196205274,
1644
+ "grad_norm": 1.7721327632872044,
1645
+ "learning_rate": 2.3618376447994633e-05,
1646
+ "loss": 0.6561,
1647
+ "step": 2340
1648
+ },
1649
+ {
1650
+ "epoch": 0.31344826436359996,
1651
+ "grad_norm": 2.171596262218261,
1652
+ "learning_rate": 2.3566333367587737e-05,
1653
+ "loss": 0.7056,
1654
+ "step": 2350
1655
+ },
1656
+ {
1657
+ "epoch": 0.31478208676514724,
1658
+ "grad_norm": 36.36759112960759,
1659
+ "learning_rate": 2.3514136807047318e-05,
1660
+ "loss": 0.7015,
1661
+ "step": 2360
1662
+ },
1663
+ {
1664
+ "epoch": 0.31611590916669446,
1665
+ "grad_norm": 2.1581747337993993,
1666
+ "learning_rate": 2.3461787701561724e-05,
1667
+ "loss": 0.7136,
1668
+ "step": 2370
1669
+ },
1670
+ {
1671
+ "epoch": 0.3174497315682417,
1672
+ "grad_norm": 4.627684355473749,
1673
+ "learning_rate": 2.340928698905239e-05,
1674
+ "loss": 0.6856,
1675
+ "step": 2380
1676
+ },
1677
+ {
1678
+ "epoch": 0.3187835539697889,
1679
+ "grad_norm": 23.008362949697226,
1680
+ "learning_rate": 2.335663561015704e-05,
1681
+ "loss": 0.6535,
1682
+ "step": 2390
1683
+ },
1684
+ {
1685
+ "epoch": 0.32011737637133614,
1686
+ "grad_norm": 1.4483305067525718,
1687
+ "learning_rate": 2.3303834508212845e-05,
1688
+ "loss": 0.6929,
1689
+ "step": 2400
1690
+ },
1691
+ {
1692
+ "epoch": 0.3214511987728834,
1693
+ "grad_norm": 2.4145146906270134,
1694
+ "learning_rate": 2.325088462923951e-05,
1695
+ "loss": 0.6859,
1696
+ "step": 2410
1697
+ },
1698
+ {
1699
+ "epoch": 0.32278502117443064,
1700
+ "grad_norm": 1.6323554754616574,
1701
+ "learning_rate": 2.319778692192233e-05,
1702
+ "loss": 0.6885,
1703
+ "step": 2420
1704
+ },
1705
+ {
1706
+ "epoch": 0.32411884357597787,
1707
+ "grad_norm": 4.22840962266095,
1708
+ "learning_rate": 2.3144542337595196e-05,
1709
+ "loss": 0.6799,
1710
+ "step": 2430
1711
+ },
1712
+ {
1713
+ "epoch": 0.3254526659775251,
1714
+ "grad_norm": 9.344155395286913,
1715
+ "learning_rate": 2.3091151830223537e-05,
1716
+ "loss": 0.7125,
1717
+ "step": 2440
1718
+ },
1719
+ {
1720
+ "epoch": 0.3267864883790723,
1721
+ "grad_norm": 6.426921263330431,
1722
+ "learning_rate": 2.3037616356387237e-05,
1723
+ "loss": 0.6844,
1724
+ "step": 2450
1725
+ },
1726
+ {
1727
+ "epoch": 0.32812031078061954,
1728
+ "grad_norm": 2.508003013718542,
1729
+ "learning_rate": 2.2983936875263495e-05,
1730
+ "loss": 0.6695,
1731
+ "step": 2460
1732
+ },
1733
+ {
1734
+ "epoch": 0.3294541331821668,
1735
+ "grad_norm": 5.676502368376463,
1736
+ "learning_rate": 2.2930114348609655e-05,
1737
+ "loss": 0.656,
1738
+ "step": 2470
1739
+ },
1740
+ {
1741
+ "epoch": 0.33078795558371404,
1742
+ "grad_norm": 2.9206555111493597,
1743
+ "learning_rate": 2.2876149740745935e-05,
1744
+ "loss": 0.7357,
1745
+ "step": 2480
1746
+ },
1747
+ {
1748
+ "epoch": 0.33212177798526127,
1749
+ "grad_norm": 3.944142868858924,
1750
+ "learning_rate": 2.28220440185382e-05,
1751
+ "loss": 0.709,
1752
+ "step": 2490
1753
+ },
1754
+ {
1755
+ "epoch": 0.3334556003868085,
1756
+ "grad_norm": 15.707981311418816,
1757
+ "learning_rate": 2.2767798151380597e-05,
1758
+ "loss": 0.685,
1759
+ "step": 2500
1760
+ },
1761
+ {
1762
+ "epoch": 0.3347894227883557,
1763
+ "grad_norm": 4.850415178599164,
1764
+ "learning_rate": 2.27134131111782e-05,
1765
+ "loss": 0.7258,
1766
+ "step": 2510
1767
+ },
1768
+ {
1769
+ "epoch": 0.33612324518990294,
1770
+ "grad_norm": 1.5546940942365648,
1771
+ "learning_rate": 2.2658889872329628e-05,
1772
+ "loss": 0.6849,
1773
+ "step": 2520
1774
+ },
1775
+ {
1776
+ "epoch": 0.3374570675914502,
1777
+ "grad_norm": 7.608111830875221,
1778
+ "learning_rate": 2.2604229411709518e-05,
1779
+ "loss": 0.689,
1780
+ "step": 2530
1781
+ },
1782
+ {
1783
+ "epoch": 0.33879088999299745,
1784
+ "grad_norm": 3.5234760696769922,
1785
+ "learning_rate": 2.25494327086511e-05,
1786
+ "loss": 0.7067,
1787
+ "step": 2540
1788
+ },
1789
+ {
1790
+ "epoch": 0.34012471239454467,
1791
+ "grad_norm": 3.050573051274483,
1792
+ "learning_rate": 2.2494500744928583e-05,
1793
+ "loss": 0.6535,
1794
+ "step": 2550
1795
+ },
1796
+ {
1797
+ "epoch": 0.3414585347960919,
1798
+ "grad_norm": 1.8068660121233675,
1799
+ "learning_rate": 2.243943450473963e-05,
1800
+ "loss": 0.7171,
1801
+ "step": 2560
1802
+ },
1803
+ {
1804
+ "epoch": 0.3427923571976391,
1805
+ "grad_norm": 5.878470836242925,
1806
+ "learning_rate": 2.2384234974687658e-05,
1807
+ "loss": 0.6732,
1808
+ "step": 2570
1809
+ },
1810
+ {
1811
+ "epoch": 0.34412617959918634,
1812
+ "grad_norm": 2.419571993629191,
1813
+ "learning_rate": 2.2328903143764216e-05,
1814
+ "loss": 0.6553,
1815
+ "step": 2580
1816
+ },
1817
+ {
1818
+ "epoch": 0.3454600020007336,
1819
+ "grad_norm": 5.4457590803918166,
1820
+ "learning_rate": 2.2273440003331237e-05,
1821
+ "loss": 0.6963,
1822
+ "step": 2590
1823
+ },
1824
+ {
1825
+ "epoch": 0.34679382440228085,
1826
+ "grad_norm": 1.611198577297041,
1827
+ "learning_rate": 2.2217846547103275e-05,
1828
+ "loss": 0.7075,
1829
+ "step": 2600
1830
+ },
1831
+ {
1832
+ "epoch": 0.3481276468038281,
1833
+ "grad_norm": 1.5058788491262667,
1834
+ "learning_rate": 2.216212377112972e-05,
1835
+ "loss": 0.67,
1836
+ "step": 2610
1837
+ },
1838
+ {
1839
+ "epoch": 0.3494614692053753,
1840
+ "grad_norm": 12.998071380752632,
1841
+ "learning_rate": 2.2106272673776934e-05,
1842
+ "loss": 0.6878,
1843
+ "step": 2620
1844
+ },
1845
+ {
1846
+ "epoch": 0.3507952916069225,
1847
+ "grad_norm": 15.038084891115052,
1848
+ "learning_rate": 2.2050294255710375e-05,
1849
+ "loss": 0.7104,
1850
+ "step": 2630
1851
+ },
1852
+ {
1853
+ "epoch": 0.35212911400846975,
1854
+ "grad_norm": 2.6541518957025474,
1855
+ "learning_rate": 2.1994189519876663e-05,
1856
+ "loss": 0.6643,
1857
+ "step": 2640
1858
+ },
1859
+ {
1860
+ "epoch": 0.353462936410017,
1861
+ "grad_norm": 2.9600345573470044,
1862
+ "learning_rate": 2.19379594714856e-05,
1863
+ "loss": 0.7138,
1864
+ "step": 2650
1865
+ },
1866
+ {
1867
+ "epoch": 0.35479675881156425,
1868
+ "grad_norm": 3.6541318055900107,
1869
+ "learning_rate": 2.188160511799219e-05,
1870
+ "loss": 0.7098,
1871
+ "step": 2660
1872
+ },
1873
+ {
1874
+ "epoch": 0.3561305812131115,
1875
+ "grad_norm": 3.5267634965358887,
1876
+ "learning_rate": 2.1825127469078555e-05,
1877
+ "loss": 0.69,
1878
+ "step": 2670
1879
+ },
1880
+ {
1881
+ "epoch": 0.3574644036146587,
1882
+ "grad_norm": 4.281448471035455,
1883
+ "learning_rate": 2.1768527536635868e-05,
1884
+ "loss": 0.66,
1885
+ "step": 2680
1886
+ },
1887
+ {
1888
+ "epoch": 0.3587982260162059,
1889
+ "grad_norm": 1.8979727093671077,
1890
+ "learning_rate": 2.171180633474621e-05,
1891
+ "loss": 0.6655,
1892
+ "step": 2690
1893
+ },
1894
+ {
1895
+ "epoch": 0.3601320484177532,
1896
+ "grad_norm": 38.260538083174715,
1897
+ "learning_rate": 2.1654964879664407e-05,
1898
+ "loss": 0.6907,
1899
+ "step": 2700
1900
+ },
1901
+ {
1902
+ "epoch": 0.3614658708193004,
1903
+ "grad_norm": 4.096124324506218,
1904
+ "learning_rate": 2.1598004189799826e-05,
1905
+ "loss": 0.674,
1906
+ "step": 2710
1907
+ },
1908
+ {
1909
+ "epoch": 0.36279969322084765,
1910
+ "grad_norm": 3.481931505321646,
1911
+ "learning_rate": 2.1540925285698122e-05,
1912
+ "loss": 0.7113,
1913
+ "step": 2720
1914
+ },
1915
+ {
1916
+ "epoch": 0.3641335156223949,
1917
+ "grad_norm": 2.0982067444268684,
1918
+ "learning_rate": 2.148372919002295e-05,
1919
+ "loss": 0.696,
1920
+ "step": 2730
1921
+ },
1922
+ {
1923
+ "epoch": 0.3654673380239421,
1924
+ "grad_norm": 3.3092373308718934,
1925
+ "learning_rate": 2.142641692753765e-05,
1926
+ "loss": 0.6315,
1927
+ "step": 2740
1928
+ },
1929
+ {
1930
+ "epoch": 0.3668011604254893,
1931
+ "grad_norm": 3.186705396754706,
1932
+ "learning_rate": 2.1368989525086893e-05,
1933
+ "loss": 0.6677,
1934
+ "step": 2750
1935
+ },
1936
+ {
1937
+ "epoch": 0.3681349828270366,
1938
+ "grad_norm": 6.164987008853785,
1939
+ "learning_rate": 2.1311448011578255e-05,
1940
+ "loss": 0.6913,
1941
+ "step": 2760
1942
+ },
1943
+ {
1944
+ "epoch": 0.36946880522858383,
1945
+ "grad_norm": 1.3538856165789028,
1946
+ "learning_rate": 2.125379341796382e-05,
1947
+ "loss": 0.7127,
1948
+ "step": 2770
1949
+ },
1950
+ {
1951
+ "epoch": 0.37080262763013105,
1952
+ "grad_norm": 5.87554442135488,
1953
+ "learning_rate": 2.1196026777221684e-05,
1954
+ "loss": 0.7016,
1955
+ "step": 2780
1956
+ },
1957
+ {
1958
+ "epoch": 0.3721364500316783,
1959
+ "grad_norm": 1.7777083138299317,
1960
+ "learning_rate": 2.1138149124337448e-05,
1961
+ "loss": 0.7295,
1962
+ "step": 2790
1963
+ },
1964
+ {
1965
+ "epoch": 0.3734702724332255,
1966
+ "grad_norm": 2.385434968259032,
1967
+ "learning_rate": 2.108016149628569e-05,
1968
+ "loss": 0.6589,
1969
+ "step": 2800
1970
+ },
1971
+ {
1972
+ "epoch": 0.3748040948347727,
1973
+ "grad_norm": 6.6511179163621446,
1974
+ "learning_rate": 2.102206493201137e-05,
1975
+ "loss": 0.659,
1976
+ "step": 2810
1977
+ },
1978
+ {
1979
+ "epoch": 0.37613791723632,
1980
+ "grad_norm": 2.7897652560781876,
1981
+ "learning_rate": 2.096386047241123e-05,
1982
+ "loss": 0.6482,
1983
+ "step": 2820
1984
+ },
1985
+ {
1986
+ "epoch": 0.37747173963786723,
1987
+ "grad_norm": 2.953800669076903,
1988
+ "learning_rate": 2.0905549160315116e-05,
1989
+ "loss": 0.675,
1990
+ "step": 2830
1991
+ },
1992
+ {
1993
+ "epoch": 0.37880556203941446,
1994
+ "grad_norm": 13.84190642263868,
1995
+ "learning_rate": 2.084713204046734e-05,
1996
+ "loss": 0.6558,
1997
+ "step": 2840
1998
+ },
1999
+ {
2000
+ "epoch": 0.3801393844409617,
2001
+ "grad_norm": 4.521658802993485,
2002
+ "learning_rate": 2.078861015950793e-05,
2003
+ "loss": 0.6861,
2004
+ "step": 2850
2005
+ },
2006
+ {
2007
+ "epoch": 0.3814732068425089,
2008
+ "grad_norm": 2.5357270384377646,
2009
+ "learning_rate": 2.072998456595387e-05,
2010
+ "loss": 0.6781,
2011
+ "step": 2860
2012
+ },
2013
+ {
2014
+ "epoch": 0.38280702924405613,
2015
+ "grad_norm": 1.694744102718741,
2016
+ "learning_rate": 2.0671256310180334e-05,
2017
+ "loss": 0.6923,
2018
+ "step": 2870
2019
+ },
2020
+ {
2021
+ "epoch": 0.3841408516456034,
2022
+ "grad_norm": 2.335196387059918,
2023
+ "learning_rate": 2.0612426444401874e-05,
2024
+ "loss": 0.6473,
2025
+ "step": 2880
2026
+ },
2027
+ {
2028
+ "epoch": 0.38547467404715063,
2029
+ "grad_norm": 4.2888708297621605,
2030
+ "learning_rate": 2.0553496022653535e-05,
2031
+ "loss": 0.7116,
2032
+ "step": 2890
2033
+ },
2034
+ {
2035
+ "epoch": 0.38680849644869786,
2036
+ "grad_norm": 2.586725028631528,
2037
+ "learning_rate": 2.0494466100772006e-05,
2038
+ "loss": 0.6388,
2039
+ "step": 2900
2040
+ },
2041
+ {
2042
+ "epoch": 0.3881423188502451,
2043
+ "grad_norm": 6.7632601439437625,
2044
+ "learning_rate": 2.0435337736376677e-05,
2045
+ "loss": 0.6663,
2046
+ "step": 2910
2047
+ },
2048
+ {
2049
+ "epoch": 0.3894761412517923,
2050
+ "grad_norm": 5.0897618694848115,
2051
+ "learning_rate": 2.03761119888507e-05,
2052
+ "loss": 0.686,
2053
+ "step": 2920
2054
+ },
2055
+ {
2056
+ "epoch": 0.39080996365333953,
2057
+ "grad_norm": 3.84563964047397,
2058
+ "learning_rate": 2.031678991932201e-05,
2059
+ "loss": 0.6763,
2060
+ "step": 2930
2061
+ },
2062
+ {
2063
+ "epoch": 0.3921437860548868,
2064
+ "grad_norm": 1.3602515715842554,
2065
+ "learning_rate": 2.0257372590644314e-05,
2066
+ "loss": 0.7278,
2067
+ "step": 2940
2068
+ },
2069
+ {
2070
+ "epoch": 0.39347760845643404,
2071
+ "grad_norm": 3.0517076558378293,
2072
+ "learning_rate": 2.0197861067378044e-05,
2073
+ "loss": 0.6734,
2074
+ "step": 2950
2075
+ },
2076
+ {
2077
+ "epoch": 0.39481143085798126,
2078
+ "grad_norm": 2.0564006437177826,
2079
+ "learning_rate": 2.0138256415771275e-05,
2080
+ "loss": 0.6861,
2081
+ "step": 2960
2082
+ },
2083
+ {
2084
+ "epoch": 0.3961452532595285,
2085
+ "grad_norm": 3.0150939790113522,
2086
+ "learning_rate": 2.0078559703740654e-05,
2087
+ "loss": 0.6745,
2088
+ "step": 2970
2089
+ },
2090
+ {
2091
+ "epoch": 0.3974790756610757,
2092
+ "grad_norm": 2.3696016899226495,
2093
+ "learning_rate": 2.0018772000852216e-05,
2094
+ "loss": 0.6812,
2095
+ "step": 2980
2096
+ },
2097
+ {
2098
+ "epoch": 0.398812898062623,
2099
+ "grad_norm": 1.8364192249492444,
2100
+ "learning_rate": 1.9958894378302265e-05,
2101
+ "loss": 0.6631,
2102
+ "step": 2990
2103
+ },
2104
+ {
2105
+ "epoch": 0.4001467204641702,
2106
+ "grad_norm": 2.3337386891806595,
2107
+ "learning_rate": 1.989892790889817e-05,
2108
+ "loss": 0.7276,
2109
+ "step": 3000
2110
+ },
2111
+ {
2112
+ "epoch": 0.40148054286571744,
2113
+ "grad_norm": 2.225372210455397,
2114
+ "learning_rate": 1.9838873667039134e-05,
2115
+ "loss": 0.6782,
2116
+ "step": 3010
2117
+ },
2118
+ {
2119
+ "epoch": 0.40281436526726466,
2120
+ "grad_norm": 4.9169439894919265,
2121
+ "learning_rate": 1.9778732728696937e-05,
2122
+ "loss": 0.6787,
2123
+ "step": 3020
2124
+ },
2125
+ {
2126
+ "epoch": 0.4041481876688119,
2127
+ "grad_norm": 2.335178009370853,
2128
+ "learning_rate": 1.9718506171396694e-05,
2129
+ "loss": 0.7036,
2130
+ "step": 3030
2131
+ },
2132
+ {
2133
+ "epoch": 0.4054820100703591,
2134
+ "grad_norm": 2.2036269677969327,
2135
+ "learning_rate": 1.965819507419751e-05,
2136
+ "loss": 0.6649,
2137
+ "step": 3040
2138
+ },
2139
+ {
2140
+ "epoch": 0.4068158324719064,
2141
+ "grad_norm": 4.7578038809060885,
2142
+ "learning_rate": 1.9597800517673165e-05,
2143
+ "loss": 0.6682,
2144
+ "step": 3050
2145
+ },
2146
+ {
2147
+ "epoch": 0.4081496548734536,
2148
+ "grad_norm": 16.429386587829253,
2149
+ "learning_rate": 1.9537323583892753e-05,
2150
+ "loss": 0.6875,
2151
+ "step": 3060
2152
+ },
2153
+ {
2154
+ "epoch": 0.40948347727500084,
2155
+ "grad_norm": 2.9803066784550953,
2156
+ "learning_rate": 1.9476765356401304e-05,
2157
+ "loss": 0.7022,
2158
+ "step": 3070
2159
+ },
2160
+ {
2161
+ "epoch": 0.41081729967654806,
2162
+ "grad_norm": 1.7088579752363298,
2163
+ "learning_rate": 1.9416126920200344e-05,
2164
+ "loss": 0.7335,
2165
+ "step": 3080
2166
+ },
2167
+ {
2168
+ "epoch": 0.4121511220780953,
2169
+ "grad_norm": 1.8641120769753927,
2170
+ "learning_rate": 1.9355409361728482e-05,
2171
+ "loss": 0.7195,
2172
+ "step": 3090
2173
+ },
2174
+ {
2175
+ "epoch": 0.4134849444796425,
2176
+ "grad_norm": 2.887584179728711,
2177
+ "learning_rate": 1.9294613768841932e-05,
2178
+ "loss": 0.6694,
2179
+ "step": 3100
2180
+ },
2181
+ {
2182
+ "epoch": 0.4148187668811898,
2183
+ "grad_norm": 3.3517946816430486,
2184
+ "learning_rate": 1.9233741230795022e-05,
2185
+ "loss": 0.6763,
2186
+ "step": 3110
2187
+ },
2188
+ {
2189
+ "epoch": 0.416152589282737,
2190
+ "grad_norm": 15.627345038869116,
2191
+ "learning_rate": 1.9172792838220686e-05,
2192
+ "loss": 0.6508,
2193
+ "step": 3120
2194
+ },
2195
+ {
2196
+ "epoch": 0.41748641168428424,
2197
+ "grad_norm": 8.325170511784908,
2198
+ "learning_rate": 1.9111769683110914e-05,
2199
+ "loss": 0.6638,
2200
+ "step": 3130
2201
+ },
2202
+ {
2203
+ "epoch": 0.41882023408583147,
2204
+ "grad_norm": 2.2227223257423576,
2205
+ "learning_rate": 1.905067285879719e-05,
2206
+ "loss": 0.6702,
2207
+ "step": 3140
2208
+ },
2209
+ {
2210
+ "epoch": 0.4201540564873787,
2211
+ "grad_norm": 2.151930199856048,
2212
+ "learning_rate": 1.8989503459930908e-05,
2213
+ "loss": 0.6702,
2214
+ "step": 3150
2215
+ },
2216
+ {
2217
+ "epoch": 0.4214878788889259,
2218
+ "grad_norm": 2.0853572424146445,
2219
+ "learning_rate": 1.892826258246376e-05,
2220
+ "loss": 0.6687,
2221
+ "step": 3160
2222
+ },
2223
+ {
2224
+ "epoch": 0.4228217012904732,
2225
+ "grad_norm": 12.723619283478206,
2226
+ "learning_rate": 1.886695132362808e-05,
2227
+ "loss": 0.6736,
2228
+ "step": 3170
2229
+ },
2230
+ {
2231
+ "epoch": 0.4241555236920204,
2232
+ "grad_norm": 2.2341551772603223,
2233
+ "learning_rate": 1.8805570781917228e-05,
2234
+ "loss": 0.6979,
2235
+ "step": 3180
2236
+ },
2237
+ {
2238
+ "epoch": 0.42548934609356764,
2239
+ "grad_norm": 2.7010833438865482,
2240
+ "learning_rate": 1.8744122057065856e-05,
2241
+ "loss": 0.6589,
2242
+ "step": 3190
2243
+ },
2244
+ {
2245
+ "epoch": 0.42682316849511487,
2246
+ "grad_norm": 3.764700495141046,
2247
+ "learning_rate": 1.868260625003024e-05,
2248
+ "loss": 0.6951,
2249
+ "step": 3200
2250
+ }
2251
+ ],
2252
+ "logging_steps": 10,
2253
+ "max_steps": 7497,
2254
+ "num_input_tokens_seen": 0,
2255
+ "num_train_epochs": 1,
2256
+ "save_steps": 400,
2257
+ "stateful_callbacks": {
2258
+ "TrainerControl": {
2259
+ "args": {
2260
+ "should_epoch_stop": false,
2261
+ "should_evaluate": false,
2262
+ "should_log": false,
2263
+ "should_save": true,
2264
+ "should_training_stop": false
2265
+ },
2266
+ "attributes": {}
2267
+ }
2268
+ },
2269
+ "total_flos": 1.4579043147972608e+19,
2270
+ "train_batch_size": 4,
2271
+ "trial_name": null,
2272
+ "trial_params": null
2273
+ }
checkpoint-3200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3917773d9035785be322ba4c24f526b38823b179d4e738034f481213b6bfc50c
3
+ size 6520
checkpoint-3200/zero_to_fp32.py ADDED
@@ -0,0 +1,587 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
252
+ param_shapes = zero_model_states[0].param_shapes
253
+
254
+ # Reconstruction protocol:
255
+ #
256
+ # XXX: document this
257
+
258
+ if debug:
259
+ for i in range(world_size):
260
+ for j in range(len(fp32_flat_groups[0])):
261
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
262
+
263
+ # XXX: memory usage doubles here (zero2)
264
+ num_param_groups = len(fp32_flat_groups[0])
265
+ merged_single_partition_of_fp32_groups = []
266
+ for i in range(num_param_groups):
267
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
268
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
269
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
270
+ avail_numel = sum(
271
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
272
+
273
+ if debug:
274
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
275
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
276
+ # not asserting if there is a mismatch due to possible padding
277
+ print(f"Have {avail_numel} numels to process.")
278
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
279
+
280
+ # params
281
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
282
+ # out-of-core computing solution
283
+ total_numel = 0
284
+ total_params = 0
285
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
286
+ offset = 0
287
+ avail_numel = full_single_fp32_vector.numel()
288
+ for name, shape in shapes.items():
289
+
290
+ unpartitioned_numel = shape.numel()
291
+ total_numel += unpartitioned_numel
292
+ total_params += 1
293
+
294
+ if debug:
295
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
296
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
297
+ offset += unpartitioned_numel
298
+
299
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
300
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
301
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
302
+ # live optimizer object, so we are checking that the numbers are within the right range
303
+ align_to = 2 * world_size
304
+
305
+ def zero2_align(x):
306
+ return align_to * math.ceil(x / align_to)
307
+
308
+ if debug:
309
+ print(f"original offset={offset}, avail_numel={avail_numel}")
310
+
311
+ offset = zero2_align(offset)
312
+ avail_numel = zero2_align(avail_numel)
313
+
314
+ if debug:
315
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
316
+
317
+ # Sanity check
318
+ if offset != avail_numel:
319
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
320
+
321
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
322
+
323
+
324
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
325
+ state_dict = OrderedDict()
326
+
327
+ # buffers
328
+ buffers = zero_model_states[0].buffers
329
+ state_dict.update(buffers)
330
+ if debug:
331
+ print(f"added {len(buffers)} buffers")
332
+
333
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
334
+
335
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
336
+
337
+ # recover shared parameters
338
+ for pair in zero_model_states[0].shared_params:
339
+ if pair[1] in state_dict:
340
+ state_dict[pair[0]] = state_dict[pair[1]]
341
+
342
+ return state_dict
343
+
344
+
345
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
346
+ remainder = unpartitioned_numel % world_size
347
+ padding_numel = (world_size - remainder) if remainder else 0
348
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
349
+ return partitioned_numel, padding_numel
350
+
351
+
352
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
353
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
354
+ return
355
+
356
+ if debug:
357
+ for i in range(world_size):
358
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
359
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
360
+
361
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
362
+ wanted_params = len(frozen_param_shapes)
363
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
364
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
365
+ print(f'Frozen params: Have {avail_numel} numels to process.')
366
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
367
+
368
+ total_params = 0
369
+ total_numel = 0
370
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
371
+ total_params += 1
372
+ unpartitioned_numel = shape.numel()
373
+ total_numel += unpartitioned_numel
374
+
375
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
376
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
377
+
378
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
379
+
380
+ if debug:
381
+ print(
382
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
383
+ )
384
+
385
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
386
+
387
+
388
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
389
+ param_shapes = zero_model_states[0].param_shapes
390
+ avail_numel = fp32_flat_groups[0].numel() * world_size
391
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
392
+ # param, re-consolidating each param, while dealing with padding if any
393
+
394
+ # merge list of dicts, preserving order
395
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
396
+
397
+ if debug:
398
+ for i in range(world_size):
399
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
400
+
401
+ wanted_params = len(param_shapes)
402
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
403
+ # not asserting if there is a mismatch due to possible padding
404
+ avail_numel = fp32_flat_groups[0].numel() * world_size
405
+ print(f"Trainable params: Have {avail_numel} numels to process.")
406
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
407
+
408
+ # params
409
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
410
+ # out-of-core computing solution
411
+ offset = 0
412
+ total_numel = 0
413
+ total_params = 0
414
+ for name, shape in param_shapes.items():
415
+
416
+ unpartitioned_numel = shape.numel()
417
+ total_numel += unpartitioned_numel
418
+ total_params += 1
419
+
420
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
421
+
422
+ if debug:
423
+ print(
424
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
425
+ )
426
+
427
+ # XXX: memory usage doubles here
428
+ state_dict[name] = torch.cat(
429
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
430
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
431
+ offset += partitioned_numel
432
+
433
+ offset *= world_size
434
+
435
+ # Sanity check
436
+ if offset != avail_numel:
437
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
438
+
439
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
440
+
441
+
442
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
443
+ state_dict = OrderedDict()
444
+
445
+ # buffers
446
+ buffers = zero_model_states[0].buffers
447
+ state_dict.update(buffers)
448
+ if debug:
449
+ print(f"added {len(buffers)} buffers")
450
+
451
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
452
+
453
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
454
+
455
+ # recover shared parameters
456
+ for pair in zero_model_states[0].shared_params:
457
+ if pair[1] in state_dict:
458
+ state_dict[pair[0]] = state_dict[pair[1]]
459
+
460
+ return state_dict
461
+
462
+
463
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
464
+ """
465
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
466
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
467
+ via a model hub.
468
+
469
+ Args:
470
+ - ``checkpoint_dir``: path to the desired checkpoint folder
471
+ - ``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``
472
+
473
+ Returns:
474
+ - pytorch ``state_dict``
475
+
476
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
477
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
478
+ the checkpoint.
479
+
480
+ A typical usage might be ::
481
+
482
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
483
+ # do the training and checkpoint saving
484
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
485
+ model = model.cpu() # move to cpu
486
+ model.load_state_dict(state_dict)
487
+ # submit to model hub or save the model to share with others
488
+
489
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
490
+ application. i.e. you will need to re-initialize the deepspeed engine, since
491
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
492
+
493
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
494
+
495
+ """
496
+ if tag is None:
497
+ latest_path = os.path.join(checkpoint_dir, 'latest')
498
+ if os.path.isfile(latest_path):
499
+ with open(latest_path, 'r') as fd:
500
+ tag = fd.read().strip()
501
+ else:
502
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
503
+
504
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
505
+
506
+ if not os.path.isdir(ds_checkpoint_dir):
507
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
508
+
509
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
510
+
511
+
512
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
513
+ """
514
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
515
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
516
+
517
+ Args:
518
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
519
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
520
+ - ``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``
521
+ """
522
+
523
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
524
+ print(f"Saving fp32 state dict to {output_file}")
525
+ torch.save(state_dict, output_file)
526
+
527
+
528
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
529
+ """
530
+ 1. Put the provided model to cpu
531
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
532
+ 3. Load it into the provided model
533
+
534
+ Args:
535
+ - ``model``: the model object to update
536
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
537
+ - ``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``
538
+
539
+ Returns:
540
+ - ``model`: modified model
541
+
542
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
543
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
544
+ conveniently placed for you in the checkpoint folder.
545
+
546
+ A typical usage might be ::
547
+
548
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
549
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
550
+ # submit to model hub or save the model to share with others
551
+
552
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
553
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
554
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
555
+
556
+ """
557
+ logger.info(f"Extracting fp32 weights")
558
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
559
+
560
+ logger.info(f"Overwriting model with fp32 weights")
561
+ model = model.cpu()
562
+ model.load_state_dict(state_dict, strict=False)
563
+
564
+ return model
565
+
566
+
567
+ if __name__ == "__main__":
568
+
569
+ parser = argparse.ArgumentParser()
570
+ parser.add_argument("checkpoint_dir",
571
+ type=str,
572
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
573
+ parser.add_argument(
574
+ "output_file",
575
+ type=str,
576
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
577
+ parser.add_argument("-t",
578
+ "--tag",
579
+ type=str,
580
+ default=None,
581
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
582
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
583
+ args = parser.parse_args()
584
+
585
+ debug = args.debug
586
+
587
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
checkpoint-4800/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: Qwen/Qwen-VL-Chat
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 0.10.0
checkpoint-4800/adapter_config.json ADDED
@@ -0,0 +1,380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 64,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.25.mlp.w2",
24
+ "transformer.visual.transformer.resblocks.39.mlp.c_proj",
25
+ "transformer.visual.transformer.resblocks.21.attn.in_proj",
26
+ "transformer.visual.transformer.resblocks.44.mlp.c_proj",
27
+ "transformer.h.15.mlp.w2",
28
+ "transformer.visual.transformer.resblocks.7.mlp.c_proj",
29
+ "transformer.h.3.mlp.w2",
30
+ "transformer.visual.transformer.resblocks.23.mlp.c_fc",
31
+ "transformer.visual.transformer.resblocks.10.mlp.c_fc",
32
+ "transformer.h.20.attn.c_attn",
33
+ "transformer.h.23.mlp.c_proj",
34
+ "transformer.visual.transformer.resblocks.34.mlp.c_fc",
35
+ "transformer.h.0.attn.c_proj",
36
+ "transformer.h.9.mlp.w1",
37
+ "transformer.visual.transformer.resblocks.12.attn.out_proj",
38
+ "transformer.visual.transformer.resblocks.22.mlp.c_fc",
39
+ "transformer.visual.transformer.resblocks.23.attn.out_proj",
40
+ "transformer.h.27.mlp.c_proj",
41
+ "transformer.visual.transformer.resblocks.1.mlp.c_proj",
42
+ "transformer.h.22.attn.c_proj",
43
+ "transformer.h.25.mlp.w1",
44
+ "transformer.visual.transformer.resblocks.0.attn.in_proj",
45
+ "transformer.visual.transformer.resblocks.2.attn.in_proj",
46
+ "transformer.h.1.attn.c_proj",
47
+ "transformer.visual.transformer.resblocks.45.mlp.c_proj",
48
+ "transformer.h.29.attn.c_attn",
49
+ "transformer.visual.transformer.resblocks.43.attn.in_proj",
50
+ "transformer.visual.transformer.resblocks.46.mlp.c_proj",
51
+ "transformer.h.28.mlp.w2",
52
+ "transformer.h.20.attn.c_proj",
53
+ "transformer.h.12.attn.c_proj",
54
+ "transformer.visual.transformer.resblocks.27.attn.out_proj",
55
+ "transformer.visual.transformer.resblocks.17.mlp.c_proj",
56
+ "transformer.visual.transformer.resblocks.33.attn.out_proj",
57
+ "transformer.h.4.attn.c_proj",
58
+ "transformer.visual.transformer.resblocks.29.attn.in_proj",
59
+ "transformer.visual.transformer.resblocks.2.attn.out_proj",
60
+ "transformer.h.4.mlp.w2",
61
+ "transformer.h.30.mlp.w1",
62
+ "transformer.h.31.attn.c_attn",
63
+ "transformer.h.10.attn.c_proj",
64
+ "transformer.h.18.mlp.w2",
65
+ "transformer.visual.transformer.resblocks.12.attn.in_proj",
66
+ "transformer.h.20.mlp.c_proj",
67
+ "transformer.visual.transformer.resblocks.36.attn.out_proj",
68
+ "transformer.visual.transformer.resblocks.31.attn.out_proj",
69
+ "transformer.h.1.attn.c_attn",
70
+ "transformer.visual.transformer.resblocks.17.attn.out_proj",
71
+ "transformer.visual.transformer.resblocks.9.mlp.c_fc",
72
+ "transformer.h.22.attn.c_attn",
73
+ "transformer.h.16.attn.c_attn",
74
+ "transformer.visual.transformer.resblocks.20.mlp.c_fc",
75
+ "transformer.visual.transformer.resblocks.31.attn.in_proj",
76
+ "transformer.h.29.mlp.w1",
77
+ "transformer.h.27.attn.c_attn",
78
+ "transformer.visual.transformer.resblocks.18.mlp.c_proj",
79
+ "transformer.visual.transformer.resblocks.8.mlp.c_proj",
80
+ "transformer.visual.transformer.resblocks.44.attn.out_proj",
81
+ "transformer.h.5.attn.c_proj",
82
+ "transformer.h.20.mlp.w1",
83
+ "transformer.h.0.mlp.c_proj",
84
+ "transformer.visual.transformer.resblocks.9.attn.out_proj",
85
+ "transformer.h.2.mlp.w1",
86
+ "transformer.visual.transformer.resblocks.3.mlp.c_proj",
87
+ "transformer.visual.transformer.resblocks.30.mlp.c_fc",
88
+ "transformer.visual.transformer.resblocks.36.mlp.c_proj",
89
+ "transformer.h.17.mlp.w2",
90
+ "transformer.h.19.mlp.w2",
91
+ "transformer.visual.transformer.resblocks.30.attn.out_proj",
92
+ "transformer.visual.transformer.resblocks.40.mlp.c_proj",
93
+ "transformer.visual.transformer.resblocks.1.mlp.c_fc",
94
+ "transformer.h.10.mlp.w1",
95
+ "transformer.h.6.attn.c_attn",
96
+ "transformer.visual.transformer.resblocks.14.mlp.c_fc",
97
+ "transformer.h.24.mlp.w2",
98
+ "transformer.visual.transformer.resblocks.28.attn.out_proj",
99
+ "transformer.visual.transformer.resblocks.27.attn.in_proj",
100
+ "transformer.h.10.attn.c_attn",
101
+ "transformer.h.17.mlp.c_proj",
102
+ "transformer.visual.transformer.resblocks.0.mlp.c_fc",
103
+ "transformer.visual.transformer.resblocks.35.attn.out_proj",
104
+ "transformer.h.6.attn.c_proj",
105
+ "transformer.h.6.mlp.w2",
106
+ "transformer.visual.transformer.resblocks.21.mlp.c_fc",
107
+ "transformer.h.22.mlp.c_proj",
108
+ "transformer.h.4.mlp.w1",
109
+ "transformer.h.14.mlp.w1",
110
+ "transformer.h.2.mlp.c_proj",
111
+ "transformer.visual.transformer.resblocks.31.mlp.c_fc",
112
+ "transformer.h.28.attn.c_proj",
113
+ "transformer.visual.transformer.resblocks.19.attn.in_proj",
114
+ "transformer.h.26.attn.c_proj",
115
+ "transformer.visual.transformer.resblocks.47.mlp.c_fc",
116
+ "transformer.h.3.attn.c_proj",
117
+ "transformer.visual.transformer.resblocks.33.attn.in_proj",
118
+ "transformer.visual.transformer.resblocks.45.attn.out_proj",
119
+ "transformer.visual.transformer.resblocks.5.attn.out_proj",
120
+ "transformer.h.7.attn.c_attn",
121
+ "transformer.h.1.mlp.w1",
122
+ "transformer.h.19.mlp.c_proj",
123
+ "transformer.visual.transformer.resblocks.39.attn.in_proj",
124
+ "transformer.visual.transformer.resblocks.7.attn.out_proj",
125
+ "transformer.h.8.mlp.w1",
126
+ "transformer.h.22.mlp.w1",
127
+ "transformer.h.7.attn.c_proj",
128
+ "transformer.visual.transformer.resblocks.0.mlp.c_proj",
129
+ "transformer.h.7.mlp.w2",
130
+ "transformer.h.25.attn.c_proj",
131
+ "transformer.visual.transformer.resblocks.17.attn.in_proj",
132
+ "transformer.visual.transformer.resblocks.24.attn.out_proj",
133
+ "transformer.h.24.mlp.c_proj",
134
+ "transformer.visual.transformer.resblocks.19.mlp.c_fc",
135
+ "transformer.h.21.mlp.w2",
136
+ "transformer.h.9.mlp.w2",
137
+ "transformer.h.27.attn.c_proj",
138
+ "transformer.h.2.attn.c_proj",
139
+ "transformer.h.5.mlp.c_proj",
140
+ "transformer.visual.transformer.resblocks.18.attn.in_proj",
141
+ "transformer.h.18.attn.c_attn",
142
+ "transformer.visual.transformer.resblocks.37.attn.in_proj",
143
+ "transformer.visual.transformer.resblocks.15.attn.out_proj",
144
+ "transformer.h.9.mlp.c_proj",
145
+ "transformer.visual.transformer.resblocks.37.mlp.c_proj",
146
+ "transformer.h.17.attn.c_proj",
147
+ "transformer.visual.transformer.resblocks.4.attn.out_proj",
148
+ "transformer.h.11.mlp.w2",
149
+ "transformer.h.3.mlp.w1",
150
+ "transformer.visual.transformer.resblocks.32.attn.out_proj",
151
+ "transformer.h.30.attn.c_attn",
152
+ "transformer.h.7.mlp.c_proj",
153
+ "transformer.visual.transformer.resblocks.38.mlp.c_proj",
154
+ "transformer.visual.transformer.resblocks.22.mlp.c_proj",
155
+ "transformer.h.23.mlp.w2",
156
+ "transformer.visual.transformer.resblocks.8.mlp.c_fc",
157
+ "transformer.h.24.attn.c_attn",
158
+ "transformer.visual.transformer.resblocks.21.attn.out_proj",
159
+ "transformer.visual.transformer.resblocks.16.mlp.c_fc",
160
+ "transformer.visual.transformer.resblocks.30.mlp.c_proj",
161
+ "transformer.visual.transformer.resblocks.29.attn.out_proj",
162
+ "transformer.visual.transformer.resblocks.19.attn.out_proj",
163
+ "transformer.visual.transformer.resblocks.1.attn.in_proj",
164
+ "transformer.h.27.mlp.w1",
165
+ "transformer.h.1.mlp.w2",
166
+ "transformer.h.17.attn.c_attn",
167
+ "transformer.visual.transformer.resblocks.3.mlp.c_fc",
168
+ "transformer.visual.transformer.resblocks.24.attn.in_proj",
169
+ "transformer.visual.transformer.resblocks.44.attn.in_proj",
170
+ "transformer.h.6.mlp.c_proj",
171
+ "transformer.visual.transformer.resblocks.3.attn.in_proj",
172
+ "transformer.visual.transformer.resblocks.19.mlp.c_proj",
173
+ "transformer.visual.transformer.resblocks.38.mlp.c_fc",
174
+ "transformer.h.14.mlp.w2",
175
+ "transformer.h.8.mlp.w2",
176
+ "transformer.h.19.mlp.w1",
177
+ "transformer.visual.transformer.resblocks.15.mlp.c_fc",
178
+ "transformer.h.19.attn.c_attn",
179
+ "transformer.h.13.attn.c_proj",
180
+ "transformer.visual.transformer.resblocks.26.mlp.c_fc",
181
+ "transformer.visual.transformer.resblocks.35.mlp.c_proj",
182
+ "transformer.h.15.mlp.w1",
183
+ "transformer.h.23.attn.c_attn",
184
+ "transformer.visual.transformer.resblocks.6.attn.out_proj",
185
+ "transformer.visual.transformer.resblocks.24.mlp.c_fc",
186
+ "transformer.visual.transformer.resblocks.18.attn.out_proj",
187
+ "transformer.visual.transformer.resblocks.20.attn.in_proj",
188
+ "transformer.visual.transformer.resblocks.31.mlp.c_proj",
189
+ "transformer.visual.transformer.resblocks.43.attn.out_proj",
190
+ "transformer.h.5.attn.c_attn",
191
+ "transformer.visual.transformer.resblocks.40.attn.out_proj",
192
+ "transformer.visual.transformer.resblocks.29.mlp.c_fc",
193
+ "transformer.h.12.mlp.w1",
194
+ "transformer.visual.transformer.resblocks.13.mlp.c_proj",
195
+ "transformer.visual.transformer.resblocks.43.mlp.c_proj",
196
+ "transformer.visual.transformer.resblocks.11.mlp.c_proj",
197
+ "transformer.visual.transformer.resblocks.37.attn.out_proj",
198
+ "transformer.visual.transformer.resblocks.40.mlp.c_fc",
199
+ "transformer.h.16.mlp.w2",
200
+ "transformer.h.9.attn.c_proj",
201
+ "transformer.visual.transformer.resblocks.26.attn.out_proj",
202
+ "transformer.visual.transformer.resblocks.4.attn.in_proj",
203
+ "transformer.visual.transformer.resblocks.10.attn.out_proj",
204
+ "transformer.h.13.mlp.c_proj",
205
+ "transformer.visual.transformer.resblocks.32.attn.in_proj",
206
+ "transformer.h.25.mlp.c_proj",
207
+ "transformer.h.24.mlp.w1",
208
+ "transformer.h.4.mlp.c_proj",
209
+ "transformer.h.21.attn.c_attn",
210
+ "transformer.visual.transformer.resblocks.37.mlp.c_fc",
211
+ "transformer.h.11.attn.c_proj",
212
+ "transformer.h.30.mlp.c_proj",
213
+ "transformer.visual.transformer.resblocks.23.mlp.c_proj",
214
+ "transformer.visual.transformer.resblocks.32.mlp.c_proj",
215
+ "transformer.h.21.attn.c_proj",
216
+ "transformer.h.3.attn.c_attn",
217
+ "transformer.visual.transformer.resblocks.5.mlp.c_proj",
218
+ "transformer.visual.transformer.resblocks.12.mlp.c_fc",
219
+ "transformer.visual.transformer.resblocks.45.attn.in_proj",
220
+ "transformer.h.4.attn.c_attn",
221
+ "transformer.visual.transformer.resblocks.33.mlp.c_proj",
222
+ "transformer.h.19.attn.c_proj",
223
+ "transformer.h.5.mlp.w1",
224
+ "transformer.visual.transformer.resblocks.14.attn.in_proj",
225
+ "transformer.h.9.attn.c_attn",
226
+ "transformer.visual.transformer.resblocks.45.mlp.c_fc",
227
+ "transformer.h.22.mlp.w2",
228
+ "transformer.visual.transformer.resblocks.41.attn.in_proj",
229
+ "transformer.h.3.mlp.c_proj",
230
+ "transformer.visual.transformer.resblocks.5.mlp.c_fc",
231
+ "transformer.h.26.attn.c_attn",
232
+ "transformer.h.30.attn.c_proj",
233
+ "transformer.visual.transformer.resblocks.10.attn.in_proj",
234
+ "transformer.visual.transformer.resblocks.24.mlp.c_proj",
235
+ "transformer.visual.transformer.resblocks.4.mlp.c_proj",
236
+ "transformer.visual.transformer.resblocks.44.mlp.c_fc",
237
+ "transformer.visual.transformer.resblocks.38.attn.in_proj",
238
+ "transformer.visual.transformer.resblocks.13.mlp.c_fc",
239
+ "transformer.visual.transformer.resblocks.8.attn.out_proj",
240
+ "transformer.visual.transformer.resblocks.23.attn.in_proj",
241
+ "transformer.visual.transformer.resblocks.41.mlp.c_proj",
242
+ "transformer.visual.transformer.resblocks.42.mlp.c_fc",
243
+ "transformer.visual.transformer.resblocks.39.mlp.c_fc",
244
+ "transformer.h.17.mlp.w1",
245
+ "transformer.visual.transformer.resblocks.42.attn.out_proj",
246
+ "transformer.visual.transformer.resblocks.8.attn.in_proj",
247
+ "transformer.h.25.attn.c_attn",
248
+ "transformer.visual.transformer.resblocks.1.attn.out_proj",
249
+ "transformer.h.6.mlp.w1",
250
+ "transformer.visual.transformer.resblocks.46.attn.out_proj",
251
+ "transformer.h.8.attn.c_attn",
252
+ "transformer.h.13.mlp.w2",
253
+ "transformer.visual.transformer.resblocks.21.mlp.c_proj",
254
+ "transformer.h.2.mlp.w2",
255
+ "transformer.visual.transformer.resblocks.30.attn.in_proj",
256
+ "transformer.h.7.mlp.w1",
257
+ "transformer.visual.transformer.resblocks.25.mlp.c_fc",
258
+ "transformer.h.11.mlp.c_proj",
259
+ "transformer.visual.transformer.resblocks.36.mlp.c_fc",
260
+ "transformer.visual.transformer.resblocks.42.mlp.c_proj",
261
+ "transformer.h.11.attn.c_attn",
262
+ "transformer.h.29.mlp.c_proj",
263
+ "transformer.visual.transformer.resblocks.36.attn.in_proj",
264
+ "transformer.visual.transformer.resblocks.40.attn.in_proj",
265
+ "transformer.visual.transformer.resblocks.27.mlp.c_fc",
266
+ "transformer.h.11.mlp.w1",
267
+ "transformer.visual.transformer.resblocks.39.attn.out_proj",
268
+ "transformer.visual.transformer.resblocks.10.mlp.c_proj",
269
+ "transformer.visual.transformer.resblocks.46.mlp.c_fc",
270
+ "transformer.h.20.mlp.w2",
271
+ "transformer.h.18.attn.c_proj",
272
+ "transformer.visual.transformer.resblocks.4.mlp.c_fc",
273
+ "transformer.visual.transformer.resblocks.17.mlp.c_fc",
274
+ "transformer.visual.transformer.resblocks.2.mlp.c_proj",
275
+ "transformer.visual.transformer.resblocks.6.attn.in_proj",
276
+ "transformer.visual.transformer.resblocks.11.mlp.c_fc",
277
+ "transformer.h.12.attn.c_attn",
278
+ "transformer.visual.transformer.resblocks.16.attn.in_proj",
279
+ "transformer.h.14.attn.c_attn",
280
+ "transformer.h.31.mlp.c_proj",
281
+ "transformer.visual.transformer.resblocks.16.attn.out_proj",
282
+ "transformer.h.0.mlp.w1",
283
+ "transformer.visual.transformer.resblocks.43.mlp.c_fc",
284
+ "transformer.visual.transformer.resblocks.5.attn.in_proj",
285
+ "transformer.h.13.attn.c_attn",
286
+ "transformer.visual.transformer.resblocks.41.mlp.c_fc",
287
+ "transformer.h.23.attn.c_proj",
288
+ "transformer.visual.transformer.resblocks.35.mlp.c_fc",
289
+ "transformer.visual.transformer.resblocks.12.mlp.c_proj",
290
+ "transformer.h.8.attn.c_proj",
291
+ "transformer.h.31.attn.c_proj",
292
+ "transformer.visual.transformer.resblocks.26.mlp.c_proj",
293
+ "transformer.h.21.mlp.w1",
294
+ "transformer.visual.transformer.resblocks.16.mlp.c_proj",
295
+ "transformer.h.31.mlp.w1",
296
+ "transformer.visual.transformer.resblocks.13.attn.in_proj",
297
+ "transformer.h.1.mlp.c_proj",
298
+ "transformer.h.12.mlp.c_proj",
299
+ "transformer.h.30.mlp.w2",
300
+ "transformer.h.14.attn.c_proj",
301
+ "transformer.visual.transformer.resblocks.15.mlp.c_proj",
302
+ "transformer.visual.transformer.resblocks.7.attn.in_proj",
303
+ "transformer.visual.transformer.resblocks.22.attn.out_proj",
304
+ "transformer.visual.transformer.resblocks.0.attn.out_proj",
305
+ "transformer.h.31.mlp.w2",
306
+ "transformer.visual.transformer.resblocks.27.mlp.c_proj",
307
+ "transformer.visual.transformer.resblocks.25.attn.out_proj",
308
+ "transformer.visual.transformer.resblocks.22.attn.in_proj",
309
+ "transformer.visual.transformer.resblocks.33.mlp.c_fc",
310
+ "transformer.h.13.mlp.w1",
311
+ "transformer.visual.transformer.resblocks.6.mlp.c_fc",
312
+ "transformer.visual.transformer.resblocks.11.attn.in_proj",
313
+ "transformer.visual.transformer.resblocks.9.attn.in_proj",
314
+ "transformer.visual.transformer.resblocks.20.mlp.c_proj",
315
+ "transformer.h.16.mlp.c_proj",
316
+ "transformer.visual.transformer.resblocks.15.attn.in_proj",
317
+ "transformer.h.26.mlp.w2",
318
+ "transformer.h.28.attn.c_attn",
319
+ "transformer.visual.transformer.resblocks.29.mlp.c_proj",
320
+ "transformer.h.23.mlp.w1",
321
+ "transformer.h.14.mlp.c_proj",
322
+ "transformer.visual.transformer.resblocks.25.mlp.c_proj",
323
+ "transformer.visual.transformer.resblocks.9.mlp.c_proj",
324
+ "transformer.visual.transformer.resblocks.34.attn.in_proj",
325
+ "transformer.h.5.mlp.w2",
326
+ "transformer.h.26.mlp.c_proj",
327
+ "transformer.h.24.attn.c_proj",
328
+ "transformer.visual.transformer.resblocks.47.attn.in_proj",
329
+ "transformer.visual.transformer.resblocks.7.mlp.c_fc",
330
+ "transformer.h.2.attn.c_attn",
331
+ "transformer.visual.conv1",
332
+ "transformer.visual.transformer.resblocks.28.attn.in_proj",
333
+ "transformer.visual.transformer.resblocks.25.attn.in_proj",
334
+ "transformer.visual.transformer.resblocks.47.attn.out_proj",
335
+ "transformer.visual.transformer.resblocks.18.mlp.c_fc",
336
+ "transformer.h.28.mlp.w1",
337
+ "transformer.h.10.mlp.w2",
338
+ "transformer.h.15.attn.c_proj",
339
+ "transformer.h.21.mlp.c_proj",
340
+ "transformer.h.29.mlp.w2",
341
+ "transformer.visual.transformer.resblocks.32.mlp.c_fc",
342
+ "transformer.h.18.mlp.w1",
343
+ "transformer.visual.transformer.resblocks.41.attn.out_proj",
344
+ "transformer.h.8.mlp.c_proj",
345
+ "transformer.visual.transformer.resblocks.47.mlp.c_proj",
346
+ "transformer.visual.transformer.resblocks.28.mlp.c_fc",
347
+ "transformer.visual.transformer.resblocks.34.attn.out_proj",
348
+ "transformer.h.29.attn.c_proj",
349
+ "transformer.visual.transformer.resblocks.38.attn.out_proj",
350
+ "transformer.visual.transformer.resblocks.14.attn.out_proj",
351
+ "transformer.h.10.mlp.c_proj",
352
+ "transformer.visual.transformer.resblocks.6.mlp.c_proj",
353
+ "transformer.h.0.mlp.w2",
354
+ "transformer.visual.transformer.resblocks.26.attn.in_proj",
355
+ "transformer.visual.transformer.resblocks.42.attn.in_proj",
356
+ "transformer.h.15.attn.c_attn",
357
+ "transformer.h.16.mlp.w1",
358
+ "transformer.visual.transformer.resblocks.3.attn.out_proj",
359
+ "transformer.visual.transformer.resblocks.11.attn.out_proj",
360
+ "transformer.visual.transformer.resblocks.28.mlp.c_proj",
361
+ "transformer.visual.transformer.resblocks.14.mlp.c_proj",
362
+ "transformer.h.0.attn.c_attn",
363
+ "transformer.h.18.mlp.c_proj",
364
+ "transformer.h.12.mlp.w2",
365
+ "transformer.visual.transformer.resblocks.35.attn.in_proj",
366
+ "transformer.visual.transformer.resblocks.2.mlp.c_fc",
367
+ "transformer.visual.transformer.resblocks.34.mlp.c_proj",
368
+ "transformer.h.26.mlp.w1",
369
+ "transformer.visual.transformer.resblocks.13.attn.out_proj",
370
+ "transformer.visual.transformer.resblocks.20.attn.out_proj",
371
+ "transformer.visual.transformer.resblocks.46.attn.in_proj",
372
+ "transformer.h.16.attn.c_proj",
373
+ "transformer.h.15.mlp.c_proj",
374
+ "transformer.h.27.mlp.w2",
375
+ "transformer.h.28.mlp.c_proj"
376
+ ],
377
+ "task_type": "CAUSAL_LM",
378
+ "use_dora": false,
379
+ "use_rslora": false
380
+ }
checkpoint-4800/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0e663a759cee99fe8865cfc04d1586ae87193182ead52ffdfc5ebaa525d6bfc
3
+ size 469105640
checkpoint-4800/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step4800
checkpoint-4800/qwen.tiktoken ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-4800/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28473162a49089d0972e2dcfe5dcc41585a36185c2eccf2bc969ec4b88f33b7c
3
+ size 14960
checkpoint-4800/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5981604d61cf445079ec50c068a212676f141b42a6e0322ebcac084bdfbb777c
3
+ size 14960
checkpoint-4800/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:977afebc33d4b242f4bcf939d40d870833863a8e1600a19418174c0ea6ce276a
3
+ size 14960
checkpoint-4800/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c6b94f2869af4b5c95e4b0d398ecfea289ce3ae4e646870d5f4470ed6d9f954
3
+ size 14960
checkpoint-4800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a664c5bf6eafd9d56904b210471088e839c3781b9ce1fe29f293f89479420e96
3
+ size 1064
checkpoint-4800/special_tokens_map.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "pad_token": "<|endoftext|>"
3
+ }
checkpoint-4800/tokenizer_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {},
3
+ "auto_map": {
4
+ "AutoTokenizer": [
5
+ "Qwen/Qwen-VL-Chat--tokenization_qwen.QWenTokenizer",
6
+ null
7
+ ]
8
+ },
9
+ "clean_up_tokenization_spaces": true,
10
+ "model_max_length": 1280,
11
+ "pad_token": "<|endoftext|>",
12
+ "padding_side": "right",
13
+ "tokenizer_class": "QWenTokenizer"
14
+ }
checkpoint-4800/trainer_state.json ADDED
@@ -0,0 +1,3393 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.6402347527426723,
5
+ "eval_steps": 500,
6
+ "global_step": 4800,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001333822401547234,
13
+ "grad_norm": 72.53171450734499,
14
+ "learning_rate": 4e-06,
15
+ "loss": 1.0439,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.002667644803094468,
20
+ "grad_norm": 7.72170877312484,
21
+ "learning_rate": 8e-06,
22
+ "loss": 1.0385,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.004001467204641702,
27
+ "grad_norm": 28.037294890647996,
28
+ "learning_rate": 1.2e-05,
29
+ "loss": 1.0328,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.005335289606188936,
34
+ "grad_norm": 10.980963099753343,
35
+ "learning_rate": 1.6e-05,
36
+ "loss": 0.9608,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.00666911200773617,
41
+ "grad_norm": 9.853684997615185,
42
+ "learning_rate": 1.9999999999999998e-05,
43
+ "loss": 0.8941,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.008002934409283404,
48
+ "grad_norm": 33.396994669126855,
49
+ "learning_rate": 2.4e-05,
50
+ "loss": 0.9174,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.009336756810830639,
55
+ "grad_norm": 5.621190624979587,
56
+ "learning_rate": 2.8e-05,
57
+ "loss": 0.8146,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.010670579212377872,
62
+ "grad_norm": 2.4646688601781843,
63
+ "learning_rate": 2.9999966406213696e-05,
64
+ "loss": 0.7774,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.012004401613925107,
69
+ "grad_norm": 2.1272107013027215,
70
+ "learning_rate": 2.9999697656826056e-05,
71
+ "loss": 0.7426,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01333822401547234,
76
+ "grad_norm": 2.0001927521328575,
77
+ "learning_rate": 2.9999160162865885e-05,
78
+ "loss": 0.7251,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.014672046417019574,
83
+ "grad_norm": 4.278494734910313,
84
+ "learning_rate": 2.9998353933963273e-05,
85
+ "loss": 0.7283,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.016005868818566808,
90
+ "grad_norm": 84.74138622339441,
91
+ "learning_rate": 2.999727898456315e-05,
92
+ "loss": 0.721,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.017339691220114042,
97
+ "grad_norm": 2.5596285292704533,
98
+ "learning_rate": 2.999593533392503e-05,
99
+ "loss": 0.7476,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.018673513621661277,
104
+ "grad_norm": 3.167163073798878,
105
+ "learning_rate": 2.9994323006122654e-05,
106
+ "loss": 0.7701,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.02000733602320851,
111
+ "grad_norm": 12.240132620099228,
112
+ "learning_rate": 2.9992442030043557e-05,
113
+ "loss": 0.7042,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.021341158424755743,
118
+ "grad_norm": 2.266079140233148,
119
+ "learning_rate": 2.9990292439388565e-05,
120
+ "loss": 0.6945,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.022674980826302978,
125
+ "grad_norm": 4.493189214061043,
126
+ "learning_rate": 2.9987874272671168e-05,
127
+ "loss": 0.7811,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.024008803227850213,
132
+ "grad_norm": 3.8513915234854132,
133
+ "learning_rate": 2.9985187573216855e-05,
134
+ "loss": 0.7229,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.025342625629397444,
139
+ "grad_norm": 2.1061250870336896,
140
+ "learning_rate": 2.998223238916232e-05,
141
+ "loss": 0.7253,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.02667644803094468,
146
+ "grad_norm": 2.8326879963181373,
147
+ "learning_rate": 2.9979008773454618e-05,
148
+ "loss": 0.7576,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.028010270432491914,
153
+ "grad_norm": 3.0294324007099402,
154
+ "learning_rate": 2.997551678385019e-05,
155
+ "loss": 0.7362,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02934409283403915,
160
+ "grad_norm": 1.5691302199679913,
161
+ "learning_rate": 2.997175648291384e-05,
162
+ "loss": 0.7212,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.03067791523558638,
167
+ "grad_norm": 4.379709379404857,
168
+ "learning_rate": 2.996772793801763e-05,
169
+ "loss": 0.7382,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.032011737637133615,
174
+ "grad_norm": 1.899348727085658,
175
+ "learning_rate": 2.996343122133965e-05,
176
+ "loss": 0.7056,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.033345560038680847,
181
+ "grad_norm": 21.481584335801955,
182
+ "learning_rate": 2.9958866409862745e-05,
183
+ "loss": 0.697,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.034679382440228085,
188
+ "grad_norm": 3.022034785435926,
189
+ "learning_rate": 2.9954033585373108e-05,
190
+ "loss": 0.69,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.036013204841775316,
195
+ "grad_norm": 2.869274575136443,
196
+ "learning_rate": 2.994893283445885e-05,
197
+ "loss": 0.7026,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.037347027243322554,
202
+ "grad_norm": 7.8785627268474,
203
+ "learning_rate": 2.9943564248508415e-05,
204
+ "loss": 0.7021,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.038680849644869786,
209
+ "grad_norm": 4.450751706186503,
210
+ "learning_rate": 2.9937927923708966e-05,
211
+ "loss": 0.6814,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.04001467204641702,
216
+ "grad_norm": 11.613541400971123,
217
+ "learning_rate": 2.993202396104465e-05,
218
+ "loss": 0.7043,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.041348494447964255,
223
+ "grad_norm": 2.0602161594051536,
224
+ "learning_rate": 2.9925852466294795e-05,
225
+ "loss": 0.7073,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.04268231684951149,
230
+ "grad_norm": 19.383133688390824,
231
+ "learning_rate": 2.9919413550032014e-05,
232
+ "loss": 0.6965,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.04401613925105872,
237
+ "grad_norm": 7.9603948136759906,
238
+ "learning_rate": 2.991270732762022e-05,
239
+ "loss": 0.7029,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.045349961652605957,
244
+ "grad_norm": 2.810319267000959,
245
+ "learning_rate": 2.990573391921255e-05,
246
+ "loss": 0.7412,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.04668378405415319,
251
+ "grad_norm": 6.082487526538632,
252
+ "learning_rate": 2.989849344974924e-05,
253
+ "loss": 0.6913,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.048017606455700426,
258
+ "grad_norm": 1.3651265332328906,
259
+ "learning_rate": 2.9890986048955368e-05,
260
+ "loss": 0.6802,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.04935142885724766,
265
+ "grad_norm": 2.7766095141894023,
266
+ "learning_rate": 2.9883211851338516e-05,
267
+ "loss": 0.7254,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.05068525125879489,
272
+ "grad_norm": 2.2284551774435033,
273
+ "learning_rate": 2.9875170996186392e-05,
274
+ "loss": 0.7026,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.05201907366034213,
279
+ "grad_norm": 2.1609680967998957,
280
+ "learning_rate": 2.986686362756431e-05,
281
+ "loss": 0.7415,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.05335289606188936,
286
+ "grad_norm": 2.071743222471935,
287
+ "learning_rate": 2.9858289894312617e-05,
288
+ "loss": 0.7081,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.0546867184634366,
293
+ "grad_norm": 7.349693198515479,
294
+ "learning_rate": 2.9849449950044036e-05,
295
+ "loss": 0.7212,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.05602054086498383,
300
+ "grad_norm": 12.990346405367537,
301
+ "learning_rate": 2.984034395314088e-05,
302
+ "loss": 0.7187,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.05735436326653106,
307
+ "grad_norm": 2.4106321189642723,
308
+ "learning_rate": 2.983097206675227e-05,
309
+ "loss": 0.6881,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0586881856680783,
314
+ "grad_norm": 3.1647778114563163,
315
+ "learning_rate": 2.9821334458791156e-05,
316
+ "loss": 0.7529,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.06002200806962553,
321
+ "grad_norm": 2.7769936556882713,
322
+ "learning_rate": 2.9811431301931344e-05,
323
+ "loss": 0.7073,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.06135583047117276,
328
+ "grad_norm": 2.834916196253906,
329
+ "learning_rate": 2.9801262773604377e-05,
330
+ "loss": 0.6649,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.06268965287271999,
335
+ "grad_norm": 2.431783869557187,
336
+ "learning_rate": 2.9790829055996398e-05,
337
+ "loss": 0.6889,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.06402347527426723,
342
+ "grad_norm": 3.7325728788625825,
343
+ "learning_rate": 2.978013033604483e-05,
344
+ "loss": 0.6771,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.06535729767581447,
349
+ "grad_norm": 10.194734403509758,
350
+ "learning_rate": 2.976916680543506e-05,
351
+ "loss": 0.6917,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.06669112007736169,
356
+ "grad_norm": 6.4699962355353655,
357
+ "learning_rate": 2.975793866059701e-05,
358
+ "loss": 0.6656,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.06802494247890893,
363
+ "grad_norm": 7.664735661456941,
364
+ "learning_rate": 2.9746446102701606e-05,
365
+ "loss": 0.7268,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.06935876488045617,
370
+ "grad_norm": 2.926294421836929,
371
+ "learning_rate": 2.9734689337657157e-05,
372
+ "loss": 0.7498,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.07069258728200341,
377
+ "grad_norm": 8.95556138792813,
378
+ "learning_rate": 2.9722668576105703e-05,
379
+ "loss": 0.7237,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.07202640968355063,
384
+ "grad_norm": 2.725769364825004,
385
+ "learning_rate": 2.971038403341921e-05,
386
+ "loss": 0.701,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.07336023208509787,
391
+ "grad_norm": 2.458130413236654,
392
+ "learning_rate": 2.9697835929695727e-05,
393
+ "loss": 0.6775,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.07469405448664511,
398
+ "grad_norm": 1.908510150241819,
399
+ "learning_rate": 2.968502448975544e-05,
400
+ "loss": 0.7042,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.07602787688819233,
405
+ "grad_norm": 26.340984665432142,
406
+ "learning_rate": 2.967194994313663e-05,
407
+ "loss": 0.7234,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.07736169928973957,
412
+ "grad_norm": 5.677162249188698,
413
+ "learning_rate": 2.9658612524091594e-05,
414
+ "loss": 0.7159,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.07869552169128681,
419
+ "grad_norm": 1.7975075976512287,
420
+ "learning_rate": 2.9645012471582406e-05,
421
+ "loss": 0.7268,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.08002934409283403,
426
+ "grad_norm": 3.031655038731278,
427
+ "learning_rate": 2.9631150029276662e-05,
428
+ "loss": 0.6904,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.08136316649438127,
433
+ "grad_norm": 13.019769162242698,
434
+ "learning_rate": 2.9617025445543114e-05,
435
+ "loss": 0.683,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.08269698889592851,
440
+ "grad_norm": 5.982097939523946,
441
+ "learning_rate": 2.9602638973447218e-05,
442
+ "loss": 0.7511,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.08403081129747574,
447
+ "grad_norm": 2.7343884277014054,
448
+ "learning_rate": 2.9587990870746574e-05,
449
+ "loss": 0.7114,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.08536463369902297,
454
+ "grad_norm": 5.586409371459215,
455
+ "learning_rate": 2.9573081399886356e-05,
456
+ "loss": 0.6907,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.08669845610057021,
461
+ "grad_norm": 9.87255558712134,
462
+ "learning_rate": 2.9557910827994568e-05,
463
+ "loss": 0.7319,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.08803227850211744,
468
+ "grad_norm": 3.9867937712251664,
469
+ "learning_rate": 2.9542479426877283e-05,
470
+ "loss": 0.6759,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.08936610090366467,
475
+ "grad_norm": 3.111751395898768,
476
+ "learning_rate": 2.9526787473013753e-05,
477
+ "loss": 0.7086,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.09069992330521191,
482
+ "grad_norm": 3.0578210081080535,
483
+ "learning_rate": 2.9510835247551485e-05,
484
+ "loss": 0.7209,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.09203374570675915,
489
+ "grad_norm": 16.575220829800145,
490
+ "learning_rate": 2.949462303630116e-05,
491
+ "loss": 0.7125,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.09336756810830638,
496
+ "grad_norm": 1.8236956939037314,
497
+ "learning_rate": 2.9478151129731567e-05,
498
+ "loss": 0.6591,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.09470139050985361,
503
+ "grad_norm": 1.9746052093914703,
504
+ "learning_rate": 2.9461419822964348e-05,
505
+ "loss": 0.7048,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.09603521291140085,
510
+ "grad_norm": 5.9710918845264,
511
+ "learning_rate": 2.9444429415768726e-05,
512
+ "loss": 0.725,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.09736903531294808,
517
+ "grad_norm": 9.961068856364284,
518
+ "learning_rate": 2.942718021255617e-05,
519
+ "loss": 0.7155,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.09870285771449532,
524
+ "grad_norm": 9.195914431100427,
525
+ "learning_rate": 2.940967252237488e-05,
526
+ "loss": 0.7171,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.10003668011604255,
531
+ "grad_norm": 5.972259927477824,
532
+ "learning_rate": 2.9391906658904296e-05,
533
+ "loss": 0.7088,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.10137050251758978,
538
+ "grad_norm": 3.0581670197906226,
539
+ "learning_rate": 2.937388294044946e-05,
540
+ "loss": 0.6645,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.10270432491913702,
545
+ "grad_norm": 3.2614276687575945,
546
+ "learning_rate": 2.9355601689935315e-05,
547
+ "loss": 0.6997,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.10403814732068425,
552
+ "grad_norm": 3.212661762840432,
553
+ "learning_rate": 2.933706323490092e-05,
554
+ "loss": 0.7005,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.10537196972223148,
559
+ "grad_norm": 10.732038885056795,
560
+ "learning_rate": 2.9318267907493583e-05,
561
+ "loss": 0.6588,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.10670579212377872,
566
+ "grad_norm": 2.495380329507228,
567
+ "learning_rate": 2.9299216044462903e-05,
568
+ "loss": 0.6965,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.10803961452532596,
573
+ "grad_norm": 4.60295956813534,
574
+ "learning_rate": 2.927990798715475e-05,
575
+ "loss": 0.6959,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.1093734369268732,
580
+ "grad_norm": 5.95010483593304,
581
+ "learning_rate": 2.926034408150513e-05,
582
+ "loss": 0.7083,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.11070725932842042,
587
+ "grad_norm": 10.48663715000287,
588
+ "learning_rate": 2.9240524678034016e-05,
589
+ "loss": 0.6977,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.11204108172996766,
594
+ "grad_norm": 3.159424497856185,
595
+ "learning_rate": 2.9220450131839037e-05,
596
+ "loss": 0.7129,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.1133749041315149,
601
+ "grad_norm": 3.984530217756339,
602
+ "learning_rate": 2.920012080258912e-05,
603
+ "loss": 0.6618,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.11470872653306212,
608
+ "grad_norm": 4.212436954015185,
609
+ "learning_rate": 2.9179537054518085e-05,
610
+ "loss": 0.6958,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.11604254893460936,
615
+ "grad_norm": 2.5397356754848657,
616
+ "learning_rate": 2.9158699256418056e-05,
617
+ "loss": 0.6593,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.1173763713361566,
622
+ "grad_norm": 7.8192219616904115,
623
+ "learning_rate": 2.9137607781632913e-05,
624
+ "loss": 0.6932,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.11871019373770382,
629
+ "grad_norm": 3.890974772100691,
630
+ "learning_rate": 2.911626300805155e-05,
631
+ "loss": 0.674,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.12004401613925106,
636
+ "grad_norm": 2.241784461612369,
637
+ "learning_rate": 2.9094665318101155e-05,
638
+ "loss": 0.6808,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.1213778385407983,
643
+ "grad_norm": 2.8295407624221456,
644
+ "learning_rate": 2.9072815098740326e-05,
645
+ "loss": 0.7074,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.12271166094234552,
650
+ "grad_norm": 2.8191065019379455,
651
+ "learning_rate": 2.9050712741452136e-05,
652
+ "loss": 0.7218,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.12404548334389276,
657
+ "grad_norm": 2.1348105213857362,
658
+ "learning_rate": 2.902835864223715e-05,
659
+ "loss": 0.6901,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.12537930574543998,
664
+ "grad_norm": 2.547712785034466,
665
+ "learning_rate": 2.9005753201606287e-05,
666
+ "loss": 0.7024,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.12671312814698724,
671
+ "grad_norm": 6.0714243848770195,
672
+ "learning_rate": 2.8982896824573678e-05,
673
+ "loss": 0.665,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.12804695054853446,
678
+ "grad_norm": 2.928297684665459,
679
+ "learning_rate": 2.8959789920649394e-05,
680
+ "loss": 0.7178,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.12938077295008168,
685
+ "grad_norm": 9.456759373673547,
686
+ "learning_rate": 2.893643290383212e-05,
687
+ "loss": 0.6759,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.13071459535162894,
692
+ "grad_norm": 3.9700721280634648,
693
+ "learning_rate": 2.891282619260172e-05,
694
+ "loss": 0.6701,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.13204841775317616,
699
+ "grad_norm": 1.9216628539066036,
700
+ "learning_rate": 2.8888970209911754e-05,
701
+ "loss": 0.6725,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.13338224015472339,
706
+ "grad_norm": 6.51104150647371,
707
+ "learning_rate": 2.8864865383181893e-05,
708
+ "loss": 0.6811,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.13471606255627064,
713
+ "grad_norm": 7.1765693144596705,
714
+ "learning_rate": 2.8840512144290273e-05,
715
+ "loss": 0.7085,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.13604988495781786,
720
+ "grad_norm": 4.136786758396937,
721
+ "learning_rate": 2.8815910929565734e-05,
722
+ "loss": 0.7128,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.1373837073593651,
727
+ "grad_norm": 3.407893493053672,
728
+ "learning_rate": 2.879106217978002e-05,
729
+ "loss": 0.6782,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.13871752976091234,
734
+ "grad_norm": 2.5759617833329544,
735
+ "learning_rate": 2.8765966340139892e-05,
736
+ "loss": 0.7387,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.14005135216245956,
741
+ "grad_norm": 8.496713788438253,
742
+ "learning_rate": 2.8740623860279116e-05,
743
+ "loss": 0.6988,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.14138517456400682,
748
+ "grad_norm": 1.438197837288145,
749
+ "learning_rate": 2.871503519425044e-05,
750
+ "loss": 0.6797,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.14271899696555404,
755
+ "grad_norm": 3.305947931404651,
756
+ "learning_rate": 2.8689200800517448e-05,
757
+ "loss": 0.727,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.14405281936710126,
762
+ "grad_norm": 4.384984204254061,
763
+ "learning_rate": 2.866312114194634e-05,
764
+ "loss": 0.7073,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.14538664176864852,
769
+ "grad_norm": 3.0424755710445455,
770
+ "learning_rate": 2.8636796685797657e-05,
771
+ "loss": 0.6712,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.14672046417019574,
776
+ "grad_norm": 3.7562374375017415,
777
+ "learning_rate": 2.8610227903717876e-05,
778
+ "loss": 0.6811,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.14805428657174297,
783
+ "grad_norm": 18.148320631618972,
784
+ "learning_rate": 2.8583415271730994e-05,
785
+ "loss": 0.669,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.14938810897329022,
790
+ "grad_norm": 10.095266883651389,
791
+ "learning_rate": 2.855635927022998e-05,
792
+ "loss": 0.7034,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.15072193137483744,
797
+ "grad_norm": 6.298167661594497,
798
+ "learning_rate": 2.8529060383968175e-05,
799
+ "loss": 0.6736,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.15205575377638467,
804
+ "grad_norm": 1.7112633381805116,
805
+ "learning_rate": 2.850151910205061e-05,
806
+ "loss": 0.7072,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.15338957617793192,
811
+ "grad_norm": 4.478017946548926,
812
+ "learning_rate": 2.847373591792523e-05,
813
+ "loss": 0.679,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.15472339857947914,
818
+ "grad_norm": 6.208223721690807,
819
+ "learning_rate": 2.844571132937407e-05,
820
+ "loss": 0.688,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.15605722098102637,
825
+ "grad_norm": 3.5734211349612974,
826
+ "learning_rate": 2.841744583850431e-05,
827
+ "loss": 0.7054,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.15739104338257362,
832
+ "grad_norm": 2.9418941665585874,
833
+ "learning_rate": 2.838893995173932e-05,
834
+ "loss": 0.7051,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.15872486578412084,
839
+ "grad_norm": 2.8796543096657055,
840
+ "learning_rate": 2.836019417980955e-05,
841
+ "loss": 0.6856,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.16005868818566807,
846
+ "grad_norm": 1.3218998593276705,
847
+ "learning_rate": 2.8331209037743387e-05,
848
+ "loss": 0.69,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.16139251058721532,
853
+ "grad_norm": 1.566163155963456,
854
+ "learning_rate": 2.8301985044857947e-05,
855
+ "loss": 0.6384,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.16272633298876255,
860
+ "grad_norm": 6.731577281026319,
861
+ "learning_rate": 2.8272522724749743e-05,
862
+ "loss": 0.7241,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.16406015539030977,
867
+ "grad_norm": 3.3521245738531813,
868
+ "learning_rate": 2.8242822605285323e-05,
869
+ "loss": 0.6931,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.16539397779185702,
874
+ "grad_norm": 11.826633616347152,
875
+ "learning_rate": 2.8212885218591812e-05,
876
+ "loss": 0.702,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.16672780019340425,
881
+ "grad_norm": 1.826627133990473,
882
+ "learning_rate": 2.8182711101047362e-05,
883
+ "loss": 0.7061,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.16806162259495147,
888
+ "grad_norm": 2.573810458184169,
889
+ "learning_rate": 2.815230079327156e-05,
890
+ "loss": 0.7369,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.16939544499649872,
895
+ "grad_norm": 3.2839708945300665,
896
+ "learning_rate": 2.8121654840115734e-05,
897
+ "loss": 0.6966,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.17072926739804595,
902
+ "grad_norm": 2.3799212816187136,
903
+ "learning_rate": 2.809077379065319e-05,
904
+ "loss": 0.6586,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.17206308979959317,
909
+ "grad_norm": 8.449506341181594,
910
+ "learning_rate": 2.805965819816937e-05,
911
+ "loss": 0.6759,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.17339691220114042,
916
+ "grad_norm": 19.59186920922371,
917
+ "learning_rate": 2.802830862015196e-05,
918
+ "loss": 0.6537,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.17473073460268765,
923
+ "grad_norm": 1.9822479021376653,
924
+ "learning_rate": 2.799672561828087e-05,
925
+ "loss": 0.6854,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.17606455700423487,
930
+ "grad_norm": 34.5695029919346,
931
+ "learning_rate": 2.79649097584182e-05,
932
+ "loss": 0.696,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.17739837940578213,
937
+ "grad_norm": 1.886275024045834,
938
+ "learning_rate": 2.7932861610598077e-05,
939
+ "loss": 0.7308,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.17873220180732935,
944
+ "grad_norm": 12.128310547971918,
945
+ "learning_rate": 2.7900581749016466e-05,
946
+ "loss": 0.6775,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.1800660242088766,
951
+ "grad_norm": 5.164562576250181,
952
+ "learning_rate": 2.7868070752020865e-05,
953
+ "loss": 0.7096,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.18139984661042383,
958
+ "grad_norm": 10.603161437547913,
959
+ "learning_rate": 2.7835329202099944e-05,
960
+ "loss": 0.7128,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.18273366901197105,
965
+ "grad_norm": 4.184026615605208,
966
+ "learning_rate": 2.7802357685873117e-05,
967
+ "loss": 0.6958,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.1840674914135183,
972
+ "grad_norm": 4.135408851494439,
973
+ "learning_rate": 2.7769156794080033e-05,
974
+ "loss": 0.6933,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.18540131381506553,
979
+ "grad_norm": 1.7535535971850658,
980
+ "learning_rate": 2.7735727121569967e-05,
981
+ "loss": 0.7088,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.18673513621661275,
986
+ "grad_norm": 5.354882317397714,
987
+ "learning_rate": 2.770206926729121e-05,
988
+ "loss": 0.6777,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.18806895861816,
993
+ "grad_norm": 3.2593978573810523,
994
+ "learning_rate": 2.7668183834280284e-05,
995
+ "loss": 0.6535,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.18940278101970723,
1000
+ "grad_norm": 2.48929554913285,
1001
+ "learning_rate": 2.763407142965117e-05,
1002
+ "loss": 0.6853,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.19073660342125445,
1007
+ "grad_norm": 7.550246976134938,
1008
+ "learning_rate": 2.759973266458444e-05,
1009
+ "loss": 0.679,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1920704258228017,
1014
+ "grad_norm": 5.033109901639584,
1015
+ "learning_rate": 2.756516815431627e-05,
1016
+ "loss": 0.6875,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.19340424822434893,
1021
+ "grad_norm": 6.1146492204468545,
1022
+ "learning_rate": 2.7530378518127445e-05,
1023
+ "loss": 0.6762,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.19473807062589615,
1028
+ "grad_norm": 72.48946246157955,
1029
+ "learning_rate": 2.7495364379332256e-05,
1030
+ "loss": 0.6766,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.1960718930274434,
1035
+ "grad_norm": 2.9676182245937426,
1036
+ "learning_rate": 2.7460126365267335e-05,
1037
+ "loss": 0.6731,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.19740571542899063,
1042
+ "grad_norm": 2.31300235137622,
1043
+ "learning_rate": 2.7424665107280402e-05,
1044
+ "loss": 0.6936,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.19873953783053785,
1049
+ "grad_norm": 2.594823775876976,
1050
+ "learning_rate": 2.738898124071898e-05,
1051
+ "loss": 0.6703,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.2000733602320851,
1056
+ "grad_norm": 2.8354629666044575,
1057
+ "learning_rate": 2.735307540491898e-05,
1058
+ "loss": 0.7022,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.20140718263363233,
1063
+ "grad_norm": 4.553033073507926,
1064
+ "learning_rate": 2.7316948243193273e-05,
1065
+ "loss": 0.6948,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.20274100503517956,
1070
+ "grad_norm": 10.052746707722667,
1071
+ "learning_rate": 2.7280600402820146e-05,
1072
+ "loss": 0.7216,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.2040748274367268,
1077
+ "grad_norm": 2.3783106567256977,
1078
+ "learning_rate": 2.724403253503171e-05,
1079
+ "loss": 0.6955,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.20540864983827403,
1084
+ "grad_norm": 2.26171864458522,
1085
+ "learning_rate": 2.7207245295002242e-05,
1086
+ "loss": 0.7062,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.20674247223982126,
1091
+ "grad_norm": 3.223919319241339,
1092
+ "learning_rate": 2.7170239341836436e-05,
1093
+ "loss": 0.6659,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.2080762946413685,
1098
+ "grad_norm": 1.6810934171118763,
1099
+ "learning_rate": 2.7133015338557585e-05,
1100
+ "loss": 0.7069,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.20941011704291573,
1105
+ "grad_norm": 2.1432477935562515,
1106
+ "learning_rate": 2.7095573952095727e-05,
1107
+ "loss": 0.703,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.21074393944446296,
1112
+ "grad_norm": 1.5654642597457342,
1113
+ "learning_rate": 2.705791585327568e-05,
1114
+ "loss": 0.7037,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.2120777618460102,
1119
+ "grad_norm": 2.5797486712147992,
1120
+ "learning_rate": 2.7020041716805014e-05,
1121
+ "loss": 0.6945,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.21341158424755743,
1126
+ "grad_norm": 6.993511462468879,
1127
+ "learning_rate": 2.6981952221261986e-05,
1128
+ "loss": 0.6965,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.21474540664910466,
1133
+ "grad_norm": 1.9963649259483505,
1134
+ "learning_rate": 2.6943648049083366e-05,
1135
+ "loss": 0.7052,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.2160792290506519,
1140
+ "grad_norm": 13.563069274486924,
1141
+ "learning_rate": 2.6905129886552208e-05,
1142
+ "loss": 0.6925,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.21741305145219914,
1147
+ "grad_norm": 1.8811719402961335,
1148
+ "learning_rate": 2.6866398423785568e-05,
1149
+ "loss": 0.6739,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.2187468738537464,
1154
+ "grad_norm": 3.400454288874876,
1155
+ "learning_rate": 2.682745435472212e-05,
1156
+ "loss": 0.7115,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.2200806962552936,
1161
+ "grad_norm": 1.7511437629316988,
1162
+ "learning_rate": 2.6788298377109748e-05,
1163
+ "loss": 0.6939,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.22141451865684084,
1168
+ "grad_norm": 7.03338780434775,
1169
+ "learning_rate": 2.6748931192493017e-05,
1170
+ "loss": 0.7081,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.2227483410583881,
1175
+ "grad_norm": 7.485978307196749,
1176
+ "learning_rate": 2.670935350620063e-05,
1177
+ "loss": 0.7483,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.2240821634599353,
1182
+ "grad_norm": 11.472423495255576,
1183
+ "learning_rate": 2.6669566027332767e-05,
1184
+ "loss": 0.6653,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.22541598586148254,
1189
+ "grad_norm": 2.8967622506628388,
1190
+ "learning_rate": 2.6629569468748404e-05,
1191
+ "loss": 0.6994,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.2267498082630298,
1196
+ "grad_norm": 3.689173173355099,
1197
+ "learning_rate": 2.658936454705251e-05,
1198
+ "loss": 0.7121,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.22808363066457701,
1203
+ "grad_norm": 2.5714663434660414,
1204
+ "learning_rate": 2.6548951982583246e-05,
1205
+ "loss": 0.679,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.22941745306612424,
1210
+ "grad_norm": 2.6005673266384854,
1211
+ "learning_rate": 2.650833249939903e-05,
1212
+ "loss": 0.6826,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.2307512754676715,
1217
+ "grad_norm": 25.941043620884358,
1218
+ "learning_rate": 2.6467506825265573e-05,
1219
+ "loss": 0.7202,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.23208509786921871,
1224
+ "grad_norm": 2.9541333611408396,
1225
+ "learning_rate": 2.642647569164284e-05,
1226
+ "loss": 0.6855,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.23341892027076594,
1231
+ "grad_norm": 1.6808077602808311,
1232
+ "learning_rate": 2.638523983367194e-05,
1233
+ "loss": 0.6901,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.2347527426723132,
1238
+ "grad_norm": 8.145841190370119,
1239
+ "learning_rate": 2.634379999016198e-05,
1240
+ "loss": 0.7,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.23608656507386042,
1245
+ "grad_norm": 17.930965708795497,
1246
+ "learning_rate": 2.6302156903576784e-05,
1247
+ "loss": 0.6315,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.23742038747540764,
1252
+ "grad_norm": 3.8449192904768186,
1253
+ "learning_rate": 2.6260311320021628e-05,
1254
+ "loss": 0.6369,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.2387542098769549,
1259
+ "grad_norm": 1.7681330868700245,
1260
+ "learning_rate": 2.6218263989229855e-05,
1261
+ "loss": 0.674,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.24008803227850212,
1266
+ "grad_norm": 2.1491823327138317,
1267
+ "learning_rate": 2.617601566454944e-05,
1268
+ "loss": 0.6979,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.24142185468004934,
1273
+ "grad_norm": 4.667708666590814,
1274
+ "learning_rate": 2.613356710292951e-05,
1275
+ "loss": 0.6932,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.2427556770815966,
1280
+ "grad_norm": 2.70251113889306,
1281
+ "learning_rate": 2.6090919064906766e-05,
1282
+ "loss": 0.6842,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.24408949948314382,
1287
+ "grad_norm": 4.187423249865094,
1288
+ "learning_rate": 2.6048072314591854e-05,
1289
+ "loss": 0.6642,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.24542332188469104,
1294
+ "grad_norm": 1.6832068899399832,
1295
+ "learning_rate": 2.600502761965569e-05,
1296
+ "loss": 0.6984,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.2467571442862383,
1301
+ "grad_norm": 5.37697139339205,
1302
+ "learning_rate": 2.59617857513157e-05,
1303
+ "loss": 0.6489,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.24809096668778552,
1308
+ "grad_norm": 2.6302546632419186,
1309
+ "learning_rate": 2.591834748432198e-05,
1310
+ "loss": 0.6923,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.24942478908933274,
1315
+ "grad_norm": 3.8259011796471927,
1316
+ "learning_rate": 2.5874713596943465e-05,
1317
+ "loss": 0.7163,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.25075861149087997,
1322
+ "grad_norm": 5.322548910178761,
1323
+ "learning_rate": 2.5830884870953933e-05,
1324
+ "loss": 0.7188,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.25209243389242725,
1329
+ "grad_norm": 1.5202627897221725,
1330
+ "learning_rate": 2.578686209161803e-05,
1331
+ "loss": 0.6975,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.2534262562939745,
1336
+ "grad_norm": 2.7405063442524966,
1337
+ "learning_rate": 2.5742646047677186e-05,
1338
+ "loss": 0.7257,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.2547600786955217,
1343
+ "grad_norm": 3.7267206506339225,
1344
+ "learning_rate": 2.5698237531335493e-05,
1345
+ "loss": 0.7078,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.2560939010970689,
1350
+ "grad_norm": 1.7323421214775496,
1351
+ "learning_rate": 2.56536373382455e-05,
1352
+ "loss": 0.6739,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.25742772349861615,
1357
+ "grad_norm": 5.812235836466133,
1358
+ "learning_rate": 2.5608846267493974e-05,
1359
+ "loss": 0.668,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.25876154590016337,
1364
+ "grad_norm": 5.482474929884452,
1365
+ "learning_rate": 2.5563865121587563e-05,
1366
+ "loss": 0.6633,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.26009536830171065,
1371
+ "grad_norm": 10.990988768370446,
1372
+ "learning_rate": 2.5518694706438445e-05,
1373
+ "loss": 0.7351,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.2614291907032579,
1378
+ "grad_norm": 5.066143279063377,
1379
+ "learning_rate": 2.5473335831349842e-05,
1380
+ "loss": 0.6479,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.2627630131048051,
1385
+ "grad_norm": 1.8538054067880405,
1386
+ "learning_rate": 2.5427789309001577e-05,
1387
+ "loss": 0.691,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.2640968355063523,
1392
+ "grad_norm": 3.642801968239499,
1393
+ "learning_rate": 2.538205595543548e-05,
1394
+ "loss": 0.6981,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.26543065790789955,
1399
+ "grad_norm": 4.027276498286695,
1400
+ "learning_rate": 2.5336136590040767e-05,
1401
+ "loss": 0.6471,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.26676448030944677,
1406
+ "grad_norm": 5.225146913060103,
1407
+ "learning_rate": 2.529003203553937e-05,
1408
+ "loss": 0.6768,
1409
+ "step": 2000
1410
+ },
1411
+ {
1412
+ "epoch": 0.26809830271099405,
1413
+ "grad_norm": 3.253526388391669,
1414
+ "learning_rate": 2.5243743117971186e-05,
1415
+ "loss": 0.7292,
1416
+ "step": 2010
1417
+ },
1418
+ {
1419
+ "epoch": 0.2694321251125413,
1420
+ "grad_norm": 3.595165338865758,
1421
+ "learning_rate": 2.5197270666679295e-05,
1422
+ "loss": 0.7153,
1423
+ "step": 2020
1424
+ },
1425
+ {
1426
+ "epoch": 0.2707659475140885,
1427
+ "grad_norm": 3.041241411415701,
1428
+ "learning_rate": 2.515061551429509e-05,
1429
+ "loss": 0.6911,
1430
+ "step": 2030
1431
+ },
1432
+ {
1433
+ "epoch": 0.2720997699156357,
1434
+ "grad_norm": 1.8998842731166448,
1435
+ "learning_rate": 2.5103778496723334e-05,
1436
+ "loss": 0.7129,
1437
+ "step": 2040
1438
+ },
1439
+ {
1440
+ "epoch": 0.27343359231718295,
1441
+ "grad_norm": 2.421232470526233,
1442
+ "learning_rate": 2.5056760453127242e-05,
1443
+ "loss": 0.7009,
1444
+ "step": 2050
1445
+ },
1446
+ {
1447
+ "epoch": 0.2747674147187302,
1448
+ "grad_norm": 1.6780177844017854,
1449
+ "learning_rate": 2.5009562225913385e-05,
1450
+ "loss": 0.6772,
1451
+ "step": 2060
1452
+ },
1453
+ {
1454
+ "epoch": 0.27610123712027745,
1455
+ "grad_norm": 4.807962830195143,
1456
+ "learning_rate": 2.4962184660716645e-05,
1457
+ "loss": 0.7037,
1458
+ "step": 2070
1459
+ },
1460
+ {
1461
+ "epoch": 0.2774350595218247,
1462
+ "grad_norm": 2.9514583345109253,
1463
+ "learning_rate": 2.4914628606385022e-05,
1464
+ "loss": 0.7142,
1465
+ "step": 2080
1466
+ },
1467
+ {
1468
+ "epoch": 0.2787688819233719,
1469
+ "grad_norm": 7.40404674094443,
1470
+ "learning_rate": 2.4866894914964462e-05,
1471
+ "loss": 0.6811,
1472
+ "step": 2090
1473
+ },
1474
+ {
1475
+ "epoch": 0.2801027043249191,
1476
+ "grad_norm": 2.925771960255195,
1477
+ "learning_rate": 2.481898444168357e-05,
1478
+ "loss": 0.6545,
1479
+ "step": 2100
1480
+ },
1481
+ {
1482
+ "epoch": 0.28143652672646635,
1483
+ "grad_norm": 5.444798449384416,
1484
+ "learning_rate": 2.4770898044938284e-05,
1485
+ "loss": 0.7043,
1486
+ "step": 2110
1487
+ },
1488
+ {
1489
+ "epoch": 0.28277034912801363,
1490
+ "grad_norm": 9.858110881841988,
1491
+ "learning_rate": 2.4722636586276522e-05,
1492
+ "loss": 0.665,
1493
+ "step": 2120
1494
+ },
1495
+ {
1496
+ "epoch": 0.28410417152956086,
1497
+ "grad_norm": 5.298660202543338,
1498
+ "learning_rate": 2.4674200930382712e-05,
1499
+ "loss": 0.6881,
1500
+ "step": 2130
1501
+ },
1502
+ {
1503
+ "epoch": 0.2854379939311081,
1504
+ "grad_norm": 3.705317238676521,
1505
+ "learning_rate": 2.4625591945062326e-05,
1506
+ "loss": 0.7149,
1507
+ "step": 2140
1508
+ },
1509
+ {
1510
+ "epoch": 0.2867718163326553,
1511
+ "grad_norm": 3.4572860020453855,
1512
+ "learning_rate": 2.4576810501226318e-05,
1513
+ "loss": 0.6714,
1514
+ "step": 2150
1515
+ },
1516
+ {
1517
+ "epoch": 0.28810563873420253,
1518
+ "grad_norm": 2.603696480584208,
1519
+ "learning_rate": 2.4527857472875515e-05,
1520
+ "loss": 0.6706,
1521
+ "step": 2160
1522
+ },
1523
+ {
1524
+ "epoch": 0.28943946113574975,
1525
+ "grad_norm": 3.6276512973605524,
1526
+ "learning_rate": 2.447873373708498e-05,
1527
+ "loss": 0.6915,
1528
+ "step": 2170
1529
+ },
1530
+ {
1531
+ "epoch": 0.29077328353729703,
1532
+ "grad_norm": 3.2247665616500414,
1533
+ "learning_rate": 2.4429440173988275e-05,
1534
+ "loss": 0.698,
1535
+ "step": 2180
1536
+ },
1537
+ {
1538
+ "epoch": 0.29210710593884426,
1539
+ "grad_norm": 2.902123587923174,
1540
+ "learning_rate": 2.43799776667617e-05,
1541
+ "loss": 0.6566,
1542
+ "step": 2190
1543
+ },
1544
+ {
1545
+ "epoch": 0.2934409283403915,
1546
+ "grad_norm": 2.1614031926364214,
1547
+ "learning_rate": 2.4330347101608492e-05,
1548
+ "loss": 0.7077,
1549
+ "step": 2200
1550
+ },
1551
+ {
1552
+ "epoch": 0.2947747507419387,
1553
+ "grad_norm": 9.153437367973524,
1554
+ "learning_rate": 2.428054936774289e-05,
1555
+ "loss": 0.6163,
1556
+ "step": 2210
1557
+ },
1558
+ {
1559
+ "epoch": 0.29610857314348593,
1560
+ "grad_norm": 1.9839793734516538,
1561
+ "learning_rate": 2.423058535737427e-05,
1562
+ "loss": 0.6525,
1563
+ "step": 2220
1564
+ },
1565
+ {
1566
+ "epoch": 0.29744239554503316,
1567
+ "grad_norm": 7.536682448957617,
1568
+ "learning_rate": 2.418045596569111e-05,
1569
+ "loss": 0.6892,
1570
+ "step": 2230
1571
+ },
1572
+ {
1573
+ "epoch": 0.29877621794658044,
1574
+ "grad_norm": 7.678647037780194,
1575
+ "learning_rate": 2.4130162090844976e-05,
1576
+ "loss": 0.6787,
1577
+ "step": 2240
1578
+ },
1579
+ {
1580
+ "epoch": 0.30011004034812766,
1581
+ "grad_norm": 4.791025776780419,
1582
+ "learning_rate": 2.4079704633934427e-05,
1583
+ "loss": 0.6458,
1584
+ "step": 2250
1585
+ },
1586
+ {
1587
+ "epoch": 0.3014438627496749,
1588
+ "grad_norm": 2.274268244339417,
1589
+ "learning_rate": 2.4029084498988864e-05,
1590
+ "loss": 0.6962,
1591
+ "step": 2260
1592
+ },
1593
+ {
1594
+ "epoch": 0.3027776851512221,
1595
+ "grad_norm": 2.156854296054033,
1596
+ "learning_rate": 2.3978302592952332e-05,
1597
+ "loss": 0.6852,
1598
+ "step": 2270
1599
+ },
1600
+ {
1601
+ "epoch": 0.30411150755276933,
1602
+ "grad_norm": 3.294748786762216,
1603
+ "learning_rate": 2.392735982566728e-05,
1604
+ "loss": 0.7508,
1605
+ "step": 2280
1606
+ },
1607
+ {
1608
+ "epoch": 0.30544532995431656,
1609
+ "grad_norm": 1.6409137290114766,
1610
+ "learning_rate": 2.387625710985826e-05,
1611
+ "loss": 0.7167,
1612
+ "step": 2290
1613
+ },
1614
+ {
1615
+ "epoch": 0.30677915235586384,
1616
+ "grad_norm": 3.5000955023678713,
1617
+ "learning_rate": 2.3824995361115552e-05,
1618
+ "loss": 0.6903,
1619
+ "step": 2300
1620
+ },
1621
+ {
1622
+ "epoch": 0.30811297475741106,
1623
+ "grad_norm": 1.085755861309128,
1624
+ "learning_rate": 2.3773575497878784e-05,
1625
+ "loss": 0.7182,
1626
+ "step": 2310
1627
+ },
1628
+ {
1629
+ "epoch": 0.3094467971589583,
1630
+ "grad_norm": 2.6834548847634445,
1631
+ "learning_rate": 2.372199844142048e-05,
1632
+ "loss": 0.6984,
1633
+ "step": 2320
1634
+ },
1635
+ {
1636
+ "epoch": 0.3107806195605055,
1637
+ "grad_norm": 2.493536604926798,
1638
+ "learning_rate": 2.3670265115829523e-05,
1639
+ "loss": 0.6536,
1640
+ "step": 2330
1641
+ },
1642
+ {
1643
+ "epoch": 0.31211444196205274,
1644
+ "grad_norm": 1.7721327632872044,
1645
+ "learning_rate": 2.3618376447994633e-05,
1646
+ "loss": 0.6561,
1647
+ "step": 2340
1648
+ },
1649
+ {
1650
+ "epoch": 0.31344826436359996,
1651
+ "grad_norm": 2.171596262218261,
1652
+ "learning_rate": 2.3566333367587737e-05,
1653
+ "loss": 0.7056,
1654
+ "step": 2350
1655
+ },
1656
+ {
1657
+ "epoch": 0.31478208676514724,
1658
+ "grad_norm": 36.36759112960759,
1659
+ "learning_rate": 2.3514136807047318e-05,
1660
+ "loss": 0.7015,
1661
+ "step": 2360
1662
+ },
1663
+ {
1664
+ "epoch": 0.31611590916669446,
1665
+ "grad_norm": 2.1581747337993993,
1666
+ "learning_rate": 2.3461787701561724e-05,
1667
+ "loss": 0.7136,
1668
+ "step": 2370
1669
+ },
1670
+ {
1671
+ "epoch": 0.3174497315682417,
1672
+ "grad_norm": 4.627684355473749,
1673
+ "learning_rate": 2.340928698905239e-05,
1674
+ "loss": 0.6856,
1675
+ "step": 2380
1676
+ },
1677
+ {
1678
+ "epoch": 0.3187835539697889,
1679
+ "grad_norm": 23.008362949697226,
1680
+ "learning_rate": 2.335663561015704e-05,
1681
+ "loss": 0.6535,
1682
+ "step": 2390
1683
+ },
1684
+ {
1685
+ "epoch": 0.32011737637133614,
1686
+ "grad_norm": 1.4483305067525718,
1687
+ "learning_rate": 2.3303834508212845e-05,
1688
+ "loss": 0.6929,
1689
+ "step": 2400
1690
+ },
1691
+ {
1692
+ "epoch": 0.3214511987728834,
1693
+ "grad_norm": 2.4145146906270134,
1694
+ "learning_rate": 2.325088462923951e-05,
1695
+ "loss": 0.6859,
1696
+ "step": 2410
1697
+ },
1698
+ {
1699
+ "epoch": 0.32278502117443064,
1700
+ "grad_norm": 1.6323554754616574,
1701
+ "learning_rate": 2.319778692192233e-05,
1702
+ "loss": 0.6885,
1703
+ "step": 2420
1704
+ },
1705
+ {
1706
+ "epoch": 0.32411884357597787,
1707
+ "grad_norm": 4.22840962266095,
1708
+ "learning_rate": 2.3144542337595196e-05,
1709
+ "loss": 0.6799,
1710
+ "step": 2430
1711
+ },
1712
+ {
1713
+ "epoch": 0.3254526659775251,
1714
+ "grad_norm": 9.344155395286913,
1715
+ "learning_rate": 2.3091151830223537e-05,
1716
+ "loss": 0.7125,
1717
+ "step": 2440
1718
+ },
1719
+ {
1720
+ "epoch": 0.3267864883790723,
1721
+ "grad_norm": 6.426921263330431,
1722
+ "learning_rate": 2.3037616356387237e-05,
1723
+ "loss": 0.6844,
1724
+ "step": 2450
1725
+ },
1726
+ {
1727
+ "epoch": 0.32812031078061954,
1728
+ "grad_norm": 2.508003013718542,
1729
+ "learning_rate": 2.2983936875263495e-05,
1730
+ "loss": 0.6695,
1731
+ "step": 2460
1732
+ },
1733
+ {
1734
+ "epoch": 0.3294541331821668,
1735
+ "grad_norm": 5.676502368376463,
1736
+ "learning_rate": 2.2930114348609655e-05,
1737
+ "loss": 0.656,
1738
+ "step": 2470
1739
+ },
1740
+ {
1741
+ "epoch": 0.33078795558371404,
1742
+ "grad_norm": 2.9206555111493597,
1743
+ "learning_rate": 2.2876149740745935e-05,
1744
+ "loss": 0.7357,
1745
+ "step": 2480
1746
+ },
1747
+ {
1748
+ "epoch": 0.33212177798526127,
1749
+ "grad_norm": 3.944142868858924,
1750
+ "learning_rate": 2.28220440185382e-05,
1751
+ "loss": 0.709,
1752
+ "step": 2490
1753
+ },
1754
+ {
1755
+ "epoch": 0.3334556003868085,
1756
+ "grad_norm": 15.707981311418816,
1757
+ "learning_rate": 2.2767798151380597e-05,
1758
+ "loss": 0.685,
1759
+ "step": 2500
1760
+ },
1761
+ {
1762
+ "epoch": 0.3347894227883557,
1763
+ "grad_norm": 4.850415178599164,
1764
+ "learning_rate": 2.27134131111782e-05,
1765
+ "loss": 0.7258,
1766
+ "step": 2510
1767
+ },
1768
+ {
1769
+ "epoch": 0.33612324518990294,
1770
+ "grad_norm": 1.5546940942365648,
1771
+ "learning_rate": 2.2658889872329628e-05,
1772
+ "loss": 0.6849,
1773
+ "step": 2520
1774
+ },
1775
+ {
1776
+ "epoch": 0.3374570675914502,
1777
+ "grad_norm": 7.608111830875221,
1778
+ "learning_rate": 2.2604229411709518e-05,
1779
+ "loss": 0.689,
1780
+ "step": 2530
1781
+ },
1782
+ {
1783
+ "epoch": 0.33879088999299745,
1784
+ "grad_norm": 3.5234760696769922,
1785
+ "learning_rate": 2.25494327086511e-05,
1786
+ "loss": 0.7067,
1787
+ "step": 2540
1788
+ },
1789
+ {
1790
+ "epoch": 0.34012471239454467,
1791
+ "grad_norm": 3.050573051274483,
1792
+ "learning_rate": 2.2494500744928583e-05,
1793
+ "loss": 0.6535,
1794
+ "step": 2550
1795
+ },
1796
+ {
1797
+ "epoch": 0.3414585347960919,
1798
+ "grad_norm": 1.8068660121233675,
1799
+ "learning_rate": 2.243943450473963e-05,
1800
+ "loss": 0.7171,
1801
+ "step": 2560
1802
+ },
1803
+ {
1804
+ "epoch": 0.3427923571976391,
1805
+ "grad_norm": 5.878470836242925,
1806
+ "learning_rate": 2.2384234974687658e-05,
1807
+ "loss": 0.6732,
1808
+ "step": 2570
1809
+ },
1810
+ {
1811
+ "epoch": 0.34412617959918634,
1812
+ "grad_norm": 2.419571993629191,
1813
+ "learning_rate": 2.2328903143764216e-05,
1814
+ "loss": 0.6553,
1815
+ "step": 2580
1816
+ },
1817
+ {
1818
+ "epoch": 0.3454600020007336,
1819
+ "grad_norm": 5.4457590803918166,
1820
+ "learning_rate": 2.2273440003331237e-05,
1821
+ "loss": 0.6963,
1822
+ "step": 2590
1823
+ },
1824
+ {
1825
+ "epoch": 0.34679382440228085,
1826
+ "grad_norm": 1.611198577297041,
1827
+ "learning_rate": 2.2217846547103275e-05,
1828
+ "loss": 0.7075,
1829
+ "step": 2600
1830
+ },
1831
+ {
1832
+ "epoch": 0.3481276468038281,
1833
+ "grad_norm": 1.5058788491262667,
1834
+ "learning_rate": 2.216212377112972e-05,
1835
+ "loss": 0.67,
1836
+ "step": 2610
1837
+ },
1838
+ {
1839
+ "epoch": 0.3494614692053753,
1840
+ "grad_norm": 12.998071380752632,
1841
+ "learning_rate": 2.2106272673776934e-05,
1842
+ "loss": 0.6878,
1843
+ "step": 2620
1844
+ },
1845
+ {
1846
+ "epoch": 0.3507952916069225,
1847
+ "grad_norm": 15.038084891115052,
1848
+ "learning_rate": 2.2050294255710375e-05,
1849
+ "loss": 0.7104,
1850
+ "step": 2630
1851
+ },
1852
+ {
1853
+ "epoch": 0.35212911400846975,
1854
+ "grad_norm": 2.6541518957025474,
1855
+ "learning_rate": 2.1994189519876663e-05,
1856
+ "loss": 0.6643,
1857
+ "step": 2640
1858
+ },
1859
+ {
1860
+ "epoch": 0.353462936410017,
1861
+ "grad_norm": 2.9600345573470044,
1862
+ "learning_rate": 2.19379594714856e-05,
1863
+ "loss": 0.7138,
1864
+ "step": 2650
1865
+ },
1866
+ {
1867
+ "epoch": 0.35479675881156425,
1868
+ "grad_norm": 3.6541318055900107,
1869
+ "learning_rate": 2.188160511799219e-05,
1870
+ "loss": 0.7098,
1871
+ "step": 2660
1872
+ },
1873
+ {
1874
+ "epoch": 0.3561305812131115,
1875
+ "grad_norm": 3.5267634965358887,
1876
+ "learning_rate": 2.1825127469078555e-05,
1877
+ "loss": 0.69,
1878
+ "step": 2670
1879
+ },
1880
+ {
1881
+ "epoch": 0.3574644036146587,
1882
+ "grad_norm": 4.281448471035455,
1883
+ "learning_rate": 2.1768527536635868e-05,
1884
+ "loss": 0.66,
1885
+ "step": 2680
1886
+ },
1887
+ {
1888
+ "epoch": 0.3587982260162059,
1889
+ "grad_norm": 1.8979727093671077,
1890
+ "learning_rate": 2.171180633474621e-05,
1891
+ "loss": 0.6655,
1892
+ "step": 2690
1893
+ },
1894
+ {
1895
+ "epoch": 0.3601320484177532,
1896
+ "grad_norm": 38.260538083174715,
1897
+ "learning_rate": 2.1654964879664407e-05,
1898
+ "loss": 0.6907,
1899
+ "step": 2700
1900
+ },
1901
+ {
1902
+ "epoch": 0.3614658708193004,
1903
+ "grad_norm": 4.096124324506218,
1904
+ "learning_rate": 2.1598004189799826e-05,
1905
+ "loss": 0.674,
1906
+ "step": 2710
1907
+ },
1908
+ {
1909
+ "epoch": 0.36279969322084765,
1910
+ "grad_norm": 3.481931505321646,
1911
+ "learning_rate": 2.1540925285698122e-05,
1912
+ "loss": 0.7113,
1913
+ "step": 2720
1914
+ },
1915
+ {
1916
+ "epoch": 0.3641335156223949,
1917
+ "grad_norm": 2.0982067444268684,
1918
+ "learning_rate": 2.148372919002295e-05,
1919
+ "loss": 0.696,
1920
+ "step": 2730
1921
+ },
1922
+ {
1923
+ "epoch": 0.3654673380239421,
1924
+ "grad_norm": 3.3092373308718934,
1925
+ "learning_rate": 2.142641692753765e-05,
1926
+ "loss": 0.6315,
1927
+ "step": 2740
1928
+ },
1929
+ {
1930
+ "epoch": 0.3668011604254893,
1931
+ "grad_norm": 3.186705396754706,
1932
+ "learning_rate": 2.1368989525086893e-05,
1933
+ "loss": 0.6677,
1934
+ "step": 2750
1935
+ },
1936
+ {
1937
+ "epoch": 0.3681349828270366,
1938
+ "grad_norm": 6.164987008853785,
1939
+ "learning_rate": 2.1311448011578255e-05,
1940
+ "loss": 0.6913,
1941
+ "step": 2760
1942
+ },
1943
+ {
1944
+ "epoch": 0.36946880522858383,
1945
+ "grad_norm": 1.3538856165789028,
1946
+ "learning_rate": 2.125379341796382e-05,
1947
+ "loss": 0.7127,
1948
+ "step": 2770
1949
+ },
1950
+ {
1951
+ "epoch": 0.37080262763013105,
1952
+ "grad_norm": 5.87554442135488,
1953
+ "learning_rate": 2.1196026777221684e-05,
1954
+ "loss": 0.7016,
1955
+ "step": 2780
1956
+ },
1957
+ {
1958
+ "epoch": 0.3721364500316783,
1959
+ "grad_norm": 1.7777083138299317,
1960
+ "learning_rate": 2.1138149124337448e-05,
1961
+ "loss": 0.7295,
1962
+ "step": 2790
1963
+ },
1964
+ {
1965
+ "epoch": 0.3734702724332255,
1966
+ "grad_norm": 2.385434968259032,
1967
+ "learning_rate": 2.108016149628569e-05,
1968
+ "loss": 0.6589,
1969
+ "step": 2800
1970
+ },
1971
+ {
1972
+ "epoch": 0.3748040948347727,
1973
+ "grad_norm": 6.6511179163621446,
1974
+ "learning_rate": 2.102206493201137e-05,
1975
+ "loss": 0.659,
1976
+ "step": 2810
1977
+ },
1978
+ {
1979
+ "epoch": 0.37613791723632,
1980
+ "grad_norm": 2.7897652560781876,
1981
+ "learning_rate": 2.096386047241123e-05,
1982
+ "loss": 0.6482,
1983
+ "step": 2820
1984
+ },
1985
+ {
1986
+ "epoch": 0.37747173963786723,
1987
+ "grad_norm": 2.953800669076903,
1988
+ "learning_rate": 2.0905549160315116e-05,
1989
+ "loss": 0.675,
1990
+ "step": 2830
1991
+ },
1992
+ {
1993
+ "epoch": 0.37880556203941446,
1994
+ "grad_norm": 13.84190642263868,
1995
+ "learning_rate": 2.084713204046734e-05,
1996
+ "loss": 0.6558,
1997
+ "step": 2840
1998
+ },
1999
+ {
2000
+ "epoch": 0.3801393844409617,
2001
+ "grad_norm": 4.521658802993485,
2002
+ "learning_rate": 2.078861015950793e-05,
2003
+ "loss": 0.6861,
2004
+ "step": 2850
2005
+ },
2006
+ {
2007
+ "epoch": 0.3814732068425089,
2008
+ "grad_norm": 2.5357270384377646,
2009
+ "learning_rate": 2.072998456595387e-05,
2010
+ "loss": 0.6781,
2011
+ "step": 2860
2012
+ },
2013
+ {
2014
+ "epoch": 0.38280702924405613,
2015
+ "grad_norm": 1.694744102718741,
2016
+ "learning_rate": 2.0671256310180334e-05,
2017
+ "loss": 0.6923,
2018
+ "step": 2870
2019
+ },
2020
+ {
2021
+ "epoch": 0.3841408516456034,
2022
+ "grad_norm": 2.335196387059918,
2023
+ "learning_rate": 2.0612426444401874e-05,
2024
+ "loss": 0.6473,
2025
+ "step": 2880
2026
+ },
2027
+ {
2028
+ "epoch": 0.38547467404715063,
2029
+ "grad_norm": 4.2888708297621605,
2030
+ "learning_rate": 2.0553496022653535e-05,
2031
+ "loss": 0.7116,
2032
+ "step": 2890
2033
+ },
2034
+ {
2035
+ "epoch": 0.38680849644869786,
2036
+ "grad_norm": 2.586725028631528,
2037
+ "learning_rate": 2.0494466100772006e-05,
2038
+ "loss": 0.6388,
2039
+ "step": 2900
2040
+ },
2041
+ {
2042
+ "epoch": 0.3881423188502451,
2043
+ "grad_norm": 6.7632601439437625,
2044
+ "learning_rate": 2.0435337736376677e-05,
2045
+ "loss": 0.6663,
2046
+ "step": 2910
2047
+ },
2048
+ {
2049
+ "epoch": 0.3894761412517923,
2050
+ "grad_norm": 5.0897618694848115,
2051
+ "learning_rate": 2.03761119888507e-05,
2052
+ "loss": 0.686,
2053
+ "step": 2920
2054
+ },
2055
+ {
2056
+ "epoch": 0.39080996365333953,
2057
+ "grad_norm": 3.84563964047397,
2058
+ "learning_rate": 2.031678991932201e-05,
2059
+ "loss": 0.6763,
2060
+ "step": 2930
2061
+ },
2062
+ {
2063
+ "epoch": 0.3921437860548868,
2064
+ "grad_norm": 1.3602515715842554,
2065
+ "learning_rate": 2.0257372590644314e-05,
2066
+ "loss": 0.7278,
2067
+ "step": 2940
2068
+ },
2069
+ {
2070
+ "epoch": 0.39347760845643404,
2071
+ "grad_norm": 3.0517076558378293,
2072
+ "learning_rate": 2.0197861067378044e-05,
2073
+ "loss": 0.6734,
2074
+ "step": 2950
2075
+ },
2076
+ {
2077
+ "epoch": 0.39481143085798126,
2078
+ "grad_norm": 2.0564006437177826,
2079
+ "learning_rate": 2.0138256415771275e-05,
2080
+ "loss": 0.6861,
2081
+ "step": 2960
2082
+ },
2083
+ {
2084
+ "epoch": 0.3961452532595285,
2085
+ "grad_norm": 3.0150939790113522,
2086
+ "learning_rate": 2.0078559703740654e-05,
2087
+ "loss": 0.6745,
2088
+ "step": 2970
2089
+ },
2090
+ {
2091
+ "epoch": 0.3974790756610757,
2092
+ "grad_norm": 2.3696016899226495,
2093
+ "learning_rate": 2.0018772000852216e-05,
2094
+ "loss": 0.6812,
2095
+ "step": 2980
2096
+ },
2097
+ {
2098
+ "epoch": 0.398812898062623,
2099
+ "grad_norm": 1.8364192249492444,
2100
+ "learning_rate": 1.9958894378302265e-05,
2101
+ "loss": 0.6631,
2102
+ "step": 2990
2103
+ },
2104
+ {
2105
+ "epoch": 0.4001467204641702,
2106
+ "grad_norm": 2.3337386891806595,
2107
+ "learning_rate": 1.989892790889817e-05,
2108
+ "loss": 0.7276,
2109
+ "step": 3000
2110
+ },
2111
+ {
2112
+ "epoch": 0.40148054286571744,
2113
+ "grad_norm": 2.225372210455397,
2114
+ "learning_rate": 1.9838873667039134e-05,
2115
+ "loss": 0.6782,
2116
+ "step": 3010
2117
+ },
2118
+ {
2119
+ "epoch": 0.40281436526726466,
2120
+ "grad_norm": 4.9169439894919265,
2121
+ "learning_rate": 1.9778732728696937e-05,
2122
+ "loss": 0.6787,
2123
+ "step": 3020
2124
+ },
2125
+ {
2126
+ "epoch": 0.4041481876688119,
2127
+ "grad_norm": 2.335178009370853,
2128
+ "learning_rate": 1.9718506171396694e-05,
2129
+ "loss": 0.7036,
2130
+ "step": 3030
2131
+ },
2132
+ {
2133
+ "epoch": 0.4054820100703591,
2134
+ "grad_norm": 2.2036269677969327,
2135
+ "learning_rate": 1.965819507419751e-05,
2136
+ "loss": 0.6649,
2137
+ "step": 3040
2138
+ },
2139
+ {
2140
+ "epoch": 0.4068158324719064,
2141
+ "grad_norm": 4.7578038809060885,
2142
+ "learning_rate": 1.9597800517673165e-05,
2143
+ "loss": 0.6682,
2144
+ "step": 3050
2145
+ },
2146
+ {
2147
+ "epoch": 0.4081496548734536,
2148
+ "grad_norm": 16.429386587829253,
2149
+ "learning_rate": 1.9537323583892753e-05,
2150
+ "loss": 0.6875,
2151
+ "step": 3060
2152
+ },
2153
+ {
2154
+ "epoch": 0.40948347727500084,
2155
+ "grad_norm": 2.9803066784550953,
2156
+ "learning_rate": 1.9476765356401304e-05,
2157
+ "loss": 0.7022,
2158
+ "step": 3070
2159
+ },
2160
+ {
2161
+ "epoch": 0.41081729967654806,
2162
+ "grad_norm": 1.7088579752363298,
2163
+ "learning_rate": 1.9416126920200344e-05,
2164
+ "loss": 0.7335,
2165
+ "step": 3080
2166
+ },
2167
+ {
2168
+ "epoch": 0.4121511220780953,
2169
+ "grad_norm": 1.8641120769753927,
2170
+ "learning_rate": 1.9355409361728482e-05,
2171
+ "loss": 0.7195,
2172
+ "step": 3090
2173
+ },
2174
+ {
2175
+ "epoch": 0.4134849444796425,
2176
+ "grad_norm": 2.887584179728711,
2177
+ "learning_rate": 1.9294613768841932e-05,
2178
+ "loss": 0.6694,
2179
+ "step": 3100
2180
+ },
2181
+ {
2182
+ "epoch": 0.4148187668811898,
2183
+ "grad_norm": 3.3517946816430486,
2184
+ "learning_rate": 1.9233741230795022e-05,
2185
+ "loss": 0.6763,
2186
+ "step": 3110
2187
+ },
2188
+ {
2189
+ "epoch": 0.416152589282737,
2190
+ "grad_norm": 15.627345038869116,
2191
+ "learning_rate": 1.9172792838220686e-05,
2192
+ "loss": 0.6508,
2193
+ "step": 3120
2194
+ },
2195
+ {
2196
+ "epoch": 0.41748641168428424,
2197
+ "grad_norm": 8.325170511784908,
2198
+ "learning_rate": 1.9111769683110914e-05,
2199
+ "loss": 0.6638,
2200
+ "step": 3130
2201
+ },
2202
+ {
2203
+ "epoch": 0.41882023408583147,
2204
+ "grad_norm": 2.2227223257423576,
2205
+ "learning_rate": 1.905067285879719e-05,
2206
+ "loss": 0.6702,
2207
+ "step": 3140
2208
+ },
2209
+ {
2210
+ "epoch": 0.4201540564873787,
2211
+ "grad_norm": 2.151930199856048,
2212
+ "learning_rate": 1.8989503459930908e-05,
2213
+ "loss": 0.6702,
2214
+ "step": 3150
2215
+ },
2216
+ {
2217
+ "epoch": 0.4214878788889259,
2218
+ "grad_norm": 2.0853572424146445,
2219
+ "learning_rate": 1.892826258246376e-05,
2220
+ "loss": 0.6687,
2221
+ "step": 3160
2222
+ },
2223
+ {
2224
+ "epoch": 0.4228217012904732,
2225
+ "grad_norm": 12.723619283478206,
2226
+ "learning_rate": 1.886695132362808e-05,
2227
+ "loss": 0.6736,
2228
+ "step": 3170
2229
+ },
2230
+ {
2231
+ "epoch": 0.4241555236920204,
2232
+ "grad_norm": 2.2341551772603223,
2233
+ "learning_rate": 1.8805570781917228e-05,
2234
+ "loss": 0.6979,
2235
+ "step": 3180
2236
+ },
2237
+ {
2238
+ "epoch": 0.42548934609356764,
2239
+ "grad_norm": 2.7010833438865482,
2240
+ "learning_rate": 1.8744122057065856e-05,
2241
+ "loss": 0.6589,
2242
+ "step": 3190
2243
+ },
2244
+ {
2245
+ "epoch": 0.42682316849511487,
2246
+ "grad_norm": 3.764700495141046,
2247
+ "learning_rate": 1.868260625003024e-05,
2248
+ "loss": 0.6951,
2249
+ "step": 3200
2250
+ },
2251
+ {
2252
+ "epoch": 0.4281569908966621,
2253
+ "grad_norm": 1.5720755325934035,
2254
+ "learning_rate": 1.8621024462968553e-05,
2255
+ "loss": 0.6837,
2256
+ "step": 3210
2257
+ },
2258
+ {
2259
+ "epoch": 0.4294908132982093,
2260
+ "grad_norm": 3.9200536138410382,
2261
+ "learning_rate": 1.85593777992211e-05,
2262
+ "loss": 0.7135,
2263
+ "step": 3220
2264
+ },
2265
+ {
2266
+ "epoch": 0.4308246356997566,
2267
+ "grad_norm": 2.1410220258311203,
2268
+ "learning_rate": 1.849766736329056e-05,
2269
+ "loss": 0.6935,
2270
+ "step": 3230
2271
+ },
2272
+ {
2273
+ "epoch": 0.4321584581013038,
2274
+ "grad_norm": 3.306117548750314,
2275
+ "learning_rate": 1.8435894260822208e-05,
2276
+ "loss": 0.6947,
2277
+ "step": 3240
2278
+ },
2279
+ {
2280
+ "epoch": 0.43349228050285105,
2281
+ "grad_norm": 3.906418209268947,
2282
+ "learning_rate": 1.8374059598584084e-05,
2283
+ "loss": 0.657,
2284
+ "step": 3250
2285
+ },
2286
+ {
2287
+ "epoch": 0.43482610290439827,
2288
+ "grad_norm": 1.3332307398585346,
2289
+ "learning_rate": 1.831216448444717e-05,
2290
+ "loss": 0.6456,
2291
+ "step": 3260
2292
+ },
2293
+ {
2294
+ "epoch": 0.4361599253059455,
2295
+ "grad_norm": 20.93039175046958,
2296
+ "learning_rate": 1.8250210027365562e-05,
2297
+ "loss": 0.7233,
2298
+ "step": 3270
2299
+ },
2300
+ {
2301
+ "epoch": 0.4374937477074928,
2302
+ "grad_norm": 4.495787238072399,
2303
+ "learning_rate": 1.818819733735657e-05,
2304
+ "loss": 0.7438,
2305
+ "step": 3280
2306
+ },
2307
+ {
2308
+ "epoch": 0.43882757010904,
2309
+ "grad_norm": 2.8059359776214166,
2310
+ "learning_rate": 1.812612752548084e-05,
2311
+ "loss": 0.654,
2312
+ "step": 3290
2313
+ },
2314
+ {
2315
+ "epoch": 0.4401613925105872,
2316
+ "grad_norm": 2.226805342431407,
2317
+ "learning_rate": 1.806400170382246e-05,
2318
+ "loss": 0.6823,
2319
+ "step": 3300
2320
+ },
2321
+ {
2322
+ "epoch": 0.44149521491213445,
2323
+ "grad_norm": 1.5989661035338567,
2324
+ "learning_rate": 1.8001820985469026e-05,
2325
+ "loss": 0.6665,
2326
+ "step": 3310
2327
+ },
2328
+ {
2329
+ "epoch": 0.4428290373136817,
2330
+ "grad_norm": 44.45501423929167,
2331
+ "learning_rate": 1.7939586484491704e-05,
2332
+ "loss": 0.6901,
2333
+ "step": 3320
2334
+ },
2335
+ {
2336
+ "epoch": 0.4441628597152289,
2337
+ "grad_norm": 3.526154343998946,
2338
+ "learning_rate": 1.787729931592525e-05,
2339
+ "loss": 0.734,
2340
+ "step": 3330
2341
+ },
2342
+ {
2343
+ "epoch": 0.4454966821167762,
2344
+ "grad_norm": 1.4841336505836662,
2345
+ "learning_rate": 1.781496059574807e-05,
2346
+ "loss": 0.7174,
2347
+ "step": 3340
2348
+ },
2349
+ {
2350
+ "epoch": 0.4468305045183234,
2351
+ "grad_norm": 1.7136084206360864,
2352
+ "learning_rate": 1.7752571440862178e-05,
2353
+ "loss": 0.6697,
2354
+ "step": 3350
2355
+ },
2356
+ {
2357
+ "epoch": 0.4481643269198706,
2358
+ "grad_norm": 1.1822744135781105,
2359
+ "learning_rate": 1.7690132969073223e-05,
2360
+ "loss": 0.6754,
2361
+ "step": 3360
2362
+ },
2363
+ {
2364
+ "epoch": 0.44949814932141785,
2365
+ "grad_norm": 3.1896570424325943,
2366
+ "learning_rate": 1.7627646299070457e-05,
2367
+ "loss": 0.7047,
2368
+ "step": 3370
2369
+ },
2370
+ {
2371
+ "epoch": 0.4508319717229651,
2372
+ "grad_norm": 3.246445692977348,
2373
+ "learning_rate": 1.7565112550406663e-05,
2374
+ "loss": 0.6758,
2375
+ "step": 3380
2376
+ },
2377
+ {
2378
+ "epoch": 0.4521657941245123,
2379
+ "grad_norm": 1.8894085083670678,
2380
+ "learning_rate": 1.7502532843478134e-05,
2381
+ "loss": 0.626,
2382
+ "step": 3390
2383
+ },
2384
+ {
2385
+ "epoch": 0.4534996165260596,
2386
+ "grad_norm": 1.9836472505954585,
2387
+ "learning_rate": 1.743990829950458e-05,
2388
+ "loss": 0.7355,
2389
+ "step": 3400
2390
+ },
2391
+ {
2392
+ "epoch": 0.4548334389276068,
2393
+ "grad_norm": 2.0265270785939817,
2394
+ "learning_rate": 1.737724004050903e-05,
2395
+ "loss": 0.6602,
2396
+ "step": 3410
2397
+ },
2398
+ {
2399
+ "epoch": 0.45616726132915403,
2400
+ "grad_norm": 37.752331390135694,
2401
+ "learning_rate": 1.731452918929774e-05,
2402
+ "loss": 0.7096,
2403
+ "step": 3420
2404
+ },
2405
+ {
2406
+ "epoch": 0.45750108373070125,
2407
+ "grad_norm": 3.069288243097065,
2408
+ "learning_rate": 1.7251776869440097e-05,
2409
+ "loss": 0.6894,
2410
+ "step": 3430
2411
+ },
2412
+ {
2413
+ "epoch": 0.4588349061322485,
2414
+ "grad_norm": 3.806122244959967,
2415
+ "learning_rate": 1.718898420524845e-05,
2416
+ "loss": 0.6717,
2417
+ "step": 3440
2418
+ },
2419
+ {
2420
+ "epoch": 0.4601687285337957,
2421
+ "grad_norm": 5.422119486900359,
2422
+ "learning_rate": 1.7126152321757985e-05,
2423
+ "loss": 0.7123,
2424
+ "step": 3450
2425
+ },
2426
+ {
2427
+ "epoch": 0.461502550935343,
2428
+ "grad_norm": 8.913213423628779,
2429
+ "learning_rate": 1.7063282344706577e-05,
2430
+ "loss": 0.6943,
2431
+ "step": 3460
2432
+ },
2433
+ {
2434
+ "epoch": 0.4628363733368902,
2435
+ "grad_norm": 2.36239995996906,
2436
+ "learning_rate": 1.7000375400514602e-05,
2437
+ "loss": 0.6472,
2438
+ "step": 3470
2439
+ },
2440
+ {
2441
+ "epoch": 0.46417019573843743,
2442
+ "grad_norm": 2.256697435474481,
2443
+ "learning_rate": 1.693743261626476e-05,
2444
+ "loss": 0.6981,
2445
+ "step": 3480
2446
+ },
2447
+ {
2448
+ "epoch": 0.46550401813998465,
2449
+ "grad_norm": 4.819999301324691,
2450
+ "learning_rate": 1.68744551196819e-05,
2451
+ "loss": 0.6436,
2452
+ "step": 3490
2453
+ },
2454
+ {
2455
+ "epoch": 0.4668378405415319,
2456
+ "grad_norm": 1.6840417535820236,
2457
+ "learning_rate": 1.6811444039112787e-05,
2458
+ "loss": 0.6993,
2459
+ "step": 3500
2460
+ },
2461
+ {
2462
+ "epoch": 0.4681716629430791,
2463
+ "grad_norm": 2.544379480255388,
2464
+ "learning_rate": 1.6748400503505905e-05,
2465
+ "loss": 0.6722,
2466
+ "step": 3510
2467
+ },
2468
+ {
2469
+ "epoch": 0.4695054853446264,
2470
+ "grad_norm": 3.7390416566783493,
2471
+ "learning_rate": 1.6685325642391223e-05,
2472
+ "loss": 0.6785,
2473
+ "step": 3520
2474
+ },
2475
+ {
2476
+ "epoch": 0.4708393077461736,
2477
+ "grad_norm": 3.855083968142357,
2478
+ "learning_rate": 1.662222058585996e-05,
2479
+ "loss": 0.6952,
2480
+ "step": 3530
2481
+ },
2482
+ {
2483
+ "epoch": 0.47217313014772083,
2484
+ "grad_norm": 9.342710099035981,
2485
+ "learning_rate": 1.6559086464544334e-05,
2486
+ "loss": 0.6772,
2487
+ "step": 3540
2488
+ },
2489
+ {
2490
+ "epoch": 0.47350695254926806,
2491
+ "grad_norm": 1.7318075611168393,
2492
+ "learning_rate": 1.6495924409597305e-05,
2493
+ "loss": 0.6757,
2494
+ "step": 3550
2495
+ },
2496
+ {
2497
+ "epoch": 0.4748407749508153,
2498
+ "grad_norm": 5.54593274347221,
2499
+ "learning_rate": 1.6432735552672317e-05,
2500
+ "loss": 0.704,
2501
+ "step": 3560
2502
+ },
2503
+ {
2504
+ "epoch": 0.4761745973523625,
2505
+ "grad_norm": 9.782086516371113,
2506
+ "learning_rate": 1.636952102590301e-05,
2507
+ "loss": 0.6993,
2508
+ "step": 3570
2509
+ },
2510
+ {
2511
+ "epoch": 0.4775084197539098,
2512
+ "grad_norm": 3.0266652780549257,
2513
+ "learning_rate": 1.630628196188295e-05,
2514
+ "loss": 0.7198,
2515
+ "step": 3580
2516
+ },
2517
+ {
2518
+ "epoch": 0.478842242155457,
2519
+ "grad_norm": 5.519301272688082,
2520
+ "learning_rate": 1.6243019493645315e-05,
2521
+ "loss": 0.7222,
2522
+ "step": 3590
2523
+ },
2524
+ {
2525
+ "epoch": 0.48017606455700423,
2526
+ "grad_norm": 2.2909343474167008,
2527
+ "learning_rate": 1.617973475464262e-05,
2528
+ "loss": 0.6569,
2529
+ "step": 3600
2530
+ },
2531
+ {
2532
+ "epoch": 0.48150988695855146,
2533
+ "grad_norm": 7.436045528492052,
2534
+ "learning_rate": 1.6116428878726396e-05,
2535
+ "loss": 0.6938,
2536
+ "step": 3610
2537
+ },
2538
+ {
2539
+ "epoch": 0.4828437093600987,
2540
+ "grad_norm": 2.658185186612875,
2541
+ "learning_rate": 1.6053103000126874e-05,
2542
+ "loss": 0.667,
2543
+ "step": 3620
2544
+ },
2545
+ {
2546
+ "epoch": 0.48417753176164596,
2547
+ "grad_norm": 3.1099903513767915,
2548
+ "learning_rate": 1.598975825343267e-05,
2549
+ "loss": 0.6666,
2550
+ "step": 3630
2551
+ },
2552
+ {
2553
+ "epoch": 0.4855113541631932,
2554
+ "grad_norm": 2.844884960349376,
2555
+ "learning_rate": 1.5926395773570447e-05,
2556
+ "loss": 0.6715,
2557
+ "step": 3640
2558
+ },
2559
+ {
2560
+ "epoch": 0.4868451765647404,
2561
+ "grad_norm": 3.5867549321919827,
2562
+ "learning_rate": 1.5863016695784604e-05,
2563
+ "loss": 0.6938,
2564
+ "step": 3650
2565
+ },
2566
+ {
2567
+ "epoch": 0.48817899896628764,
2568
+ "grad_norm": 6.440074516955184,
2569
+ "learning_rate": 1.5799622155616887e-05,
2570
+ "loss": 0.6575,
2571
+ "step": 3660
2572
+ },
2573
+ {
2574
+ "epoch": 0.48951282136783486,
2575
+ "grad_norm": 5.454278599829281,
2576
+ "learning_rate": 1.5736213288886112e-05,
2577
+ "loss": 0.6925,
2578
+ "step": 3670
2579
+ },
2580
+ {
2581
+ "epoch": 0.4908466437693821,
2582
+ "grad_norm": 1.3038190132520797,
2583
+ "learning_rate": 1.567279123166776e-05,
2584
+ "loss": 0.6827,
2585
+ "step": 3680
2586
+ },
2587
+ {
2588
+ "epoch": 0.49218046617092936,
2589
+ "grad_norm": 2.2843915654324127,
2590
+ "learning_rate": 1.560935712027364e-05,
2591
+ "loss": 0.7272,
2592
+ "step": 3690
2593
+ },
2594
+ {
2595
+ "epoch": 0.4935142885724766,
2596
+ "grad_norm": 5.235529803525943,
2597
+ "learning_rate": 1.5545912091231543e-05,
2598
+ "loss": 0.6859,
2599
+ "step": 3700
2600
+ },
2601
+ {
2602
+ "epoch": 0.4948481109740238,
2603
+ "grad_norm": 67.2928715868646,
2604
+ "learning_rate": 1.548245728126486e-05,
2605
+ "loss": 0.6755,
2606
+ "step": 3710
2607
+ },
2608
+ {
2609
+ "epoch": 0.49618193337557104,
2610
+ "grad_norm": 1.8808319311625479,
2611
+ "learning_rate": 1.5418993827272224e-05,
2612
+ "loss": 0.6827,
2613
+ "step": 3720
2614
+ },
2615
+ {
2616
+ "epoch": 0.49751575577711826,
2617
+ "grad_norm": 3.665691607326232,
2618
+ "learning_rate": 1.5355522866307144e-05,
2619
+ "loss": 0.729,
2620
+ "step": 3730
2621
+ },
2622
+ {
2623
+ "epoch": 0.4988495781786655,
2624
+ "grad_norm": 3.6256043818215864,
2625
+ "learning_rate": 1.529204553555762e-05,
2626
+ "loss": 0.674,
2627
+ "step": 3740
2628
+ },
2629
+ {
2630
+ "epoch": 0.5001834005802127,
2631
+ "grad_norm": 5.335543393148527,
2632
+ "learning_rate": 1.522856297232579e-05,
2633
+ "loss": 0.683,
2634
+ "step": 3750
2635
+ },
2636
+ {
2637
+ "epoch": 0.5015172229817599,
2638
+ "grad_norm": 1.768660779664017,
2639
+ "learning_rate": 1.5165076314007529e-05,
2640
+ "loss": 0.6607,
2641
+ "step": 3760
2642
+ },
2643
+ {
2644
+ "epoch": 0.5028510453833072,
2645
+ "grad_norm": 3.6829482612300515,
2646
+ "learning_rate": 1.5101586698072095e-05,
2647
+ "loss": 0.6778,
2648
+ "step": 3770
2649
+ },
2650
+ {
2651
+ "epoch": 0.5041848677848545,
2652
+ "grad_norm": 2.016271324919982,
2653
+ "learning_rate": 1.5038095262041725e-05,
2654
+ "loss": 0.6758,
2655
+ "step": 3780
2656
+ },
2657
+ {
2658
+ "epoch": 0.5055186901864017,
2659
+ "grad_norm": 4.828462163022634,
2660
+ "learning_rate": 1.4974603143471268e-05,
2661
+ "loss": 0.7128,
2662
+ "step": 3790
2663
+ },
2664
+ {
2665
+ "epoch": 0.506852512587949,
2666
+ "grad_norm": 2.1222493938328792,
2667
+ "learning_rate": 1.4911111479927804e-05,
2668
+ "loss": 0.6626,
2669
+ "step": 3800
2670
+ },
2671
+ {
2672
+ "epoch": 0.5081863349894962,
2673
+ "grad_norm": 7.554329738410342,
2674
+ "learning_rate": 1.4847621408970266e-05,
2675
+ "loss": 0.691,
2676
+ "step": 3810
2677
+ },
2678
+ {
2679
+ "epoch": 0.5095201573910434,
2680
+ "grad_norm": 1.1718744405109343,
2681
+ "learning_rate": 1.4784134068129043e-05,
2682
+ "loss": 0.7085,
2683
+ "step": 3820
2684
+ },
2685
+ {
2686
+ "epoch": 0.5108539797925906,
2687
+ "grad_norm": 4.3974979135617875,
2688
+ "learning_rate": 1.4720650594885614e-05,
2689
+ "loss": 0.6888,
2690
+ "step": 3830
2691
+ },
2692
+ {
2693
+ "epoch": 0.5121878021941378,
2694
+ "grad_norm": 3.056822672044886,
2695
+ "learning_rate": 1.4657172126652167e-05,
2696
+ "loss": 0.7112,
2697
+ "step": 3840
2698
+ },
2699
+ {
2700
+ "epoch": 0.5135216245956851,
2701
+ "grad_norm": 2.9519833573179786,
2702
+ "learning_rate": 1.459369980075121e-05,
2703
+ "loss": 0.7143,
2704
+ "step": 3850
2705
+ },
2706
+ {
2707
+ "epoch": 0.5148554469972323,
2708
+ "grad_norm": 2.7879172145478277,
2709
+ "learning_rate": 1.4530234754395207e-05,
2710
+ "loss": 0.73,
2711
+ "step": 3860
2712
+ },
2713
+ {
2714
+ "epoch": 0.5161892693987795,
2715
+ "grad_norm": 2.689401796470298,
2716
+ "learning_rate": 1.4466778124666192e-05,
2717
+ "loss": 0.6802,
2718
+ "step": 3870
2719
+ },
2720
+ {
2721
+ "epoch": 0.5175230918003267,
2722
+ "grad_norm": 2.1741383786138297,
2723
+ "learning_rate": 1.4403331048495404e-05,
2724
+ "loss": 0.6404,
2725
+ "step": 3880
2726
+ },
2727
+ {
2728
+ "epoch": 0.5188569142018741,
2729
+ "grad_norm": 6.3241834633627985,
2730
+ "learning_rate": 1.4339894662642914e-05,
2731
+ "loss": 0.678,
2732
+ "step": 3890
2733
+ },
2734
+ {
2735
+ "epoch": 0.5201907366034213,
2736
+ "grad_norm": 6.491040731017525,
2737
+ "learning_rate": 1.4276470103677257e-05,
2738
+ "loss": 0.6339,
2739
+ "step": 3900
2740
+ },
2741
+ {
2742
+ "epoch": 0.5215245590049685,
2743
+ "grad_norm": 2.043861558768385,
2744
+ "learning_rate": 1.4213058507955072e-05,
2745
+ "loss": 0.6655,
2746
+ "step": 3910
2747
+ },
2748
+ {
2749
+ "epoch": 0.5228583814065157,
2750
+ "grad_norm": 1.6513869669590875,
2751
+ "learning_rate": 1.4149661011600734e-05,
2752
+ "loss": 0.688,
2753
+ "step": 3920
2754
+ },
2755
+ {
2756
+ "epoch": 0.524192203808063,
2757
+ "grad_norm": 3.541890141037579,
2758
+ "learning_rate": 1.4086278750486017e-05,
2759
+ "loss": 0.6899,
2760
+ "step": 3930
2761
+ },
2762
+ {
2763
+ "epoch": 0.5255260262096102,
2764
+ "grad_norm": 6.194603743545731,
2765
+ "learning_rate": 1.4022912860209709e-05,
2766
+ "loss": 0.669,
2767
+ "step": 3940
2768
+ },
2769
+ {
2770
+ "epoch": 0.5268598486111574,
2771
+ "grad_norm": 2.0486629917737873,
2772
+ "learning_rate": 1.3959564476077308e-05,
2773
+ "loss": 0.7003,
2774
+ "step": 3950
2775
+ },
2776
+ {
2777
+ "epoch": 0.5281936710127046,
2778
+ "grad_norm": 2.7696033032598693,
2779
+ "learning_rate": 1.389623473308065e-05,
2780
+ "loss": 0.7083,
2781
+ "step": 3960
2782
+ },
2783
+ {
2784
+ "epoch": 0.5295274934142519,
2785
+ "grad_norm": 2.4390436810800566,
2786
+ "learning_rate": 1.3832924765877587e-05,
2787
+ "loss": 0.6827,
2788
+ "step": 3970
2789
+ },
2790
+ {
2791
+ "epoch": 0.5308613158157991,
2792
+ "grad_norm": 2.152867968184263,
2793
+ "learning_rate": 1.3769635708771654e-05,
2794
+ "loss": 0.6739,
2795
+ "step": 3980
2796
+ },
2797
+ {
2798
+ "epoch": 0.5321951382173463,
2799
+ "grad_norm": 7.177870428132333,
2800
+ "learning_rate": 1.3706368695691745e-05,
2801
+ "loss": 0.6652,
2802
+ "step": 3990
2803
+ },
2804
+ {
2805
+ "epoch": 0.5335289606188935,
2806
+ "grad_norm": 2.2804181398144987,
2807
+ "learning_rate": 1.3643124860171801e-05,
2808
+ "loss": 0.6592,
2809
+ "step": 4000
2810
+ },
2811
+ {
2812
+ "epoch": 0.5348627830204409,
2813
+ "grad_norm": 2.0999660171302312,
2814
+ "learning_rate": 1.35799053353305e-05,
2815
+ "loss": 0.6905,
2816
+ "step": 4010
2817
+ },
2818
+ {
2819
+ "epoch": 0.5361966054219881,
2820
+ "grad_norm": 1.8299441945056365,
2821
+ "learning_rate": 1.3516711253850949e-05,
2822
+ "loss": 0.6714,
2823
+ "step": 4020
2824
+ },
2825
+ {
2826
+ "epoch": 0.5375304278235353,
2827
+ "grad_norm": 5.370538678704669,
2828
+ "learning_rate": 1.3453543747960393e-05,
2829
+ "loss": 0.7027,
2830
+ "step": 4030
2831
+ },
2832
+ {
2833
+ "epoch": 0.5388642502250826,
2834
+ "grad_norm": 5.933915833300909,
2835
+ "learning_rate": 1.3390403949409943e-05,
2836
+ "loss": 0.7126,
2837
+ "step": 4040
2838
+ },
2839
+ {
2840
+ "epoch": 0.5401980726266298,
2841
+ "grad_norm": 2.927835677762918,
2842
+ "learning_rate": 1.3327292989454273e-05,
2843
+ "loss": 0.652,
2844
+ "step": 4050
2845
+ },
2846
+ {
2847
+ "epoch": 0.541531895028177,
2848
+ "grad_norm": 2.3060298824971297,
2849
+ "learning_rate": 1.3264211998831374e-05,
2850
+ "loss": 0.6363,
2851
+ "step": 4060
2852
+ },
2853
+ {
2854
+ "epoch": 0.5428657174297242,
2855
+ "grad_norm": 8.939144383392009,
2856
+ "learning_rate": 1.3201162107742285e-05,
2857
+ "loss": 0.6518,
2858
+ "step": 4070
2859
+ },
2860
+ {
2861
+ "epoch": 0.5441995398312715,
2862
+ "grad_norm": 2.1217426691379253,
2863
+ "learning_rate": 1.3138144445830841e-05,
2864
+ "loss": 0.6764,
2865
+ "step": 4080
2866
+ },
2867
+ {
2868
+ "epoch": 0.5455333622328187,
2869
+ "grad_norm": 3.455459703040737,
2870
+ "learning_rate": 1.3075160142163442e-05,
2871
+ "loss": 0.669,
2872
+ "step": 4090
2873
+ },
2874
+ {
2875
+ "epoch": 0.5468671846343659,
2876
+ "grad_norm": 1.0556174805250833,
2877
+ "learning_rate": 1.3012210325208818e-05,
2878
+ "loss": 0.6653,
2879
+ "step": 4100
2880
+ },
2881
+ {
2882
+ "epoch": 0.5482010070359131,
2883
+ "grad_norm": 2.5622275603077846,
2884
+ "learning_rate": 1.2949296122817813e-05,
2885
+ "loss": 0.6999,
2886
+ "step": 4110
2887
+ },
2888
+ {
2889
+ "epoch": 0.5495348294374603,
2890
+ "grad_norm": 1.311339273822971,
2891
+ "learning_rate": 1.2886418662203174e-05,
2892
+ "loss": 0.6542,
2893
+ "step": 4120
2894
+ },
2895
+ {
2896
+ "epoch": 0.5508686518390077,
2897
+ "grad_norm": 2.0314733529170548,
2898
+ "learning_rate": 1.282357906991936e-05,
2899
+ "loss": 0.7086,
2900
+ "step": 4130
2901
+ },
2902
+ {
2903
+ "epoch": 0.5522024742405549,
2904
+ "grad_norm": 1.3992167597958347,
2905
+ "learning_rate": 1.276077847184236e-05,
2906
+ "loss": 0.6647,
2907
+ "step": 4140
2908
+ },
2909
+ {
2910
+ "epoch": 0.5535362966421021,
2911
+ "grad_norm": 2.8142228677605243,
2912
+ "learning_rate": 1.2698017993149504e-05,
2913
+ "loss": 0.6679,
2914
+ "step": 4150
2915
+ },
2916
+ {
2917
+ "epoch": 0.5548701190436494,
2918
+ "grad_norm": 53.450181243846124,
2919
+ "learning_rate": 1.2635298758299336e-05,
2920
+ "loss": 0.6677,
2921
+ "step": 4160
2922
+ },
2923
+ {
2924
+ "epoch": 0.5562039414451966,
2925
+ "grad_norm": 34.73442771588442,
2926
+ "learning_rate": 1.2572621891011426e-05,
2927
+ "loss": 0.6358,
2928
+ "step": 4170
2929
+ },
2930
+ {
2931
+ "epoch": 0.5575377638467438,
2932
+ "grad_norm": 2.25939959543013,
2933
+ "learning_rate": 1.2509988514246272e-05,
2934
+ "loss": 0.7153,
2935
+ "step": 4180
2936
+ },
2937
+ {
2938
+ "epoch": 0.558871586248291,
2939
+ "grad_norm": 28.104396750316823,
2940
+ "learning_rate": 1.2447399750185166e-05,
2941
+ "loss": 0.7164,
2942
+ "step": 4190
2943
+ },
2944
+ {
2945
+ "epoch": 0.5602054086498383,
2946
+ "grad_norm": 2.2840282297009913,
2947
+ "learning_rate": 1.2384856720210086e-05,
2948
+ "loss": 0.6526,
2949
+ "step": 4200
2950
+ },
2951
+ {
2952
+ "epoch": 0.5615392310513855,
2953
+ "grad_norm": 1.6880048773636342,
2954
+ "learning_rate": 1.2322360544883608e-05,
2955
+ "loss": 0.6928,
2956
+ "step": 4210
2957
+ },
2958
+ {
2959
+ "epoch": 0.5628730534529327,
2960
+ "grad_norm": 2.507578235748276,
2961
+ "learning_rate": 1.2259912343928831e-05,
2962
+ "loss": 0.6805,
2963
+ "step": 4220
2964
+ },
2965
+ {
2966
+ "epoch": 0.5642068758544799,
2967
+ "grad_norm": 2.2070032708749183,
2968
+ "learning_rate": 1.2197513236209312e-05,
2969
+ "loss": 0.6833,
2970
+ "step": 4230
2971
+ },
2972
+ {
2973
+ "epoch": 0.5655406982560273,
2974
+ "grad_norm": 3.2628445264349515,
2975
+ "learning_rate": 1.213516433970902e-05,
2976
+ "loss": 0.6706,
2977
+ "step": 4240
2978
+ },
2979
+ {
2980
+ "epoch": 0.5668745206575745,
2981
+ "grad_norm": 5.645591683747128,
2982
+ "learning_rate": 1.2072866771512306e-05,
2983
+ "loss": 0.6812,
2984
+ "step": 4250
2985
+ },
2986
+ {
2987
+ "epoch": 0.5682083430591217,
2988
+ "grad_norm": 2.7073660859499684,
2989
+ "learning_rate": 1.201062164778389e-05,
2990
+ "loss": 0.6623,
2991
+ "step": 4260
2992
+ },
2993
+ {
2994
+ "epoch": 0.5695421654606689,
2995
+ "grad_norm": 5.201872987152117,
2996
+ "learning_rate": 1.1948430083748864e-05,
2997
+ "loss": 0.6562,
2998
+ "step": 4270
2999
+ },
3000
+ {
3001
+ "epoch": 0.5708759878622162,
3002
+ "grad_norm": 1.3559544817056064,
3003
+ "learning_rate": 1.1886293193672707e-05,
3004
+ "loss": 0.6906,
3005
+ "step": 4280
3006
+ },
3007
+ {
3008
+ "epoch": 0.5722098102637634,
3009
+ "grad_norm": 24.227173434331696,
3010
+ "learning_rate": 1.1824212090841321e-05,
3011
+ "loss": 0.642,
3012
+ "step": 4290
3013
+ },
3014
+ {
3015
+ "epoch": 0.5735436326653106,
3016
+ "grad_norm": 5.123165453557631,
3017
+ "learning_rate": 1.1762187887541088e-05,
3018
+ "loss": 0.7297,
3019
+ "step": 4300
3020
+ },
3021
+ {
3022
+ "epoch": 0.5748774550668578,
3023
+ "grad_norm": 5.724684674103886,
3024
+ "learning_rate": 1.1700221695038944e-05,
3025
+ "loss": 0.6524,
3026
+ "step": 4310
3027
+ },
3028
+ {
3029
+ "epoch": 0.5762112774684051,
3030
+ "grad_norm": 9.965228089412879,
3031
+ "learning_rate": 1.1638314623562459e-05,
3032
+ "loss": 0.704,
3033
+ "step": 4320
3034
+ },
3035
+ {
3036
+ "epoch": 0.5775450998699523,
3037
+ "grad_norm": 3.232350328466124,
3038
+ "learning_rate": 1.1576467782279953e-05,
3039
+ "loss": 0.6391,
3040
+ "step": 4330
3041
+ },
3042
+ {
3043
+ "epoch": 0.5788789222714995,
3044
+ "grad_norm": 3.1098394765854707,
3045
+ "learning_rate": 1.1514682279280621e-05,
3046
+ "loss": 0.7083,
3047
+ "step": 4340
3048
+ },
3049
+ {
3050
+ "epoch": 0.5802127446730467,
3051
+ "grad_norm": 4.287688599823072,
3052
+ "learning_rate": 1.1452959221554684e-05,
3053
+ "loss": 0.7003,
3054
+ "step": 4350
3055
+ },
3056
+ {
3057
+ "epoch": 0.5815465670745941,
3058
+ "grad_norm": 3.889095266103746,
3059
+ "learning_rate": 1.1391299714973553e-05,
3060
+ "loss": 0.6915,
3061
+ "step": 4360
3062
+ },
3063
+ {
3064
+ "epoch": 0.5828803894761413,
3065
+ "grad_norm": 28.63862172481202,
3066
+ "learning_rate": 1.1329704864270005e-05,
3067
+ "loss": 0.639,
3068
+ "step": 4370
3069
+ },
3070
+ {
3071
+ "epoch": 0.5842142118776885,
3072
+ "grad_norm": 2.8068799133869353,
3073
+ "learning_rate": 1.1268175773018409e-05,
3074
+ "loss": 0.675,
3075
+ "step": 4380
3076
+ },
3077
+ {
3078
+ "epoch": 0.5855480342792357,
3079
+ "grad_norm": 10.272361368305473,
3080
+ "learning_rate": 1.1206713543614942e-05,
3081
+ "loss": 0.707,
3082
+ "step": 4390
3083
+ },
3084
+ {
3085
+ "epoch": 0.586881856680783,
3086
+ "grad_norm": 1.9200541236321964,
3087
+ "learning_rate": 1.1145319277257834e-05,
3088
+ "loss": 0.7177,
3089
+ "step": 4400
3090
+ },
3091
+ {
3092
+ "epoch": 0.5882156790823302,
3093
+ "grad_norm": 2.933892165040151,
3094
+ "learning_rate": 1.108399407392765e-05,
3095
+ "loss": 0.6991,
3096
+ "step": 4410
3097
+ },
3098
+ {
3099
+ "epoch": 0.5895495014838774,
3100
+ "grad_norm": 2.898606287727756,
3101
+ "learning_rate": 1.1022739032367572e-05,
3102
+ "loss": 0.6697,
3103
+ "step": 4420
3104
+ },
3105
+ {
3106
+ "epoch": 0.5908833238854246,
3107
+ "grad_norm": 1.8428061704250374,
3108
+ "learning_rate": 1.0961555250063718e-05,
3109
+ "loss": 0.6939,
3110
+ "step": 4430
3111
+ },
3112
+ {
3113
+ "epoch": 0.5922171462869719,
3114
+ "grad_norm": 2.93674459281119,
3115
+ "learning_rate": 1.090044382322548e-05,
3116
+ "loss": 0.6926,
3117
+ "step": 4440
3118
+ },
3119
+ {
3120
+ "epoch": 0.5935509686885191,
3121
+ "grad_norm": 1.50823929038625,
3122
+ "learning_rate": 1.083940584676588e-05,
3123
+ "loss": 0.6757,
3124
+ "step": 4450
3125
+ },
3126
+ {
3127
+ "epoch": 0.5948847910900663,
3128
+ "grad_norm": 2.0214873235561224,
3129
+ "learning_rate": 1.077844241428195e-05,
3130
+ "loss": 0.6794,
3131
+ "step": 4460
3132
+ },
3133
+ {
3134
+ "epoch": 0.5962186134916136,
3135
+ "grad_norm": 1.87436887927708,
3136
+ "learning_rate": 1.071755461803515e-05,
3137
+ "loss": 0.6882,
3138
+ "step": 4470
3139
+ },
3140
+ {
3141
+ "epoch": 0.5975524358931609,
3142
+ "grad_norm": 1.4982260705038484,
3143
+ "learning_rate": 1.0656743548931784e-05,
3144
+ "loss": 0.6843,
3145
+ "step": 4480
3146
+ },
3147
+ {
3148
+ "epoch": 0.5988862582947081,
3149
+ "grad_norm": 1.8602173123516976,
3150
+ "learning_rate": 1.0596010296503469e-05,
3151
+ "loss": 0.6553,
3152
+ "step": 4490
3153
+ },
3154
+ {
3155
+ "epoch": 0.6002200806962553,
3156
+ "grad_norm": 13.381873530435502,
3157
+ "learning_rate": 1.0535355948887598e-05,
3158
+ "loss": 0.6708,
3159
+ "step": 4500
3160
+ },
3161
+ {
3162
+ "epoch": 0.6015539030978025,
3163
+ "grad_norm": 1.9185796879534898,
3164
+ "learning_rate": 1.0474781592807854e-05,
3165
+ "loss": 0.6806,
3166
+ "step": 4510
3167
+ },
3168
+ {
3169
+ "epoch": 0.6028877254993498,
3170
+ "grad_norm": 3.364824709584871,
3171
+ "learning_rate": 1.0414288313554746e-05,
3172
+ "loss": 0.6847,
3173
+ "step": 4520
3174
+ },
3175
+ {
3176
+ "epoch": 0.604221547900897,
3177
+ "grad_norm": 3.497292970388317,
3178
+ "learning_rate": 1.0353877194966152e-05,
3179
+ "loss": 0.6635,
3180
+ "step": 4530
3181
+ },
3182
+ {
3183
+ "epoch": 0.6055553703024442,
3184
+ "grad_norm": 4.099439787389805,
3185
+ "learning_rate": 1.0293549319407901e-05,
3186
+ "loss": 0.7056,
3187
+ "step": 4540
3188
+ },
3189
+ {
3190
+ "epoch": 0.6068891927039914,
3191
+ "grad_norm": 4.726371219011721,
3192
+ "learning_rate": 1.0233305767754391e-05,
3193
+ "loss": 0.6746,
3194
+ "step": 4550
3195
+ },
3196
+ {
3197
+ "epoch": 0.6082230151055387,
3198
+ "grad_norm": 1.9002791525912257,
3199
+ "learning_rate": 1.0173147619369212e-05,
3200
+ "loss": 0.6843,
3201
+ "step": 4560
3202
+ },
3203
+ {
3204
+ "epoch": 0.6095568375070859,
3205
+ "grad_norm": 2.1361565277382026,
3206
+ "learning_rate": 1.0113075952085815e-05,
3207
+ "loss": 0.716,
3208
+ "step": 4570
3209
+ },
3210
+ {
3211
+ "epoch": 0.6108906599086331,
3212
+ "grad_norm": 2.7505558492527284,
3213
+ "learning_rate": 1.0053091842188196e-05,
3214
+ "loss": 0.6676,
3215
+ "step": 4580
3216
+ },
3217
+ {
3218
+ "epoch": 0.6122244823101805,
3219
+ "grad_norm": 7.936649787603975,
3220
+ "learning_rate": 9.993196364391614e-06,
3221
+ "loss": 0.7174,
3222
+ "step": 4590
3223
+ },
3224
+ {
3225
+ "epoch": 0.6135583047117277,
3226
+ "grad_norm": 1.9962614403326984,
3227
+ "learning_rate": 9.93339059182334e-06,
3228
+ "loss": 0.7116,
3229
+ "step": 4600
3230
+ },
3231
+ {
3232
+ "epoch": 0.6148921271132749,
3233
+ "grad_norm": 1.6832414510723148,
3234
+ "learning_rate": 9.873675596003424e-06,
3235
+ "loss": 0.631,
3236
+ "step": 4610
3237
+ },
3238
+ {
3239
+ "epoch": 0.6162259495148221,
3240
+ "grad_norm": 1.943682374421793,
3241
+ "learning_rate": 9.8140524468255e-06,
3242
+ "loss": 0.7083,
3243
+ "step": 4620
3244
+ },
3245
+ {
3246
+ "epoch": 0.6175597719163693,
3247
+ "grad_norm": 3.6516090347661323,
3248
+ "learning_rate": 9.754522212537614e-06,
3249
+ "loss": 0.6983,
3250
+ "step": 4630
3251
+ },
3252
+ {
3253
+ "epoch": 0.6188935943179166,
3254
+ "grad_norm": 2.245519627711479,
3255
+ "learning_rate": 9.695085959723088e-06,
3256
+ "loss": 0.664,
3257
+ "step": 4640
3258
+ },
3259
+ {
3260
+ "epoch": 0.6202274167194638,
3261
+ "grad_norm": 1.9531032713635088,
3262
+ "learning_rate": 9.63574475328141e-06,
3263
+ "loss": 0.6853,
3264
+ "step": 4650
3265
+ },
3266
+ {
3267
+ "epoch": 0.621561239121011,
3268
+ "grad_norm": 5.427236407190242,
3269
+ "learning_rate": 9.576499656409158e-06,
3270
+ "loss": 0.694,
3271
+ "step": 4660
3272
+ },
3273
+ {
3274
+ "epoch": 0.6228950615225582,
3275
+ "grad_norm": 2.000287154564092,
3276
+ "learning_rate": 9.517351730580939e-06,
3277
+ "loss": 0.7371,
3278
+ "step": 4670
3279
+ },
3280
+ {
3281
+ "epoch": 0.6242288839241055,
3282
+ "grad_norm": 1.8847269162961595,
3283
+ "learning_rate": 9.458302035530384e-06,
3284
+ "loss": 0.6643,
3285
+ "step": 4680
3286
+ },
3287
+ {
3288
+ "epoch": 0.6255627063256527,
3289
+ "grad_norm": 2.7122993676306564,
3290
+ "learning_rate": 9.399351629231154e-06,
3291
+ "loss": 0.6912,
3292
+ "step": 4690
3293
+ },
3294
+ {
3295
+ "epoch": 0.6268965287271999,
3296
+ "grad_norm": 3.1005319569612078,
3297
+ "learning_rate": 9.340501567877989e-06,
3298
+ "loss": 0.6798,
3299
+ "step": 4700
3300
+ },
3301
+ {
3302
+ "epoch": 0.6282303511287473,
3303
+ "grad_norm": 3.9244802871969133,
3304
+ "learning_rate": 9.281752905867778e-06,
3305
+ "loss": 0.6383,
3306
+ "step": 4710
3307
+ },
3308
+ {
3309
+ "epoch": 0.6295641735302945,
3310
+ "grad_norm": 2.435298414489433,
3311
+ "learning_rate": 9.223106695780677e-06,
3312
+ "loss": 0.6892,
3313
+ "step": 4720
3314
+ },
3315
+ {
3316
+ "epoch": 0.6308979959318417,
3317
+ "grad_norm": 8.27385430371254,
3318
+ "learning_rate": 9.164563988361242e-06,
3319
+ "loss": 0.6359,
3320
+ "step": 4730
3321
+ },
3322
+ {
3323
+ "epoch": 0.6322318183333889,
3324
+ "grad_norm": 1.53362121712898,
3325
+ "learning_rate": 9.106125832499604e-06,
3326
+ "loss": 0.6611,
3327
+ "step": 4740
3328
+ },
3329
+ {
3330
+ "epoch": 0.6335656407349362,
3331
+ "grad_norm": 3.683488506541024,
3332
+ "learning_rate": 9.047793275212686e-06,
3333
+ "loss": 0.6755,
3334
+ "step": 4750
3335
+ },
3336
+ {
3337
+ "epoch": 0.6348994631364834,
3338
+ "grad_norm": 5.130363634989403,
3339
+ "learning_rate": 8.989567361625427e-06,
3340
+ "loss": 0.6753,
3341
+ "step": 4760
3342
+ },
3343
+ {
3344
+ "epoch": 0.6362332855380306,
3345
+ "grad_norm": 5.687746482790019,
3346
+ "learning_rate": 8.931449134952075e-06,
3347
+ "loss": 0.6841,
3348
+ "step": 4770
3349
+ },
3350
+ {
3351
+ "epoch": 0.6375671079395778,
3352
+ "grad_norm": 6.9830588651752,
3353
+ "learning_rate": 8.873439636477484e-06,
3354
+ "loss": 0.7064,
3355
+ "step": 4780
3356
+ },
3357
+ {
3358
+ "epoch": 0.638900930341125,
3359
+ "grad_norm": 6.829790484006074,
3360
+ "learning_rate": 8.815539905538459e-06,
3361
+ "loss": 0.6757,
3362
+ "step": 4790
3363
+ },
3364
+ {
3365
+ "epoch": 0.6402347527426723,
3366
+ "grad_norm": 3.4577918643740753,
3367
+ "learning_rate": 8.757750979505137e-06,
3368
+ "loss": 0.6936,
3369
+ "step": 4800
3370
+ }
3371
+ ],
3372
+ "logging_steps": 10,
3373
+ "max_steps": 7497,
3374
+ "num_input_tokens_seen": 0,
3375
+ "num_train_epochs": 1,
3376
+ "save_steps": 400,
3377
+ "stateful_callbacks": {
3378
+ "TrainerControl": {
3379
+ "args": {
3380
+ "should_epoch_stop": false,
3381
+ "should_evaluate": false,
3382
+ "should_log": false,
3383
+ "should_save": true,
3384
+ "should_training_stop": false
3385
+ },
3386
+ "attributes": {}
3387
+ }
3388
+ },
3389
+ "total_flos": 2.186856472195891e+19,
3390
+ "train_batch_size": 4,
3391
+ "trial_name": null,
3392
+ "trial_params": null
3393
+ }
checkpoint-4800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3917773d9035785be322ba4c24f526b38823b179d4e738034f481213b6bfc50c
3
+ size 6520
checkpoint-4800/zero_to_fp32.py ADDED
@@ -0,0 +1,587 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
252
+ param_shapes = zero_model_states[0].param_shapes
253
+
254
+ # Reconstruction protocol:
255
+ #
256
+ # XXX: document this
257
+
258
+ if debug:
259
+ for i in range(world_size):
260
+ for j in range(len(fp32_flat_groups[0])):
261
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
262
+
263
+ # XXX: memory usage doubles here (zero2)
264
+ num_param_groups = len(fp32_flat_groups[0])
265
+ merged_single_partition_of_fp32_groups = []
266
+ for i in range(num_param_groups):
267
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
268
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
269
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
270
+ avail_numel = sum(
271
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
272
+
273
+ if debug:
274
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
275
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
276
+ # not asserting if there is a mismatch due to possible padding
277
+ print(f"Have {avail_numel} numels to process.")
278
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
279
+
280
+ # params
281
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
282
+ # out-of-core computing solution
283
+ total_numel = 0
284
+ total_params = 0
285
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
286
+ offset = 0
287
+ avail_numel = full_single_fp32_vector.numel()
288
+ for name, shape in shapes.items():
289
+
290
+ unpartitioned_numel = shape.numel()
291
+ total_numel += unpartitioned_numel
292
+ total_params += 1
293
+
294
+ if debug:
295
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
296
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
297
+ offset += unpartitioned_numel
298
+
299
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
300
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
301
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
302
+ # live optimizer object, so we are checking that the numbers are within the right range
303
+ align_to = 2 * world_size
304
+
305
+ def zero2_align(x):
306
+ return align_to * math.ceil(x / align_to)
307
+
308
+ if debug:
309
+ print(f"original offset={offset}, avail_numel={avail_numel}")
310
+
311
+ offset = zero2_align(offset)
312
+ avail_numel = zero2_align(avail_numel)
313
+
314
+ if debug:
315
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
316
+
317
+ # Sanity check
318
+ if offset != avail_numel:
319
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
320
+
321
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
322
+
323
+
324
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
325
+ state_dict = OrderedDict()
326
+
327
+ # buffers
328
+ buffers = zero_model_states[0].buffers
329
+ state_dict.update(buffers)
330
+ if debug:
331
+ print(f"added {len(buffers)} buffers")
332
+
333
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
334
+
335
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
336
+
337
+ # recover shared parameters
338
+ for pair in zero_model_states[0].shared_params:
339
+ if pair[1] in state_dict:
340
+ state_dict[pair[0]] = state_dict[pair[1]]
341
+
342
+ return state_dict
343
+
344
+
345
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
346
+ remainder = unpartitioned_numel % world_size
347
+ padding_numel = (world_size - remainder) if remainder else 0
348
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
349
+ return partitioned_numel, padding_numel
350
+
351
+
352
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
353
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
354
+ return
355
+
356
+ if debug:
357
+ for i in range(world_size):
358
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
359
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
360
+
361
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
362
+ wanted_params = len(frozen_param_shapes)
363
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
364
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
365
+ print(f'Frozen params: Have {avail_numel} numels to process.')
366
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
367
+
368
+ total_params = 0
369
+ total_numel = 0
370
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
371
+ total_params += 1
372
+ unpartitioned_numel = shape.numel()
373
+ total_numel += unpartitioned_numel
374
+
375
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
376
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
377
+
378
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
379
+
380
+ if debug:
381
+ print(
382
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
383
+ )
384
+
385
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
386
+
387
+
388
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
389
+ param_shapes = zero_model_states[0].param_shapes
390
+ avail_numel = fp32_flat_groups[0].numel() * world_size
391
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
392
+ # param, re-consolidating each param, while dealing with padding if any
393
+
394
+ # merge list of dicts, preserving order
395
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
396
+
397
+ if debug:
398
+ for i in range(world_size):
399
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
400
+
401
+ wanted_params = len(param_shapes)
402
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
403
+ # not asserting if there is a mismatch due to possible padding
404
+ avail_numel = fp32_flat_groups[0].numel() * world_size
405
+ print(f"Trainable params: Have {avail_numel} numels to process.")
406
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
407
+
408
+ # params
409
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
410
+ # out-of-core computing solution
411
+ offset = 0
412
+ total_numel = 0
413
+ total_params = 0
414
+ for name, shape in param_shapes.items():
415
+
416
+ unpartitioned_numel = shape.numel()
417
+ total_numel += unpartitioned_numel
418
+ total_params += 1
419
+
420
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
421
+
422
+ if debug:
423
+ print(
424
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
425
+ )
426
+
427
+ # XXX: memory usage doubles here
428
+ state_dict[name] = torch.cat(
429
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
430
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
431
+ offset += partitioned_numel
432
+
433
+ offset *= world_size
434
+
435
+ # Sanity check
436
+ if offset != avail_numel:
437
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
438
+
439
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
440
+
441
+
442
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
443
+ state_dict = OrderedDict()
444
+
445
+ # buffers
446
+ buffers = zero_model_states[0].buffers
447
+ state_dict.update(buffers)
448
+ if debug:
449
+ print(f"added {len(buffers)} buffers")
450
+
451
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
452
+
453
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
454
+
455
+ # recover shared parameters
456
+ for pair in zero_model_states[0].shared_params:
457
+ if pair[1] in state_dict:
458
+ state_dict[pair[0]] = state_dict[pair[1]]
459
+
460
+ return state_dict
461
+
462
+
463
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
464
+ """
465
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
466
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
467
+ via a model hub.
468
+
469
+ Args:
470
+ - ``checkpoint_dir``: path to the desired checkpoint folder
471
+ - ``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``
472
+
473
+ Returns:
474
+ - pytorch ``state_dict``
475
+
476
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
477
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
478
+ the checkpoint.
479
+
480
+ A typical usage might be ::
481
+
482
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
483
+ # do the training and checkpoint saving
484
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
485
+ model = model.cpu() # move to cpu
486
+ model.load_state_dict(state_dict)
487
+ # submit to model hub or save the model to share with others
488
+
489
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
490
+ application. i.e. you will need to re-initialize the deepspeed engine, since
491
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
492
+
493
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
494
+
495
+ """
496
+ if tag is None:
497
+ latest_path = os.path.join(checkpoint_dir, 'latest')
498
+ if os.path.isfile(latest_path):
499
+ with open(latest_path, 'r') as fd:
500
+ tag = fd.read().strip()
501
+ else:
502
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
503
+
504
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
505
+
506
+ if not os.path.isdir(ds_checkpoint_dir):
507
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
508
+
509
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
510
+
511
+
512
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
513
+ """
514
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
515
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
516
+
517
+ Args:
518
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
519
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
520
+ - ``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``
521
+ """
522
+
523
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
524
+ print(f"Saving fp32 state dict to {output_file}")
525
+ torch.save(state_dict, output_file)
526
+
527
+
528
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
529
+ """
530
+ 1. Put the provided model to cpu
531
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
532
+ 3. Load it into the provided model
533
+
534
+ Args:
535
+ - ``model``: the model object to update
536
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
537
+ - ``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``
538
+
539
+ Returns:
540
+ - ``model`: modified model
541
+
542
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
543
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
544
+ conveniently placed for you in the checkpoint folder.
545
+
546
+ A typical usage might be ::
547
+
548
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
549
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
550
+ # submit to model hub or save the model to share with others
551
+
552
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
553
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
554
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
555
+
556
+ """
557
+ logger.info(f"Extracting fp32 weights")
558
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
559
+
560
+ logger.info(f"Overwriting model with fp32 weights")
561
+ model = model.cpu()
562
+ model.load_state_dict(state_dict, strict=False)
563
+
564
+ return model
565
+
566
+
567
+ if __name__ == "__main__":
568
+
569
+ parser = argparse.ArgumentParser()
570
+ parser.add_argument("checkpoint_dir",
571
+ type=str,
572
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
573
+ parser.add_argument(
574
+ "output_file",
575
+ type=str,
576
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
577
+ parser.add_argument("-t",
578
+ "--tag",
579
+ type=str,
580
+ default=None,
581
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
582
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
583
+ args = parser.parse_args()
584
+
585
+ debug = args.debug
586
+
587
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
checkpoint-6400/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: Qwen/Qwen-VL-Chat
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 0.10.0
checkpoint-6400/adapter_config.json ADDED
@@ -0,0 +1,380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "Qwen/Qwen-VL-Chat",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 64,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.25.mlp.w2",
24
+ "transformer.visual.transformer.resblocks.39.mlp.c_proj",
25
+ "transformer.visual.transformer.resblocks.21.attn.in_proj",
26
+ "transformer.visual.transformer.resblocks.44.mlp.c_proj",
27
+ "transformer.h.15.mlp.w2",
28
+ "transformer.visual.transformer.resblocks.7.mlp.c_proj",
29
+ "transformer.h.3.mlp.w2",
30
+ "transformer.visual.transformer.resblocks.23.mlp.c_fc",
31
+ "transformer.visual.transformer.resblocks.10.mlp.c_fc",
32
+ "transformer.h.20.attn.c_attn",
33
+ "transformer.h.23.mlp.c_proj",
34
+ "transformer.visual.transformer.resblocks.34.mlp.c_fc",
35
+ "transformer.h.0.attn.c_proj",
36
+ "transformer.h.9.mlp.w1",
37
+ "transformer.visual.transformer.resblocks.12.attn.out_proj",
38
+ "transformer.visual.transformer.resblocks.22.mlp.c_fc",
39
+ "transformer.visual.transformer.resblocks.23.attn.out_proj",
40
+ "transformer.h.27.mlp.c_proj",
41
+ "transformer.visual.transformer.resblocks.1.mlp.c_proj",
42
+ "transformer.h.22.attn.c_proj",
43
+ "transformer.h.25.mlp.w1",
44
+ "transformer.visual.transformer.resblocks.0.attn.in_proj",
45
+ "transformer.visual.transformer.resblocks.2.attn.in_proj",
46
+ "transformer.h.1.attn.c_proj",
47
+ "transformer.visual.transformer.resblocks.45.mlp.c_proj",
48
+ "transformer.h.29.attn.c_attn",
49
+ "transformer.visual.transformer.resblocks.43.attn.in_proj",
50
+ "transformer.visual.transformer.resblocks.46.mlp.c_proj",
51
+ "transformer.h.28.mlp.w2",
52
+ "transformer.h.20.attn.c_proj",
53
+ "transformer.h.12.attn.c_proj",
54
+ "transformer.visual.transformer.resblocks.27.attn.out_proj",
55
+ "transformer.visual.transformer.resblocks.17.mlp.c_proj",
56
+ "transformer.visual.transformer.resblocks.33.attn.out_proj",
57
+ "transformer.h.4.attn.c_proj",
58
+ "transformer.visual.transformer.resblocks.29.attn.in_proj",
59
+ "transformer.visual.transformer.resblocks.2.attn.out_proj",
60
+ "transformer.h.4.mlp.w2",
61
+ "transformer.h.30.mlp.w1",
62
+ "transformer.h.31.attn.c_attn",
63
+ "transformer.h.10.attn.c_proj",
64
+ "transformer.h.18.mlp.w2",
65
+ "transformer.visual.transformer.resblocks.12.attn.in_proj",
66
+ "transformer.h.20.mlp.c_proj",
67
+ "transformer.visual.transformer.resblocks.36.attn.out_proj",
68
+ "transformer.visual.transformer.resblocks.31.attn.out_proj",
69
+ "transformer.h.1.attn.c_attn",
70
+ "transformer.visual.transformer.resblocks.17.attn.out_proj",
71
+ "transformer.visual.transformer.resblocks.9.mlp.c_fc",
72
+ "transformer.h.22.attn.c_attn",
73
+ "transformer.h.16.attn.c_attn",
74
+ "transformer.visual.transformer.resblocks.20.mlp.c_fc",
75
+ "transformer.visual.transformer.resblocks.31.attn.in_proj",
76
+ "transformer.h.29.mlp.w1",
77
+ "transformer.h.27.attn.c_attn",
78
+ "transformer.visual.transformer.resblocks.18.mlp.c_proj",
79
+ "transformer.visual.transformer.resblocks.8.mlp.c_proj",
80
+ "transformer.visual.transformer.resblocks.44.attn.out_proj",
81
+ "transformer.h.5.attn.c_proj",
82
+ "transformer.h.20.mlp.w1",
83
+ "transformer.h.0.mlp.c_proj",
84
+ "transformer.visual.transformer.resblocks.9.attn.out_proj",
85
+ "transformer.h.2.mlp.w1",
86
+ "transformer.visual.transformer.resblocks.3.mlp.c_proj",
87
+ "transformer.visual.transformer.resblocks.30.mlp.c_fc",
88
+ "transformer.visual.transformer.resblocks.36.mlp.c_proj",
89
+ "transformer.h.17.mlp.w2",
90
+ "transformer.h.19.mlp.w2",
91
+ "transformer.visual.transformer.resblocks.30.attn.out_proj",
92
+ "transformer.visual.transformer.resblocks.40.mlp.c_proj",
93
+ "transformer.visual.transformer.resblocks.1.mlp.c_fc",
94
+ "transformer.h.10.mlp.w1",
95
+ "transformer.h.6.attn.c_attn",
96
+ "transformer.visual.transformer.resblocks.14.mlp.c_fc",
97
+ "transformer.h.24.mlp.w2",
98
+ "transformer.visual.transformer.resblocks.28.attn.out_proj",
99
+ "transformer.visual.transformer.resblocks.27.attn.in_proj",
100
+ "transformer.h.10.attn.c_attn",
101
+ "transformer.h.17.mlp.c_proj",
102
+ "transformer.visual.transformer.resblocks.0.mlp.c_fc",
103
+ "transformer.visual.transformer.resblocks.35.attn.out_proj",
104
+ "transformer.h.6.attn.c_proj",
105
+ "transformer.h.6.mlp.w2",
106
+ "transformer.visual.transformer.resblocks.21.mlp.c_fc",
107
+ "transformer.h.22.mlp.c_proj",
108
+ "transformer.h.4.mlp.w1",
109
+ "transformer.h.14.mlp.w1",
110
+ "transformer.h.2.mlp.c_proj",
111
+ "transformer.visual.transformer.resblocks.31.mlp.c_fc",
112
+ "transformer.h.28.attn.c_proj",
113
+ "transformer.visual.transformer.resblocks.19.attn.in_proj",
114
+ "transformer.h.26.attn.c_proj",
115
+ "transformer.visual.transformer.resblocks.47.mlp.c_fc",
116
+ "transformer.h.3.attn.c_proj",
117
+ "transformer.visual.transformer.resblocks.33.attn.in_proj",
118
+ "transformer.visual.transformer.resblocks.45.attn.out_proj",
119
+ "transformer.visual.transformer.resblocks.5.attn.out_proj",
120
+ "transformer.h.7.attn.c_attn",
121
+ "transformer.h.1.mlp.w1",
122
+ "transformer.h.19.mlp.c_proj",
123
+ "transformer.visual.transformer.resblocks.39.attn.in_proj",
124
+ "transformer.visual.transformer.resblocks.7.attn.out_proj",
125
+ "transformer.h.8.mlp.w1",
126
+ "transformer.h.22.mlp.w1",
127
+ "transformer.h.7.attn.c_proj",
128
+ "transformer.visual.transformer.resblocks.0.mlp.c_proj",
129
+ "transformer.h.7.mlp.w2",
130
+ "transformer.h.25.attn.c_proj",
131
+ "transformer.visual.transformer.resblocks.17.attn.in_proj",
132
+ "transformer.visual.transformer.resblocks.24.attn.out_proj",
133
+ "transformer.h.24.mlp.c_proj",
134
+ "transformer.visual.transformer.resblocks.19.mlp.c_fc",
135
+ "transformer.h.21.mlp.w2",
136
+ "transformer.h.9.mlp.w2",
137
+ "transformer.h.27.attn.c_proj",
138
+ "transformer.h.2.attn.c_proj",
139
+ "transformer.h.5.mlp.c_proj",
140
+ "transformer.visual.transformer.resblocks.18.attn.in_proj",
141
+ "transformer.h.18.attn.c_attn",
142
+ "transformer.visual.transformer.resblocks.37.attn.in_proj",
143
+ "transformer.visual.transformer.resblocks.15.attn.out_proj",
144
+ "transformer.h.9.mlp.c_proj",
145
+ "transformer.visual.transformer.resblocks.37.mlp.c_proj",
146
+ "transformer.h.17.attn.c_proj",
147
+ "transformer.visual.transformer.resblocks.4.attn.out_proj",
148
+ "transformer.h.11.mlp.w2",
149
+ "transformer.h.3.mlp.w1",
150
+ "transformer.visual.transformer.resblocks.32.attn.out_proj",
151
+ "transformer.h.30.attn.c_attn",
152
+ "transformer.h.7.mlp.c_proj",
153
+ "transformer.visual.transformer.resblocks.38.mlp.c_proj",
154
+ "transformer.visual.transformer.resblocks.22.mlp.c_proj",
155
+ "transformer.h.23.mlp.w2",
156
+ "transformer.visual.transformer.resblocks.8.mlp.c_fc",
157
+ "transformer.h.24.attn.c_attn",
158
+ "transformer.visual.transformer.resblocks.21.attn.out_proj",
159
+ "transformer.visual.transformer.resblocks.16.mlp.c_fc",
160
+ "transformer.visual.transformer.resblocks.30.mlp.c_proj",
161
+ "transformer.visual.transformer.resblocks.29.attn.out_proj",
162
+ "transformer.visual.transformer.resblocks.19.attn.out_proj",
163
+ "transformer.visual.transformer.resblocks.1.attn.in_proj",
164
+ "transformer.h.27.mlp.w1",
165
+ "transformer.h.1.mlp.w2",
166
+ "transformer.h.17.attn.c_attn",
167
+ "transformer.visual.transformer.resblocks.3.mlp.c_fc",
168
+ "transformer.visual.transformer.resblocks.24.attn.in_proj",
169
+ "transformer.visual.transformer.resblocks.44.attn.in_proj",
170
+ "transformer.h.6.mlp.c_proj",
171
+ "transformer.visual.transformer.resblocks.3.attn.in_proj",
172
+ "transformer.visual.transformer.resblocks.19.mlp.c_proj",
173
+ "transformer.visual.transformer.resblocks.38.mlp.c_fc",
174
+ "transformer.h.14.mlp.w2",
175
+ "transformer.h.8.mlp.w2",
176
+ "transformer.h.19.mlp.w1",
177
+ "transformer.visual.transformer.resblocks.15.mlp.c_fc",
178
+ "transformer.h.19.attn.c_attn",
179
+ "transformer.h.13.attn.c_proj",
180
+ "transformer.visual.transformer.resblocks.26.mlp.c_fc",
181
+ "transformer.visual.transformer.resblocks.35.mlp.c_proj",
182
+ "transformer.h.15.mlp.w1",
183
+ "transformer.h.23.attn.c_attn",
184
+ "transformer.visual.transformer.resblocks.6.attn.out_proj",
185
+ "transformer.visual.transformer.resblocks.24.mlp.c_fc",
186
+ "transformer.visual.transformer.resblocks.18.attn.out_proj",
187
+ "transformer.visual.transformer.resblocks.20.attn.in_proj",
188
+ "transformer.visual.transformer.resblocks.31.mlp.c_proj",
189
+ "transformer.visual.transformer.resblocks.43.attn.out_proj",
190
+ "transformer.h.5.attn.c_attn",
191
+ "transformer.visual.transformer.resblocks.40.attn.out_proj",
192
+ "transformer.visual.transformer.resblocks.29.mlp.c_fc",
193
+ "transformer.h.12.mlp.w1",
194
+ "transformer.visual.transformer.resblocks.13.mlp.c_proj",
195
+ "transformer.visual.transformer.resblocks.43.mlp.c_proj",
196
+ "transformer.visual.transformer.resblocks.11.mlp.c_proj",
197
+ "transformer.visual.transformer.resblocks.37.attn.out_proj",
198
+ "transformer.visual.transformer.resblocks.40.mlp.c_fc",
199
+ "transformer.h.16.mlp.w2",
200
+ "transformer.h.9.attn.c_proj",
201
+ "transformer.visual.transformer.resblocks.26.attn.out_proj",
202
+ "transformer.visual.transformer.resblocks.4.attn.in_proj",
203
+ "transformer.visual.transformer.resblocks.10.attn.out_proj",
204
+ "transformer.h.13.mlp.c_proj",
205
+ "transformer.visual.transformer.resblocks.32.attn.in_proj",
206
+ "transformer.h.25.mlp.c_proj",
207
+ "transformer.h.24.mlp.w1",
208
+ "transformer.h.4.mlp.c_proj",
209
+ "transformer.h.21.attn.c_attn",
210
+ "transformer.visual.transformer.resblocks.37.mlp.c_fc",
211
+ "transformer.h.11.attn.c_proj",
212
+ "transformer.h.30.mlp.c_proj",
213
+ "transformer.visual.transformer.resblocks.23.mlp.c_proj",
214
+ "transformer.visual.transformer.resblocks.32.mlp.c_proj",
215
+ "transformer.h.21.attn.c_proj",
216
+ "transformer.h.3.attn.c_attn",
217
+ "transformer.visual.transformer.resblocks.5.mlp.c_proj",
218
+ "transformer.visual.transformer.resblocks.12.mlp.c_fc",
219
+ "transformer.visual.transformer.resblocks.45.attn.in_proj",
220
+ "transformer.h.4.attn.c_attn",
221
+ "transformer.visual.transformer.resblocks.33.mlp.c_proj",
222
+ "transformer.h.19.attn.c_proj",
223
+ "transformer.h.5.mlp.w1",
224
+ "transformer.visual.transformer.resblocks.14.attn.in_proj",
225
+ "transformer.h.9.attn.c_attn",
226
+ "transformer.visual.transformer.resblocks.45.mlp.c_fc",
227
+ "transformer.h.22.mlp.w2",
228
+ "transformer.visual.transformer.resblocks.41.attn.in_proj",
229
+ "transformer.h.3.mlp.c_proj",
230
+ "transformer.visual.transformer.resblocks.5.mlp.c_fc",
231
+ "transformer.h.26.attn.c_attn",
232
+ "transformer.h.30.attn.c_proj",
233
+ "transformer.visual.transformer.resblocks.10.attn.in_proj",
234
+ "transformer.visual.transformer.resblocks.24.mlp.c_proj",
235
+ "transformer.visual.transformer.resblocks.4.mlp.c_proj",
236
+ "transformer.visual.transformer.resblocks.44.mlp.c_fc",
237
+ "transformer.visual.transformer.resblocks.38.attn.in_proj",
238
+ "transformer.visual.transformer.resblocks.13.mlp.c_fc",
239
+ "transformer.visual.transformer.resblocks.8.attn.out_proj",
240
+ "transformer.visual.transformer.resblocks.23.attn.in_proj",
241
+ "transformer.visual.transformer.resblocks.41.mlp.c_proj",
242
+ "transformer.visual.transformer.resblocks.42.mlp.c_fc",
243
+ "transformer.visual.transformer.resblocks.39.mlp.c_fc",
244
+ "transformer.h.17.mlp.w1",
245
+ "transformer.visual.transformer.resblocks.42.attn.out_proj",
246
+ "transformer.visual.transformer.resblocks.8.attn.in_proj",
247
+ "transformer.h.25.attn.c_attn",
248
+ "transformer.visual.transformer.resblocks.1.attn.out_proj",
249
+ "transformer.h.6.mlp.w1",
250
+ "transformer.visual.transformer.resblocks.46.attn.out_proj",
251
+ "transformer.h.8.attn.c_attn",
252
+ "transformer.h.13.mlp.w2",
253
+ "transformer.visual.transformer.resblocks.21.mlp.c_proj",
254
+ "transformer.h.2.mlp.w2",
255
+ "transformer.visual.transformer.resblocks.30.attn.in_proj",
256
+ "transformer.h.7.mlp.w1",
257
+ "transformer.visual.transformer.resblocks.25.mlp.c_fc",
258
+ "transformer.h.11.mlp.c_proj",
259
+ "transformer.visual.transformer.resblocks.36.mlp.c_fc",
260
+ "transformer.visual.transformer.resblocks.42.mlp.c_proj",
261
+ "transformer.h.11.attn.c_attn",
262
+ "transformer.h.29.mlp.c_proj",
263
+ "transformer.visual.transformer.resblocks.36.attn.in_proj",
264
+ "transformer.visual.transformer.resblocks.40.attn.in_proj",
265
+ "transformer.visual.transformer.resblocks.27.mlp.c_fc",
266
+ "transformer.h.11.mlp.w1",
267
+ "transformer.visual.transformer.resblocks.39.attn.out_proj",
268
+ "transformer.visual.transformer.resblocks.10.mlp.c_proj",
269
+ "transformer.visual.transformer.resblocks.46.mlp.c_fc",
270
+ "transformer.h.20.mlp.w2",
271
+ "transformer.h.18.attn.c_proj",
272
+ "transformer.visual.transformer.resblocks.4.mlp.c_fc",
273
+ "transformer.visual.transformer.resblocks.17.mlp.c_fc",
274
+ "transformer.visual.transformer.resblocks.2.mlp.c_proj",
275
+ "transformer.visual.transformer.resblocks.6.attn.in_proj",
276
+ "transformer.visual.transformer.resblocks.11.mlp.c_fc",
277
+ "transformer.h.12.attn.c_attn",
278
+ "transformer.visual.transformer.resblocks.16.attn.in_proj",
279
+ "transformer.h.14.attn.c_attn",
280
+ "transformer.h.31.mlp.c_proj",
281
+ "transformer.visual.transformer.resblocks.16.attn.out_proj",
282
+ "transformer.h.0.mlp.w1",
283
+ "transformer.visual.transformer.resblocks.43.mlp.c_fc",
284
+ "transformer.visual.transformer.resblocks.5.attn.in_proj",
285
+ "transformer.h.13.attn.c_attn",
286
+ "transformer.visual.transformer.resblocks.41.mlp.c_fc",
287
+ "transformer.h.23.attn.c_proj",
288
+ "transformer.visual.transformer.resblocks.35.mlp.c_fc",
289
+ "transformer.visual.transformer.resblocks.12.mlp.c_proj",
290
+ "transformer.h.8.attn.c_proj",
291
+ "transformer.h.31.attn.c_proj",
292
+ "transformer.visual.transformer.resblocks.26.mlp.c_proj",
293
+ "transformer.h.21.mlp.w1",
294
+ "transformer.visual.transformer.resblocks.16.mlp.c_proj",
295
+ "transformer.h.31.mlp.w1",
296
+ "transformer.visual.transformer.resblocks.13.attn.in_proj",
297
+ "transformer.h.1.mlp.c_proj",
298
+ "transformer.h.12.mlp.c_proj",
299
+ "transformer.h.30.mlp.w2",
300
+ "transformer.h.14.attn.c_proj",
301
+ "transformer.visual.transformer.resblocks.15.mlp.c_proj",
302
+ "transformer.visual.transformer.resblocks.7.attn.in_proj",
303
+ "transformer.visual.transformer.resblocks.22.attn.out_proj",
304
+ "transformer.visual.transformer.resblocks.0.attn.out_proj",
305
+ "transformer.h.31.mlp.w2",
306
+ "transformer.visual.transformer.resblocks.27.mlp.c_proj",
307
+ "transformer.visual.transformer.resblocks.25.attn.out_proj",
308
+ "transformer.visual.transformer.resblocks.22.attn.in_proj",
309
+ "transformer.visual.transformer.resblocks.33.mlp.c_fc",
310
+ "transformer.h.13.mlp.w1",
311
+ "transformer.visual.transformer.resblocks.6.mlp.c_fc",
312
+ "transformer.visual.transformer.resblocks.11.attn.in_proj",
313
+ "transformer.visual.transformer.resblocks.9.attn.in_proj",
314
+ "transformer.visual.transformer.resblocks.20.mlp.c_proj",
315
+ "transformer.h.16.mlp.c_proj",
316
+ "transformer.visual.transformer.resblocks.15.attn.in_proj",
317
+ "transformer.h.26.mlp.w2",
318
+ "transformer.h.28.attn.c_attn",
319
+ "transformer.visual.transformer.resblocks.29.mlp.c_proj",
320
+ "transformer.h.23.mlp.w1",
321
+ "transformer.h.14.mlp.c_proj",
322
+ "transformer.visual.transformer.resblocks.25.mlp.c_proj",
323
+ "transformer.visual.transformer.resblocks.9.mlp.c_proj",
324
+ "transformer.visual.transformer.resblocks.34.attn.in_proj",
325
+ "transformer.h.5.mlp.w2",
326
+ "transformer.h.26.mlp.c_proj",
327
+ "transformer.h.24.attn.c_proj",
328
+ "transformer.visual.transformer.resblocks.47.attn.in_proj",
329
+ "transformer.visual.transformer.resblocks.7.mlp.c_fc",
330
+ "transformer.h.2.attn.c_attn",
331
+ "transformer.visual.conv1",
332
+ "transformer.visual.transformer.resblocks.28.attn.in_proj",
333
+ "transformer.visual.transformer.resblocks.25.attn.in_proj",
334
+ "transformer.visual.transformer.resblocks.47.attn.out_proj",
335
+ "transformer.visual.transformer.resblocks.18.mlp.c_fc",
336
+ "transformer.h.28.mlp.w1",
337
+ "transformer.h.10.mlp.w2",
338
+ "transformer.h.15.attn.c_proj",
339
+ "transformer.h.21.mlp.c_proj",
340
+ "transformer.h.29.mlp.w2",
341
+ "transformer.visual.transformer.resblocks.32.mlp.c_fc",
342
+ "transformer.h.18.mlp.w1",
343
+ "transformer.visual.transformer.resblocks.41.attn.out_proj",
344
+ "transformer.h.8.mlp.c_proj",
345
+ "transformer.visual.transformer.resblocks.47.mlp.c_proj",
346
+ "transformer.visual.transformer.resblocks.28.mlp.c_fc",
347
+ "transformer.visual.transformer.resblocks.34.attn.out_proj",
348
+ "transformer.h.29.attn.c_proj",
349
+ "transformer.visual.transformer.resblocks.38.attn.out_proj",
350
+ "transformer.visual.transformer.resblocks.14.attn.out_proj",
351
+ "transformer.h.10.mlp.c_proj",
352
+ "transformer.visual.transformer.resblocks.6.mlp.c_proj",
353
+ "transformer.h.0.mlp.w2",
354
+ "transformer.visual.transformer.resblocks.26.attn.in_proj",
355
+ "transformer.visual.transformer.resblocks.42.attn.in_proj",
356
+ "transformer.h.15.attn.c_attn",
357
+ "transformer.h.16.mlp.w1",
358
+ "transformer.visual.transformer.resblocks.3.attn.out_proj",
359
+ "transformer.visual.transformer.resblocks.11.attn.out_proj",
360
+ "transformer.visual.transformer.resblocks.28.mlp.c_proj",
361
+ "transformer.visual.transformer.resblocks.14.mlp.c_proj",
362
+ "transformer.h.0.attn.c_attn",
363
+ "transformer.h.18.mlp.c_proj",
364
+ "transformer.h.12.mlp.w2",
365
+ "transformer.visual.transformer.resblocks.35.attn.in_proj",
366
+ "transformer.visual.transformer.resblocks.2.mlp.c_fc",
367
+ "transformer.visual.transformer.resblocks.34.mlp.c_proj",
368
+ "transformer.h.26.mlp.w1",
369
+ "transformer.visual.transformer.resblocks.13.attn.out_proj",
370
+ "transformer.visual.transformer.resblocks.20.attn.out_proj",
371
+ "transformer.visual.transformer.resblocks.46.attn.in_proj",
372
+ "transformer.h.16.attn.c_proj",
373
+ "transformer.h.15.mlp.c_proj",
374
+ "transformer.h.27.mlp.w2",
375
+ "transformer.h.28.mlp.c_proj"
376
+ ],
377
+ "task_type": "CAUSAL_LM",
378
+ "use_dora": false,
379
+ "use_rslora": false
380
+ }