rat45 commited on
Commit
b507220
·
verified ·
1 Parent(s): 88d95a6

Upload folder using huggingface_hub

Browse files
LoRA_model/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
3
+ library_name: peft
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+ ### Framework versions
201
+
202
+ - PEFT 0.15.1
LoRA_model/adapter_config.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "TinyLlama/TinyLlama-1.1B-Chat-v1.0",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 16,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 8,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "k_proj",
28
+ "v_proj",
29
+ "q_proj",
30
+ "o_proj"
31
+ ],
32
+ "task_type": "CAUSAL_LM",
33
+ "trainable_token_indices": null,
34
+ "use_dora": false,
35
+ "use_rslora": false
36
+ }
LoRA_model/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e84604d98b6a677f55ca05b22c93254d612b414566abee48fc47c8ba460c97a9
3
+ size 9034304
LoRA_model/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b81a2972294949815272010b6a86d7bf43fc9acf4fcf54e51c4f513f3dc6248
3
+ size 18165306
LoRA_model/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2ffd2e67a5e106bcbed7941bfeb9076b68a447bab53d273e41d280ee2e05187
3
+ size 13990
LoRA_model/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8eadd6817a663d1f9bf7712eda9b171e5dcac162c47465954b243da74701b474
3
+ size 1064
LoRA_model/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "</s>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
LoRA_model/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
LoRA_model/tokenizer_config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n' + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "extra_special_tokens": {},
36
+ "legacy": false,
37
+ "model_max_length": 2048,
38
+ "pad_token": "</s>",
39
+ "padding_side": "right",
40
+ "sp_model_kwargs": {},
41
+ "tokenizer_class": "LlamaTokenizer",
42
+ "unk_token": "<unk>",
43
+ "use_default_system_prompt": false
44
+ }
LoRA_model/trainer_state.json ADDED
@@ -0,0 +1,2283 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1250,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.008,
13
+ "grad_norm": 0.8926979899406433,
14
+ "learning_rate": 1.9999649082784807e-05,
15
+ "loss": 1.915,
16
+ "mean_token_accuracy": 0.6236007302999497,
17
+ "num_tokens": 6210.0,
18
+ "step": 5
19
+ },
20
+ {
21
+ "epoch": 0.016,
22
+ "grad_norm": 1.0845613479614258,
23
+ "learning_rate": 1.9998596355767805e-05,
24
+ "loss": 1.9663,
25
+ "mean_token_accuracy": 0.6149427682161331,
26
+ "num_tokens": 12009.0,
27
+ "step": 10
28
+ },
29
+ {
30
+ "epoch": 0.024,
31
+ "grad_norm": 1.2039991617202759,
32
+ "learning_rate": 1.9996841892833e-05,
33
+ "loss": 1.9187,
34
+ "mean_token_accuracy": 0.6216587990522384,
35
+ "num_tokens": 17962.0,
36
+ "step": 15
37
+ },
38
+ {
39
+ "epoch": 0.032,
40
+ "grad_norm": 1.0329039096832275,
41
+ "learning_rate": 1.9994385817114644e-05,
42
+ "loss": 1.8672,
43
+ "mean_token_accuracy": 0.6195997446775436,
44
+ "num_tokens": 23774.0,
45
+ "step": 20
46
+ },
47
+ {
48
+ "epoch": 0.04,
49
+ "grad_norm": 1.345142126083374,
50
+ "learning_rate": 1.9991228300988586e-05,
51
+ "loss": 1.8149,
52
+ "mean_token_accuracy": 0.6287452057003975,
53
+ "num_tokens": 29787.0,
54
+ "step": 25
55
+ },
56
+ {
57
+ "epoch": 0.048,
58
+ "grad_norm": 1.151061773300171,
59
+ "learning_rate": 1.998736956606018e-05,
60
+ "loss": 1.7157,
61
+ "mean_token_accuracy": 0.6416823953390122,
62
+ "num_tokens": 35996.0,
63
+ "step": 30
64
+ },
65
+ {
66
+ "epoch": 0.056,
67
+ "grad_norm": 1.244079828262329,
68
+ "learning_rate": 1.998280988314872e-05,
69
+ "loss": 1.7079,
70
+ "mean_token_accuracy": 0.643532133102417,
71
+ "num_tokens": 42228.0,
72
+ "step": 35
73
+ },
74
+ {
75
+ "epoch": 0.064,
76
+ "grad_norm": 1.2181051969528198,
77
+ "learning_rate": 1.997754957226847e-05,
78
+ "loss": 1.6285,
79
+ "mean_token_accuracy": 0.6439681276679039,
80
+ "num_tokens": 48201.0,
81
+ "step": 40
82
+ },
83
+ {
84
+ "epoch": 0.072,
85
+ "grad_norm": 1.4243274927139282,
86
+ "learning_rate": 1.997158900260614e-05,
87
+ "loss": 1.5656,
88
+ "mean_token_accuracy": 0.6529261693358421,
89
+ "num_tokens": 54156.0,
90
+ "step": 45
91
+ },
92
+ {
93
+ "epoch": 0.08,
94
+ "grad_norm": 1.655982255935669,
95
+ "learning_rate": 1.9964928592495046e-05,
96
+ "loss": 1.5318,
97
+ "mean_token_accuracy": 0.6556347042322159,
98
+ "num_tokens": 59789.0,
99
+ "step": 50
100
+ },
101
+ {
102
+ "epoch": 0.088,
103
+ "grad_norm": 1.7871781587600708,
104
+ "learning_rate": 1.9957568809385693e-05,
105
+ "loss": 1.4525,
106
+ "mean_token_accuracy": 0.6752936288714408,
107
+ "num_tokens": 65529.0,
108
+ "step": 55
109
+ },
110
+ {
111
+ "epoch": 0.096,
112
+ "grad_norm": 1.5810030698776245,
113
+ "learning_rate": 1.9949510169813006e-05,
114
+ "loss": 1.3936,
115
+ "mean_token_accuracy": 0.6881851211190224,
116
+ "num_tokens": 71580.0,
117
+ "step": 60
118
+ },
119
+ {
120
+ "epoch": 0.104,
121
+ "grad_norm": 1.2322911024093628,
122
+ "learning_rate": 1.9940753239360047e-05,
123
+ "loss": 1.2931,
124
+ "mean_token_accuracy": 0.705792248249054,
125
+ "num_tokens": 77696.0,
126
+ "step": 65
127
+ },
128
+ {
129
+ "epoch": 0.112,
130
+ "grad_norm": 1.869781255722046,
131
+ "learning_rate": 1.9931298632618355e-05,
132
+ "loss": 1.2388,
133
+ "mean_token_accuracy": 0.7188379809260368,
134
+ "num_tokens": 83593.0,
135
+ "step": 70
136
+ },
137
+ {
138
+ "epoch": 0.12,
139
+ "grad_norm": 1.4321820735931396,
140
+ "learning_rate": 1.9921147013144782e-05,
141
+ "loss": 1.1519,
142
+ "mean_token_accuracy": 0.7377950385212898,
143
+ "num_tokens": 89420.0,
144
+ "step": 75
145
+ },
146
+ {
147
+ "epoch": 0.128,
148
+ "grad_norm": 1.0898200273513794,
149
+ "learning_rate": 1.991029909341493e-05,
150
+ "loss": 1.1292,
151
+ "mean_token_accuracy": 0.7410064041614532,
152
+ "num_tokens": 95764.0,
153
+ "step": 80
154
+ },
155
+ {
156
+ "epoch": 0.136,
157
+ "grad_norm": 0.8793467283248901,
158
+ "learning_rate": 1.989875563477316e-05,
159
+ "loss": 1.0849,
160
+ "mean_token_accuracy": 0.7562480002641678,
161
+ "num_tokens": 101572.0,
162
+ "step": 85
163
+ },
164
+ {
165
+ "epoch": 0.144,
166
+ "grad_norm": 1.2151330709457397,
167
+ "learning_rate": 1.988651744737914e-05,
168
+ "loss": 1.0784,
169
+ "mean_token_accuracy": 0.7531677842140198,
170
+ "num_tokens": 107521.0,
171
+ "step": 90
172
+ },
173
+ {
174
+ "epoch": 0.152,
175
+ "grad_norm": 1.0950583219528198,
176
+ "learning_rate": 1.9873585390151003e-05,
177
+ "loss": 1.025,
178
+ "mean_token_accuracy": 0.7635247632861137,
179
+ "num_tokens": 113770.0,
180
+ "step": 95
181
+ },
182
+ {
183
+ "epoch": 0.16,
184
+ "grad_norm": 0.9222788214683533,
185
+ "learning_rate": 1.985996037070505e-05,
186
+ "loss": 1.0575,
187
+ "mean_token_accuracy": 0.7597984328866005,
188
+ "num_tokens": 119764.0,
189
+ "step": 100
190
+ },
191
+ {
192
+ "epoch": 0.168,
193
+ "grad_norm": 0.8580851554870605,
194
+ "learning_rate": 1.9845643345292055e-05,
195
+ "loss": 0.9968,
196
+ "mean_token_accuracy": 0.7612782716751099,
197
+ "num_tokens": 125792.0,
198
+ "step": 105
199
+ },
200
+ {
201
+ "epoch": 0.176,
202
+ "grad_norm": 1.1582268476486206,
203
+ "learning_rate": 1.9830635318730155e-05,
204
+ "loss": 1.0357,
205
+ "mean_token_accuracy": 0.7526804327964782,
206
+ "num_tokens": 131722.0,
207
+ "step": 110
208
+ },
209
+ {
210
+ "epoch": 0.184,
211
+ "grad_norm": 1.106162428855896,
212
+ "learning_rate": 1.981493734433433e-05,
213
+ "loss": 1.0262,
214
+ "mean_token_accuracy": 0.7645935282111168,
215
+ "num_tokens": 137469.0,
216
+ "step": 115
217
+ },
218
+ {
219
+ "epoch": 0.192,
220
+ "grad_norm": 1.4415435791015625,
221
+ "learning_rate": 1.979855052384247e-05,
222
+ "loss": 0.955,
223
+ "mean_token_accuracy": 0.7726871728897095,
224
+ "num_tokens": 143726.0,
225
+ "step": 120
226
+ },
227
+ {
228
+ "epoch": 0.2,
229
+ "grad_norm": 1.089171290397644,
230
+ "learning_rate": 1.9781476007338058e-05,
231
+ "loss": 0.9509,
232
+ "mean_token_accuracy": 0.778785240650177,
233
+ "num_tokens": 149903.0,
234
+ "step": 125
235
+ },
236
+ {
237
+ "epoch": 0.208,
238
+ "grad_norm": 1.1151140928268433,
239
+ "learning_rate": 1.976371499316945e-05,
240
+ "loss": 0.8961,
241
+ "mean_token_accuracy": 0.7866656824946403,
242
+ "num_tokens": 156017.0,
243
+ "step": 130
244
+ },
245
+ {
246
+ "epoch": 0.216,
247
+ "grad_norm": 0.8687840104103088,
248
+ "learning_rate": 1.9745268727865774e-05,
249
+ "loss": 0.9853,
250
+ "mean_token_accuracy": 0.7754098773002625,
251
+ "num_tokens": 162023.0,
252
+ "step": 135
253
+ },
254
+ {
255
+ "epoch": 0.224,
256
+ "grad_norm": 1.1764326095581055,
257
+ "learning_rate": 1.9726138506049438e-05,
258
+ "loss": 0.9626,
259
+ "mean_token_accuracy": 0.7783279910683631,
260
+ "num_tokens": 167680.0,
261
+ "step": 140
262
+ },
263
+ {
264
+ "epoch": 0.232,
265
+ "grad_norm": 1.0557278394699097,
266
+ "learning_rate": 1.9706325670345276e-05,
267
+ "loss": 0.9459,
268
+ "mean_token_accuracy": 0.782708041369915,
269
+ "num_tokens": 173399.0,
270
+ "step": 145
271
+ },
272
+ {
273
+ "epoch": 0.24,
274
+ "grad_norm": 0.9362039566040039,
275
+ "learning_rate": 1.9685831611286312e-05,
276
+ "loss": 0.9103,
277
+ "mean_token_accuracy": 0.781058345735073,
278
+ "num_tokens": 179707.0,
279
+ "step": 150
280
+ },
281
+ {
282
+ "epoch": 0.248,
283
+ "grad_norm": 1.031844973564148,
284
+ "learning_rate": 1.9664657767216176e-05,
285
+ "loss": 0.9124,
286
+ "mean_token_accuracy": 0.7841480255126954,
287
+ "num_tokens": 185557.0,
288
+ "step": 155
289
+ },
290
+ {
291
+ "epoch": 0.256,
292
+ "grad_norm": 0.9480335712432861,
293
+ "learning_rate": 1.964280562418815e-05,
294
+ "loss": 0.9323,
295
+ "mean_token_accuracy": 0.7855196356773376,
296
+ "num_tokens": 191644.0,
297
+ "step": 160
298
+ },
299
+ {
300
+ "epoch": 0.264,
301
+ "grad_norm": 0.8738415837287903,
302
+ "learning_rate": 1.962027671586086e-05,
303
+ "loss": 0.9483,
304
+ "mean_token_accuracy": 0.777952316403389,
305
+ "num_tokens": 197484.0,
306
+ "step": 165
307
+ },
308
+ {
309
+ "epoch": 0.272,
310
+ "grad_norm": 0.8345744609832764,
311
+ "learning_rate": 1.9597072623390668e-05,
312
+ "loss": 0.8611,
313
+ "mean_token_accuracy": 0.7970519348978996,
314
+ "num_tokens": 203437.0,
315
+ "step": 170
316
+ },
317
+ {
318
+ "epoch": 0.28,
319
+ "grad_norm": 0.9600830674171448,
320
+ "learning_rate": 1.9573194975320672e-05,
321
+ "loss": 0.8807,
322
+ "mean_token_accuracy": 0.7881375521421432,
323
+ "num_tokens": 209364.0,
324
+ "step": 175
325
+ },
326
+ {
327
+ "epoch": 0.288,
328
+ "grad_norm": 1.1065053939819336,
329
+ "learning_rate": 1.9548645447466433e-05,
330
+ "loss": 0.8945,
331
+ "mean_token_accuracy": 0.7865997895598411,
332
+ "num_tokens": 215644.0,
333
+ "step": 180
334
+ },
335
+ {
336
+ "epoch": 0.296,
337
+ "grad_norm": 1.0388072729110718,
338
+ "learning_rate": 1.9523425762798328e-05,
339
+ "loss": 0.9352,
340
+ "mean_token_accuracy": 0.7840969815850258,
341
+ "num_tokens": 221329.0,
342
+ "step": 185
343
+ },
344
+ {
345
+ "epoch": 0.304,
346
+ "grad_norm": 1.0316041707992554,
347
+ "learning_rate": 1.949753769132067e-05,
348
+ "loss": 0.9323,
349
+ "mean_token_accuracy": 0.7798065140843391,
350
+ "num_tokens": 227125.0,
351
+ "step": 190
352
+ },
353
+ {
354
+ "epoch": 0.312,
355
+ "grad_norm": 0.925582230091095,
356
+ "learning_rate": 1.9470983049947446e-05,
357
+ "loss": 0.9131,
358
+ "mean_token_accuracy": 0.7816194474697113,
359
+ "num_tokens": 233075.0,
360
+ "step": 195
361
+ },
362
+ {
363
+ "epoch": 0.32,
364
+ "grad_norm": 1.0397557020187378,
365
+ "learning_rate": 1.944376370237481e-05,
366
+ "loss": 0.8624,
367
+ "mean_token_accuracy": 0.7949413478374481,
368
+ "num_tokens": 239090.0,
369
+ "step": 200
370
+ },
371
+ {
372
+ "epoch": 0.328,
373
+ "grad_norm": 0.9452941417694092,
374
+ "learning_rate": 1.9415881558950302e-05,
375
+ "loss": 0.8869,
376
+ "mean_token_accuracy": 0.7879748582839966,
377
+ "num_tokens": 245136.0,
378
+ "step": 205
379
+ },
380
+ {
381
+ "epoch": 0.336,
382
+ "grad_norm": 1.2039729356765747,
383
+ "learning_rate": 1.9387338576538743e-05,
384
+ "loss": 0.8851,
385
+ "mean_token_accuracy": 0.7879695892333984,
386
+ "num_tokens": 251118.0,
387
+ "step": 210
388
+ },
389
+ {
390
+ "epoch": 0.344,
391
+ "grad_norm": 0.9607964754104614,
392
+ "learning_rate": 1.935813675838491e-05,
393
+ "loss": 0.8707,
394
+ "mean_token_accuracy": 0.7867416545748711,
395
+ "num_tokens": 257070.0,
396
+ "step": 215
397
+ },
398
+ {
399
+ "epoch": 0.352,
400
+ "grad_norm": 1.0176879167556763,
401
+ "learning_rate": 1.9328278153972947e-05,
402
+ "loss": 0.9031,
403
+ "mean_token_accuracy": 0.7807257235050201,
404
+ "num_tokens": 263183.0,
405
+ "step": 220
406
+ },
407
+ {
408
+ "epoch": 0.36,
409
+ "grad_norm": 0.8885599970817566,
410
+ "learning_rate": 1.9297764858882516e-05,
411
+ "loss": 0.8765,
412
+ "mean_token_accuracy": 0.7853849545121193,
413
+ "num_tokens": 269103.0,
414
+ "step": 225
415
+ },
416
+ {
417
+ "epoch": 0.368,
418
+ "grad_norm": 1.1129798889160156,
419
+ "learning_rate": 1.9266599014641724e-05,
420
+ "loss": 0.8507,
421
+ "mean_token_accuracy": 0.7872389897704124,
422
+ "num_tokens": 274991.0,
423
+ "step": 230
424
+ },
425
+ {
426
+ "epoch": 0.376,
427
+ "grad_norm": 1.0610594749450684,
428
+ "learning_rate": 1.9234782808576823e-05,
429
+ "loss": 0.9264,
430
+ "mean_token_accuracy": 0.7737136602401733,
431
+ "num_tokens": 281227.0,
432
+ "step": 235
433
+ },
434
+ {
435
+ "epoch": 0.384,
436
+ "grad_norm": 1.1191679239273071,
437
+ "learning_rate": 1.9202318473658707e-05,
438
+ "loss": 0.9039,
439
+ "mean_token_accuracy": 0.7833364680409431,
440
+ "num_tokens": 287074.0,
441
+ "step": 240
442
+ },
443
+ {
444
+ "epoch": 0.392,
445
+ "grad_norm": 1.1191157102584839,
446
+ "learning_rate": 1.9169208288346168e-05,
447
+ "loss": 0.8519,
448
+ "mean_token_accuracy": 0.7942998081445694,
449
+ "num_tokens": 292840.0,
450
+ "step": 245
451
+ },
452
+ {
453
+ "epoch": 0.4,
454
+ "grad_norm": 1.3284661769866943,
455
+ "learning_rate": 1.913545457642601e-05,
456
+ "loss": 0.8732,
457
+ "mean_token_accuracy": 0.7943138211965561,
458
+ "num_tokens": 298270.0,
459
+ "step": 250
460
+ },
461
+ {
462
+ "epoch": 0.408,
463
+ "grad_norm": 1.1186164617538452,
464
+ "learning_rate": 1.9101059706849957e-05,
465
+ "loss": 0.9022,
466
+ "mean_token_accuracy": 0.7804520472884178,
467
+ "num_tokens": 304256.0,
468
+ "step": 255
469
+ },
470
+ {
471
+ "epoch": 0.416,
472
+ "grad_norm": 1.570860505104065,
473
+ "learning_rate": 1.906602609356838e-05,
474
+ "loss": 0.8653,
475
+ "mean_token_accuracy": 0.7954010605812073,
476
+ "num_tokens": 309886.0,
477
+ "step": 260
478
+ },
479
+ {
480
+ "epoch": 0.424,
481
+ "grad_norm": 1.4769788980484009,
482
+ "learning_rate": 1.9030356195360875e-05,
483
+ "loss": 0.9436,
484
+ "mean_token_accuracy": 0.7720165103673935,
485
+ "num_tokens": 315600.0,
486
+ "step": 265
487
+ },
488
+ {
489
+ "epoch": 0.432,
490
+ "grad_norm": 1.3115391731262207,
491
+ "learning_rate": 1.899405251566371e-05,
492
+ "loss": 0.8888,
493
+ "mean_token_accuracy": 0.7815406247973442,
494
+ "num_tokens": 321505.0,
495
+ "step": 270
496
+ },
497
+ {
498
+ "epoch": 0.44,
499
+ "grad_norm": 1.1410883665084839,
500
+ "learning_rate": 1.895711760239413e-05,
501
+ "loss": 0.8594,
502
+ "mean_token_accuracy": 0.7837367206811905,
503
+ "num_tokens": 327551.0,
504
+ "step": 275
505
+ },
506
+ {
507
+ "epoch": 0.448,
508
+ "grad_norm": 0.9591237902641296,
509
+ "learning_rate": 1.8919554047771508e-05,
510
+ "loss": 0.8754,
511
+ "mean_token_accuracy": 0.7869080483913422,
512
+ "num_tokens": 333256.0,
513
+ "step": 280
514
+ },
515
+ {
516
+ "epoch": 0.456,
517
+ "grad_norm": 1.1006039381027222,
518
+ "learning_rate": 1.8881364488135448e-05,
519
+ "loss": 0.8767,
520
+ "mean_token_accuracy": 0.7911439999938011,
521
+ "num_tokens": 339470.0,
522
+ "step": 285
523
+ },
524
+ {
525
+ "epoch": 0.464,
526
+ "grad_norm": 1.1295944452285767,
527
+ "learning_rate": 1.8842551603760725e-05,
528
+ "loss": 0.8486,
529
+ "mean_token_accuracy": 0.791024886071682,
530
+ "num_tokens": 345384.0,
531
+ "step": 290
532
+ },
533
+ {
534
+ "epoch": 0.472,
535
+ "grad_norm": 0.8775473833084106,
536
+ "learning_rate": 1.8803118118669203e-05,
537
+ "loss": 0.8566,
538
+ "mean_token_accuracy": 0.7951827242970466,
539
+ "num_tokens": 351347.0,
540
+ "step": 295
541
+ },
542
+ {
543
+ "epoch": 0.48,
544
+ "grad_norm": 1.2100735902786255,
545
+ "learning_rate": 1.8763066800438638e-05,
546
+ "loss": 0.8716,
547
+ "mean_token_accuracy": 0.789273151755333,
548
+ "num_tokens": 357564.0,
549
+ "step": 300
550
+ },
551
+ {
552
+ "epoch": 0.488,
553
+ "grad_norm": 1.5061770677566528,
554
+ "learning_rate": 1.8722400460008437e-05,
555
+ "loss": 0.9002,
556
+ "mean_token_accuracy": 0.7856774225831031,
557
+ "num_tokens": 363267.0,
558
+ "step": 305
559
+ },
560
+ {
561
+ "epoch": 0.496,
562
+ "grad_norm": 1.0282052755355835,
563
+ "learning_rate": 1.8681121951482397e-05,
564
+ "loss": 0.8413,
565
+ "mean_token_accuracy": 0.7916343569755554,
566
+ "num_tokens": 369749.0,
567
+ "step": 310
568
+ },
569
+ {
570
+ "epoch": 0.504,
571
+ "grad_norm": 1.0946552753448486,
572
+ "learning_rate": 1.8639234171928355e-05,
573
+ "loss": 0.9081,
574
+ "mean_token_accuracy": 0.784088309109211,
575
+ "num_tokens": 375785.0,
576
+ "step": 315
577
+ },
578
+ {
579
+ "epoch": 0.512,
580
+ "grad_norm": 1.677815556526184,
581
+ "learning_rate": 1.8596740061174912e-05,
582
+ "loss": 0.8838,
583
+ "mean_token_accuracy": 0.7858014374971389,
584
+ "num_tokens": 381448.0,
585
+ "step": 320
586
+ },
587
+ {
588
+ "epoch": 0.52,
589
+ "grad_norm": 1.0580253601074219,
590
+ "learning_rate": 1.855364260160507e-05,
591
+ "loss": 0.8555,
592
+ "mean_token_accuracy": 0.7877993151545525,
593
+ "num_tokens": 387194.0,
594
+ "step": 325
595
+ },
596
+ {
597
+ "epoch": 0.528,
598
+ "grad_norm": 1.1060526371002197,
599
+ "learning_rate": 1.850994481794692e-05,
600
+ "loss": 0.8072,
601
+ "mean_token_accuracy": 0.803234039247036,
602
+ "num_tokens": 393263.0,
603
+ "step": 330
604
+ },
605
+ {
606
+ "epoch": 0.536,
607
+ "grad_norm": 1.0966763496398926,
608
+ "learning_rate": 1.8465649777061377e-05,
609
+ "loss": 0.8491,
610
+ "mean_token_accuracy": 0.7901261404156685,
611
+ "num_tokens": 399258.0,
612
+ "step": 335
613
+ },
614
+ {
615
+ "epoch": 0.544,
616
+ "grad_norm": 1.3221973180770874,
617
+ "learning_rate": 1.8420760587726925e-05,
618
+ "loss": 0.8998,
619
+ "mean_token_accuracy": 0.7855966106057167,
620
+ "num_tokens": 404461.0,
621
+ "step": 340
622
+ },
623
+ {
624
+ "epoch": 0.552,
625
+ "grad_norm": 1.2236474752426147,
626
+ "learning_rate": 1.837528040042142e-05,
627
+ "loss": 0.8653,
628
+ "mean_token_accuracy": 0.7893021360039711,
629
+ "num_tokens": 410148.0,
630
+ "step": 345
631
+ },
632
+ {
633
+ "epoch": 0.56,
634
+ "grad_norm": 1.3368154764175415,
635
+ "learning_rate": 1.8329212407100996e-05,
636
+ "loss": 0.8955,
637
+ "mean_token_accuracy": 0.7823142364621163,
638
+ "num_tokens": 415591.0,
639
+ "step": 350
640
+ },
641
+ {
642
+ "epoch": 0.568,
643
+ "grad_norm": 1.0354520082473755,
644
+ "learning_rate": 1.8282559840976043e-05,
645
+ "loss": 0.861,
646
+ "mean_token_accuracy": 0.7877244621515274,
647
+ "num_tokens": 421706.0,
648
+ "step": 355
649
+ },
650
+ {
651
+ "epoch": 0.576,
652
+ "grad_norm": 1.7042676210403442,
653
+ "learning_rate": 1.8235325976284276e-05,
654
+ "loss": 0.8673,
655
+ "mean_token_accuracy": 0.7852049991488457,
656
+ "num_tokens": 427637.0,
657
+ "step": 360
658
+ },
659
+ {
660
+ "epoch": 0.584,
661
+ "grad_norm": 1.429971694946289,
662
+ "learning_rate": 1.8187514128060946e-05,
663
+ "loss": 0.8536,
664
+ "mean_token_accuracy": 0.7903238639235497,
665
+ "num_tokens": 433715.0,
666
+ "step": 365
667
+ },
668
+ {
669
+ "epoch": 0.592,
670
+ "grad_norm": 1.0030739307403564,
671
+ "learning_rate": 1.8139127651906183e-05,
672
+ "loss": 0.815,
673
+ "mean_token_accuracy": 0.795821775496006,
674
+ "num_tokens": 439724.0,
675
+ "step": 370
676
+ },
677
+ {
678
+ "epoch": 0.6,
679
+ "grad_norm": 1.2871325016021729,
680
+ "learning_rate": 1.8090169943749477e-05,
681
+ "loss": 0.8762,
682
+ "mean_token_accuracy": 0.7886411756277084,
683
+ "num_tokens": 446378.0,
684
+ "step": 375
685
+ },
686
+ {
687
+ "epoch": 0.608,
688
+ "grad_norm": 1.5208072662353516,
689
+ "learning_rate": 1.8040644439611348e-05,
690
+ "loss": 0.8861,
691
+ "mean_token_accuracy": 0.7832151293754578,
692
+ "num_tokens": 452313.0,
693
+ "step": 380
694
+ },
695
+ {
696
+ "epoch": 0.616,
697
+ "grad_norm": 1.665307641029358,
698
+ "learning_rate": 1.79905546153622e-05,
699
+ "loss": 0.851,
700
+ "mean_token_accuracy": 0.7918971106410027,
701
+ "num_tokens": 458505.0,
702
+ "step": 385
703
+ },
704
+ {
705
+ "epoch": 0.624,
706
+ "grad_norm": 1.1428550481796265,
707
+ "learning_rate": 1.7939903986478354e-05,
708
+ "loss": 0.8525,
709
+ "mean_token_accuracy": 0.7893514275550843,
710
+ "num_tokens": 464226.0,
711
+ "step": 390
712
+ },
713
+ {
714
+ "epoch": 0.632,
715
+ "grad_norm": 1.140528917312622,
716
+ "learning_rate": 1.7888696107795343e-05,
717
+ "loss": 0.8131,
718
+ "mean_token_accuracy": 0.8022583290934563,
719
+ "num_tokens": 470086.0,
720
+ "step": 395
721
+ },
722
+ {
723
+ "epoch": 0.64,
724
+ "grad_norm": 1.0871620178222656,
725
+ "learning_rate": 1.78369345732584e-05,
726
+ "loss": 0.8211,
727
+ "mean_token_accuracy": 0.8043899014592171,
728
+ "num_tokens": 475819.0,
729
+ "step": 400
730
+ },
731
+ {
732
+ "epoch": 0.648,
733
+ "grad_norm": 1.1760526895523071,
734
+ "learning_rate": 1.7784623015670237e-05,
735
+ "loss": 0.8811,
736
+ "mean_token_accuracy": 0.7891304656863213,
737
+ "num_tokens": 482091.0,
738
+ "step": 405
739
+ },
740
+ {
741
+ "epoch": 0.656,
742
+ "grad_norm": 1.117303729057312,
743
+ "learning_rate": 1.7731765106436073e-05,
744
+ "loss": 0.845,
745
+ "mean_token_accuracy": 0.7931285366415978,
746
+ "num_tokens": 488068.0,
747
+ "step": 410
748
+ },
749
+ {
750
+ "epoch": 0.664,
751
+ "grad_norm": 1.2734001874923706,
752
+ "learning_rate": 1.767836455530598e-05,
753
+ "loss": 0.907,
754
+ "mean_token_accuracy": 0.783031564950943,
755
+ "num_tokens": 493871.0,
756
+ "step": 415
757
+ },
758
+ {
759
+ "epoch": 0.672,
760
+ "grad_norm": 1.4908583164215088,
761
+ "learning_rate": 1.762442511011448e-05,
762
+ "loss": 0.8853,
763
+ "mean_token_accuracy": 0.7815811723470688,
764
+ "num_tokens": 499751.0,
765
+ "step": 420
766
+ },
767
+ {
768
+ "epoch": 0.68,
769
+ "grad_norm": 1.2894372940063477,
770
+ "learning_rate": 1.7569950556517566e-05,
771
+ "loss": 0.8461,
772
+ "mean_token_accuracy": 0.7941734075546265,
773
+ "num_tokens": 505816.0,
774
+ "step": 425
775
+ },
776
+ {
777
+ "epoch": 0.688,
778
+ "grad_norm": 1.237874150276184,
779
+ "learning_rate": 1.7514944717726962e-05,
780
+ "loss": 0.8388,
781
+ "mean_token_accuracy": 0.7915323451161385,
782
+ "num_tokens": 511745.0,
783
+ "step": 430
784
+ },
785
+ {
786
+ "epoch": 0.696,
787
+ "grad_norm": 1.0999034643173218,
788
+ "learning_rate": 1.7459411454241822e-05,
789
+ "loss": 0.836,
790
+ "mean_token_accuracy": 0.7994311302900314,
791
+ "num_tokens": 517968.0,
792
+ "step": 435
793
+ },
794
+ {
795
+ "epoch": 0.704,
796
+ "grad_norm": 1.4175597429275513,
797
+ "learning_rate": 1.7403354663577782e-05,
798
+ "loss": 0.8139,
799
+ "mean_token_accuracy": 0.799339534342289,
800
+ "num_tokens": 524079.0,
801
+ "step": 440
802
+ },
803
+ {
804
+ "epoch": 0.712,
805
+ "grad_norm": 1.1123442649841309,
806
+ "learning_rate": 1.7346778279993417e-05,
807
+ "loss": 0.8159,
808
+ "mean_token_accuracy": 0.7988538891077042,
809
+ "num_tokens": 530403.0,
810
+ "step": 445
811
+ },
812
+ {
813
+ "epoch": 0.72,
814
+ "grad_norm": 1.0958361625671387,
815
+ "learning_rate": 1.7289686274214116e-05,
816
+ "loss": 0.8934,
817
+ "mean_token_accuracy": 0.783437828719616,
818
+ "num_tokens": 536165.0,
819
+ "step": 450
820
+ },
821
+ {
822
+ "epoch": 0.728,
823
+ "grad_norm": 1.0283623933792114,
824
+ "learning_rate": 1.7232082653153422e-05,
825
+ "loss": 0.8398,
826
+ "mean_token_accuracy": 0.784825636446476,
827
+ "num_tokens": 542277.0,
828
+ "step": 455
829
+ },
830
+ {
831
+ "epoch": 0.736,
832
+ "grad_norm": 1.2006138563156128,
833
+ "learning_rate": 1.717397145963179e-05,
834
+ "loss": 0.8515,
835
+ "mean_token_accuracy": 0.7904125943779945,
836
+ "num_tokens": 548374.0,
837
+ "step": 460
838
+ },
839
+ {
840
+ "epoch": 0.744,
841
+ "grad_norm": 1.0756789445877075,
842
+ "learning_rate": 1.7115356772092854e-05,
843
+ "loss": 0.8633,
844
+ "mean_token_accuracy": 0.7910906136035919,
845
+ "num_tokens": 554387.0,
846
+ "step": 465
847
+ },
848
+ {
849
+ "epoch": 0.752,
850
+ "grad_norm": 1.108975887298584,
851
+ "learning_rate": 1.705624270431721e-05,
852
+ "loss": 0.8358,
853
+ "mean_token_accuracy": 0.7929062396287918,
854
+ "num_tokens": 560326.0,
855
+ "step": 470
856
+ },
857
+ {
858
+ "epoch": 0.76,
859
+ "grad_norm": 1.1952314376831055,
860
+ "learning_rate": 1.6996633405133656e-05,
861
+ "loss": 0.8402,
862
+ "mean_token_accuracy": 0.7873096525669098,
863
+ "num_tokens": 566113.0,
864
+ "step": 475
865
+ },
866
+ {
867
+ "epoch": 0.768,
868
+ "grad_norm": 1.0153251886367798,
869
+ "learning_rate": 1.693653305812805e-05,
870
+ "loss": 0.8155,
871
+ "mean_token_accuracy": 0.7994796469807625,
872
+ "num_tokens": 572252.0,
873
+ "step": 480
874
+ },
875
+ {
876
+ "epoch": 0.776,
877
+ "grad_norm": 1.128273844718933,
878
+ "learning_rate": 1.6875945881349676e-05,
879
+ "loss": 0.8804,
880
+ "mean_token_accuracy": 0.7862005636096001,
881
+ "num_tokens": 578388.0,
882
+ "step": 485
883
+ },
884
+ {
885
+ "epoch": 0.784,
886
+ "grad_norm": 1.0263817310333252,
887
+ "learning_rate": 1.68148761270152e-05,
888
+ "loss": 0.8718,
889
+ "mean_token_accuracy": 0.778036293387413,
890
+ "num_tokens": 584367.0,
891
+ "step": 490
892
+ },
893
+ {
894
+ "epoch": 0.792,
895
+ "grad_norm": 1.2925713062286377,
896
+ "learning_rate": 1.6753328081210244e-05,
897
+ "loss": 0.8579,
898
+ "mean_token_accuracy": 0.7924200773239136,
899
+ "num_tokens": 590341.0,
900
+ "step": 495
901
+ },
902
+ {
903
+ "epoch": 0.8,
904
+ "grad_norm": 1.036569356918335,
905
+ "learning_rate": 1.6691306063588583e-05,
906
+ "loss": 0.8756,
907
+ "mean_token_accuracy": 0.7841189652681351,
908
+ "num_tokens": 596177.0,
909
+ "step": 500
910
+ },
911
+ {
912
+ "epoch": 0.808,
913
+ "grad_norm": 1.3352292776107788,
914
+ "learning_rate": 1.6628814427068954e-05,
915
+ "loss": 0.8344,
916
+ "mean_token_accuracy": 0.792496457695961,
917
+ "num_tokens": 602474.0,
918
+ "step": 505
919
+ },
920
+ {
921
+ "epoch": 0.816,
922
+ "grad_norm": 1.142318606376648,
923
+ "learning_rate": 1.6565857557529567e-05,
924
+ "loss": 0.8211,
925
+ "mean_token_accuracy": 0.7938537418842315,
926
+ "num_tokens": 608412.0,
927
+ "step": 510
928
+ },
929
+ {
930
+ "epoch": 0.824,
931
+ "grad_norm": 1.1519471406936646,
932
+ "learning_rate": 1.650243987350029e-05,
933
+ "loss": 0.84,
934
+ "mean_token_accuracy": 0.7874479576945305,
935
+ "num_tokens": 614525.0,
936
+ "step": 515
937
+ },
938
+ {
939
+ "epoch": 0.832,
940
+ "grad_norm": 1.3666894435882568,
941
+ "learning_rate": 1.643856582585254e-05,
942
+ "loss": 0.8436,
943
+ "mean_token_accuracy": 0.7965899407863617,
944
+ "num_tokens": 620623.0,
945
+ "step": 520
946
+ },
947
+ {
948
+ "epoch": 0.84,
949
+ "grad_norm": 1.048234462738037,
950
+ "learning_rate": 1.63742398974869e-05,
951
+ "loss": 0.8252,
952
+ "mean_token_accuracy": 0.7954581871628761,
953
+ "num_tokens": 626599.0,
954
+ "step": 525
955
+ },
956
+ {
957
+ "epoch": 0.848,
958
+ "grad_norm": 1.2072833776474,
959
+ "learning_rate": 1.6309466603018497e-05,
960
+ "loss": 0.8156,
961
+ "mean_token_accuracy": 0.7952276915311813,
962
+ "num_tokens": 632517.0,
963
+ "step": 530
964
+ },
965
+ {
966
+ "epoch": 0.856,
967
+ "grad_norm": 1.244985818862915,
968
+ "learning_rate": 1.624425048846016e-05,
969
+ "loss": 0.858,
970
+ "mean_token_accuracy": 0.7910035625100136,
971
+ "num_tokens": 638566.0,
972
+ "step": 535
973
+ },
974
+ {
975
+ "epoch": 0.864,
976
+ "grad_norm": 1.4564027786254883,
977
+ "learning_rate": 1.6178596130903345e-05,
978
+ "loss": 0.818,
979
+ "mean_token_accuracy": 0.7960182785987854,
980
+ "num_tokens": 644332.0,
981
+ "step": 540
982
+ },
983
+ {
984
+ "epoch": 0.872,
985
+ "grad_norm": 1.243749737739563,
986
+ "learning_rate": 1.611250813819692e-05,
987
+ "loss": 0.8009,
988
+ "mean_token_accuracy": 0.7983238011598587,
989
+ "num_tokens": 650329.0,
990
+ "step": 545
991
+ },
992
+ {
993
+ "epoch": 0.88,
994
+ "grad_norm": 1.1024765968322754,
995
+ "learning_rate": 1.6045991148623752e-05,
996
+ "loss": 0.8407,
997
+ "mean_token_accuracy": 0.7944921687245369,
998
+ "num_tokens": 656266.0,
999
+ "step": 550
1000
+ },
1001
+ {
1002
+ "epoch": 0.888,
1003
+ "grad_norm": 1.2258902788162231,
1004
+ "learning_rate": 1.597904983057519e-05,
1005
+ "loss": 0.8158,
1006
+ "mean_token_accuracy": 0.7947202190756798,
1007
+ "num_tokens": 661977.0,
1008
+ "step": 555
1009
+ },
1010
+ {
1011
+ "epoch": 0.896,
1012
+ "grad_norm": 1.008183240890503,
1013
+ "learning_rate": 1.591168888222342e-05,
1014
+ "loss": 0.8302,
1015
+ "mean_token_accuracy": 0.7910092756152153,
1016
+ "num_tokens": 668076.0,
1017
+ "step": 560
1018
+ },
1019
+ {
1020
+ "epoch": 0.904,
1021
+ "grad_norm": 1.5126904249191284,
1022
+ "learning_rate": 1.5843913031191722e-05,
1023
+ "loss": 0.831,
1024
+ "mean_token_accuracy": 0.7950059458613395,
1025
+ "num_tokens": 674078.0,
1026
+ "step": 565
1027
+ },
1028
+ {
1029
+ "epoch": 0.912,
1030
+ "grad_norm": 1.405203104019165,
1031
+ "learning_rate": 1.5775727034222675e-05,
1032
+ "loss": 0.8442,
1033
+ "mean_token_accuracy": 0.7927875980734825,
1034
+ "num_tokens": 679908.0,
1035
+ "step": 570
1036
+ },
1037
+ {
1038
+ "epoch": 0.92,
1039
+ "grad_norm": 1.0306074619293213,
1040
+ "learning_rate": 1.570713567684432e-05,
1041
+ "loss": 0.8556,
1042
+ "mean_token_accuracy": 0.7945606961846352,
1043
+ "num_tokens": 685791.0,
1044
+ "step": 575
1045
+ },
1046
+ {
1047
+ "epoch": 0.928,
1048
+ "grad_norm": 1.0438185930252075,
1049
+ "learning_rate": 1.5638143773034268e-05,
1050
+ "loss": 0.8518,
1051
+ "mean_token_accuracy": 0.7933464452624321,
1052
+ "num_tokens": 691871.0,
1053
+ "step": 580
1054
+ },
1055
+ {
1056
+ "epoch": 0.936,
1057
+ "grad_norm": 1.2007946968078613,
1058
+ "learning_rate": 1.556875616488188e-05,
1059
+ "loss": 0.8222,
1060
+ "mean_token_accuracy": 0.7947555348277092,
1061
+ "num_tokens": 697838.0,
1062
+ "step": 585
1063
+ },
1064
+ {
1065
+ "epoch": 0.944,
1066
+ "grad_norm": 1.1926593780517578,
1067
+ "learning_rate": 1.54989777222484e-05,
1068
+ "loss": 0.8399,
1069
+ "mean_token_accuracy": 0.7944862857460976,
1070
+ "num_tokens": 703760.0,
1071
+ "step": 590
1072
+ },
1073
+ {
1074
+ "epoch": 0.952,
1075
+ "grad_norm": 1.5483673810958862,
1076
+ "learning_rate": 1.5428813342425177e-05,
1077
+ "loss": 0.8439,
1078
+ "mean_token_accuracy": 0.7925432935357094,
1079
+ "num_tokens": 709673.0,
1080
+ "step": 595
1081
+ },
1082
+ {
1083
+ "epoch": 0.96,
1084
+ "grad_norm": 1.1998529434204102,
1085
+ "learning_rate": 1.5358267949789968e-05,
1086
+ "loss": 0.8173,
1087
+ "mean_token_accuracy": 0.8004210472106934,
1088
+ "num_tokens": 715807.0,
1089
+ "step": 600
1090
+ },
1091
+ {
1092
+ "epoch": 0.968,
1093
+ "grad_norm": 1.2061858177185059,
1094
+ "learning_rate": 1.528734649546132e-05,
1095
+ "loss": 0.8557,
1096
+ "mean_token_accuracy": 0.7907719686627388,
1097
+ "num_tokens": 721659.0,
1098
+ "step": 605
1099
+ },
1100
+ {
1101
+ "epoch": 0.976,
1102
+ "grad_norm": 1.0802125930786133,
1103
+ "learning_rate": 1.5216053956951081e-05,
1104
+ "loss": 0.8225,
1105
+ "mean_token_accuracy": 0.7847193196415901,
1106
+ "num_tokens": 727941.0,
1107
+ "step": 610
1108
+ },
1109
+ {
1110
+ "epoch": 0.984,
1111
+ "grad_norm": 1.1718847751617432,
1112
+ "learning_rate": 1.5144395337815066e-05,
1113
+ "loss": 0.8139,
1114
+ "mean_token_accuracy": 0.8009869039058686,
1115
+ "num_tokens": 734053.0,
1116
+ "step": 615
1117
+ },
1118
+ {
1119
+ "epoch": 0.992,
1120
+ "grad_norm": 1.477616310119629,
1121
+ "learning_rate": 1.507237566730189e-05,
1122
+ "loss": 0.8273,
1123
+ "mean_token_accuracy": 0.7916375547647476,
1124
+ "num_tokens": 740050.0,
1125
+ "step": 620
1126
+ },
1127
+ {
1128
+ "epoch": 1.0,
1129
+ "grad_norm": 1.284795880317688,
1130
+ "learning_rate": 1.5000000000000002e-05,
1131
+ "loss": 0.8571,
1132
+ "mean_token_accuracy": 0.7930364921689034,
1133
+ "num_tokens": 745874.0,
1134
+ "step": 625
1135
+ },
1136
+ {
1137
+ "epoch": 1.008,
1138
+ "grad_norm": 1.4859092235565186,
1139
+ "learning_rate": 1.4927273415482916e-05,
1140
+ "loss": 0.8252,
1141
+ "mean_token_accuracy": 0.8047172293066979,
1142
+ "num_tokens": 752111.0,
1143
+ "step": 630
1144
+ },
1145
+ {
1146
+ "epoch": 1.016,
1147
+ "grad_norm": 1.6376901865005493,
1148
+ "learning_rate": 1.485420101795274e-05,
1149
+ "loss": 0.7909,
1150
+ "mean_token_accuracy": 0.8041815027594567,
1151
+ "num_tokens": 758152.0,
1152
+ "step": 635
1153
+ },
1154
+ {
1155
+ "epoch": 1.024,
1156
+ "grad_norm": 1.3351467847824097,
1157
+ "learning_rate": 1.4780787935881925e-05,
1158
+ "loss": 0.8404,
1159
+ "mean_token_accuracy": 0.7902583315968513,
1160
+ "num_tokens": 764197.0,
1161
+ "step": 640
1162
+ },
1163
+ {
1164
+ "epoch": 1.032,
1165
+ "grad_norm": 1.3304749727249146,
1166
+ "learning_rate": 1.470703932165333e-05,
1167
+ "loss": 0.8212,
1168
+ "mean_token_accuracy": 0.7978306338191032,
1169
+ "num_tokens": 770075.0,
1170
+ "step": 645
1171
+ },
1172
+ {
1173
+ "epoch": 1.04,
1174
+ "grad_norm": 1.0387990474700928,
1175
+ "learning_rate": 1.463296035119862e-05,
1176
+ "loss": 0.8376,
1177
+ "mean_token_accuracy": 0.7961583107709884,
1178
+ "num_tokens": 776252.0,
1179
+ "step": 650
1180
+ },
1181
+ {
1182
+ "epoch": 1.048,
1183
+ "grad_norm": 1.4199416637420654,
1184
+ "learning_rate": 1.4558556223635004e-05,
1185
+ "loss": 0.8514,
1186
+ "mean_token_accuracy": 0.7941056564450264,
1187
+ "num_tokens": 781797.0,
1188
+ "step": 655
1189
+ },
1190
+ {
1191
+ "epoch": 1.056,
1192
+ "grad_norm": 0.9742839336395264,
1193
+ "learning_rate": 1.4483832160900326e-05,
1194
+ "loss": 0.84,
1195
+ "mean_token_accuracy": 0.7879764214158058,
1196
+ "num_tokens": 787607.0,
1197
+ "step": 660
1198
+ },
1199
+ {
1200
+ "epoch": 1.064,
1201
+ "grad_norm": 1.1152559518814087,
1202
+ "learning_rate": 1.4408793407386587e-05,
1203
+ "loss": 0.8068,
1204
+ "mean_token_accuracy": 0.7992551028728485,
1205
+ "num_tokens": 793146.0,
1206
+ "step": 665
1207
+ },
1208
+ {
1209
+ "epoch": 1.072,
1210
+ "grad_norm": 1.1599243879318237,
1211
+ "learning_rate": 1.4333445229571874e-05,
1212
+ "loss": 0.833,
1213
+ "mean_token_accuracy": 0.7881477907299995,
1214
+ "num_tokens": 799067.0,
1215
+ "step": 670
1216
+ },
1217
+ {
1218
+ "epoch": 1.08,
1219
+ "grad_norm": 1.4914982318878174,
1220
+ "learning_rate": 1.4257792915650728e-05,
1221
+ "loss": 0.8199,
1222
+ "mean_token_accuracy": 0.7969856977462768,
1223
+ "num_tokens": 805016.0,
1224
+ "step": 675
1225
+ },
1226
+ {
1227
+ "epoch": 1.088,
1228
+ "grad_norm": 1.4309738874435425,
1229
+ "learning_rate": 1.4181841775163014e-05,
1230
+ "loss": 0.8052,
1231
+ "mean_token_accuracy": 0.800905755162239,
1232
+ "num_tokens": 810625.0,
1233
+ "step": 680
1234
+ },
1235
+ {
1236
+ "epoch": 1.096,
1237
+ "grad_norm": 1.109816312789917,
1238
+ "learning_rate": 1.4105597138621281e-05,
1239
+ "loss": 0.8151,
1240
+ "mean_token_accuracy": 0.795543110370636,
1241
+ "num_tokens": 816596.0,
1242
+ "step": 685
1243
+ },
1244
+ {
1245
+ "epoch": 1.104,
1246
+ "grad_norm": 1.2419342994689941,
1247
+ "learning_rate": 1.4029064357136628e-05,
1248
+ "loss": 0.8082,
1249
+ "mean_token_accuracy": 0.8007919058203697,
1250
+ "num_tokens": 822855.0,
1251
+ "step": 690
1252
+ },
1253
+ {
1254
+ "epoch": 1.112,
1255
+ "grad_norm": 1.24587082862854,
1256
+ "learning_rate": 1.3952248802043166e-05,
1257
+ "loss": 0.8344,
1258
+ "mean_token_accuracy": 0.796556057035923,
1259
+ "num_tokens": 828830.0,
1260
+ "step": 695
1261
+ },
1262
+ {
1263
+ "epoch": 1.12,
1264
+ "grad_norm": 1.063744068145752,
1265
+ "learning_rate": 1.3875155864521031e-05,
1266
+ "loss": 0.7801,
1267
+ "mean_token_accuracy": 0.8025736212730408,
1268
+ "num_tokens": 834959.0,
1269
+ "step": 700
1270
+ },
1271
+ {
1272
+ "epoch": 1.1280000000000001,
1273
+ "grad_norm": 1.1293323040008545,
1274
+ "learning_rate": 1.3797790955218014e-05,
1275
+ "loss": 0.7981,
1276
+ "mean_token_accuracy": 0.7934223636984825,
1277
+ "num_tokens": 841276.0,
1278
+ "step": 705
1279
+ },
1280
+ {
1281
+ "epoch": 1.1360000000000001,
1282
+ "grad_norm": 1.3948642015457153,
1283
+ "learning_rate": 1.3720159503869816e-05,
1284
+ "loss": 0.8723,
1285
+ "mean_token_accuracy": 0.7871071428060532,
1286
+ "num_tokens": 847154.0,
1287
+ "step": 710
1288
+ },
1289
+ {
1290
+ "epoch": 1.144,
1291
+ "grad_norm": 1.2734830379486084,
1292
+ "learning_rate": 1.3642266958918985e-05,
1293
+ "loss": 0.8406,
1294
+ "mean_token_accuracy": 0.7929588705301285,
1295
+ "num_tokens": 853193.0,
1296
+ "step": 715
1297
+ },
1298
+ {
1299
+ "epoch": 1.152,
1300
+ "grad_norm": 1.2411038875579834,
1301
+ "learning_rate": 1.3564118787132507e-05,
1302
+ "loss": 0.8195,
1303
+ "mean_token_accuracy": 0.8023369893431663,
1304
+ "num_tokens": 859180.0,
1305
+ "step": 720
1306
+ },
1307
+ {
1308
+ "epoch": 1.16,
1309
+ "grad_norm": 1.1774182319641113,
1310
+ "learning_rate": 1.3485720473218153e-05,
1311
+ "loss": 0.8442,
1312
+ "mean_token_accuracy": 0.7901751175522804,
1313
+ "num_tokens": 865016.0,
1314
+ "step": 725
1315
+ },
1316
+ {
1317
+ "epoch": 1.168,
1318
+ "grad_norm": 1.3339149951934814,
1319
+ "learning_rate": 1.340707751943952e-05,
1320
+ "loss": 0.8455,
1321
+ "mean_token_accuracy": 0.79264917075634,
1322
+ "num_tokens": 870974.0,
1323
+ "step": 730
1324
+ },
1325
+ {
1326
+ "epoch": 1.176,
1327
+ "grad_norm": 1.2754383087158203,
1328
+ "learning_rate": 1.3328195445229869e-05,
1329
+ "loss": 0.784,
1330
+ "mean_token_accuracy": 0.809282261133194,
1331
+ "num_tokens": 876702.0,
1332
+ "step": 735
1333
+ },
1334
+ {
1335
+ "epoch": 1.184,
1336
+ "grad_norm": 1.0842269659042358,
1337
+ "learning_rate": 1.3249079786804765e-05,
1338
+ "loss": 0.8386,
1339
+ "mean_token_accuracy": 0.795893557369709,
1340
+ "num_tokens": 882614.0,
1341
+ "step": 740
1342
+ },
1343
+ {
1344
+ "epoch": 1.192,
1345
+ "grad_norm": 1.134906530380249,
1346
+ "learning_rate": 1.316973609677352e-05,
1347
+ "loss": 0.8076,
1348
+ "mean_token_accuracy": 0.8011891514062881,
1349
+ "num_tokens": 888767.0,
1350
+ "step": 745
1351
+ },
1352
+ {
1353
+ "epoch": 1.2,
1354
+ "grad_norm": 1.200283169746399,
1355
+ "learning_rate": 1.3090169943749475e-05,
1356
+ "loss": 0.8324,
1357
+ "mean_token_accuracy": 0.7995044961571693,
1358
+ "num_tokens": 894065.0,
1359
+ "step": 750
1360
+ },
1361
+ {
1362
+ "epoch": 1.208,
1363
+ "grad_norm": 1.1556944847106934,
1364
+ "learning_rate": 1.3010386911959207e-05,
1365
+ "loss": 0.846,
1366
+ "mean_token_accuracy": 0.7956264927983284,
1367
+ "num_tokens": 899852.0,
1368
+ "step": 755
1369
+ },
1370
+ {
1371
+ "epoch": 1.216,
1372
+ "grad_norm": 1.079065203666687,
1373
+ "learning_rate": 1.2930392600850574e-05,
1374
+ "loss": 0.8138,
1375
+ "mean_token_accuracy": 0.8026395171880722,
1376
+ "num_tokens": 905691.0,
1377
+ "step": 760
1378
+ },
1379
+ {
1380
+ "epoch": 1.224,
1381
+ "grad_norm": 1.322657823562622,
1382
+ "learning_rate": 1.2850192624699762e-05,
1383
+ "loss": 0.8469,
1384
+ "mean_token_accuracy": 0.7925262525677681,
1385
+ "num_tokens": 911383.0,
1386
+ "step": 765
1387
+ },
1388
+ {
1389
+ "epoch": 1.232,
1390
+ "grad_norm": 1.3103306293487549,
1391
+ "learning_rate": 1.2769792612217224e-05,
1392
+ "loss": 0.8376,
1393
+ "mean_token_accuracy": 0.7923985093832016,
1394
+ "num_tokens": 917061.0,
1395
+ "step": 770
1396
+ },
1397
+ {
1398
+ "epoch": 1.24,
1399
+ "grad_norm": 1.2759097814559937,
1400
+ "learning_rate": 1.2689198206152657e-05,
1401
+ "loss": 0.7747,
1402
+ "mean_token_accuracy": 0.8024341821670532,
1403
+ "num_tokens": 922637.0,
1404
+ "step": 775
1405
+ },
1406
+ {
1407
+ "epoch": 1.248,
1408
+ "grad_norm": 1.4393937587738037,
1409
+ "learning_rate": 1.2608415062898971e-05,
1410
+ "loss": 0.8061,
1411
+ "mean_token_accuracy": 0.7970004111528397,
1412
+ "num_tokens": 929037.0,
1413
+ "step": 780
1414
+ },
1415
+ {
1416
+ "epoch": 1.256,
1417
+ "grad_norm": 1.2215389013290405,
1418
+ "learning_rate": 1.2527448852095295e-05,
1419
+ "loss": 0.8134,
1420
+ "mean_token_accuracy": 0.7974464222788811,
1421
+ "num_tokens": 935182.0,
1422
+ "step": 785
1423
+ },
1424
+ {
1425
+ "epoch": 1.264,
1426
+ "grad_norm": 1.6864231824874878,
1427
+ "learning_rate": 1.2446305256229074e-05,
1428
+ "loss": 0.8433,
1429
+ "mean_token_accuracy": 0.7880089089274407,
1430
+ "num_tokens": 940915.0,
1431
+ "step": 790
1432
+ },
1433
+ {
1434
+ "epoch": 1.272,
1435
+ "grad_norm": 1.2228368520736694,
1436
+ "learning_rate": 1.236498997023725e-05,
1437
+ "loss": 0.8315,
1438
+ "mean_token_accuracy": 0.7921820938587188,
1439
+ "num_tokens": 946964.0,
1440
+ "step": 795
1441
+ },
1442
+ {
1443
+ "epoch": 1.28,
1444
+ "grad_norm": 1.1751633882522583,
1445
+ "learning_rate": 1.2283508701106559e-05,
1446
+ "loss": 0.8258,
1447
+ "mean_token_accuracy": 0.7954315572977066,
1448
+ "num_tokens": 952867.0,
1449
+ "step": 800
1450
+ },
1451
+ {
1452
+ "epoch": 1.288,
1453
+ "grad_norm": 1.2997604608535767,
1454
+ "learning_rate": 1.2201867167473015e-05,
1455
+ "loss": 0.7963,
1456
+ "mean_token_accuracy": 0.79445910602808,
1457
+ "num_tokens": 959081.0,
1458
+ "step": 805
1459
+ },
1460
+ {
1461
+ "epoch": 1.296,
1462
+ "grad_norm": 1.10807466506958,
1463
+ "learning_rate": 1.212007109922055e-05,
1464
+ "loss": 0.8128,
1465
+ "mean_token_accuracy": 0.7948173075914383,
1466
+ "num_tokens": 965090.0,
1467
+ "step": 810
1468
+ },
1469
+ {
1470
+ "epoch": 1.304,
1471
+ "grad_norm": 1.3159791231155396,
1472
+ "learning_rate": 1.203812623707885e-05,
1473
+ "loss": 0.829,
1474
+ "mean_token_accuracy": 0.7960757419466973,
1475
+ "num_tokens": 971348.0,
1476
+ "step": 815
1477
+ },
1478
+ {
1479
+ "epoch": 1.312,
1480
+ "grad_norm": 1.5525962114334106,
1481
+ "learning_rate": 1.1956038332220484e-05,
1482
+ "loss": 0.8072,
1483
+ "mean_token_accuracy": 0.8035811170935631,
1484
+ "num_tokens": 976874.0,
1485
+ "step": 820
1486
+ },
1487
+ {
1488
+ "epoch": 1.32,
1489
+ "grad_norm": 1.1736457347869873,
1490
+ "learning_rate": 1.187381314585725e-05,
1491
+ "loss": 0.791,
1492
+ "mean_token_accuracy": 0.8109298884868622,
1493
+ "num_tokens": 983026.0,
1494
+ "step": 825
1495
+ },
1496
+ {
1497
+ "epoch": 1.328,
1498
+ "grad_norm": 1.191405177116394,
1499
+ "learning_rate": 1.1791456448835825e-05,
1500
+ "loss": 0.8268,
1501
+ "mean_token_accuracy": 0.7918137550354004,
1502
+ "num_tokens": 989001.0,
1503
+ "step": 830
1504
+ },
1505
+ {
1506
+ "epoch": 1.336,
1507
+ "grad_norm": 1.3512290716171265,
1508
+ "learning_rate": 1.1708974021232768e-05,
1509
+ "loss": 0.8444,
1510
+ "mean_token_accuracy": 0.790115873515606,
1511
+ "num_tokens": 995043.0,
1512
+ "step": 835
1513
+ },
1514
+ {
1515
+ "epoch": 1.3439999999999999,
1516
+ "grad_norm": 1.274695634841919,
1517
+ "learning_rate": 1.1626371651948839e-05,
1518
+ "loss": 0.8171,
1519
+ "mean_token_accuracy": 0.7923313453793526,
1520
+ "num_tokens": 1000882.0,
1521
+ "step": 840
1522
+ },
1523
+ {
1524
+ "epoch": 1.3519999999999999,
1525
+ "grad_norm": 1.3958570957183838,
1526
+ "learning_rate": 1.1543655138302714e-05,
1527
+ "loss": 0.8531,
1528
+ "mean_token_accuracy": 0.7938149958848953,
1529
+ "num_tokens": 1006642.0,
1530
+ "step": 845
1531
+ },
1532
+ {
1533
+ "epoch": 1.3599999999999999,
1534
+ "grad_norm": 1.266419529914856,
1535
+ "learning_rate": 1.1460830285624119e-05,
1536
+ "loss": 0.7806,
1537
+ "mean_token_accuracy": 0.8017867222428322,
1538
+ "num_tokens": 1012973.0,
1539
+ "step": 850
1540
+ },
1541
+ {
1542
+ "epoch": 1.3679999999999999,
1543
+ "grad_norm": 1.3229976892471313,
1544
+ "learning_rate": 1.137790290684638e-05,
1545
+ "loss": 0.8035,
1546
+ "mean_token_accuracy": 0.8037858188152314,
1547
+ "num_tokens": 1019010.0,
1548
+ "step": 855
1549
+ },
1550
+ {
1551
+ "epoch": 1.376,
1552
+ "grad_norm": 1.2124758958816528,
1553
+ "learning_rate": 1.129487882209847e-05,
1554
+ "loss": 0.8166,
1555
+ "mean_token_accuracy": 0.7993422210216522,
1556
+ "num_tokens": 1025168.0,
1557
+ "step": 860
1558
+ },
1559
+ {
1560
+ "epoch": 1.384,
1561
+ "grad_norm": 1.2201908826828003,
1562
+ "learning_rate": 1.1211763858296507e-05,
1563
+ "loss": 0.8128,
1564
+ "mean_token_accuracy": 0.7982980251312256,
1565
+ "num_tokens": 1031047.0,
1566
+ "step": 865
1567
+ },
1568
+ {
1569
+ "epoch": 1.392,
1570
+ "grad_norm": 1.0084983110427856,
1571
+ "learning_rate": 1.1128563848734817e-05,
1572
+ "loss": 0.8105,
1573
+ "mean_token_accuracy": 0.8003023475408554,
1574
+ "num_tokens": 1037575.0,
1575
+ "step": 870
1576
+ },
1577
+ {
1578
+ "epoch": 1.4,
1579
+ "grad_norm": 1.1894408464431763,
1580
+ "learning_rate": 1.1045284632676535e-05,
1581
+ "loss": 0.7904,
1582
+ "mean_token_accuracy": 0.8060181707143783,
1583
+ "num_tokens": 1043577.0,
1584
+ "step": 875
1585
+ },
1586
+ {
1587
+ "epoch": 1.408,
1588
+ "grad_norm": 1.233709454536438,
1589
+ "learning_rate": 1.0961932054943778e-05,
1590
+ "loss": 0.8299,
1591
+ "mean_token_accuracy": 0.7969600349664688,
1592
+ "num_tokens": 1049735.0,
1593
+ "step": 880
1594
+ },
1595
+ {
1596
+ "epoch": 1.416,
1597
+ "grad_norm": 1.170153260231018,
1598
+ "learning_rate": 1.0878511965507435e-05,
1599
+ "loss": 0.801,
1600
+ "mean_token_accuracy": 0.7926143258810043,
1601
+ "num_tokens": 1055885.0,
1602
+ "step": 885
1603
+ },
1604
+ {
1605
+ "epoch": 1.424,
1606
+ "grad_norm": 1.1802887916564941,
1607
+ "learning_rate": 1.07950302190766e-05,
1608
+ "loss": 0.8271,
1609
+ "mean_token_accuracy": 0.7875034034252166,
1610
+ "num_tokens": 1062382.0,
1611
+ "step": 890
1612
+ },
1613
+ {
1614
+ "epoch": 1.432,
1615
+ "grad_norm": 1.1903148889541626,
1616
+ "learning_rate": 1.071149267468767e-05,
1617
+ "loss": 0.8153,
1618
+ "mean_token_accuracy": 0.8013403192162514,
1619
+ "num_tokens": 1068374.0,
1620
+ "step": 895
1621
+ },
1622
+ {
1623
+ "epoch": 1.44,
1624
+ "grad_norm": 1.1974633932113647,
1625
+ "learning_rate": 1.0627905195293135e-05,
1626
+ "loss": 0.7853,
1627
+ "mean_token_accuracy": 0.7982048079371452,
1628
+ "num_tokens": 1074097.0,
1629
+ "step": 900
1630
+ },
1631
+ {
1632
+ "epoch": 1.448,
1633
+ "grad_norm": 1.3149733543395996,
1634
+ "learning_rate": 1.0544273647350091e-05,
1635
+ "loss": 0.7884,
1636
+ "mean_token_accuracy": 0.7991619855165482,
1637
+ "num_tokens": 1080138.0,
1638
+ "step": 905
1639
+ },
1640
+ {
1641
+ "epoch": 1.456,
1642
+ "grad_norm": 1.117723822593689,
1643
+ "learning_rate": 1.0460603900408523e-05,
1644
+ "loss": 0.8301,
1645
+ "mean_token_accuracy": 0.7960488602519036,
1646
+ "num_tokens": 1086033.0,
1647
+ "step": 910
1648
+ },
1649
+ {
1650
+ "epoch": 1.464,
1651
+ "grad_norm": 1.0806896686553955,
1652
+ "learning_rate": 1.0376901826699349e-05,
1653
+ "loss": 0.7418,
1654
+ "mean_token_accuracy": 0.815102542936802,
1655
+ "num_tokens": 1092009.0,
1656
+ "step": 915
1657
+ },
1658
+ {
1659
+ "epoch": 1.472,
1660
+ "grad_norm": 1.0784786939620972,
1661
+ "learning_rate": 1.0293173300722286e-05,
1662
+ "loss": 0.8307,
1663
+ "mean_token_accuracy": 0.7905350834131241,
1664
+ "num_tokens": 1097979.0,
1665
+ "step": 920
1666
+ },
1667
+ {
1668
+ "epoch": 1.48,
1669
+ "grad_norm": 1.1907199621200562,
1670
+ "learning_rate": 1.0209424198833571e-05,
1671
+ "loss": 0.8569,
1672
+ "mean_token_accuracy": 0.7896250411868095,
1673
+ "num_tokens": 1103820.0,
1674
+ "step": 925
1675
+ },
1676
+ {
1677
+ "epoch": 1.488,
1678
+ "grad_norm": 1.1053500175476074,
1679
+ "learning_rate": 1.0125660398833528e-05,
1680
+ "loss": 0.8264,
1681
+ "mean_token_accuracy": 0.799218937754631,
1682
+ "num_tokens": 1109796.0,
1683
+ "step": 930
1684
+ },
1685
+ {
1686
+ "epoch": 1.496,
1687
+ "grad_norm": 1.2176475524902344,
1688
+ "learning_rate": 1.0041887779554041e-05,
1689
+ "loss": 0.8187,
1690
+ "mean_token_accuracy": 0.7931439965963364,
1691
+ "num_tokens": 1115648.0,
1692
+ "step": 935
1693
+ },
1694
+ {
1695
+ "epoch": 1.504,
1696
+ "grad_norm": 1.3423031568527222,
1697
+ "learning_rate": 9.958112220445964e-06,
1698
+ "loss": 0.8589,
1699
+ "mean_token_accuracy": 0.7898055583238601,
1700
+ "num_tokens": 1121574.0,
1701
+ "step": 940
1702
+ },
1703
+ {
1704
+ "epoch": 1.512,
1705
+ "grad_norm": 1.2004626989364624,
1706
+ "learning_rate": 9.874339601166474e-06,
1707
+ "loss": 0.7966,
1708
+ "mean_token_accuracy": 0.801232923567295,
1709
+ "num_tokens": 1127148.0,
1710
+ "step": 945
1711
+ },
1712
+ {
1713
+ "epoch": 1.52,
1714
+ "grad_norm": 1.4939032793045044,
1715
+ "learning_rate": 9.790575801166432e-06,
1716
+ "loss": 0.8169,
1717
+ "mean_token_accuracy": 0.7929247871041298,
1718
+ "num_tokens": 1133222.0,
1719
+ "step": 950
1720
+ },
1721
+ {
1722
+ "epoch": 1.528,
1723
+ "grad_norm": 1.6087186336517334,
1724
+ "learning_rate": 9.706826699277719e-06,
1725
+ "loss": 0.7953,
1726
+ "mean_token_accuracy": 0.802008081972599,
1727
+ "num_tokens": 1139797.0,
1728
+ "step": 955
1729
+ },
1730
+ {
1731
+ "epoch": 1.536,
1732
+ "grad_norm": 1.4525336027145386,
1733
+ "learning_rate": 9.623098173300655e-06,
1734
+ "loss": 0.7951,
1735
+ "mean_token_accuracy": 0.8010875299572945,
1736
+ "num_tokens": 1145770.0,
1737
+ "step": 960
1738
+ },
1739
+ {
1740
+ "epoch": 1.544,
1741
+ "grad_norm": 1.455733060836792,
1742
+ "learning_rate": 9.539396099591477e-06,
1743
+ "loss": 0.7943,
1744
+ "mean_token_accuracy": 0.8002624407410621,
1745
+ "num_tokens": 1151537.0,
1746
+ "step": 965
1747
+ },
1748
+ {
1749
+ "epoch": 1.552,
1750
+ "grad_norm": 1.620224952697754,
1751
+ "learning_rate": 9.45572635264991e-06,
1752
+ "loss": 0.8281,
1753
+ "mean_token_accuracy": 0.7949922427535057,
1754
+ "num_tokens": 1157720.0,
1755
+ "step": 970
1756
+ },
1757
+ {
1758
+ "epoch": 1.56,
1759
+ "grad_norm": 1.1966170072555542,
1760
+ "learning_rate": 9.372094804706867e-06,
1761
+ "loss": 0.8129,
1762
+ "mean_token_accuracy": 0.799052669107914,
1763
+ "num_tokens": 1163859.0,
1764
+ "step": 975
1765
+ },
1766
+ {
1767
+ "epoch": 1.568,
1768
+ "grad_norm": 1.2910419702529907,
1769
+ "learning_rate": 9.288507325312334e-06,
1770
+ "loss": 0.7883,
1771
+ "mean_token_accuracy": 0.7996888637542725,
1772
+ "num_tokens": 1169977.0,
1773
+ "step": 980
1774
+ },
1775
+ {
1776
+ "epoch": 1.576,
1777
+ "grad_norm": 1.3536475896835327,
1778
+ "learning_rate": 9.204969780923404e-06,
1779
+ "loss": 0.7976,
1780
+ "mean_token_accuracy": 0.8028262749314308,
1781
+ "num_tokens": 1175928.0,
1782
+ "step": 985
1783
+ },
1784
+ {
1785
+ "epoch": 1.584,
1786
+ "grad_norm": 1.3092838525772095,
1787
+ "learning_rate": 9.121488034492569e-06,
1788
+ "loss": 0.8005,
1789
+ "mean_token_accuracy": 0.8030487224459648,
1790
+ "num_tokens": 1182385.0,
1791
+ "step": 990
1792
+ },
1793
+ {
1794
+ "epoch": 1.592,
1795
+ "grad_norm": 1.634954810142517,
1796
+ "learning_rate": 9.038067945056229e-06,
1797
+ "loss": 0.834,
1798
+ "mean_token_accuracy": 0.793434987962246,
1799
+ "num_tokens": 1188030.0,
1800
+ "step": 995
1801
+ },
1802
+ {
1803
+ "epoch": 1.6,
1804
+ "grad_norm": 1.4819631576538086,
1805
+ "learning_rate": 8.954715367323468e-06,
1806
+ "loss": 0.8065,
1807
+ "mean_token_accuracy": 0.7991513565182686,
1808
+ "num_tokens": 1194062.0,
1809
+ "step": 1000
1810
+ },
1811
+ {
1812
+ "epoch": 1.608,
1813
+ "grad_norm": 1.420319676399231,
1814
+ "learning_rate": 8.871436151265183e-06,
1815
+ "loss": 0.8174,
1816
+ "mean_token_accuracy": 0.7973307102918625,
1817
+ "num_tokens": 1199833.0,
1818
+ "step": 1005
1819
+ },
1820
+ {
1821
+ "epoch": 1.616,
1822
+ "grad_norm": 1.1758671998977661,
1823
+ "learning_rate": 8.788236141703498e-06,
1824
+ "loss": 0.7976,
1825
+ "mean_token_accuracy": 0.7996651351451873,
1826
+ "num_tokens": 1206242.0,
1827
+ "step": 1010
1828
+ },
1829
+ {
1830
+ "epoch": 1.624,
1831
+ "grad_norm": 1.2172248363494873,
1832
+ "learning_rate": 8.705121177901532e-06,
1833
+ "loss": 0.8028,
1834
+ "mean_token_accuracy": 0.8009381666779518,
1835
+ "num_tokens": 1212296.0,
1836
+ "step": 1015
1837
+ },
1838
+ {
1839
+ "epoch": 1.6320000000000001,
1840
+ "grad_norm": 1.362866759300232,
1841
+ "learning_rate": 8.62209709315362e-06,
1842
+ "loss": 0.8503,
1843
+ "mean_token_accuracy": 0.7932586327195168,
1844
+ "num_tokens": 1218390.0,
1845
+ "step": 1020
1846
+ },
1847
+ {
1848
+ "epoch": 1.6400000000000001,
1849
+ "grad_norm": 1.5994294881820679,
1850
+ "learning_rate": 8.539169714375885e-06,
1851
+ "loss": 0.791,
1852
+ "mean_token_accuracy": 0.7985045969486236,
1853
+ "num_tokens": 1224581.0,
1854
+ "step": 1025
1855
+ },
1856
+ {
1857
+ "epoch": 1.6480000000000001,
1858
+ "grad_norm": 1.3143961429595947,
1859
+ "learning_rate": 8.45634486169729e-06,
1860
+ "loss": 0.7825,
1861
+ "mean_token_accuracy": 0.7993462473154068,
1862
+ "num_tokens": 1230921.0,
1863
+ "step": 1030
1864
+ },
1865
+ {
1866
+ "epoch": 1.6560000000000001,
1867
+ "grad_norm": 1.2672239542007446,
1868
+ "learning_rate": 8.373628348051165e-06,
1869
+ "loss": 0.816,
1870
+ "mean_token_accuracy": 0.8000213339924812,
1871
+ "num_tokens": 1236774.0,
1872
+ "step": 1035
1873
+ },
1874
+ {
1875
+ "epoch": 1.6640000000000001,
1876
+ "grad_norm": 1.2891995906829834,
1877
+ "learning_rate": 8.291025978767236e-06,
1878
+ "loss": 0.8203,
1879
+ "mean_token_accuracy": 0.7939956024289131,
1880
+ "num_tokens": 1242458.0,
1881
+ "step": 1040
1882
+ },
1883
+ {
1884
+ "epoch": 1.6720000000000002,
1885
+ "grad_norm": 1.859931230545044,
1886
+ "learning_rate": 8.208543551164178e-06,
1887
+ "loss": 0.843,
1888
+ "mean_token_accuracy": 0.7980906665325165,
1889
+ "num_tokens": 1247832.0,
1890
+ "step": 1045
1891
+ },
1892
+ {
1893
+ "epoch": 1.6800000000000002,
1894
+ "grad_norm": 1.249145269393921,
1895
+ "learning_rate": 8.126186854142752e-06,
1896
+ "loss": 0.7933,
1897
+ "mean_token_accuracy": 0.8037418410181999,
1898
+ "num_tokens": 1253519.0,
1899
+ "step": 1050
1900
+ },
1901
+ {
1902
+ "epoch": 1.688,
1903
+ "grad_norm": 1.2496392726898193,
1904
+ "learning_rate": 8.04396166777952e-06,
1905
+ "loss": 0.8268,
1906
+ "mean_token_accuracy": 0.7900358602404595,
1907
+ "num_tokens": 1259478.0,
1908
+ "step": 1055
1909
+ },
1910
+ {
1911
+ "epoch": 1.696,
1912
+ "grad_norm": 1.3781346082687378,
1913
+ "learning_rate": 7.961873762921153e-06,
1914
+ "loss": 0.8215,
1915
+ "mean_token_accuracy": 0.7916676893830299,
1916
+ "num_tokens": 1265272.0,
1917
+ "step": 1060
1918
+ },
1919
+ {
1920
+ "epoch": 1.704,
1921
+ "grad_norm": 1.5332448482513428,
1922
+ "learning_rate": 7.879928900779457e-06,
1923
+ "loss": 0.795,
1924
+ "mean_token_accuracy": 0.7954872667789459,
1925
+ "num_tokens": 1271220.0,
1926
+ "step": 1065
1927
+ },
1928
+ {
1929
+ "epoch": 1.712,
1930
+ "grad_norm": 1.3455692529678345,
1931
+ "learning_rate": 7.798132832526986e-06,
1932
+ "loss": 0.81,
1933
+ "mean_token_accuracy": 0.7959314361214638,
1934
+ "num_tokens": 1277316.0,
1935
+ "step": 1070
1936
+ },
1937
+ {
1938
+ "epoch": 1.72,
1939
+ "grad_norm": 1.2499018907546997,
1940
+ "learning_rate": 7.716491298893443e-06,
1941
+ "loss": 0.7599,
1942
+ "mean_token_accuracy": 0.8050675049424172,
1943
+ "num_tokens": 1283633.0,
1944
+ "step": 1075
1945
+ },
1946
+ {
1947
+ "epoch": 1.728,
1948
+ "grad_norm": 1.2576276063919067,
1949
+ "learning_rate": 7.635010029762755e-06,
1950
+ "loss": 0.8125,
1951
+ "mean_token_accuracy": 0.8032173991203309,
1952
+ "num_tokens": 1289403.0,
1953
+ "step": 1080
1954
+ },
1955
+ {
1956
+ "epoch": 1.736,
1957
+ "grad_norm": 1.2021043300628662,
1958
+ "learning_rate": 7.553694743770928e-06,
1959
+ "loss": 0.8135,
1960
+ "mean_token_accuracy": 0.7988339021801949,
1961
+ "num_tokens": 1295570.0,
1962
+ "step": 1085
1963
+ },
1964
+ {
1965
+ "epoch": 1.744,
1966
+ "grad_norm": 1.3602943420410156,
1967
+ "learning_rate": 7.472551147904708e-06,
1968
+ "loss": 0.7765,
1969
+ "mean_token_accuracy": 0.808410918712616,
1970
+ "num_tokens": 1301628.0,
1971
+ "step": 1090
1972
+ },
1973
+ {
1974
+ "epoch": 1.752,
1975
+ "grad_norm": 1.2860238552093506,
1976
+ "learning_rate": 7.391584937101034e-06,
1977
+ "loss": 0.8073,
1978
+ "mean_token_accuracy": 0.7989855810999871,
1979
+ "num_tokens": 1307458.0,
1980
+ "step": 1095
1981
+ },
1982
+ {
1983
+ "epoch": 1.76,
1984
+ "grad_norm": 1.5155887603759766,
1985
+ "learning_rate": 7.310801793847344e-06,
1986
+ "loss": 0.7845,
1987
+ "mean_token_accuracy": 0.8080469697713852,
1988
+ "num_tokens": 1313341.0,
1989
+ "step": 1100
1990
+ },
1991
+ {
1992
+ "epoch": 1.768,
1993
+ "grad_norm": 1.6270219087600708,
1994
+ "learning_rate": 7.2302073877827775e-06,
1995
+ "loss": 0.7778,
1996
+ "mean_token_accuracy": 0.8048338174819947,
1997
+ "num_tokens": 1319272.0,
1998
+ "step": 1105
1999
+ },
2000
+ {
2001
+ "epoch": 1.776,
2002
+ "grad_norm": 1.132137417793274,
2003
+ "learning_rate": 7.149807375300239e-06,
2004
+ "loss": 0.7922,
2005
+ "mean_token_accuracy": 0.8009102180600166,
2006
+ "num_tokens": 1325093.0,
2007
+ "step": 1110
2008
+ },
2009
+ {
2010
+ "epoch": 1.784,
2011
+ "grad_norm": 1.3153914213180542,
2012
+ "learning_rate": 7.069607399149427e-06,
2013
+ "loss": 0.7907,
2014
+ "mean_token_accuracy": 0.8020437583327293,
2015
+ "num_tokens": 1331145.0,
2016
+ "step": 1115
2017
+ },
2018
+ {
2019
+ "epoch": 1.792,
2020
+ "grad_norm": 1.4287455081939697,
2021
+ "learning_rate": 6.9896130880407965e-06,
2022
+ "loss": 0.7904,
2023
+ "mean_token_accuracy": 0.8002470403909683,
2024
+ "num_tokens": 1336725.0,
2025
+ "step": 1120
2026
+ },
2027
+ {
2028
+ "epoch": 1.8,
2029
+ "grad_norm": 1.140147089958191,
2030
+ "learning_rate": 6.909830056250527e-06,
2031
+ "loss": 0.7818,
2032
+ "mean_token_accuracy": 0.7999431058764458,
2033
+ "num_tokens": 1342919.0,
2034
+ "step": 1125
2035
+ },
2036
+ {
2037
+ "epoch": 1.808,
2038
+ "grad_norm": 1.2778555154800415,
2039
+ "learning_rate": 6.830263903226483e-06,
2040
+ "loss": 0.7946,
2041
+ "mean_token_accuracy": 0.8025822728872299,
2042
+ "num_tokens": 1349048.0,
2043
+ "step": 1130
2044
+ },
2045
+ {
2046
+ "epoch": 1.8159999999999998,
2047
+ "grad_norm": 1.253410816192627,
2048
+ "learning_rate": 6.750920213195238e-06,
2049
+ "loss": 0.8111,
2050
+ "mean_token_accuracy": 0.7975707486271858,
2051
+ "num_tokens": 1354965.0,
2052
+ "step": 1135
2053
+ },
2054
+ {
2055
+ "epoch": 1.8239999999999998,
2056
+ "grad_norm": 1.3343291282653809,
2057
+ "learning_rate": 6.671804554770135e-06,
2058
+ "loss": 0.7876,
2059
+ "mean_token_accuracy": 0.8030346512794495,
2060
+ "num_tokens": 1361210.0,
2061
+ "step": 1140
2062
+ },
2063
+ {
2064
+ "epoch": 1.8319999999999999,
2065
+ "grad_norm": 1.3073536157608032,
2066
+ "learning_rate": 6.5929224805604845e-06,
2067
+ "loss": 0.8279,
2068
+ "mean_token_accuracy": 0.7967935264110565,
2069
+ "num_tokens": 1366856.0,
2070
+ "step": 1145
2071
+ },
2072
+ {
2073
+ "epoch": 1.8399999999999999,
2074
+ "grad_norm": 1.2902213335037231,
2075
+ "learning_rate": 6.5142795267818505e-06,
2076
+ "loss": 0.8045,
2077
+ "mean_token_accuracy": 0.7988436847925187,
2078
+ "num_tokens": 1372706.0,
2079
+ "step": 1150
2080
+ },
2081
+ {
2082
+ "epoch": 1.8479999999999999,
2083
+ "grad_norm": 1.038400650024414,
2084
+ "learning_rate": 6.435881212867494e-06,
2085
+ "loss": 0.7736,
2086
+ "mean_token_accuracy": 0.8005026668310166,
2087
+ "num_tokens": 1379305.0,
2088
+ "step": 1155
2089
+ },
2090
+ {
2091
+ "epoch": 1.8559999999999999,
2092
+ "grad_norm": 1.340643286705017,
2093
+ "learning_rate": 6.357733041081018e-06,
2094
+ "loss": 0.7819,
2095
+ "mean_token_accuracy": 0.8061650961637497,
2096
+ "num_tokens": 1385146.0,
2097
+ "step": 1160
2098
+ },
2099
+ {
2100
+ "epoch": 1.8639999999999999,
2101
+ "grad_norm": 1.423438549041748,
2102
+ "learning_rate": 6.27984049613019e-06,
2103
+ "loss": 0.8054,
2104
+ "mean_token_accuracy": 0.8006222054362298,
2105
+ "num_tokens": 1390600.0,
2106
+ "step": 1165
2107
+ },
2108
+ {
2109
+ "epoch": 1.8719999999999999,
2110
+ "grad_norm": 1.2361985445022583,
2111
+ "learning_rate": 6.202209044781991e-06,
2112
+ "loss": 0.8295,
2113
+ "mean_token_accuracy": 0.7897411197423935,
2114
+ "num_tokens": 1396485.0,
2115
+ "step": 1170
2116
+ },
2117
+ {
2118
+ "epoch": 1.88,
2119
+ "grad_norm": 1.0449228286743164,
2120
+ "learning_rate": 6.124844135478971e-06,
2121
+ "loss": 0.8185,
2122
+ "mean_token_accuracy": 0.8001339569687843,
2123
+ "num_tokens": 1402781.0,
2124
+ "step": 1175
2125
+ },
2126
+ {
2127
+ "epoch": 1.888,
2128
+ "grad_norm": 1.530038833618164,
2129
+ "learning_rate": 6.047751197956838e-06,
2130
+ "loss": 0.805,
2131
+ "mean_token_accuracy": 0.8008620426058769,
2132
+ "num_tokens": 1408853.0,
2133
+ "step": 1180
2134
+ },
2135
+ {
2136
+ "epoch": 1.896,
2137
+ "grad_norm": 1.1717791557312012,
2138
+ "learning_rate": 5.970935642863375e-06,
2139
+ "loss": 0.7928,
2140
+ "mean_token_accuracy": 0.8000158056616783,
2141
+ "num_tokens": 1414603.0,
2142
+ "step": 1185
2143
+ },
2144
+ {
2145
+ "epoch": 1.904,
2146
+ "grad_norm": 1.2921278476715088,
2147
+ "learning_rate": 5.894402861378721e-06,
2148
+ "loss": 0.8487,
2149
+ "mean_token_accuracy": 0.7888679310679436,
2150
+ "num_tokens": 1421159.0,
2151
+ "step": 1190
2152
+ },
2153
+ {
2154
+ "epoch": 1.912,
2155
+ "grad_norm": 1.2942895889282227,
2156
+ "learning_rate": 5.818158224836987e-06,
2157
+ "loss": 0.8275,
2158
+ "mean_token_accuracy": 0.7944061666727066,
2159
+ "num_tokens": 1427020.0,
2160
+ "step": 1195
2161
+ },
2162
+ {
2163
+ "epoch": 1.92,
2164
+ "grad_norm": 1.3930217027664185,
2165
+ "learning_rate": 5.742207084349274e-06,
2166
+ "loss": 0.7942,
2167
+ "mean_token_accuracy": 0.8061083048582077,
2168
+ "num_tokens": 1433286.0,
2169
+ "step": 1200
2170
+ },
2171
+ {
2172
+ "epoch": 1.928,
2173
+ "grad_norm": 1.2439182996749878,
2174
+ "learning_rate": 5.666554770428129e-06,
2175
+ "loss": 0.7791,
2176
+ "mean_token_accuracy": 0.806618258357048,
2177
+ "num_tokens": 1439282.0,
2178
+ "step": 1205
2179
+ },
2180
+ {
2181
+ "epoch": 1.936,
2182
+ "grad_norm": 1.2223875522613525,
2183
+ "learning_rate": 5.591206592613416e-06,
2184
+ "loss": 0.8201,
2185
+ "mean_token_accuracy": 0.7910259455442429,
2186
+ "num_tokens": 1445214.0,
2187
+ "step": 1210
2188
+ },
2189
+ {
2190
+ "epoch": 1.944,
2191
+ "grad_norm": 1.293550729751587,
2192
+ "learning_rate": 5.516167839099679e-06,
2193
+ "loss": 0.8075,
2194
+ "mean_token_accuracy": 0.8063176274299622,
2195
+ "num_tokens": 1451100.0,
2196
+ "step": 1215
2197
+ },
2198
+ {
2199
+ "epoch": 1.952,
2200
+ "grad_norm": 1.467372179031372,
2201
+ "learning_rate": 5.441443776365003e-06,
2202
+ "loss": 0.8237,
2203
+ "mean_token_accuracy": 0.7898676633834839,
2204
+ "num_tokens": 1456778.0,
2205
+ "step": 1220
2206
+ },
2207
+ {
2208
+ "epoch": 1.96,
2209
+ "grad_norm": 1.4951013326644897,
2210
+ "learning_rate": 5.367039648801386e-06,
2211
+ "loss": 0.7994,
2212
+ "mean_token_accuracy": 0.7988926216959953,
2213
+ "num_tokens": 1462707.0,
2214
+ "step": 1225
2215
+ },
2216
+ {
2217
+ "epoch": 1.968,
2218
+ "grad_norm": 1.392849087715149,
2219
+ "learning_rate": 5.292960678346674e-06,
2220
+ "loss": 0.7739,
2221
+ "mean_token_accuracy": 0.8094693034887314,
2222
+ "num_tokens": 1468764.0,
2223
+ "step": 1230
2224
+ },
2225
+ {
2226
+ "epoch": 1.976,
2227
+ "grad_norm": 1.2680391073226929,
2228
+ "learning_rate": 5.219212064118079e-06,
2229
+ "loss": 0.8329,
2230
+ "mean_token_accuracy": 0.7990137442946434,
2231
+ "num_tokens": 1474638.0,
2232
+ "step": 1235
2233
+ },
2234
+ {
2235
+ "epoch": 1.984,
2236
+ "grad_norm": 1.7272734642028809,
2237
+ "learning_rate": 5.145798982047261e-06,
2238
+ "loss": 0.8139,
2239
+ "mean_token_accuracy": 0.7936960220336914,
2240
+ "num_tokens": 1480240.0,
2241
+ "step": 1240
2242
+ },
2243
+ {
2244
+ "epoch": 1.992,
2245
+ "grad_norm": 1.3061951398849487,
2246
+ "learning_rate": 5.072726584517086e-06,
2247
+ "loss": 0.811,
2248
+ "mean_token_accuracy": 0.7989570170640945,
2249
+ "num_tokens": 1485959.0,
2250
+ "step": 1245
2251
+ },
2252
+ {
2253
+ "epoch": 2.0,
2254
+ "grad_norm": 1.4654656648635864,
2255
+ "learning_rate": 5.000000000000003e-06,
2256
+ "loss": 0.8079,
2257
+ "mean_token_accuracy": 0.7938532695174217,
2258
+ "num_tokens": 1491748.0,
2259
+ "step": 1250
2260
+ }
2261
+ ],
2262
+ "logging_steps": 5,
2263
+ "max_steps": 1875,
2264
+ "num_input_tokens_seen": 0,
2265
+ "num_train_epochs": 3,
2266
+ "save_steps": 500,
2267
+ "stateful_callbacks": {
2268
+ "TrainerControl": {
2269
+ "args": {
2270
+ "should_epoch_stop": false,
2271
+ "should_evaluate": false,
2272
+ "should_log": false,
2273
+ "should_save": true,
2274
+ "should_training_stop": false
2275
+ },
2276
+ "attributes": {}
2277
+ }
2278
+ },
2279
+ "total_flos": 1.1030265057337344e+16,
2280
+ "train_batch_size": 2,
2281
+ "trial_name": null,
2282
+ "trial_params": null
2283
+ }
LoRA_model/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:509e4017301ac12fd4abee99c872e42138d7dac0ca0270f72c57dceb8f4f67c5
3
+ size 5624
app.py CHANGED
@@ -1,13 +1,17 @@
1
  import gradio as gr
2
  import torch
 
3
  from transformers import AutoTokenizer, AutoModelForCausalLM
4
 
5
  # Select device: GPU if available, else CPU
6
  device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
7
 
8
  # Load tokenizer and model from local directory
9
- tokenizer = AutoTokenizer.from_pretrained("rat45/sql-lora-fp32")
10
- model = AutoModelForCausalLM.from_pretrained("rat45/sql-lora-fp32").to(device)
 
 
 
11
 
12
 
13
  # Define generation function
@@ -20,20 +24,21 @@ def generate_sql(prompt):
20
  temperature=0.7,
21
  top_p=0.95,
22
  eos_token_id=tokenizer.eos_token_id,
23
- early_stopping=True
 
24
  )
25
  full_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
26
- return full_output[len(prompt):].strip() # remove prompt from beginning
27
 
28
 
29
  # Gradio UI
30
  interface = gr.Interface(
31
  fn=generate_sql,
32
- inputs=gr.Textbox(lines=3, placeholder="Enter instruction, e.g. 'Show all users with age > 30'"),
33
  outputs="text",
34
- title="🧠 SQL Generator",
35
  description="Type a natural language prompt and get a SQL query generated by the fine-tuned TinyLlama model.",
36
  theme="default"
37
  )
38
 
39
- interface.launch(share=True)
 
1
  import gradio as gr
2
  import torch
3
+ from peft import PeftModel
4
  from transformers import AutoTokenizer, AutoModelForCausalLM
5
 
6
  # Select device: GPU if available, else CPU
7
  device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
8
 
9
  # Load tokenizer and model from local directory
10
+ tokenizer = AutoTokenizer.from_pretrained("TinyLlama/TinyLlama-1.1B-Chat-v1.0")
11
+ model = AutoModelForCausalLM.from_pretrained("TinyLlama/TinyLlama-1.1B-Chat-v1.0").to(device)
12
+
13
+ # Load LoRA adapter
14
+ model = PeftModel.from_pretrained(model, "LoRA_model")
15
 
16
 
17
  # Define generation function
 
24
  temperature=0.7,
25
  top_p=0.95,
26
  eos_token_id=tokenizer.eos_token_id,
27
+ early_stopping=True,
28
+ num_beams=5,
29
  )
30
  full_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
31
+ return full_output[len(prompt):].strip().split(';', 1)[0] + ';' # remove prompt from beginning and only the first SQL statement
32
 
33
 
34
  # Gradio UI
35
  interface = gr.Interface(
36
  fn=generate_sql,
37
+ inputs=gr.Textbox(lines=3, placeholder="Enter instruction, e.g. 'Show all users with age > 30' or 'Show all users where gender is female.'"),
38
  outputs="text",
39
+ title="SQL Generator",
40
  description="Type a natural language prompt and get a SQL query generated by the fine-tuned TinyLlama model.",
41
  theme="default"
42
  )
43
 
44
+ interface.launch(share=True)
requirements.txt CHANGED
@@ -1,3 +1,4 @@
1
  transformers
2
  torch
3
  gradio
 
 
1
  transformers
2
  torch
3
  gradio
4
+ peft