antonpolishko commited on
Commit
61e2fe7
·
verified ·
1 Parent(s): 66e34d6

Model save

Browse files
README.md ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: mistralai/Mistral-7B-v0.3
3
+ library_name: transformers
4
+ model_name: mistral-7b-overfit-ua
5
+ tags:
6
+ - generated_from_trainer
7
+ - trl
8
+ - sft
9
+ licence: license
10
+ ---
11
+
12
+ # Model Card for mistral-7b-overfit-ua
13
+
14
+ This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3).
15
+ It has been trained using [TRL](https://github.com/huggingface/trl).
16
+
17
+ ## Quick start
18
+
19
+ ```python
20
+ from transformers import pipeline
21
+
22
+ question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
23
+ generator = pipeline("text-generation", model="antonpolishko/mistral-7b-overfit-ua", device="cuda")
24
+ output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
25
+ print(output["generated_text"])
26
+ ```
27
+
28
+ ## Training procedure
29
+
30
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/polyagent/huggingface/runs/gc8y9lrf)
31
+
32
+ This model was trained with SFT.
33
+
34
+ ### Framework versions
35
+
36
+ - TRL: 0.12.1
37
+ - Transformers: 4.46.3
38
+ - Pytorch: 2.6.0.dev20241113+rocm6.2
39
+ - Datasets: 3.1.0
40
+ - Tokenizers: 0.20.3
41
+
42
+ ## Citations
43
+
44
+
45
+
46
+ Cite TRL as:
47
+
48
+ ```bibtex
49
+ @misc{vonwerra2022trl,
50
+ title = {{TRL: Transformer Reinforcement Learning}},
51
+ author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
52
+ year = 2020,
53
+ journal = {GitHub repository},
54
+ publisher = {GitHub},
55
+ howpublished = {\url{https://github.com/huggingface/trl}}
56
+ }
57
+ ```
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 8.072899219804914e+18,
4
+ "train_loss": 1.2002355462548142,
5
+ "train_runtime": 6547.9575,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 14.077,
8
+ "train_steps_per_second": 0.22
9
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.46.3"
6
+ }
runs/Dec26_18-40-03_mia1-gpu-109/events.out.tfevents.1735238501.mia1-gpu-109.3917105.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0732e07e1aa74a8fa50b6740d74237e3511eee8e4f73577bc2415c814d394d00
3
- size 67673
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aab9fde4d4ea187122d7c1aca7f2d2b634fd8ebf81fa2b835778eba263bf2130
3
+ size 68027
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 8.072899219804914e+18,
4
+ "train_loss": 1.2002355462548142,
5
+ "train_runtime": 6547.9575,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 14.077,
8
+ "train_steps_per_second": 0.22
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,2097 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 300,
6
+ "global_step": 1443,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.002079002079002079,
13
+ "grad_norm": 21.0,
14
+ "learning_rate": 1e-06,
15
+ "loss": 1.4443,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.010395010395010396,
20
+ "grad_norm": 7.46875,
21
+ "learning_rate": 1.9999786113807535e-06,
22
+ "loss": 1.4249,
23
+ "step": 5
24
+ },
25
+ {
26
+ "epoch": 0.02079002079002079,
27
+ "grad_norm": 5.1875,
28
+ "learning_rate": 1.999847906465395e-06,
29
+ "loss": 1.3779,
30
+ "step": 10
31
+ },
32
+ {
33
+ "epoch": 0.031185031185031187,
34
+ "grad_norm": 4.125,
35
+ "learning_rate": 1.999598394713099e-06,
36
+ "loss": 1.3446,
37
+ "step": 15
38
+ },
39
+ {
40
+ "epoch": 0.04158004158004158,
41
+ "grad_norm": 3.859375,
42
+ "learning_rate": 1.9992301057720783e-06,
43
+ "loss": 1.3445,
44
+ "step": 20
45
+ },
46
+ {
47
+ "epoch": 0.05197505197505198,
48
+ "grad_norm": 3.75,
49
+ "learning_rate": 1.9987430834042318e-06,
50
+ "loss": 1.3221,
51
+ "step": 25
52
+ },
53
+ {
54
+ "epoch": 0.062370062370062374,
55
+ "grad_norm": 3.703125,
56
+ "learning_rate": 1.9981373854799487e-06,
57
+ "loss": 1.3312,
58
+ "step": 30
59
+ },
60
+ {
61
+ "epoch": 0.07276507276507277,
62
+ "grad_norm": 3.515625,
63
+ "learning_rate": 1.99741308397123e-06,
64
+ "loss": 1.3108,
65
+ "step": 35
66
+ },
67
+ {
68
+ "epoch": 0.08316008316008316,
69
+ "grad_norm": 3.703125,
70
+ "learning_rate": 1.9965702649431362e-06,
71
+ "loss": 1.3148,
72
+ "step": 40
73
+ },
74
+ {
75
+ "epoch": 0.09355509355509356,
76
+ "grad_norm": 3.5625,
77
+ "learning_rate": 1.9956090285435628e-06,
78
+ "loss": 1.3075,
79
+ "step": 45
80
+ },
81
+ {
82
+ "epoch": 0.10395010395010396,
83
+ "grad_norm": 3.703125,
84
+ "learning_rate": 1.9945294889913377e-06,
85
+ "loss": 1.3029,
86
+ "step": 50
87
+ },
88
+ {
89
+ "epoch": 0.11434511434511435,
90
+ "grad_norm": 3.671875,
91
+ "learning_rate": 1.9933317745626504e-06,
92
+ "loss": 1.3022,
93
+ "step": 55
94
+ },
95
+ {
96
+ "epoch": 0.12474012474012475,
97
+ "grad_norm": 3.671875,
98
+ "learning_rate": 1.9920160275758096e-06,
99
+ "loss": 1.2885,
100
+ "step": 60
101
+ },
102
+ {
103
+ "epoch": 0.13513513513513514,
104
+ "grad_norm": 3.515625,
105
+ "learning_rate": 1.9905824043743316e-06,
106
+ "loss": 1.2944,
107
+ "step": 65
108
+ },
109
+ {
110
+ "epoch": 0.14553014553014554,
111
+ "grad_norm": 3.453125,
112
+ "learning_rate": 1.9890310753083633e-06,
113
+ "loss": 1.2993,
114
+ "step": 70
115
+ },
116
+ {
117
+ "epoch": 0.15592515592515593,
118
+ "grad_norm": 3.484375,
119
+ "learning_rate": 1.987362224714441e-06,
120
+ "loss": 1.2885,
121
+ "step": 75
122
+ },
123
+ {
124
+ "epoch": 0.16632016632016633,
125
+ "grad_norm": 3.5,
126
+ "learning_rate": 1.985576050893585e-06,
127
+ "loss": 1.2862,
128
+ "step": 80
129
+ },
130
+ {
131
+ "epoch": 0.17671517671517672,
132
+ "grad_norm": 3.578125,
133
+ "learning_rate": 1.983672766087738e-06,
134
+ "loss": 1.2854,
135
+ "step": 85
136
+ },
137
+ {
138
+ "epoch": 0.18711018711018712,
139
+ "grad_norm": 3.546875,
140
+ "learning_rate": 1.9816525964545446e-06,
141
+ "loss": 1.2765,
142
+ "step": 90
143
+ },
144
+ {
145
+ "epoch": 0.19750519750519752,
146
+ "grad_norm": 3.640625,
147
+ "learning_rate": 1.979515782040478e-06,
148
+ "loss": 1.2874,
149
+ "step": 95
150
+ },
151
+ {
152
+ "epoch": 0.2079002079002079,
153
+ "grad_norm": 3.546875,
154
+ "learning_rate": 1.977262576752319e-06,
155
+ "loss": 1.2808,
156
+ "step": 100
157
+ },
158
+ {
159
+ "epoch": 0.2182952182952183,
160
+ "grad_norm": 3.59375,
161
+ "learning_rate": 1.97489324832698e-06,
162
+ "loss": 1.2768,
163
+ "step": 105
164
+ },
165
+ {
166
+ "epoch": 0.2286902286902287,
167
+ "grad_norm": 3.59375,
168
+ "learning_rate": 1.972408078299698e-06,
169
+ "loss": 1.2757,
170
+ "step": 110
171
+ },
172
+ {
173
+ "epoch": 0.2390852390852391,
174
+ "grad_norm": 3.53125,
175
+ "learning_rate": 1.969807361970575e-06,
176
+ "loss": 1.2897,
177
+ "step": 115
178
+ },
179
+ {
180
+ "epoch": 0.2494802494802495,
181
+ "grad_norm": 3.5,
182
+ "learning_rate": 1.9670914083694948e-06,
183
+ "loss": 1.2796,
184
+ "step": 120
185
+ },
186
+ {
187
+ "epoch": 0.2598752598752599,
188
+ "grad_norm": 3.515625,
189
+ "learning_rate": 1.9642605402193976e-06,
190
+ "loss": 1.2684,
191
+ "step": 125
192
+ },
193
+ {
194
+ "epoch": 0.2702702702702703,
195
+ "grad_norm": 3.53125,
196
+ "learning_rate": 1.9613150938979343e-06,
197
+ "loss": 1.2694,
198
+ "step": 130
199
+ },
200
+ {
201
+ "epoch": 0.2806652806652807,
202
+ "grad_norm": 3.546875,
203
+ "learning_rate": 1.958255419397498e-06,
204
+ "loss": 1.263,
205
+ "step": 135
206
+ },
207
+ {
208
+ "epoch": 0.2910602910602911,
209
+ "grad_norm": 3.578125,
210
+ "learning_rate": 1.9550818802836337e-06,
211
+ "loss": 1.271,
212
+ "step": 140
213
+ },
214
+ {
215
+ "epoch": 0.30145530145530147,
216
+ "grad_norm": 3.640625,
217
+ "learning_rate": 1.9517948536518397e-06,
218
+ "loss": 1.2638,
219
+ "step": 145
220
+ },
221
+ {
222
+ "epoch": 0.31185031185031187,
223
+ "grad_norm": 3.59375,
224
+ "learning_rate": 1.9483947300827577e-06,
225
+ "loss": 1.2633,
226
+ "step": 150
227
+ },
228
+ {
229
+ "epoch": 0.32224532224532226,
230
+ "grad_norm": 3.484375,
231
+ "learning_rate": 1.944881913595762e-06,
232
+ "loss": 1.2648,
233
+ "step": 155
234
+ },
235
+ {
236
+ "epoch": 0.33264033264033266,
237
+ "grad_norm": 3.46875,
238
+ "learning_rate": 1.941256821600955e-06,
239
+ "loss": 1.2588,
240
+ "step": 160
241
+ },
242
+ {
243
+ "epoch": 0.34303534303534305,
244
+ "grad_norm": 3.5,
245
+ "learning_rate": 1.9375198848495637e-06,
246
+ "loss": 1.258,
247
+ "step": 165
248
+ },
249
+ {
250
+ "epoch": 0.35343035343035345,
251
+ "grad_norm": 3.53125,
252
+ "learning_rate": 1.9336715473827594e-06,
253
+ "loss": 1.2588,
254
+ "step": 170
255
+ },
256
+ {
257
+ "epoch": 0.36382536382536385,
258
+ "grad_norm": 3.515625,
259
+ "learning_rate": 1.929712266478893e-06,
260
+ "loss": 1.2538,
261
+ "step": 175
262
+ },
263
+ {
264
+ "epoch": 0.37422037422037424,
265
+ "grad_norm": 3.765625,
266
+ "learning_rate": 1.9256425125991596e-06,
267
+ "loss": 1.2539,
268
+ "step": 180
269
+ },
270
+ {
271
+ "epoch": 0.38461538461538464,
272
+ "grad_norm": 3.546875,
273
+ "learning_rate": 1.9214627693316952e-06,
274
+ "loss": 1.2556,
275
+ "step": 185
276
+ },
277
+ {
278
+ "epoch": 0.39501039501039503,
279
+ "grad_norm": 3.59375,
280
+ "learning_rate": 1.9171735333341148e-06,
281
+ "loss": 1.2524,
282
+ "step": 190
283
+ },
284
+ {
285
+ "epoch": 0.40540540540540543,
286
+ "grad_norm": 3.59375,
287
+ "learning_rate": 1.9127753142744977e-06,
288
+ "loss": 1.2471,
289
+ "step": 195
290
+ },
291
+ {
292
+ "epoch": 0.4158004158004158,
293
+ "grad_norm": 3.515625,
294
+ "learning_rate": 1.9082686347708253e-06,
295
+ "loss": 1.246,
296
+ "step": 200
297
+ },
298
+ {
299
+ "epoch": 0.4261954261954262,
300
+ "grad_norm": 3.421875,
301
+ "learning_rate": 1.9036540303288815e-06,
302
+ "loss": 1.247,
303
+ "step": 205
304
+ },
305
+ {
306
+ "epoch": 0.4365904365904366,
307
+ "grad_norm": 3.546875,
308
+ "learning_rate": 1.898932049278621e-06,
309
+ "loss": 1.2496,
310
+ "step": 210
311
+ },
312
+ {
313
+ "epoch": 0.446985446985447,
314
+ "grad_norm": 3.40625,
315
+ "learning_rate": 1.8941032527090148e-06,
316
+ "loss": 1.2492,
317
+ "step": 215
318
+ },
319
+ {
320
+ "epoch": 0.4573804573804574,
321
+ "grad_norm": 3.515625,
322
+ "learning_rate": 1.8891682144013777e-06,
323
+ "loss": 1.2476,
324
+ "step": 220
325
+ },
326
+ {
327
+ "epoch": 0.4677754677754678,
328
+ "grad_norm": 3.40625,
329
+ "learning_rate": 1.8841275207611895e-06,
330
+ "loss": 1.2426,
331
+ "step": 225
332
+ },
333
+ {
334
+ "epoch": 0.4781704781704782,
335
+ "grad_norm": 3.5,
336
+ "learning_rate": 1.8789817707484162e-06,
337
+ "loss": 1.2388,
338
+ "step": 230
339
+ },
340
+ {
341
+ "epoch": 0.4885654885654886,
342
+ "grad_norm": 3.5625,
343
+ "learning_rate": 1.8737315758063364e-06,
344
+ "loss": 1.2374,
345
+ "step": 235
346
+ },
347
+ {
348
+ "epoch": 0.498960498960499,
349
+ "grad_norm": 3.453125,
350
+ "learning_rate": 1.8683775597888886e-06,
351
+ "loss": 1.2437,
352
+ "step": 240
353
+ },
354
+ {
355
+ "epoch": 0.5093555093555093,
356
+ "grad_norm": 3.5,
357
+ "learning_rate": 1.8629203588865419e-06,
358
+ "loss": 1.2335,
359
+ "step": 245
360
+ },
361
+ {
362
+ "epoch": 0.5197505197505198,
363
+ "grad_norm": 3.53125,
364
+ "learning_rate": 1.8573606215506985e-06,
365
+ "loss": 1.2355,
366
+ "step": 250
367
+ },
368
+ {
369
+ "epoch": 0.5301455301455301,
370
+ "grad_norm": 3.4375,
371
+ "learning_rate": 1.8516990084166443e-06,
372
+ "loss": 1.2384,
373
+ "step": 255
374
+ },
375
+ {
376
+ "epoch": 0.5405405405405406,
377
+ "grad_norm": 3.625,
378
+ "learning_rate": 1.8459361922250469e-06,
379
+ "loss": 1.232,
380
+ "step": 260
381
+ },
382
+ {
383
+ "epoch": 0.5509355509355509,
384
+ "grad_norm": 3.515625,
385
+ "learning_rate": 1.8400728577420187e-06,
386
+ "loss": 1.2356,
387
+ "step": 265
388
+ },
389
+ {
390
+ "epoch": 0.5613305613305614,
391
+ "grad_norm": 3.515625,
392
+ "learning_rate": 1.8341097016777484e-06,
393
+ "loss": 1.2408,
394
+ "step": 270
395
+ },
396
+ {
397
+ "epoch": 0.5717255717255717,
398
+ "grad_norm": 3.46875,
399
+ "learning_rate": 1.8280474326037155e-06,
400
+ "loss": 1.2465,
401
+ "step": 275
402
+ },
403
+ {
404
+ "epoch": 0.5821205821205822,
405
+ "grad_norm": 3.53125,
406
+ "learning_rate": 1.8218867708684937e-06,
407
+ "loss": 1.2331,
408
+ "step": 280
409
+ },
410
+ {
411
+ "epoch": 0.5925155925155925,
412
+ "grad_norm": 3.46875,
413
+ "learning_rate": 1.8156284485121556e-06,
414
+ "loss": 1.2323,
415
+ "step": 285
416
+ },
417
+ {
418
+ "epoch": 0.6029106029106029,
419
+ "grad_norm": 3.4375,
420
+ "learning_rate": 1.8092732091792884e-06,
421
+ "loss": 1.2365,
422
+ "step": 290
423
+ },
424
+ {
425
+ "epoch": 0.6133056133056133,
426
+ "grad_norm": 3.5625,
427
+ "learning_rate": 1.8028218080306302e-06,
428
+ "loss": 1.2394,
429
+ "step": 295
430
+ },
431
+ {
432
+ "epoch": 0.6237006237006237,
433
+ "grad_norm": 3.609375,
434
+ "learning_rate": 1.7962750116533387e-06,
435
+ "loss": 1.2263,
436
+ "step": 300
437
+ },
438
+ {
439
+ "epoch": 0.6237006237006237,
440
+ "eval_loss": 1.2331745624542236,
441
+ "eval_runtime": 10.8501,
442
+ "eval_samples_per_second": 85.713,
443
+ "eval_steps_per_second": 2.765,
444
+ "step": 300
445
+ },
446
+ {
447
+ "epoch": 0.6340956340956341,
448
+ "grad_norm": 3.5625,
449
+ "learning_rate": 1.7896335979699001e-06,
450
+ "loss": 1.2311,
451
+ "step": 305
452
+ },
453
+ {
454
+ "epoch": 0.6444906444906445,
455
+ "grad_norm": 3.46875,
456
+ "learning_rate": 1.782898356145694e-06,
457
+ "loss": 1.2236,
458
+ "step": 310
459
+ },
460
+ {
461
+ "epoch": 0.6548856548856549,
462
+ "grad_norm": 4.25,
463
+ "learning_rate": 1.7760700864952205e-06,
464
+ "loss": 1.2376,
465
+ "step": 315
466
+ },
467
+ {
468
+ "epoch": 0.6652806652806653,
469
+ "grad_norm": 3.484375,
470
+ "learning_rate": 1.7691496003870018e-06,
471
+ "loss": 1.2239,
472
+ "step": 320
473
+ },
474
+ {
475
+ "epoch": 0.6756756756756757,
476
+ "grad_norm": 3.578125,
477
+ "learning_rate": 1.7621377201471735e-06,
478
+ "loss": 1.2311,
479
+ "step": 325
480
+ },
481
+ {
482
+ "epoch": 0.6860706860706861,
483
+ "grad_norm": 3.796875,
484
+ "learning_rate": 1.75503527896177e-06,
485
+ "loss": 1.226,
486
+ "step": 330
487
+ },
488
+ {
489
+ "epoch": 0.6964656964656964,
490
+ "grad_norm": 3.46875,
491
+ "learning_rate": 1.7478431207777215e-06,
492
+ "loss": 1.228,
493
+ "step": 335
494
+ },
495
+ {
496
+ "epoch": 0.7068607068607069,
497
+ "grad_norm": 3.484375,
498
+ "learning_rate": 1.7405621002025735e-06,
499
+ "loss": 1.2216,
500
+ "step": 340
501
+ },
502
+ {
503
+ "epoch": 0.7172557172557172,
504
+ "grad_norm": 3.46875,
505
+ "learning_rate": 1.733193082402936e-06,
506
+ "loss": 1.2223,
507
+ "step": 345
508
+ },
509
+ {
510
+ "epoch": 0.7276507276507277,
511
+ "grad_norm": 3.53125,
512
+ "learning_rate": 1.7257369430016817e-06,
513
+ "loss": 1.2158,
514
+ "step": 350
515
+ },
516
+ {
517
+ "epoch": 0.738045738045738,
518
+ "grad_norm": 3.4375,
519
+ "learning_rate": 1.7181945679739003e-06,
520
+ "loss": 1.2206,
521
+ "step": 355
522
+ },
523
+ {
524
+ "epoch": 0.7484407484407485,
525
+ "grad_norm": 3.59375,
526
+ "learning_rate": 1.7105668535416205e-06,
527
+ "loss": 1.2292,
528
+ "step": 360
529
+ },
530
+ {
531
+ "epoch": 0.7588357588357588,
532
+ "grad_norm": 3.578125,
533
+ "learning_rate": 1.7028547060673197e-06,
534
+ "loss": 1.2272,
535
+ "step": 365
536
+ },
537
+ {
538
+ "epoch": 0.7692307692307693,
539
+ "grad_norm": 3.609375,
540
+ "learning_rate": 1.6950590419462229e-06,
541
+ "loss": 1.2264,
542
+ "step": 370
543
+ },
544
+ {
545
+ "epoch": 0.7796257796257796,
546
+ "grad_norm": 3.578125,
547
+ "learning_rate": 1.687180787497413e-06,
548
+ "loss": 1.2155,
549
+ "step": 375
550
+ },
551
+ {
552
+ "epoch": 0.7900207900207901,
553
+ "grad_norm": 3.421875,
554
+ "learning_rate": 1.6792208788537617e-06,
555
+ "loss": 1.2154,
556
+ "step": 380
557
+ },
558
+ {
559
+ "epoch": 0.8004158004158004,
560
+ "grad_norm": 3.625,
561
+ "learning_rate": 1.6711802618506926e-06,
562
+ "loss": 1.2134,
563
+ "step": 385
564
+ },
565
+ {
566
+ "epoch": 0.8108108108108109,
567
+ "grad_norm": 3.546875,
568
+ "learning_rate": 1.663059891913793e-06,
569
+ "loss": 1.2211,
570
+ "step": 390
571
+ },
572
+ {
573
+ "epoch": 0.8212058212058212,
574
+ "grad_norm": 3.734375,
575
+ "learning_rate": 1.6548607339452852e-06,
576
+ "loss": 1.2222,
577
+ "step": 395
578
+ },
579
+ {
580
+ "epoch": 0.8316008316008316,
581
+ "grad_norm": 3.515625,
582
+ "learning_rate": 1.6465837622093722e-06,
583
+ "loss": 1.2188,
584
+ "step": 400
585
+ },
586
+ {
587
+ "epoch": 0.841995841995842,
588
+ "grad_norm": 3.515625,
589
+ "learning_rate": 1.6382299602164706e-06,
590
+ "loss": 1.2189,
591
+ "step": 405
592
+ },
593
+ {
594
+ "epoch": 0.8523908523908524,
595
+ "grad_norm": 3.5625,
596
+ "learning_rate": 1.6298003206063466e-06,
597
+ "loss": 1.2145,
598
+ "step": 410
599
+ },
600
+ {
601
+ "epoch": 0.8627858627858628,
602
+ "grad_norm": 3.53125,
603
+ "learning_rate": 1.6212958450301625e-06,
604
+ "loss": 1.2168,
605
+ "step": 415
606
+ },
607
+ {
608
+ "epoch": 0.8731808731808732,
609
+ "grad_norm": 3.515625,
610
+ "learning_rate": 1.6127175440314594e-06,
611
+ "loss": 1.2083,
612
+ "step": 420
613
+ },
614
+ {
615
+ "epoch": 0.8835758835758836,
616
+ "grad_norm": 3.484375,
617
+ "learning_rate": 1.6040664369260758e-06,
618
+ "loss": 1.2183,
619
+ "step": 425
620
+ },
621
+ {
622
+ "epoch": 0.893970893970894,
623
+ "grad_norm": 3.671875,
624
+ "learning_rate": 1.5953435516810303e-06,
625
+ "loss": 1.2134,
626
+ "step": 430
627
+ },
628
+ {
629
+ "epoch": 0.9043659043659044,
630
+ "grad_norm": 3.515625,
631
+ "learning_rate": 1.586549924792372e-06,
632
+ "loss": 1.2145,
633
+ "step": 435
634
+ },
635
+ {
636
+ "epoch": 0.9147609147609148,
637
+ "grad_norm": 3.53125,
638
+ "learning_rate": 1.5776866011620198e-06,
639
+ "loss": 1.2148,
640
+ "step": 440
641
+ },
642
+ {
643
+ "epoch": 0.9251559251559252,
644
+ "grad_norm": 3.6875,
645
+ "learning_rate": 1.5687546339736013e-06,
646
+ "loss": 1.212,
647
+ "step": 445
648
+ },
649
+ {
650
+ "epoch": 0.9355509355509356,
651
+ "grad_norm": 3.578125,
652
+ "learning_rate": 1.559755084567309e-06,
653
+ "loss": 1.2183,
654
+ "step": 450
655
+ },
656
+ {
657
+ "epoch": 0.9459459459459459,
658
+ "grad_norm": 3.640625,
659
+ "learning_rate": 1.5506890223137857e-06,
660
+ "loss": 1.2123,
661
+ "step": 455
662
+ },
663
+ {
664
+ "epoch": 0.9563409563409564,
665
+ "grad_norm": 3.5625,
666
+ "learning_rate": 1.5415575244870578e-06,
667
+ "loss": 1.214,
668
+ "step": 460
669
+ },
670
+ {
671
+ "epoch": 0.9667359667359667,
672
+ "grad_norm": 3.59375,
673
+ "learning_rate": 1.5323616761365278e-06,
674
+ "loss": 1.2081,
675
+ "step": 465
676
+ },
677
+ {
678
+ "epoch": 0.9771309771309772,
679
+ "grad_norm": 3.515625,
680
+ "learning_rate": 1.5231025699580427e-06,
681
+ "loss": 1.1995,
682
+ "step": 470
683
+ },
684
+ {
685
+ "epoch": 0.9875259875259875,
686
+ "grad_norm": 3.484375,
687
+ "learning_rate": 1.513781306164056e-06,
688
+ "loss": 1.2226,
689
+ "step": 475
690
+ },
691
+ {
692
+ "epoch": 0.997920997920998,
693
+ "grad_norm": 3.5,
694
+ "learning_rate": 1.5043989923528937e-06,
695
+ "loss": 1.2183,
696
+ "step": 480
697
+ },
698
+ {
699
+ "epoch": 1.0083160083160083,
700
+ "grad_norm": 3.453125,
701
+ "learning_rate": 1.4949567433771448e-06,
702
+ "loss": 1.1826,
703
+ "step": 485
704
+ },
705
+ {
706
+ "epoch": 1.0187110187110187,
707
+ "grad_norm": 3.609375,
708
+ "learning_rate": 1.4854556812111887e-06,
709
+ "loss": 1.1925,
710
+ "step": 490
711
+ },
712
+ {
713
+ "epoch": 1.0291060291060292,
714
+ "grad_norm": 3.5,
715
+ "learning_rate": 1.4758969348178766e-06,
716
+ "loss": 1.19,
717
+ "step": 495
718
+ },
719
+ {
720
+ "epoch": 1.0395010395010396,
721
+ "grad_norm": 3.53125,
722
+ "learning_rate": 1.4662816400143836e-06,
723
+ "loss": 1.1963,
724
+ "step": 500
725
+ },
726
+ {
727
+ "epoch": 1.04989604989605,
728
+ "grad_norm": 3.640625,
729
+ "learning_rate": 1.4566109393372433e-06,
730
+ "loss": 1.1872,
731
+ "step": 505
732
+ },
733
+ {
734
+ "epoch": 1.0602910602910602,
735
+ "grad_norm": 3.578125,
736
+ "learning_rate": 1.4468859819065882e-06,
737
+ "loss": 1.1833,
738
+ "step": 510
739
+ },
740
+ {
741
+ "epoch": 1.0706860706860706,
742
+ "grad_norm": 3.5625,
743
+ "learning_rate": 1.4371079232896044e-06,
744
+ "loss": 1.1815,
745
+ "step": 515
746
+ },
747
+ {
748
+ "epoch": 1.0810810810810811,
749
+ "grad_norm": 3.53125,
750
+ "learning_rate": 1.4272779253632212e-06,
751
+ "loss": 1.1855,
752
+ "step": 520
753
+ },
754
+ {
755
+ "epoch": 1.0914760914760915,
756
+ "grad_norm": 3.59375,
757
+ "learning_rate": 1.4173971561760518e-06,
758
+ "loss": 1.188,
759
+ "step": 525
760
+ },
761
+ {
762
+ "epoch": 1.1018711018711018,
763
+ "grad_norm": 3.609375,
764
+ "learning_rate": 1.4074667898096009e-06,
765
+ "loss": 1.1873,
766
+ "step": 530
767
+ },
768
+ {
769
+ "epoch": 1.1122661122661124,
770
+ "grad_norm": 3.53125,
771
+ "learning_rate": 1.397488006238752e-06,
772
+ "loss": 1.1945,
773
+ "step": 535
774
+ },
775
+ {
776
+ "epoch": 1.1226611226611227,
777
+ "grad_norm": 3.5,
778
+ "learning_rate": 1.387461991191559e-06,
779
+ "loss": 1.1856,
780
+ "step": 540
781
+ },
782
+ {
783
+ "epoch": 1.133056133056133,
784
+ "grad_norm": 3.609375,
785
+ "learning_rate": 1.3773899360083524e-06,
786
+ "loss": 1.1868,
787
+ "step": 545
788
+ },
789
+ {
790
+ "epoch": 1.1434511434511434,
791
+ "grad_norm": 3.609375,
792
+ "learning_rate": 1.3672730375001773e-06,
793
+ "loss": 1.1791,
794
+ "step": 550
795
+ },
796
+ {
797
+ "epoch": 1.1538461538461537,
798
+ "grad_norm": 3.65625,
799
+ "learning_rate": 1.357112497806582e-06,
800
+ "loss": 1.1969,
801
+ "step": 555
802
+ },
803
+ {
804
+ "epoch": 1.1642411642411643,
805
+ "grad_norm": 3.609375,
806
+ "learning_rate": 1.3469095242527764e-06,
807
+ "loss": 1.1828,
808
+ "step": 560
809
+ },
810
+ {
811
+ "epoch": 1.1746361746361746,
812
+ "grad_norm": 3.515625,
813
+ "learning_rate": 1.3366653292061682e-06,
814
+ "loss": 1.1803,
815
+ "step": 565
816
+ },
817
+ {
818
+ "epoch": 1.185031185031185,
819
+ "grad_norm": 3.53125,
820
+ "learning_rate": 1.3263811299323063e-06,
821
+ "loss": 1.1803,
822
+ "step": 570
823
+ },
824
+ {
825
+ "epoch": 1.1954261954261955,
826
+ "grad_norm": 3.6875,
827
+ "learning_rate": 1.3160581484502382e-06,
828
+ "loss": 1.1823,
829
+ "step": 575
830
+ },
831
+ {
832
+ "epoch": 1.2058212058212059,
833
+ "grad_norm": 3.71875,
834
+ "learning_rate": 1.3056976113873037e-06,
835
+ "loss": 1.1832,
836
+ "step": 580
837
+ },
838
+ {
839
+ "epoch": 1.2162162162162162,
840
+ "grad_norm": 3.625,
841
+ "learning_rate": 1.2953007498333807e-06,
842
+ "loss": 1.1841,
843
+ "step": 585
844
+ },
845
+ {
846
+ "epoch": 1.2266112266112266,
847
+ "grad_norm": 3.59375,
848
+ "learning_rate": 1.284868799194602e-06,
849
+ "loss": 1.1804,
850
+ "step": 590
851
+ },
852
+ {
853
+ "epoch": 1.237006237006237,
854
+ "grad_norm": 3.5,
855
+ "learning_rate": 1.2744029990465574e-06,
856
+ "loss": 1.1811,
857
+ "step": 595
858
+ },
859
+ {
860
+ "epoch": 1.2474012474012475,
861
+ "grad_norm": 3.5625,
862
+ "learning_rate": 1.2639045929870018e-06,
863
+ "loss": 1.1794,
864
+ "step": 600
865
+ },
866
+ {
867
+ "epoch": 1.2474012474012475,
868
+ "eval_loss": 1.2036519050598145,
869
+ "eval_runtime": 10.8624,
870
+ "eval_samples_per_second": 85.616,
871
+ "eval_steps_per_second": 2.762,
872
+ "step": 600
873
+ },
874
+ {
875
+ "epoch": 1.2577962577962578,
876
+ "grad_norm": 3.609375,
877
+ "learning_rate": 1.2533748284880842e-06,
878
+ "loss": 1.1905,
879
+ "step": 605
880
+ },
881
+ {
882
+ "epoch": 1.2681912681912682,
883
+ "grad_norm": 3.5625,
884
+ "learning_rate": 1.2428149567481184e-06,
885
+ "loss": 1.1836,
886
+ "step": 610
887
+ },
888
+ {
889
+ "epoch": 1.2785862785862787,
890
+ "grad_norm": 3.5625,
891
+ "learning_rate": 1.2322262325429063e-06,
892
+ "loss": 1.1823,
893
+ "step": 615
894
+ },
895
+ {
896
+ "epoch": 1.288981288981289,
897
+ "grad_norm": 3.5625,
898
+ "learning_rate": 1.2216099140766436e-06,
899
+ "loss": 1.1836,
900
+ "step": 620
901
+ },
902
+ {
903
+ "epoch": 1.2993762993762994,
904
+ "grad_norm": 3.71875,
905
+ "learning_rate": 1.2109672628324104e-06,
906
+ "loss": 1.1837,
907
+ "step": 625
908
+ },
909
+ {
910
+ "epoch": 1.3097713097713097,
911
+ "grad_norm": 3.578125,
912
+ "learning_rate": 1.2002995434222767e-06,
913
+ "loss": 1.1827,
914
+ "step": 630
915
+ },
916
+ {
917
+ "epoch": 1.32016632016632,
918
+ "grad_norm": 3.6875,
919
+ "learning_rate": 1.1896080234370355e-06,
920
+ "loss": 1.1803,
921
+ "step": 635
922
+ },
923
+ {
924
+ "epoch": 1.3305613305613306,
925
+ "grad_norm": 3.515625,
926
+ "learning_rate": 1.178893973295581e-06,
927
+ "loss": 1.1788,
928
+ "step": 640
929
+ },
930
+ {
931
+ "epoch": 1.340956340956341,
932
+ "grad_norm": 3.640625,
933
+ "learning_rate": 1.1681586660939504e-06,
934
+ "loss": 1.1918,
935
+ "step": 645
936
+ },
937
+ {
938
+ "epoch": 1.3513513513513513,
939
+ "grad_norm": 3.578125,
940
+ "learning_rate": 1.1574033774540505e-06,
941
+ "loss": 1.1796,
942
+ "step": 650
943
+ },
944
+ {
945
+ "epoch": 1.3617463617463619,
946
+ "grad_norm": 3.65625,
947
+ "learning_rate": 1.1466293853720795e-06,
948
+ "loss": 1.1837,
949
+ "step": 655
950
+ },
951
+ {
952
+ "epoch": 1.3721413721413722,
953
+ "grad_norm": 3.609375,
954
+ "learning_rate": 1.1358379700666703e-06,
955
+ "loss": 1.1776,
956
+ "step": 660
957
+ },
958
+ {
959
+ "epoch": 1.3825363825363826,
960
+ "grad_norm": 3.546875,
961
+ "learning_rate": 1.1250304138267701e-06,
962
+ "loss": 1.181,
963
+ "step": 665
964
+ },
965
+ {
966
+ "epoch": 1.392931392931393,
967
+ "grad_norm": 3.6875,
968
+ "learning_rate": 1.11420800085927e-06,
969
+ "loss": 1.1764,
970
+ "step": 670
971
+ },
972
+ {
973
+ "epoch": 1.4033264033264032,
974
+ "grad_norm": 3.65625,
975
+ "learning_rate": 1.1033720171364108e-06,
976
+ "loss": 1.1792,
977
+ "step": 675
978
+ },
979
+ {
980
+ "epoch": 1.4137214137214138,
981
+ "grad_norm": 3.546875,
982
+ "learning_rate": 1.092523750242977e-06,
983
+ "loss": 1.1784,
984
+ "step": 680
985
+ },
986
+ {
987
+ "epoch": 1.4241164241164241,
988
+ "grad_norm": 3.65625,
989
+ "learning_rate": 1.0816644892232997e-06,
990
+ "loss": 1.1855,
991
+ "step": 685
992
+ },
993
+ {
994
+ "epoch": 1.4345114345114345,
995
+ "grad_norm": 3.640625,
996
+ "learning_rate": 1.070795524428086e-06,
997
+ "loss": 1.1782,
998
+ "step": 690
999
+ },
1000
+ {
1001
+ "epoch": 1.444906444906445,
1002
+ "grad_norm": 3.5625,
1003
+ "learning_rate": 1.0599181473610938e-06,
1004
+ "loss": 1.1837,
1005
+ "step": 695
1006
+ },
1007
+ {
1008
+ "epoch": 1.4553014553014554,
1009
+ "grad_norm": 3.59375,
1010
+ "learning_rate": 1.049033650525668e-06,
1011
+ "loss": 1.1786,
1012
+ "step": 700
1013
+ },
1014
+ {
1015
+ "epoch": 1.4656964656964657,
1016
+ "grad_norm": 3.5625,
1017
+ "learning_rate": 1.0381433272711585e-06,
1018
+ "loss": 1.1747,
1019
+ "step": 705
1020
+ },
1021
+ {
1022
+ "epoch": 1.476091476091476,
1023
+ "grad_norm": 3.71875,
1024
+ "learning_rate": 1.0272484716392408e-06,
1025
+ "loss": 1.1854,
1026
+ "step": 710
1027
+ },
1028
+ {
1029
+ "epoch": 1.4864864864864864,
1030
+ "grad_norm": 3.59375,
1031
+ "learning_rate": 1.0163503782101484e-06,
1032
+ "loss": 1.1755,
1033
+ "step": 715
1034
+ },
1035
+ {
1036
+ "epoch": 1.496881496881497,
1037
+ "grad_norm": 3.578125,
1038
+ "learning_rate": 1.0054503419488454e-06,
1039
+ "loss": 1.1795,
1040
+ "step": 720
1041
+ },
1042
+ {
1043
+ "epoch": 1.5072765072765073,
1044
+ "grad_norm": 3.578125,
1045
+ "learning_rate": 9.945496580511543e-07,
1046
+ "loss": 1.1846,
1047
+ "step": 725
1048
+ },
1049
+ {
1050
+ "epoch": 1.5176715176715176,
1051
+ "grad_norm": 3.6875,
1052
+ "learning_rate": 9.836496217898518e-07,
1053
+ "loss": 1.1806,
1054
+ "step": 730
1055
+ },
1056
+ {
1057
+ "epoch": 1.5280665280665282,
1058
+ "grad_norm": 3.609375,
1059
+ "learning_rate": 9.72751528360759e-07,
1060
+ "loss": 1.1667,
1061
+ "step": 735
1062
+ },
1063
+ {
1064
+ "epoch": 1.5384615384615383,
1065
+ "grad_norm": 3.65625,
1066
+ "learning_rate": 9.618566727288414e-07,
1067
+ "loss": 1.185,
1068
+ "step": 740
1069
+ },
1070
+ {
1071
+ "epoch": 1.5488565488565489,
1072
+ "grad_norm": 3.703125,
1073
+ "learning_rate": 9.509663494743321e-07,
1074
+ "loss": 1.191,
1075
+ "step": 745
1076
+ },
1077
+ {
1078
+ "epoch": 1.5592515592515592,
1079
+ "grad_norm": 4.9375,
1080
+ "learning_rate": 9.400818526389062e-07,
1081
+ "loss": 1.1769,
1082
+ "step": 750
1083
+ },
1084
+ {
1085
+ "epoch": 1.5696465696465696,
1086
+ "grad_norm": 3.546875,
1087
+ "learning_rate": 9.292044755719138e-07,
1088
+ "loss": 1.1741,
1089
+ "step": 755
1090
+ },
1091
+ {
1092
+ "epoch": 1.5800415800415801,
1093
+ "grad_norm": 3.546875,
1094
+ "learning_rate": 9.183355107767003e-07,
1095
+ "loss": 1.1771,
1096
+ "step": 760
1097
+ },
1098
+ {
1099
+ "epoch": 1.5904365904365905,
1100
+ "grad_norm": 3.59375,
1101
+ "learning_rate": 9.07476249757023e-07,
1102
+ "loss": 1.1725,
1103
+ "step": 765
1104
+ },
1105
+ {
1106
+ "epoch": 1.6008316008316008,
1107
+ "grad_norm": 3.59375,
1108
+ "learning_rate": 8.966279828635894e-07,
1109
+ "loss": 1.1801,
1110
+ "step": 770
1111
+ },
1112
+ {
1113
+ "epoch": 1.6112266112266114,
1114
+ "grad_norm": 3.609375,
1115
+ "learning_rate": 8.8579199914073e-07,
1116
+ "loss": 1.1697,
1117
+ "step": 775
1118
+ },
1119
+ {
1120
+ "epoch": 1.6216216216216215,
1121
+ "grad_norm": 3.546875,
1122
+ "learning_rate": 8.749695861732299e-07,
1123
+ "loss": 1.175,
1124
+ "step": 780
1125
+ },
1126
+ {
1127
+ "epoch": 1.632016632016632,
1128
+ "grad_norm": 3.578125,
1129
+ "learning_rate": 8.641620299333295e-07,
1130
+ "loss": 1.1818,
1131
+ "step": 785
1132
+ },
1133
+ {
1134
+ "epoch": 1.6424116424116424,
1135
+ "grad_norm": 3.65625,
1136
+ "learning_rate": 8.533706146279207e-07,
1137
+ "loss": 1.1697,
1138
+ "step": 790
1139
+ },
1140
+ {
1141
+ "epoch": 1.6528066528066527,
1142
+ "grad_norm": 3.59375,
1143
+ "learning_rate": 8.425966225459493e-07,
1144
+ "loss": 1.1702,
1145
+ "step": 795
1146
+ },
1147
+ {
1148
+ "epoch": 1.6632016632016633,
1149
+ "grad_norm": 3.65625,
1150
+ "learning_rate": 8.318413339060495e-07,
1151
+ "loss": 1.1793,
1152
+ "step": 800
1153
+ },
1154
+ {
1155
+ "epoch": 1.6735966735966736,
1156
+ "grad_norm": 3.65625,
1157
+ "learning_rate": 8.21106026704419e-07,
1158
+ "loss": 1.1801,
1159
+ "step": 805
1160
+ },
1161
+ {
1162
+ "epoch": 1.683991683991684,
1163
+ "grad_norm": 3.59375,
1164
+ "learning_rate": 8.103919765629645e-07,
1165
+ "loss": 1.1797,
1166
+ "step": 810
1167
+ },
1168
+ {
1169
+ "epoch": 1.6943866943866945,
1170
+ "grad_norm": 3.5625,
1171
+ "learning_rate": 7.997004565777233e-07,
1172
+ "loss": 1.1659,
1173
+ "step": 815
1174
+ },
1175
+ {
1176
+ "epoch": 1.7047817047817047,
1177
+ "grad_norm": 3.671875,
1178
+ "learning_rate": 7.890327371675895e-07,
1179
+ "loss": 1.1743,
1180
+ "step": 820
1181
+ },
1182
+ {
1183
+ "epoch": 1.7151767151767152,
1184
+ "grad_norm": 3.671875,
1185
+ "learning_rate": 7.783900859233562e-07,
1186
+ "loss": 1.1849,
1187
+ "step": 825
1188
+ },
1189
+ {
1190
+ "epoch": 1.7255717255717256,
1191
+ "grad_norm": 3.625,
1192
+ "learning_rate": 7.677737674570936e-07,
1193
+ "loss": 1.1791,
1194
+ "step": 830
1195
+ },
1196
+ {
1197
+ "epoch": 1.735966735966736,
1198
+ "grad_norm": 3.578125,
1199
+ "learning_rate": 7.571850432518819e-07,
1200
+ "loss": 1.1789,
1201
+ "step": 835
1202
+ },
1203
+ {
1204
+ "epoch": 1.7463617463617465,
1205
+ "grad_norm": 3.5625,
1206
+ "learning_rate": 7.466251715119156e-07,
1207
+ "loss": 1.1773,
1208
+ "step": 840
1209
+ },
1210
+ {
1211
+ "epoch": 1.7567567567567568,
1212
+ "grad_norm": 3.609375,
1213
+ "learning_rate": 7.360954070129981e-07,
1214
+ "loss": 1.175,
1215
+ "step": 845
1216
+ },
1217
+ {
1218
+ "epoch": 1.7671517671517671,
1219
+ "grad_norm": 3.609375,
1220
+ "learning_rate": 7.255970009534425e-07,
1221
+ "loss": 1.1816,
1222
+ "step": 850
1223
+ },
1224
+ {
1225
+ "epoch": 1.7775467775467777,
1226
+ "grad_norm": 3.734375,
1227
+ "learning_rate": 7.151312008053979e-07,
1228
+ "loss": 1.1848,
1229
+ "step": 855
1230
+ },
1231
+ {
1232
+ "epoch": 1.7879417879417878,
1233
+ "grad_norm": 3.5625,
1234
+ "learning_rate": 7.046992501666195e-07,
1235
+ "loss": 1.175,
1236
+ "step": 860
1237
+ },
1238
+ {
1239
+ "epoch": 1.7983367983367984,
1240
+ "grad_norm": 3.625,
1241
+ "learning_rate": 6.943023886126965e-07,
1242
+ "loss": 1.1753,
1243
+ "step": 865
1244
+ },
1245
+ {
1246
+ "epoch": 1.8087318087318087,
1247
+ "grad_norm": 3.5625,
1248
+ "learning_rate": 6.839418515497618e-07,
1249
+ "loss": 1.1718,
1250
+ "step": 870
1251
+ },
1252
+ {
1253
+ "epoch": 1.819126819126819,
1254
+ "grad_norm": 3.59375,
1255
+ "learning_rate": 6.736188700676935e-07,
1256
+ "loss": 1.164,
1257
+ "step": 875
1258
+ },
1259
+ {
1260
+ "epoch": 1.8295218295218296,
1261
+ "grad_norm": 3.59375,
1262
+ "learning_rate": 6.633346707938319e-07,
1263
+ "loss": 1.1761,
1264
+ "step": 880
1265
+ },
1266
+ {
1267
+ "epoch": 1.83991683991684,
1268
+ "grad_norm": 3.671875,
1269
+ "learning_rate": 6.530904757472236e-07,
1270
+ "loss": 1.1869,
1271
+ "step": 885
1272
+ },
1273
+ {
1274
+ "epoch": 1.8503118503118503,
1275
+ "grad_norm": 3.59375,
1276
+ "learning_rate": 6.42887502193418e-07,
1277
+ "loss": 1.1836,
1278
+ "step": 890
1279
+ },
1280
+ {
1281
+ "epoch": 1.8607068607068609,
1282
+ "grad_norm": 3.578125,
1283
+ "learning_rate": 6.327269624998227e-07,
1284
+ "loss": 1.1699,
1285
+ "step": 895
1286
+ },
1287
+ {
1288
+ "epoch": 1.871101871101871,
1289
+ "grad_norm": 3.65625,
1290
+ "learning_rate": 6.226100639916474e-07,
1291
+ "loss": 1.1743,
1292
+ "step": 900
1293
+ },
1294
+ {
1295
+ "epoch": 1.871101871101871,
1296
+ "eval_loss": 1.1942965984344482,
1297
+ "eval_runtime": 10.786,
1298
+ "eval_samples_per_second": 86.223,
1299
+ "eval_steps_per_second": 2.781,
1300
+ "step": 900
1301
+ },
1302
+ {
1303
+ "epoch": 1.8814968814968815,
1304
+ "grad_norm": 3.53125,
1305
+ "learning_rate": 6.125380088084408e-07,
1306
+ "loss": 1.1797,
1307
+ "step": 905
1308
+ },
1309
+ {
1310
+ "epoch": 1.8918918918918919,
1311
+ "grad_norm": 3.71875,
1312
+ "learning_rate": 6.025119937612481e-07,
1313
+ "loss": 1.1758,
1314
+ "step": 910
1315
+ },
1316
+ {
1317
+ "epoch": 1.9022869022869022,
1318
+ "grad_norm": 3.65625,
1319
+ "learning_rate": 5.925332101903994e-07,
1320
+ "loss": 1.1783,
1321
+ "step": 915
1322
+ },
1323
+ {
1324
+ "epoch": 1.9126819126819128,
1325
+ "grad_norm": 3.71875,
1326
+ "learning_rate": 5.826028438239479e-07,
1327
+ "loss": 1.1763,
1328
+ "step": 920
1329
+ },
1330
+ {
1331
+ "epoch": 1.9230769230769231,
1332
+ "grad_norm": 3.546875,
1333
+ "learning_rate": 5.727220746367791e-07,
1334
+ "loss": 1.18,
1335
+ "step": 925
1336
+ },
1337
+ {
1338
+ "epoch": 1.9334719334719335,
1339
+ "grad_norm": 3.546875,
1340
+ "learning_rate": 5.628920767103957e-07,
1341
+ "loss": 1.1782,
1342
+ "step": 930
1343
+ },
1344
+ {
1345
+ "epoch": 1.943866943866944,
1346
+ "grad_norm": 3.65625,
1347
+ "learning_rate": 5.531140180934119e-07,
1348
+ "loss": 1.1772,
1349
+ "step": 935
1350
+ },
1351
+ {
1352
+ "epoch": 1.9542619542619541,
1353
+ "grad_norm": 3.546875,
1354
+ "learning_rate": 5.433890606627568e-07,
1355
+ "loss": 1.175,
1356
+ "step": 940
1357
+ },
1358
+ {
1359
+ "epoch": 1.9646569646569647,
1360
+ "grad_norm": 3.59375,
1361
+ "learning_rate": 5.337183599856164e-07,
1362
+ "loss": 1.1745,
1363
+ "step": 945
1364
+ },
1365
+ {
1366
+ "epoch": 1.975051975051975,
1367
+ "grad_norm": 3.5625,
1368
+ "learning_rate": 5.241030651821231e-07,
1369
+ "loss": 1.1662,
1370
+ "step": 950
1371
+ },
1372
+ {
1373
+ "epoch": 1.9854469854469854,
1374
+ "grad_norm": 3.609375,
1375
+ "learning_rate": 5.145443187888114e-07,
1376
+ "loss": 1.171,
1377
+ "step": 955
1378
+ },
1379
+ {
1380
+ "epoch": 1.995841995841996,
1381
+ "grad_norm": 3.640625,
1382
+ "learning_rate": 5.050432566228552e-07,
1383
+ "loss": 1.1831,
1384
+ "step": 960
1385
+ },
1386
+ {
1387
+ "epoch": 2.006237006237006,
1388
+ "grad_norm": 3.609375,
1389
+ "learning_rate": 4.956010076471065e-07,
1390
+ "loss": 1.1592,
1391
+ "step": 965
1392
+ },
1393
+ {
1394
+ "epoch": 2.0166320166320166,
1395
+ "grad_norm": 3.53125,
1396
+ "learning_rate": 4.862186938359441e-07,
1397
+ "loss": 1.1667,
1398
+ "step": 970
1399
+ },
1400
+ {
1401
+ "epoch": 2.027027027027027,
1402
+ "grad_norm": 3.5625,
1403
+ "learning_rate": 4.768974300419573e-07,
1404
+ "loss": 1.1678,
1405
+ "step": 975
1406
+ },
1407
+ {
1408
+ "epoch": 2.0374220374220373,
1409
+ "grad_norm": 3.625,
1410
+ "learning_rate": 4.6763832386347214e-07,
1411
+ "loss": 1.1723,
1412
+ "step": 980
1413
+ },
1414
+ {
1415
+ "epoch": 2.047817047817048,
1416
+ "grad_norm": 3.578125,
1417
+ "learning_rate": 4.5844247551294224e-07,
1418
+ "loss": 1.1642,
1419
+ "step": 985
1420
+ },
1421
+ {
1422
+ "epoch": 2.0582120582120584,
1423
+ "grad_norm": 3.609375,
1424
+ "learning_rate": 4.493109776862143e-07,
1425
+ "loss": 1.17,
1426
+ "step": 990
1427
+ },
1428
+ {
1429
+ "epoch": 2.0686070686070686,
1430
+ "grad_norm": 3.640625,
1431
+ "learning_rate": 4.402449154326913e-07,
1432
+ "loss": 1.1651,
1433
+ "step": 995
1434
+ },
1435
+ {
1436
+ "epoch": 2.079002079002079,
1437
+ "grad_norm": 3.640625,
1438
+ "learning_rate": 4.312453660263987e-07,
1439
+ "loss": 1.1783,
1440
+ "step": 1000
1441
+ },
1442
+ {
1443
+ "epoch": 2.0893970893970892,
1444
+ "grad_norm": 3.578125,
1445
+ "learning_rate": 4.2231339883798025e-07,
1446
+ "loss": 1.1711,
1447
+ "step": 1005
1448
+ },
1449
+ {
1450
+ "epoch": 2.0997920997921,
1451
+ "grad_norm": 3.5625,
1452
+ "learning_rate": 4.13450075207628e-07,
1453
+ "loss": 1.163,
1454
+ "step": 1010
1455
+ },
1456
+ {
1457
+ "epoch": 2.1101871101871104,
1458
+ "grad_norm": 3.515625,
1459
+ "learning_rate": 4.0465644831897006e-07,
1460
+ "loss": 1.17,
1461
+ "step": 1015
1462
+ },
1463
+ {
1464
+ "epoch": 2.1205821205821205,
1465
+ "grad_norm": 3.5625,
1466
+ "learning_rate": 3.9593356307392436e-07,
1467
+ "loss": 1.1733,
1468
+ "step": 1020
1469
+ },
1470
+ {
1471
+ "epoch": 2.130977130977131,
1472
+ "grad_norm": 3.59375,
1473
+ "learning_rate": 3.872824559685409e-07,
1474
+ "loss": 1.1762,
1475
+ "step": 1025
1476
+ },
1477
+ {
1478
+ "epoch": 2.141372141372141,
1479
+ "grad_norm": 3.671875,
1480
+ "learning_rate": 3.7870415496983743e-07,
1481
+ "loss": 1.1734,
1482
+ "step": 1030
1483
+ },
1484
+ {
1485
+ "epoch": 2.1517671517671517,
1486
+ "grad_norm": 3.578125,
1487
+ "learning_rate": 3.701996793936535e-07,
1488
+ "loss": 1.1724,
1489
+ "step": 1035
1490
+ },
1491
+ {
1492
+ "epoch": 2.1621621621621623,
1493
+ "grad_norm": 3.5625,
1494
+ "learning_rate": 3.6177003978352917e-07,
1495
+ "loss": 1.1718,
1496
+ "step": 1040
1497
+ },
1498
+ {
1499
+ "epoch": 2.1725571725571724,
1500
+ "grad_norm": 3.5625,
1501
+ "learning_rate": 3.5341623779062813e-07,
1502
+ "loss": 1.1688,
1503
+ "step": 1045
1504
+ },
1505
+ {
1506
+ "epoch": 2.182952182952183,
1507
+ "grad_norm": 3.609375,
1508
+ "learning_rate": 3.45139266054715e-07,
1509
+ "loss": 1.1732,
1510
+ "step": 1050
1511
+ },
1512
+ {
1513
+ "epoch": 2.1933471933471935,
1514
+ "grad_norm": 3.59375,
1515
+ "learning_rate": 3.3694010808620733e-07,
1516
+ "loss": 1.1619,
1517
+ "step": 1055
1518
+ },
1519
+ {
1520
+ "epoch": 2.2037422037422036,
1521
+ "grad_norm": 3.625,
1522
+ "learning_rate": 3.288197381493075e-07,
1523
+ "loss": 1.1673,
1524
+ "step": 1060
1525
+ },
1526
+ {
1527
+ "epoch": 2.214137214137214,
1528
+ "grad_norm": 3.609375,
1529
+ "learning_rate": 3.207791211462383e-07,
1530
+ "loss": 1.1725,
1531
+ "step": 1065
1532
+ },
1533
+ {
1534
+ "epoch": 2.2245322245322248,
1535
+ "grad_norm": 3.546875,
1536
+ "learning_rate": 3.128192125025869e-07,
1537
+ "loss": 1.1673,
1538
+ "step": 1070
1539
+ },
1540
+ {
1541
+ "epoch": 2.234927234927235,
1542
+ "grad_norm": 3.578125,
1543
+ "learning_rate": 3.049409580537773e-07,
1544
+ "loss": 1.1735,
1545
+ "step": 1075
1546
+ },
1547
+ {
1548
+ "epoch": 2.2453222453222454,
1549
+ "grad_norm": 3.5625,
1550
+ "learning_rate": 2.9714529393268016e-07,
1551
+ "loss": 1.1583,
1552
+ "step": 1080
1553
+ },
1554
+ {
1555
+ "epoch": 2.2557172557172556,
1556
+ "grad_norm": 3.484375,
1557
+ "learning_rate": 2.8943314645837955e-07,
1558
+ "loss": 1.1715,
1559
+ "step": 1085
1560
+ },
1561
+ {
1562
+ "epoch": 2.266112266112266,
1563
+ "grad_norm": 3.609375,
1564
+ "learning_rate": 2.8180543202609984e-07,
1565
+ "loss": 1.164,
1566
+ "step": 1090
1567
+ },
1568
+ {
1569
+ "epoch": 2.2765072765072767,
1570
+ "grad_norm": 3.53125,
1571
+ "learning_rate": 2.742630569983182e-07,
1572
+ "loss": 1.1695,
1573
+ "step": 1095
1574
+ },
1575
+ {
1576
+ "epoch": 2.286902286902287,
1577
+ "grad_norm": 3.578125,
1578
+ "learning_rate": 2.66806917597064e-07,
1579
+ "loss": 1.169,
1580
+ "step": 1100
1581
+ },
1582
+ {
1583
+ "epoch": 2.2972972972972974,
1584
+ "grad_norm": 3.578125,
1585
+ "learning_rate": 2.594378997974267e-07,
1586
+ "loss": 1.1615,
1587
+ "step": 1105
1588
+ },
1589
+ {
1590
+ "epoch": 2.3076923076923075,
1591
+ "grad_norm": 3.6875,
1592
+ "learning_rate": 2.5215687922227845e-07,
1593
+ "loss": 1.1712,
1594
+ "step": 1110
1595
+ },
1596
+ {
1597
+ "epoch": 2.318087318087318,
1598
+ "grad_norm": 3.515625,
1599
+ "learning_rate": 2.4496472103823027e-07,
1600
+ "loss": 1.1688,
1601
+ "step": 1115
1602
+ },
1603
+ {
1604
+ "epoch": 2.3284823284823286,
1605
+ "grad_norm": 3.609375,
1606
+ "learning_rate": 2.378622798528266e-07,
1607
+ "loss": 1.1631,
1608
+ "step": 1120
1609
+ },
1610
+ {
1611
+ "epoch": 2.3388773388773387,
1612
+ "grad_norm": 3.6875,
1613
+ "learning_rate": 2.3085039961299814e-07,
1614
+ "loss": 1.1671,
1615
+ "step": 1125
1616
+ },
1617
+ {
1618
+ "epoch": 2.3492723492723493,
1619
+ "grad_norm": 3.640625,
1620
+ "learning_rate": 2.239299135047794e-07,
1621
+ "loss": 1.1623,
1622
+ "step": 1130
1623
+ },
1624
+ {
1625
+ "epoch": 2.35966735966736,
1626
+ "grad_norm": 3.578125,
1627
+ "learning_rate": 2.1710164385430585e-07,
1628
+ "loss": 1.1716,
1629
+ "step": 1135
1630
+ },
1631
+ {
1632
+ "epoch": 2.37006237006237,
1633
+ "grad_norm": 3.546875,
1634
+ "learning_rate": 2.103664020300997e-07,
1635
+ "loss": 1.1674,
1636
+ "step": 1140
1637
+ },
1638
+ {
1639
+ "epoch": 2.3804573804573805,
1640
+ "grad_norm": 3.453125,
1641
+ "learning_rate": 2.037249883466614e-07,
1642
+ "loss": 1.1623,
1643
+ "step": 1145
1644
+ },
1645
+ {
1646
+ "epoch": 2.390852390852391,
1647
+ "grad_norm": 3.65625,
1648
+ "learning_rate": 1.971781919693697e-07,
1649
+ "loss": 1.1808,
1650
+ "step": 1150
1651
+ },
1652
+ {
1653
+ "epoch": 2.401247401247401,
1654
+ "grad_norm": 3.625,
1655
+ "learning_rate": 1.9072679082071163e-07,
1656
+ "loss": 1.169,
1657
+ "step": 1155
1658
+ },
1659
+ {
1660
+ "epoch": 2.4116424116424118,
1661
+ "grad_norm": 3.671875,
1662
+ "learning_rate": 1.8437155148784433e-07,
1663
+ "loss": 1.1717,
1664
+ "step": 1160
1665
+ },
1666
+ {
1667
+ "epoch": 2.422037422037422,
1668
+ "grad_norm": 3.59375,
1669
+ "learning_rate": 1.781132291315064e-07,
1670
+ "loss": 1.1706,
1671
+ "step": 1165
1672
+ },
1673
+ {
1674
+ "epoch": 2.4324324324324325,
1675
+ "grad_norm": 3.625,
1676
+ "learning_rate": 1.7195256739628439e-07,
1677
+ "loss": 1.1722,
1678
+ "step": 1170
1679
+ },
1680
+ {
1681
+ "epoch": 2.442827442827443,
1682
+ "grad_norm": 3.71875,
1683
+ "learning_rate": 1.6589029832225155e-07,
1684
+ "loss": 1.1615,
1685
+ "step": 1175
1686
+ },
1687
+ {
1688
+ "epoch": 2.453222453222453,
1689
+ "grad_norm": 3.625,
1690
+ "learning_rate": 1.599271422579812e-07,
1691
+ "loss": 1.1691,
1692
+ "step": 1180
1693
+ },
1694
+ {
1695
+ "epoch": 2.4636174636174637,
1696
+ "grad_norm": 3.640625,
1697
+ "learning_rate": 1.5406380777495297e-07,
1698
+ "loss": 1.1647,
1699
+ "step": 1185
1700
+ },
1701
+ {
1702
+ "epoch": 2.474012474012474,
1703
+ "grad_norm": 3.59375,
1704
+ "learning_rate": 1.4830099158335563e-07,
1705
+ "loss": 1.1707,
1706
+ "step": 1190
1707
+ },
1708
+ {
1709
+ "epoch": 2.4844074844074844,
1710
+ "grad_norm": 3.578125,
1711
+ "learning_rate": 1.426393784493015e-07,
1712
+ "loss": 1.1564,
1713
+ "step": 1195
1714
+ },
1715
+ {
1716
+ "epoch": 2.494802494802495,
1717
+ "grad_norm": 3.59375,
1718
+ "learning_rate": 1.3707964111345805e-07,
1719
+ "loss": 1.1721,
1720
+ "step": 1200
1721
+ },
1722
+ {
1723
+ "epoch": 2.494802494802495,
1724
+ "eval_loss": 1.1930803060531616,
1725
+ "eval_runtime": 10.8355,
1726
+ "eval_samples_per_second": 85.829,
1727
+ "eval_steps_per_second": 2.769,
1728
+ "step": 1200
1729
+ },
1730
+ {
1731
+ "epoch": 2.505197505197505,
1732
+ "grad_norm": 3.65625,
1733
+ "learning_rate": 1.3162244021111123e-07,
1734
+ "loss": 1.1677,
1735
+ "step": 1205
1736
+ },
1737
+ {
1738
+ "epoch": 2.5155925155925156,
1739
+ "grad_norm": 3.59375,
1740
+ "learning_rate": 1.2626842419366369e-07,
1741
+ "loss": 1.1551,
1742
+ "step": 1210
1743
+ },
1744
+ {
1745
+ "epoch": 2.525987525987526,
1746
+ "grad_norm": 3.640625,
1747
+ "learning_rate": 1.2101822925158378e-07,
1748
+ "loss": 1.1678,
1749
+ "step": 1215
1750
+ },
1751
+ {
1752
+ "epoch": 2.5363825363825363,
1753
+ "grad_norm": 3.65625,
1754
+ "learning_rate": 1.1587247923881016e-07,
1755
+ "loss": 1.1731,
1756
+ "step": 1220
1757
+ },
1758
+ {
1759
+ "epoch": 2.546777546777547,
1760
+ "grad_norm": 3.6875,
1761
+ "learning_rate": 1.1083178559862227e-07,
1762
+ "loss": 1.1707,
1763
+ "step": 1225
1764
+ },
1765
+ {
1766
+ "epoch": 2.5571725571725574,
1767
+ "grad_norm": 3.671875,
1768
+ "learning_rate": 1.0589674729098507e-07,
1769
+ "loss": 1.1733,
1770
+ "step": 1230
1771
+ },
1772
+ {
1773
+ "epoch": 2.5675675675675675,
1774
+ "grad_norm": 3.578125,
1775
+ "learning_rate": 1.0106795072137896e-07,
1776
+ "loss": 1.1741,
1777
+ "step": 1235
1778
+ },
1779
+ {
1780
+ "epoch": 2.577962577962578,
1781
+ "grad_norm": 3.625,
1782
+ "learning_rate": 9.634596967111853e-08,
1783
+ "loss": 1.1704,
1784
+ "step": 1240
1785
+ },
1786
+ {
1787
+ "epoch": 2.5883575883575882,
1788
+ "grad_norm": 3.640625,
1789
+ "learning_rate": 9.173136522917457e-08,
1790
+ "loss": 1.1679,
1791
+ "step": 1245
1792
+ },
1793
+ {
1794
+ "epoch": 2.598752598752599,
1795
+ "grad_norm": 3.546875,
1796
+ "learning_rate": 8.722468572550213e-08,
1797
+ "loss": 1.1682,
1798
+ "step": 1250
1799
+ },
1800
+ {
1801
+ "epoch": 2.609147609147609,
1802
+ "grad_norm": 3.640625,
1803
+ "learning_rate": 8.28264666658851e-08,
1804
+ "loss": 1.1653,
1805
+ "step": 1255
1806
+ },
1807
+ {
1808
+ "epoch": 2.6195426195426195,
1809
+ "grad_norm": 3.734375,
1810
+ "learning_rate": 7.853723066830486e-08,
1811
+ "loss": 1.1672,
1812
+ "step": 1260
1813
+ },
1814
+ {
1815
+ "epoch": 2.62993762993763,
1816
+ "grad_norm": 3.640625,
1817
+ "learning_rate": 7.435748740084046e-08,
1818
+ "loss": 1.1606,
1819
+ "step": 1265
1820
+ },
1821
+ {
1822
+ "epoch": 2.64033264033264,
1823
+ "grad_norm": 3.65625,
1824
+ "learning_rate": 7.028773352110684e-08,
1825
+ "loss": 1.1634,
1826
+ "step": 1270
1827
+ },
1828
+ {
1829
+ "epoch": 2.6507276507276507,
1830
+ "grad_norm": 3.515625,
1831
+ "learning_rate": 6.632845261724051e-08,
1832
+ "loss": 1.1635,
1833
+ "step": 1275
1834
+ },
1835
+ {
1836
+ "epoch": 2.6611226611226613,
1837
+ "grad_norm": 4.1875,
1838
+ "learning_rate": 6.248011515043617e-08,
1839
+ "loss": 1.1641,
1840
+ "step": 1280
1841
+ },
1842
+ {
1843
+ "epoch": 2.6715176715176714,
1844
+ "grad_norm": 3.65625,
1845
+ "learning_rate": 5.8743178399044966e-08,
1846
+ "loss": 1.1642,
1847
+ "step": 1285
1848
+ },
1849
+ {
1850
+ "epoch": 2.681912681912682,
1851
+ "grad_norm": 3.625,
1852
+ "learning_rate": 5.511808640423765e-08,
1853
+ "loss": 1.1727,
1854
+ "step": 1290
1855
+ },
1856
+ {
1857
+ "epoch": 2.6923076923076925,
1858
+ "grad_norm": 3.59375,
1859
+ "learning_rate": 5.160526991724246e-08,
1860
+ "loss": 1.1732,
1861
+ "step": 1295
1862
+ },
1863
+ {
1864
+ "epoch": 2.7027027027027026,
1865
+ "grad_norm": 3.609375,
1866
+ "learning_rate": 4.8205146348160195e-08,
1867
+ "loss": 1.1699,
1868
+ "step": 1300
1869
+ },
1870
+ {
1871
+ "epoch": 2.713097713097713,
1872
+ "grad_norm": 3.6875,
1873
+ "learning_rate": 4.491811971636605e-08,
1874
+ "loss": 1.166,
1875
+ "step": 1305
1876
+ },
1877
+ {
1878
+ "epoch": 2.7234927234927238,
1879
+ "grad_norm": 3.625,
1880
+ "learning_rate": 4.174458060250208e-08,
1881
+ "loss": 1.1712,
1882
+ "step": 1310
1883
+ },
1884
+ {
1885
+ "epoch": 2.733887733887734,
1886
+ "grad_norm": 3.59375,
1887
+ "learning_rate": 3.868490610206565e-08,
1888
+ "loss": 1.1595,
1889
+ "step": 1315
1890
+ },
1891
+ {
1892
+ "epoch": 2.7442827442827444,
1893
+ "grad_norm": 3.578125,
1894
+ "learning_rate": 3.5739459780602665e-08,
1895
+ "loss": 1.1684,
1896
+ "step": 1320
1897
+ },
1898
+ {
1899
+ "epoch": 2.7546777546777546,
1900
+ "grad_norm": 3.578125,
1901
+ "learning_rate": 3.290859163050508e-08,
1902
+ "loss": 1.1744,
1903
+ "step": 1325
1904
+ },
1905
+ {
1906
+ "epoch": 2.765072765072765,
1907
+ "grad_norm": 3.671875,
1908
+ "learning_rate": 3.0192638029424735e-08,
1909
+ "loss": 1.1664,
1910
+ "step": 1330
1911
+ },
1912
+ {
1913
+ "epoch": 2.7754677754677752,
1914
+ "grad_norm": 3.53125,
1915
+ "learning_rate": 2.7591921700302222e-08,
1916
+ "loss": 1.1612,
1917
+ "step": 1335
1918
+ },
1919
+ {
1920
+ "epoch": 2.785862785862786,
1921
+ "grad_norm": 3.609375,
1922
+ "learning_rate": 2.5106751673020012e-08,
1923
+ "loss": 1.174,
1924
+ "step": 1340
1925
+ },
1926
+ {
1927
+ "epoch": 2.7962577962577964,
1928
+ "grad_norm": 3.59375,
1929
+ "learning_rate": 2.273742324768124e-08,
1930
+ "loss": 1.1602,
1931
+ "step": 1345
1932
+ },
1933
+ {
1934
+ "epoch": 2.8066528066528065,
1935
+ "grad_norm": 3.546875,
1936
+ "learning_rate": 2.048421795952171e-08,
1937
+ "loss": 1.1708,
1938
+ "step": 1350
1939
+ },
1940
+ {
1941
+ "epoch": 2.817047817047817,
1942
+ "grad_norm": 3.609375,
1943
+ "learning_rate": 1.8347403545455497e-08,
1944
+ "loss": 1.1622,
1945
+ "step": 1355
1946
+ },
1947
+ {
1948
+ "epoch": 2.8274428274428276,
1949
+ "grad_norm": 3.640625,
1950
+ "learning_rate": 1.6327233912261984e-08,
1951
+ "loss": 1.1668,
1952
+ "step": 1360
1953
+ },
1954
+ {
1955
+ "epoch": 2.8378378378378377,
1956
+ "grad_norm": 3.609375,
1957
+ "learning_rate": 1.4423949106414868e-08,
1958
+ "loss": 1.1708,
1959
+ "step": 1365
1960
+ },
1961
+ {
1962
+ "epoch": 2.8482328482328483,
1963
+ "grad_norm": 3.640625,
1964
+ "learning_rate": 1.2637775285558983e-08,
1965
+ "loss": 1.1663,
1966
+ "step": 1370
1967
+ },
1968
+ {
1969
+ "epoch": 2.858627858627859,
1970
+ "grad_norm": 3.5625,
1971
+ "learning_rate": 1.0968924691636572e-08,
1972
+ "loss": 1.1621,
1973
+ "step": 1375
1974
+ },
1975
+ {
1976
+ "epoch": 2.869022869022869,
1977
+ "grad_norm": 3.671875,
1978
+ "learning_rate": 9.417595625668462e-09,
1979
+ "loss": 1.1769,
1980
+ "step": 1380
1981
+ },
1982
+ {
1983
+ "epoch": 2.8794178794178795,
1984
+ "grad_norm": 3.71875,
1985
+ "learning_rate": 7.983972424190354e-09,
1986
+ "loss": 1.1784,
1987
+ "step": 1385
1988
+ },
1989
+ {
1990
+ "epoch": 2.88981288981289,
1991
+ "grad_norm": 3.65625,
1992
+ "learning_rate": 6.668225437349351e-09,
1993
+ "loss": 1.1734,
1994
+ "step": 1390
1995
+ },
1996
+ {
1997
+ "epoch": 2.9002079002079,
1998
+ "grad_norm": 3.75,
1999
+ "learning_rate": 5.470511008662026e-09,
2000
+ "loss": 1.1747,
2001
+ "step": 1395
2002
+ },
2003
+ {
2004
+ "epoch": 2.9106029106029108,
2005
+ "grad_norm": 3.671875,
2006
+ "learning_rate": 4.390971456437076e-09,
2007
+ "loss": 1.171,
2008
+ "step": 1400
2009
+ },
2010
+ {
2011
+ "epoch": 2.920997920997921,
2012
+ "grad_norm": 3.671875,
2013
+ "learning_rate": 3.429735056863725e-09,
2014
+ "loss": 1.1667,
2015
+ "step": 1405
2016
+ },
2017
+ {
2018
+ "epoch": 2.9313929313929314,
2019
+ "grad_norm": 3.671875,
2020
+ "learning_rate": 2.5869160287702586e-09,
2021
+ "loss": 1.1683,
2022
+ "step": 1410
2023
+ },
2024
+ {
2025
+ "epoch": 2.9417879417879416,
2026
+ "grad_norm": 3.625,
2027
+ "learning_rate": 1.8626145200513199e-09,
2028
+ "loss": 1.1702,
2029
+ "step": 1415
2030
+ },
2031
+ {
2032
+ "epoch": 2.952182952182952,
2033
+ "grad_norm": 3.5,
2034
+ "learning_rate": 1.2569165957680983e-09,
2035
+ "loss": 1.1712,
2036
+ "step": 1420
2037
+ },
2038
+ {
2039
+ "epoch": 2.9625779625779627,
2040
+ "grad_norm": 3.640625,
2041
+ "learning_rate": 7.698942279216192e-10,
2042
+ "loss": 1.1725,
2043
+ "step": 1425
2044
+ },
2045
+ {
2046
+ "epoch": 2.972972972972973,
2047
+ "grad_norm": 3.5,
2048
+ "learning_rate": 4.016052869005859e-10,
2049
+ "loss": 1.1693,
2050
+ "step": 1430
2051
+ },
2052
+ {
2053
+ "epoch": 2.9833679833679834,
2054
+ "grad_norm": 3.578125,
2055
+ "learning_rate": 1.520935346051022e-10,
2056
+ "loss": 1.1682,
2057
+ "step": 1435
2058
+ },
2059
+ {
2060
+ "epoch": 2.993762993762994,
2061
+ "grad_norm": 3.609375,
2062
+ "learning_rate": 2.1388619246498486e-11,
2063
+ "loss": 1.1627,
2064
+ "step": 1440
2065
+ },
2066
+ {
2067
+ "epoch": 3.0,
2068
+ "step": 1443,
2069
+ "total_flos": 8.072899219804914e+18,
2070
+ "train_loss": 1.2002355462548142,
2071
+ "train_runtime": 6547.9575,
2072
+ "train_samples_per_second": 14.077,
2073
+ "train_steps_per_second": 0.22
2074
+ }
2075
+ ],
2076
+ "logging_steps": 5,
2077
+ "max_steps": 1443,
2078
+ "num_input_tokens_seen": 0,
2079
+ "num_train_epochs": 3,
2080
+ "save_steps": 300,
2081
+ "stateful_callbacks": {
2082
+ "TrainerControl": {
2083
+ "args": {
2084
+ "should_epoch_stop": false,
2085
+ "should_evaluate": false,
2086
+ "should_log": false,
2087
+ "should_save": true,
2088
+ "should_training_stop": true
2089
+ },
2090
+ "attributes": {}
2091
+ }
2092
+ },
2093
+ "total_flos": 8.072899219804914e+18,
2094
+ "train_batch_size": 8,
2095
+ "trial_name": null,
2096
+ "trial_params": null
2097
+ }