Model save
Browse files- README.md +1 -1
- all_results.json +3 -3
- eval/results/data_exp_Qwen-2.5-7B-LoRA-r256a512-Simple-RL_mpl1024-mcl1024/results_2025-03-02T22-36-22.692760.json +98 -0
- model-00001-of-00004.safetensors +1 -1
- model-00002-of-00004.safetensors +1 -1
- model-00003-of-00004.safetensors +1 -1
- model-00004-of-00004.safetensors +1 -1
- train_results.json +3 -3
- trainer_state.json +506 -856
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -27,7 +27,7 @@ print(output["generated_text"])
|
|
| 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/jun-liang-chen-the-hong-kong-polytechnic-university/huggingface/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
|
|
|
| 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/jun-liang-chen-the-hong-kong-polytechnic-university/huggingface/runs/cc1xx7bd)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 7500,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": -0.0028305855572698046,
|
| 4 |
+
"train_runtime": 255598.0178,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.029,
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
eval/results/data_exp_Qwen-2.5-7B-LoRA-r256a512-Simple-RL_mpl1024-mcl1024/results_2025-03-02T22-36-22.692760.json
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"config_general": {
|
| 3 |
+
"lighteval_sha": "?",
|
| 4 |
+
"num_fewshot_seeds": 1,
|
| 5 |
+
"override_batch_size": -1,
|
| 6 |
+
"max_samples": null,
|
| 7 |
+
"job_id": 0,
|
| 8 |
+
"start_time": 9119170.875753852,
|
| 9 |
+
"end_time": 9119323.87025024,
|
| 10 |
+
"total_evaluation_time_secondes": "152.99449638836086",
|
| 11 |
+
"model_name": "data_exp_Qwen-2.5-7B-LoRA-r256a512-Simple-RL_mpl1024-mcl1024",
|
| 12 |
+
"model_sha": "",
|
| 13 |
+
"model_dtype": null,
|
| 14 |
+
"model_size": null
|
| 15 |
+
},
|
| 16 |
+
"results": {
|
| 17 |
+
"custom|math_500|0": {
|
| 18 |
+
"extractive_match": 0.744,
|
| 19 |
+
"extractive_match_stderr": 0.019536923574747605
|
| 20 |
+
},
|
| 21 |
+
"all": {
|
| 22 |
+
"extractive_match": 0.744,
|
| 23 |
+
"extractive_match_stderr": 0.019536923574747605
|
| 24 |
+
}
|
| 25 |
+
},
|
| 26 |
+
"versions": {
|
| 27 |
+
"custom|math_500|0": 1
|
| 28 |
+
},
|
| 29 |
+
"config_tasks": {
|
| 30 |
+
"custom|math_500": {
|
| 31 |
+
"name": "math_500",
|
| 32 |
+
"prompt_function": "math_prompt_fn",
|
| 33 |
+
"hf_repo": "HuggingFaceH4/MATH-500",
|
| 34 |
+
"hf_subset": "default",
|
| 35 |
+
"metric": [
|
| 36 |
+
{
|
| 37 |
+
"metric_name": "extractive_match",
|
| 38 |
+
"higher_is_better": true,
|
| 39 |
+
"category": "3",
|
| 40 |
+
"use_case": "1",
|
| 41 |
+
"sample_level_fn": "sample_level_fn",
|
| 42 |
+
"corpus_level_fn": "mean"
|
| 43 |
+
}
|
| 44 |
+
],
|
| 45 |
+
"hf_revision": null,
|
| 46 |
+
"hf_filter": null,
|
| 47 |
+
"hf_avail_splits": [
|
| 48 |
+
"test"
|
| 49 |
+
],
|
| 50 |
+
"trust_dataset": false,
|
| 51 |
+
"evaluation_splits": [
|
| 52 |
+
"test"
|
| 53 |
+
],
|
| 54 |
+
"few_shots_split": null,
|
| 55 |
+
"few_shots_select": null,
|
| 56 |
+
"generation_size": 32768,
|
| 57 |
+
"generation_grammar": null,
|
| 58 |
+
"stop_sequence": [],
|
| 59 |
+
"num_samples": null,
|
| 60 |
+
"suite": [
|
| 61 |
+
"custom"
|
| 62 |
+
],
|
| 63 |
+
"original_num_docs": 500,
|
| 64 |
+
"effective_num_docs": 500,
|
| 65 |
+
"must_remove_duplicate_docs": false,
|
| 66 |
+
"version": 1
|
| 67 |
+
}
|
| 68 |
+
},
|
| 69 |
+
"summary_tasks": {
|
| 70 |
+
"custom|math_500|0": {
|
| 71 |
+
"hashes": {
|
| 72 |
+
"hash_examples": "adf0cc8311011db2",
|
| 73 |
+
"hash_full_prompts": "361f515b5e9f6586",
|
| 74 |
+
"hash_input_tokens": "a8fc04351e3cac24",
|
| 75 |
+
"hash_cont_tokens": "00e383eb0bc9fed4"
|
| 76 |
+
},
|
| 77 |
+
"truncated": 0,
|
| 78 |
+
"non_truncated": 500,
|
| 79 |
+
"padded": 0,
|
| 80 |
+
"non_padded": 500,
|
| 81 |
+
"effective_few_shots": 0.0,
|
| 82 |
+
"num_truncated_few_shots": 0
|
| 83 |
+
}
|
| 84 |
+
},
|
| 85 |
+
"summary_general": {
|
| 86 |
+
"hashes": {
|
| 87 |
+
"hash_examples": "bfaad1993ff37a60",
|
| 88 |
+
"hash_full_prompts": "1034f05d8530f544",
|
| 89 |
+
"hash_input_tokens": "d61e53dd898aa8b5",
|
| 90 |
+
"hash_cont_tokens": "862adbb30247d743"
|
| 91 |
+
},
|
| 92 |
+
"truncated": 0,
|
| 93 |
+
"non_truncated": 500,
|
| 94 |
+
"padded": 0,
|
| 95 |
+
"non_padded": 500,
|
| 96 |
+
"num_truncated_few_shots": 0
|
| 97 |
+
}
|
| 98 |
+
}
|
model-00001-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4877660776
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20ddd578f7ba02967882b6b1a6ee1713d1e8175c167a6396b6215aa62f335a0c
|
| 3 |
size 4877660776
|
model-00002-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4932751008
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:53d2190ccdaf6cdd91a502322c095c49eb7ea96305a5502738f63272e751f764
|
| 3 |
size 4932751008
|
model-00003-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4330865200
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:89b98c13cf5f9547bdf3e911547b28e46abb518bf8424c4d4054c14a4f7cd657
|
| 3 |
size 4330865200
|
model-00004-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1089994880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36fdb9373e4f639847e1323c6b2479392a455e607795106e96b6030c41bc5716
|
| 3 |
size 1089994880
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 7500,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": -0.0028305855572698046,
|
| 4 |
+
"train_runtime": 255598.0178,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.029,
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1146 +1,796 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"clip_ratio": 0.0,
|
| 13 |
-
"completion_length":
|
| 14 |
-
"epoch": 0.
|
| 15 |
-
"grad_norm": 0.
|
| 16 |
-
"kl":
|
| 17 |
-
"learning_rate":
|
| 18 |
-
"loss": 0.
|
| 19 |
-
"reward": 0.
|
| 20 |
-
"reward_std": 0.
|
| 21 |
-
"rewards/accuracy_reward": 0.
|
| 22 |
"rewards/format_reward": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"clip_ratio": 0.0,
|
| 27 |
-
"completion_length":
|
| 28 |
-
"epoch": 0.
|
| 29 |
-
"grad_norm": 0.
|
| 30 |
-
"kl": 0.
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 0.
|
| 33 |
-
"reward": 0.
|
| 34 |
-
"reward_std": 0.
|
| 35 |
-
"rewards/accuracy_reward": 0.
|
| 36 |
"rewards/format_reward": 0.0,
|
| 37 |
"step": 5
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"clip_ratio": 0.0,
|
| 41 |
-
"completion_length":
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"grad_norm": 0.
|
| 44 |
-
"kl": 0.
|
| 45 |
-
"learning_rate":
|
| 46 |
-
"loss": 0.
|
| 47 |
-
"reward": 0.
|
| 48 |
-
"reward_std": 0.
|
| 49 |
-
"rewards/accuracy_reward": 0.
|
| 50 |
"rewards/format_reward": 0.0,
|
| 51 |
"step": 10
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"clip_ratio": 0.0,
|
| 55 |
-
"completion_length":
|
| 56 |
-
"epoch": 0.
|
| 57 |
-
"grad_norm":
|
| 58 |
-
"kl": 0.
|
| 59 |
-
"learning_rate": 1.
|
| 60 |
-
"loss": -0.
|
| 61 |
-
"reward": 0.
|
| 62 |
-
"reward_std": 0.
|
| 63 |
-
"rewards/accuracy_reward": 0.
|
| 64 |
"rewards/format_reward": 0.0,
|
| 65 |
"step": 15
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"clip_ratio": 0.0,
|
| 69 |
-
"completion_length":
|
| 70 |
-
"epoch": 0.
|
| 71 |
-
"grad_norm":
|
| 72 |
-
"kl": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
-
"reward": 0.
|
| 76 |
-
"reward_std": 0.
|
| 77 |
-
"rewards/accuracy_reward": 0.
|
| 78 |
"rewards/format_reward": 0.0,
|
| 79 |
"step": 20
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"clip_ratio": 0.0,
|
| 83 |
-
"completion_length":
|
| 84 |
-
"epoch": 0.
|
| 85 |
-
"grad_norm": 0.
|
| 86 |
-
"kl": 0.
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"loss": -0.
|
| 89 |
-
"reward": 0.
|
| 90 |
-
"reward_std": 0.
|
| 91 |
-
"rewards/accuracy_reward": 0.
|
| 92 |
"rewards/format_reward": 0.0,
|
| 93 |
"step": 25
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"clip_ratio": 0.0,
|
| 97 |
-
"completion_length":
|
| 98 |
-
"epoch": 0.
|
| 99 |
-
"grad_norm": 0.
|
| 100 |
-
"kl": 0.
|
| 101 |
-
"learning_rate": 2.
|
| 102 |
-
"loss": 0.
|
| 103 |
-
"reward": 0.
|
| 104 |
-
"reward_std": 0.
|
| 105 |
-
"rewards/accuracy_reward": 0.
|
| 106 |
"rewards/format_reward": 0.0,
|
| 107 |
"step": 30
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"clip_ratio": 0.0,
|
| 111 |
-
"completion_length":
|
| 112 |
-
"epoch": 0.
|
| 113 |
-
"grad_norm": 0.
|
| 114 |
-
"kl":
|
| 115 |
-
"learning_rate": 2.
|
| 116 |
-
"loss": 0.
|
| 117 |
-
"reward": 0.
|
| 118 |
-
"reward_std": 0.
|
| 119 |
-
"rewards/accuracy_reward": 0.
|
| 120 |
"rewards/format_reward": 0.0,
|
| 121 |
"step": 35
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"clip_ratio": 0.0,
|
| 125 |
-
"completion_length":
|
| 126 |
-
"epoch": 0.
|
| 127 |
-
"grad_norm":
|
| 128 |
-
"kl": 0.
|
| 129 |
-
"learning_rate":
|
| 130 |
-
"loss": 0.
|
| 131 |
-
"reward": 0.
|
| 132 |
-
"reward_std": 0.
|
| 133 |
-
"rewards/accuracy_reward": 0.
|
| 134 |
-
"rewards/format_reward": 0.
|
| 135 |
"step": 40
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"clip_ratio": 0.0,
|
| 139 |
-
"completion_length":
|
| 140 |
-
"epoch": 0.
|
| 141 |
-
"grad_norm": 0.
|
| 142 |
-
"kl": 0.
|
| 143 |
-
"learning_rate": 2.
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"reward": 0.
|
| 146 |
-
"reward_std": 0.
|
| 147 |
-
"rewards/accuracy_reward": 0.
|
| 148 |
"rewards/format_reward": 0.0,
|
| 149 |
"step": 45
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"clip_ratio": 0.0,
|
| 153 |
-
"completion_length":
|
| 154 |
-
"epoch": 0.
|
| 155 |
-
"grad_norm":
|
| 156 |
-
"kl": 0.
|
| 157 |
-
"learning_rate": 2.
|
| 158 |
-
"loss":
|
| 159 |
-
"reward": 0.
|
| 160 |
-
"reward_std": 0.
|
| 161 |
-
"rewards/accuracy_reward": 0.
|
| 162 |
"rewards/format_reward": 0.0,
|
| 163 |
"step": 50
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"clip_ratio": 0.0,
|
| 167 |
-
"completion_length":
|
| 168 |
-
"epoch": 0.
|
| 169 |
-
"grad_norm": 0.
|
| 170 |
-
"kl": 0.
|
| 171 |
-
"learning_rate": 2.
|
| 172 |
-
"loss": 0.
|
| 173 |
-
"reward": 0.
|
| 174 |
-
"reward_std": 0.
|
| 175 |
-
"rewards/accuracy_reward": 0.
|
| 176 |
"rewards/format_reward": 0.0,
|
| 177 |
"step": 55
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"clip_ratio": 0.0,
|
| 181 |
-
"completion_length":
|
| 182 |
-
"epoch": 0.
|
| 183 |
-
"grad_norm": 0.
|
| 184 |
-
"kl": 0.
|
| 185 |
-
"learning_rate": 2.
|
| 186 |
-
"loss": 0.
|
| 187 |
-
"reward": 0.
|
| 188 |
-
"reward_std": 0.
|
| 189 |
-
"rewards/accuracy_reward": 0.
|
| 190 |
"rewards/format_reward": 0.0,
|
| 191 |
"step": 60
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"clip_ratio": 0.0,
|
| 195 |
-
"completion_length":
|
| 196 |
-
"epoch": 0.
|
| 197 |
-
"grad_norm": 0.
|
| 198 |
-
"kl":
|
| 199 |
-
"learning_rate": 2.
|
| 200 |
-
"loss": 0.
|
| 201 |
-
"reward": 0.
|
| 202 |
-
"reward_std": 0.
|
| 203 |
-
"rewards/accuracy_reward": 0.
|
| 204 |
"rewards/format_reward": 0.0,
|
| 205 |
"step": 65
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"clip_ratio": 0.0,
|
| 209 |
-
"completion_length":
|
| 210 |
-
"epoch": 0.
|
| 211 |
-
"grad_norm": 0.
|
| 212 |
-
"kl": 0.
|
| 213 |
-
"learning_rate": 2.
|
| 214 |
-
"loss": 0.
|
| 215 |
-
"reward": 0.
|
| 216 |
-
"reward_std": 0.
|
| 217 |
-
"rewards/accuracy_reward": 0.
|
| 218 |
"rewards/format_reward": 0.0,
|
| 219 |
"step": 70
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"clip_ratio": 0.0,
|
| 223 |
-
"completion_length":
|
| 224 |
-
"epoch": 0.
|
| 225 |
-
"grad_norm": 0.
|
| 226 |
-
"kl": 0.
|
| 227 |
-
"learning_rate": 2.
|
| 228 |
-
"loss": 0.
|
| 229 |
-
"reward": 0.
|
| 230 |
-
"reward_std": 0.
|
| 231 |
-
"rewards/accuracy_reward": 0.
|
| 232 |
"rewards/format_reward": 0.0,
|
| 233 |
"step": 75
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"clip_ratio": 0.0,
|
| 237 |
-
"completion_length":
|
| 238 |
-
"epoch": 0.
|
| 239 |
-
"grad_norm":
|
| 240 |
-
"kl": 0.
|
| 241 |
-
"learning_rate": 2.
|
| 242 |
-
"loss": 0.
|
| 243 |
-
"reward": 0.
|
| 244 |
-
"reward_std": 0.
|
| 245 |
-
"rewards/accuracy_reward": 0.
|
| 246 |
"rewards/format_reward": 0.0,
|
| 247 |
"step": 80
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"clip_ratio": 0.0,
|
| 251 |
-
"completion_length":
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"grad_norm": 0.
|
| 254 |
-
"kl": 0.
|
| 255 |
-
"learning_rate": 2.
|
| 256 |
-
"loss": 0.
|
| 257 |
-
"reward": 0.
|
| 258 |
-
"reward_std": 0.
|
| 259 |
-
"rewards/accuracy_reward": 0.
|
| 260 |
"rewards/format_reward": 0.0,
|
| 261 |
"step": 85
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"clip_ratio": 0.0,
|
| 265 |
-
"completion_length":
|
| 266 |
-
"epoch": 0.
|
| 267 |
-
"grad_norm": 0.
|
| 268 |
-
"kl": 0.
|
| 269 |
-
"learning_rate": 2.
|
| 270 |
-
"loss": 0.
|
| 271 |
-
"reward": 0.
|
| 272 |
-
"reward_std": 0.
|
| 273 |
-
"rewards/accuracy_reward": 0.
|
| 274 |
"rewards/format_reward": 0.0,
|
| 275 |
"step": 90
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"clip_ratio": 0.0,
|
| 279 |
-
"completion_length":
|
| 280 |
-
"epoch": 0.
|
| 281 |
-
"grad_norm": 0.
|
| 282 |
-
"kl": 0.
|
| 283 |
-
"learning_rate": 2.
|
| 284 |
-
"loss": 0.
|
| 285 |
-
"reward": 0.
|
| 286 |
-
"reward_std": 0.
|
| 287 |
-
"rewards/accuracy_reward": 0.
|
| 288 |
"rewards/format_reward": 0.0,
|
| 289 |
"step": 95
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"clip_ratio": 0.0,
|
| 293 |
-
"completion_length":
|
| 294 |
-
"epoch": 0.
|
| 295 |
-
"grad_norm": 0.
|
| 296 |
-
"kl": 0.
|
| 297 |
-
"learning_rate": 2.
|
| 298 |
-
"loss": 0.
|
| 299 |
-
"reward": 0.
|
| 300 |
-
"reward_std": 0.
|
| 301 |
-
"rewards/accuracy_reward": 0.
|
| 302 |
"rewards/format_reward": 0.0,
|
| 303 |
"step": 100
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"clip_ratio": 0.0,
|
| 307 |
-
"completion_length":
|
| 308 |
-
"epoch": 0.
|
| 309 |
-
"grad_norm":
|
| 310 |
-
"kl": 0.
|
| 311 |
-
"learning_rate": 2.
|
| 312 |
-
"loss": 0.
|
| 313 |
-
"reward": 0.
|
| 314 |
-
"reward_std": 0.
|
| 315 |
-
"rewards/accuracy_reward": 0.
|
| 316 |
"rewards/format_reward": 0.0,
|
| 317 |
"step": 105
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"clip_ratio": 0.0,
|
| 321 |
-
"completion_length":
|
| 322 |
-
"epoch": 0.
|
| 323 |
-
"grad_norm": 0.
|
| 324 |
-
"kl": 0.
|
| 325 |
-
"learning_rate": 2.
|
| 326 |
-
"loss": 0.
|
| 327 |
-
"reward": 0.
|
| 328 |
-
"reward_std": 0.
|
| 329 |
-
"rewards/accuracy_reward": 0.
|
| 330 |
"rewards/format_reward": 0.0,
|
| 331 |
"step": 110
|
| 332 |
},
|
| 333 |
{
|
| 334 |
"clip_ratio": 0.0,
|
| 335 |
-
"completion_length":
|
| 336 |
-
"epoch": 0.
|
| 337 |
-
"grad_norm": 0.
|
| 338 |
-
"kl": 0.
|
| 339 |
-
"learning_rate": 2.
|
| 340 |
-
"loss": 0.
|
| 341 |
-
"reward": 0.
|
| 342 |
-
"reward_std": 0.
|
| 343 |
-
"rewards/accuracy_reward": 0.
|
| 344 |
"rewards/format_reward": 0.0,
|
| 345 |
"step": 115
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"clip_ratio": 0.0,
|
| 349 |
-
"completion_length":
|
| 350 |
-
"epoch": 0.
|
| 351 |
-
"grad_norm":
|
| 352 |
-
"kl": 0.
|
| 353 |
-
"learning_rate": 2.
|
| 354 |
-
"loss": 0.
|
| 355 |
-
"reward": 0.
|
| 356 |
-
"reward_std": 0.
|
| 357 |
-
"rewards/accuracy_reward": 0.
|
| 358 |
"rewards/format_reward": 0.0,
|
| 359 |
"step": 120
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"clip_ratio": 0.0,
|
| 363 |
-
"completion_length":
|
| 364 |
-
"epoch": 0.
|
| 365 |
-
"grad_norm": 0.
|
| 366 |
-
"kl": 0.
|
| 367 |
-
"learning_rate":
|
| 368 |
-
"loss": 0.
|
| 369 |
-
"reward": 0.
|
| 370 |
-
"reward_std": 0.
|
| 371 |
-
"rewards/accuracy_reward": 0.
|
| 372 |
"rewards/format_reward": 0.0,
|
| 373 |
"step": 125
|
| 374 |
},
|
| 375 |
{
|
| 376 |
"clip_ratio": 0.0,
|
| 377 |
-
"completion_length":
|
| 378 |
-
"epoch": 0.
|
| 379 |
-
"grad_norm": 0.
|
| 380 |
-
"kl": 0.
|
| 381 |
-
"learning_rate":
|
| 382 |
-
"loss": 0.
|
| 383 |
-
"reward": 0.
|
| 384 |
-
"reward_std": 0.
|
| 385 |
-
"rewards/accuracy_reward": 0.
|
| 386 |
"rewards/format_reward": 0.0,
|
| 387 |
"step": 130
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"clip_ratio": 0.0,
|
| 391 |
-
"completion_length":
|
| 392 |
-
"epoch": 0.
|
| 393 |
-
"grad_norm": 0.
|
| 394 |
-
"kl": 0.
|
| 395 |
-
"learning_rate":
|
| 396 |
-
"loss": 0.
|
| 397 |
-
"reward": 0.
|
| 398 |
-
"reward_std": 0.
|
| 399 |
-
"rewards/accuracy_reward": 0.
|
| 400 |
"rewards/format_reward": 0.0,
|
| 401 |
"step": 135
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"clip_ratio": 0.0,
|
| 405 |
-
"completion_length":
|
| 406 |
-
"epoch": 0.
|
| 407 |
-
"grad_norm": 0.
|
| 408 |
-
"kl": 0.
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"loss": -0.
|
| 411 |
-
"reward": 0.
|
| 412 |
-
"reward_std": 0.
|
| 413 |
-
"rewards/accuracy_reward": 0.
|
| 414 |
"rewards/format_reward": 0.0,
|
| 415 |
"step": 140
|
| 416 |
},
|
| 417 |
{
|
| 418 |
"clip_ratio": 0.0,
|
| 419 |
-
"completion_length":
|
| 420 |
-
"epoch": 0.
|
| 421 |
-
"grad_norm": 0.
|
| 422 |
-
"kl":
|
| 423 |
-
"learning_rate":
|
| 424 |
-
"loss": 0.
|
| 425 |
-
"reward": 0.
|
| 426 |
-
"reward_std": 0.
|
| 427 |
-
"rewards/accuracy_reward": 0.
|
| 428 |
"rewards/format_reward": 0.0,
|
| 429 |
"step": 145
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"clip_ratio": 0.0,
|
| 433 |
-
"completion_length":
|
| 434 |
-
"epoch": 0.
|
| 435 |
-
"grad_norm": 0.
|
| 436 |
-
"kl": 0.
|
| 437 |
-
"learning_rate":
|
| 438 |
-
"loss": 0.
|
| 439 |
-
"reward": 0.
|
| 440 |
-
"reward_std": 0.
|
| 441 |
-
"rewards/accuracy_reward": 0.
|
| 442 |
"rewards/format_reward": 0.0,
|
| 443 |
"step": 150
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"clip_ratio": 0.0,
|
| 447 |
-
"completion_length":
|
| 448 |
-
"epoch": 0.
|
| 449 |
-
"grad_norm": 0.
|
| 450 |
-
"kl": 0.
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss": 0.
|
| 453 |
-
"reward": 0.
|
| 454 |
-
"reward_std": 0.
|
| 455 |
-
"rewards/accuracy_reward": 0.
|
| 456 |
"rewards/format_reward": 0.0,
|
| 457 |
"step": 155
|
| 458 |
},
|
| 459 |
{
|
| 460 |
"clip_ratio": 0.0,
|
| 461 |
-
"completion_length":
|
| 462 |
-
"epoch": 0.
|
| 463 |
-
"grad_norm": 0.
|
| 464 |
-
"kl": 0.
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"loss": 0.
|
| 467 |
-
"reward": 0.
|
| 468 |
-
"reward_std": 0.
|
| 469 |
-
"rewards/accuracy_reward": 0.
|
| 470 |
"rewards/format_reward": 0.0,
|
| 471 |
"step": 160
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"clip_ratio": 0.0,
|
| 475 |
-
"completion_length":
|
| 476 |
-
"epoch": 0.
|
| 477 |
-
"grad_norm": 0.
|
| 478 |
-
"kl": 0.
|
| 479 |
-
"learning_rate":
|
| 480 |
-
"loss": 0.
|
| 481 |
-
"reward": 0.
|
| 482 |
-
"reward_std": 0.
|
| 483 |
-
"rewards/accuracy_reward": 0.
|
| 484 |
"rewards/format_reward": 0.0,
|
| 485 |
"step": 165
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"clip_ratio": 0.0,
|
| 489 |
-
"completion_length":
|
| 490 |
-
"epoch": 0.
|
| 491 |
-
"grad_norm": 0.
|
| 492 |
-
"kl": 0.
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss": 0.
|
| 495 |
-
"reward": 0.
|
| 496 |
-
"reward_std": 0.
|
| 497 |
-
"rewards/accuracy_reward": 0.
|
| 498 |
"rewards/format_reward": 0.0,
|
| 499 |
"step": 170
|
| 500 |
},
|
| 501 |
{
|
| 502 |
"clip_ratio": 0.0,
|
| 503 |
-
"completion_length":
|
| 504 |
-
"epoch": 0.
|
| 505 |
-
"grad_norm":
|
| 506 |
-
"kl": 0.
|
| 507 |
-
"learning_rate":
|
| 508 |
-
"loss":
|
| 509 |
-
"reward": 0.
|
| 510 |
-
"reward_std": 0.
|
| 511 |
-
"rewards/accuracy_reward": 0.
|
| 512 |
"rewards/format_reward": 0.0,
|
| 513 |
"step": 175
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"clip_ratio": 0.0,
|
| 517 |
-
"completion_length":
|
| 518 |
-
"epoch": 0.
|
| 519 |
-
"grad_norm":
|
| 520 |
-
"kl":
|
| 521 |
-
"learning_rate":
|
| 522 |
-
"loss": 0.
|
| 523 |
-
"reward": 0.
|
| 524 |
-
"reward_std": 0.
|
| 525 |
-
"rewards/accuracy_reward": 0.
|
| 526 |
"rewards/format_reward": 0.0,
|
| 527 |
"step": 180
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"clip_ratio": 0.0,
|
| 531 |
-
"completion_length":
|
| 532 |
-
"epoch": 0.
|
| 533 |
-
"grad_norm":
|
| 534 |
-
"kl":
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss": 0.
|
| 537 |
-
"reward": 0.
|
| 538 |
-
"reward_std": 0.
|
| 539 |
-
"rewards/accuracy_reward": 0.
|
| 540 |
"rewards/format_reward": 0.0,
|
| 541 |
"step": 185
|
| 542 |
},
|
| 543 |
{
|
| 544 |
"clip_ratio": 0.0,
|
| 545 |
-
"completion_length":
|
| 546 |
-
"epoch": 0.
|
| 547 |
-
"grad_norm":
|
| 548 |
-
"kl":
|
| 549 |
-
"learning_rate":
|
| 550 |
-
"loss": 0.
|
| 551 |
-
"reward": 0.
|
| 552 |
-
"reward_std": 0.
|
| 553 |
-
"rewards/accuracy_reward": 0.
|
| 554 |
"rewards/format_reward": 0.0,
|
| 555 |
"step": 190
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"clip_ratio": 0.0,
|
| 559 |
-
"completion_length":
|
| 560 |
-
"epoch": 0.
|
| 561 |
-
"grad_norm": 0.
|
| 562 |
-
"kl": 0.
|
| 563 |
-
"learning_rate":
|
| 564 |
-
"loss": 0.
|
| 565 |
-
"reward": 0.
|
| 566 |
-
"reward_std": 0.
|
| 567 |
-
"rewards/accuracy_reward": 0.
|
| 568 |
"rewards/format_reward": 0.0,
|
| 569 |
"step": 195
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch": 0.
|
| 573 |
-
"grad_norm": 0.
|
| 574 |
-
"learning_rate":
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 200
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch": 0.
|
| 580 |
"eval_clip_ratio": 0.0,
|
| 581 |
-
"eval_completion_length":
|
| 582 |
-
"eval_kl":
|
| 583 |
-
"eval_loss": 0.
|
| 584 |
-
"eval_reward": 0.
|
| 585 |
-
"eval_reward_std": 0.
|
| 586 |
-
"eval_rewards/accuracy_reward": 0.
|
| 587 |
"eval_rewards/format_reward": 0.0,
|
| 588 |
-
"eval_runtime":
|
| 589 |
-
"eval_samples_per_second": 0.
|
| 590 |
-
"eval_steps_per_second": 0.
|
| 591 |
"step": 200
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"clip_ratio": 0.0,
|
| 595 |
-
"completion_length":
|
| 596 |
-
"epoch": 0.
|
| 597 |
-
"grad_norm": 0.
|
| 598 |
-
"kl": 0.
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"loss": 0.
|
| 601 |
-
"reward": 0.
|
| 602 |
-
"reward_std": 0.
|
| 603 |
-
"rewards/accuracy_reward": 0.
|
| 604 |
"rewards/format_reward": 0.0,
|
| 605 |
"step": 205
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"clip_ratio": 0.0,
|
| 609 |
-
"completion_length":
|
| 610 |
-
"epoch": 0.
|
| 611 |
-
"grad_norm":
|
| 612 |
-
"kl": 0.
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 0.
|
| 615 |
-
"reward": 0.
|
| 616 |
-
"reward_std": 0.
|
| 617 |
-
"rewards/accuracy_reward": 0.
|
| 618 |
-
"rewards/format_reward": 0.
|
| 619 |
"step": 210
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"clip_ratio": 0.0,
|
| 623 |
-
"completion_length":
|
| 624 |
-
"epoch": 0.
|
| 625 |
-
"grad_norm": 0.
|
| 626 |
-
"kl": 0.
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"loss": 0.
|
| 629 |
-
"reward": 0.
|
| 630 |
-
"reward_std": 0.
|
| 631 |
-
"rewards/accuracy_reward": 0.
|
| 632 |
"rewards/format_reward": 0.0,
|
| 633 |
"step": 215
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"clip_ratio": 0.0,
|
| 637 |
-
"completion_length":
|
| 638 |
-
"epoch": 0.
|
| 639 |
-
"grad_norm":
|
| 640 |
-
"kl": 0.
|
| 641 |
-
"learning_rate":
|
| 642 |
-
"loss": 0.
|
| 643 |
-
"reward": 0.
|
| 644 |
-
"reward_std": 0.
|
| 645 |
-
"rewards/accuracy_reward": 0.
|
| 646 |
"rewards/format_reward": 0.0,
|
| 647 |
"step": 220
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"clip_ratio": 0.0,
|
| 651 |
-
"completion_length":
|
| 652 |
-
"epoch": 0.
|
| 653 |
-
"grad_norm": 0.
|
| 654 |
-
"kl": 0.
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
-
"reward": 0.
|
| 658 |
-
"reward_std": 0.
|
| 659 |
-
"rewards/accuracy_reward": 0.
|
| 660 |
"rewards/format_reward": 0.0,
|
| 661 |
"step": 225
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"clip_ratio": 0.0,
|
| 665 |
-
"completion_length":
|
| 666 |
-
"epoch": 0.
|
| 667 |
-
"grad_norm":
|
| 668 |
-
"kl": 0.
|
| 669 |
-
"learning_rate": 1.
|
| 670 |
-
"loss": 0.
|
| 671 |
-
"reward": 0.
|
| 672 |
-
"reward_std": 0.
|
| 673 |
-
"rewards/accuracy_reward": 0.
|
| 674 |
"rewards/format_reward": 0.0,
|
| 675 |
"step": 230
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"clip_ratio": 0.0,
|
| 679 |
-
"completion_length":
|
| 680 |
-
"epoch": 0.
|
| 681 |
-
"grad_norm": 0.
|
| 682 |
-
"kl": 0.
|
| 683 |
-
"learning_rate": 1.
|
| 684 |
-
"loss": 0.
|
| 685 |
-
"reward": 0.
|
| 686 |
-
"reward_std": 0.
|
| 687 |
-
"rewards/accuracy_reward": 0.
|
| 688 |
"rewards/format_reward": 0.0,
|
| 689 |
"step": 235
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"clip_ratio": 0.0,
|
| 693 |
-
"completion_length":
|
| 694 |
-
"epoch": 0.
|
| 695 |
-
"grad_norm": 0.
|
| 696 |
-
"kl":
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss": 0.
|
| 699 |
-
"reward": 0.
|
| 700 |
-
"reward_std": 0.
|
| 701 |
-
"rewards/accuracy_reward": 0.
|
| 702 |
"rewards/format_reward": 0.0,
|
| 703 |
"step": 240
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"clip_ratio": 0.0,
|
| 707 |
-
"completion_length": 598.
|
| 708 |
-
"epoch": 0.
|
| 709 |
-
"grad_norm":
|
| 710 |
-
"kl":
|
| 711 |
-
"learning_rate":
|
| 712 |
-
"loss": 0.
|
| 713 |
-
"reward": 0.
|
| 714 |
-
"reward_std": 0.
|
| 715 |
-
"rewards/accuracy_reward": 0.
|
| 716 |
"rewards/format_reward": 0.0,
|
| 717 |
"step": 245
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"clip_ratio": 0.0,
|
| 721 |
-
"completion_length":
|
| 722 |
-
"epoch": 0.
|
| 723 |
-
"grad_norm":
|
| 724 |
-
"kl": 0.
|
| 725 |
-
"learning_rate":
|
| 726 |
-
"loss": 0.
|
| 727 |
-
"reward": 0.
|
| 728 |
-
"reward_std": 0.
|
| 729 |
-
"rewards/accuracy_reward": 0.
|
| 730 |
"rewards/format_reward": 0.0,
|
| 731 |
"step": 250
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"clip_ratio": 0.0,
|
| 735 |
-
"completion_length":
|
| 736 |
-
"epoch": 0.
|
| 737 |
-
"grad_norm":
|
| 738 |
-
"kl":
|
| 739 |
-
"learning_rate": 1.
|
| 740 |
-
"loss": 0.
|
| 741 |
-
"reward": 0.
|
| 742 |
-
"reward_std": 0.
|
| 743 |
-
"rewards/accuracy_reward": 0.
|
| 744 |
"rewards/format_reward": 0.0,
|
| 745 |
"step": 255
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"clip_ratio": 0.0,
|
| 749 |
-
"completion_length":
|
| 750 |
-
"epoch": 0.
|
| 751 |
-
"grad_norm":
|
| 752 |
-
"kl": 0.
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"loss": 0.
|
| 755 |
-
"reward": 0.
|
| 756 |
-
"reward_std": 0.
|
| 757 |
-
"rewards/accuracy_reward": 0.
|
| 758 |
"rewards/format_reward": 0.0,
|
| 759 |
"step": 260
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"clip_ratio": 0.0,
|
| 763 |
-
"completion_length":
|
| 764 |
-
"epoch": 0.
|
| 765 |
-
"grad_norm":
|
| 766 |
-
"kl":
|
| 767 |
-
"learning_rate":
|
| 768 |
-
"loss":
|
| 769 |
-
"reward": 0.
|
| 770 |
-
"reward_std": 0.
|
| 771 |
-
"rewards/accuracy_reward": 0.
|
| 772 |
"rewards/format_reward": 0.0,
|
| 773 |
"step": 265
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"clip_ratio": 0.0,
|
| 777 |
-
"completion_length":
|
| 778 |
-
"epoch": 0.
|
| 779 |
-
"
|
| 780 |
-
"
|
| 781 |
-
"
|
| 782 |
-
"
|
| 783 |
-
"reward": 0.7385965019464493,
|
| 784 |
-
"reward_std": 0.20718254509725068,
|
| 785 |
-
"rewards/accuracy_reward": 0.7385965019464493,
|
| 786 |
-
"rewards/format_reward": 0.0,
|
| 787 |
-
"step": 270
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"clip_ratio": 0.0,
|
| 791 |
-
"completion_length": 614.8351022820724,
|
| 792 |
-
"epoch": 0.6966666666666667,
|
| 793 |
-
"grad_norm": 31.654590606689453,
|
| 794 |
-
"kl": 0.7249486019736842,
|
| 795 |
-
"learning_rate": 7.615577684223272e-07,
|
| 796 |
-
"loss": 0.0309,
|
| 797 |
-
"reward": 0.7473684353263754,
|
| 798 |
-
"reward_std": 0.241305032529329,
|
| 799 |
-
"rewards/accuracy_reward": 0.7473684353263754,
|
| 800 |
-
"rewards/format_reward": 0.0,
|
| 801 |
-
"step": 275
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"clip_ratio": 0.0,
|
| 805 |
-
"completion_length": 598.9438759251644,
|
| 806 |
-
"epoch": 0.7093333333333334,
|
| 807 |
-
"grad_norm": 5.547906875610352,
|
| 808 |
-
"kl": 0.7660654168379935,
|
| 809 |
-
"learning_rate": 7.043686378203864e-07,
|
| 810 |
-
"loss": 0.0408,
|
| 811 |
-
"reward": 0.7596491347802313,
|
| 812 |
-
"reward_std": 0.25206103419002734,
|
| 813 |
-
"rewards/accuracy_reward": 0.7596491347802313,
|
| 814 |
-
"rewards/format_reward": 0.0,
|
| 815 |
-
"step": 280
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"clip_ratio": 0.0,
|
| 819 |
-
"completion_length": 620.8754567999589,
|
| 820 |
-
"epoch": 0.722,
|
| 821 |
-
"grad_norm": 4.756077289581299,
|
| 822 |
-
"kl": 0.8680445620888158,
|
| 823 |
-
"learning_rate": 6.487458033099425e-07,
|
| 824 |
-
"loss": 0.0573,
|
| 825 |
-
"reward": 0.6789473770480406,
|
| 826 |
-
"reward_std": 0.2216498475325735,
|
| 827 |
-
"rewards/accuracy_reward": 0.6789473770480406,
|
| 828 |
-
"rewards/format_reward": 0.0,
|
| 829 |
-
"step": 285
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"clip_ratio": 0.0,
|
| 833 |
-
"completion_length": 642.0228240966796,
|
| 834 |
-
"epoch": 0.7346666666666667,
|
| 835 |
-
"grad_norm": 3.3402116298675537,
|
| 836 |
-
"kl": 1.7112895764802631,
|
| 837 |
-
"learning_rate": 5.947987651349942e-07,
|
| 838 |
-
"loss": 0.0737,
|
| 839 |
-
"reward": 0.6964912395728262,
|
| 840 |
-
"reward_std": 0.26836833514665304,
|
| 841 |
-
"rewards/accuracy_reward": 0.6964912395728262,
|
| 842 |
-
"rewards/format_reward": 0.0,
|
| 843 |
-
"step": 290
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"clip_ratio": 0.0,
|
| 847 |
-
"completion_length": 597.8894905491879,
|
| 848 |
-
"epoch": 0.7473333333333333,
|
| 849 |
-
"grad_norm": 2.271803617477417,
|
| 850 |
-
"kl": 0.4571268182051809,
|
| 851 |
-
"learning_rate": 5.426337245327703e-07,
|
| 852 |
-
"loss": 0.0476,
|
| 853 |
-
"reward": 0.673684225427477,
|
| 854 |
-
"reward_std": 0.23058271408081055,
|
| 855 |
-
"rewards/accuracy_reward": 0.673684225427477,
|
| 856 |
-
"rewards/format_reward": 0.0,
|
| 857 |
-
"step": 295
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"clip_ratio": 0.0,
|
| 861 |
-
"completion_length": 586.7842289974815,
|
| 862 |
-
"epoch": 0.76,
|
| 863 |
-
"grad_norm": 0.23953944444656372,
|
| 864 |
-
"kl": 0.30870281018708884,
|
| 865 |
-
"learning_rate": 4.923533746638108e-07,
|
| 866 |
-
"loss": 0.0216,
|
| 867 |
-
"reward": 0.7614035162486528,
|
| 868 |
-
"reward_std": 0.2001630961894989,
|
| 869 |
-
"rewards/accuracy_reward": 0.7614035162486528,
|
| 870 |
-
"rewards/format_reward": 0.0,
|
| 871 |
-
"step": 300
|
| 872 |
-
},
|
| 873 |
-
{
|
| 874 |
-
"clip_ratio": 0.0,
|
| 875 |
-
"completion_length": 571.1228213661595,
|
| 876 |
-
"epoch": 0.7726666666666666,
|
| 877 |
-
"grad_norm": 0.22501805424690247,
|
| 878 |
-
"kl": 0.35944968775699015,
|
| 879 |
-
"learning_rate": 4.440566984481256e-07,
|
| 880 |
-
"loss": 0.0413,
|
| 881 |
-
"reward": 0.7157894881148087,
|
| 882 |
-
"reward_std": 0.2280832472600435,
|
| 883 |
-
"rewards/accuracy_reward": 0.7157894881148087,
|
| 884 |
-
"rewards/format_reward": 0.0,
|
| 885 |
-
"step": 305
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"clip_ratio": 0.0,
|
| 889 |
-
"completion_length": 657.6000173468339,
|
| 890 |
-
"epoch": 0.7853333333333333,
|
| 891 |
-
"grad_norm": 0.6775135397911072,
|
| 892 |
-
"kl": 0.7142669677734375,
|
| 893 |
-
"learning_rate": 3.978387737053994e-07,
|
| 894 |
-
"loss": 0.0583,
|
| 895 |
-
"reward": 0.6701754515108309,
|
| 896 |
-
"reward_std": 0.2519539353094603,
|
| 897 |
-
"rewards/accuracy_reward": 0.6701754515108309,
|
| 898 |
-
"rewards/format_reward": 0.0,
|
| 899 |
-
"step": 310
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"clip_ratio": 0.0,
|
| 903 |
-
"completion_length": 611.8982592130963,
|
| 904 |
-
"epoch": 0.798,
|
| 905 |
-
"grad_norm": 0.3379327952861786,
|
| 906 |
-
"kl": 0.28816158896998356,
|
| 907 |
-
"learning_rate": 3.5379058598286167e-07,
|
| 908 |
-
"loss": 0.0266,
|
| 909 |
-
"reward": 0.6771929922856783,
|
| 910 |
-
"reward_std": 0.22066064414225126,
|
| 911 |
-
"rewards/accuracy_reward": 0.6771929922856783,
|
| 912 |
-
"rewards/format_reward": 0.0,
|
| 913 |
-
"step": 315
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"clip_ratio": 0.0,
|
| 917 |
-
"completion_length": 576.1263292814556,
|
| 918 |
-
"epoch": 0.8106666666666666,
|
| 919 |
-
"grad_norm": 0.1547779142856598,
|
| 920 |
-
"kl": 0.282763671875,
|
| 921 |
-
"learning_rate": 3.119988494392894e-07,
|
| 922 |
-
"loss": 0.0458,
|
| 923 |
-
"reward": 0.7543859776697661,
|
| 924 |
-
"reward_std": 0.21713829605202925,
|
| 925 |
-
"rewards/accuracy_reward": 0.7543859776697661,
|
| 926 |
-
"rewards/format_reward": 0.0,
|
| 927 |
-
"step": 320
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"clip_ratio": 0.0,
|
| 931 |
-
"completion_length": 581.9789603785465,
|
| 932 |
-
"epoch": 0.8233333333333334,
|
| 933 |
-
"grad_norm": 1.6545398235321045,
|
| 934 |
-
"kl": 1.8802037289268092,
|
| 935 |
-
"learning_rate": 2.725458361377465e-07,
|
| 936 |
-
"loss": 0.0836,
|
| 937 |
-
"reward": 0.7596491349370856,
|
| 938 |
-
"reward_std": 0.21056411203585174,
|
| 939 |
-
"rewards/accuracy_reward": 0.7596491349370856,
|
| 940 |
-
"rewards/format_reward": 0.0,
|
| 941 |
-
"step": 325
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"clip_ratio": 0.0,
|
| 945 |
-
"completion_length": 624.1122974596525,
|
| 946 |
-
"epoch": 0.836,
|
| 947 |
-
"grad_norm": 0.04716293886303902,
|
| 948 |
-
"kl": 0.2931084883840461,
|
| 949 |
-
"learning_rate": 2.3550921408312737e-07,
|
| 950 |
-
"loss": 0.0158,
|
| 951 |
-
"reward": 0.6947368513596686,
|
| 952 |
-
"reward_std": 0.215224878411544,
|
| 953 |
-
"rewards/accuracy_reward": 0.6947368513596686,
|
| 954 |
-
"rewards/format_reward": 0.0,
|
| 955 |
-
"step": 330
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"clip_ratio": 0.0,
|
| 959 |
-
"completion_length": 604.3649275930304,
|
| 960 |
-
"epoch": 0.8486666666666667,
|
| 961 |
-
"grad_norm": 14.650071144104004,
|
| 962 |
-
"kl": 0.46654261538856906,
|
| 963 |
-
"learning_rate": 2.0096189432334195e-07,
|
| 964 |
-
"loss": 0.0406,
|
| 965 |
-
"reward": 0.7333333448359841,
|
| 966 |
-
"reward_std": 0.21933933402362624,
|
| 967 |
-
"rewards/accuracy_reward": 0.7333333448359841,
|
| 968 |
-
"rewards/format_reward": 0.0,
|
| 969 |
-
"step": 335
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"clip_ratio": 0.0,
|
| 973 |
-
"completion_length": 598.4035267076994,
|
| 974 |
-
"epoch": 0.8613333333333333,
|
| 975 |
-
"grad_norm": 0.0046224575489759445,
|
| 976 |
-
"kl": 0.14371498509457237,
|
| 977 |
-
"learning_rate": 1.6897188741514286e-07,
|
| 978 |
-
"loss": 0.029,
|
| 979 |
-
"reward": 0.701754395585311,
|
| 980 |
-
"reward_std": 0.1958320642772474,
|
| 981 |
-
"rewards/accuracy_reward": 0.701754395585311,
|
| 982 |
-
"rewards/format_reward": 0.0,
|
| 983 |
-
"step": 340
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"clip_ratio": 0.0,
|
| 987 |
-
"completion_length": 569.3929971795333,
|
| 988 |
-
"epoch": 0.874,
|
| 989 |
-
"grad_norm": 25.004467010498047,
|
| 990 |
-
"kl": 0.42442787571957236,
|
| 991 |
-
"learning_rate": 1.396021695371582e-07,
|
| 992 |
-
"loss": 0.0334,
|
| 993 |
-
"reward": 0.7789473804988359,
|
| 994 |
-
"reward_std": 0.22242484782871447,
|
| 995 |
-
"rewards/accuracy_reward": 0.7789473804988359,
|
| 996 |
-
"rewards/format_reward": 0.0,
|
| 997 |
-
"step": 345
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"clip_ratio": 0.0,
|
| 1001 |
-
"completion_length": 618.5912446674548,
|
| 1002 |
-
"epoch": 0.8866666666666667,
|
| 1003 |
-
"grad_norm": 12.571819305419922,
|
| 1004 |
-
"kl": 0.7060623972039474,
|
| 1005 |
-
"learning_rate": 1.1291055851370623e-07,
|
| 1006 |
-
"loss": 0.0507,
|
| 1007 |
-
"reward": 0.6964912398865348,
|
| 1008 |
-
"reward_std": 0.2621238843390816,
|
| 1009 |
-
"rewards/accuracy_reward": 0.6964912398865348,
|
| 1010 |
-
"rewards/format_reward": 0.0,
|
| 1011 |
-
"step": 350
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"clip_ratio": 0.0,
|
| 1015 |
-
"completion_length": 580.6228229723479,
|
| 1016 |
-
"epoch": 0.8993333333333333,
|
| 1017 |
-
"grad_norm": 0.06348630040884018,
|
| 1018 |
-
"kl": 0.45227805689761513,
|
| 1019 |
-
"learning_rate": 8.894959999345015e-08,
|
| 1020 |
-
"loss": 0.0257,
|
| 1021 |
-
"reward": 0.7157894854482851,
|
| 1022 |
-
"reward_std": 0.21632961285741706,
|
| 1023 |
-
"rewards/accuracy_reward": 0.7157894854482851,
|
| 1024 |
-
"rewards/format_reward": 0.0,
|
| 1025 |
-
"step": 355
|
| 1026 |
-
},
|
| 1027 |
-
{
|
| 1028 |
-
"clip_ratio": 0.0,
|
| 1029 |
-
"completion_length": 588.1842253032484,
|
| 1030 |
-
"epoch": 0.912,
|
| 1031 |
-
"grad_norm": 2.36680006980896,
|
| 1032 |
-
"kl": 0.27915199681332237,
|
| 1033 |
-
"learning_rate": 6.776646400696212e-08,
|
| 1034 |
-
"loss": 0.0277,
|
| 1035 |
-
"reward": 0.7140350958234385,
|
| 1036 |
-
"reward_std": 0.18397132754325868,
|
| 1037 |
-
"rewards/accuracy_reward": 0.7140350958234385,
|
| 1038 |
-
"rewards/format_reward": 0.0,
|
| 1039 |
-
"step": 360
|
| 1040 |
-
},
|
| 1041 |
-
{
|
| 1042 |
-
"clip_ratio": 0.0,
|
| 1043 |
-
"completion_length": 599.5210685328433,
|
| 1044 |
-
"epoch": 0.9246666666666666,
|
| 1045 |
-
"grad_norm": 5.9008965492248535,
|
| 1046 |
-
"kl": 0.44251114694695726,
|
| 1047 |
-
"learning_rate": 4.940285210684375e-08,
|
| 1048 |
-
"loss": 0.0327,
|
| 1049 |
-
"reward": 0.7438596629782727,
|
| 1050 |
-
"reward_std": 0.25195393279979106,
|
| 1051 |
-
"rewards/accuracy_reward": 0.7438596629782727,
|
| 1052 |
-
"rewards/format_reward": 0.0,
|
| 1053 |
-
"step": 365
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"clip_ratio": 0.0,
|
| 1057 |
-
"completion_length": 622.9333529823705,
|
| 1058 |
-
"epoch": 0.9373333333333334,
|
| 1059 |
-
"grad_norm": 0.37333959341049194,
|
| 1060 |
-
"kl": 0.4884661222759046,
|
| 1061 |
-
"learning_rate": 3.389491527319999e-08,
|
| 1062 |
-
"loss": 0.0296,
|
| 1063 |
-
"reward": 0.78070176604547,
|
| 1064 |
-
"reward_std": 0.22106379866600037,
|
| 1065 |
-
"rewards/accuracy_reward": 0.78070176604547,
|
| 1066 |
-
"rewards/format_reward": 0.0,
|
| 1067 |
-
"step": 370
|
| 1068 |
-
},
|
| 1069 |
-
{
|
| 1070 |
-
"clip_ratio": 0.0,
|
| 1071 |
-
"completion_length": 604.7368583277652,
|
| 1072 |
-
"epoch": 0.95,
|
| 1073 |
-
"grad_norm": 0.755801260471344,
|
| 1074 |
-
"kl": 0.6571317973889803,
|
| 1075 |
-
"learning_rate": 2.127318274608381e-08,
|
| 1076 |
-
"loss": 0.0439,
|
| 1077 |
-
"reward": 0.7228070301444907,
|
| 1078 |
-
"reward_std": 0.2571670673395458,
|
| 1079 |
-
"rewards/accuracy_reward": 0.7228070301444907,
|
| 1080 |
-
"rewards/format_reward": 0.0,
|
| 1081 |
-
"step": 375
|
| 1082 |
-
},
|
| 1083 |
-
{
|
| 1084 |
-
"clip_ratio": 0.0,
|
| 1085 |
-
"completion_length": 571.0789629484477,
|
| 1086 |
-
"epoch": 0.9626666666666667,
|
| 1087 |
-
"grad_norm": 5.0163397789001465,
|
| 1088 |
-
"kl": 1.0549049778988486,
|
| 1089 |
-
"learning_rate": 1.1562501925013125e-08,
|
| 1090 |
-
"loss": 0.0801,
|
| 1091 |
-
"reward": 0.7771929958933278,
|
| 1092 |
-
"reward_std": 0.213723046214957,
|
| 1093 |
-
"rewards/accuracy_reward": 0.7771929958933278,
|
| 1094 |
-
"rewards/format_reward": 0.0,
|
| 1095 |
-
"step": 380
|
| 1096 |
-
},
|
| 1097 |
-
{
|
| 1098 |
-
"clip_ratio": 0.0,
|
| 1099 |
-
"completion_length": 611.6526505319696,
|
| 1100 |
-
"epoch": 0.9753333333333334,
|
| 1101 |
-
"grad_norm": 0.42803478240966797,
|
| 1102 |
-
"kl": 0.45795432642886513,
|
| 1103 |
-
"learning_rate": 4.781989453874814e-09,
|
| 1104 |
-
"loss": 0.0535,
|
| 1105 |
-
"reward": 0.7070175552054455,
|
| 1106 |
-
"reward_std": 0.26146441823557803,
|
| 1107 |
-
"rewards/accuracy_reward": 0.7070175552054455,
|
| 1108 |
-
"rewards/format_reward": 0.0,
|
| 1109 |
-
"step": 385
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"clip_ratio": 0.0,
|
| 1113 |
-
"completion_length": 607.4087858501233,
|
| 1114 |
-
"epoch": 0.988,
|
| 1115 |
-
"grad_norm": 0.9341120719909668,
|
| 1116 |
-
"kl": 0.3894335295024671,
|
| 1117 |
-
"learning_rate": 9.44993587509657e-10,
|
| 1118 |
-
"loss": 0.0488,
|
| 1119 |
-
"reward": 0.6964912394159719,
|
| 1120 |
-
"reward_std": 0.2111501599612989,
|
| 1121 |
-
"rewards/accuracy_reward": 0.6964912394159719,
|
| 1122 |
-
"rewards/format_reward": 0.0,
|
| 1123 |
-
"step": 390
|
| 1124 |
-
},
|
| 1125 |
-
{
|
| 1126 |
-
"clip_ratio": 0.0,
|
| 1127 |
-
"completion_length": 596.2105405707108,
|
| 1128 |
-
"epoch": 0.9981333333333333,
|
| 1129 |
-
"kl": 0.275941949141653,
|
| 1130 |
-
"reward": 0.712719306937958,
|
| 1131 |
-
"reward_std": 0.2265615369144239,
|
| 1132 |
-
"rewards/accuracy_reward": 0.712719306937958,
|
| 1133 |
"rewards/format_reward": 0.0,
|
| 1134 |
-
"step":
|
| 1135 |
"total_flos": 0.0,
|
| 1136 |
-
"train_loss": 0.
|
| 1137 |
-
"train_runtime":
|
| 1138 |
-
"train_samples_per_second": 0.
|
| 1139 |
"train_steps_per_second": 0.001
|
| 1140 |
}
|
| 1141 |
],
|
| 1142 |
"logging_steps": 5,
|
| 1143 |
-
"max_steps":
|
| 1144 |
"num_input_tokens_seen": 0,
|
| 1145 |
"num_train_epochs": 1,
|
| 1146 |
"save_steps": 500,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9968,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 267,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"clip_ratio": 0.0,
|
| 13 |
+
"completion_length": 584.2053571428571,
|
| 14 |
+
"epoch": 0.0037333333333333333,
|
| 15 |
+
"grad_norm": 0.0,
|
| 16 |
+
"kl": 3.123922007424491e-07,
|
| 17 |
+
"learning_rate": 1.1111111111111111e-07,
|
| 18 |
+
"loss": -0.0051,
|
| 19 |
+
"reward": 0.6517857142857143,
|
| 20 |
+
"reward_std": 0.35586076974868774,
|
| 21 |
+
"rewards/accuracy_reward": 0.6517857142857143,
|
| 22 |
"rewards/format_reward": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"clip_ratio": 0.0,
|
| 27 |
+
"completion_length": 590.6540178571429,
|
| 28 |
+
"epoch": 0.018666666666666668,
|
| 29 |
+
"grad_norm": 0.1855999380350113,
|
| 30 |
+
"kl": 0.00017639675310679844,
|
| 31 |
+
"learning_rate": 5.555555555555555e-07,
|
| 32 |
+
"loss": -0.024,
|
| 33 |
+
"reward": 0.6004464285714286,
|
| 34 |
+
"reward_std": 0.30399111764771597,
|
| 35 |
+
"rewards/accuracy_reward": 0.6004464285714286,
|
| 36 |
"rewards/format_reward": 0.0,
|
| 37 |
"step": 5
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"clip_ratio": 0.0,
|
| 41 |
+
"completion_length": 577.3446428571428,
|
| 42 |
+
"epoch": 0.037333333333333336,
|
| 43 |
+
"grad_norm": 0.10475708544254303,
|
| 44 |
+
"kl": 0.00021860003471374512,
|
| 45 |
+
"learning_rate": 1.111111111111111e-06,
|
| 46 |
+
"loss": -0.0247,
|
| 47 |
+
"reward": 0.6017857142857143,
|
| 48 |
+
"reward_std": 0.32809826050485885,
|
| 49 |
+
"rewards/accuracy_reward": 0.6017857142857143,
|
| 50 |
"rewards/format_reward": 0.0,
|
| 51 |
"step": 10
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"clip_ratio": 0.0,
|
| 55 |
+
"completion_length": 635.5607142857143,
|
| 56 |
+
"epoch": 0.056,
|
| 57 |
+
"grad_norm": 0.022834666073322296,
|
| 58 |
+
"kl": 0.00047294582639421733,
|
| 59 |
+
"learning_rate": 1.6666666666666669e-06,
|
| 60 |
+
"loss": -0.0105,
|
| 61 |
+
"reward": 0.5839285714285715,
|
| 62 |
+
"reward_std": 0.32206040705953326,
|
| 63 |
+
"rewards/accuracy_reward": 0.5839285714285715,
|
| 64 |
"rewards/format_reward": 0.0,
|
| 65 |
"step": 15
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"clip_ratio": 0.0,
|
| 69 |
+
"completion_length": 608.0660714285714,
|
| 70 |
+
"epoch": 0.07466666666666667,
|
| 71 |
+
"grad_norm": 1.4037145376205444,
|
| 72 |
+
"kl": 0.003454787390572684,
|
| 73 |
+
"learning_rate": 2.222222222222222e-06,
|
| 74 |
+
"loss": -0.0408,
|
| 75 |
+
"reward": 0.5660714285714286,
|
| 76 |
+
"reward_std": 0.30615685411861965,
|
| 77 |
+
"rewards/accuracy_reward": 0.5660714285714286,
|
| 78 |
"rewards/format_reward": 0.0,
|
| 79 |
"step": 20
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"clip_ratio": 0.0,
|
| 83 |
+
"completion_length": 585.6625,
|
| 84 |
+
"epoch": 0.09333333333333334,
|
| 85 |
+
"grad_norm": 0.04715263098478317,
|
| 86 |
+
"kl": 0.006245636940002441,
|
| 87 |
+
"learning_rate": 2.777777777777778e-06,
|
| 88 |
+
"loss": -0.0257,
|
| 89 |
+
"reward": 0.5785714285714286,
|
| 90 |
+
"reward_std": 0.3421275573117392,
|
| 91 |
+
"rewards/accuracy_reward": 0.5785714285714286,
|
| 92 |
"rewards/format_reward": 0.0,
|
| 93 |
"step": 25
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"clip_ratio": 0.0,
|
| 97 |
+
"completion_length": 592.3142857142857,
|
| 98 |
+
"epoch": 0.112,
|
| 99 |
+
"grad_norm": 0.022860102355480194,
|
| 100 |
+
"kl": 0.3629974705832345,
|
| 101 |
+
"learning_rate": 2.9988435543610844e-06,
|
| 102 |
+
"loss": -0.0083,
|
| 103 |
+
"reward": 0.6392857142857142,
|
| 104 |
+
"reward_std": 0.30530826789992194,
|
| 105 |
+
"rewards/accuracy_reward": 0.6392857142857142,
|
| 106 |
"rewards/format_reward": 0.0,
|
| 107 |
"step": 30
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"clip_ratio": 0.0,
|
| 111 |
+
"completion_length": 606.3267857142857,
|
| 112 |
+
"epoch": 0.13066666666666665,
|
| 113 |
+
"grad_norm": 0.11820275336503983,
|
| 114 |
+
"kl": 6.243857077189854,
|
| 115 |
+
"learning_rate": 2.99178284305241e-06,
|
| 116 |
+
"loss": 0.3173,
|
| 117 |
+
"reward": 0.5857142857142857,
|
| 118 |
+
"reward_std": 0.3432325610092708,
|
| 119 |
+
"rewards/accuracy_reward": 0.5857142857142857,
|
| 120 |
"rewards/format_reward": 0.0,
|
| 121 |
"step": 35
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"clip_ratio": 0.0,
|
| 125 |
+
"completion_length": 572.0892857142857,
|
| 126 |
+
"epoch": 0.14933333333333335,
|
| 127 |
+
"grad_norm": 21.97744369506836,
|
| 128 |
+
"kl": 0.33153715814862933,
|
| 129 |
+
"learning_rate": 2.978334088587117e-06,
|
| 130 |
+
"loss": -0.0104,
|
| 131 |
+
"reward": 0.6625,
|
| 132 |
+
"reward_std": 0.33634612134524755,
|
| 133 |
+
"rewards/accuracy_reward": 0.6607142857142857,
|
| 134 |
+
"rewards/format_reward": 0.0017857142857142857,
|
| 135 |
"step": 40
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"clip_ratio": 0.0,
|
| 139 |
+
"completion_length": 600.6875,
|
| 140 |
+
"epoch": 0.168,
|
| 141 |
+
"grad_norm": 0.17174075543880463,
|
| 142 |
+
"kl": 0.3202719347817557,
|
| 143 |
+
"learning_rate": 2.958554880596515e-06,
|
| 144 |
+
"loss": 0.0072,
|
| 145 |
+
"reward": 0.6,
|
| 146 |
+
"reward_std": 0.3338796964713505,
|
| 147 |
+
"rewards/accuracy_reward": 0.6,
|
| 148 |
"rewards/format_reward": 0.0,
|
| 149 |
"step": 45
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"clip_ratio": 0.0,
|
| 153 |
+
"completion_length": 616.7535714285714,
|
| 154 |
+
"epoch": 0.18666666666666668,
|
| 155 |
+
"grad_norm": 0.09551213681697845,
|
| 156 |
+
"kl": 0.33372737339564734,
|
| 157 |
+
"learning_rate": 2.9325299166857803e-06,
|
| 158 |
+
"loss": 0.0081,
|
| 159 |
+
"reward": 0.6071428571428571,
|
| 160 |
+
"reward_std": 0.3335836121014186,
|
| 161 |
+
"rewards/accuracy_reward": 0.6071428571428571,
|
| 162 |
"rewards/format_reward": 0.0,
|
| 163 |
"step": 50
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"clip_ratio": 0.0,
|
| 167 |
+
"completion_length": 564.6660714285714,
|
| 168 |
+
"epoch": 0.20533333333333334,
|
| 169 |
+
"grad_norm": 0.12027700990438461,
|
| 170 |
+
"kl": 0.20207279750279017,
|
| 171 |
+
"learning_rate": 2.9003706397458025e-06,
|
| 172 |
+
"loss": 0.0048,
|
| 173 |
+
"reward": 0.6446428571428572,
|
| 174 |
+
"reward_std": 0.3487179126058306,
|
| 175 |
+
"rewards/accuracy_reward": 0.6446428571428572,
|
| 176 |
"rewards/format_reward": 0.0,
|
| 177 |
"step": 55
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"clip_ratio": 0.0,
|
| 181 |
+
"completion_length": 576.2642857142857,
|
| 182 |
+
"epoch": 0.224,
|
| 183 |
+
"grad_norm": 0.594131588935852,
|
| 184 |
+
"kl": 0.24056614467075893,
|
| 185 |
+
"learning_rate": 2.862214760737622e-06,
|
| 186 |
+
"loss": -0.0414,
|
| 187 |
+
"reward": 0.6232142857142857,
|
| 188 |
+
"reward_std": 0.33634612134524755,
|
| 189 |
+
"rewards/accuracy_reward": 0.6232142857142857,
|
| 190 |
"rewards/format_reward": 0.0,
|
| 191 |
"step": 60
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"clip_ratio": 0.0,
|
| 195 |
+
"completion_length": 578.2357142857143,
|
| 196 |
+
"epoch": 0.24266666666666667,
|
| 197 |
+
"grad_norm": 0.3320864140987396,
|
| 198 |
+
"kl": 0.17519269670758927,
|
| 199 |
+
"learning_rate": 2.818225668992948e-06,
|
| 200 |
+
"loss": -0.0211,
|
| 201 |
+
"reward": 0.6125,
|
| 202 |
+
"reward_std": 0.32176432268960137,
|
| 203 |
+
"rewards/accuracy_reward": 0.6125,
|
| 204 |
"rewards/format_reward": 0.0,
|
| 205 |
"step": 65
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"clip_ratio": 0.0,
|
| 209 |
+
"completion_length": 569.9196428571429,
|
| 210 |
+
"epoch": 0.2613333333333333,
|
| 211 |
+
"grad_norm": 0.42868152260780334,
|
| 212 |
+
"kl": 0.07821709769112724,
|
| 213 |
+
"learning_rate": 2.7685917325559604e-06,
|
| 214 |
+
"loss": -0.0073,
|
| 215 |
+
"reward": 0.6,
|
| 216 |
+
"reward_std": 0.35862327899251667,
|
| 217 |
+
"rewards/accuracy_reward": 0.6,
|
| 218 |
"rewards/format_reward": 0.0,
|
| 219 |
"step": 70
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"clip_ratio": 0.0,
|
| 223 |
+
"completion_length": 592.0357142857143,
|
| 224 |
+
"epoch": 0.28,
|
| 225 |
+
"grad_norm": 0.0943068340420723,
|
| 226 |
+
"kl": 0.06971386500767299,
|
| 227 |
+
"learning_rate": 2.713525491562421e-06,
|
| 228 |
+
"loss": -0.0088,
|
| 229 |
+
"reward": 0.6053571428571428,
|
| 230 |
+
"reward_std": 0.32537541815212795,
|
| 231 |
+
"rewards/accuracy_reward": 0.6053571428571428,
|
| 232 |
"rewards/format_reward": 0.0,
|
| 233 |
"step": 75
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"clip_ratio": 0.0,
|
| 237 |
+
"completion_length": 578.4678571428572,
|
| 238 |
+
"epoch": 0.2986666666666667,
|
| 239 |
+
"grad_norm": 0.011618987657129765,
|
| 240 |
+
"kl": 0.08212520054408483,
|
| 241 |
+
"learning_rate": 2.6532627481101893e-06,
|
| 242 |
+
"loss": -0.0353,
|
| 243 |
+
"reward": 0.6267857142857143,
|
| 244 |
+
"reward_std": 0.2574786084038871,
|
| 245 |
+
"rewards/accuracy_reward": 0.6267857142857143,
|
| 246 |
"rewards/format_reward": 0.0,
|
| 247 |
"step": 80
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"clip_ratio": 0.0,
|
| 251 |
+
"completion_length": 622.4571428571429,
|
| 252 |
+
"epoch": 0.31733333333333336,
|
| 253 |
+
"grad_norm": 0.7904458045959473,
|
| 254 |
+
"kl": 0.05238505772181919,
|
| 255 |
+
"learning_rate": 2.5880615565184313e-06,
|
| 256 |
+
"loss": -0.0003,
|
| 257 |
+
"reward": 0.6267857142857143,
|
| 258 |
+
"reward_std": 0.2849450332777841,
|
| 259 |
+
"rewards/accuracy_reward": 0.6267857142857143,
|
| 260 |
"rewards/format_reward": 0.0,
|
| 261 |
"step": 85
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"clip_ratio": 0.0,
|
| 265 |
+
"completion_length": 557.3285714285714,
|
| 266 |
+
"epoch": 0.336,
|
| 267 |
+
"grad_norm": 0.15280544757843018,
|
| 268 |
+
"kl": 0.17704980032784598,
|
| 269 |
+
"learning_rate": 2.518201118299413e-06,
|
| 270 |
+
"loss": 0.0103,
|
| 271 |
+
"reward": 0.6214285714285714,
|
| 272 |
+
"reward_std": 0.3481654107570648,
|
| 273 |
+
"rewards/accuracy_reward": 0.6214285714285714,
|
| 274 |
"rewards/format_reward": 0.0,
|
| 275 |
"step": 90
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"clip_ratio": 0.0,
|
| 279 |
+
"completion_length": 613.6553571428572,
|
| 280 |
+
"epoch": 0.3546666666666667,
|
| 281 |
+
"grad_norm": 0.21528230607509613,
|
| 282 |
+
"kl": 0.09126150948660715,
|
| 283 |
+
"learning_rate": 2.4439805865747562e-06,
|
| 284 |
+
"loss": -0.0161,
|
| 285 |
+
"reward": 0.6196428571428572,
|
| 286 |
+
"reward_std": 0.3066696890762874,
|
| 287 |
+
"rewards/accuracy_reward": 0.6196428571428572,
|
| 288 |
"rewards/format_reward": 0.0,
|
| 289 |
"step": 95
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"clip_ratio": 0.0,
|
| 293 |
+
"completion_length": 579.3107142857143,
|
| 294 |
+
"epoch": 0.37333333333333335,
|
| 295 |
+
"grad_norm": 0.20065245032310486,
|
| 296 |
+
"kl": 0.17014105660574777,
|
| 297 |
+
"learning_rate": 2.3657177850558505e-06,
|
| 298 |
+
"loss": -0.0211,
|
| 299 |
+
"reward": 0.6392857142857142,
|
| 300 |
+
"reward_std": 0.3195939821856362,
|
| 301 |
+
"rewards/accuracy_reward": 0.6392857142857142,
|
| 302 |
"rewards/format_reward": 0.0,
|
| 303 |
"step": 100
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"clip_ratio": 0.0,
|
| 307 |
+
"completion_length": 617.9732142857143,
|
| 308 |
+
"epoch": 0.392,
|
| 309 |
+
"grad_norm": 0.010423385538160801,
|
| 310 |
+
"kl": 0.06536189488002232,
|
| 311 |
+
"learning_rate": 2.2837478470739234e-06,
|
| 312 |
+
"loss": -0.0072,
|
| 313 |
+
"reward": 0.5732142857142857,
|
| 314 |
+
"reward_std": 0.34157505546297345,
|
| 315 |
+
"rewards/accuracy_reward": 0.5732142857142857,
|
| 316 |
"rewards/format_reward": 0.0,
|
| 317 |
"step": 105
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"clip_ratio": 0.0,
|
| 321 |
+
"completion_length": 587.2910714285714,
|
| 322 |
+
"epoch": 0.4106666666666667,
|
| 323 |
+
"grad_norm": 0.14102862775325775,
|
| 324 |
+
"kl": 0.1574784415108817,
|
| 325 |
+
"learning_rate": 2.198421780487667e-06,
|
| 326 |
+
"loss": -0.0464,
|
| 327 |
+
"reward": 0.6053571428571428,
|
| 328 |
+
"reward_std": 0.3215475721018655,
|
| 329 |
+
"rewards/accuracy_reward": 0.6053571428571428,
|
| 330 |
"rewards/format_reward": 0.0,
|
| 331 |
"step": 110
|
| 332 |
},
|
| 333 |
{
|
| 334 |
"clip_ratio": 0.0,
|
| 335 |
+
"completion_length": 592.3178571428572,
|
| 336 |
+
"epoch": 0.42933333333333334,
|
| 337 |
+
"grad_norm": 0.10852164775133133,
|
| 338 |
+
"kl": 0.11443628583635602,
|
| 339 |
+
"learning_rate": 2.1101049646137005e-06,
|
| 340 |
+
"loss": -0.0282,
|
| 341 |
+
"reward": 0.5714285714285714,
|
| 342 |
+
"reward_std": 0.3459554033620017,
|
| 343 |
+
"rewards/accuracy_reward": 0.5714285714285714,
|
| 344 |
"rewards/format_reward": 0.0,
|
| 345 |
"step": 115
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"clip_ratio": 0.0,
|
| 349 |
+
"completion_length": 618.7821428571428,
|
| 350 |
+
"epoch": 0.448,
|
| 351 |
+
"grad_norm": 0.6055614948272705,
|
| 352 |
+
"kl": 0.16327661786760603,
|
| 353 |
+
"learning_rate": 2.0191755856162397e-06,
|
| 354 |
+
"loss": -0.0015,
|
| 355 |
+
"reward": 0.6017857142857143,
|
| 356 |
+
"reward_std": 0.30254575865609307,
|
| 357 |
+
"rewards/accuracy_reward": 0.6017857142857143,
|
| 358 |
"rewards/format_reward": 0.0,
|
| 359 |
"step": 120
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"clip_ratio": 0.0,
|
| 363 |
+
"completion_length": 586.575,
|
| 364 |
+
"epoch": 0.4666666666666667,
|
| 365 |
+
"grad_norm": 0.33487093448638916,
|
| 366 |
+
"kl": 0.2866717747279576,
|
| 367 |
+
"learning_rate": 1.9260230170558845e-06,
|
| 368 |
+
"loss": -0.0131,
|
| 369 |
+
"reward": 0.5660714285714286,
|
| 370 |
+
"reward_std": 0.34650790521076746,
|
| 371 |
+
"rewards/accuracy_reward": 0.5660714285714286,
|
| 372 |
"rewards/format_reward": 0.0,
|
| 373 |
"step": 125
|
| 374 |
},
|
| 375 |
{
|
| 376 |
"clip_ratio": 0.0,
|
| 377 |
+
"completion_length": 598.3553571428571,
|
| 378 |
+
"epoch": 0.48533333333333334,
|
| 379 |
+
"grad_norm": 0.4094862639904022,
|
| 380 |
+
"kl": 0.2726052965436663,
|
| 381 |
+
"learning_rate": 1.8310461525322523e-06,
|
| 382 |
+
"loss": -0.0308,
|
| 383 |
+
"reward": 0.6160714285714286,
|
| 384 |
+
"reward_std": 0.34790899327823094,
|
| 385 |
+
"rewards/accuracy_reward": 0.6160714285714286,
|
| 386 |
"rewards/format_reward": 0.0,
|
| 387 |
"step": 130
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"clip_ratio": 0.0,
|
| 391 |
+
"completion_length": 615.3178571428572,
|
| 392 |
+
"epoch": 0.504,
|
| 393 |
+
"grad_norm": 0.5523812770843506,
|
| 394 |
+
"kl": 0.39636241367885044,
|
| 395 |
+
"learning_rate": 1.7346516975603465e-06,
|
| 396 |
+
"loss": -0.0353,
|
| 397 |
+
"reward": 0.5767857142857142,
|
| 398 |
+
"reward_std": 0.33332719462258475,
|
| 399 |
+
"rewards/accuracy_reward": 0.5767857142857142,
|
| 400 |
"rewards/format_reward": 0.0,
|
| 401 |
"step": 135
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"clip_ratio": 0.0,
|
| 405 |
+
"completion_length": 626.4767857142857,
|
| 406 |
+
"epoch": 0.5226666666666666,
|
| 407 |
+
"grad_norm": 0.08616114407777786,
|
| 408 |
+
"kl": 0.4847276960100446,
|
| 409 |
+
"learning_rate": 1.637252427995104e-06,
|
| 410 |
+
"loss": -0.0036,
|
| 411 |
+
"reward": 0.5392857142857143,
|
| 412 |
+
"reward_std": 0.30309826050485883,
|
| 413 |
+
"rewards/accuracy_reward": 0.5392857142857143,
|
| 414 |
"rewards/format_reward": 0.0,
|
| 415 |
"step": 140
|
| 416 |
},
|
| 417 |
{
|
| 418 |
"clip_ratio": 0.0,
|
| 419 |
+
"completion_length": 605.8607142857143,
|
| 420 |
+
"epoch": 0.5413333333333333,
|
| 421 |
+
"grad_norm": 0.35899537801742554,
|
| 422 |
+
"kl": 1.201324244907924,
|
| 423 |
+
"learning_rate": 1.53926542246181e-06,
|
| 424 |
+
"loss": -0.0055,
|
| 425 |
+
"reward": 0.5928571428571429,
|
| 426 |
+
"reward_std": 0.3091361139501844,
|
| 427 |
+
"rewards/accuracy_reward": 0.5928571428571429,
|
| 428 |
"rewards/format_reward": 0.0,
|
| 429 |
"step": 145
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"clip_ratio": 0.0,
|
| 433 |
+
"completion_length": 596.8107142857143,
|
| 434 |
+
"epoch": 0.56,
|
| 435 |
+
"grad_norm": 0.16239966452121735,
|
| 436 |
+
"kl": 0.4776475633893694,
|
| 437 |
+
"learning_rate": 1.4411102763613975e-06,
|
| 438 |
+
"loss": -0.0096,
|
| 439 |
+
"reward": 0.5910714285714286,
|
| 440 |
+
"reward_std": 0.36189862319401334,
|
| 441 |
+
"rewards/accuracy_reward": 0.5910714285714286,
|
| 442 |
"rewards/format_reward": 0.0,
|
| 443 |
"step": 150
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"clip_ratio": 0.0,
|
| 447 |
+
"completion_length": 579.8357142857143,
|
| 448 |
+
"epoch": 0.5786666666666667,
|
| 449 |
+
"grad_norm": 0.2588573098182678,
|
| 450 |
+
"kl": 0.37055963788713725,
|
| 451 |
+
"learning_rate": 1.3432073050985201e-06,
|
| 452 |
+
"loss": -0.0182,
|
| 453 |
+
"reward": 0.5892857142857143,
|
| 454 |
+
"reward_std": 0.3256318356309618,
|
| 455 |
+
"rewards/accuracy_reward": 0.5892857142857143,
|
| 456 |
"rewards/format_reward": 0.0,
|
| 457 |
"step": 155
|
| 458 |
},
|
| 459 |
{
|
| 460 |
"clip_ratio": 0.0,
|
| 461 |
+
"completion_length": 636.9660714285715,
|
| 462 |
+
"epoch": 0.5973333333333334,
|
| 463 |
+
"grad_norm": 0.3046307861804962,
|
| 464 |
+
"kl": 0.18770294189453124,
|
| 465 |
+
"learning_rate": 1.245975744226463e-06,
|
| 466 |
+
"loss": -0.0251,
|
| 467 |
+
"reward": 0.6071428571428571,
|
| 468 |
+
"reward_std": 0.32452683193343024,
|
| 469 |
+
"rewards/accuracy_reward": 0.6071428571428571,
|
| 470 |
"rewards/format_reward": 0.0,
|
| 471 |
"step": 160
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"clip_ratio": 0.0,
|
| 475 |
+
"completion_length": 591.6803571428571,
|
| 476 |
+
"epoch": 0.616,
|
| 477 |
+
"grad_norm": 0.19371181726455688,
|
| 478 |
+
"kl": 0.3922314235142299,
|
| 479 |
+
"learning_rate": 1.1498319542161423e-06,
|
| 480 |
+
"loss": -0.0093,
|
| 481 |
+
"reward": 0.5875,
|
| 482 |
+
"reward_std": 0.3490139969757625,
|
| 483 |
+
"rewards/accuracy_reward": 0.5875,
|
| 484 |
"rewards/format_reward": 0.0,
|
| 485 |
"step": 165
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"clip_ratio": 0.0,
|
| 489 |
+
"completion_length": 584.7196428571428,
|
| 490 |
+
"epoch": 0.6346666666666667,
|
| 491 |
+
"grad_norm": 0.10712730884552002,
|
| 492 |
+
"kl": 0.3678185599190848,
|
| 493 |
+
"learning_rate": 1.0551876375366437e-06,
|
| 494 |
+
"loss": -0.0021,
|
| 495 |
+
"reward": 0.5875,
|
| 496 |
+
"reward_std": 0.33443219832011634,
|
| 497 |
+
"rewards/accuracy_reward": 0.5875,
|
| 498 |
"rewards/format_reward": 0.0,
|
| 499 |
"step": 170
|
| 500 |
},
|
| 501 |
{
|
| 502 |
"clip_ratio": 0.0,
|
| 503 |
+
"completion_length": 596.0017857142857,
|
| 504 |
+
"epoch": 0.6533333333333333,
|
| 505 |
+
"grad_norm": 1.3254014253616333,
|
| 506 |
+
"kl": 0.6625612531389509,
|
| 507 |
+
"learning_rate": 9.624480756820497e-07,
|
| 508 |
+
"loss": 0.0012,
|
| 509 |
+
"reward": 0.625,
|
| 510 |
+
"reward_std": 0.3338796964713505,
|
| 511 |
+
"rewards/accuracy_reward": 0.625,
|
| 512 |
"rewards/format_reward": 0.0,
|
| 513 |
"step": 175
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"clip_ratio": 0.0,
|
| 517 |
+
"completion_length": 586.9053571428572,
|
| 518 |
+
"epoch": 0.672,
|
| 519 |
+
"grad_norm": 5.647360324859619,
|
| 520 |
+
"kl": 1.4718305315290179,
|
| 521 |
+
"learning_rate": 8.720103936938583e-07,
|
| 522 |
+
"loss": 0.0312,
|
| 523 |
+
"reward": 0.5928571428571429,
|
| 524 |
+
"reward_std": 0.3011843374797276,
|
| 525 |
+
"rewards/accuracy_reward": 0.5928571428571429,
|
| 526 |
"rewards/format_reward": 0.0,
|
| 527 |
"step": 180
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"clip_ratio": 0.0,
|
| 531 |
+
"completion_length": 624.6,
|
| 532 |
+
"epoch": 0.6906666666666667,
|
| 533 |
+
"grad_norm": 0.3304588198661804,
|
| 534 |
+
"kl": 1.4476038251604353,
|
| 535 |
+
"learning_rate": 7.842618596105873e-07,
|
| 536 |
+
"loss": -0.0088,
|
| 537 |
+
"reward": 0.5571428571428572,
|
| 538 |
+
"reward_std": 0.3657661361353738,
|
| 539 |
+
"rewards/accuracy_reward": 0.5571428571428572,
|
| 540 |
"rewards/format_reward": 0.0,
|
| 541 |
"step": 185
|
| 542 |
},
|
| 543 |
{
|
| 544 |
"clip_ratio": 0.0,
|
| 545 |
+
"completion_length": 609.3267857142857,
|
| 546 |
+
"epoch": 0.7093333333333334,
|
| 547 |
+
"grad_norm": 0.0052251736633479595,
|
| 548 |
+
"kl": 0.7994631086077009,
|
| 549 |
+
"learning_rate": 6.995782261265828e-07,
|
| 550 |
+
"loss": -0.0033,
|
| 551 |
+
"reward": 0.6571428571428571,
|
| 552 |
+
"reward_std": 0.3195939821856362,
|
| 553 |
+
"rewards/accuracy_reward": 0.6571428571428571,
|
| 554 |
"rewards/format_reward": 0.0,
|
| 555 |
"step": 190
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"clip_ratio": 0.0,
|
| 559 |
+
"completion_length": 627.0696428571429,
|
| 560 |
+
"epoch": 0.728,
|
| 561 |
+
"grad_norm": 0.8703639507293701,
|
| 562 |
+
"kl": 0.4258202144077846,
|
| 563 |
+
"learning_rate": 6.183221215612905e-07,
|
| 564 |
+
"loss": -0.0207,
|
| 565 |
+
"reward": 0.5607142857142857,
|
| 566 |
+
"reward_std": 0.3360897038664137,
|
| 567 |
+
"rewards/accuracy_reward": 0.5607142857142857,
|
| 568 |
"rewards/format_reward": 0.0,
|
| 569 |
"step": 195
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 0.7466666666666667,
|
| 573 |
+
"grad_norm": 0.2712614834308624,
|
| 574 |
+
"learning_rate": 5.40841497029123e-07,
|
| 575 |
+
"loss": 0.0225,
|
| 576 |
"step": 200
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"epoch": 0.7466666666666667,
|
| 580 |
"eval_clip_ratio": 0.0,
|
| 581 |
+
"eval_completion_length": 581.896,
|
| 582 |
+
"eval_kl": 0.6586514877319336,
|
| 583 |
+
"eval_loss": -0.014948751777410507,
|
| 584 |
+
"eval_reward": 0.49185,
|
| 585 |
+
"eval_reward_std": 0.3328778264760971,
|
| 586 |
+
"eval_rewards/accuracy_reward": 0.49185,
|
| 587 |
"eval_rewards/format_reward": 0.0,
|
| 588 |
+
"eval_runtime": 80126.5226,
|
| 589 |
+
"eval_samples_per_second": 0.062,
|
| 590 |
+
"eval_steps_per_second": 0.016,
|
| 591 |
"step": 200
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"clip_ratio": 0.0,
|
| 595 |
+
"completion_length": 596.6607142857143,
|
| 596 |
+
"epoch": 0.7653333333333333,
|
| 597 |
+
"grad_norm": 0.17807097733020782,
|
| 598 |
+
"kl": 0.6059337615966797,
|
| 599 |
+
"learning_rate": 4.674681364593688e-07,
|
| 600 |
+
"loss": 0.0047,
|
| 601 |
+
"reward": 0.5991071428571428,
|
| 602 |
+
"reward_std": 0.30915594739573343,
|
| 603 |
+
"rewards/accuracy_reward": 0.5991071428571428,
|
| 604 |
"rewards/format_reward": 0.0,
|
| 605 |
"step": 205
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"clip_ratio": 0.0,
|
| 609 |
+
"completion_length": 645.5696428571429,
|
| 610 |
+
"epoch": 0.784,
|
| 611 |
+
"grad_norm": 1.1000945568084717,
|
| 612 |
+
"kl": 0.31501290457589287,
|
| 613 |
+
"learning_rate": 3.98516235846472e-07,
|
| 614 |
+
"loss": 0.0128,
|
| 615 |
+
"reward": 0.6053571428571428,
|
| 616 |
+
"reward_std": 0.3121470332145691,
|
| 617 |
+
"rewards/accuracy_reward": 0.6035714285714285,
|
| 618 |
+
"rewards/format_reward": 0.0017857142857142857,
|
| 619 |
"step": 210
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"clip_ratio": 0.0,
|
| 623 |
+
"completion_length": 614.0910714285715,
|
| 624 |
+
"epoch": 0.8026666666666666,
|
| 625 |
+
"grad_norm": 0.3984079360961914,
|
| 626 |
+
"kl": 0.5104325430733817,
|
| 627 |
+
"learning_rate": 3.3428105781454364e-07,
|
| 628 |
+
"loss": 0.0147,
|
| 629 |
+
"reward": 0.575,
|
| 630 |
+
"reward_std": 0.3561171872275216,
|
| 631 |
+
"rewards/accuracy_reward": 0.575,
|
| 632 |
"rewards/format_reward": 0.0,
|
| 633 |
"step": 215
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"clip_ratio": 0.0,
|
| 637 |
+
"completion_length": 577.9607142857143,
|
| 638 |
+
"epoch": 0.8213333333333334,
|
| 639 |
+
"grad_norm": 0.9381574392318726,
|
| 640 |
+
"kl": 0.5813296726771764,
|
| 641 |
+
"learning_rate": 2.750376672574816e-07,
|
| 642 |
+
"loss": -0.0266,
|
| 643 |
+
"reward": 0.6232142857142857,
|
| 644 |
+
"reward_std": 0.32235649142946515,
|
| 645 |
+
"rewards/accuracy_reward": 0.6232142857142857,
|
| 646 |
"rewards/format_reward": 0.0,
|
| 647 |
"step": 220
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"clip_ratio": 0.0,
|
| 651 |
+
"completion_length": 625.3107142857143,
|
| 652 |
+
"epoch": 0.84,
|
| 653 |
+
"grad_norm": 0.2192346155643463,
|
| 654 |
+
"kl": 0.5947923932756697,
|
| 655 |
+
"learning_rate": 2.2103975346886175e-07,
|
| 656 |
+
"loss": 0.0184,
|
| 657 |
+
"reward": 0.6017857142857143,
|
| 658 |
+
"reward_std": 0.3096886157989502,
|
| 659 |
+
"rewards/accuracy_reward": 0.6017857142857143,
|
| 660 |
"rewards/format_reward": 0.0,
|
| 661 |
"step": 225
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"clip_ratio": 0.0,
|
| 665 |
+
"completion_length": 609.7071428571429,
|
| 666 |
+
"epoch": 0.8586666666666667,
|
| 667 |
+
"grad_norm": 1.8841651678085327,
|
| 668 |
+
"kl": 0.9176664079938616,
|
| 669 |
+
"learning_rate": 1.7251854380543735e-07,
|
| 670 |
+
"loss": -0.0191,
|
| 671 |
+
"reward": 0.5982142857142857,
|
| 672 |
+
"reward_std": 0.3261843374797276,
|
| 673 |
+
"rewards/accuracy_reward": 0.5982142857142857,
|
| 674 |
"rewards/format_reward": 0.0,
|
| 675 |
"step": 230
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"clip_ratio": 0.0,
|
| 679 |
+
"completion_length": 614.4732142857143,
|
| 680 |
+
"epoch": 0.8773333333333333,
|
| 681 |
+
"grad_norm": 0.18197353184223175,
|
| 682 |
+
"kl": 0.8367650713239397,
|
| 683 |
+
"learning_rate": 1.2968181353609853e-07,
|
| 684 |
+
"loss": 0.0089,
|
| 685 |
+
"reward": 0.6214285714285714,
|
| 686 |
+
"reward_std": 0.28277469277381895,
|
| 687 |
+
"rewards/accuracy_reward": 0.6214285714285714,
|
| 688 |
"rewards/format_reward": 0.0,
|
| 689 |
"step": 235
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"clip_ratio": 0.0,
|
| 693 |
+
"completion_length": 587.1607142857143,
|
| 694 |
+
"epoch": 0.896,
|
| 695 |
+
"grad_norm": 0.9800770282745361,
|
| 696 |
+
"kl": 1.4477369035993304,
|
| 697 |
+
"learning_rate": 9.271299611627392e-08,
|
| 698 |
+
"loss": -0.0131,
|
| 699 |
+
"reward": 0.6142857142857143,
|
| 700 |
+
"reward_std": 0.3237179126058306,
|
| 701 |
+
"rewards/accuracy_reward": 0.6142857142857143,
|
| 702 |
"rewards/format_reward": 0.0,
|
| 703 |
"step": 240
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"clip_ratio": 0.0,
|
| 707 |
+
"completion_length": 598.5571428571428,
|
| 708 |
+
"epoch": 0.9146666666666666,
|
| 709 |
+
"grad_norm": 1.9400585889816284,
|
| 710 |
+
"kl": 1.4993116106305804,
|
| 711 |
+
"learning_rate": 6.177039769771042e-08,
|
| 712 |
+
"loss": -0.0213,
|
| 713 |
+
"reward": 0.5803571428571429,
|
| 714 |
+
"reward_std": 0.36108970386641365,
|
| 715 |
+
"rewards/accuracy_reward": 0.5803571428571429,
|
| 716 |
"rewards/format_reward": 0.0,
|
| 717 |
"step": 245
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"clip_ratio": 0.0,
|
| 721 |
+
"completion_length": 625.3357142857143,
|
| 722 |
+
"epoch": 0.9333333333333333,
|
| 723 |
+
"grad_norm": 1.5598750114440918,
|
| 724 |
+
"kl": 0.9994749886648996,
|
| 725 |
+
"learning_rate": 3.698651923723101e-08,
|
| 726 |
+
"loss": -0.0048,
|
| 727 |
+
"reward": 0.6071428571428571,
|
| 728 |
+
"reward_std": 0.39846149512699675,
|
| 729 |
+
"rewards/accuracy_reward": 0.6071428571428571,
|
| 730 |
"rewards/format_reward": 0.0,
|
| 731 |
"step": 250
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"clip_ratio": 0.0,
|
| 735 |
+
"completion_length": 589.4857142857143,
|
| 736 |
+
"epoch": 0.952,
|
| 737 |
+
"grad_norm": 1.0319033861160278,
|
| 738 |
+
"kl": 1.174580601283482,
|
| 739 |
+
"learning_rate": 1.846748910729351e-08,
|
| 740 |
+
"loss": -0.0229,
|
| 741 |
+
"reward": 0.6053571428571428,
|
| 742 |
+
"reward_std": 0.3490139969757625,
|
| 743 |
+
"rewards/accuracy_reward": 0.6053571428571428,
|
| 744 |
"rewards/format_reward": 0.0,
|
| 745 |
"step": 255
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"clip_ratio": 0.0,
|
| 749 |
+
"completion_length": 602.9089285714285,
|
| 750 |
+
"epoch": 0.9706666666666667,
|
| 751 |
+
"grad_norm": 0.4708018898963928,
|
| 752 |
+
"kl": 0.8759209769112724,
|
| 753 |
+
"learning_rate": 6.292608638007513e-09,
|
| 754 |
+
"loss": 0.0268,
|
| 755 |
+
"reward": 0.6160714285714286,
|
| 756 |
+
"reward_std": 0.32809826050485885,
|
| 757 |
+
"rewards/accuracy_reward": 0.6160714285714286,
|
| 758 |
"rewards/format_reward": 0.0,
|
| 759 |
"step": 260
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"clip_ratio": 0.0,
|
| 763 |
+
"completion_length": 609.9160714285714,
|
| 764 |
+
"epoch": 0.9893333333333333,
|
| 765 |
+
"grad_norm": 0.7122711539268494,
|
| 766 |
+
"kl": 1.1589093889508928,
|
| 767 |
+
"learning_rate": 5.140125366641102e-10,
|
| 768 |
+
"loss": 0.008,
|
| 769 |
+
"reward": 0.6017857142857143,
|
| 770 |
+
"reward_std": 0.3146214655467442,
|
| 771 |
+
"rewards/accuracy_reward": 0.6017857142857143,
|
| 772 |
"rewards/format_reward": 0.0,
|
| 773 |
"step": 265
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"clip_ratio": 0.0,
|
| 777 |
+
"completion_length": 585.875,
|
| 778 |
+
"epoch": 0.9968,
|
| 779 |
+
"kl": 1.7515335083007812,
|
| 780 |
+
"reward": 0.5803571428571429,
|
| 781 |
+
"reward_std": 0.327693800841059,
|
| 782 |
+
"rewards/accuracy_reward": 0.5803571428571429,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 783 |
"rewards/format_reward": 0.0,
|
| 784 |
+
"step": 267,
|
| 785 |
"total_flos": 0.0,
|
| 786 |
+
"train_loss": -0.0028305855572698046,
|
| 787 |
+
"train_runtime": 255598.0178,
|
| 788 |
+
"train_samples_per_second": 0.029,
|
| 789 |
"train_steps_per_second": 0.001
|
| 790 |
}
|
| 791 |
],
|
| 792 |
"logging_steps": 5,
|
| 793 |
+
"max_steps": 267,
|
| 794 |
"num_input_tokens_seen": 0,
|
| 795 |
"num_train_epochs": 1,
|
| 796 |
"save_steps": 500,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8056
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b96515b2f249966954c886884a845d62fb82eb793b98e31490e2eebeff1e8f8
|
| 3 |
size 8056
|