Model save
Browse files- README.md +1 -1
- all_results.json +4 -4
- 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 +4 -4
- trainer_state.json +453 -115
- training_args.bin +2 -2
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/swy666-georgia-institute-of-technology/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/swy666-georgia-institute-of-technology/huggingface/runs/v2hi6eqw)
|
| 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":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.01644946206223318,
|
| 4 |
+
"train_runtime": 16174.6118,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.464,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
}
|
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 sha256:44956b4af20af3b78a6801cf5477dcdc1ec8fa2480bbefc697fa8471268401cf
|
| 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:c1fd717a4ae6b501f23d286c75784ba610dfdd4dec2e22d657a95c35e7aaeb0d
|
| 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 sha256:76b8c3f7b6418f7ff4dfa72adaeabfda3debcf80e28efab75fd938ce093c478d
|
| 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 sha256:3015980f2935746b3ecdfaeca65d25413a76ad7e6216af2c3f4212790a97c433
|
| 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":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.01644946206223318,
|
| 4 |
+
"train_runtime": 16174.6118,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.464,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,174 +1,512 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"eval_steps":
|
| 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 |
-
"completion_length":
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm":
|
| 15 |
-
"kl": 0.
|
| 16 |
-
"learning_rate":
|
| 17 |
"loss": 0.0,
|
| 18 |
-
"reward": 0.
|
| 19 |
-
"reward_std": 0.
|
| 20 |
-
"rewards/accuracy_reward": 0.
|
| 21 |
-
"rewards/format_reward": 0.
|
| 22 |
"step": 5
|
| 23 |
},
|
| 24 |
{
|
| 25 |
-
"completion_length":
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"kl": 0.
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss": 0.
|
| 31 |
-
"reward": 0.
|
| 32 |
-
"reward_std": 0.
|
| 33 |
-
"rewards/accuracy_reward": 0.
|
| 34 |
-
"rewards/format_reward": 0.
|
| 35 |
"step": 10
|
| 36 |
},
|
| 37 |
{
|
| 38 |
-
"completion_length":
|
| 39 |
-
"epoch": 0.
|
| 40 |
-
"grad_norm": 0.
|
| 41 |
-
"kl": 0.
|
| 42 |
-
"learning_rate": 2.
|
| 43 |
"loss": 0.0001,
|
| 44 |
-
"reward": 0.
|
| 45 |
-
"reward_std": 0.
|
| 46 |
-
"rewards/accuracy_reward": 0.
|
| 47 |
"rewards/format_reward": 0.0,
|
| 48 |
"step": 15
|
| 49 |
},
|
| 50 |
{
|
| 51 |
-
"completion_length":
|
| 52 |
-
"epoch": 0.
|
| 53 |
-
"grad_norm": 0.
|
| 54 |
-
"kl": 0.
|
| 55 |
-
"learning_rate": 2.
|
| 56 |
-
"loss": 0.
|
| 57 |
-
"reward": 0.
|
| 58 |
-
"reward_std": 0.
|
| 59 |
-
"rewards/accuracy_reward": 0.
|
| 60 |
"rewards/format_reward": 0.0,
|
| 61 |
"step": 20
|
| 62 |
},
|
| 63 |
{
|
| 64 |
-
"completion_length":
|
| 65 |
-
"epoch": 0.
|
| 66 |
-
"grad_norm":
|
| 67 |
-
"kl":
|
| 68 |
-
"learning_rate": 2.
|
| 69 |
-
"loss": 0.
|
| 70 |
-
"reward": 0.
|
| 71 |
-
"reward_std": 0.
|
| 72 |
-
"rewards/accuracy_reward": 0.
|
| 73 |
"rewards/format_reward": 0.0,
|
| 74 |
"step": 25
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"completion_length": 618.
|
| 78 |
-
"epoch": 0.
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"kl":
|
| 81 |
-
"learning_rate":
|
| 82 |
-
"loss": 0.
|
| 83 |
-
"reward": 0.
|
| 84 |
-
"reward_std": 0.
|
| 85 |
-
"rewards/accuracy_reward": 0.
|
| 86 |
"rewards/format_reward": 0.0,
|
| 87 |
"step": 30
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"completion_length":
|
| 91 |
-
"epoch": 0.
|
| 92 |
-
"grad_norm":
|
| 93 |
-
"kl": 0.
|
| 94 |
-
"learning_rate":
|
| 95 |
-
"loss": 0.
|
| 96 |
-
"reward": 0.
|
| 97 |
-
"reward_std": 0.
|
| 98 |
-
"rewards/accuracy_reward": 0.
|
| 99 |
"rewards/format_reward": 0.0,
|
| 100 |
"step": 35
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"completion_length":
|
| 104 |
-
"epoch": 0.
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
-
"kl": 0.
|
| 107 |
-
"learning_rate":
|
| 108 |
-
"loss": 0.
|
| 109 |
-
"reward": 0.
|
| 110 |
-
"reward_std": 0.
|
| 111 |
-
"rewards/accuracy_reward": 0.
|
| 112 |
"rewards/format_reward": 0.0,
|
| 113 |
"step": 40
|
| 114 |
},
|
| 115 |
{
|
| 116 |
-
"completion_length":
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"kl": 0.
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
-
"reward": 0.
|
| 123 |
-
"reward_std": 0.
|
| 124 |
-
"rewards/accuracy_reward": 0.
|
| 125 |
"rewards/format_reward": 0.0,
|
| 126 |
"step": 45
|
| 127 |
},
|
| 128 |
{
|
| 129 |
-
"completion_length":
|
| 130 |
-
"epoch": 0.
|
| 131 |
-
"grad_norm":
|
| 132 |
-
"kl": 0.
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
-
"reward": 0.
|
| 136 |
-
"reward_std": 0.
|
| 137 |
-
"rewards/accuracy_reward": 0.
|
| 138 |
"rewards/format_reward": 0.0,
|
| 139 |
"step": 50
|
| 140 |
},
|
| 141 |
{
|
| 142 |
-
"completion_length":
|
| 143 |
-
"epoch": 0.
|
| 144 |
-
"grad_norm": 0.
|
| 145 |
-
"kl": 0.
|
| 146 |
-
"learning_rate": 2.
|
| 147 |
-
"loss": 0.
|
| 148 |
-
"reward": 0.
|
| 149 |
-
"reward_std": 0.
|
| 150 |
-
"rewards/accuracy_reward": 0.
|
| 151 |
"rewards/format_reward": 0.0,
|
| 152 |
"step": 55
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"completion_length":
|
| 156 |
-
"epoch": 0.
|
| 157 |
-
"
|
| 158 |
-
"
|
| 159 |
-
"
|
| 160 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 161 |
"rewards/format_reward": 0.0,
|
| 162 |
-
"step":
|
| 163 |
"total_flos": 0.0,
|
| 164 |
-
"train_loss": 0.
|
| 165 |
-
"train_runtime":
|
| 166 |
-
"train_samples_per_second":
|
| 167 |
-
"train_steps_per_second": 0.
|
| 168 |
}
|
| 169 |
],
|
| 170 |
"logging_steps": 5,
|
| 171 |
-
"max_steps":
|
| 172 |
"num_input_tokens_seen": 0,
|
| 173 |
"num_train_epochs": 1,
|
| 174 |
"save_steps": 500,
|
|
@@ -185,7 +523,7 @@
|
|
| 185 |
}
|
| 186 |
},
|
| 187 |
"total_flos": 0.0,
|
| 188 |
-
"train_batch_size":
|
| 189 |
"trial_name": null,
|
| 190 |
"trial_params": null
|
| 191 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9973333333333333,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 187,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"completion_length": 598.9556373596191,
|
| 13 |
+
"epoch": 0.02666666666666667,
|
| 14 |
+
"grad_norm": 9.657825469970703,
|
| 15 |
+
"kl": 0.00011155605316162109,
|
| 16 |
+
"learning_rate": 7.894736842105263e-07,
|
| 17 |
"loss": 0.0,
|
| 18 |
+
"reward": 0.6343750096857548,
|
| 19 |
+
"reward_std": 0.32928847298026087,
|
| 20 |
+
"rewards/accuracy_reward": 0.6337500102818012,
|
| 21 |
+
"rewards/format_reward": 0.0006250000093132258,
|
| 22 |
"step": 5
|
| 23 |
},
|
| 24 |
{
|
| 25 |
+
"completion_length": 605.9112594604492,
|
| 26 |
+
"epoch": 0.05333333333333334,
|
| 27 |
+
"grad_norm": 1.3849772214889526,
|
| 28 |
+
"kl": 0.00035059452056884766,
|
| 29 |
+
"learning_rate": 1.5789473684210526e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"reward": 0.6206250086426734,
|
| 32 |
+
"reward_std": 0.30763257276266814,
|
| 33 |
+
"rewards/accuracy_reward": 0.6200000092387199,
|
| 34 |
+
"rewards/format_reward": 0.0006250000093132258,
|
| 35 |
"step": 10
|
| 36 |
},
|
| 37 |
{
|
| 38 |
+
"completion_length": 640.480638885498,
|
| 39 |
+
"epoch": 0.08,
|
| 40 |
+
"grad_norm": 0.2819180488586426,
|
| 41 |
+
"kl": 0.0017528533935546875,
|
| 42 |
+
"learning_rate": 2.368421052631579e-06,
|
| 43 |
"loss": 0.0001,
|
| 44 |
+
"reward": 0.6893750086426735,
|
| 45 |
+
"reward_std": 0.28186101242899897,
|
| 46 |
+
"rewards/accuracy_reward": 0.6893750086426735,
|
| 47 |
"rewards/format_reward": 0.0,
|
| 48 |
"step": 15
|
| 49 |
},
|
| 50 |
{
|
| 51 |
+
"completion_length": 641.2750106811524,
|
| 52 |
+
"epoch": 0.10666666666666667,
|
| 53 |
+
"grad_norm": 0.2619169354438782,
|
| 54 |
+
"kl": 0.003242778778076172,
|
| 55 |
+
"learning_rate": 2.99973774136912e-06,
|
| 56 |
+
"loss": 0.0001,
|
| 57 |
+
"reward": 0.6918750137090683,
|
| 58 |
+
"reward_std": 0.280092066898942,
|
| 59 |
+
"rewards/accuracy_reward": 0.6918750137090683,
|
| 60 |
"rewards/format_reward": 0.0,
|
| 61 |
"step": 20
|
| 62 |
},
|
| 63 |
{
|
| 64 |
+
"completion_length": 651.3437652587891,
|
| 65 |
+
"epoch": 0.13333333333333333,
|
| 66 |
+
"grad_norm": 0.7606603503227234,
|
| 67 |
+
"kl": 0.0034598350524902345,
|
| 68 |
+
"learning_rate": 2.9905683148398643e-06,
|
| 69 |
+
"loss": 0.0001,
|
| 70 |
+
"reward": 0.6881250128149986,
|
| 71 |
+
"reward_std": 0.2426654167473316,
|
| 72 |
+
"rewards/accuracy_reward": 0.6881250128149986,
|
| 73 |
"rewards/format_reward": 0.0,
|
| 74 |
"step": 25
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"completion_length": 618.8025131225586,
|
| 78 |
+
"epoch": 0.16,
|
| 79 |
+
"grad_norm": 18.173175811767578,
|
| 80 |
+
"kl": 1.1331443786621094,
|
| 81 |
+
"learning_rate": 2.968377518846473e-06,
|
| 82 |
+
"loss": 0.0453,
|
| 83 |
+
"reward": 0.7312500067055225,
|
| 84 |
+
"reward_std": 0.21250597070902585,
|
| 85 |
+
"rewards/accuracy_reward": 0.7312500067055225,
|
| 86 |
"rewards/format_reward": 0.0,
|
| 87 |
"step": 30
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"completion_length": 594.2168823242188,
|
| 91 |
+
"epoch": 0.18666666666666668,
|
| 92 |
+
"grad_norm": 50.74555587768555,
|
| 93 |
+
"kl": 0.35082550048828126,
|
| 94 |
+
"learning_rate": 2.933359208679211e-06,
|
| 95 |
+
"loss": 0.014,
|
| 96 |
+
"reward": 0.7662500098347664,
|
| 97 |
+
"reward_std": 0.19725151248276235,
|
| 98 |
+
"rewards/accuracy_reward": 0.7662500098347664,
|
| 99 |
"rewards/format_reward": 0.0,
|
| 100 |
"step": 35
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"completion_length": 605.8262619018554,
|
| 104 |
+
"epoch": 0.21333333333333335,
|
| 105 |
+
"grad_norm": 0.8362184762954712,
|
| 106 |
+
"kl": 0.11929092407226563,
|
| 107 |
+
"learning_rate": 2.88581929876693e-06,
|
| 108 |
+
"loss": 0.0048,
|
| 109 |
+
"reward": 0.7368750087916851,
|
| 110 |
+
"reward_std": 0.19084826521575451,
|
| 111 |
+
"rewards/accuracy_reward": 0.7368750087916851,
|
| 112 |
"rewards/format_reward": 0.0,
|
| 113 |
"step": 40
|
| 114 |
},
|
| 115 |
{
|
| 116 |
+
"completion_length": 587.2862632751464,
|
| 117 |
+
"epoch": 0.24,
|
| 118 |
+
"grad_norm": 17.15748405456543,
|
| 119 |
+
"kl": 0.3153179168701172,
|
| 120 |
+
"learning_rate": 2.8261730902569146e-06,
|
| 121 |
+
"loss": 0.0126,
|
| 122 |
+
"reward": 0.771250007301569,
|
| 123 |
+
"reward_std": 0.1733900425955653,
|
| 124 |
+
"rewards/accuracy_reward": 0.771250007301569,
|
| 125 |
"rewards/format_reward": 0.0,
|
| 126 |
"step": 45
|
| 127 |
},
|
| 128 |
{
|
| 129 |
+
"completion_length": 587.2462539672852,
|
| 130 |
+
"epoch": 0.26666666666666666,
|
| 131 |
+
"grad_norm": 1.1150826215744019,
|
| 132 |
+
"kl": 0.0309173583984375,
|
| 133 |
+
"learning_rate": 2.7549416430096295e-06,
|
| 134 |
+
"loss": 0.0012,
|
| 135 |
+
"reward": 0.7643750108778476,
|
| 136 |
+
"reward_std": 0.20076387487351893,
|
| 137 |
+
"rewards/accuracy_reward": 0.7643750108778476,
|
| 138 |
"rewards/format_reward": 0.0,
|
| 139 |
"step": 50
|
| 140 |
},
|
| 141 |
{
|
| 142 |
+
"completion_length": 596.9281379699707,
|
| 143 |
+
"epoch": 0.29333333333333333,
|
| 144 |
+
"grad_norm": 0.9277231693267822,
|
| 145 |
+
"kl": 0.026005935668945313,
|
| 146 |
+
"learning_rate": 2.6727472237020448e-06,
|
| 147 |
+
"loss": 0.001,
|
| 148 |
+
"reward": 0.7562500096857547,
|
| 149 |
+
"reward_std": 0.20265843532979488,
|
| 150 |
+
"rewards/accuracy_reward": 0.7562500096857547,
|
| 151 |
"rewards/format_reward": 0.0,
|
| 152 |
"step": 55
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"completion_length": 557.7656349182129,
|
| 156 |
+
"epoch": 0.32,
|
| 157 |
+
"grad_norm": 78.5399398803711,
|
| 158 |
+
"kl": 0.11486587524414063,
|
| 159 |
+
"learning_rate": 2.58030786980419e-06,
|
| 160 |
+
"loss": 0.0046,
|
| 161 |
+
"reward": 0.7831250101327896,
|
| 162 |
+
"reward_std": 0.16795916706323624,
|
| 163 |
+
"rewards/accuracy_reward": 0.7831250101327896,
|
| 164 |
+
"rewards/format_reward": 0.0,
|
| 165 |
+
"step": 60
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 540.9593856811523,
|
| 169 |
+
"epoch": 0.3466666666666667,
|
| 170 |
+
"grad_norm": 2.0994198322296143,
|
| 171 |
+
"kl": 0.02927398681640625,
|
| 172 |
+
"learning_rate": 2.4784311169171817e-06,
|
| 173 |
+
"loss": 0.0012,
|
| 174 |
+
"reward": 0.8068750128149986,
|
| 175 |
+
"reward_std": 0.1683789264410734,
|
| 176 |
+
"rewards/accuracy_reward": 0.8068750128149986,
|
| 177 |
+
"rewards/format_reward": 0.0,
|
| 178 |
+
"step": 65
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 608.050633239746,
|
| 182 |
+
"epoch": 0.37333333333333335,
|
| 183 |
+
"grad_norm": 0.15975548326969147,
|
| 184 |
+
"kl": 11.121460723876954,
|
| 185 |
+
"learning_rate": 2.368006944269709e-06,
|
| 186 |
+
"loss": 0.4441,
|
| 187 |
+
"reward": 0.756250013038516,
|
| 188 |
+
"reward_std": 0.19869754556566477,
|
| 189 |
+
"rewards/accuracy_reward": 0.756250013038516,
|
| 190 |
+
"rewards/format_reward": 0.0,
|
| 191 |
+
"step": 70
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"completion_length": 556.0112609863281,
|
| 195 |
+
"epoch": 0.4,
|
| 196 |
+
"grad_norm": 0.257069855928421,
|
| 197 |
+
"kl": 0.0086212158203125,
|
| 198 |
+
"learning_rate": 2.25e-06,
|
| 199 |
+
"loss": 0.0003,
|
| 200 |
+
"reward": 0.8018750116229058,
|
| 201 |
+
"reward_std": 0.16034170743077994,
|
| 202 |
+
"rewards/accuracy_reward": 0.8018750116229058,
|
| 203 |
+
"rewards/format_reward": 0.0,
|
| 204 |
+
"step": 75
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"completion_length": 582.3343910217285,
|
| 208 |
+
"epoch": 0.4266666666666667,
|
| 209 |
+
"grad_norm": 0.13077034056186676,
|
| 210 |
+
"kl": 0.006656646728515625,
|
| 211 |
+
"learning_rate": 2.1254411741419925e-06,
|
| 212 |
+
"loss": 0.0003,
|
| 213 |
+
"reward": 0.7481250077486038,
|
| 214 |
+
"reward_std": 0.1724422551691532,
|
| 215 |
+
"rewards/accuracy_reward": 0.7481250077486038,
|
| 216 |
+
"rewards/format_reward": 0.0,
|
| 217 |
+
"step": 80
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"completion_length": 565.938136291504,
|
| 221 |
+
"epoch": 0.4533333333333333,
|
| 222 |
+
"grad_norm": 0.8749185800552368,
|
| 223 |
+
"kl": 1.2233882904052735,
|
| 224 |
+
"learning_rate": 1.9954185929327507e-06,
|
| 225 |
+
"loss": 0.0489,
|
| 226 |
+
"reward": 0.7856250107288361,
|
| 227 |
+
"reward_std": 0.17018861770629884,
|
| 228 |
+
"rewards/accuracy_reward": 0.7856250107288361,
|
| 229 |
+
"rewards/format_reward": 0.0,
|
| 230 |
+
"step": 85
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 597.7000122070312,
|
| 234 |
+
"epoch": 0.48,
|
| 235 |
+
"grad_norm": 0.44738760590553284,
|
| 236 |
+
"kl": 0.14383926391601562,
|
| 237 |
+
"learning_rate": 1.8610681131134598e-06,
|
| 238 |
+
"loss": 0.0058,
|
| 239 |
+
"reward": 0.7462500132620334,
|
| 240 |
+
"reward_std": 0.18847194407135248,
|
| 241 |
+
"rewards/accuracy_reward": 0.7462500132620334,
|
| 242 |
+
"rewards/format_reward": 0.0,
|
| 243 |
+
"step": 90
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"completion_length": 594.4781372070313,
|
| 247 |
+
"epoch": 0.5066666666666667,
|
| 248 |
+
"grad_norm": 0.20932912826538086,
|
| 249 |
+
"kl": 0.019523239135742186,
|
| 250 |
+
"learning_rate": 1.7235633992642616e-06,
|
| 251 |
+
"loss": 0.0008,
|
| 252 |
+
"reward": 0.7800000101327896,
|
| 253 |
+
"reward_std": 0.18672875873744488,
|
| 254 |
+
"rewards/accuracy_reward": 0.7800000101327896,
|
| 255 |
+
"rewards/format_reward": 0.0,
|
| 256 |
+
"step": 95
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"completion_length": 567.590631866455,
|
| 260 |
+
"epoch": 0.5333333333333333,
|
| 261 |
+
"grad_norm": 0.26683634519577026,
|
| 262 |
+
"kl": 0.00815887451171875,
|
| 263 |
+
"learning_rate": 1.5841056708557877e-06,
|
| 264 |
+
"loss": 0.0003,
|
| 265 |
+
"reward": 0.8012500151991844,
|
| 266 |
+
"reward_std": 0.15466174986213446,
|
| 267 |
+
"rewards/accuracy_reward": 0.8012500151991844,
|
| 268 |
+
"rewards/format_reward": 0.0,
|
| 269 |
+
"step": 100
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"completion_length": 602.0687614440918,
|
| 273 |
+
"epoch": 0.56,
|
| 274 |
+
"grad_norm": 0.24808332324028015,
|
| 275 |
+
"kl": 0.046283721923828125,
|
| 276 |
+
"learning_rate": 1.4439132085855116e-06,
|
| 277 |
+
"loss": 0.0018,
|
| 278 |
+
"reward": 0.7306250102818013,
|
| 279 |
+
"reward_std": 0.2066970258951187,
|
| 280 |
+
"rewards/accuracy_reward": 0.7306250102818013,
|
| 281 |
+
"rewards/format_reward": 0.0,
|
| 282 |
+
"step": 105
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"completion_length": 561.5981369018555,
|
| 286 |
+
"epoch": 0.5866666666666667,
|
| 287 |
+
"grad_norm": 0.13664592802524567,
|
| 288 |
+
"kl": 0.00771636962890625,
|
| 289 |
+
"learning_rate": 1.304210711669923e-06,
|
| 290 |
+
"loss": 0.0003,
|
| 291 |
+
"reward": 0.7881250090897083,
|
| 292 |
+
"reward_std": 0.16074271276593208,
|
| 293 |
+
"rewards/accuracy_reward": 0.7881250090897083,
|
| 294 |
+
"rewards/format_reward": 0.0,
|
| 295 |
+
"step": 110
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"completion_length": 572.8368896484375,
|
| 299 |
+
"epoch": 0.6133333333333333,
|
| 300 |
+
"grad_norm": 61.19581985473633,
|
| 301 |
+
"kl": 0.08488540649414063,
|
| 302 |
+
"learning_rate": 1.1662185990655286e-06,
|
| 303 |
+
"loss": 0.0034,
|
| 304 |
+
"reward": 0.7856250144541264,
|
| 305 |
+
"reward_std": 0.15731600020080805,
|
| 306 |
+
"rewards/accuracy_reward": 0.7856250144541264,
|
| 307 |
+
"rewards/format_reward": 0.0,
|
| 308 |
+
"step": 115
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"completion_length": 566.652507019043,
|
| 312 |
+
"epoch": 0.64,
|
| 313 |
+
"grad_norm": 0.6947761178016663,
|
| 314 |
+
"kl": 0.00809326171875,
|
| 315 |
+
"learning_rate": 1.0311423480815335e-06,
|
| 316 |
+
"loss": 0.0003,
|
| 317 |
+
"reward": 0.7475000098347664,
|
| 318 |
+
"reward_std": 0.17467193938791753,
|
| 319 |
+
"rewards/accuracy_reward": 0.7475000098347664,
|
| 320 |
+
"rewards/format_reward": 0.0,
|
| 321 |
+
"step": 120
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"completion_length": 578.0793846130371,
|
| 325 |
+
"epoch": 0.6666666666666666,
|
| 326 |
+
"grad_norm": 0.2171238362789154,
|
| 327 |
+
"kl": 0.4027862548828125,
|
| 328 |
+
"learning_rate": 9.001619635203888e-07,
|
| 329 |
+
"loss": 0.0162,
|
| 330 |
+
"reward": 0.7456250101327896,
|
| 331 |
+
"reward_std": 0.15293692518025637,
|
| 332 |
+
"rewards/accuracy_reward": 0.7456250101327896,
|
| 333 |
+
"rewards/format_reward": 0.0,
|
| 334 |
+
"step": 125
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"completion_length": 565.3787605285645,
|
| 338 |
+
"epoch": 0.6933333333333334,
|
| 339 |
+
"grad_norm": 0.3446357548236847,
|
| 340 |
+
"kl": 0.038055801391601564,
|
| 341 |
+
"learning_rate": 7.744216693421403e-07,
|
| 342 |
+
"loss": 0.0015,
|
| 343 |
+
"reward": 0.7831250131130219,
|
| 344 |
+
"reward_std": 0.1700244778767228,
|
| 345 |
+
"rewards/accuracy_reward": 0.7831250131130219,
|
| 346 |
+
"rewards/format_reward": 0.0,
|
| 347 |
+
"step": 130
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"completion_length": 582.8950141906738,
|
| 351 |
+
"epoch": 0.72,
|
| 352 |
+
"grad_norm": 0.15175624191761017,
|
| 353 |
+
"kl": 0.008423233032226562,
|
| 354 |
+
"learning_rate": 6.550199129045669e-07,
|
| 355 |
+
"loss": 0.0003,
|
| 356 |
+
"reward": 0.7750000104308128,
|
| 357 |
+
"reward_std": 0.16627665907144545,
|
| 358 |
+
"rewards/accuracy_reward": 0.7750000104308128,
|
| 359 |
+
"rewards/format_reward": 0.0,
|
| 360 |
+
"step": 135
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"completion_length": 574.6862632751465,
|
| 364 |
+
"epoch": 0.7466666666666667,
|
| 365 |
+
"grad_norm": 0.24874858558177948,
|
| 366 |
+
"kl": 0.018523406982421876,
|
| 367 |
+
"learning_rate": 5.429997691004874e-07,
|
| 368 |
+
"loss": 0.0007,
|
| 369 |
+
"reward": 0.7637500140815974,
|
| 370 |
+
"reward_std": 0.17574753649532796,
|
| 371 |
+
"rewards/accuracy_reward": 0.7637500140815974,
|
| 372 |
+
"rewards/format_reward": 0.0,
|
| 373 |
+
"step": 140
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"completion_length": 582.8781341552734,
|
| 377 |
+
"epoch": 0.7733333333333333,
|
| 378 |
+
"grad_norm": 0.22961439192295074,
|
| 379 |
+
"kl": 0.00735931396484375,
|
| 380 |
+
"learning_rate": 4.3933982822017883e-07,
|
| 381 |
+
"loss": 0.0003,
|
| 382 |
+
"reward": 0.7487500086426735,
|
| 383 |
+
"reward_std": 0.16670059841126203,
|
| 384 |
+
"rewards/accuracy_reward": 0.7487500086426735,
|
| 385 |
+
"rewards/format_reward": 0.0,
|
| 386 |
+
"step": 145
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"completion_length": 591.0300117492676,
|
| 390 |
+
"epoch": 0.8,
|
| 391 |
+
"grad_norm": 0.1450783908367157,
|
| 392 |
+
"kl": 0.0070438385009765625,
|
| 393 |
+
"learning_rate": 3.449456471411058e-07,
|
| 394 |
+
"loss": 0.0003,
|
| 395 |
+
"reward": 0.7575000070035458,
|
| 396 |
+
"reward_std": 0.17982883434742689,
|
| 397 |
+
"rewards/accuracy_reward": 0.7575000070035458,
|
| 398 |
+
"rewards/format_reward": 0.0,
|
| 399 |
+
"step": 150
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"completion_length": 595.1437629699707,
|
| 403 |
+
"epoch": 0.8266666666666667,
|
| 404 |
+
"grad_norm": 0.1316715031862259,
|
| 405 |
+
"kl": 0.012836456298828125,
|
| 406 |
+
"learning_rate": 2.6064183852600797e-07,
|
| 407 |
+
"loss": 0.0005,
|
| 408 |
+
"reward": 0.7687500096857548,
|
| 409 |
+
"reward_std": 0.1724598340690136,
|
| 410 |
+
"rewards/accuracy_reward": 0.7687500096857548,
|
| 411 |
+
"rewards/format_reward": 0.0,
|
| 412 |
+
"step": 155
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"completion_length": 572.5243888854981,
|
| 416 |
+
"epoch": 0.8533333333333334,
|
| 417 |
+
"grad_norm": 0.17138485610485077,
|
| 418 |
+
"kl": 0.007579803466796875,
|
| 419 |
+
"learning_rate": 1.8716486713686948e-07,
|
| 420 |
+
"loss": 0.0003,
|
| 421 |
+
"reward": 0.7725000090897083,
|
| 422 |
+
"reward_std": 0.18208168745040892,
|
| 423 |
+
"rewards/accuracy_reward": 0.7725000090897083,
|
| 424 |
+
"rewards/format_reward": 0.0,
|
| 425 |
+
"step": 160
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"completion_length": 587.6368865966797,
|
| 429 |
+
"epoch": 0.88,
|
| 430 |
+
"grad_norm": 0.37863653898239136,
|
| 431 |
+
"kl": 0.00740203857421875,
|
| 432 |
+
"learning_rate": 1.251566161950357e-07,
|
| 433 |
+
"loss": 0.0003,
|
| 434 |
+
"reward": 0.7600000105798245,
|
| 435 |
+
"reward_std": 0.19197216629981995,
|
| 436 |
+
"rewards/accuracy_reward": 0.7600000105798245,
|
| 437 |
+
"rewards/format_reward": 0.0,
|
| 438 |
+
"step": 165
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"completion_length": 590.2731330871582,
|
| 442 |
+
"epoch": 0.9066666666666666,
|
| 443 |
+
"grad_norm": 0.17426247894763947,
|
| 444 |
+
"kl": 0.008208465576171876,
|
| 445 |
+
"learning_rate": 7.515877999073101e-08,
|
| 446 |
+
"loss": 0.0003,
|
| 447 |
+
"reward": 0.7662500113248825,
|
| 448 |
+
"reward_std": 0.19529659319669007,
|
| 449 |
+
"rewards/accuracy_reward": 0.7662500113248825,
|
| 450 |
+
"rewards/format_reward": 0.0,
|
| 451 |
+
"step": 170
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"completion_length": 580.357511138916,
|
| 455 |
+
"epoch": 0.9333333333333333,
|
| 456 |
+
"grad_norm": 3.413297176361084,
|
| 457 |
+
"kl": 0.05416526794433594,
|
| 458 |
+
"learning_rate": 3.7608131727264573e-08,
|
| 459 |
+
"loss": 0.0022,
|
| 460 |
+
"reward": 0.7906250134110451,
|
| 461 |
+
"reward_std": 0.17527884822338818,
|
| 462 |
+
"rewards/accuracy_reward": 0.7906250134110451,
|
| 463 |
+
"rewards/format_reward": 0.0,
|
| 464 |
+
"step": 175
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"completion_length": 597.2575073242188,
|
| 468 |
+
"epoch": 0.96,
|
| 469 |
+
"grad_norm": 0.16546441614627838,
|
| 470 |
+
"kl": 0.00909271240234375,
|
| 471 |
+
"learning_rate": 1.2832707939284426e-08,
|
| 472 |
+
"loss": 0.0004,
|
| 473 |
+
"reward": 0.7381250090897084,
|
| 474 |
+
"reward_std": 0.1613092703744769,
|
| 475 |
+
"rewards/accuracy_reward": 0.7381250090897084,
|
| 476 |
+
"rewards/format_reward": 0.0,
|
| 477 |
+
"step": 180
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"completion_length": 548.148137664795,
|
| 481 |
+
"epoch": 0.9866666666666667,
|
| 482 |
+
"grad_norm": 0.48810043931007385,
|
| 483 |
+
"kl": 0.007534408569335937,
|
| 484 |
+
"learning_rate": 1.0489428174020875e-09,
|
| 485 |
+
"loss": 0.0003,
|
| 486 |
+
"reward": 0.7881250143051147,
|
| 487 |
+
"reward_std": 0.1571642367169261,
|
| 488 |
+
"rewards/accuracy_reward": 0.7881250143051147,
|
| 489 |
+
"rewards/format_reward": 0.0,
|
| 490 |
+
"step": 185
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"completion_length": 565.9296970367432,
|
| 494 |
+
"epoch": 0.9973333333333333,
|
| 495 |
+
"kl": 0.0072765350341796875,
|
| 496 |
+
"reward": 0.8171875178813934,
|
| 497 |
+
"reward_std": 0.1507129198871553,
|
| 498 |
+
"rewards/accuracy_reward": 0.8171875178813934,
|
| 499 |
"rewards/format_reward": 0.0,
|
| 500 |
+
"step": 187,
|
| 501 |
"total_flos": 0.0,
|
| 502 |
+
"train_loss": 0.01644946206223318,
|
| 503 |
+
"train_runtime": 16174.6118,
|
| 504 |
+
"train_samples_per_second": 0.464,
|
| 505 |
+
"train_steps_per_second": 0.012
|
| 506 |
}
|
| 507 |
],
|
| 508 |
"logging_steps": 5,
|
| 509 |
+
"max_steps": 187,
|
| 510 |
"num_input_tokens_seen": 0,
|
| 511 |
"num_train_epochs": 1,
|
| 512 |
"save_steps": 500,
|
|
|
|
| 523 |
}
|
| 524 |
},
|
| 525 |
"total_flos": 0.0,
|
| 526 |
+
"train_batch_size": 8,
|
| 527 |
"trial_name": null,
|
| 528 |
"trial_params": null
|
| 529 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f89ddda7b07eaac0237eb7b492024455239dbe07a59368b11b00655c059a7f84
|
| 3 |
+
size 7544
|