End of training
Browse files- README.md +2 -1
- adapter_config.json +39 -0
- adapter_model.safetensors +3 -0
- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +142 -902
- training_args.bin +2 -2
README.md
CHANGED
|
@@ -8,6 +8,7 @@ tags:
|
|
| 8 |
- grpo
|
| 9 |
- gensyn
|
| 10 |
- I am bipedal exotic pelican
|
|
|
|
| 11 |
- trl
|
| 12 |
licence: license
|
| 13 |
---
|
|
@@ -39,7 +40,7 @@ This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing
|
|
| 39 |
|
| 40 |
- TRL: 0.15.2
|
| 41 |
- Transformers: 4.51.3
|
| 42 |
-
- Pytorch: 2.
|
| 43 |
- Datasets: 3.5.1
|
| 44 |
- Tokenizers: 0.21.1
|
| 45 |
|
|
|
|
| 8 |
- grpo
|
| 9 |
- gensyn
|
| 10 |
- I am bipedal exotic pelican
|
| 11 |
+
- unsloth
|
| 12 |
- trl
|
| 13 |
licence: license
|
| 14 |
---
|
|
|
|
| 40 |
|
| 41 |
- TRL: 0.15.2
|
| 42 |
- Transformers: 4.51.3
|
| 43 |
+
- Pytorch: 2.6.0
|
| 44 |
- Datasets: 3.5.1
|
| 45 |
- Tokenizers: 0.21.1
|
| 46 |
|
adapter_config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "Gensyn/Qwen2.5-0.5B-Instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"corda_config": null,
|
| 7 |
+
"eva_config": null,
|
| 8 |
+
"exclude_modules": null,
|
| 9 |
+
"fan_in_fan_out": false,
|
| 10 |
+
"inference_mode": true,
|
| 11 |
+
"init_lora_weights": true,
|
| 12 |
+
"layer_replication": null,
|
| 13 |
+
"layers_pattern": null,
|
| 14 |
+
"layers_to_transform": null,
|
| 15 |
+
"loftq_config": {},
|
| 16 |
+
"lora_alpha": 16,
|
| 17 |
+
"lora_bias": false,
|
| 18 |
+
"lora_dropout": 0,
|
| 19 |
+
"megatron_config": null,
|
| 20 |
+
"megatron_core": "megatron.core",
|
| 21 |
+
"modules_to_save": null,
|
| 22 |
+
"peft_type": "LORA",
|
| 23 |
+
"r": 16,
|
| 24 |
+
"rank_pattern": {},
|
| 25 |
+
"revision": null,
|
| 26 |
+
"target_modules": [
|
| 27 |
+
"up_proj",
|
| 28 |
+
"o_proj",
|
| 29 |
+
"k_proj",
|
| 30 |
+
"v_proj",
|
| 31 |
+
"down_proj",
|
| 32 |
+
"gate_proj",
|
| 33 |
+
"q_proj"
|
| 34 |
+
],
|
| 35 |
+
"task_type": "CAUSAL_LM",
|
| 36 |
+
"trainable_token_indices": null,
|
| 37 |
+
"use_dora": false,
|
| 38 |
+
"use_rslora": false
|
| 39 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8995ee0ce70d515dc683e56f972f42413eb9b3d3b356e6821e9f2e385d8c4d8f
|
| 3 |
+
size 35237104
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 2.47737436893658e-06,
|
| 4 |
+
"train_runtime": 471.1429,
|
| 5 |
+
"train_samples": 19,
|
| 6 |
+
"train_samples_per_second": 0.679,
|
| 7 |
+
"train_steps_per_second": 0.042
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 2.47737436893658e-06,
|
| 4 |
+
"train_runtime": 471.1429,
|
| 5 |
+
"train_samples": 19,
|
| 6 |
+
"train_samples_per_second": 0.679,
|
| 7 |
+
"train_steps_per_second": 0.042
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -2,977 +2,217 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"completion_length":
|
| 14 |
-
"epoch": 0.
|
| 15 |
-
"grad_norm":
|
| 16 |
-
"kl": 0.
|
| 17 |
-
"learning_rate":
|
| 18 |
-
"loss":
|
| 19 |
-
"reward":
|
| 20 |
-
"reward_std":
|
| 21 |
-
"rewards/concensus_correctness_reward_func":
|
| 22 |
-
"rewards/consensus_reward_func":
|
| 23 |
"rewards/cumulative_reward_2": 0.0,
|
| 24 |
"rewards/final_correctness_reward_func": 0.0,
|
| 25 |
-
"rewards/question_recreation_reward_func": 0.
|
| 26 |
"rewards/soft_format_reward_func": 0.0,
|
| 27 |
-
"rewards/strict_format_reward_func": 0.
|
| 28 |
-
"rewards/xmlcount_reward_func": 0.
|
| 29 |
"step": 2
|
| 30 |
},
|
| 31 |
{
|
| 32 |
-
"completion_length":
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"kl": 0.
|
| 36 |
-
"learning_rate":
|
| 37 |
"loss": 0.0,
|
| 38 |
-
"reward":
|
| 39 |
-
"reward_std": 0.
|
| 40 |
-
"rewards/concensus_correctness_reward_func":
|
| 41 |
-
"rewards/consensus_reward_func":
|
| 42 |
"rewards/cumulative_reward_2": 0.0,
|
| 43 |
-
"rewards/final_correctness_reward_func": 0.
|
| 44 |
-
"rewards/question_recreation_reward_func": 0.
|
| 45 |
"rewards/soft_format_reward_func": 0.0,
|
| 46 |
-
"rewards/strict_format_reward_func": 0.
|
| 47 |
-
"rewards/xmlcount_reward_func":
|
| 48 |
"step": 4
|
| 49 |
},
|
| 50 |
{
|
| 51 |
-
"completion_length":
|
| 52 |
-
"epoch":
|
| 53 |
-
"grad_norm":
|
| 54 |
-
"kl": 0.
|
| 55 |
-
"learning_rate": 4.
|
| 56 |
"loss": 0.0,
|
| 57 |
-
"reward":
|
| 58 |
-
"reward_std": 0.
|
| 59 |
-
"rewards/concensus_correctness_reward_func":
|
| 60 |
-
"rewards/consensus_reward_func":
|
| 61 |
"rewards/cumulative_reward_2": 0.0,
|
| 62 |
-
"rewards/final_correctness_reward_func": 0.
|
| 63 |
-
"rewards/question_recreation_reward_func":
|
| 64 |
"rewards/soft_format_reward_func": 0.0,
|
| 65 |
-
"rewards/strict_format_reward_func": 0.
|
| 66 |
-
"rewards/xmlcount_reward_func":
|
| 67 |
"step": 6
|
| 68 |
},
|
| 69 |
{
|
| 70 |
-
"completion_length":
|
| 71 |
-
"epoch":
|
| 72 |
-
"grad_norm":
|
| 73 |
-
"kl": 0.
|
| 74 |
-
"learning_rate":
|
| 75 |
"loss": 0.0,
|
| 76 |
-
"reward":
|
| 77 |
-
"reward_std": 0.
|
| 78 |
-
"rewards/concensus_correctness_reward_func":
|
| 79 |
-
"rewards/consensus_reward_func":
|
| 80 |
"rewards/cumulative_reward_2": 0.0,
|
| 81 |
-
"rewards/final_correctness_reward_func": 0.
|
| 82 |
-
"rewards/question_recreation_reward_func":
|
| 83 |
"rewards/soft_format_reward_func": 0.0,
|
| 84 |
-
"rewards/strict_format_reward_func": 0.
|
| 85 |
-
"rewards/xmlcount_reward_func":
|
| 86 |
"step": 8
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"completion_length":
|
| 90 |
-
"epoch": 0
|
| 91 |
-
"grad_norm": 0.
|
| 92 |
-
"kl": 0.
|
| 93 |
-
"learning_rate":
|
| 94 |
"loss": 0.0,
|
| 95 |
-
"reward":
|
| 96 |
-
"reward_std": 0.
|
| 97 |
-
"rewards/concensus_correctness_reward_func":
|
| 98 |
-
"rewards/consensus_reward_func":
|
| 99 |
"rewards/cumulative_reward_2": 0.0,
|
| 100 |
-
"rewards/final_correctness_reward_func": 0.
|
| 101 |
-
"rewards/question_recreation_reward_func": 0.
|
| 102 |
"rewards/soft_format_reward_func": 0.0,
|
| 103 |
-
"rewards/strict_format_reward_func": 0.
|
| 104 |
-
"rewards/xmlcount_reward_func":
|
| 105 |
"step": 10
|
| 106 |
},
|
| 107 |
{
|
| 108 |
-
"completion_length":
|
| 109 |
-
"epoch":
|
| 110 |
-
"grad_norm":
|
| 111 |
-
"kl":
|
| 112 |
-
"learning_rate":
|
| 113 |
-
"loss": 0.007,
|
| 114 |
-
"reward": 7.247430235147476,
|
| 115 |
-
"reward_std": 0.0,
|
| 116 |
-
"rewards/concensus_correctness_reward_func": 2.2944999784231186,
|
| 117 |
-
"rewards/consensus_reward_func": 1.875,
|
| 118 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 119 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 120 |
-
"rewards/question_recreation_reward_func": 0.9998052977025509,
|
| 121 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 122 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 123 |
-
"rewards/xmlcount_reward_func": 1.234375,
|
| 124 |
-
"step": 12
|
| 125 |
-
},
|
| 126 |
-
{
|
| 127 |
-
"completion_length": 122.0,
|
| 128 |
-
"epoch": 0.8,
|
| 129 |
-
"grad_norm": 1.0056698322296143,
|
| 130 |
-
"kl": 0.013116561494825874,
|
| 131 |
-
"learning_rate": 4.870022949890676e-07,
|
| 132 |
"loss": 0.0,
|
| 133 |
-
"reward":
|
| 134 |
-
"reward_std": 0.
|
| 135 |
-
"rewards/concensus_correctness_reward_func":
|
| 136 |
-
"rewards/consensus_reward_func":
|
| 137 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 138 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 139 |
-
"rewards/question_recreation_reward_func": 0.9946808516979218,
|
| 140 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 141 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 142 |
-
"rewards/xmlcount_reward_func": 1.2421875,
|
| 143 |
-
"step": 14
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"completion_length": 117.96875,
|
| 147 |
-
"epoch": 0.9142857142857143,
|
| 148 |
-
"grad_norm": 55.910499572753906,
|
| 149 |
-
"kl": 0.029010365306021413,
|
| 150 |
-
"learning_rate": 4.81355307410676e-07,
|
| 151 |
-
"loss": 0.0,
|
| 152 |
-
"reward": 6.665663808584213,
|
| 153 |
-
"reward_std": 0.08780299872159958,
|
| 154 |
-
"rewards/concensus_correctness_reward_func": 1.977749988436699,
|
| 155 |
-
"rewards/consensus_reward_func": 1.875,
|
| 156 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 157 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 158 |
-
"rewards/question_recreation_reward_func": 0.9691639058291912,
|
| 159 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 160 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 161 |
-
"rewards/xmlcount_reward_func": 1.234375,
|
| 162 |
-
"step": 16
|
| 163 |
-
},
|
| 164 |
-
{
|
| 165 |
-
"completion_length": 116.91666666666667,
|
| 166 |
-
"epoch": 1.0,
|
| 167 |
-
"grad_norm": 0.018898706883192062,
|
| 168 |
-
"kl": 0.20654772468454516,
|
| 169 |
-
"learning_rate": 4.747379352713488e-07,
|
| 170 |
-
"loss": 0.0002,
|
| 171 |
-
"reward": 6.508333325386047,
|
| 172 |
-
"reward_std": 0.0,
|
| 173 |
-
"rewards/concensus_correctness_reward_func": 1.7583333055178325,
|
| 174 |
-
"rewards/consensus_reward_func": 2.0,
|
| 175 |
"rewards/cumulative_reward_2": 0.0,
|
| 176 |
"rewards/final_correctness_reward_func": 0.0,
|
| 177 |
-
"rewards/question_recreation_reward_func":
|
| 178 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 179 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 180 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 181 |
-
"step": 18
|
| 182 |
-
},
|
| 183 |
-
{
|
| 184 |
-
"completion_length": 119.65625,
|
| 185 |
-
"epoch": 1.1142857142857143,
|
| 186 |
-
"grad_norm": 131673767936.0,
|
| 187 |
-
"kl": 1484174298.078927,
|
| 188 |
-
"learning_rate": 4.6717793412953776e-07,
|
| 189 |
-
"loss": 1484174.375,
|
| 190 |
-
"reward": 6.440662741661072,
|
| 191 |
-
"reward_std": 0.3215883672237396,
|
| 192 |
-
"rewards/concensus_correctness_reward_func": 1.8583749793469906,
|
| 193 |
-
"rewards/consensus_reward_func": 1.9375,
|
| 194 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 195 |
-
"rewards/final_correctness_reward_func": 0.0,
|
| 196 |
-
"rewards/question_recreation_reward_func": 0.9690064564347267,
|
| 197 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 198 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 199 |
-
"rewards/xmlcount_reward_func": 1.20703125,
|
| 200 |
-
"step": 20
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"completion_length": 113.53125,
|
| 204 |
-
"epoch": 1.2285714285714286,
|
| 205 |
-
"grad_norm": 0.2772182524204254,
|
| 206 |
-
"kl": 0.598538670794369,
|
| 207 |
-
"learning_rate": 4.5870701325731773e-07,
|
| 208 |
-
"loss": 0.0006,
|
| 209 |
-
"reward": 7.0761168003082275,
|
| 210 |
-
"reward_std": 0.06973974592983723,
|
| 211 |
-
"rewards/concensus_correctness_reward_func": 2.1724999845027924,
|
| 212 |
-
"rewards/consensus_reward_func": 1.875,
|
| 213 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 214 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 215 |
-
"rewards/question_recreation_reward_func": 0.9934605993330479,
|
| 216 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 217 |
-
"rewards/strict_format_reward_func": 0.453125,
|
| 218 |
-
"rewards/xmlcount_reward_func": 1.20703125,
|
| 219 |
-
"step": 22
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"completion_length": 111.625,
|
| 223 |
-
"epoch": 1.342857142857143,
|
| 224 |
-
"grad_norm": 0.3359204828739166,
|
| 225 |
-
"kl": 0.025825566772255115,
|
| 226 |
-
"learning_rate": 4.4936070264068016e-07,
|
| 227 |
-
"loss": 0.0,
|
| 228 |
-
"reward": 7.169749945402145,
|
| 229 |
-
"reward_std": 0.0,
|
| 230 |
-
"rewards/concensus_correctness_reward_func": 2.169749990105629,
|
| 231 |
-
"rewards/consensus_reward_func": 2.0,
|
| 232 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 233 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 234 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 235 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 236 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 237 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 238 |
-
"step": 24
|
| 239 |
-
},
|
| 240 |
-
{
|
| 241 |
-
"completion_length": 113.6875,
|
| 242 |
-
"epoch": 1.457142857142857,
|
| 243 |
-
"grad_norm": 0.005876157432794571,
|
| 244 |
-
"kl": 0.0050323337000008905,
|
| 245 |
-
"learning_rate": 4.391782039544238e-07,
|
| 246 |
-
"loss": 0.0,
|
| 247 |
-
"reward": 7.012999922037125,
|
| 248 |
-
"reward_std": 0.0,
|
| 249 |
-
"rewards/concensus_correctness_reward_func": 2.0129999965429306,
|
| 250 |
-
"rewards/consensus_reward_func": 2.0,
|
| 251 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 252 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 253 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 254 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 255 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 256 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 257 |
-
"step": 26
|
| 258 |
-
},
|
| 259 |
-
{
|
| 260 |
-
"completion_length": 121.6875,
|
| 261 |
-
"epoch": 1.5714285714285714,
|
| 262 |
-
"grad_norm": 0.3083041310310364,
|
| 263 |
-
"kl": 0.2358870167809073,
|
| 264 |
-
"learning_rate": 4.282022261367073e-07,
|
| 265 |
-
"loss": 0.0002,
|
| 266 |
-
"reward": 6.6188749223947525,
|
| 267 |
-
"reward_std": 0.11048543266952038,
|
| 268 |
-
"rewards/concensus_correctness_reward_func": 1.946999991312623,
|
| 269 |
-
"rewards/consensus_reward_func": 1.875,
|
| 270 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 271 |
-
"rewards/final_correctness_reward_func": 0.0625,
|
| 272 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 273 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 274 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 275 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 276 |
-
"step": 28
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"completion_length": 117.34375,
|
| 280 |
-
"epoch": 1.6857142857142857,
|
| 281 |
-
"grad_norm": 0.00944291427731514,
|
| 282 |
-
"kl": 0.005240324962869636,
|
| 283 |
-
"learning_rate": 4.1647880625292027e-07,
|
| 284 |
-
"loss": 0.0,
|
| 285 |
-
"reward": 6.917999982833862,
|
| 286 |
-
"reward_std": 0.0,
|
| 287 |
-
"rewards/concensus_correctness_reward_func": 2.0429999604821205,
|
| 288 |
-
"rewards/consensus_reward_func": 2.0,
|
| 289 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 290 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 291 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 292 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 293 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 294 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 295 |
-
"step": 30
|
| 296 |
-
},
|
| 297 |
-
{
|
| 298 |
-
"completion_length": 118.59375,
|
| 299 |
-
"epoch": 1.8,
|
| 300 |
-
"grad_norm": 1.7269436120986938,
|
| 301 |
-
"kl": 0.01583864638087107,
|
| 302 |
-
"learning_rate": 4.040571164002318e-07,
|
| 303 |
-
"loss": 0.0,
|
| 304 |
-
"reward": 7.481499880552292,
|
| 305 |
-
"reward_std": 0.0883883461356163,
|
| 306 |
-
"rewards/concensus_correctness_reward_func": 2.419000007212162,
|
| 307 |
-
"rewards/consensus_reward_func": 1.875,
|
| 308 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 309 |
-
"rewards/final_correctness_reward_func": 0.4375,
|
| 310 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 311 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 312 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 313 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 314 |
-
"step": 32
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"completion_length": 122.4375,
|
| 318 |
-
"epoch": 1.9142857142857141,
|
| 319 |
-
"grad_norm": 0.019974946975708008,
|
| 320 |
-
"kl": 0.048992674514011014,
|
| 321 |
-
"learning_rate": 3.909892574627266e-07,
|
| 322 |
-
"loss": 0.0,
|
| 323 |
-
"reward": 6.924062520265579,
|
| 324 |
-
"reward_std": 0.12153397500514984,
|
| 325 |
-
"rewards/concensus_correctness_reward_func": 1.9162499904632568,
|
| 326 |
-
"rewards/consensus_reward_func": 2.0,
|
| 327 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 328 |
-
"rewards/final_correctness_reward_func": 0.3125,
|
| 329 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 330 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 331 |
-
"rewards/strict_format_reward_func": 0.453125,
|
| 332 |
-
"rewards/xmlcount_reward_func": 1.2421875,
|
| 333 |
-
"step": 34
|
| 334 |
-
},
|
| 335 |
-
{
|
| 336 |
-
"completion_length": 121.91666666666667,
|
| 337 |
-
"epoch": 2.0,
|
| 338 |
-
"grad_norm": 0.011273320764303207,
|
| 339 |
-
"kl": 0.027701640533147536,
|
| 340 |
-
"learning_rate": 3.773300405821908e-07,
|
| 341 |
-
"loss": 0.0,
|
| 342 |
-
"reward": 6.87549106280009,
|
| 343 |
-
"reward_std": 0.033116184175014496,
|
| 344 |
-
"rewards/concensus_correctness_reward_func": 1.9893333216508229,
|
| 345 |
-
"rewards/consensus_reward_func": 2.0,
|
| 346 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 347 |
-
"rewards/final_correctness_reward_func": 0.16666666666666666,
|
| 348 |
-
"rewards/question_recreation_reward_func": 0.9929078022638956,
|
| 349 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 350 |
-
"rewards/strict_format_reward_func": 0.4791666666666667,
|
| 351 |
-
"rewards/xmlcount_reward_func": 1.2474166651566823,
|
| 352 |
-
"step": 36
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"completion_length": 113.9375,
|
| 356 |
-
"epoch": 2.1142857142857143,
|
| 357 |
-
"grad_norm": 0.11432693153619766,
|
| 358 |
-
"kl": 0.04086480487239896,
|
| 359 |
-
"learning_rate": 3.6313675726113475e-07,
|
| 360 |
-
"loss": 0.0,
|
| 361 |
-
"reward": 7.168903559446335,
|
| 362 |
-
"reward_std": 4.442277713678777e-05,
|
| 363 |
-
"rewards/concensus_correctness_reward_func": 2.169124983251095,
|
| 364 |
-
"rewards/consensus_reward_func": 2.0,
|
| 365 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 366 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 367 |
-
"rewards/question_recreation_reward_func": 0.9997786208987236,
|
| 368 |
"rewards/soft_format_reward_func": 0.0,
|
| 369 |
-
"rewards/strict_format_reward_func": 0.
|
| 370 |
-
"rewards/xmlcount_reward_func":
|
| 371 |
-
"step":
|
| 372 |
},
|
| 373 |
{
|
| 374 |
-
"completion_length":
|
| 375 |
-
"epoch": 2.
|
| 376 |
-
"grad_norm":
|
| 377 |
-
"kl": 0.
|
| 378 |
-
"learning_rate":
|
| 379 |
"loss": 0.0,
|
| 380 |
-
"reward":
|
| 381 |
-
"reward_std": 0.
|
| 382 |
-
"rewards/concensus_correctness_reward_func":
|
| 383 |
-
"rewards/consensus_reward_func":
|
| 384 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 385 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 386 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 387 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 388 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 389 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 390 |
-
"step": 40
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"completion_length": 135.125,
|
| 394 |
-
"epoch": 2.342857142857143,
|
| 395 |
-
"grad_norm": 3.4787302017211914,
|
| 396 |
-
"kl": 0.593309123571089,
|
| 397 |
-
"learning_rate": 3.3338810791270517e-07,
|
| 398 |
-
"loss": 0.0006,
|
| 399 |
-
"reward": 7.257124900817871,
|
| 400 |
-
"reward_std": 0.0,
|
| 401 |
-
"rewards/concensus_correctness_reward_func": 2.1321250051259995,
|
| 402 |
-
"rewards/consensus_reward_func": 2.0,
|
| 403 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 404 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 405 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 406 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 407 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 408 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 409 |
-
"step": 42
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"completion_length": 121.03125,
|
| 413 |
-
"epoch": 2.4571428571428573,
|
| 414 |
-
"grad_norm": 1.0152610540390015,
|
| 415 |
-
"kl": 0.24847944835892122,
|
| 416 |
-
"learning_rate": 3.179575180590857e-07,
|
| 417 |
-
"loss": 0.0002,
|
| 418 |
-
"reward": 7.024739623069763,
|
| 419 |
-
"reward_std": 0.011151571234222502,
|
| 420 |
-
"rewards/concensus_correctness_reward_func": 2.0638749971985817,
|
| 421 |
-
"rewards/consensus_reward_func": 2.0,
|
| 422 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 423 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 424 |
-
"rewards/question_recreation_reward_func": 0.9999271556735039,
|
| 425 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 426 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 427 |
-
"rewards/xmlcount_reward_func": 1.2421875,
|
| 428 |
-
"step": 44
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"completion_length": 124.9375,
|
| 432 |
-
"epoch": 2.571428571428571,
|
| 433 |
-
"grad_norm": 7.650447845458984,
|
| 434 |
-
"kl": 0.8674181794049218,
|
| 435 |
-
"learning_rate": 3.022418907578188e-07,
|
| 436 |
-
"loss": 0.0009,
|
| 437 |
-
"reward": 6.307055249810219,
|
| 438 |
-
"reward_std": 0.04419417306780815,
|
| 439 |
-
"rewards/concensus_correctness_reward_func": 1.7134999874979258,
|
| 440 |
-
"rewards/consensus_reward_func": 1.875,
|
| 441 |
"rewards/cumulative_reward_2": 0.0,
|
| 442 |
"rewards/final_correctness_reward_func": 0.0,
|
| 443 |
-
"rewards/question_recreation_reward_func": 0.
|
| 444 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 445 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 446 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 447 |
-
"step": 46
|
| 448 |
-
},
|
| 449 |
-
{
|
| 450 |
-
"completion_length": 120.9375,
|
| 451 |
-
"epoch": 2.685714285714286,
|
| 452 |
-
"grad_norm": 1.830073356628418,
|
| 453 |
-
"kl": 0.1826353092528734,
|
| 454 |
-
"learning_rate": 2.863071428113726e-07,
|
| 455 |
-
"loss": 0.0002,
|
| 456 |
-
"reward": 6.91051921248436,
|
| 457 |
-
"reward_std": 0.011993602442089468,
|
| 458 |
-
"rewards/concensus_correctness_reward_func": 2.0439999848604202,
|
| 459 |
-
"rewards/consensus_reward_func": 2.0,
|
| 460 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 461 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 462 |
-
"rewards/question_recreation_reward_func": 0.9915192350745201,
|
| 463 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 464 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 465 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 466 |
-
"step": 48
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"completion_length": 112.4375,
|
| 470 |
-
"epoch": 2.8,
|
| 471 |
-
"grad_norm": 772774.3125,
|
| 472 |
-
"kl": 61294.761235032754,
|
| 473 |
-
"learning_rate": 2.7022011009035107e-07,
|
| 474 |
-
"loss": 61.2948,
|
| 475 |
-
"reward": 6.738324627280235,
|
| 476 |
-
"reward_std": 0.09737744927406311,
|
| 477 |
-
"rewards/concensus_correctness_reward_func": 2.064999971538782,
|
| 478 |
-
"rewards/consensus_reward_func": 1.75,
|
| 479 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 480 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 481 |
-
"rewards/question_recreation_reward_func": 0.9897308498620987,
|
| 482 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 483 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 484 |
-
"rewards/xmlcount_reward_func": 1.21484375,
|
| 485 |
-
"step": 50
|
| 486 |
-
},
|
| 487 |
-
{
|
| 488 |
-
"completion_length": 113.0,
|
| 489 |
-
"epoch": 2.914285714285714,
|
| 490 |
-
"grad_norm": 0.027667926624417305,
|
| 491 |
-
"kl": 0.02456555592652876,
|
| 492 |
-
"learning_rate": 2.540482672006254e-07,
|
| 493 |
-
"loss": 0.0,
|
| 494 |
-
"reward": 7.299999892711639,
|
| 495 |
-
"reward_std": 0.0,
|
| 496 |
-
"rewards/concensus_correctness_reward_func": 2.1749999970197678,
|
| 497 |
-
"rewards/consensus_reward_func": 2.0,
|
| 498 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 499 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 500 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 501 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 502 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 503 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 504 |
-
"step": 52
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"completion_length": 108.95833333333333,
|
| 508 |
-
"epoch": 3.0,
|
| 509 |
-
"grad_norm": 0.08859401941299438,
|
| 510 |
-
"kl": 0.022353365551680326,
|
| 511 |
-
"learning_rate": 2.37859444471388e-07,
|
| 512 |
-
"loss": 0.0,
|
| 513 |
-
"reward": 6.915333350499471,
|
| 514 |
-
"reward_std": 0.0,
|
| 515 |
-
"rewards/concensus_correctness_reward_func": 1.9986666440963745,
|
| 516 |
-
"rewards/consensus_reward_func": 2.0,
|
| 517 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 518 |
-
"rewards/final_correctness_reward_func": 0.16666666666666666,
|
| 519 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 520 |
"rewards/soft_format_reward_func": 0.0,
|
| 521 |
-
"rewards/strict_format_reward_func": 0.
|
| 522 |
-
"rewards/xmlcount_reward_func":
|
| 523 |
-
"step":
|
| 524 |
-
},
|
| 525 |
-
{
|
| 526 |
-
"completion_length": 113.40625,
|
| 527 |
-
"epoch": 3.1142857142857143,
|
| 528 |
-
"grad_norm": 0.04827781766653061,
|
| 529 |
-
"kl": 0.1669018538814271,
|
| 530 |
-
"learning_rate": 2.2172154345117894e-07,
|
| 531 |
-
"loss": 0.0002,
|
| 532 |
-
"reward": 6.828624963760376,
|
| 533 |
-
"reward_std": 0.0,
|
| 534 |
-
"rewards/concensus_correctness_reward_func": 1.953624963760376,
|
| 535 |
-
"rewards/consensus_reward_func": 2.0,
|
| 536 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 537 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 538 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 539 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 540 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 541 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 542 |
-
"step": 56
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"completion_length": 111.3125,
|
| 546 |
-
"epoch": 3.2285714285714286,
|
| 547 |
-
"grad_norm": 0.08809684216976166,
|
| 548 |
-
"kl": 0.09562112473940942,
|
| 549 |
-
"learning_rate": 2.0570225210519433e-07,
|
| 550 |
-
"loss": 0.0001,
|
| 551 |
-
"reward": 7.075749933719635,
|
| 552 |
-
"reward_std": 0.02863781340420246,
|
| 553 |
-
"rewards/concensus_correctness_reward_func": 2.0960000082850456,
|
| 554 |
-
"rewards/consensus_reward_func": 2.0,
|
| 555 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 556 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 557 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 558 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 559 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 560 |
-
"rewards/xmlcount_reward_func": 1.2453749999403954,
|
| 561 |
-
"step": 58
|
| 562 |
-
},
|
| 563 |
-
{
|
| 564 |
-
"completion_length": 116.65625,
|
| 565 |
-
"epoch": 3.342857142857143,
|
| 566 |
-
"grad_norm": 5.16734504699707,
|
| 567 |
-
"kl": 9.673246745584038,
|
| 568 |
-
"learning_rate": 1.8986876090843664e-07,
|
| 569 |
-
"loss": 0.0097,
|
| 570 |
-
"reward": 6.642811328172684,
|
| 571 |
-
"reward_std": 0.14822890423238277,
|
| 572 |
-
"rewards/concensus_correctness_reward_func": 2.0444999784231186,
|
| 573 |
-
"rewards/consensus_reward_func": 1.75,
|
| 574 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 575 |
-
"rewards/final_correctness_reward_func": 0.1875,
|
| 576 |
-
"rewards/question_recreation_reward_func": 0.9889363348484039,
|
| 577 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 578 |
-
"rewards/strict_format_reward_func": 0.453125,
|
| 579 |
-
"rewards/xmlcount_reward_func": 1.21875,
|
| 580 |
-
"step": 60
|
| 581 |
-
},
|
| 582 |
-
{
|
| 583 |
-
"completion_length": 114.09375,
|
| 584 |
-
"epoch": 3.4571428571428573,
|
| 585 |
-
"grad_norm": 8.348511695861816,
|
| 586 |
-
"kl": 1.3997792335576378,
|
| 587 |
-
"learning_rate": 1.7428748102551234e-07,
|
| 588 |
-
"loss": 0.0014,
|
| 589 |
-
"reward": 6.94431246817112,
|
| 590 |
-
"reward_std": 0.19701763801276684,
|
| 591 |
-
"rewards/concensus_correctness_reward_func": 2.045875007286668,
|
| 592 |
-
"rewards/consensus_reward_func": 1.8125,
|
| 593 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 594 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 595 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 596 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 597 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 598 |
-
"rewards/xmlcount_reward_func": 1.2421875,
|
| 599 |
-
"step": 62
|
| 600 |
-
},
|
| 601 |
-
{
|
| 602 |
-
"completion_length": 118.75,
|
| 603 |
-
"epoch": 3.571428571428571,
|
| 604 |
-
"grad_norm": 28.351816177368164,
|
| 605 |
-
"kl": 1.849376610138279,
|
| 606 |
-
"learning_rate": 1.5902376575912814e-07,
|
| 607 |
-
"loss": 0.0018,
|
| 608 |
-
"reward": 7.588562428951263,
|
| 609 |
-
"reward_std": 0.09943689405918121,
|
| 610 |
-
"rewards/concensus_correctness_reward_func": 2.315124996006489,
|
| 611 |
-
"rewards/consensus_reward_func": 2.0,
|
| 612 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 613 |
-
"rewards/final_correctness_reward_func": 0.5625,
|
| 614 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 615 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 616 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 617 |
-
"rewards/xmlcount_reward_func": 1.2421875,
|
| 618 |
-
"step": 64
|
| 619 |
-
},
|
| 620 |
-
{
|
| 621 |
-
"completion_length": 125.0,
|
| 622 |
-
"epoch": 3.685714285714286,
|
| 623 |
-
"grad_norm": 0.1395985186100006,
|
| 624 |
-
"kl": 0.08357869584142463,
|
| 625 |
-
"learning_rate": 1.4414163643562753e-07,
|
| 626 |
-
"loss": 0.0001,
|
| 627 |
-
"reward": 7.033588498830795,
|
| 628 |
-
"reward_std": 0.04468413017457351,
|
| 629 |
-
"rewards/concensus_correctness_reward_func": 2.0653749853372574,
|
| 630 |
-
"rewards/consensus_reward_func": 2.0,
|
| 631 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 632 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 633 |
-
"rewards/question_recreation_reward_func": 0.999463576823473,
|
| 634 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 635 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 636 |
-
"rewards/xmlcount_reward_func": 1.234375,
|
| 637 |
-
"step": 66
|
| 638 |
},
|
| 639 |
{
|
| 640 |
-
"completion_length":
|
| 641 |
-
"epoch": 3.
|
| 642 |
-
"grad_norm":
|
| 643 |
-
"kl": 0.
|
| 644 |
-
"learning_rate":
|
| 645 |
"loss": 0.0,
|
| 646 |
-
"reward":
|
| 647 |
-
"reward_std": 0.
|
| 648 |
-
"rewards/concensus_correctness_reward_func":
|
| 649 |
-
"rewards/consensus_reward_func":
|
| 650 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 651 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 652 |
-
"rewards/question_recreation_reward_func": 0.9999413713812828,
|
| 653 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 654 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 655 |
-
"rewards/xmlcount_reward_func": 1.23828125,
|
| 656 |
-
"step": 68
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"completion_length": 118.40625,
|
| 660 |
-
"epoch": 3.914285714285714,
|
| 661 |
-
"grad_norm": 0.014703701250255108,
|
| 662 |
-
"kl": 610.1349981201129,
|
| 663 |
-
"learning_rate": 1.1576995658775404e-07,
|
| 664 |
-
"loss": 0.6101,
|
| 665 |
-
"reward": 6.650367766618729,
|
| 666 |
-
"reward_std": 0.025721000507473946,
|
| 667 |
-
"rewards/concensus_correctness_reward_func": 1.9187499731779099,
|
| 668 |
-
"rewards/consensus_reward_func": 2.0,
|
| 669 |
"rewards/cumulative_reward_2": 0.0,
|
| 670 |
"rewards/final_correctness_reward_func": 0.0,
|
| 671 |
-
"rewards/question_recreation_reward_func": 0.
|
| 672 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 673 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 674 |
-
"rewards/xmlcount_reward_func": 1.2474374994635582,
|
| 675 |
-
"step": 70
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"completion_length": 117.20833333333333,
|
| 679 |
-
"epoch": 4.0,
|
| 680 |
-
"grad_norm": 0.14886708557605743,
|
| 681 |
-
"kl": 0.06357832976694529,
|
| 682 |
-
"learning_rate": 1.0239940674851941e-07,
|
| 683 |
-
"loss": 0.0,
|
| 684 |
-
"reward": 6.836241086324056,
|
| 685 |
-
"reward_std": 0.0,
|
| 686 |
-
"rewards/concensus_correctness_reward_func": 1.9266666571299236,
|
| 687 |
-
"rewards/consensus_reward_func": 2.0,
|
| 688 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 689 |
-
"rewards/final_correctness_reward_func": 0.16666666666666666,
|
| 690 |
-
"rewards/question_recreation_reward_func": 0.9929078022638956,
|
| 691 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 692 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 693 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 694 |
-
"step": 72
|
| 695 |
-
},
|
| 696 |
-
{
|
| 697 |
-
"completion_length": 120.96875,
|
| 698 |
-
"epoch": 4.114285714285714,
|
| 699 |
-
"grad_norm": 2.8220157623291016,
|
| 700 |
-
"kl": 0.02625376718060579,
|
| 701 |
-
"learning_rate": 8.964794509221507e-08,
|
| 702 |
-
"loss": 0.0,
|
| 703 |
-
"reward": 6.904249906539917,
|
| 704 |
-
"reward_std": 0.022627420723438263,
|
| 705 |
-
"rewards/concensus_correctness_reward_func": 2.0452499985694885,
|
| 706 |
-
"rewards/consensus_reward_func": 2.0,
|
| 707 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 708 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 709 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 710 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 711 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 712 |
-
"rewards/xmlcount_reward_func": 1.2496249973773956,
|
| 713 |
-
"step": 74
|
| 714 |
-
},
|
| 715 |
-
{
|
| 716 |
-
"completion_length": 112.25,
|
| 717 |
-
"epoch": 4.228571428571429,
|
| 718 |
-
"grad_norm": 13.022274017333984,
|
| 719 |
-
"kl": 0.10997068358119577,
|
| 720 |
-
"learning_rate": 7.756905568047392e-08,
|
| 721 |
-
"loss": 0.0001,
|
| 722 |
-
"reward": 6.940118342638016,
|
| 723 |
-
"reward_std": 0.011048543266952038,
|
| 724 |
-
"rewards/concensus_correctness_reward_func": 2.1251249834895134,
|
| 725 |
-
"rewards/consensus_reward_func": 1.875,
|
| 726 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 727 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 728 |
-
"rewards/question_recreation_reward_func": 0.9946808516979218,
|
| 729 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 730 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 731 |
-
"rewards/xmlcount_reward_func": 1.2265625,
|
| 732 |
-
"step": 76
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"completion_length": 116.78125,
|
| 736 |
-
"epoch": 4.3428571428571425,
|
| 737 |
-
"grad_norm": 0.11638541519641876,
|
| 738 |
-
"kl": 0.03174898200086318,
|
| 739 |
-
"learning_rate": 6.621340157319996e-08,
|
| 740 |
-
"loss": 0.0,
|
| 741 |
-
"reward": 7.31488761305809,
|
| 742 |
-
"reward_std": 0.0003356575034558773,
|
| 743 |
-
"rewards/concensus_correctness_reward_func": 2.190124988555908,
|
| 744 |
-
"rewards/consensus_reward_func": 2.0,
|
| 745 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 746 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 747 |
-
"rewards/question_recreation_reward_func": 0.9997626580297947,
|
| 748 |
"rewards/soft_format_reward_func": 0.0,
|
| 749 |
-
"rewards/strict_format_reward_func": 0.
|
| 750 |
-
"rewards/xmlcount_reward_func":
|
| 751 |
-
"step":
|
| 752 |
},
|
| 753 |
{
|
| 754 |
-
"completion_length":
|
| 755 |
-
"epoch":
|
| 756 |
-
"grad_norm":
|
| 757 |
-
"kl": 0.
|
| 758 |
-
"learning_rate":
|
| 759 |
"loss": 0.0,
|
| 760 |
-
"reward":
|
| 761 |
-
"reward_std": 0.
|
| 762 |
-
"rewards/concensus_correctness_reward_func":
|
| 763 |
-
"rewards/consensus_reward_func":
|
| 764 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 765 |
-
"rewards/final_correctness_reward_func": 0.5,
|
| 766 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 767 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 768 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 769 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 770 |
-
"step": 80
|
| 771 |
-
},
|
| 772 |
-
{
|
| 773 |
-
"completion_length": 130.3125,
|
| 774 |
-
"epoch": 4.571428571428571,
|
| 775 |
-
"grad_norm": 11.376300811767578,
|
| 776 |
-
"kl": 0.08390317361772759,
|
| 777 |
-
"learning_rate": 4.5859084235697235e-08,
|
| 778 |
-
"loss": 0.0001,
|
| 779 |
-
"reward": 6.354187414050102,
|
| 780 |
-
"reward_std": 0.02324613742530346,
|
| 781 |
-
"rewards/concensus_correctness_reward_func": 1.7456249836832285,
|
| 782 |
-
"rewards/consensus_reward_func": 1.875,
|
| 783 |
"rewards/cumulative_reward_2": 0.0,
|
| 784 |
"rewards/final_correctness_reward_func": 0.0,
|
| 785 |
-
"rewards/question_recreation_reward_func":
|
| 786 |
"rewards/soft_format_reward_func": 0.0,
|
| 787 |
-
"rewards/strict_format_reward_func": 0.
|
| 788 |
-
"rewards/xmlcount_reward_func":
|
| 789 |
-
"step":
|
| 790 |
-
},
|
| 791 |
-
{
|
| 792 |
-
"completion_length": 120.53125,
|
| 793 |
-
"epoch": 4.685714285714286,
|
| 794 |
-
"grad_norm": 0.025102941319346428,
|
| 795 |
-
"kl": 654.5053322186013,
|
| 796 |
-
"learning_rate": 3.6945794086007705e-08,
|
| 797 |
-
"loss": 0.6545,
|
| 798 |
-
"reward": 7.0527812242507935,
|
| 799 |
-
"reward_std": 0.10496116429567337,
|
| 800 |
-
"rewards/concensus_correctness_reward_func": 2.0332499966025352,
|
| 801 |
-
"rewards/consensus_reward_func": 1.875,
|
| 802 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 803 |
-
"rewards/final_correctness_reward_func": 0.4375,
|
| 804 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 805 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 806 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 807 |
-
"rewards/xmlcount_reward_func": 1.23828125,
|
| 808 |
-
"step": 84
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"completion_length": 114.5625,
|
| 812 |
-
"epoch": 4.8,
|
| 813 |
-
"grad_norm": 13.666620254516602,
|
| 814 |
-
"kl": 2.7016793186194263,
|
| 815 |
-
"learning_rate": 2.892612731749414e-08,
|
| 816 |
-
"loss": 0.0027,
|
| 817 |
-
"reward": 6.9134474992752075,
|
| 818 |
-
"reward_std": 0.0069684546906501055,
|
| 819 |
-
"rewards/concensus_correctness_reward_func": 2.043374978005886,
|
| 820 |
-
"rewards/consensus_reward_func": 2.0,
|
| 821 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 822 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 823 |
-
"rewards/question_recreation_reward_func": 0.9950725585222244,
|
| 824 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 825 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 826 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 827 |
-
"step": 86
|
| 828 |
-
},
|
| 829 |
-
{
|
| 830 |
-
"completion_length": 114.96875,
|
| 831 |
-
"epoch": 4.914285714285715,
|
| 832 |
-
"grad_norm": 0.09325135499238968,
|
| 833 |
-
"kl": 10.726173710958392,
|
| 834 |
-
"learning_rate": 2.183372119961499e-08,
|
| 835 |
-
"loss": 0.0107,
|
| 836 |
-
"reward": 6.717237025499344,
|
| 837 |
-
"reward_std": 0.2833777070045471,
|
| 838 |
-
"rewards/concensus_correctness_reward_func": 1.9829999767243862,
|
| 839 |
-
"rewards/consensus_reward_func": 1.9375,
|
| 840 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 841 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 842 |
-
"rewards/question_recreation_reward_func": 0.9764245375990868,
|
| 843 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 844 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 845 |
-
"rewards/xmlcount_reward_func": 1.2109375,
|
| 846 |
-
"step": 88
|
| 847 |
-
},
|
| 848 |
-
{
|
| 849 |
-
"completion_length": 113.625,
|
| 850 |
-
"epoch": 5.0,
|
| 851 |
-
"grad_norm": 0.05840963497757912,
|
| 852 |
-
"kl": 0.028356703667668626,
|
| 853 |
-
"learning_rate": 1.5698323748414122e-08,
|
| 854 |
-
"loss": 0.0,
|
| 855 |
-
"reward": 6.843000014623006,
|
| 856 |
-
"reward_std": 0.0,
|
| 857 |
-
"rewards/concensus_correctness_reward_func": 1.9263333181540172,
|
| 858 |
-
"rewards/consensus_reward_func": 2.0,
|
| 859 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 860 |
-
"rewards/final_correctness_reward_func": 0.16666666666666666,
|
| 861 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 862 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 863 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 864 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 865 |
-
"step": 90
|
| 866 |
-
},
|
| 867 |
-
{
|
| 868 |
-
"completion_length": 115.71875,
|
| 869 |
-
"epoch": 5.114285714285714,
|
| 870 |
-
"grad_norm": 20.699186325073242,
|
| 871 |
-
"kl": 0.19849324076494668,
|
| 872 |
-
"learning_rate": 1.054566895300324e-08,
|
| 873 |
-
"loss": 0.0002,
|
| 874 |
-
"reward": 6.53057761490345,
|
| 875 |
-
"reward_std": 0.16040338575839996,
|
| 876 |
-
"rewards/concensus_correctness_reward_func": 1.8652499951422215,
|
| 877 |
-
"rewards/consensus_reward_func": 1.875,
|
| 878 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 879 |
-
"rewards/final_correctness_reward_func": 0.125,
|
| 880 |
-
"rewards/question_recreation_reward_func": 0.9700151830911636,
|
| 881 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 882 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 883 |
-
"rewards/xmlcount_reward_func": 1.2265625,
|
| 884 |
-
"step": 92
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"completion_length": 123.6875,
|
| 888 |
-
"epoch": 5.228571428571429,
|
| 889 |
-
"grad_norm": 8.814971923828125,
|
| 890 |
-
"kl": 1.0889026026070496,
|
| 891 |
-
"learning_rate": 6.397368838268496e-09,
|
| 892 |
-
"loss": 0.0011,
|
| 893 |
-
"reward": 6.928559973835945,
|
| 894 |
-
"reward_std": 0.022097086533904076,
|
| 895 |
-
"rewards/concensus_correctness_reward_func": 2.069374980404973,
|
| 896 |
-
"rewards/consensus_reward_func": 1.875,
|
| 897 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 898 |
-
"rewards/final_correctness_reward_func": 0.25,
|
| 899 |
-
"rewards/question_recreation_reward_func": 0.9998100288212299,
|
| 900 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 901 |
-
"rewards/strict_format_reward_func": 0.484375,
|
| 902 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 903 |
-
"step": 94
|
| 904 |
-
},
|
| 905 |
-
{
|
| 906 |
-
"completion_length": 115.28125,
|
| 907 |
-
"epoch": 5.3428571428571425,
|
| 908 |
-
"grad_norm": 7.976017475128174,
|
| 909 |
-
"kl": 0.14256246562763408,
|
| 910 |
-
"learning_rate": 3.2708228165273244e-09,
|
| 911 |
-
"loss": 0.0001,
|
| 912 |
-
"reward": 6.844999939203262,
|
| 913 |
-
"reward_std": 0.0883883461356163,
|
| 914 |
-
"rewards/concensus_correctness_reward_func": 1.9387499913573265,
|
| 915 |
-
"rewards/consensus_reward_func": 2.0,
|
| 916 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 917 |
-
"rewards/final_correctness_reward_func": 0.1875,
|
| 918 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 919 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 920 |
-
"rewards/strict_format_reward_func": 0.46875,
|
| 921 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 922 |
-
"step": 96
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"completion_length": 114.75,
|
| 926 |
-
"epoch": 5.457142857142857,
|
| 927 |
-
"grad_norm": 0.04377632215619087,
|
| 928 |
-
"kl": 0.06355893767249654,
|
| 929 |
-
"learning_rate": 1.1791447083465133e-09,
|
| 930 |
-
"loss": 0.0001,
|
| 931 |
-
"reward": 7.034624993801117,
|
| 932 |
-
"reward_std": 0.0,
|
| 933 |
-
"rewards/concensus_correctness_reward_func": 2.034624993801117,
|
| 934 |
-
"rewards/consensus_reward_func": 1.875,
|
| 935 |
-
"rewards/cumulative_reward_2": 0.0,
|
| 936 |
-
"rewards/final_correctness_reward_func": 0.375,
|
| 937 |
-
"rewards/question_recreation_reward_func": 1.0,
|
| 938 |
-
"rewards/soft_format_reward_func": 0.0,
|
| 939 |
-
"rewards/strict_format_reward_func": 0.5,
|
| 940 |
-
"rewards/xmlcount_reward_func": 1.25,
|
| 941 |
-
"step": 98
|
| 942 |
},
|
| 943 |
{
|
| 944 |
-
"completion_length":
|
| 945 |
-
"epoch":
|
| 946 |
-
"grad_norm":
|
| 947 |
-
"kl": 0.
|
| 948 |
-
"learning_rate":
|
| 949 |
"loss": 0.0,
|
| 950 |
-
"reward":
|
| 951 |
-
"reward_std": 0.
|
| 952 |
-
"rewards/concensus_correctness_reward_func":
|
| 953 |
-
"rewards/consensus_reward_func":
|
| 954 |
"rewards/cumulative_reward_2": 0.0,
|
| 955 |
-
"rewards/final_correctness_reward_func": 0.
|
| 956 |
-
"rewards/question_recreation_reward_func": 0.
|
| 957 |
"rewards/soft_format_reward_func": 0.0,
|
| 958 |
-
"rewards/strict_format_reward_func": 0.
|
| 959 |
-
"rewards/xmlcount_reward_func":
|
| 960 |
-
"step":
|
| 961 |
},
|
| 962 |
{
|
| 963 |
-
"epoch":
|
| 964 |
-
"step":
|
| 965 |
"total_flos": 0.0,
|
| 966 |
-
"train_loss":
|
| 967 |
-
"train_runtime":
|
| 968 |
-
"train_samples_per_second":
|
| 969 |
-
"train_steps_per_second": 0.
|
| 970 |
}
|
| 971 |
],
|
| 972 |
"logging_steps": 2,
|
| 973 |
-
"max_steps":
|
| 974 |
"num_input_tokens_seen": 0,
|
| 975 |
-
"num_train_epochs":
|
| 976 |
"save_steps": 25,
|
| 977 |
"stateful_callbacks": {
|
| 978 |
"TrainerControl": {
|
|
@@ -987,7 +227,7 @@
|
|
| 987 |
}
|
| 988 |
},
|
| 989 |
"total_flos": 0.0,
|
| 990 |
-
"train_batch_size":
|
| 991 |
"trial_name": null,
|
| 992 |
"trial_params": null
|
| 993 |
}
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 4.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 20,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"completion_length": 294.875,
|
| 14 |
+
"epoch": 0.42105263157894735,
|
| 15 |
+
"grad_norm": 7.388230323791504,
|
| 16 |
+
"kl": 0.0018326228964724578,
|
| 17 |
+
"learning_rate": 5e-07,
|
| 18 |
+
"loss": 0.0,
|
| 19 |
+
"reward": 0.23368853772990406,
|
| 20 |
+
"reward_std": 0.298145150532946,
|
| 21 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 22 |
+
"rewards/consensus_reward_func": 0.0625,
|
| 23 |
"rewards/cumulative_reward_2": 0.0,
|
| 24 |
"rewards/final_correctness_reward_func": 0.0,
|
| 25 |
+
"rewards/question_recreation_reward_func": 0.09978228050749749,
|
| 26 |
"rewards/soft_format_reward_func": 0.0,
|
| 27 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 28 |
+
"rewards/xmlcount_reward_func": 0.07140624802559614,
|
| 29 |
"step": 2
|
| 30 |
},
|
| 31 |
{
|
| 32 |
+
"completion_length": 365.21875,
|
| 33 |
+
"epoch": 0.8421052631578947,
|
| 34 |
+
"grad_norm": 23.853408813476562,
|
| 35 |
+
"kl": 0.005862652775249444,
|
| 36 |
+
"learning_rate": 4.864543104251586e-07,
|
| 37 |
"loss": 0.0,
|
| 38 |
+
"reward": 0.23603129759430885,
|
| 39 |
+
"reward_std": 0.30246378132142127,
|
| 40 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 41 |
+
"rewards/consensus_reward_func": 0.0,
|
| 42 |
"rewards/cumulative_reward_2": 0.0,
|
| 43 |
+
"rewards/final_correctness_reward_func": 0.0,
|
| 44 |
+
"rewards/question_recreation_reward_func": 0.12512504076585174,
|
| 45 |
"rewards/soft_format_reward_func": 0.0,
|
| 46 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 47 |
+
"rewards/xmlcount_reward_func": 0.11090624984353781,
|
| 48 |
"step": 4
|
| 49 |
},
|
| 50 |
{
|
| 51 |
+
"completion_length": 226.0,
|
| 52 |
+
"epoch": 1.2105263157894737,
|
| 53 |
+
"grad_norm": 7.582596778869629,
|
| 54 |
+
"kl": 0.0031690104099522743,
|
| 55 |
+
"learning_rate": 4.472851273490984e-07,
|
| 56 |
"loss": 0.0,
|
| 57 |
+
"reward": 0.16822473385504313,
|
| 58 |
+
"reward_std": 0.5507601164281368,
|
| 59 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 60 |
+
"rewards/consensus_reward_func": 0.0,
|
| 61 |
"rewards/cumulative_reward_2": 0.0,
|
| 62 |
+
"rewards/final_correctness_reward_func": 0.07142857142857142,
|
| 63 |
+
"rewards/question_recreation_reward_func": 0.1537961567352925,
|
| 64 |
"rewards/soft_format_reward_func": 0.0,
|
| 65 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 66 |
+
"rewards/xmlcount_reward_func": -0.05699999790106501,
|
| 67 |
"step": 6
|
| 68 |
},
|
| 69 |
{
|
| 70 |
+
"completion_length": 385.09375,
|
| 71 |
+
"epoch": 1.631578947368421,
|
| 72 |
+
"grad_norm": 2.9946677684783936,
|
| 73 |
+
"kl": 0.001347492725471966,
|
| 74 |
+
"learning_rate": 3.867370395306068e-07,
|
| 75 |
"loss": 0.0,
|
| 76 |
+
"reward": 0.21872268104925752,
|
| 77 |
+
"reward_std": 0.6904346485389397,
|
| 78 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 79 |
+
"rewards/consensus_reward_func": 0.125,
|
| 80 |
"rewards/cumulative_reward_2": 0.0,
|
| 81 |
+
"rewards/final_correctness_reward_func": 0.0,
|
| 82 |
+
"rewards/question_recreation_reward_func": 0.2022851686924696,
|
| 83 |
"rewards/soft_format_reward_func": 0.0,
|
| 84 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 85 |
+
"rewards/xmlcount_reward_func": -0.10856249555945396,
|
| 86 |
"step": 8
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"completion_length": 446.85714285714283,
|
| 90 |
+
"epoch": 2.0,
|
| 91 |
+
"grad_norm": 0.9212186336517334,
|
| 92 |
+
"kl": 0.0022978045121167918,
|
| 93 |
+
"learning_rate": 3.1137137178519977e-07,
|
| 94 |
"loss": 0.0,
|
| 95 |
+
"reward": 0.13073402217456273,
|
| 96 |
+
"reward_std": 0.3137416091880628,
|
| 97 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 98 |
+
"rewards/consensus_reward_func": 0.0,
|
| 99 |
"rewards/cumulative_reward_2": 0.0,
|
| 100 |
+
"rewards/final_correctness_reward_func": 0.0,
|
| 101 |
+
"rewards/question_recreation_reward_func": 0.13244829353477275,
|
| 102 |
"rewards/soft_format_reward_func": 0.0,
|
| 103 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 104 |
+
"rewards/xmlcount_reward_func": -0.001714284930910383,
|
| 105 |
"step": 10
|
| 106 |
},
|
| 107 |
{
|
| 108 |
+
"completion_length": 368.5625,
|
| 109 |
+
"epoch": 2.4210526315789473,
|
| 110 |
+
"grad_norm": 1.7672561407089233,
|
| 111 |
+
"kl": 0.0013554280303651467,
|
| 112 |
+
"learning_rate": 2.2935516363191693e-07,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
"loss": 0.0,
|
| 114 |
+
"reward": -0.1324465568177402,
|
| 115 |
+
"reward_std": 0.6439249363029376,
|
| 116 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 117 |
+
"rewards/consensus_reward_func": 0.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 118 |
"rewards/cumulative_reward_2": 0.0,
|
| 119 |
"rewards/final_correctness_reward_func": 0.0,
|
| 120 |
+
"rewards/question_recreation_reward_func": 0.09061592211946845,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
"rewards/soft_format_reward_func": 0.0,
|
| 122 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 123 |
+
"rewards/xmlcount_reward_func": -0.22306250617839396,
|
| 124 |
+
"step": 12
|
| 125 |
},
|
| 126 |
{
|
| 127 |
+
"completion_length": 217.71875,
|
| 128 |
+
"epoch": 2.8421052631578947,
|
| 129 |
+
"grad_norm": 51.95741653442383,
|
| 130 |
+
"kl": 0.0025395552984264214,
|
| 131 |
+
"learning_rate": 1.4957614383675767e-07,
|
| 132 |
"loss": 0.0,
|
| 133 |
+
"reward": 0.18613064312376082,
|
| 134 |
+
"reward_std": 0.20095090114045888,
|
| 135 |
+
"rewards/concensus_correctness_reward_func": 0.0013749999925494194,
|
| 136 |
+
"rewards/consensus_reward_func": 0.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 137 |
"rewards/cumulative_reward_2": 0.0,
|
| 138 |
"rewards/final_correctness_reward_func": 0.0,
|
| 139 |
+
"rewards/question_recreation_reward_func": 0.1167868955526501,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
"rewards/soft_format_reward_func": 0.0,
|
| 141 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 142 |
+
"rewards/xmlcount_reward_func": 0.06796875037252903,
|
| 143 |
+
"step": 14
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 144 |
},
|
| 145 |
{
|
| 146 |
+
"completion_length": 276.2142857142857,
|
| 147 |
+
"epoch": 3.2105263157894735,
|
| 148 |
+
"grad_norm": 29.23552131652832,
|
| 149 |
+
"kl": 0.0017278649694552378,
|
| 150 |
+
"learning_rate": 8.067960709356478e-08,
|
| 151 |
"loss": 0.0,
|
| 152 |
+
"reward": 0.2764262727328709,
|
| 153 |
+
"reward_std": 0.45395882214818684,
|
| 154 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 155 |
+
"rewards/consensus_reward_func": 0.07142857142857142,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 156 |
"rewards/cumulative_reward_2": 0.0,
|
| 157 |
"rewards/final_correctness_reward_func": 0.0,
|
| 158 |
+
"rewards/question_recreation_reward_func": 0.16878341936639377,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
"rewards/soft_format_reward_func": 0.0,
|
| 160 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 161 |
+
"rewards/xmlcount_reward_func": 0.03621428566319602,
|
| 162 |
+
"step": 16
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"completion_length": 319.46875,
|
| 166 |
+
"epoch": 3.6315789473684212,
|
| 167 |
+
"grad_norm": 52.476932525634766,
|
| 168 |
+
"kl": 0.004187825550616253,
|
| 169 |
+
"learning_rate": 3.013156219837776e-08,
|
| 170 |
"loss": 0.0,
|
| 171 |
+
"reward": 0.058828551205806434,
|
| 172 |
+
"reward_std": 0.11632925440790132,
|
| 173 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 174 |
+
"rewards/consensus_reward_func": 0.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 175 |
"rewards/cumulative_reward_2": 0.0,
|
| 176 |
"rewards/final_correctness_reward_func": 0.0,
|
| 177 |
+
"rewards/question_recreation_reward_func": 0.0568285504123196,
|
| 178 |
"rewards/soft_format_reward_func": 0.0,
|
| 179 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 180 |
+
"rewards/xmlcount_reward_func": 0.0020000003278255463,
|
| 181 |
+
"step": 18
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 182 |
},
|
| 183 |
{
|
| 184 |
+
"completion_length": 451.92857142857144,
|
| 185 |
+
"epoch": 4.0,
|
| 186 |
+
"grad_norm": 6.4608612060546875,
|
| 187 |
+
"kl": 0.0014391274640469679,
|
| 188 |
+
"learning_rate": 3.4096741493194193e-09,
|
| 189 |
"loss": 0.0,
|
| 190 |
+
"reward": 0.09713409947497505,
|
| 191 |
+
"reward_std": 0.34493599299873623,
|
| 192 |
+
"rewards/concensus_correctness_reward_func": 0.0,
|
| 193 |
+
"rewards/consensus_reward_func": 0.0,
|
| 194 |
"rewards/cumulative_reward_2": 0.0,
|
| 195 |
+
"rewards/final_correctness_reward_func": 0.07142857142857142,
|
| 196 |
+
"rewards/question_recreation_reward_func": 0.1079555196421487,
|
| 197 |
"rewards/soft_format_reward_func": 0.0,
|
| 198 |
+
"rewards/strict_format_reward_func": 0.0,
|
| 199 |
+
"rewards/xmlcount_reward_func": -0.08224999904632568,
|
| 200 |
+
"step": 20
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 4.0,
|
| 204 |
+
"step": 20,
|
| 205 |
"total_flos": 0.0,
|
| 206 |
+
"train_loss": 2.47737436893658e-06,
|
| 207 |
+
"train_runtime": 471.1429,
|
| 208 |
+
"train_samples_per_second": 0.679,
|
| 209 |
+
"train_steps_per_second": 0.042
|
| 210 |
}
|
| 211 |
],
|
| 212 |
"logging_steps": 2,
|
| 213 |
+
"max_steps": 20,
|
| 214 |
"num_input_tokens_seen": 0,
|
| 215 |
+
"num_train_epochs": 5,
|
| 216 |
"save_steps": 25,
|
| 217 |
"stateful_callbacks": {
|
| 218 |
"TrainerControl": {
|
|
|
|
| 227 |
}
|
| 228 |
},
|
| 229 |
"total_flos": 0.0,
|
| 230 |
+
"train_batch_size": 4,
|
| 231 |
"trial_name": null,
|
| 232 |
"trial_params": null
|
| 233 |
}
|
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 sha256:590cd41cfd14a5b1b8e84d269c5bcc6f22c8809140576b81ac98d0782e5a6e5e
|
| 3 |
+
size 6008
|