End of training
Browse files- README.md +3 -2
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +327 -0
- training_loss.png +0 -0
- training_rewards_accuracies.png +0 -0
README.md
CHANGED
|
@@ -3,9 +3,10 @@ library_name: peft
|
|
| 3 |
license: llama3.1
|
| 4 |
base_model: meta-llama/Llama-3.1-8B-Instruct
|
| 5 |
tags:
|
|
|
|
|
|
|
| 6 |
- trl
|
| 7 |
- dpo
|
| 8 |
-
- llama-factory
|
| 9 |
- generated_from_trainer
|
| 10 |
model-index:
|
| 11 |
- name: Llama-3.1-8B-Instruct_dpo_sg_values
|
|
@@ -17,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 17 |
|
| 18 |
# Llama-3.1-8B-Instruct_dpo_sg_values
|
| 19 |
|
| 20 |
-
This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 3 |
license: llama3.1
|
| 4 |
base_model: meta-llama/Llama-3.1-8B-Instruct
|
| 5 |
tags:
|
| 6 |
+
- llama-factory
|
| 7 |
+
- lora
|
| 8 |
- trl
|
| 9 |
- dpo
|
|
|
|
| 10 |
- generated_from_trainer
|
| 11 |
model-index:
|
| 12 |
- name: Llama-3.1-8B-Instruct_dpo_sg_values
|
|
|
|
| 18 |
|
| 19 |
# Llama-3.1-8B-Instruct_dpo_sg_values
|
| 20 |
|
| 21 |
+
This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the dpo_sg_values dataset.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996261682242991,
|
| 3 |
+
"total_flos": 1.8353312732676096e+17,
|
| 4 |
+
"train_loss": 0.2906364963049307,
|
| 5 |
+
"train_runtime": 4209.7283,
|
| 6 |
+
"train_samples_per_second": 2.542,
|
| 7 |
+
"train_steps_per_second": 0.318
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996261682242991,
|
| 3 |
+
"total_flos": 1.8353312732676096e+17,
|
| 4 |
+
"train_loss": 0.2906364963049307,
|
| 5 |
+
"train_runtime": 4209.7283,
|
| 6 |
+
"train_samples_per_second": 2.542,
|
| 7 |
+
"train_steps_per_second": 0.318
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,327 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9996261682242991,
|
| 5 |
+
"eval_steps": 134,
|
| 6 |
+
"global_step": 1337,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05009345794392523,
|
| 13 |
+
"grad_norm": 1.3687894344329834,
|
| 14 |
+
"learning_rate": 5e-07,
|
| 15 |
+
"logits/chosen": -0.6576648950576782,
|
| 16 |
+
"logits/rejected": -0.6672008633613586,
|
| 17 |
+
"logps/chosen": -2.227020502090454,
|
| 18 |
+
"logps/rejected": -5.950113296508789,
|
| 19 |
+
"loss": 0.6927,
|
| 20 |
+
"rewards/accuracies": 0.5,
|
| 21 |
+
"rewards/chosen": 0.0005052318447269499,
|
| 22 |
+
"rewards/margins": 0.0008711821283213794,
|
| 23 |
+
"rewards/rejected": -0.0003659502835944295,
|
| 24 |
+
"step": 67
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.10018691588785046,
|
| 28 |
+
"grad_norm": 2.2226035594940186,
|
| 29 |
+
"learning_rate": 1e-06,
|
| 30 |
+
"logits/chosen": -0.643012285232544,
|
| 31 |
+
"logits/rejected": -0.6527656316757202,
|
| 32 |
+
"logps/chosen": -2.1846628189086914,
|
| 33 |
+
"logps/rejected": -6.394010066986084,
|
| 34 |
+
"loss": 0.6883,
|
| 35 |
+
"rewards/accuracies": 0.7276118993759155,
|
| 36 |
+
"rewards/chosen": -0.0008213059627451003,
|
| 37 |
+
"rewards/margins": 0.00972337368875742,
|
| 38 |
+
"rewards/rejected": -0.01054468099027872,
|
| 39 |
+
"step": 134
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.1502803738317757,
|
| 43 |
+
"grad_norm": 2.133711576461792,
|
| 44 |
+
"learning_rate": 9.443059019118869e-07,
|
| 45 |
+
"logits/chosen": -0.6600874662399292,
|
| 46 |
+
"logits/rejected": -0.6701396703720093,
|
| 47 |
+
"logps/chosen": -2.2199361324310303,
|
| 48 |
+
"logps/rejected": -7.209847450256348,
|
| 49 |
+
"loss": 0.6524,
|
| 50 |
+
"rewards/accuracies": 0.8563432693481445,
|
| 51 |
+
"rewards/chosen": -0.0032122363336384296,
|
| 52 |
+
"rewards/margins": 0.08549854904413223,
|
| 53 |
+
"rewards/rejected": -0.08871078491210938,
|
| 54 |
+
"step": 201
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.20037383177570092,
|
| 58 |
+
"grad_norm": 3.0340538024902344,
|
| 59 |
+
"learning_rate": 8.886118038237738e-07,
|
| 60 |
+
"logits/chosen": -0.6653879284858704,
|
| 61 |
+
"logits/rejected": -0.677828848361969,
|
| 62 |
+
"logps/chosen": -2.533212184906006,
|
| 63 |
+
"logps/rejected": -10.03713321685791,
|
| 64 |
+
"loss": 0.5533,
|
| 65 |
+
"rewards/accuracies": 0.8488805890083313,
|
| 66 |
+
"rewards/chosen": -0.02541249990463257,
|
| 67 |
+
"rewards/margins": 0.32844945788383484,
|
| 68 |
+
"rewards/rejected": -0.3538619577884674,
|
| 69 |
+
"step": 268
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.2504672897196262,
|
| 73 |
+
"grad_norm": 2.396055221557617,
|
| 74 |
+
"learning_rate": 8.329177057356608e-07,
|
| 75 |
+
"logits/chosen": -0.6770755052566528,
|
| 76 |
+
"logits/rejected": -0.6937317848205566,
|
| 77 |
+
"logps/chosen": -3.257599353790283,
|
| 78 |
+
"logps/rejected": -14.972702980041504,
|
| 79 |
+
"loss": 0.4397,
|
| 80 |
+
"rewards/accuracies": 0.8526118993759155,
|
| 81 |
+
"rewards/chosen": -0.10955464839935303,
|
| 82 |
+
"rewards/margins": 0.7154061198234558,
|
| 83 |
+
"rewards/rejected": -0.8249607682228088,
|
| 84 |
+
"step": 335
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.3005607476635514,
|
| 88 |
+
"grad_norm": 2.1619019508361816,
|
| 89 |
+
"learning_rate": 7.772236076475478e-07,
|
| 90 |
+
"logits/chosen": -0.7039727568626404,
|
| 91 |
+
"logits/rejected": -0.7222999334335327,
|
| 92 |
+
"logps/chosen": -3.5498507022857666,
|
| 93 |
+
"logps/rejected": -20.35832405090332,
|
| 94 |
+
"loss": 0.3224,
|
| 95 |
+
"rewards/accuracies": 0.9029850363731384,
|
| 96 |
+
"rewards/chosen": -0.15284313261508942,
|
| 97 |
+
"rewards/margins": 1.2502268552780151,
|
| 98 |
+
"rewards/rejected": -1.4030699729919434,
|
| 99 |
+
"step": 402
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.3506542056074766,
|
| 103 |
+
"grad_norm": 4.72946834564209,
|
| 104 |
+
"learning_rate": 7.215295095594347e-07,
|
| 105 |
+
"logits/chosen": -0.7181271910667419,
|
| 106 |
+
"logits/rejected": -0.7374780774116516,
|
| 107 |
+
"logps/chosen": -5.401639461517334,
|
| 108 |
+
"logps/rejected": -25.78311538696289,
|
| 109 |
+
"loss": 0.2782,
|
| 110 |
+
"rewards/accuracies": 0.9048507213592529,
|
| 111 |
+
"rewards/chosen": -0.3204249441623688,
|
| 112 |
+
"rewards/margins": 1.6534833908081055,
|
| 113 |
+
"rewards/rejected": -1.9739083051681519,
|
| 114 |
+
"step": 469
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.40074766355140184,
|
| 118 |
+
"grad_norm": 0.9766824245452881,
|
| 119 |
+
"learning_rate": 6.658354114713217e-07,
|
| 120 |
+
"logits/chosen": -0.756517231464386,
|
| 121 |
+
"logits/rejected": -0.7776155471801758,
|
| 122 |
+
"logps/chosen": -5.15533971786499,
|
| 123 |
+
"logps/rejected": -30.16547203063965,
|
| 124 |
+
"loss": 0.2283,
|
| 125 |
+
"rewards/accuracies": 0.9123134016990662,
|
| 126 |
+
"rewards/chosen": -0.3061079978942871,
|
| 127 |
+
"rewards/margins": 2.072856903076172,
|
| 128 |
+
"rewards/rejected": -2.37896466255188,
|
| 129 |
+
"step": 536
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.4508411214953271,
|
| 133 |
+
"grad_norm": 4.220199108123779,
|
| 134 |
+
"learning_rate": 6.101413133832086e-07,
|
| 135 |
+
"logits/chosen": -0.7682312726974487,
|
| 136 |
+
"logits/rejected": -0.7901257872581482,
|
| 137 |
+
"logps/chosen": -6.049707412719727,
|
| 138 |
+
"logps/rejected": -31.82540512084961,
|
| 139 |
+
"loss": 0.2276,
|
| 140 |
+
"rewards/accuracies": 0.9048507213592529,
|
| 141 |
+
"rewards/chosen": -0.37660378217697144,
|
| 142 |
+
"rewards/margins": 2.2089908123016357,
|
| 143 |
+
"rewards/rejected": -2.585594415664673,
|
| 144 |
+
"step": 603
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.5009345794392523,
|
| 148 |
+
"grad_norm": 6.615551471710205,
|
| 149 |
+
"learning_rate": 5.544472152950955e-07,
|
| 150 |
+
"logits/chosen": -0.7774823904037476,
|
| 151 |
+
"logits/rejected": -0.8002229332923889,
|
| 152 |
+
"logps/chosen": -6.285848617553711,
|
| 153 |
+
"logps/rejected": -35.35755157470703,
|
| 154 |
+
"loss": 0.1959,
|
| 155 |
+
"rewards/accuracies": 0.9197760820388794,
|
| 156 |
+
"rewards/chosen": -0.40754997730255127,
|
| 157 |
+
"rewards/margins": 2.4701759815216064,
|
| 158 |
+
"rewards/rejected": -2.8777260780334473,
|
| 159 |
+
"step": 670
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.5510280373831775,
|
| 163 |
+
"grad_norm": 3.6465256214141846,
|
| 164 |
+
"learning_rate": 4.987531172069825e-07,
|
| 165 |
+
"logits/chosen": -0.7746462821960449,
|
| 166 |
+
"logits/rejected": -0.7996317148208618,
|
| 167 |
+
"logps/chosen": -7.126999378204346,
|
| 168 |
+
"logps/rejected": -37.35933303833008,
|
| 169 |
+
"loss": 0.1813,
|
| 170 |
+
"rewards/accuracies": 0.9309701323509216,
|
| 171 |
+
"rewards/chosen": -0.4826628267765045,
|
| 172 |
+
"rewards/margins": 2.6208343505859375,
|
| 173 |
+
"rewards/rejected": -3.10349702835083,
|
| 174 |
+
"step": 737
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.6011214953271028,
|
| 178 |
+
"grad_norm": 9.422818183898926,
|
| 179 |
+
"learning_rate": 4.4305901911886947e-07,
|
| 180 |
+
"logits/chosen": -0.7760910391807556,
|
| 181 |
+
"logits/rejected": -0.802481472492218,
|
| 182 |
+
"logps/chosen": -6.253903388977051,
|
| 183 |
+
"logps/rejected": -37.6789436340332,
|
| 184 |
+
"loss": 0.1769,
|
| 185 |
+
"rewards/accuracies": 0.9402984976768494,
|
| 186 |
+
"rewards/chosen": -0.3954426348209381,
|
| 187 |
+
"rewards/margins": 2.7266414165496826,
|
| 188 |
+
"rewards/rejected": -3.122084140777588,
|
| 189 |
+
"step": 804
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.6512149532710281,
|
| 193 |
+
"grad_norm": 6.800887107849121,
|
| 194 |
+
"learning_rate": 3.873649210307564e-07,
|
| 195 |
+
"logits/chosen": -0.7794383764266968,
|
| 196 |
+
"logits/rejected": -0.8046081066131592,
|
| 197 |
+
"logps/chosen": -6.785355567932129,
|
| 198 |
+
"logps/rejected": -38.823604583740234,
|
| 199 |
+
"loss": 0.1737,
|
| 200 |
+
"rewards/accuracies": 0.9291044473648071,
|
| 201 |
+
"rewards/chosen": -0.43714043498039246,
|
| 202 |
+
"rewards/margins": 2.833800792694092,
|
| 203 |
+
"rewards/rejected": -3.2709412574768066,
|
| 204 |
+
"step": 871
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.7013084112149532,
|
| 208 |
+
"grad_norm": 4.8703718185424805,
|
| 209 |
+
"learning_rate": 3.3167082294264335e-07,
|
| 210 |
+
"logits/chosen": -0.7899920344352722,
|
| 211 |
+
"logits/rejected": -0.8165501952171326,
|
| 212 |
+
"logps/chosen": -7.20438814163208,
|
| 213 |
+
"logps/rejected": -39.9392204284668,
|
| 214 |
+
"loss": 0.1891,
|
| 215 |
+
"rewards/accuracies": 0.9272387623786926,
|
| 216 |
+
"rewards/chosen": -0.47980982065200806,
|
| 217 |
+
"rewards/margins": 2.8853275775909424,
|
| 218 |
+
"rewards/rejected": -3.3651373386383057,
|
| 219 |
+
"step": 938
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.7514018691588785,
|
| 223 |
+
"grad_norm": 0.2955164611339569,
|
| 224 |
+
"learning_rate": 2.7597672485453034e-07,
|
| 225 |
+
"logits/chosen": -0.7796534895896912,
|
| 226 |
+
"logits/rejected": -0.8110276460647583,
|
| 227 |
+
"logps/chosen": -4.945315361022949,
|
| 228 |
+
"logps/rejected": -41.536590576171875,
|
| 229 |
+
"loss": 0.1215,
|
| 230 |
+
"rewards/accuracies": 0.9570895433425903,
|
| 231 |
+
"rewards/chosen": -0.2934176027774811,
|
| 232 |
+
"rewards/margins": 3.1789510250091553,
|
| 233 |
+
"rewards/rejected": -3.4723687171936035,
|
| 234 |
+
"step": 1005
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.8014953271028037,
|
| 238 |
+
"grad_norm": 6.8209075927734375,
|
| 239 |
+
"learning_rate": 2.2028262676641728e-07,
|
| 240 |
+
"logits/chosen": -0.7921628952026367,
|
| 241 |
+
"logits/rejected": -0.8196142315864563,
|
| 242 |
+
"logps/chosen": -6.230679512023926,
|
| 243 |
+
"logps/rejected": -41.968841552734375,
|
| 244 |
+
"loss": 0.1502,
|
| 245 |
+
"rewards/accuracies": 0.9477611780166626,
|
| 246 |
+
"rewards/chosen": -0.4138028919696808,
|
| 247 |
+
"rewards/margins": 3.128687858581543,
|
| 248 |
+
"rewards/rejected": -3.5424907207489014,
|
| 249 |
+
"step": 1072
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.851588785046729,
|
| 253 |
+
"grad_norm": 6.143770217895508,
|
| 254 |
+
"learning_rate": 1.6458852867830422e-07,
|
| 255 |
+
"logits/chosen": -0.7991640567779541,
|
| 256 |
+
"logits/rejected": -0.8256679773330688,
|
| 257 |
+
"logps/chosen": -6.564340114593506,
|
| 258 |
+
"logps/rejected": -42.35133361816406,
|
| 259 |
+
"loss": 0.1459,
|
| 260 |
+
"rewards/accuracies": 0.9402984976768494,
|
| 261 |
+
"rewards/chosen": -0.44414621591567993,
|
| 262 |
+
"rewards/margins": 3.182438611984253,
|
| 263 |
+
"rewards/rejected": -3.626584529876709,
|
| 264 |
+
"step": 1139
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.9016822429906542,
|
| 268 |
+
"grad_norm": 5.307474613189697,
|
| 269 |
+
"learning_rate": 1.0889443059019118e-07,
|
| 270 |
+
"logits/chosen": -0.7985506057739258,
|
| 271 |
+
"logits/rejected": -0.8279980421066284,
|
| 272 |
+
"logps/chosen": -5.217541217803955,
|
| 273 |
+
"logps/rejected": -43.41652297973633,
|
| 274 |
+
"loss": 0.1275,
|
| 275 |
+
"rewards/accuracies": 0.9496268630027771,
|
| 276 |
+
"rewards/chosen": -0.3293641209602356,
|
| 277 |
+
"rewards/margins": 3.362889289855957,
|
| 278 |
+
"rewards/rejected": -3.692253351211548,
|
| 279 |
+
"step": 1206
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.9517757009345794,
|
| 283 |
+
"grad_norm": 5.2688984870910645,
|
| 284 |
+
"learning_rate": 5.320033250207814e-08,
|
| 285 |
+
"logits/chosen": -0.7938746809959412,
|
| 286 |
+
"logits/rejected": -0.8232108354568481,
|
| 287 |
+
"logps/chosen": -5.571805477142334,
|
| 288 |
+
"logps/rejected": -43.23577880859375,
|
| 289 |
+
"loss": 0.1313,
|
| 290 |
+
"rewards/accuracies": 0.9589552283287048,
|
| 291 |
+
"rewards/chosen": -0.33297136425971985,
|
| 292 |
+
"rewards/margins": 3.3113536834716797,
|
| 293 |
+
"rewards/rejected": -3.644325017929077,
|
| 294 |
+
"step": 1273
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.9996261682242991,
|
| 298 |
+
"step": 1337,
|
| 299 |
+
"total_flos": 1.8353312732676096e+17,
|
| 300 |
+
"train_loss": 0.2906364963049307,
|
| 301 |
+
"train_runtime": 4209.7283,
|
| 302 |
+
"train_samples_per_second": 2.542,
|
| 303 |
+
"train_steps_per_second": 0.318
|
| 304 |
+
}
|
| 305 |
+
],
|
| 306 |
+
"logging_steps": 67,
|
| 307 |
+
"max_steps": 1337,
|
| 308 |
+
"num_input_tokens_seen": 0,
|
| 309 |
+
"num_train_epochs": 1,
|
| 310 |
+
"save_steps": 500,
|
| 311 |
+
"stateful_callbacks": {
|
| 312 |
+
"TrainerControl": {
|
| 313 |
+
"args": {
|
| 314 |
+
"should_epoch_stop": false,
|
| 315 |
+
"should_evaluate": false,
|
| 316 |
+
"should_log": false,
|
| 317 |
+
"should_save": true,
|
| 318 |
+
"should_training_stop": true
|
| 319 |
+
},
|
| 320 |
+
"attributes": {}
|
| 321 |
+
}
|
| 322 |
+
},
|
| 323 |
+
"total_flos": 1.8353312732676096e+17,
|
| 324 |
+
"train_batch_size": 2,
|
| 325 |
+
"trial_name": null,
|
| 326 |
+
"trial_params": null
|
| 327 |
+
}
|
training_loss.png
ADDED
|
training_rewards_accuracies.png
ADDED
|