Model save
Browse files- README.md +1 -21
- adapter_config.json +4 -4
- adapter_model.safetensors +1 -1
- all_results.json +6 -19
- train_results.json +6 -6
- trainer_state.json +15 -737
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -2,13 +2,9 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
-
- alignment-handbook
|
| 6 |
-
- generated_from_trainer
|
| 7 |
- trl
|
| 8 |
- dpo
|
| 9 |
- generated_from_trainer
|
| 10 |
-
datasets:
|
| 11 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
| 12 |
base_model: mistralai/Mistral-7B-v0.1
|
| 13 |
model-index:
|
| 14 |
- name: zephyr-7b-dpo-qlora
|
|
@@ -20,17 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 20 |
|
| 21 |
# zephyr-7b-dpo-qlora
|
| 22 |
|
| 23 |
-
This model is a fine-tuned version of [
|
| 24 |
-
It achieves the following results on the evaluation set:
|
| 25 |
-
- Loss: 1721.1201
|
| 26 |
-
- Rewards/chosen: -0.0627
|
| 27 |
-
- Rewards/rejected: -0.2250
|
| 28 |
-
- Rewards/accuracies: 0.7738
|
| 29 |
-
- Rewards/margins: 0.1623
|
| 30 |
-
- Logps/rejected: -267.2721
|
| 31 |
-
- Logps/chosen: -271.2979
|
| 32 |
-
- Logits/rejected: -2.0354
|
| 33 |
-
- Logits/chosen: -2.0918
|
| 34 |
|
| 35 |
## Model description
|
| 36 |
|
|
@@ -65,12 +51,6 @@ The following hyperparameters were used during training:
|
|
| 65 |
|
| 66 |
### Training results
|
| 67 |
|
| 68 |
-
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 69 |
-
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 70 |
-
| 1797.9404 | 0.21 | 100 | 1887.4103 | 0.0131 | -0.1197 | 0.7520 | 0.1328 | -256.7424 | -263.7133 | -2.1486 | -2.1969 |
|
| 71 |
-
| 1700.9055 | 0.42 | 200 | 1784.6598 | -0.0464 | -0.2062 | 0.7619 | 0.1598 | -265.3905 | -269.6655 | -2.1081 | -2.1618 |
|
| 72 |
-
| 1767.2219 | 0.63 | 300 | 1735.5183 | -0.0467 | -0.2001 | 0.7698 | 0.1534 | -264.7795 | -269.6956 | -2.1057 | -2.1587 |
|
| 73 |
-
| 1717.4336 | 0.84 | 400 | 1721.6765 | -0.0691 | -0.2309 | 0.7718 | 0.1618 | -267.8569 | -271.9333 | -2.0322 | -2.0885 |
|
| 74 |
|
| 75 |
|
| 76 |
### Framework versions
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- trl
|
| 6 |
- dpo
|
| 7 |
- generated_from_trainer
|
|
|
|
|
|
|
| 8 |
base_model: mistralai/Mistral-7B-v0.1
|
| 9 |
model-index:
|
| 10 |
- name: zephyr-7b-dpo-qlora
|
|
|
|
| 16 |
|
| 17 |
# zephyr-7b-dpo-qlora
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
|
|
|
| 51 |
|
| 52 |
### Training results
|
| 53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
|
| 55 |
|
| 56 |
### Framework versions
|
adapter_config.json
CHANGED
|
@@ -20,12 +20,12 @@
|
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
| 22 |
"q_proj",
|
| 23 |
-
"
|
| 24 |
-
"up_proj",
|
| 25 |
"k_proj",
|
| 26 |
-
"down_proj",
|
| 27 |
"o_proj",
|
| 28 |
-
"
|
|
|
|
|
|
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM"
|
| 31 |
}
|
|
|
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
| 22 |
"q_proj",
|
| 23 |
+
"v_proj",
|
|
|
|
| 24 |
"k_proj",
|
|
|
|
| 25 |
"o_proj",
|
| 26 |
+
"up_proj",
|
| 27 |
+
"gate_proj",
|
| 28 |
+
"down_proj"
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM"
|
| 31 |
}
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52b42f1363dc4a7b4a73efab14d317e8879e92c4e95ca6b669018483353b9ad3
|
| 3 |
size 671150064
|
all_results.json
CHANGED
|
@@ -1,21 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
| 8 |
-
"eval_rewards/accuracies": 0.773809552192688,
|
| 9 |
-
"eval_rewards/chosen": -0.06273359060287476,
|
| 10 |
-
"eval_rewards/margins": 0.16227789223194122,
|
| 11 |
-
"eval_rewards/rejected": -0.22501146793365479,
|
| 12 |
-
"eval_runtime": 548.8776,
|
| 13 |
-
"eval_samples": 2000,
|
| 14 |
-
"eval_samples_per_second": 3.644,
|
| 15 |
-
"eval_steps_per_second": 0.115,
|
| 16 |
-
"train_loss": 1826.8015694608227,
|
| 17 |
-
"train_runtime": 32379.7062,
|
| 18 |
-
"train_samples": 61135,
|
| 19 |
-
"train_samples_per_second": 1.888,
|
| 20 |
-
"train_steps_per_second": 0.015
|
| 21 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.82,
|
| 3 |
+
"train_loss": 0.6931473016738892,
|
| 4 |
+
"train_runtime": 124.4347,
|
| 5 |
+
"train_samples": 305,
|
| 6 |
+
"train_samples_per_second": 2.451,
|
| 7 |
+
"train_steps_per_second": 0.016
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.82,
|
| 3 |
+
"train_loss": 0.6931473016738892,
|
| 4 |
+
"train_runtime": 124.4347,
|
| 5 |
+
"train_samples": 305,
|
| 6 |
+
"train_samples_per_second": 2.451,
|
| 7 |
+
"train_steps_per_second": 0.016
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,761 +1,39 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 100,
|
| 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 |
-
"epoch": 0.
|
| 13 |
-
"learning_rate":
|
| 14 |
-
"logits/chosen": -2.
|
| 15 |
-
"logits/rejected": -2.
|
| 16 |
-
"logps/chosen": -
|
| 17 |
-
"logps/rejected": -
|
| 18 |
-
"loss":
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
| 22 |
"rewards/rejected": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
-
{
|
| 26 |
-
"epoch": 0.02,
|
| 27 |
-
"learning_rate": 1.0416666666666667e-06,
|
| 28 |
-
"logits/chosen": -2.585383176803589,
|
| 29 |
-
"logits/rejected": -2.6190898418426514,
|
| 30 |
-
"logps/chosen": -265.6199035644531,
|
| 31 |
-
"logps/rejected": -261.3590393066406,
|
| 32 |
-
"loss": 2489.4685,
|
| 33 |
-
"rewards/accuracies": 0.4548611044883728,
|
| 34 |
-
"rewards/chosen": 0.006730278953909874,
|
| 35 |
-
"rewards/margins": 0.0007296364055946469,
|
| 36 |
-
"rewards/rejected": 0.006000642664730549,
|
| 37 |
-
"step": 10
|
| 38 |
-
},
|
| 39 |
-
{
|
| 40 |
-
"epoch": 0.04,
|
| 41 |
-
"learning_rate": 2.0833333333333334e-06,
|
| 42 |
-
"logits/chosen": -2.616151809692383,
|
| 43 |
-
"logits/rejected": -2.599904775619507,
|
| 44 |
-
"logps/chosen": -253.3858184814453,
|
| 45 |
-
"logps/rejected": -245.82345581054688,
|
| 46 |
-
"loss": 2411.3754,
|
| 47 |
-
"rewards/accuracies": 0.6000000238418579,
|
| 48 |
-
"rewards/chosen": 0.036651305854320526,
|
| 49 |
-
"rewards/margins": 0.009106594137847424,
|
| 50 |
-
"rewards/rejected": 0.02754470705986023,
|
| 51 |
-
"step": 20
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.06,
|
| 55 |
-
"learning_rate": 3.125e-06,
|
| 56 |
-
"logits/chosen": -2.617845058441162,
|
| 57 |
-
"logits/rejected": -2.6118521690368652,
|
| 58 |
-
"logps/chosen": -250.7469482421875,
|
| 59 |
-
"logps/rejected": -223.05172729492188,
|
| 60 |
-
"loss": 2306.1311,
|
| 61 |
-
"rewards/accuracies": 0.671875,
|
| 62 |
-
"rewards/chosen": 0.04942930489778519,
|
| 63 |
-
"rewards/margins": 0.023983022198081017,
|
| 64 |
-
"rewards/rejected": 0.02544628083705902,
|
| 65 |
-
"step": 30
|
| 66 |
-
},
|
| 67 |
-
{
|
| 68 |
-
"epoch": 0.08,
|
| 69 |
-
"learning_rate": 4.166666666666667e-06,
|
| 70 |
-
"logits/chosen": -2.6323208808898926,
|
| 71 |
-
"logits/rejected": -2.608524799346924,
|
| 72 |
-
"logps/chosen": -276.45947265625,
|
| 73 |
-
"logps/rejected": -238.35391235351562,
|
| 74 |
-
"loss": 2100.6182,
|
| 75 |
-
"rewards/accuracies": 0.6968749761581421,
|
| 76 |
-
"rewards/chosen": 0.05112973973155022,
|
| 77 |
-
"rewards/margins": 0.05378426983952522,
|
| 78 |
-
"rewards/rejected": -0.002654529409483075,
|
| 79 |
-
"step": 40
|
| 80 |
-
},
|
| 81 |
-
{
|
| 82 |
-
"epoch": 0.1,
|
| 83 |
-
"learning_rate": 4.999731868769027e-06,
|
| 84 |
-
"logits/chosen": -2.552873373031616,
|
| 85 |
-
"logits/rejected": -2.5477213859558105,
|
| 86 |
-
"logps/chosen": -253.2111358642578,
|
| 87 |
-
"logps/rejected": -248.1074676513672,
|
| 88 |
-
"loss": 2103.8223,
|
| 89 |
-
"rewards/accuracies": 0.6781250238418579,
|
| 90 |
-
"rewards/chosen": 0.022746428847312927,
|
| 91 |
-
"rewards/margins": 0.07937721163034439,
|
| 92 |
-
"rewards/rejected": -0.05663077160716057,
|
| 93 |
-
"step": 50
|
| 94 |
-
},
|
| 95 |
-
{
|
| 96 |
-
"epoch": 0.13,
|
| 97 |
-
"learning_rate": 4.9903533134293035e-06,
|
| 98 |
-
"logits/chosen": -2.556926727294922,
|
| 99 |
-
"logits/rejected": -2.551504611968994,
|
| 100 |
-
"logps/chosen": -261.6982116699219,
|
| 101 |
-
"logps/rejected": -240.27059936523438,
|
| 102 |
-
"loss": 2054.3434,
|
| 103 |
-
"rewards/accuracies": 0.6781250238418579,
|
| 104 |
-
"rewards/chosen": 0.023721303790807724,
|
| 105 |
-
"rewards/margins": 0.08682042360305786,
|
| 106 |
-
"rewards/rejected": -0.06309913098812103,
|
| 107 |
-
"step": 60
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 0.15,
|
| 111 |
-
"learning_rate": 4.967625656594782e-06,
|
| 112 |
-
"logits/chosen": -2.5740597248077393,
|
| 113 |
-
"logits/rejected": -2.553145408630371,
|
| 114 |
-
"logps/chosen": -278.0965270996094,
|
| 115 |
-
"logps/rejected": -267.19586181640625,
|
| 116 |
-
"loss": 1971.1375,
|
| 117 |
-
"rewards/accuracies": 0.6875,
|
| 118 |
-
"rewards/chosen": 0.015462947078049183,
|
| 119 |
-
"rewards/margins": 0.08871600031852722,
|
| 120 |
-
"rewards/rejected": -0.07325305044651031,
|
| 121 |
-
"step": 70
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.17,
|
| 125 |
-
"learning_rate": 4.93167072587771e-06,
|
| 126 |
-
"logits/chosen": -2.5298993587493896,
|
| 127 |
-
"logits/rejected": -2.5009925365448,
|
| 128 |
-
"logps/chosen": -258.5903015136719,
|
| 129 |
-
"logps/rejected": -263.52850341796875,
|
| 130 |
-
"loss": 1933.0076,
|
| 131 |
-
"rewards/accuracies": 0.731249988079071,
|
| 132 |
-
"rewards/chosen": 0.029317494481801987,
|
| 133 |
-
"rewards/margins": 0.11680855602025986,
|
| 134 |
-
"rewards/rejected": -0.08749105781316757,
|
| 135 |
-
"step": 80
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.19,
|
| 139 |
-
"learning_rate": 4.882681251368549e-06,
|
| 140 |
-
"logits/chosen": -2.5273003578186035,
|
| 141 |
-
"logits/rejected": -2.493241548538208,
|
| 142 |
-
"logps/chosen": -247.47506713867188,
|
| 143 |
-
"logps/rejected": -260.76678466796875,
|
| 144 |
-
"loss": 1845.5582,
|
| 145 |
-
"rewards/accuracies": 0.7281249761581421,
|
| 146 |
-
"rewards/chosen": -0.03806694597005844,
|
| 147 |
-
"rewards/margins": 0.11302463710308075,
|
| 148 |
-
"rewards/rejected": -0.15109160542488098,
|
| 149 |
-
"step": 90
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.21,
|
| 153 |
-
"learning_rate": 4.8209198325401815e-06,
|
| 154 |
-
"logits/chosen": -2.5287628173828125,
|
| 155 |
-
"logits/rejected": -2.5358829498291016,
|
| 156 |
-
"logps/chosen": -272.0884704589844,
|
| 157 |
-
"logps/rejected": -275.2580871582031,
|
| 158 |
-
"loss": 1797.9404,
|
| 159 |
-
"rewards/accuracies": 0.762499988079071,
|
| 160 |
-
"rewards/chosen": -0.0045492262579500675,
|
| 161 |
-
"rewards/margins": 0.13472509384155273,
|
| 162 |
-
"rewards/rejected": -0.13927432894706726,
|
| 163 |
-
"step": 100
|
| 164 |
-
},
|
| 165 |
-
{
|
| 166 |
-
"epoch": 0.21,
|
| 167 |
-
"eval_logits/chosen": -2.196876287460327,
|
| 168 |
-
"eval_logits/rejected": -2.1486356258392334,
|
| 169 |
-
"eval_logps/chosen": -263.71331787109375,
|
| 170 |
-
"eval_logps/rejected": -256.7424011230469,
|
| 171 |
-
"eval_loss": 1887.4102783203125,
|
| 172 |
-
"eval_rewards/accuracies": 0.7519841194152832,
|
| 173 |
-
"eval_rewards/chosen": 0.013112416490912437,
|
| 174 |
-
"eval_rewards/margins": 0.13282696902751923,
|
| 175 |
-
"eval_rewards/rejected": -0.11971456557512283,
|
| 176 |
-
"eval_runtime": 549.9966,
|
| 177 |
-
"eval_samples_per_second": 3.636,
|
| 178 |
-
"eval_steps_per_second": 0.115,
|
| 179 |
-
"step": 100
|
| 180 |
-
},
|
| 181 |
-
{
|
| 182 |
-
"epoch": 0.23,
|
| 183 |
-
"learning_rate": 4.746717530629565e-06,
|
| 184 |
-
"logits/chosen": -2.480510711669922,
|
| 185 |
-
"logits/rejected": -2.4668211936950684,
|
| 186 |
-
"logps/chosen": -267.04180908203125,
|
| 187 |
-
"logps/rejected": -262.5838317871094,
|
| 188 |
-
"loss": 1870.9051,
|
| 189 |
-
"rewards/accuracies": 0.734375,
|
| 190 |
-
"rewards/chosen": -0.020576762035489082,
|
| 191 |
-
"rewards/margins": 0.125274196267128,
|
| 192 |
-
"rewards/rejected": -0.14585095643997192,
|
| 193 |
-
"step": 110
|
| 194 |
-
},
|
| 195 |
-
{
|
| 196 |
-
"epoch": 0.25,
|
| 197 |
-
"learning_rate": 4.660472094042121e-06,
|
| 198 |
-
"logits/chosen": -2.44077205657959,
|
| 199 |
-
"logits/rejected": -2.4053845405578613,
|
| 200 |
-
"logps/chosen": -256.12939453125,
|
| 201 |
-
"logps/rejected": -248.28060913085938,
|
| 202 |
-
"loss": 1855.318,
|
| 203 |
-
"rewards/accuracies": 0.684374988079071,
|
| 204 |
-
"rewards/chosen": -0.049732744693756104,
|
| 205 |
-
"rewards/margins": 0.10872016102075577,
|
| 206 |
-
"rewards/rejected": -0.15845291316509247,
|
| 207 |
-
"step": 120
|
| 208 |
-
},
|
| 209 |
-
{
|
| 210 |
-
"epoch": 0.27,
|
| 211 |
-
"learning_rate": 4.5626458262912745e-06,
|
| 212 |
-
"logits/chosen": -2.395805597305298,
|
| 213 |
-
"logits/rejected": -2.383305311203003,
|
| 214 |
-
"logps/chosen": -280.74053955078125,
|
| 215 |
-
"logps/rejected": -270.37860107421875,
|
| 216 |
-
"loss": 1811.4148,
|
| 217 |
-
"rewards/accuracies": 0.7437499761581421,
|
| 218 |
-
"rewards/chosen": -0.027533594518899918,
|
| 219 |
-
"rewards/margins": 0.1342071145772934,
|
| 220 |
-
"rewards/rejected": -0.1617407202720642,
|
| 221 |
-
"step": 130
|
| 222 |
-
},
|
| 223 |
-
{
|
| 224 |
-
"epoch": 0.29,
|
| 225 |
-
"learning_rate": 4.453763107901676e-06,
|
| 226 |
-
"logits/chosen": -2.4485344886779785,
|
| 227 |
-
"logits/rejected": -2.43884015083313,
|
| 228 |
-
"logps/chosen": -243.1454315185547,
|
| 229 |
-
"logps/rejected": -255.15432739257812,
|
| 230 |
-
"loss": 1803.225,
|
| 231 |
-
"rewards/accuracies": 0.734375,
|
| 232 |
-
"rewards/chosen": -0.028790492564439774,
|
| 233 |
-
"rewards/margins": 0.1489991694688797,
|
| 234 |
-
"rewards/rejected": -0.17778967320919037,
|
| 235 |
-
"step": 140
|
| 236 |
-
},
|
| 237 |
-
{
|
| 238 |
-
"epoch": 0.31,
|
| 239 |
-
"learning_rate": 4.33440758555951e-06,
|
| 240 |
-
"logits/chosen": -2.459658622741699,
|
| 241 |
-
"logits/rejected": -2.483065605163574,
|
| 242 |
-
"logps/chosen": -267.7740478515625,
|
| 243 |
-
"logps/rejected": -243.34609985351562,
|
| 244 |
-
"loss": 1781.1752,
|
| 245 |
-
"rewards/accuracies": 0.699999988079071,
|
| 246 |
-
"rewards/chosen": -0.02288922667503357,
|
| 247 |
-
"rewards/margins": 0.12706486880779266,
|
| 248 |
-
"rewards/rejected": -0.14995409548282623,
|
| 249 |
-
"step": 150
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.33,
|
| 253 |
-
"learning_rate": 4.205219043576955e-06,
|
| 254 |
-
"logits/chosen": -2.483583688735962,
|
| 255 |
-
"logits/rejected": -2.4244942665100098,
|
| 256 |
-
"logps/chosen": -260.3743896484375,
|
| 257 |
-
"logps/rejected": -258.7478332519531,
|
| 258 |
-
"loss": 1754.5766,
|
| 259 |
-
"rewards/accuracies": 0.784375011920929,
|
| 260 |
-
"rewards/chosen": -0.0020265295170247555,
|
| 261 |
-
"rewards/margins": 0.16760031878948212,
|
| 262 |
-
"rewards/rejected": -0.169626846909523,
|
| 263 |
-
"step": 160
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 0.36,
|
| 267 |
-
"learning_rate": 4.066889974440757e-06,
|
| 268 |
-
"logits/chosen": -2.4374189376831055,
|
| 269 |
-
"logits/rejected": -2.428433656692505,
|
| 270 |
-
"logps/chosen": -264.5699768066406,
|
| 271 |
-
"logps/rejected": -252.79421997070312,
|
| 272 |
-
"loss": 1953.8818,
|
| 273 |
-
"rewards/accuracies": 0.668749988079071,
|
| 274 |
-
"rewards/chosen": -0.06210694834589958,
|
| 275 |
-
"rewards/margins": 0.13285748660564423,
|
| 276 |
-
"rewards/rejected": -0.1949644386768341,
|
| 277 |
-
"step": 170
|
| 278 |
-
},
|
| 279 |
-
{
|
| 280 |
-
"epoch": 0.38,
|
| 281 |
-
"learning_rate": 3.92016186682789e-06,
|
| 282 |
-
"logits/chosen": -2.467085361480713,
|
| 283 |
-
"logits/rejected": -2.487204074859619,
|
| 284 |
-
"logps/chosen": -262.995361328125,
|
| 285 |
-
"logps/rejected": -271.94183349609375,
|
| 286 |
-
"loss": 1848.9945,
|
| 287 |
-
"rewards/accuracies": 0.721875011920929,
|
| 288 |
-
"rewards/chosen": -0.07509048283100128,
|
| 289 |
-
"rewards/margins": 0.12671387195587158,
|
| 290 |
-
"rewards/rejected": -0.20180435478687286,
|
| 291 |
-
"step": 180
|
| 292 |
-
},
|
| 293 |
-
{
|
| 294 |
-
"epoch": 0.4,
|
| 295 |
-
"learning_rate": 3.7658212309857576e-06,
|
| 296 |
-
"logits/chosen": -2.450601816177368,
|
| 297 |
-
"logits/rejected": -2.4304168224334717,
|
| 298 |
-
"logps/chosen": -269.1886901855469,
|
| 299 |
-
"logps/rejected": -265.7490539550781,
|
| 300 |
-
"loss": 1698.6666,
|
| 301 |
-
"rewards/accuracies": 0.746874988079071,
|
| 302 |
-
"rewards/chosen": -0.09053254127502441,
|
| 303 |
-
"rewards/margins": 0.14999321103096008,
|
| 304 |
-
"rewards/rejected": -0.2405257225036621,
|
| 305 |
-
"step": 190
|
| 306 |
-
},
|
| 307 |
-
{
|
| 308 |
-
"epoch": 0.42,
|
| 309 |
-
"learning_rate": 3.604695382782159e-06,
|
| 310 |
-
"logits/chosen": -2.447007179260254,
|
| 311 |
-
"logits/rejected": -2.419039726257324,
|
| 312 |
-
"logps/chosen": -282.8253479003906,
|
| 313 |
-
"logps/rejected": -278.14508056640625,
|
| 314 |
-
"loss": 1700.9055,
|
| 315 |
-
"rewards/accuracies": 0.784375011920929,
|
| 316 |
-
"rewards/chosen": -0.07701022177934647,
|
| 317 |
-
"rewards/margins": 0.16163742542266846,
|
| 318 |
-
"rewards/rejected": -0.23864765465259552,
|
| 319 |
-
"step": 200
|
| 320 |
-
},
|
| 321 |
-
{
|
| 322 |
-
"epoch": 0.42,
|
| 323 |
-
"eval_logits/chosen": -2.161839485168457,
|
| 324 |
-
"eval_logits/rejected": -2.1081268787384033,
|
| 325 |
-
"eval_logps/chosen": -269.66546630859375,
|
| 326 |
-
"eval_logps/rejected": -265.3905029296875,
|
| 327 |
-
"eval_loss": 1784.6597900390625,
|
| 328 |
-
"eval_rewards/accuracies": 0.761904776096344,
|
| 329 |
-
"eval_rewards/chosen": -0.0464087538421154,
|
| 330 |
-
"eval_rewards/margins": 0.15978708863258362,
|
| 331 |
-
"eval_rewards/rejected": -0.2061958611011505,
|
| 332 |
-
"eval_runtime": 549.0189,
|
| 333 |
-
"eval_samples_per_second": 3.643,
|
| 334 |
-
"eval_steps_per_second": 0.115,
|
| 335 |
-
"step": 200
|
| 336 |
-
},
|
| 337 |
-
{
|
| 338 |
-
"epoch": 0.44,
|
| 339 |
-
"learning_rate": 3.437648009023905e-06,
|
| 340 |
-
"logits/chosen": -2.458688259124756,
|
| 341 |
-
"logits/rejected": -2.4217796325683594,
|
| 342 |
-
"logps/chosen": -252.5647430419922,
|
| 343 |
-
"logps/rejected": -248.326416015625,
|
| 344 |
-
"loss": 1806.0594,
|
| 345 |
-
"rewards/accuracies": 0.7562500238418579,
|
| 346 |
-
"rewards/chosen": -0.02530970238149166,
|
| 347 |
-
"rewards/margins": 0.14908090233802795,
|
| 348 |
-
"rewards/rejected": -0.17439061403274536,
|
| 349 |
-
"step": 210
|
| 350 |
-
},
|
| 351 |
-
{
|
| 352 |
-
"epoch": 0.46,
|
| 353 |
-
"learning_rate": 3.265574537815398e-06,
|
| 354 |
-
"logits/chosen": -2.4742610454559326,
|
| 355 |
-
"logits/rejected": -2.4789376258850098,
|
| 356 |
-
"logps/chosen": -286.0444030761719,
|
| 357 |
-
"logps/rejected": -261.9767150878906,
|
| 358 |
-
"loss": 1855.6273,
|
| 359 |
-
"rewards/accuracies": 0.706250011920929,
|
| 360 |
-
"rewards/chosen": -0.03405206650495529,
|
| 361 |
-
"rewards/margins": 0.11277566105127335,
|
| 362 |
-
"rewards/rejected": -0.14682772755622864,
|
| 363 |
-
"step": 220
|
| 364 |
-
},
|
| 365 |
-
{
|
| 366 |
-
"epoch": 0.48,
|
| 367 |
-
"learning_rate": 3.089397338773569e-06,
|
| 368 |
-
"logits/chosen": -2.38773775100708,
|
| 369 |
-
"logits/rejected": -2.3718185424804688,
|
| 370 |
-
"logps/chosen": -257.7181701660156,
|
| 371 |
-
"logps/rejected": -253.3428955078125,
|
| 372 |
-
"loss": 1797.9486,
|
| 373 |
-
"rewards/accuracies": 0.721875011920929,
|
| 374 |
-
"rewards/chosen": -0.06530335545539856,
|
| 375 |
-
"rewards/margins": 0.1308148354291916,
|
| 376 |
-
"rewards/rejected": -0.19611820578575134,
|
| 377 |
-
"step": 230
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 0.5,
|
| 381 |
-
"learning_rate": 2.9100607788275547e-06,
|
| 382 |
-
"logits/chosen": -2.4125852584838867,
|
| 383 |
-
"logits/rejected": -2.414628267288208,
|
| 384 |
-
"logps/chosen": -265.2156066894531,
|
| 385 |
-
"logps/rejected": -257.0289001464844,
|
| 386 |
-
"loss": 1850.0729,
|
| 387 |
-
"rewards/accuracies": 0.690625011920929,
|
| 388 |
-
"rewards/chosen": -0.04248107224702835,
|
| 389 |
-
"rewards/margins": 0.12761279940605164,
|
| 390 |
-
"rewards/rejected": -0.17009387910366058,
|
| 391 |
-
"step": 240
|
| 392 |
-
},
|
| 393 |
-
{
|
| 394 |
-
"epoch": 0.52,
|
| 395 |
-
"learning_rate": 2.72852616010567e-06,
|
| 396 |
-
"logits/chosen": -2.4339253902435303,
|
| 397 |
-
"logits/rejected": -2.4054951667785645,
|
| 398 |
-
"logps/chosen": -271.8371276855469,
|
| 399 |
-
"logps/rejected": -255.33438110351562,
|
| 400 |
-
"loss": 1766.1885,
|
| 401 |
-
"rewards/accuracies": 0.753125011920929,
|
| 402 |
-
"rewards/chosen": -0.0324532687664032,
|
| 403 |
-
"rewards/margins": 0.1579422652721405,
|
| 404 |
-
"rewards/rejected": -0.1903955340385437,
|
| 405 |
-
"step": 250
|
| 406 |
-
},
|
| 407 |
-
{
|
| 408 |
-
"epoch": 0.54,
|
| 409 |
-
"learning_rate": 2.5457665670441937e-06,
|
| 410 |
-
"logits/chosen": -2.4216437339782715,
|
| 411 |
-
"logits/rejected": -2.4156367778778076,
|
| 412 |
-
"logps/chosen": -266.7996520996094,
|
| 413 |
-
"logps/rejected": -243.180419921875,
|
| 414 |
-
"loss": 1710.8809,
|
| 415 |
-
"rewards/accuracies": 0.753125011920929,
|
| 416 |
-
"rewards/chosen": -0.035904210060834885,
|
| 417 |
-
"rewards/margins": 0.16971439123153687,
|
| 418 |
-
"rewards/rejected": -0.20561861991882324,
|
| 419 |
-
"step": 260
|
| 420 |
-
},
|
| 421 |
-
{
|
| 422 |
-
"epoch": 0.57,
|
| 423 |
-
"learning_rate": 2.3627616503391813e-06,
|
| 424 |
-
"logits/chosen": -2.4438915252685547,
|
| 425 |
-
"logits/rejected": -2.416748285293579,
|
| 426 |
-
"logps/chosen": -290.58453369140625,
|
| 427 |
-
"logps/rejected": -277.0739440917969,
|
| 428 |
-
"loss": 1714.5062,
|
| 429 |
-
"rewards/accuracies": 0.765625,
|
| 430 |
-
"rewards/chosen": -0.042676471173763275,
|
| 431 |
-
"rewards/margins": 0.17854078114032745,
|
| 432 |
-
"rewards/rejected": -0.22121724486351013,
|
| 433 |
-
"step": 270
|
| 434 |
-
},
|
| 435 |
-
{
|
| 436 |
-
"epoch": 0.59,
|
| 437 |
-
"learning_rate": 2.1804923757009885e-06,
|
| 438 |
-
"logits/chosen": -2.414602756500244,
|
| 439 |
-
"logits/rejected": -2.4200820922851562,
|
| 440 |
-
"logps/chosen": -282.95147705078125,
|
| 441 |
-
"logps/rejected": -261.1886291503906,
|
| 442 |
-
"loss": 1764.4607,
|
| 443 |
-
"rewards/accuracies": 0.7437499761581421,
|
| 444 |
-
"rewards/chosen": -0.07589195668697357,
|
| 445 |
-
"rewards/margins": 0.13762618601322174,
|
| 446 |
-
"rewards/rejected": -0.2135181427001953,
|
| 447 |
-
"step": 280
|
| 448 |
-
},
|
| 449 |
-
{
|
| 450 |
-
"epoch": 0.61,
|
| 451 |
-
"learning_rate": 1.9999357655598894e-06,
|
| 452 |
-
"logits/chosen": -2.430169105529785,
|
| 453 |
-
"logits/rejected": -2.4057881832122803,
|
| 454 |
-
"logps/chosen": -265.06805419921875,
|
| 455 |
-
"logps/rejected": -263.2739562988281,
|
| 456 |
-
"loss": 1786.2846,
|
| 457 |
-
"rewards/accuracies": 0.746874988079071,
|
| 458 |
-
"rewards/chosen": -0.010251840576529503,
|
| 459 |
-
"rewards/margins": 0.15472975373268127,
|
| 460 |
-
"rewards/rejected": -0.16498157382011414,
|
| 461 |
-
"step": 290
|
| 462 |
-
},
|
| 463 |
-
{
|
| 464 |
-
"epoch": 0.63,
|
| 465 |
-
"learning_rate": 1.8220596619089576e-06,
|
| 466 |
-
"logits/chosen": -2.392138957977295,
|
| 467 |
-
"logits/rejected": -2.3823294639587402,
|
| 468 |
-
"logps/chosen": -255.75393676757812,
|
| 469 |
-
"logps/rejected": -261.84271240234375,
|
| 470 |
-
"loss": 1767.2219,
|
| 471 |
-
"rewards/accuracies": 0.737500011920929,
|
| 472 |
-
"rewards/chosen": -0.04702477902173996,
|
| 473 |
-
"rewards/margins": 0.13455010950565338,
|
| 474 |
-
"rewards/rejected": -0.18157489597797394,
|
| 475 |
-
"step": 300
|
| 476 |
-
},
|
| 477 |
-
{
|
| 478 |
-
"epoch": 0.63,
|
| 479 |
-
"eval_logits/chosen": -2.158698797225952,
|
| 480 |
-
"eval_logits/rejected": -2.1057095527648926,
|
| 481 |
-
"eval_logps/chosen": -269.6955871582031,
|
| 482 |
-
"eval_logps/rejected": -264.77947998046875,
|
| 483 |
-
"eval_loss": 1735.518310546875,
|
| 484 |
-
"eval_rewards/accuracies": 0.7698412537574768,
|
| 485 |
-
"eval_rewards/chosen": -0.04671022295951843,
|
| 486 |
-
"eval_rewards/margins": 0.15337513387203217,
|
| 487 |
-
"eval_rewards/rejected": -0.2000853717327118,
|
| 488 |
-
"eval_runtime": 548.7136,
|
| 489 |
-
"eval_samples_per_second": 3.645,
|
| 490 |
-
"eval_steps_per_second": 0.115,
|
| 491 |
-
"step": 300
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 0.65,
|
| 495 |
-
"learning_rate": 1.647817538357072e-06,
|
| 496 |
-
"logits/chosen": -2.4140188694000244,
|
| 497 |
-
"logits/rejected": -2.4031002521514893,
|
| 498 |
-
"logps/chosen": -274.3767395019531,
|
| 499 |
-
"logps/rejected": -259.40155029296875,
|
| 500 |
-
"loss": 1673.9693,
|
| 501 |
-
"rewards/accuracies": 0.746874988079071,
|
| 502 |
-
"rewards/chosen": -0.04385297745466232,
|
| 503 |
-
"rewards/margins": 0.1517268717288971,
|
| 504 |
-
"rewards/rejected": -0.19557985663414001,
|
| 505 |
-
"step": 310
|
| 506 |
-
},
|
| 507 |
-
{
|
| 508 |
-
"epoch": 0.67,
|
| 509 |
-
"learning_rate": 1.4781433892011132e-06,
|
| 510 |
-
"logits/chosen": -2.416640520095825,
|
| 511 |
-
"logits/rejected": -2.370535135269165,
|
| 512 |
-
"logps/chosen": -252.216064453125,
|
| 513 |
-
"logps/rejected": -255.1393280029297,
|
| 514 |
-
"loss": 1673.8594,
|
| 515 |
-
"rewards/accuracies": 0.7593749761581421,
|
| 516 |
-
"rewards/chosen": -0.043931327760219574,
|
| 517 |
-
"rewards/margins": 0.16487570106983185,
|
| 518 |
-
"rewards/rejected": -0.20880703628063202,
|
| 519 |
-
"step": 320
|
| 520 |
-
},
|
| 521 |
-
{
|
| 522 |
-
"epoch": 0.69,
|
| 523 |
-
"learning_rate": 1.3139467229135999e-06,
|
| 524 |
-
"logits/chosen": -2.362358570098877,
|
| 525 |
-
"logits/rejected": -2.3449196815490723,
|
| 526 |
-
"logps/chosen": -270.876220703125,
|
| 527 |
-
"logps/rejected": -259.251953125,
|
| 528 |
-
"loss": 1731.3877,
|
| 529 |
-
"rewards/accuracies": 0.753125011920929,
|
| 530 |
-
"rewards/chosen": -0.03340950980782509,
|
| 531 |
-
"rewards/margins": 0.13931182026863098,
|
| 532 |
-
"rewards/rejected": -0.17272132635116577,
|
| 533 |
-
"step": 330
|
| 534 |
-
},
|
| 535 |
-
{
|
| 536 |
-
"epoch": 0.71,
|
| 537 |
-
"learning_rate": 1.1561076868822756e-06,
|
| 538 |
-
"logits/chosen": -2.3923397064208984,
|
| 539 |
-
"logits/rejected": -2.384582281112671,
|
| 540 |
-
"logps/chosen": -284.8360290527344,
|
| 541 |
-
"logps/rejected": -257.0713806152344,
|
| 542 |
-
"loss": 1778.2957,
|
| 543 |
-
"rewards/accuracies": 0.7718750238418579,
|
| 544 |
-
"rewards/chosen": -0.039347052574157715,
|
| 545 |
-
"rewards/margins": 0.1650826632976532,
|
| 546 |
-
"rewards/rejected": -0.20442970097064972,
|
| 547 |
-
"step": 340
|
| 548 |
-
},
|
| 549 |
-
{
|
| 550 |
-
"epoch": 0.73,
|
| 551 |
-
"learning_rate": 1.0054723495346484e-06,
|
| 552 |
-
"logits/chosen": -2.3869528770446777,
|
| 553 |
-
"logits/rejected": -2.3370375633239746,
|
| 554 |
-
"logps/chosen": -259.75189208984375,
|
| 555 |
-
"logps/rejected": -231.13577270507812,
|
| 556 |
-
"loss": 1665.3461,
|
| 557 |
-
"rewards/accuracies": 0.7437499761581421,
|
| 558 |
-
"rewards/chosen": -0.03856384754180908,
|
| 559 |
-
"rewards/margins": 0.1649591028690338,
|
| 560 |
-
"rewards/rejected": -0.2035229504108429,
|
| 561 |
-
"step": 350
|
| 562 |
-
},
|
| 563 |
-
{
|
| 564 |
-
"epoch": 0.75,
|
| 565 |
-
"learning_rate": 8.628481651367876e-07,
|
| 566 |
-
"logits/chosen": -2.4105262756347656,
|
| 567 |
-
"logits/rejected": -2.352128744125366,
|
| 568 |
-
"logps/chosen": -269.03790283203125,
|
| 569 |
-
"logps/rejected": -247.90872192382812,
|
| 570 |
-
"loss": 1665.3982,
|
| 571 |
-
"rewards/accuracies": 0.7875000238418579,
|
| 572 |
-
"rewards/chosen": -0.03317265957593918,
|
| 573 |
-
"rewards/margins": 0.1719072014093399,
|
| 574 |
-
"rewards/rejected": -0.20507986843585968,
|
| 575 |
-
"step": 360
|
| 576 |
-
},
|
| 577 |
-
{
|
| 578 |
-
"epoch": 0.77,
|
| 579 |
-
"learning_rate": 7.289996455765749e-07,
|
| 580 |
-
"logits/chosen": -2.4054064750671387,
|
| 581 |
-
"logits/rejected": -2.38871431350708,
|
| 582 |
-
"logps/chosen": -279.2740173339844,
|
| 583 |
-
"logps/rejected": -259.63690185546875,
|
| 584 |
-
"loss": 1704.7645,
|
| 585 |
-
"rewards/accuracies": 0.7124999761581421,
|
| 586 |
-
"rewards/chosen": -0.07123039662837982,
|
| 587 |
-
"rewards/margins": 0.15871620178222656,
|
| 588 |
-
"rewards/rejected": -0.22994661331176758,
|
| 589 |
-
"step": 370
|
| 590 |
-
},
|
| 591 |
-
{
|
| 592 |
-
"epoch": 0.8,
|
| 593 |
-
"learning_rate": 6.046442623320145e-07,
|
| 594 |
-
"logits/chosen": -2.3605639934539795,
|
| 595 |
-
"logits/rejected": -2.368460178375244,
|
| 596 |
-
"logps/chosen": -267.2261657714844,
|
| 597 |
-
"logps/rejected": -260.45550537109375,
|
| 598 |
-
"loss": 1647.7326,
|
| 599 |
-
"rewards/accuracies": 0.734375,
|
| 600 |
-
"rewards/chosen": -0.0854114517569542,
|
| 601 |
-
"rewards/margins": 0.15202030539512634,
|
| 602 |
-
"rewards/rejected": -0.23743176460266113,
|
| 603 |
-
"step": 380
|
| 604 |
-
},
|
| 605 |
{
|
| 606 |
"epoch": 0.82,
|
| 607 |
-
"
|
| 608 |
-
"logits/chosen": -2.4076011180877686,
|
| 609 |
-
"logits/rejected": -2.3770554065704346,
|
| 610 |
-
"logps/chosen": -292.8500061035156,
|
| 611 |
-
"logps/rejected": -292.0636291503906,
|
| 612 |
-
"loss": 1739.5414,
|
| 613 |
-
"rewards/accuracies": 0.746874988079071,
|
| 614 |
-
"rewards/chosen": -0.06398223340511322,
|
| 615 |
-
"rewards/margins": 0.15206970274448395,
|
| 616 |
-
"rewards/rejected": -0.21605193614959717,
|
| 617 |
-
"step": 390
|
| 618 |
-
},
|
| 619 |
-
{
|
| 620 |
-
"epoch": 0.84,
|
| 621 |
-
"learning_rate": 3.8702478614051353e-07,
|
| 622 |
-
"logits/chosen": -2.3384757041931152,
|
| 623 |
-
"logits/rejected": -2.3366100788116455,
|
| 624 |
-
"logps/chosen": -259.2252502441406,
|
| 625 |
-
"logps/rejected": -265.5692138671875,
|
| 626 |
-
"loss": 1717.4336,
|
| 627 |
-
"rewards/accuracies": 0.7124999761581421,
|
| 628 |
-
"rewards/chosen": -0.09164019674062729,
|
| 629 |
-
"rewards/margins": 0.14173154532909393,
|
| 630 |
-
"rewards/rejected": -0.23337173461914062,
|
| 631 |
-
"step": 400
|
| 632 |
-
},
|
| 633 |
-
{
|
| 634 |
-
"epoch": 0.84,
|
| 635 |
-
"eval_logits/chosen": -2.088451385498047,
|
| 636 |
-
"eval_logits/rejected": -2.032222032546997,
|
| 637 |
-
"eval_logps/chosen": -271.9333190917969,
|
| 638 |
-
"eval_logps/rejected": -267.85687255859375,
|
| 639 |
-
"eval_loss": 1721.676513671875,
|
| 640 |
-
"eval_rewards/accuracies": 0.77182537317276,
|
| 641 |
-
"eval_rewards/chosen": -0.06908722221851349,
|
| 642 |
-
"eval_rewards/margins": 0.1617719829082489,
|
| 643 |
-
"eval_rewards/rejected": -0.23085922002792358,
|
| 644 |
-
"eval_runtime": 548.423,
|
| 645 |
-
"eval_samples_per_second": 3.647,
|
| 646 |
-
"eval_steps_per_second": 0.115,
|
| 647 |
-
"step": 400
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.86,
|
| 651 |
-
"learning_rate": 2.9492720416985004e-07,
|
| 652 |
-
"logits/chosen": -2.3725836277008057,
|
| 653 |
-
"logits/rejected": -2.3304688930511475,
|
| 654 |
-
"logps/chosen": -296.72991943359375,
|
| 655 |
-
"logps/rejected": -266.0842590332031,
|
| 656 |
-
"loss": 1755.1898,
|
| 657 |
-
"rewards/accuracies": 0.7406250238418579,
|
| 658 |
-
"rewards/chosen": -0.0669253021478653,
|
| 659 |
-
"rewards/margins": 0.16722533106803894,
|
| 660 |
-
"rewards/rejected": -0.23415064811706543,
|
| 661 |
-
"step": 410
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.88,
|
| 665 |
-
"learning_rate": 2.1464952759020857e-07,
|
| 666 |
-
"logits/chosen": -2.3592472076416016,
|
| 667 |
-
"logits/rejected": -2.331540107727051,
|
| 668 |
-
"logps/chosen": -266.99005126953125,
|
| 669 |
-
"logps/rejected": -292.03680419921875,
|
| 670 |
-
"loss": 1730.9672,
|
| 671 |
-
"rewards/accuracies": 0.706250011920929,
|
| 672 |
-
"rewards/chosen": -0.08814045041799545,
|
| 673 |
-
"rewards/margins": 0.12336041778326035,
|
| 674 |
-
"rewards/rejected": -0.2115008533000946,
|
| 675 |
-
"step": 420
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.9,
|
| 679 |
-
"learning_rate": 1.4662207078575685e-07,
|
| 680 |
-
"logits/chosen": -2.366381883621216,
|
| 681 |
-
"logits/rejected": -2.35951566696167,
|
| 682 |
-
"logps/chosen": -280.3116149902344,
|
| 683 |
-
"logps/rejected": -281.93939208984375,
|
| 684 |
-
"loss": 1760.3617,
|
| 685 |
-
"rewards/accuracies": 0.7281249761581421,
|
| 686 |
-
"rewards/chosen": -0.08124328404664993,
|
| 687 |
-
"rewards/margins": 0.14409320056438446,
|
| 688 |
-
"rewards/rejected": -0.2253364771604538,
|
| 689 |
-
"step": 430
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.92,
|
| 693 |
-
"learning_rate": 9.120948298936422e-08,
|
| 694 |
-
"logits/chosen": -2.3266444206237793,
|
| 695 |
-
"logits/rejected": -2.2898497581481934,
|
| 696 |
-
"logps/chosen": -243.9964141845703,
|
| 697 |
-
"logps/rejected": -248.1795196533203,
|
| 698 |
-
"loss": 1711.7143,
|
| 699 |
-
"rewards/accuracies": 0.75,
|
| 700 |
-
"rewards/chosen": -0.08177933841943741,
|
| 701 |
-
"rewards/margins": 0.15183614194393158,
|
| 702 |
-
"rewards/rejected": -0.23361548781394958,
|
| 703 |
-
"step": 440
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.94,
|
| 707 |
-
"learning_rate": 4.870879364444109e-08,
|
| 708 |
-
"logits/chosen": -2.3814821243286133,
|
| 709 |
-
"logits/rejected": -2.4406635761260986,
|
| 710 |
-
"logps/chosen": -275.4070129394531,
|
| 711 |
-
"logps/rejected": -278.91082763671875,
|
| 712 |
-
"loss": 1743.9877,
|
| 713 |
-
"rewards/accuracies": 0.7406250238418579,
|
| 714 |
-
"rewards/chosen": -0.06510698050260544,
|
| 715 |
-
"rewards/margins": 0.14476314187049866,
|
| 716 |
-
"rewards/rejected": -0.2098701000213623,
|
| 717 |
-
"step": 450
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.96,
|
| 721 |
-
"learning_rate": 1.93478202307823e-08,
|
| 722 |
-
"logits/chosen": -2.34289288520813,
|
| 723 |
-
"logits/rejected": -2.346625804901123,
|
| 724 |
-
"logps/chosen": -270.3787841796875,
|
| 725 |
-
"logps/rejected": -275.61651611328125,
|
| 726 |
-
"loss": 1676.5176,
|
| 727 |
-
"rewards/accuracies": 0.737500011920929,
|
| 728 |
-
"rewards/chosen": -0.07770398259162903,
|
| 729 |
-
"rewards/margins": 0.16223737597465515,
|
| 730 |
-
"rewards/rejected": -0.23994135856628418,
|
| 731 |
-
"step": 460
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.98,
|
| 735 |
-
"learning_rate": 3.283947088983663e-09,
|
| 736 |
-
"logits/chosen": -2.3888649940490723,
|
| 737 |
-
"logits/rejected": -2.4120144844055176,
|
| 738 |
-
"logps/chosen": -261.28997802734375,
|
| 739 |
-
"logps/rejected": -261.67755126953125,
|
| 740 |
-
"loss": 1663.4154,
|
| 741 |
-
"rewards/accuracies": 0.768750011920929,
|
| 742 |
-
"rewards/chosen": -0.0687546655535698,
|
| 743 |
-
"rewards/margins": 0.16177912056446075,
|
| 744 |
-
"rewards/rejected": -0.23053380846977234,
|
| 745 |
-
"step": 470
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 1.0,
|
| 749 |
-
"step": 477,
|
| 750 |
"total_flos": 0.0,
|
| 751 |
-
"train_loss":
|
| 752 |
-
"train_runtime":
|
| 753 |
-
"train_samples_per_second":
|
| 754 |
-
"train_steps_per_second": 0.
|
| 755 |
}
|
| 756 |
],
|
| 757 |
"logging_steps": 10,
|
| 758 |
-
"max_steps":
|
| 759 |
"num_input_tokens_seen": 0,
|
| 760 |
"num_train_epochs": 1,
|
| 761 |
"save_steps": 100,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8205128205128205,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 2,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.41,
|
| 13 |
+
"learning_rate": 5e-06,
|
| 14 |
+
"logits/chosen": -2.6451878547668457,
|
| 15 |
+
"logits/rejected": -2.6389708518981934,
|
| 16 |
+
"logps/chosen": -317.35162353515625,
|
| 17 |
+
"logps/rejected": -283.9225769042969,
|
| 18 |
+
"loss": 0.6931,
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
| 22 |
"rewards/rejected": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
{
|
| 26 |
"epoch": 0.82,
|
| 27 |
+
"step": 2,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
"total_flos": 0.0,
|
| 29 |
+
"train_loss": 0.6931473016738892,
|
| 30 |
+
"train_runtime": 124.4347,
|
| 31 |
+
"train_samples_per_second": 2.451,
|
| 32 |
+
"train_steps_per_second": 0.016
|
| 33 |
}
|
| 34 |
],
|
| 35 |
"logging_steps": 10,
|
| 36 |
+
"max_steps": 2,
|
| 37 |
"num_input_tokens_seen": 0,
|
| 38 |
"num_train_epochs": 1,
|
| 39 |
"save_steps": 100,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4920
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b771063cb1469c5cb63a9d414b651fffe150d786ee263a7c6072b0561a3775f4
|
| 3 |
size 4920
|