Commit
•
e51898b
1
Parent(s):
550781f
Model save
Browse files- README.md +60 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +402 -0
README.md
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
base_model: allenai/tulu-2-7b
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: uf-tulu-2-7b-dpo
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# uf-tulu-2-7b-dpo
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [allenai/tulu-2-7b](https://huggingface.co/allenai/tulu-2-7b) on an unknown dataset.
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 5e-07
|
| 38 |
+
- train_batch_size: 8
|
| 39 |
+
- eval_batch_size: 8
|
| 40 |
+
- seed: 42
|
| 41 |
+
- distributed_type: multi-GPU
|
| 42 |
+
- num_devices: 8
|
| 43 |
+
- gradient_accumulation_steps: 4
|
| 44 |
+
- total_train_batch_size: 256
|
| 45 |
+
- total_eval_batch_size: 64
|
| 46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
+
- lr_scheduler_type: cosine
|
| 48 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 49 |
+
- num_epochs: 1
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
### Framework versions
|
| 56 |
+
|
| 57 |
+
- Transformers 4.44.1
|
| 58 |
+
- Pytorch 2.1.2+cu121
|
| 59 |
+
- Datasets 2.21.0
|
| 60 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6731121569996599,
|
| 5 |
+
"train_runtime": 6739.8654,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 9.071,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.1"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6731121569996599,
|
| 5 |
+
"train_runtime": 6739.8654,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 9.071,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,402 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 239,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0041841004184100415,
|
| 13 |
+
"grad_norm": 2.3349975667313307,
|
| 14 |
+
"learning_rate": 2.083333333333333e-08,
|
| 15 |
+
"logits/chosen": -1.168128490447998,
|
| 16 |
+
"logits/rejected": -1.1001684665679932,
|
| 17 |
+
"logps/chosen": -345.7468566894531,
|
| 18 |
+
"logps/rejected": -388.03436279296875,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04184100418410042,
|
| 28 |
+
"grad_norm": 2.2444991754758727,
|
| 29 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 30 |
+
"logits/chosen": -1.1528964042663574,
|
| 31 |
+
"logits/rejected": -1.2175092697143555,
|
| 32 |
+
"logps/chosen": -327.3994445800781,
|
| 33 |
+
"logps/rejected": -330.5547790527344,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 36 |
+
"rewards/chosen": 0.0001323310425505042,
|
| 37 |
+
"rewards/margins": 0.0002724311489146203,
|
| 38 |
+
"rewards/rejected": -0.00014010007726028562,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08368200836820083,
|
| 43 |
+
"grad_norm": 3.1970801165125993,
|
| 44 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 45 |
+
"logits/chosen": -1.1212791204452515,
|
| 46 |
+
"logits/rejected": -1.1984214782714844,
|
| 47 |
+
"logps/chosen": -337.3634033203125,
|
| 48 |
+
"logps/rejected": -310.3741455078125,
|
| 49 |
+
"loss": 0.6931,
|
| 50 |
+
"rewards/accuracies": 0.503125011920929,
|
| 51 |
+
"rewards/chosen": 0.0006343486020341516,
|
| 52 |
+
"rewards/margins": 0.00016590832092333585,
|
| 53 |
+
"rewards/rejected": 0.00046844035387039185,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.12552301255230125,
|
| 58 |
+
"grad_norm": 2.033994682703014,
|
| 59 |
+
"learning_rate": 4.990398100856366e-07,
|
| 60 |
+
"logits/chosen": -1.1630992889404297,
|
| 61 |
+
"logits/rejected": -1.1824390888214111,
|
| 62 |
+
"logps/chosen": -327.43975830078125,
|
| 63 |
+
"logps/rejected": -347.76409912109375,
|
| 64 |
+
"loss": 0.6922,
|
| 65 |
+
"rewards/accuracies": 0.546875,
|
| 66 |
+
"rewards/chosen": 0.004818023648113012,
|
| 67 |
+
"rewards/margins": 0.0015862248837947845,
|
| 68 |
+
"rewards/rejected": 0.003231799229979515,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.16736401673640167,
|
| 73 |
+
"grad_norm": 2.045259811639636,
|
| 74 |
+
"learning_rate": 4.931986719649298e-07,
|
| 75 |
+
"logits/chosen": -1.2456501722335815,
|
| 76 |
+
"logits/rejected": -1.2868419885635376,
|
| 77 |
+
"logps/chosen": -352.53753662109375,
|
| 78 |
+
"logps/rejected": -314.9628601074219,
|
| 79 |
+
"loss": 0.6901,
|
| 80 |
+
"rewards/accuracies": 0.640625,
|
| 81 |
+
"rewards/chosen": 0.013885138556361198,
|
| 82 |
+
"rewards/margins": 0.00753264594823122,
|
| 83 |
+
"rewards/rejected": 0.006352494470775127,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.20920502092050208,
|
| 88 |
+
"grad_norm": 1.9514164939530987,
|
| 89 |
+
"learning_rate": 4.821741763807186e-07,
|
| 90 |
+
"logits/chosen": -1.2360882759094238,
|
| 91 |
+
"logits/rejected": -1.3141974210739136,
|
| 92 |
+
"logps/chosen": -316.40191650390625,
|
| 93 |
+
"logps/rejected": -312.9485168457031,
|
| 94 |
+
"loss": 0.6873,
|
| 95 |
+
"rewards/accuracies": 0.671875,
|
| 96 |
+
"rewards/chosen": 0.020323345437645912,
|
| 97 |
+
"rewards/margins": 0.012846221216022968,
|
| 98 |
+
"rewards/rejected": 0.007477124221622944,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.2510460251046025,
|
| 103 |
+
"grad_norm": 2.209686138841628,
|
| 104 |
+
"learning_rate": 4.662012913161997e-07,
|
| 105 |
+
"logits/chosen": -1.2470484972000122,
|
| 106 |
+
"logits/rejected": -1.358208417892456,
|
| 107 |
+
"logps/chosen": -348.89031982421875,
|
| 108 |
+
"logps/rejected": -317.46673583984375,
|
| 109 |
+
"loss": 0.685,
|
| 110 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 111 |
+
"rewards/chosen": 0.03178168088197708,
|
| 112 |
+
"rewards/margins": 0.01709566079080105,
|
| 113 |
+
"rewards/rejected": 0.014686022885143757,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2928870292887029,
|
| 118 |
+
"grad_norm": 1.8450350094954893,
|
| 119 |
+
"learning_rate": 4.456204510851956e-07,
|
| 120 |
+
"logits/chosen": -1.2555049657821655,
|
| 121 |
+
"logits/rejected": -1.3169441223144531,
|
| 122 |
+
"logps/chosen": -344.35125732421875,
|
| 123 |
+
"logps/rejected": -338.9962158203125,
|
| 124 |
+
"loss": 0.6816,
|
| 125 |
+
"rewards/accuracies": 0.640625,
|
| 126 |
+
"rewards/chosen": 0.040135230869054794,
|
| 127 |
+
"rewards/margins": 0.024550432339310646,
|
| 128 |
+
"rewards/rejected": 0.015584799461066723,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.33472803347280333,
|
| 133 |
+
"grad_norm": 1.8585724869830902,
|
| 134 |
+
"learning_rate": 4.2087030056579986e-07,
|
| 135 |
+
"logits/chosen": -1.2849452495574951,
|
| 136 |
+
"logits/rejected": -1.3933959007263184,
|
| 137 |
+
"logps/chosen": -328.14727783203125,
|
| 138 |
+
"logps/rejected": -317.79901123046875,
|
| 139 |
+
"loss": 0.6802,
|
| 140 |
+
"rewards/accuracies": 0.612500011920929,
|
| 141 |
+
"rewards/chosen": 0.03799828886985779,
|
| 142 |
+
"rewards/margins": 0.022710641846060753,
|
| 143 |
+
"rewards/rejected": 0.015287647023797035,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.37656903765690375,
|
| 148 |
+
"grad_norm": 1.9259218744474744,
|
| 149 |
+
"learning_rate": 3.9247834624635404e-07,
|
| 150 |
+
"logits/chosen": -1.2762553691864014,
|
| 151 |
+
"logits/rejected": -1.367308497428894,
|
| 152 |
+
"logps/chosen": -303.9913024902344,
|
| 153 |
+
"logps/rejected": -290.3238830566406,
|
| 154 |
+
"loss": 0.6758,
|
| 155 |
+
"rewards/accuracies": 0.675000011920929,
|
| 156 |
+
"rewards/chosen": 0.04959210380911827,
|
| 157 |
+
"rewards/margins": 0.03906155750155449,
|
| 158 |
+
"rewards/rejected": 0.010530550964176655,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.41841004184100417,
|
| 163 |
+
"grad_norm": 2.007278325554356,
|
| 164 |
+
"learning_rate": 3.610497133404795e-07,
|
| 165 |
+
"logits/chosen": -1.3190323114395142,
|
| 166 |
+
"logits/rejected": -1.3636301755905151,
|
| 167 |
+
"logps/chosen": -319.6354064941406,
|
| 168 |
+
"logps/rejected": -312.177490234375,
|
| 169 |
+
"loss": 0.6759,
|
| 170 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 171 |
+
"rewards/chosen": 0.053187936544418335,
|
| 172 |
+
"rewards/margins": 0.04710903763771057,
|
| 173 |
+
"rewards/rejected": 0.0060789017006754875,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.4602510460251046,
|
| 178 |
+
"grad_norm": 1.8893655511763634,
|
| 179 |
+
"learning_rate": 3.272542485937368e-07,
|
| 180 |
+
"logits/chosen": -1.2609952688217163,
|
| 181 |
+
"logits/rejected": -1.3663299083709717,
|
| 182 |
+
"logps/chosen": -330.7979736328125,
|
| 183 |
+
"logps/rejected": -304.4183654785156,
|
| 184 |
+
"loss": 0.6735,
|
| 185 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 186 |
+
"rewards/chosen": 0.04585585743188858,
|
| 187 |
+
"rewards/margins": 0.048381414264440536,
|
| 188 |
+
"rewards/rejected": -0.0025255570653826,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.502092050209205,
|
| 193 |
+
"grad_norm": 15.033556125532698,
|
| 194 |
+
"learning_rate": 2.9181224366319943e-07,
|
| 195 |
+
"logits/chosen": -1.2779179811477661,
|
| 196 |
+
"logits/rejected": -1.3179519176483154,
|
| 197 |
+
"logps/chosen": -308.0982666015625,
|
| 198 |
+
"logps/rejected": -307.01239013671875,
|
| 199 |
+
"loss": 0.6698,
|
| 200 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 201 |
+
"rewards/chosen": 0.05374608188867569,
|
| 202 |
+
"rewards/margins": 0.05363842099905014,
|
| 203 |
+
"rewards/rejected": 0.00010766675404738635,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.5439330543933054,
|
| 208 |
+
"grad_norm": 1.7668289559235293,
|
| 209 |
+
"learning_rate": 2.55479083351317e-07,
|
| 210 |
+
"logits/chosen": -1.2395732402801514,
|
| 211 |
+
"logits/rejected": -1.33193039894104,
|
| 212 |
+
"logps/chosen": -348.57562255859375,
|
| 213 |
+
"logps/rejected": -316.04498291015625,
|
| 214 |
+
"loss": 0.6689,
|
| 215 |
+
"rewards/accuracies": 0.71875,
|
| 216 |
+
"rewards/chosen": 0.06320692598819733,
|
| 217 |
+
"rewards/margins": 0.07001270353794098,
|
| 218 |
+
"rewards/rejected": -0.006805785000324249,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5857740585774058,
|
| 223 |
+
"grad_norm": 2.0024361354347793,
|
| 224 |
+
"learning_rate": 2.19029145890313e-07,
|
| 225 |
+
"logits/chosen": -1.2449181079864502,
|
| 226 |
+
"logits/rejected": -1.329097867012024,
|
| 227 |
+
"logps/chosen": -326.33868408203125,
|
| 228 |
+
"logps/rejected": -311.33966064453125,
|
| 229 |
+
"loss": 0.6678,
|
| 230 |
+
"rewards/accuracies": 0.653124988079071,
|
| 231 |
+
"rewards/chosen": 0.054462842643260956,
|
| 232 |
+
"rewards/margins": 0.05262676998972893,
|
| 233 |
+
"rewards/rejected": 0.0018360689282417297,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6276150627615062,
|
| 238 |
+
"grad_norm": 1.8787432387858691,
|
| 239 |
+
"learning_rate": 1.8323929841460178e-07,
|
| 240 |
+
"logits/chosen": -1.2301727533340454,
|
| 241 |
+
"logits/rejected": -1.332716703414917,
|
| 242 |
+
"logps/chosen": -326.64910888671875,
|
| 243 |
+
"logps/rejected": -308.0601501464844,
|
| 244 |
+
"loss": 0.6659,
|
| 245 |
+
"rewards/accuracies": 0.65625,
|
| 246 |
+
"rewards/chosen": 0.035623423755168915,
|
| 247 |
+
"rewards/margins": 0.05233610421419144,
|
| 248 |
+
"rewards/rejected": -0.016712676733732224,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.6694560669456067,
|
| 253 |
+
"grad_norm": 1.9176367647768078,
|
| 254 |
+
"learning_rate": 1.488723393865766e-07,
|
| 255 |
+
"logits/chosen": -1.27090322971344,
|
| 256 |
+
"logits/rejected": -1.314906358718872,
|
| 257 |
+
"logps/chosen": -344.8522644042969,
|
| 258 |
+
"logps/rejected": -312.78271484375,
|
| 259 |
+
"loss": 0.6612,
|
| 260 |
+
"rewards/accuracies": 0.6875,
|
| 261 |
+
"rewards/chosen": 0.045879464596509933,
|
| 262 |
+
"rewards/margins": 0.0723222941160202,
|
| 263 |
+
"rewards/rejected": -0.02644282951951027,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.7112970711297071,
|
| 268 |
+
"grad_norm": 2.0586908284329564,
|
| 269 |
+
"learning_rate": 1.1666074087171627e-07,
|
| 270 |
+
"logits/chosen": -1.2575898170471191,
|
| 271 |
+
"logits/rejected": -1.2899234294891357,
|
| 272 |
+
"logps/chosen": -332.93658447265625,
|
| 273 |
+
"logps/rejected": -329.26226806640625,
|
| 274 |
+
"loss": 0.6644,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": 0.03310205787420273,
|
| 277 |
+
"rewards/margins": 0.057156242430210114,
|
| 278 |
+
"rewards/rejected": -0.024054180830717087,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7531380753138075,
|
| 283 |
+
"grad_norm": 1.8213789779768046,
|
| 284 |
+
"learning_rate": 8.729103716819111e-08,
|
| 285 |
+
"logits/chosen": -1.2486387491226196,
|
| 286 |
+
"logits/rejected": -1.2856649160385132,
|
| 287 |
+
"logps/chosen": -332.7528076171875,
|
| 288 |
+
"logps/rejected": -310.4466857910156,
|
| 289 |
+
"loss": 0.6616,
|
| 290 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 291 |
+
"rewards/chosen": 0.047333668917417526,
|
| 292 |
+
"rewards/margins": 0.07636959850788116,
|
| 293 |
+
"rewards/rejected": -0.029035940766334534,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7949790794979079,
|
| 298 |
+
"grad_norm": 1.863301500159963,
|
| 299 |
+
"learning_rate": 6.138919252022435e-08,
|
| 300 |
+
"logits/chosen": -1.2751197814941406,
|
| 301 |
+
"logits/rejected": -1.3302768468856812,
|
| 302 |
+
"logps/chosen": -313.1865539550781,
|
| 303 |
+
"logps/rejected": -310.4251708984375,
|
| 304 |
+
"loss": 0.6592,
|
| 305 |
+
"rewards/accuracies": 0.653124988079071,
|
| 306 |
+
"rewards/chosen": 0.020702695474028587,
|
| 307 |
+
"rewards/margins": 0.06860536336898804,
|
| 308 |
+
"rewards/rejected": -0.0479026660323143,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.8368200836820083,
|
| 313 |
+
"grad_norm": 4.872667925392408,
|
| 314 |
+
"learning_rate": 3.9507259776993954e-08,
|
| 315 |
+
"logits/chosen": -1.2594833374023438,
|
| 316 |
+
"logits/rejected": -1.2656358480453491,
|
| 317 |
+
"logps/chosen": -339.21563720703125,
|
| 318 |
+
"logps/rejected": -337.2076416015625,
|
| 319 |
+
"loss": 0.6604,
|
| 320 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 321 |
+
"rewards/chosen": 0.039656586945056915,
|
| 322 |
+
"rewards/margins": 0.0757637619972229,
|
| 323 |
+
"rewards/rejected": -0.036107175052165985,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8786610878661087,
|
| 328 |
+
"grad_norm": 1.9793278923515518,
|
| 329 |
+
"learning_rate": 2.2111614344599684e-08,
|
| 330 |
+
"logits/chosen": -1.1997952461242676,
|
| 331 |
+
"logits/rejected": -1.318477988243103,
|
| 332 |
+
"logps/chosen": -347.9354553222656,
|
| 333 |
+
"logps/rejected": -335.8204040527344,
|
| 334 |
+
"loss": 0.6613,
|
| 335 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 336 |
+
"rewards/chosen": 0.03624551743268967,
|
| 337 |
+
"rewards/margins": 0.0588444285094738,
|
| 338 |
+
"rewards/rejected": -0.022598912939429283,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.9205020920502092,
|
| 343 |
+
"grad_norm": 1.9125182215206764,
|
| 344 |
+
"learning_rate": 9.57301420397924e-09,
|
| 345 |
+
"logits/chosen": -1.2063685655593872,
|
| 346 |
+
"logits/rejected": -1.248859167098999,
|
| 347 |
+
"logps/chosen": -338.15325927734375,
|
| 348 |
+
"logps/rejected": -323.5889892578125,
|
| 349 |
+
"loss": 0.6584,
|
| 350 |
+
"rewards/accuracies": 0.637499988079071,
|
| 351 |
+
"rewards/chosen": 0.04115135595202446,
|
| 352 |
+
"rewards/margins": 0.07347537577152252,
|
| 353 |
+
"rewards/rejected": -0.03232400864362717,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.9623430962343096,
|
| 358 |
+
"grad_norm": 1.9573083804318165,
|
| 359 |
+
"learning_rate": 2.158697848236607e-09,
|
| 360 |
+
"logits/chosen": -1.2343095541000366,
|
| 361 |
+
"logits/rejected": -1.3073015213012695,
|
| 362 |
+
"logps/chosen": -334.5377197265625,
|
| 363 |
+
"logps/rejected": -336.1081237792969,
|
| 364 |
+
"loss": 0.6611,
|
| 365 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 366 |
+
"rewards/chosen": 0.028930548578500748,
|
| 367 |
+
"rewards/margins": 0.054905764758586884,
|
| 368 |
+
"rewards/rejected": -0.025975216180086136,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.0,
|
| 373 |
+
"step": 239,
|
| 374 |
+
"total_flos": 0.0,
|
| 375 |
+
"train_loss": 0.6731121569996599,
|
| 376 |
+
"train_runtime": 6739.8654,
|
| 377 |
+
"train_samples_per_second": 9.071,
|
| 378 |
+
"train_steps_per_second": 0.035
|
| 379 |
+
}
|
| 380 |
+
],
|
| 381 |
+
"logging_steps": 10,
|
| 382 |
+
"max_steps": 239,
|
| 383 |
+
"num_input_tokens_seen": 0,
|
| 384 |
+
"num_train_epochs": 1,
|
| 385 |
+
"save_steps": 100,
|
| 386 |
+
"stateful_callbacks": {
|
| 387 |
+
"TrainerControl": {
|
| 388 |
+
"args": {
|
| 389 |
+
"should_epoch_stop": false,
|
| 390 |
+
"should_evaluate": false,
|
| 391 |
+
"should_log": false,
|
| 392 |
+
"should_save": true,
|
| 393 |
+
"should_training_stop": true
|
| 394 |
+
},
|
| 395 |
+
"attributes": {}
|
| 396 |
+
}
|
| 397 |
+
},
|
| 398 |
+
"total_flos": 0.0,
|
| 399 |
+
"train_batch_size": 8,
|
| 400 |
+
"trial_name": null,
|
| 401 |
+
"trial_params": null
|
| 402 |
+
}
|