Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +8 -0
- train_results.json +9 -0
- trainer_state.json +447 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: gemma
|
| 4 |
+
base_model: google/gemma-2-9b-it
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: simpo-train-largest-30-abs-diff
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# simpo-train-largest-30-abs-diff
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on an unknown dataset.
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 8e-07
|
| 39 |
+
- train_batch_size: 2
|
| 40 |
+
- eval_batch_size: 4
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- num_devices: 4
|
| 44 |
+
- gradient_accumulation_steps: 16
|
| 45 |
+
- total_train_batch_size: 128
|
| 46 |
+
- total_eval_batch_size: 16
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: cosine
|
| 49 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 1
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- Transformers 4.44.2
|
| 59 |
+
- Pytorch 2.7.0+cu128
|
| 60 |
+
- Datasets 2.18.0
|
| 61 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.995910949568378,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 4.287031420826041,
|
| 5 |
+
"train_runtime": 1653.4486,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 10.647,
|
| 8 |
+
"train_steps_per_second": 0.083
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 2,
|
| 4 |
+
"cache_implementation": "hybrid",
|
| 5 |
+
"eos_token_id": 1,
|
| 6 |
+
"pad_token_id": 0,
|
| 7 |
+
"transformers_version": "4.44.2"
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.995910949568378,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 4.287031420826041,
|
| 5 |
+
"train_runtime": 1653.4486,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 10.647,
|
| 8 |
+
"train_steps_per_second": 0.083
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,447 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.995910949568378,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 137,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03634711494775102,
|
| 13 |
+
"grad_norm": 262.34647903554514,
|
| 14 |
+
"learning_rate": 2.857142857142857e-07,
|
| 15 |
+
"logits/chosen": -9.14239501953125,
|
| 16 |
+
"logits/rejected": -9.232463836669922,
|
| 17 |
+
"logps/chosen": -1.7616302967071533,
|
| 18 |
+
"logps/rejected": -1.7519614696502686,
|
| 19 |
+
"loss": 7.6689,
|
| 20 |
+
"rewards/accuracies": 0.53125,
|
| 21 |
+
"rewards/chosen": -17.61629867553711,
|
| 22 |
+
"rewards/margins": -0.09668443351984024,
|
| 23 |
+
"rewards/rejected": -17.519615173339844,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.07269422989550205,
|
| 28 |
+
"grad_norm": 160.26775090527158,
|
| 29 |
+
"learning_rate": 5.714285714285714e-07,
|
| 30 |
+
"logits/chosen": -9.558967590332031,
|
| 31 |
+
"logits/rejected": -9.606477737426758,
|
| 32 |
+
"logps/chosen": -1.4764775037765503,
|
| 33 |
+
"logps/rejected": -1.439325213432312,
|
| 34 |
+
"loss": 6.6349,
|
| 35 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 36 |
+
"rewards/chosen": -14.764776229858398,
|
| 37 |
+
"rewards/margins": -0.3715229332447052,
|
| 38 |
+
"rewards/rejected": -14.3932523727417,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.10904134484325306,
|
| 43 |
+
"grad_norm": 120.66971185794887,
|
| 44 |
+
"learning_rate": 7.998695344323425e-07,
|
| 45 |
+
"logits/chosen": -8.532218933105469,
|
| 46 |
+
"logits/rejected": -8.379935264587402,
|
| 47 |
+
"logps/chosen": -1.2477877140045166,
|
| 48 |
+
"logps/rejected": -1.3596165180206299,
|
| 49 |
+
"loss": 6.3247,
|
| 50 |
+
"rewards/accuracies": 0.5625,
|
| 51 |
+
"rewards/chosen": -12.477876663208008,
|
| 52 |
+
"rewards/margins": 1.1182864904403687,
|
| 53 |
+
"rewards/rejected": -13.596163749694824,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1453884597910041,
|
| 58 |
+
"grad_norm": 78.51354956499227,
|
| 59 |
+
"learning_rate": 7.953121695121394e-07,
|
| 60 |
+
"logits/chosen": -7.22064208984375,
|
| 61 |
+
"logits/rejected": -6.8809967041015625,
|
| 62 |
+
"logps/chosen": -0.9391318559646606,
|
| 63 |
+
"logps/rejected": -1.0672681331634521,
|
| 64 |
+
"loss": 4.9513,
|
| 65 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 66 |
+
"rewards/chosen": -9.391318321228027,
|
| 67 |
+
"rewards/margins": 1.2813628911972046,
|
| 68 |
+
"rewards/rejected": -10.67268180847168,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.18173557473875512,
|
| 73 |
+
"grad_norm": 73.47123034470297,
|
| 74 |
+
"learning_rate": 7.843163833184991e-07,
|
| 75 |
+
"logits/chosen": -6.873021125793457,
|
| 76 |
+
"logits/rejected": -6.335709095001221,
|
| 77 |
+
"logps/chosen": -0.7527015209197998,
|
| 78 |
+
"logps/rejected": -0.8773846626281738,
|
| 79 |
+
"loss": 4.5988,
|
| 80 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 81 |
+
"rewards/chosen": -7.527015686035156,
|
| 82 |
+
"rewards/margins": 1.246830701828003,
|
| 83 |
+
"rewards/rejected": -8.773846626281738,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.21808268968650613,
|
| 88 |
+
"grad_norm": 77.81006578282152,
|
| 89 |
+
"learning_rate": 7.670612634414511e-07,
|
| 90 |
+
"logits/chosen": -6.2256245613098145,
|
| 91 |
+
"logits/rejected": -6.002863883972168,
|
| 92 |
+
"logps/chosen": -0.7618826031684875,
|
| 93 |
+
"logps/rejected": -0.9232344627380371,
|
| 94 |
+
"loss": 4.4284,
|
| 95 |
+
"rewards/accuracies": 0.65625,
|
| 96 |
+
"rewards/chosen": -7.618826389312744,
|
| 97 |
+
"rewards/margins": 1.613519310951233,
|
| 98 |
+
"rewards/rejected": -9.232345581054688,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.25442980463425713,
|
| 103 |
+
"grad_norm": 61.664034648588775,
|
| 104 |
+
"learning_rate": 7.438278427948805e-07,
|
| 105 |
+
"logits/chosen": -6.400006294250488,
|
| 106 |
+
"logits/rejected": -6.255277633666992,
|
| 107 |
+
"logps/chosen": -0.7186794877052307,
|
| 108 |
+
"logps/rejected": -0.9298878908157349,
|
| 109 |
+
"loss": 4.3183,
|
| 110 |
+
"rewards/accuracies": 0.625,
|
| 111 |
+
"rewards/chosen": -7.186795234680176,
|
| 112 |
+
"rewards/margins": 2.1120834350585938,
|
| 113 |
+
"rewards/rejected": -9.29887866973877,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2907769195820082,
|
| 118 |
+
"grad_norm": 68.56556472157719,
|
| 119 |
+
"learning_rate": 7.149945224533862e-07,
|
| 120 |
+
"logits/chosen": -5.755168914794922,
|
| 121 |
+
"logits/rejected": -5.6258063316345215,
|
| 122 |
+
"logps/chosen": -0.8867130279541016,
|
| 123 |
+
"logps/rejected": -1.0443212985992432,
|
| 124 |
+
"loss": 4.4094,
|
| 125 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 126 |
+
"rewards/chosen": -8.8671293258667,
|
| 127 |
+
"rewards/margins": 1.5760822296142578,
|
| 128 |
+
"rewards/rejected": -10.443211555480957,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3271240345297592,
|
| 133 |
+
"grad_norm": 75.32781896911939,
|
| 134 |
+
"learning_rate": 6.810309086608129e-07,
|
| 135 |
+
"logits/chosen": -6.590113639831543,
|
| 136 |
+
"logits/rejected": -6.312002658843994,
|
| 137 |
+
"logps/chosen": -0.8065680265426636,
|
| 138 |
+
"logps/rejected": -1.0438225269317627,
|
| 139 |
+
"loss": 4.1676,
|
| 140 |
+
"rewards/accuracies": 0.65625,
|
| 141 |
+
"rewards/chosen": -8.065679550170898,
|
| 142 |
+
"rewards/margins": 2.3725459575653076,
|
| 143 |
+
"rewards/rejected": -10.438226699829102,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.36347114947751025,
|
| 148 |
+
"grad_norm": 74.83324411670193,
|
| 149 |
+
"learning_rate": 6.424901643866552e-07,
|
| 150 |
+
"logits/chosen": -6.4003729820251465,
|
| 151 |
+
"logits/rejected": -6.2513837814331055,
|
| 152 |
+
"logps/chosen": -0.7952122688293457,
|
| 153 |
+
"logps/rejected": -1.0395910739898682,
|
| 154 |
+
"loss": 4.1941,
|
| 155 |
+
"rewards/accuracies": 0.637499988079071,
|
| 156 |
+
"rewards/chosen": -7.952122688293457,
|
| 157 |
+
"rewards/margins": 2.443786144256592,
|
| 158 |
+
"rewards/rejected": -10.39590835571289,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.39981826442526125,
|
| 163 |
+
"grad_norm": 59.64575454875869,
|
| 164 |
+
"learning_rate": 6e-07,
|
| 165 |
+
"logits/chosen": -6.756447792053223,
|
| 166 |
+
"logits/rejected": -6.111125946044922,
|
| 167 |
+
"logps/chosen": -0.8663609623908997,
|
| 168 |
+
"logps/rejected": -1.0826358795166016,
|
| 169 |
+
"loss": 3.9943,
|
| 170 |
+
"rewards/accuracies": 0.699999988079071,
|
| 171 |
+
"rewards/chosen": -8.663609504699707,
|
| 172 |
+
"rewards/margins": 2.162750005722046,
|
| 173 |
+
"rewards/rejected": -10.826359748840332,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.43616537937301225,
|
| 178 |
+
"grad_norm": 67.48524030832144,
|
| 179 |
+
"learning_rate": 5.542524497952543e-07,
|
| 180 |
+
"logits/chosen": -6.575152397155762,
|
| 181 |
+
"logits/rejected": -6.2960124015808105,
|
| 182 |
+
"logps/chosen": -0.8099439740180969,
|
| 183 |
+
"logps/rejected": -1.0211912393569946,
|
| 184 |
+
"loss": 3.7201,
|
| 185 |
+
"rewards/accuracies": 0.668749988079071,
|
| 186 |
+
"rewards/chosen": -8.09943962097168,
|
| 187 |
+
"rewards/margins": 2.1124720573425293,
|
| 188 |
+
"rewards/rejected": -10.211912155151367,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.4725124943207633,
|
| 193 |
+
"grad_norm": 82.16104654213201,
|
| 194 |
+
"learning_rate": 5.059926008786647e-07,
|
| 195 |
+
"logits/chosen": -6.274717330932617,
|
| 196 |
+
"logits/rejected": -6.062811851501465,
|
| 197 |
+
"logps/chosen": -0.9238764643669128,
|
| 198 |
+
"logps/rejected": -1.0822433233261108,
|
| 199 |
+
"loss": 4.027,
|
| 200 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 201 |
+
"rewards/chosen": -9.238764762878418,
|
| 202 |
+
"rewards/margins": 1.5836679935455322,
|
| 203 |
+
"rewards/rejected": -10.822432518005371,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.5088596092685143,
|
| 208 |
+
"grad_norm": 69.13096378464603,
|
| 209 |
+
"learning_rate": 4.5600645798745166e-07,
|
| 210 |
+
"logits/chosen": -6.3393049240112305,
|
| 211 |
+
"logits/rejected": -6.236004829406738,
|
| 212 |
+
"logps/chosen": -0.7679723501205444,
|
| 213 |
+
"logps/rejected": -0.968636155128479,
|
| 214 |
+
"loss": 3.7213,
|
| 215 |
+
"rewards/accuracies": 0.643750011920929,
|
| 216 |
+
"rewards/chosen": -7.679723262786865,
|
| 217 |
+
"rewards/margins": 2.006639003753662,
|
| 218 |
+
"rewards/rejected": -9.686361312866211,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5452067242162654,
|
| 223 |
+
"grad_norm": 82.34301802283626,
|
| 224 |
+
"learning_rate": 4.051081418863895e-07,
|
| 225 |
+
"logits/chosen": -6.352741241455078,
|
| 226 |
+
"logits/rejected": -6.1597676277160645,
|
| 227 |
+
"logps/chosen": -0.7970303297042847,
|
| 228 |
+
"logps/rejected": -1.043393850326538,
|
| 229 |
+
"loss": 3.7425,
|
| 230 |
+
"rewards/accuracies": 0.71875,
|
| 231 |
+
"rewards/chosen": -7.970303535461426,
|
| 232 |
+
"rewards/margins": 2.4636356830596924,
|
| 233 |
+
"rewards/rejected": -10.433938980102539,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5815538391640164,
|
| 238 |
+
"grad_norm": 72.13584102425693,
|
| 239 |
+
"learning_rate": 3.541266298406398e-07,
|
| 240 |
+
"logits/chosen": -6.617664337158203,
|
| 241 |
+
"logits/rejected": -6.41934871673584,
|
| 242 |
+
"logps/chosen": -0.8421775698661804,
|
| 243 |
+
"logps/rejected": -1.0213863849639893,
|
| 244 |
+
"loss": 3.7518,
|
| 245 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 246 |
+
"rewards/chosen": -8.42177677154541,
|
| 247 |
+
"rewards/margins": 1.7920877933502197,
|
| 248 |
+
"rewards/rejected": -10.213863372802734,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.6179009541117674,
|
| 253 |
+
"grad_norm": 92.05622250652328,
|
| 254 |
+
"learning_rate": 3.0389225412181565e-07,
|
| 255 |
+
"logits/chosen": -6.101731300354004,
|
| 256 |
+
"logits/rejected": -5.849137306213379,
|
| 257 |
+
"logps/chosen": -0.7452789545059204,
|
| 258 |
+
"logps/rejected": -1.045109748840332,
|
| 259 |
+
"loss": 3.5802,
|
| 260 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 261 |
+
"rewards/chosen": -7.452790260314941,
|
| 262 |
+
"rewards/margins": 2.998307466506958,
|
| 263 |
+
"rewards/rejected": -10.45109748840332,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.6542480690595184,
|
| 268 |
+
"grad_norm": 74.65829731544908,
|
| 269 |
+
"learning_rate": 2.5522317844515273e-07,
|
| 270 |
+
"logits/chosen": -6.542179107666016,
|
| 271 |
+
"logits/rejected": -6.144691467285156,
|
| 272 |
+
"logps/chosen": -0.8069820404052734,
|
| 273 |
+
"logps/rejected": -1.0035241842269897,
|
| 274 |
+
"loss": 3.705,
|
| 275 |
+
"rewards/accuracies": 0.675000011920929,
|
| 276 |
+
"rewards/chosen": -8.069819450378418,
|
| 277 |
+
"rewards/margins": 1.9654232263565063,
|
| 278 |
+
"rewards/rejected": -10.035242080688477,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.6905951840072694,
|
| 283 |
+
"grad_norm": 75.40130974842236,
|
| 284 |
+
"learning_rate": 2.0891207259509476e-07,
|
| 285 |
+
"logits/chosen": -6.643575191497803,
|
| 286 |
+
"logits/rejected": -6.608299255371094,
|
| 287 |
+
"logps/chosen": -0.8557837605476379,
|
| 288 |
+
"logps/rejected": -1.0176724195480347,
|
| 289 |
+
"loss": 3.7581,
|
| 290 |
+
"rewards/accuracies": 0.625,
|
| 291 |
+
"rewards/chosen": -8.557836532592773,
|
| 292 |
+
"rewards/margins": 1.6188881397247314,
|
| 293 |
+
"rewards/rejected": -10.176724433898926,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7269422989550205,
|
| 298 |
+
"grad_norm": 83.85915732184031,
|
| 299 |
+
"learning_rate": 1.6571320226872206e-07,
|
| 300 |
+
"logits/chosen": -6.411711692810059,
|
| 301 |
+
"logits/rejected": -6.091231346130371,
|
| 302 |
+
"logps/chosen": -0.8256447911262512,
|
| 303 |
+
"logps/rejected": -1.1177527904510498,
|
| 304 |
+
"loss": 3.7165,
|
| 305 |
+
"rewards/accuracies": 0.737500011920929,
|
| 306 |
+
"rewards/chosen": -8.256447792053223,
|
| 307 |
+
"rewards/margins": 2.9210803508758545,
|
| 308 |
+
"rewards/rejected": -11.177528381347656,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.7632894139027715,
|
| 313 |
+
"grad_norm": 102.74344975744411,
|
| 314 |
+
"learning_rate": 1.2633014440382787e-07,
|
| 315 |
+
"logits/chosen": -7.004655361175537,
|
| 316 |
+
"logits/rejected": -6.623693943023682,
|
| 317 |
+
"logps/chosen": -0.8427413105964661,
|
| 318 |
+
"logps/rejected": -1.0215742588043213,
|
| 319 |
+
"loss": 3.8455,
|
| 320 |
+
"rewards/accuracies": 0.668749988079071,
|
| 321 |
+
"rewards/chosen": -8.427413940429688,
|
| 322 |
+
"rewards/margins": 1.7883294820785522,
|
| 323 |
+
"rewards/rejected": -10.215742111206055,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.7996365288505225,
|
| 328 |
+
"grad_norm": 84.46889768857108,
|
| 329 |
+
"learning_rate": 9.14043280712228e-08,
|
| 330 |
+
"logits/chosen": -6.523610591888428,
|
| 331 |
+
"logits/rejected": -6.330638408660889,
|
| 332 |
+
"logps/chosen": -0.9024786949157715,
|
| 333 |
+
"logps/rejected": -1.0619735717773438,
|
| 334 |
+
"loss": 3.9125,
|
| 335 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 336 |
+
"rewards/chosen": -9.024787902832031,
|
| 337 |
+
"rewards/margins": 1.5949490070343018,
|
| 338 |
+
"rewards/rejected": -10.619735717773438,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8359836437982735,
|
| 343 |
+
"grad_norm": 80.56642671916686,
|
| 344 |
+
"learning_rate": 6.150458756494239e-08,
|
| 345 |
+
"logits/chosen": -7.126093864440918,
|
| 346 |
+
"logits/rejected": -6.942577362060547,
|
| 347 |
+
"logps/chosen": -0.8365545272827148,
|
| 348 |
+
"logps/rejected": -1.044828176498413,
|
| 349 |
+
"loss": 3.6672,
|
| 350 |
+
"rewards/accuracies": 0.625,
|
| 351 |
+
"rewards/chosen": -8.365545272827148,
|
| 352 |
+
"rewards/margins": 2.082737445831299,
|
| 353 |
+
"rewards/rejected": -10.448283195495605,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.8723307587460245,
|
| 358 |
+
"grad_norm": 89.06959408393006,
|
| 359 |
+
"learning_rate": 3.711789783843522e-08,
|
| 360 |
+
"logits/chosen": -6.565405368804932,
|
| 361 |
+
"logits/rejected": -6.4343461990356445,
|
| 362 |
+
"logps/chosen": -0.8695400953292847,
|
| 363 |
+
"logps/rejected": -1.0754826068878174,
|
| 364 |
+
"loss": 3.881,
|
| 365 |
+
"rewards/accuracies": 0.612500011920929,
|
| 366 |
+
"rewards/chosen": -8.695401191711426,
|
| 367 |
+
"rewards/margins": 2.0594258308410645,
|
| 368 |
+
"rewards/rejected": -10.754826545715332,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9086778736937755,
|
| 373 |
+
"grad_norm": 74.5528985533384,
|
| 374 |
+
"learning_rate": 1.8641443178027784e-08,
|
| 375 |
+
"logits/chosen": -7.178065299987793,
|
| 376 |
+
"logits/rejected": -6.637632846832275,
|
| 377 |
+
"logps/chosen": -0.8037236332893372,
|
| 378 |
+
"logps/rejected": -0.9828687906265259,
|
| 379 |
+
"loss": 3.7674,
|
| 380 |
+
"rewards/accuracies": 0.6875,
|
| 381 |
+
"rewards/chosen": -8.037236213684082,
|
| 382 |
+
"rewards/margins": 1.791452169418335,
|
| 383 |
+
"rewards/rejected": -9.828689575195312,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.9450249886415266,
|
| 388 |
+
"grad_norm": 70.70049345302832,
|
| 389 |
+
"learning_rate": 6.376148290617145e-09,
|
| 390 |
+
"logits/chosen": -7.0094404220581055,
|
| 391 |
+
"logits/rejected": -6.915109157562256,
|
| 392 |
+
"logps/chosen": -0.957663357257843,
|
| 393 |
+
"logps/rejected": -1.1495531797409058,
|
| 394 |
+
"loss": 3.7884,
|
| 395 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 396 |
+
"rewards/chosen": -9.57663345336914,
|
| 397 |
+
"rewards/margins": 1.9188988208770752,
|
| 398 |
+
"rewards/rejected": -11.495532989501953,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.9813721035892776,
|
| 403 |
+
"grad_norm": 66.91451872853837,
|
| 404 |
+
"learning_rate": 5.217771643080127e-10,
|
| 405 |
+
"logits/chosen": -6.611492156982422,
|
| 406 |
+
"logits/rejected": -6.406693458557129,
|
| 407 |
+
"logps/chosen": -0.8434920310974121,
|
| 408 |
+
"logps/rejected": -1.1371880769729614,
|
| 409 |
+
"loss": 3.5998,
|
| 410 |
+
"rewards/accuracies": 0.699999988079071,
|
| 411 |
+
"rewards/chosen": -8.434919357299805,
|
| 412 |
+
"rewards/margins": 2.936959743499756,
|
| 413 |
+
"rewards/rejected": -11.371880531311035,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.995910949568378,
|
| 418 |
+
"step": 137,
|
| 419 |
+
"total_flos": 0.0,
|
| 420 |
+
"train_loss": 4.287031420826041,
|
| 421 |
+
"train_runtime": 1653.4486,
|
| 422 |
+
"train_samples_per_second": 10.647,
|
| 423 |
+
"train_steps_per_second": 0.083
|
| 424 |
+
}
|
| 425 |
+
],
|
| 426 |
+
"logging_steps": 5,
|
| 427 |
+
"max_steps": 137,
|
| 428 |
+
"num_input_tokens_seen": 0,
|
| 429 |
+
"num_train_epochs": 1,
|
| 430 |
+
"save_steps": 1000000,
|
| 431 |
+
"stateful_callbacks": {
|
| 432 |
+
"TrainerControl": {
|
| 433 |
+
"args": {
|
| 434 |
+
"should_epoch_stop": false,
|
| 435 |
+
"should_evaluate": false,
|
| 436 |
+
"should_log": false,
|
| 437 |
+
"should_save": true,
|
| 438 |
+
"should_training_stop": true
|
| 439 |
+
},
|
| 440 |
+
"attributes": {}
|
| 441 |
+
}
|
| 442 |
+
},
|
| 443 |
+
"total_flos": 0.0,
|
| 444 |
+
"train_batch_size": 2,
|
| 445 |
+
"trial_name": null,
|
| 446 |
+
"trial_params": null
|
| 447 |
+
}
|