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-ppl-chosen
|
| 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-ppl-chosen
|
| 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": 4.762774154217574,
|
| 5 |
+
"train_runtime": 2004.0235,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 8.785,
|
| 8 |
+
"train_steps_per_second": 0.068
|
| 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": 4.762774154217574,
|
| 5 |
+
"train_runtime": 2004.0235,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 8.785,
|
| 8 |
+
"train_steps_per_second": 0.068
|
| 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": 292.3499633268165,
|
| 14 |
+
"learning_rate": 2.857142857142857e-07,
|
| 15 |
+
"logits/chosen": -9.351981163024902,
|
| 16 |
+
"logits/rejected": -9.40494441986084,
|
| 17 |
+
"logps/chosen": -1.8519542217254639,
|
| 18 |
+
"logps/rejected": -1.6600620746612549,
|
| 19 |
+
"loss": 8.7269,
|
| 20 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 21 |
+
"rewards/chosen": -18.519542694091797,
|
| 22 |
+
"rewards/margins": -1.9189239740371704,
|
| 23 |
+
"rewards/rejected": -16.60062026977539,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.07269422989550205,
|
| 28 |
+
"grad_norm": 176.49993734667456,
|
| 29 |
+
"learning_rate": 5.714285714285714e-07,
|
| 30 |
+
"logits/chosen": -8.979061126708984,
|
| 31 |
+
"logits/rejected": -9.256832122802734,
|
| 32 |
+
"logps/chosen": -1.6596624851226807,
|
| 33 |
+
"logps/rejected": -1.5709294080734253,
|
| 34 |
+
"loss": 7.2313,
|
| 35 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 36 |
+
"rewards/chosen": -16.59662437438965,
|
| 37 |
+
"rewards/margins": -0.8873310089111328,
|
| 38 |
+
"rewards/rejected": -15.709294319152832,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.10904134484325306,
|
| 43 |
+
"grad_norm": 107.47412480741363,
|
| 44 |
+
"learning_rate": 7.998695344323425e-07,
|
| 45 |
+
"logits/chosen": -8.614570617675781,
|
| 46 |
+
"logits/rejected": -8.825468063354492,
|
| 47 |
+
"logps/chosen": -1.4021928310394287,
|
| 48 |
+
"logps/rejected": -1.2052686214447021,
|
| 49 |
+
"loss": 6.7486,
|
| 50 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 51 |
+
"rewards/chosen": -14.021929740905762,
|
| 52 |
+
"rewards/margins": -1.969242811203003,
|
| 53 |
+
"rewards/rejected": -12.052685737609863,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1453884597910041,
|
| 58 |
+
"grad_norm": 82.01335104879803,
|
| 59 |
+
"learning_rate": 7.953121695121394e-07,
|
| 60 |
+
"logits/chosen": -7.269529819488525,
|
| 61 |
+
"logits/rejected": -7.306704044342041,
|
| 62 |
+
"logps/chosen": -1.0399733781814575,
|
| 63 |
+
"logps/rejected": -1.0558375120162964,
|
| 64 |
+
"loss": 5.8003,
|
| 65 |
+
"rewards/accuracies": 0.5,
|
| 66 |
+
"rewards/chosen": -10.39973258972168,
|
| 67 |
+
"rewards/margins": 0.15864244103431702,
|
| 68 |
+
"rewards/rejected": -10.558377265930176,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.18173557473875512,
|
| 73 |
+
"grad_norm": 72.82132091492836,
|
| 74 |
+
"learning_rate": 7.843163833184991e-07,
|
| 75 |
+
"logits/chosen": -6.520535945892334,
|
| 76 |
+
"logits/rejected": -6.559422969818115,
|
| 77 |
+
"logps/chosen": -0.8438193202018738,
|
| 78 |
+
"logps/rejected": -0.8522235751152039,
|
| 79 |
+
"loss": 5.0458,
|
| 80 |
+
"rewards/accuracies": 0.5,
|
| 81 |
+
"rewards/chosen": -8.438192367553711,
|
| 82 |
+
"rewards/margins": 0.08404362201690674,
|
| 83 |
+
"rewards/rejected": -8.522235870361328,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.21808268968650613,
|
| 88 |
+
"grad_norm": 69.03727598335514,
|
| 89 |
+
"learning_rate": 7.670612634414511e-07,
|
| 90 |
+
"logits/chosen": -5.9485015869140625,
|
| 91 |
+
"logits/rejected": -5.740406513214111,
|
| 92 |
+
"logps/chosen": -0.8595390319824219,
|
| 93 |
+
"logps/rejected": -0.8615515828132629,
|
| 94 |
+
"loss": 4.7981,
|
| 95 |
+
"rewards/accuracies": 0.46875,
|
| 96 |
+
"rewards/chosen": -8.595389366149902,
|
| 97 |
+
"rewards/margins": 0.020125534385442734,
|
| 98 |
+
"rewards/rejected": -8.615514755249023,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.25442980463425713,
|
| 103 |
+
"grad_norm": 71.78320222375913,
|
| 104 |
+
"learning_rate": 7.438278427948805e-07,
|
| 105 |
+
"logits/chosen": -6.031136989593506,
|
| 106 |
+
"logits/rejected": -5.746572494506836,
|
| 107 |
+
"logps/chosen": -0.7891378998756409,
|
| 108 |
+
"logps/rejected": -0.874317467212677,
|
| 109 |
+
"loss": 4.7484,
|
| 110 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 111 |
+
"rewards/chosen": -7.891378879547119,
|
| 112 |
+
"rewards/margins": 0.8517962694168091,
|
| 113 |
+
"rewards/rejected": -8.743175506591797,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2907769195820082,
|
| 118 |
+
"grad_norm": 72.82722028196132,
|
| 119 |
+
"learning_rate": 7.149945224533862e-07,
|
| 120 |
+
"logits/chosen": -5.47674560546875,
|
| 121 |
+
"logits/rejected": -5.537765979766846,
|
| 122 |
+
"logps/chosen": -0.7930837869644165,
|
| 123 |
+
"logps/rejected": -0.904737651348114,
|
| 124 |
+
"loss": 4.6279,
|
| 125 |
+
"rewards/accuracies": 0.581250011920929,
|
| 126 |
+
"rewards/chosen": -7.930837154388428,
|
| 127 |
+
"rewards/margins": 1.1165391206741333,
|
| 128 |
+
"rewards/rejected": -9.04737663269043,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3271240345297592,
|
| 133 |
+
"grad_norm": 61.23739520942703,
|
| 134 |
+
"learning_rate": 6.810309086608129e-07,
|
| 135 |
+
"logits/chosen": -5.746423721313477,
|
| 136 |
+
"logits/rejected": -5.778194427490234,
|
| 137 |
+
"logps/chosen": -0.7503607273101807,
|
| 138 |
+
"logps/rejected": -0.8812963366508484,
|
| 139 |
+
"loss": 4.7314,
|
| 140 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 141 |
+
"rewards/chosen": -7.503607749938965,
|
| 142 |
+
"rewards/margins": 1.3093562126159668,
|
| 143 |
+
"rewards/rejected": -8.81296443939209,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.36347114947751025,
|
| 148 |
+
"grad_norm": 68.26509051330353,
|
| 149 |
+
"learning_rate": 6.424901643866552e-07,
|
| 150 |
+
"logits/chosen": -5.5795392990112305,
|
| 151 |
+
"logits/rejected": -5.432155609130859,
|
| 152 |
+
"logps/chosen": -0.816428005695343,
|
| 153 |
+
"logps/rejected": -0.9023356437683105,
|
| 154 |
+
"loss": 4.7166,
|
| 155 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 156 |
+
"rewards/chosen": -8.16427993774414,
|
| 157 |
+
"rewards/margins": 0.8590763211250305,
|
| 158 |
+
"rewards/rejected": -9.023356437683105,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.39981826442526125,
|
| 163 |
+
"grad_norm": 71.6111666575235,
|
| 164 |
+
"learning_rate": 6e-07,
|
| 165 |
+
"logits/chosen": -5.5865983963012695,
|
| 166 |
+
"logits/rejected": -5.635308265686035,
|
| 167 |
+
"logps/chosen": -0.7537165880203247,
|
| 168 |
+
"logps/rejected": -0.9402713775634766,
|
| 169 |
+
"loss": 4.4952,
|
| 170 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 171 |
+
"rewards/chosen": -7.537165641784668,
|
| 172 |
+
"rewards/margins": 1.8655481338500977,
|
| 173 |
+
"rewards/rejected": -9.402713775634766,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.43616537937301225,
|
| 178 |
+
"grad_norm": 65.19861764331375,
|
| 179 |
+
"learning_rate": 5.542524497952543e-07,
|
| 180 |
+
"logits/chosen": -5.944064140319824,
|
| 181 |
+
"logits/rejected": -5.957856178283691,
|
| 182 |
+
"logps/chosen": -0.7788850665092468,
|
| 183 |
+
"logps/rejected": -0.9255874752998352,
|
| 184 |
+
"loss": 4.3522,
|
| 185 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 186 |
+
"rewards/chosen": -7.7888503074646,
|
| 187 |
+
"rewards/margins": 1.4670238494873047,
|
| 188 |
+
"rewards/rejected": -9.255874633789062,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.4725124943207633,
|
| 193 |
+
"grad_norm": 67.86424317787042,
|
| 194 |
+
"learning_rate": 5.059926008786647e-07,
|
| 195 |
+
"logits/chosen": -5.869328498840332,
|
| 196 |
+
"logits/rejected": -5.912378787994385,
|
| 197 |
+
"logps/chosen": -0.7861538529396057,
|
| 198 |
+
"logps/rejected": -0.944945216178894,
|
| 199 |
+
"loss": 4.3518,
|
| 200 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 201 |
+
"rewards/chosen": -7.861537933349609,
|
| 202 |
+
"rewards/margins": 1.587913990020752,
|
| 203 |
+
"rewards/rejected": -9.449451446533203,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.5088596092685143,
|
| 208 |
+
"grad_norm": 70.15593128180441,
|
| 209 |
+
"learning_rate": 4.5600645798745166e-07,
|
| 210 |
+
"logits/chosen": -6.250380039215088,
|
| 211 |
+
"logits/rejected": -6.102661609649658,
|
| 212 |
+
"logps/chosen": -0.7886101603507996,
|
| 213 |
+
"logps/rejected": -0.9173639416694641,
|
| 214 |
+
"loss": 4.2954,
|
| 215 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 216 |
+
"rewards/chosen": -7.886102199554443,
|
| 217 |
+
"rewards/margins": 1.2875380516052246,
|
| 218 |
+
"rewards/rejected": -9.173639297485352,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5452067242162654,
|
| 223 |
+
"grad_norm": 72.06139720995746,
|
| 224 |
+
"learning_rate": 4.051081418863895e-07,
|
| 225 |
+
"logits/chosen": -6.077569484710693,
|
| 226 |
+
"logits/rejected": -5.909134864807129,
|
| 227 |
+
"logps/chosen": -0.8054354786872864,
|
| 228 |
+
"logps/rejected": -1.0184094905853271,
|
| 229 |
+
"loss": 4.1856,
|
| 230 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 231 |
+
"rewards/chosen": -8.054354667663574,
|
| 232 |
+
"rewards/margins": 2.129739284515381,
|
| 233 |
+
"rewards/rejected": -10.18409538269043,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5815538391640164,
|
| 238 |
+
"grad_norm": 111.49055555981266,
|
| 239 |
+
"learning_rate": 3.541266298406398e-07,
|
| 240 |
+
"logits/chosen": -6.036823749542236,
|
| 241 |
+
"logits/rejected": -5.949917793273926,
|
| 242 |
+
"logps/chosen": -0.8323701024055481,
|
| 243 |
+
"logps/rejected": -0.9793428182601929,
|
| 244 |
+
"loss": 4.0504,
|
| 245 |
+
"rewards/accuracies": 0.612500011920929,
|
| 246 |
+
"rewards/chosen": -8.323701858520508,
|
| 247 |
+
"rewards/margins": 1.4697275161743164,
|
| 248 |
+
"rewards/rejected": -9.793428421020508,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.6179009541117674,
|
| 253 |
+
"grad_norm": 81.14697844302252,
|
| 254 |
+
"learning_rate": 3.0389225412181565e-07,
|
| 255 |
+
"logits/chosen": -5.898052215576172,
|
| 256 |
+
"logits/rejected": -5.901252746582031,
|
| 257 |
+
"logps/chosen": -0.7811166644096375,
|
| 258 |
+
"logps/rejected": -1.0059183835983276,
|
| 259 |
+
"loss": 4.2063,
|
| 260 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 261 |
+
"rewards/chosen": -7.811167240142822,
|
| 262 |
+
"rewards/margins": 2.248016595840454,
|
| 263 |
+
"rewards/rejected": -10.059184074401855,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.6542480690595184,
|
| 268 |
+
"grad_norm": 77.67069971551602,
|
| 269 |
+
"learning_rate": 2.5522317844515273e-07,
|
| 270 |
+
"logits/chosen": -6.16554069519043,
|
| 271 |
+
"logits/rejected": -6.2015275955200195,
|
| 272 |
+
"logps/chosen": -0.8136342167854309,
|
| 273 |
+
"logps/rejected": -1.0194966793060303,
|
| 274 |
+
"loss": 4.0071,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": -8.136343002319336,
|
| 277 |
+
"rewards/margins": 2.058624267578125,
|
| 278 |
+
"rewards/rejected": -10.194966316223145,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.6905951840072694,
|
| 283 |
+
"grad_norm": 67.22921956049545,
|
| 284 |
+
"learning_rate": 2.0891207259509476e-07,
|
| 285 |
+
"logits/chosen": -6.2798662185668945,
|
| 286 |
+
"logits/rejected": -6.4411468505859375,
|
| 287 |
+
"logps/chosen": -0.780041515827179,
|
| 288 |
+
"logps/rejected": -0.9418380856513977,
|
| 289 |
+
"loss": 4.2225,
|
| 290 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 291 |
+
"rewards/chosen": -7.8004150390625,
|
| 292 |
+
"rewards/margins": 1.6179659366607666,
|
| 293 |
+
"rewards/rejected": -9.418381690979004,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7269422989550205,
|
| 298 |
+
"grad_norm": 71.88699575292867,
|
| 299 |
+
"learning_rate": 1.6571320226872206e-07,
|
| 300 |
+
"logits/chosen": -6.111327171325684,
|
| 301 |
+
"logits/rejected": -5.933043003082275,
|
| 302 |
+
"logps/chosen": -0.79816734790802,
|
| 303 |
+
"logps/rejected": -1.0315134525299072,
|
| 304 |
+
"loss": 4.1323,
|
| 305 |
+
"rewards/accuracies": 0.6875,
|
| 306 |
+
"rewards/chosen": -7.981672763824463,
|
| 307 |
+
"rewards/margins": 2.3334617614746094,
|
| 308 |
+
"rewards/rejected": -10.31513500213623,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.7632894139027715,
|
| 313 |
+
"grad_norm": 78.83415408435721,
|
| 314 |
+
"learning_rate": 1.2633014440382787e-07,
|
| 315 |
+
"logits/chosen": -6.445625305175781,
|
| 316 |
+
"logits/rejected": -6.3378729820251465,
|
| 317 |
+
"logps/chosen": -0.7882248163223267,
|
| 318 |
+
"logps/rejected": -0.938172459602356,
|
| 319 |
+
"loss": 4.2202,
|
| 320 |
+
"rewards/accuracies": 0.637499988079071,
|
| 321 |
+
"rewards/chosen": -7.882249355316162,
|
| 322 |
+
"rewards/margins": 1.4994752407073975,
|
| 323 |
+
"rewards/rejected": -9.38172435760498,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.7996365288505225,
|
| 328 |
+
"grad_norm": 84.93158033781451,
|
| 329 |
+
"learning_rate": 9.14043280712228e-08,
|
| 330 |
+
"logits/chosen": -5.919665813446045,
|
| 331 |
+
"logits/rejected": -5.9659295082092285,
|
| 332 |
+
"logps/chosen": -0.807255744934082,
|
| 333 |
+
"logps/rejected": -0.9430872797966003,
|
| 334 |
+
"loss": 4.3819,
|
| 335 |
+
"rewards/accuracies": 0.59375,
|
| 336 |
+
"rewards/chosen": -8.07255744934082,
|
| 337 |
+
"rewards/margins": 1.3583143949508667,
|
| 338 |
+
"rewards/rejected": -9.430871963500977,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8359836437982735,
|
| 343 |
+
"grad_norm": 81.4577593129811,
|
| 344 |
+
"learning_rate": 6.150458756494239e-08,
|
| 345 |
+
"logits/chosen": -5.945127964019775,
|
| 346 |
+
"logits/rejected": -5.908052921295166,
|
| 347 |
+
"logps/chosen": -0.8148506879806519,
|
| 348 |
+
"logps/rejected": -0.9752057790756226,
|
| 349 |
+
"loss": 4.1354,
|
| 350 |
+
"rewards/accuracies": 0.643750011920929,
|
| 351 |
+
"rewards/chosen": -8.148507118225098,
|
| 352 |
+
"rewards/margins": 1.6035501956939697,
|
| 353 |
+
"rewards/rejected": -9.752057075500488,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.8723307587460245,
|
| 358 |
+
"grad_norm": 86.67335944638306,
|
| 359 |
+
"learning_rate": 3.711789783843522e-08,
|
| 360 |
+
"logits/chosen": -5.938599109649658,
|
| 361 |
+
"logits/rejected": -6.1739959716796875,
|
| 362 |
+
"logps/chosen": -0.818856418132782,
|
| 363 |
+
"logps/rejected": -0.9494014978408813,
|
| 364 |
+
"loss": 4.1957,
|
| 365 |
+
"rewards/accuracies": 0.581250011920929,
|
| 366 |
+
"rewards/chosen": -8.18856430053711,
|
| 367 |
+
"rewards/margins": 1.305450677871704,
|
| 368 |
+
"rewards/rejected": -9.494014739990234,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9086778736937755,
|
| 373 |
+
"grad_norm": 73.2996295516593,
|
| 374 |
+
"learning_rate": 1.8641443178027784e-08,
|
| 375 |
+
"logits/chosen": -5.637792110443115,
|
| 376 |
+
"logits/rejected": -5.634159088134766,
|
| 377 |
+
"logps/chosen": -0.7925742268562317,
|
| 378 |
+
"logps/rejected": -1.0251243114471436,
|
| 379 |
+
"loss": 4.0793,
|
| 380 |
+
"rewards/accuracies": 0.643750011920929,
|
| 381 |
+
"rewards/chosen": -7.925743103027344,
|
| 382 |
+
"rewards/margins": 2.325500011444092,
|
| 383 |
+
"rewards/rejected": -10.251241683959961,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.9450249886415266,
|
| 388 |
+
"grad_norm": 75.69723178942408,
|
| 389 |
+
"learning_rate": 6.376148290617145e-09,
|
| 390 |
+
"logits/chosen": -5.835230350494385,
|
| 391 |
+
"logits/rejected": -5.948345184326172,
|
| 392 |
+
"logps/chosen": -0.820753276348114,
|
| 393 |
+
"logps/rejected": -1.0133521556854248,
|
| 394 |
+
"loss": 4.191,
|
| 395 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 396 |
+
"rewards/chosen": -8.20753288269043,
|
| 397 |
+
"rewards/margins": 1.9259881973266602,
|
| 398 |
+
"rewards/rejected": -10.13352108001709,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.9813721035892776,
|
| 403 |
+
"grad_norm": 80.16820363004321,
|
| 404 |
+
"learning_rate": 5.217771643080127e-10,
|
| 405 |
+
"logits/chosen": -6.059390068054199,
|
| 406 |
+
"logits/rejected": -6.196063041687012,
|
| 407 |
+
"logps/chosen": -0.7539313435554504,
|
| 408 |
+
"logps/rejected": -0.9142172932624817,
|
| 409 |
+
"loss": 4.1238,
|
| 410 |
+
"rewards/accuracies": 0.668749988079071,
|
| 411 |
+
"rewards/chosen": -7.539312839508057,
|
| 412 |
+
"rewards/margins": 1.602860450744629,
|
| 413 |
+
"rewards/rejected": -9.142173767089844,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.995910949568378,
|
| 418 |
+
"step": 137,
|
| 419 |
+
"total_flos": 0.0,
|
| 420 |
+
"train_loss": 4.762774154217574,
|
| 421 |
+
"train_runtime": 2004.0235,
|
| 422 |
+
"train_samples_per_second": 8.785,
|
| 423 |
+
"train_steps_per_second": 0.068
|
| 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 |
+
}
|