metadata
license: mit
base_model: HuggingFaceH4/mistral-7b-sft-beta
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model is a fine-tuned version of HuggingFaceH4/mistral-7b-sft-beta on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4965
- Rewards/chosen: -2.9708
- Rewards/rejected: -4.3017
- Rewards/accuracies: 0.7695
- Rewards/margins: 1.3309
- Logps/rejected: -687.5271
- Logps/chosen: -554.1226
- Logits/rejected: -0.1928
- Logits/chosen: -0.6531
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 3
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
---|---|---|---|---|---|---|---|---|---|---|---|
0.5326 | 0.11 | 100 | 0.6180 | -0.4024 | -0.6993 | 0.6797 | 0.2969 | -327.2873 | -297.2842 | -2.5800 | -2.5958 |
0.4709 | 0.23 | 200 | 0.5608 | -1.1383 | -1.7616 | 0.7109 | 0.6233 | -433.5121 | -370.8716 | -2.1515 | -2.1720 |
0.4289 | 0.34 | 300 | 0.5293 | -1.5404 | -2.3958 | 0.7539 | 0.8554 | -496.9380 | -411.0811 | -2.0882 | -2.1204 |
0.4195 | 0.45 | 400 | 0.5096 | -1.7916 | -2.8995 | 0.7812 | 1.1079 | -547.3041 | -436.1970 | -1.0571 | -1.2976 |
0.3891 | 0.57 | 500 | 0.5086 | -2.6047 | -3.9255 | 0.7812 | 1.3208 | -649.9016 | -517.5072 | -0.8608 | -1.1314 |
0.4182 | 0.68 | 600 | 0.4976 | -2.4968 | -3.7962 | 0.7695 | 1.2994 | -636.9742 | -506.7195 | -0.4354 | -0.8384 |
0.3845 | 0.79 | 700 | 0.4967 | -2.6976 | -4.0084 | 0.7695 | 1.3108 | -658.1885 | -526.7999 | -0.2826 | -0.7200 |
0.3896 | 0.91 | 800 | 0.4965 | -2.9708 | -4.3017 | 0.7695 | 1.3309 | -687.5271 | -554.1226 | -0.1928 | -0.6531 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1