llama-3-8b-instruct-simpo
This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the princeton-nlp/llama3-ultrafeedback dataset. It achieves the following results on the evaluation set:
- Loss: 1.3755
- Rewards/chosen: -2.9448
- Rewards/rejected: -3.6038
- Rewards/accuracies: 0.6613
- Rewards/margins: 0.6589
- Logps/rejected: -1.4415
- Logps/chosen: -1.1779
- Logits/rejected: -1.1545
- Logits/chosen: -1.1873
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: 1e-06
- train_batch_size: 2
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- total_eval_batch_size: 32
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.3975 | 0.8549 | 400 | 1.3755 | -2.9448 | -3.6038 | 0.6613 | 0.6589 | -1.4415 | -1.1779 | -1.1545 | -1.1873 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.2.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 0