metadata
license: llama3
base_model: meta-llama/Meta-Llama-3-8B-Instruct
tags:
- alignment-handbook
- generated_from_trainer
datasets:
- princeton-nlp/llama3-ultrafeedback
model-index:
- name: llama-3-8b-instruct-cpo-beta-0.1
results: []
llama-3-8b-instruct-cpo-beta-0.1
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: 1.3274
- Rewards/chosen: -13.1885
- Rewards/rejected: -14.3989
- Rewards/accuracies: 0.7016
- Rewards/margins: 1.2104
- Logps/rejected: -143.9887
- Logps/chosen: -131.8848
- Logits/rejected: 0.2135
- Logits/chosen: 0.1708
- Nll Loss: 0.3287
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: 16
- gradient_accumulation_steps: 4
- 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 | Nll Loss |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.2968 | 0.8547 | 400 | 1.3274 | -13.1885 | -14.3989 | 0.7016 | 1.2104 | -143.9887 | -131.8848 | 0.2135 | 0.1708 | 0.3287 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+rocm6.0
- Datasets 2.19.2
- Tokenizers 0.19.1