metadata
license: llama2
library_name: peft
tags:
- trl
- dpo
- generated_from_trainer
base_model: meta-llama/Llama-2-7b-chat-hf
model-index:
- name: model_hh_usp4_400
results: []
model_hh_usp4_400
This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 4.8447
- Rewards/chosen: -12.8728
- Rewards/rejected: -15.9738
- Rewards/accuracies: 0.6200
- Rewards/margins: 3.1010
- Logps/rejected: -130.8977
- Logps/chosen: -125.8617
- Logits/rejected: -0.9554
- Logits/chosen: -0.9016
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: 0.0005
- train_batch_size: 4
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0115 | 4.0 | 100 | 2.1095 | -12.3873 | -13.8364 | 0.6200 | 1.4491 | -128.5228 | -125.3223 | -0.2640 | -0.2414 |
| 0.0271 | 8.0 | 200 | 3.0011 | -0.7751 | -1.2662 | 0.5200 | 0.4911 | -114.5559 | -112.4198 | -0.1794 | -0.1752 |
| 0.0029 | 12.0 | 300 | 5.8848 | -30.6815 | -32.8502 | 0.5300 | 2.1687 | -149.6492 | -145.6491 | -0.9305 | -0.8856 |
| 0.0 | 16.0 | 400 | 4.8706 | -12.8599 | -15.9629 | 0.6200 | 3.1030 | -130.8855 | -125.8473 | -0.9564 | -0.9029 |
| 0.0 | 20.0 | 500 | 4.8654 | -12.8852 | -15.9625 | 0.6200 | 3.0773 | -130.8851 | -125.8754 | -0.9556 | -0.9016 |
| 0.0 | 24.0 | 600 | 4.8663 | -12.8617 | -15.9688 | 0.6200 | 3.1071 | -130.8921 | -125.8493 | -0.9551 | -0.9013 |
| 0.0 | 28.0 | 700 | 4.8599 | -12.8977 | -15.9664 | 0.6200 | 3.0686 | -130.8894 | -125.8893 | -0.9553 | -0.9015 |
| 0.0 | 32.0 | 800 | 4.8410 | -12.8636 | -15.9914 | 0.6200 | 3.1277 | -130.9172 | -125.8515 | -0.9553 | -0.9014 |
| 0.0 | 36.0 | 900 | 4.8425 | -12.8856 | -15.9582 | 0.6200 | 3.0726 | -130.8803 | -125.8759 | -0.9551 | -0.9014 |
| 0.0 | 40.0 | 1000 | 4.8447 | -12.8728 | -15.9738 | 0.6200 | 3.1010 | -130.8977 | -125.8617 | -0.9554 | -0.9016 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2