|
--- |
|
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: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# model_hh_usp4_400 |
|
|
|
This model is a fine-tuned version of [meta-llama/Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 4.4266 |
|
- Rewards/chosen: -7.2918 |
|
- Rewards/rejected: -9.3870 |
|
- Rewards/accuracies: 0.5500 |
|
- Rewards/margins: 2.0952 |
|
- Logps/rejected: -122.5611 |
|
- Logps/chosen: -121.2051 |
|
- Logits/rejected: -0.2787 |
|
- Logits/chosen: -0.2572 |
|
|
|
## 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 | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:| |
|
| 0.098 | 4.0 | 100 | 2.3307 | -3.2289 | -4.9359 | 0.5700 | 1.7070 | -117.6155 | -116.6908 | -0.5136 | -0.5011 | |
|
| 0.2615 | 8.0 | 200 | 3.5637 | -3.5399 | -4.5546 | 0.5700 | 1.0147 | -117.1918 | -117.0363 | -0.4837 | -0.4844 | |
|
| 0.0137 | 12.0 | 300 | 4.2146 | -3.4955 | -5.8321 | 0.5600 | 2.3366 | -118.6113 | -116.9870 | -0.3503 | -0.3327 | |
|
| 0.0 | 16.0 | 400 | 4.4247 | -7.2840 | -9.3968 | 0.5500 | 2.1128 | -122.5721 | -121.1964 | -0.2788 | -0.2574 | |
|
| 0.0 | 20.0 | 500 | 4.4045 | -7.2800 | -9.4193 | 0.5600 | 2.1393 | -122.5971 | -121.1920 | -0.2793 | -0.2578 | |
|
| 0.0 | 24.0 | 600 | 4.4242 | -7.2774 | -9.3711 | 0.5600 | 2.0936 | -122.5435 | -121.1891 | -0.2789 | -0.2573 | |
|
| 0.0 | 28.0 | 700 | 4.4048 | -7.2951 | -9.4062 | 0.5600 | 2.1110 | -122.5825 | -121.2088 | -0.2785 | -0.2570 | |
|
| 0.0 | 32.0 | 800 | 4.4098 | -7.2804 | -9.3847 | 0.5500 | 2.1043 | -122.5586 | -121.1924 | -0.2783 | -0.2569 | |
|
| 0.0 | 36.0 | 900 | 4.4251 | -7.2849 | -9.3768 | 0.5500 | 2.0918 | -122.5498 | -121.1974 | -0.2792 | -0.2575 | |
|
| 0.0 | 40.0 | 1000 | 4.4266 | -7.2918 | -9.3870 | 0.5500 | 2.0952 | -122.5611 | -121.2051 | -0.2787 | -0.2572 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.10.0 |
|
- Transformers 4.39.3 |
|
- Pytorch 2.2.2+cu121 |
|
- Datasets 2.18.0 |
|
- Tokenizers 0.15.2 |
