|
--- |
|
license: llama2 |
|
library_name: peft |
|
tags: |
|
- trl |
|
- dpo |
|
- generated_from_trainer |
|
base_model: meta-llama/Llama-2-7b-hf |
|
model-index: |
|
- name: llama_DPO_model_e2 |
|
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. --> |
|
|
|
# llama_DPO_model_e2 |
|
|
|
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0937 |
|
- Rewards/chosen: 0.4389 |
|
- Rewards/rejected: -2.0384 |
|
- Rewards/accuracies: 1.0 |
|
- Rewards/margins: 2.4774 |
|
- Logps/rejected: -205.1940 |
|
- Logps/chosen: -156.2447 |
|
- Logits/rejected: -1.0509 |
|
- Logits/chosen: -0.8587 |
|
|
|
## 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: 8e-07 |
|
- train_batch_size: 1 |
|
- eval_batch_size: 1 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 2 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:| |
|
| 0.673 | 0.1 | 25 | 0.6445 | 0.0273 | -0.0740 | 0.9000 | 0.1013 | -185.5491 | -160.3607 | -1.0521 | -0.8545 | |
|
| 0.5737 | 0.2 | 50 | 0.5485 | 0.0856 | -0.2335 | 0.9933 | 0.3190 | -187.1442 | -159.7781 | -1.0526 | -0.8551 | |
|
| 0.4843 | 0.3 | 75 | 0.4496 | 0.1470 | -0.4343 | 1.0 | 0.5814 | -189.1528 | -159.1637 | -1.0527 | -0.8571 | |
|
| 0.4006 | 0.4 | 100 | 0.3655 | 0.2043 | -0.6419 | 1.0 | 0.8462 | -191.2286 | -158.5909 | -1.0521 | -0.8556 | |
|
| 0.3417 | 0.5 | 125 | 0.2945 | 0.2551 | -0.8630 | 1.0 | 1.1180 | -193.4393 | -158.0833 | -1.0522 | -0.8562 | |
|
| 0.2601 | 0.6 | 150 | 0.2353 | 0.3032 | -1.0903 | 1.0 | 1.3935 | -195.7128 | -157.6020 | -1.0520 | -0.8597 | |
|
| 0.2197 | 0.7 | 175 | 0.1891 | 0.3442 | -1.3124 | 1.0 | 1.6565 | -197.9333 | -157.1923 | -1.0522 | -0.8579 | |
|
| 0.1675 | 0.79 | 200 | 0.1532 | 0.3815 | -1.5253 | 1.0 | 1.9067 | -200.0621 | -156.8192 | -1.0526 | -0.8582 | |
|
| 0.1417 | 0.89 | 225 | 0.1289 | 0.4011 | -1.7082 | 1.0 | 2.1094 | -201.8920 | -156.6225 | -1.0525 | -0.8585 | |
|
| 0.1203 | 0.99 | 250 | 0.1117 | 0.4214 | -1.8534 | 1.0 | 2.2748 | -203.3437 | -156.4196 | -1.0517 | -0.8603 | |
|
| 0.1156 | 1.09 | 275 | 0.1034 | 0.4296 | -1.9336 | 1.0 | 2.3633 | -204.1459 | -156.3377 | -1.0517 | -0.8590 | |
|
| 0.0942 | 1.19 | 300 | 0.0990 | 0.4310 | -1.9823 | 1.0 | 2.4133 | -204.6330 | -156.3240 | -1.0514 | -0.8577 | |
|
| 0.0903 | 1.29 | 325 | 0.0957 | 0.4380 | -2.0137 | 1.0 | 2.4517 | -204.9467 | -156.2539 | -1.0511 | -0.8593 | |
|
| 0.1023 | 1.39 | 350 | 0.0946 | 0.4384 | -2.0296 | 1.0 | 2.4680 | -205.1059 | -156.2503 | -1.0519 | -0.8587 | |
|
| 0.0984 | 1.49 | 375 | 0.0945 | 0.4352 | -2.0350 | 1.0 | 2.4702 | -205.1597 | -156.2819 | -1.0510 | -0.8580 | |
|
| 0.0899 | 1.59 | 400 | 0.0939 | 0.4360 | -2.0393 | 1.0 | 2.4752 | -205.2024 | -156.2742 | -1.0513 | -0.8594 | |
|
| 0.0883 | 1.69 | 425 | 0.0939 | 0.4374 | -2.0378 | 1.0 | 2.4752 | -205.1877 | -156.2598 | -1.0514 | -0.8590 | |
|
| 0.1011 | 1.79 | 450 | 0.0939 | 0.4368 | -2.0412 | 1.0 | 2.4781 | -205.2217 | -156.2654 | -1.0513 | -0.8583 | |
|
| 0.0962 | 1.89 | 475 | 0.0935 | 0.4403 | -2.0395 | 1.0 | 2.4798 | -205.2041 | -156.2308 | -1.0510 | -0.8574 | |
|
| 0.0971 | 1.99 | 500 | 0.0937 | 0.4389 | -2.0384 | 1.0 | 2.4774 | -205.1940 | -156.2447 | -1.0509 | -0.8587 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.8.2 |
|
- Transformers 4.38.1 |
|
- Pytorch 2.2.0+cu118 |
|
- Datasets 2.17.1 |
|
- Tokenizers 0.15.2 |