model_hh_usp1_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: 3.4291
- Rewards/chosen: -2.1852
- Rewards/rejected: -10.4536
- Rewards/accuracies: 0.6900
- Rewards/margins: 8.2684
- Logps/rejected: -125.6639
- Logps/chosen: -112.8688
- Logits/rejected: -0.9672
- Logits/chosen: -0.9637
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0025 | 4.0 | 100 | 1.8113 | -1.7152 | -5.1212 | 0.6100 | 3.4059 | -119.7389 | -112.3466 | -0.1554 | -0.1598 |
| 0.1942 | 8.0 | 200 | 3.6090 | -1.4379 | -8.2063 | 0.6100 | 6.7684 | -123.1668 | -112.0384 | -1.0994 | -1.1187 |
| 0.0502 | 12.0 | 300 | 3.3229 | -9.0906 | -16.5854 | 0.6200 | 7.4948 | -132.4769 | -120.5415 | -0.9988 | -1.0079 |
| 0.0 | 16.0 | 400 | 3.4296 | -2.1656 | -10.3972 | 0.6900 | 8.2316 | -125.6012 | -112.8470 | -0.9657 | -0.9623 |
| 0.0 | 20.0 | 500 | 3.4471 | -2.1796 | -10.4172 | 0.7100 | 8.2376 | -125.6234 | -112.8626 | -0.9676 | -0.9637 |
| 0.0 | 24.0 | 600 | 3.4031 | -2.1735 | -10.4669 | 0.7000 | 8.2933 | -125.6786 | -112.8558 | -0.9675 | -0.9640 |
| 0.0 | 28.0 | 700 | 3.4346 | -2.1542 | -10.4272 | 0.7000 | 8.2730 | -125.6345 | -112.8343 | -0.9673 | -0.9639 |
| 0.0 | 32.0 | 800 | 3.4246 | -2.1606 | -10.4103 | 0.6900 | 8.2497 | -125.6157 | -112.8415 | -0.9675 | -0.9642 |
| 0.0 | 36.0 | 900 | 3.4315 | -2.1805 | -10.4501 | 0.7000 | 8.2696 | -125.6599 | -112.8635 | -0.9674 | -0.9639 |
| 0.0 | 40.0 | 1000 | 3.4291 | -2.1852 | -10.4536 | 0.6900 | 8.2684 | -125.6639 | -112.8688 | -0.9672 | -0.9637 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 2
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Model tree for guoyu-zhang/model_hh_usp1_400
Base model
meta-llama/Llama-2-7b-chat-hf