model_shp1_dpo1
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: 2.0112
- Rewards/chosen: -9.7625
- Rewards/rejected: -9.2926
- Rewards/accuracies: 0.4700
- Rewards/margins: -0.4699
- Logps/rejected: -307.4124
- Logps/chosen: -345.0927
- Logits/rejected: -1.0692
- Logits/chosen: -1.0975
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0741 | 2.67 | 100 | 1.1825 | -3.1899 | -3.0393 | 0.4800 | -0.1506 | -244.8796 | -279.3668 | -1.2034 | -1.2620 |
| 0.0016 | 5.33 | 200 | 2.1179 | -8.9597 | -8.2224 | 0.4100 | -0.7372 | -296.7111 | -337.0645 | -1.1154 | -1.1503 |
| 0.0001 | 8.0 | 300 | 1.9624 | -9.5308 | -9.0562 | 0.4500 | -0.4746 | -305.0487 | -342.7763 | -1.0878 | -1.1168 |
| 0.0001 | 10.67 | 400 | 1.9799 | -9.6041 | -9.1296 | 0.4500 | -0.4745 | -305.7831 | -343.5089 | -1.0797 | -1.1079 |
| 0.0001 | 13.33 | 500 | 1.9938 | -9.6787 | -9.2063 | 0.4500 | -0.4724 | -306.5495 | -344.2545 | -1.0746 | -1.1031 |
| 0.0001 | 16.0 | 600 | 2.0046 | -9.7222 | -9.2446 | 0.4600 | -0.4776 | -306.9330 | -344.6898 | -1.0722 | -1.0999 |
| 0.0001 | 18.67 | 700 | 2.0079 | -9.7525 | -9.2749 | 0.4500 | -0.4776 | -307.2361 | -344.9933 | -1.0706 | -1.0984 |
| 0.0001 | 21.33 | 800 | 2.0091 | -9.7588 | -9.2867 | 0.4600 | -0.4721 | -307.3541 | -345.0561 | -1.0699 | -1.0978 |
| 0.0001 | 24.0 | 900 | 2.0158 | -9.7704 | -9.2915 | 0.4500 | -0.4789 | -307.4015 | -345.1719 | -1.0694 | -1.0975 |
| 0.0001 | 26.67 | 1000 | 2.0112 | -9.7625 | -9.2926 | 0.4700 | -0.4699 | -307.4124 | -345.0927 | -1.0692 | -1.0975 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.1
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 0
Model tree for guoyu-zhang/model_shp1_dpo1
Base model
meta-llama/Llama-2-7b-chat-hf