Llama-2-7b-spin-10k
This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1026
- Rewards/real: 7.5091
- Rewards/generated: -8.0379
- Rewards/accuracies: 1.0
- Rewards/margins: 15.5469
- Logps/generated: -363.0305
- Logps/real: -104.4774
- Logits/generated: -0.5236
- Logits/real: -0.9459
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: 5e-07
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/real | Rewards/generated | Rewards/accuracies | Rewards/margins | Logps/generated | Logps/real | Logits/generated | Logits/real |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.1495 | 0.1984 | 62 | 0.1401 | 3.7274 | -5.5913 | 1.0 | 9.3186 | -338.5644 | -142.2943 | -0.5586 | -0.5198 |
| 0.1087 | 0.3968 | 124 | 0.1060 | 7.1500 | -4.2208 | 1.0 | 11.3708 | -324.8601 | -108.0682 | -0.6577 | -1.0069 |
| 0.1056 | 0.5952 | 186 | 0.1046 | 7.2683 | -5.6243 | 1.0 | 12.8927 | -338.8952 | -106.8850 | -0.5520 | -0.9362 |
| 0.1037 | 0.7936 | 248 | 0.1041 | 7.3329 | -5.6913 | 1.0 | 13.0242 | -339.5646 | -106.2389 | -0.5560 | -0.9504 |
| 0.1041 | 0.992 | 310 | 0.1037 | 7.3755 | -6.3330 | 1.0 | 13.7085 | -345.9819 | -105.8133 | -0.5095 | -0.9077 |
| 0.0976 | 1.1904 | 372 | 0.1035 | 7.4053 | -7.5036 | 1.0 | 14.9089 | -357.6875 | -105.5148 | -0.5378 | -0.9621 |
| 0.1018 | 1.3888 | 434 | 0.1034 | 7.4118 | -7.9940 | 1.0 | 15.4059 | -362.5919 | -105.4498 | -0.5389 | -0.9673 |
| 0.0991 | 1.5872 | 496 | 0.1031 | 7.4489 | -6.9160 | 1.0 | 14.3649 | -351.8115 | -105.0788 | -0.5154 | -0.9266 |
| 0.0954 | 1.7856 | 558 | 0.1029 | 7.4703 | -7.2607 | 1.0 | 14.7310 | -355.2591 | -104.8652 | -0.5039 | -0.9100 |
| 0.0995 | 1.984 | 620 | 0.1028 | 7.4973 | -7.3534 | 1.0 | 14.8507 | -356.1862 | -104.5955 | -0.5304 | -0.9424 |
| 0.095 | 2.1824 | 682 | 0.1028 | 7.4894 | -7.6075 | 1.0 | 15.0969 | -358.7269 | -104.6745 | -0.5408 | -0.9674 |
| 0.0964 | 2.3808 | 744 | 0.1027 | 7.4957 | -7.5378 | 1.0 | 15.0335 | -358.0298 | -104.6114 | -0.5268 | -0.9397 |
| 0.1003 | 2.5792 | 806 | 0.1026 | 7.5077 | -7.7372 | 1.0 | 15.2449 | -360.0238 | -104.4909 | -0.5189 | -0.9338 |
| 0.099 | 2.7776 | 868 | 0.1026 | 7.5125 | -7.9795 | 1.0 | 15.4919 | -362.4467 | -104.4437 | -0.5214 | -0.9428 |
| 0.1053 | 2.976 | 930 | 0.1026 | 7.5091 | -8.0379 | 1.0 | 15.5469 | -363.0305 | -104.4774 | -0.5236 | -0.9459 |
Framework versions
- Transformers 4.43.3
- Pytorch 2.2.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 31
Model tree for AmberYifan/Llama-2-7b-spin-10k
Base model
meta-llama/Llama-2-7b-hf