aiql01
This model is a fine-tuned version of NousResearch/Llama-2-7b-chat-hf on the Agent Instruct dataset. It achieves the following results on the evaluation set:
- Loss: 0.3156
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: 0.0002
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.5579 | 0.09 | 20 | 0.6611 |
| 0.3651 | 0.18 | 40 | 0.4642 |
| 0.3704 | 0.26 | 60 | 0.3942 |
| 0.3868 | 0.35 | 80 | 0.3672 |
| 0.2843 | 0.44 | 100 | 0.3513 |
| 0.2535 | 0.53 | 120 | 0.3479 |
| 0.3797 | 0.61 | 140 | 0.3406 |
| 0.3702 | 0.7 | 160 | 0.3347 |
| 0.2986 | 0.79 | 180 | 0.3345 |
| 0.2952 | 0.88 | 200 | 0.3249 |
| 0.2109 | 0.96 | 220 | 0.3272 |
| 0.2744 | 1.05 | 240 | 0.3275 |
| 0.1762 | 1.14 | 260 | 0.3249 |
| 0.28 | 1.23 | 280 | 0.3247 |
| 0.2997 | 1.31 | 300 | 0.3263 |
| 0.2115 | 1.4 | 320 | 0.3212 |
| 0.252 | 1.49 | 340 | 0.3229 |
| 0.2687 | 1.58 | 360 | 0.3246 |
| 0.1996 | 1.66 | 380 | 0.3205 |
| 0.2227 | 1.75 | 400 | 0.3164 |
| 0.5474 | 1.84 | 420 | 0.3145 |
| 0.2224 | 1.93 | 440 | 0.3121 |
| 0.3695 | 2.02 | 460 | 0.3129 |
| 0.1738 | 2.1 | 480 | 0.3182 |
| 0.1299 | 2.19 | 500 | 0.3178 |
| 0.2314 | 2.28 | 520 | 0.3159 |
| 0.2391 | 2.37 | 540 | 0.3179 |
| 0.2131 | 2.45 | 560 | 0.3178 |
| 0.1541 | 2.54 | 580 | 0.3154 |
| 0.2768 | 2.63 | 600 | 0.3153 |
| 0.1362 | 2.72 | 620 | 0.3155 |
| 0.2494 | 2.8 | 640 | 0.3157 |
| 0.1641 | 2.89 | 660 | 0.3154 |
| 0.1397 | 2.98 | 680 | 0.3156 |
Framework versions
- Transformers 4.34.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1