lmind_nq_train6000_eval6489_v1_docidx_v3_1e-4_lora2
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: 5.3535
- Accuracy: 0.4374
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: 0.0001
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 50.0
Training results
Training Loss | Epoch | Step | Accuracy | Validation Loss |
---|---|---|---|---|
1.3892 | 1.0 | 341 | 0.4544 | 3.4056 |
1.3499 | 2.0 | 683 | 0.4577 | 3.4531 |
1.2427 | 3.0 | 1024 | 0.4584 | 3.6711 |
1.1231 | 4.0 | 1366 | 0.4570 | 3.8000 |
0.995 | 5.0 | 1707 | 0.4552 | 3.9532 |
0.8693 | 6.0 | 2049 | 0.4526 | 4.0766 |
0.7302 | 7.0 | 2390 | 0.4501 | 4.1717 |
0.6033 | 8.0 | 2732 | 0.448 | 4.2778 |
0.4825 | 9.0 | 3073 | 0.4462 | 4.3415 |
0.387 | 10.0 | 3415 | 0.4463 | 4.4131 |
0.2933 | 11.0 | 3756 | 0.4434 | 4.4906 |
0.2344 | 12.0 | 4098 | 0.4425 | 4.6517 |
0.1919 | 13.0 | 4439 | 0.4408 | 4.7515 |
0.1581 | 14.0 | 4781 | 0.4421 | 4.7323 |
0.1429 | 15.0 | 5122 | 0.4407 | 4.8101 |
0.1279 | 16.0 | 5464 | 0.4406 | 4.8482 |
0.1231 | 17.0 | 5805 | 0.4411 | 4.9735 |
0.1145 | 18.0 | 6147 | 0.4415 | 5.0121 |
0.1087 | 19.0 | 6488 | 0.4394 | 4.9836 |
0.1084 | 20.0 | 6830 | 0.4388 | 5.1171 |
0.1069 | 21.0 | 7171 | 0.4405 | 5.0120 |
0.1075 | 22.0 | 7513 | 0.44 | 5.2343 |
0.1024 | 23.0 | 7854 | 0.4409 | 5.1501 |
0.0981 | 24.0 | 8196 | 0.4403 | 5.0801 |
0.097 | 25.0 | 8537 | 0.4416 | 5.1037 |
0.0963 | 26.0 | 8879 | 0.4398 | 5.2064 |
0.0983 | 27.0 | 9220 | 0.4414 | 5.0664 |
0.0969 | 28.0 | 9562 | 0.4410 | 5.2559 |
0.0966 | 29.0 | 9903 | 0.4404 | 5.1960 |
0.0954 | 30.0 | 10245 | 0.4396 | 5.2238 |
0.0931 | 31.0 | 10586 | 0.4402 | 5.2195 |
0.0923 | 32.0 | 10928 | 0.4407 | 5.2871 |
0.0911 | 33.0 | 11269 | 0.4392 | 5.3201 |
0.0934 | 34.0 | 11611 | 0.4387 | 5.3628 |
0.091 | 35.0 | 11952 | 0.4390 | 5.3197 |
0.0902 | 36.0 | 12294 | 0.4391 | 5.1868 |
0.0916 | 37.0 | 12635 | 0.4424 | 5.1227 |
0.0905 | 38.0 | 12977 | 0.4367 | 5.2214 |
0.0907 | 39.0 | 13318 | 0.4412 | 5.2412 |
0.0883 | 40.0 | 13660 | 0.4395 | 5.3015 |
0.0892 | 41.0 | 14001 | 0.4392 | 5.2816 |
0.0881 | 42.0 | 14343 | 0.4351 | 5.3583 |
0.0881 | 43.0 | 14684 | 0.4365 | 5.2678 |
0.0898 | 44.0 | 15026 | 0.4372 | 5.3854 |
0.0874 | 45.0 | 15345 | 5.3568 | 0.4392 |
0.088 | 46.0 | 15687 | 5.3908 | 0.4358 |
0.0885 | 47.0 | 16028 | 5.2685 | 0.4366 |
0.0872 | 48.0 | 16370 | 5.3500 | 0.44 |
0.0869 | 49.0 | 16711 | 5.3612 | 0.4372 |
0.0864 | 49.99 | 17050 | 5.3535 | 0.4374 |
Framework versions
- Transformers 4.34.0
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.14.1
Model tree for tyzhu/lmind_nq_train6000_eval6489_v1_docidx_v3_1e-4_lora2
Base model
meta-llama/Llama-2-7b-hf