lmind_hotpot_train8000_eval7405_v1_recite_qa_Qwen_Qwen1.5-4B_lora2
This model is a fine-tuned version of Qwen/Qwen1.5-4B on the tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qa dataset. It achieves the following results on the evaluation set:
- Loss: 0.4804
- Accuracy: 0.7780
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: 1
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- 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: 20.0
Training results
Training Loss | Epoch | Step | Accuracy | Validation Loss |
---|---|---|---|---|
1.5635 | 0.9998 | 1089 | 0.6796 | 1.4615 |
1.4521 | 1.9995 | 2178 | 0.6874 | 1.3626 |
1.2848 | 2.9993 | 3267 | 0.6958 | 1.2575 |
1.1197 | 4.0 | 4357 | 0.7054 | 1.1527 |
0.9756 | 4.9998 | 5446 | 0.7143 | 1.0532 |
0.8393 | 5.9995 | 6535 | 0.7241 | 0.9538 |
0.7125 | 6.9993 | 7624 | 0.7324 | 0.8674 |
0.6144 | 8.0 | 8714 | 0.7404 | 0.7907 |
0.5355 | 8.9998 | 9803 | 0.7469 | 0.7288 |
0.4584 | 9.9977 | 10890 | 0.7531 | 0.6794 |
0.413 | 10.9998 | 11979 | 0.7577 | 0.6292 |
0.3731 | 11.9995 | 13068 | 0.7616 | 0.5926 |
0.3423 | 12.9993 | 14157 | 0.7656 | 0.5620 |
0.3185 | 14.0 | 15247 | 0.7682 | 0.5426 |
0.2924 | 14.9998 | 16336 | 0.7708 | 0.5232 |
0.2824 | 15.9995 | 17425 | 0.7727 | 0.5129 |
0.2669 | 16.9993 | 18514 | 0.7748 | 0.4988 |
0.2517 | 18.0 | 19604 | 0.7762 | 0.4892 |
0.2376 | 18.9998 | 20693 | 0.7773 | 0.4808 |
0.2316 | 19.9977 | 21780 | 0.7780 | 0.4804 |
Framework versions
- PEFT 0.5.0
- Transformers 4.40.2
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 2
Model tree for tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qa_Qwen_Qwen1.5-4B_lora2
Base model
Qwen/Qwen1.5-4B
Adapter
this model
Dataset used to train tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qa_Qwen_Qwen1.5-4B_lora2
Evaluation results
- Accuracy on tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qaself-reported0.778