Edit model card

lmind_hotpot_train8000_eval7405_v1_recite_qa_Qwen_Qwen1.5-4B_3e-4_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: 0.4756
  • Accuracy: 0.7764

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: 0.0003
  • 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 Validation Loss Accuracy
1.5356 0.9998 1089 1.3711 0.6864
1.3102 1.9995 2178 1.1753 0.7020
1.0549 2.9993 3267 1.0095 0.7164
0.8461 4.0 4357 0.8722 0.7297
0.701 4.9998 5446 0.7641 0.7406
0.5977 5.9995 6535 0.6797 0.7490
0.5238 6.9993 7624 0.6209 0.7559
0.4742 8.0 8714 0.5837 0.7600
0.438 8.9998 9803 0.5532 0.7638
0.402 9.9995 10892 0.5331 0.7664
0.383 10.9993 11981 0.5156 0.7685
0.3627 12.0 13071 0.5070 0.7702
0.3521 12.9998 14160 0.4984 0.7714
0.344 13.9995 15249 0.4925 0.7722
0.3341 14.9993 16338 0.4847 0.7736
0.3275 16.0 17428 0.4808 0.7748
0.3223 16.9998 18517 0.4776 0.7751
0.3155 17.9995 19606 0.4804 0.7758
0.3033 18.9993 20695 0.4787 0.7761
0.2989 19.9954 21780 0.4756 0.7764

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
0
Unable to determine this model’s pipeline type. Check the docs .

Adapter for

Dataset used to train tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qa_Qwen_Qwen1.5-4B_3e-4_lora2

Evaluation results

  • Accuracy on tyzhu/lmind_hotpot_train8000_eval7405_v1_recite_qa
    self-reported
    0.776