adl-hw2-qwen3

This model is a fine-tuned version of Qwen/Qwen3-4B on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0910

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: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 128
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: constant_with_warmup
  • lr_scheduler_warmup_steps: 30
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss
15.2427 0.128 10 0.3634
0.18 0.256 20 0.1202
0.1177 0.384 30 0.1100
0.1055 0.512 40 0.1040
0.1045 0.64 50 0.0994
0.0964 0.768 60 0.0965
0.0977 0.896 70 0.0948
0.0925 1.0128 80 0.0936
0.0914 1.1408 90 0.0930
0.0882 1.2688 100 0.0920
0.0868 1.3968 110 0.0919
0.0899 1.5248 120 0.0915
0.0862 1.6528 130 0.0914
0.0894 1.7808 140 0.0909
0.0891 1.9088 150 0.0910

Framework versions

  • PEFT 0.17.1
  • Transformers 4.56.1
  • Pytorch 2.8.0+cu126
  • Datasets 4.0.0
  • Tokenizers 0.22.0
Downloads last month
2
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for wangwang0204/adl-hw2-qwen3

Base model

Qwen/Qwen3-4B-Base
Finetuned
Qwen/Qwen3-4B
Adapter
(113)
this model

Evaluation results