Configuration Parsing Warning: In adapter_config.json: "peft.task_type" must be a string

Qwen3-ASR-1.7B-med-pl-lora-decoder-only

This model is a fine-tuned version of Qwen/Qwen3-ASR-1.7B on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3123
  • Wer: 26.8407
  • Cer: 10.9659

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: 8
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • 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: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.575 1.0 735 0.2944 36.3769 13.9503
0.4698 2.0 1470 0.2650 33.3117 12.4164
0.4318 3.0 2205 0.2514 34.1550 13.6125
0.3622 4.0 2940 0.2498 31.4142 12.2337
0.2891 5.0 3675 0.2525 29.8411 11.5257
0.2421 6.0 4410 0.2621 28.7869 11.4193
0.1971 7.0 5145 0.2737 28.8842 11.7478
0.1624 8.0 5880 0.2841 27.5057 11.0422
0.137 9.0 6615 0.3008 26.8570 10.8687
0.1203 10.0 7350 0.3123 26.8407 10.9659

Framework versions

  • PEFT 0.18.1
  • Transformers 4.57.6
  • Pytorch 2.8.0+cu128
  • Datasets 4.5.0
  • Tokenizers 0.22.2
Downloads last month
2
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for AleksanderObuchowski/Qwen3-ASR-1.7B-med-pl-lora-decoder-only

Adapter
(3)
this model