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

Qwen3-ASR-0.6B-med-pl-lora-decoder-only

This model is a fine-tuned version of Qwen/Qwen3-ASR-0.6B on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4958
  • Wer: 38.6529
  • Cer: 15.6408

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
0.7471 1.0 501 0.5691 52.4302 20.2179
0.6244 2.0 1002 0.4613 45.6181 17.2191
0.5278 3.0 1503 0.4229 44.3169 16.8220
0.4237 4.0 2004 0.4113 39.8775 14.4188
0.3687 5.0 2505 0.4136 39.9923 15.2691
0.3252 6.0 3006 0.4214 40.6812 15.4422
0.2576 7.0 3507 0.4430 40.1454 16.1703
0.2157 8.0 4008 0.4639 37.0455 14.3628
0.1941 9.0 4509 0.4803 36.2419 13.4973
0.1774 10.0 5010 0.4958 38.6529 15.6408

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
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for AleksanderObuchowski/Qwen3-ASR-0.6B-med-pl-lora-decoder-only

Adapter
(1)
this model