{
  "id": 583538,
  "title": "How to use LoRA on timm？peft？",
  "url": "/competitions/waveform-inversion/discussion/583538",
  "author_name": "water joe",
  "post_date": "2025-06-07T17:17:27.144000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I tried to use peft's LoRA, but it seems that timm doesn't support this library<br>\n<code>\nTraceback (most recent call last):\n[rank1]:   File \"/kaggle/working/_train.py\", line 358, in &lt;module&gt;\n[rank1]:   File \"/kaggle/working/_train.py\", line 264, in main\n[rank1]:     out = ema_model.module(x)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/kaggle/working/_model.py\", line 371, in forward\n[rank1]:     x= self.backbone(x)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/peft_model.py\", line 2721, in forward\n[rank1]:     return self.base_model(\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/tuners/tuners_utils.py\", line 197, in forward\n[rank1]:     return self.model.forward(*args, **kwargs)\n[rank1]: TypeError: FeatureListNet.forward() got an unexpected keyword argument 'input_ids'</code></p>",
  "messages": [
    {
      "id": 3219408,
      "postDate": "2025-06-07T17:17:27.143Z",
      "content": "<p>I tried to use peft's LoRA, but it seems that timm doesn't support this library<br>\n<code>\nTraceback (most recent call last):\n[rank1]:   File \"/kaggle/working/_train.py\", line 358, in &lt;module&gt;\n[rank1]:   File \"/kaggle/working/_train.py\", line 264, in main\n[rank1]:     out = ema_model.module(x)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/kaggle/working/_model.py\", line 371, in forward\n[rank1]:     x= self.backbone(x)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/peft_model.py\", line 2721, in forward\n[rank1]:     return self.base_model(\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/tuners/tuners_utils.py\", line 197, in forward\n[rank1]:     return self.model.forward(*args, **kwargs)\n[rank1]: TypeError: FeatureListNet.forward() got an unexpected keyword argument 'input_ids'</code></p>",
      "rawMarkdown": "I tried to use peft's LoRA, but it seems that timm doesn't support this library\n`\nTraceback (most recent call last):\n[rank1]:   File \"/kaggle/working/_train.py\", line 358, in <module>\n[rank1]:     main(cfg)\n[rank1]:   File \"/kaggle/working/_train.py\", line 264, in main\n[rank1]:     out = ema_model.module(x)\n[rank1]:           ^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/kaggle/working/_model.py\", line 371, in forward\n[rank1]:     x= self.backbone(x)\n[rank1]:        ^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/peft_model.py\", line 2721, in forward\n[rank1]:     return self.base_model(\n[rank1]:            ^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/tuners/tuners_utils.py\", line 197, in forward\n[rank1]:     return self.model.forward(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]: TypeError: FeatureListNet.forward() got an unexpected keyword argument 'input_ids'`",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3219408": "I tried to use peft's LoRA, but it seems that timm doesn't support this library\n`\nTraceback (most recent call last):\n[rank1]:   File \"/kaggle/working/_train.py\", line 358, in <module>\n[rank1]:     main(cfg)\n[rank1]:   File \"/kaggle/working/_train.py\", line 264, in main\n[rank1]:     out = ema_model.module(x)\n[rank1]:           ^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/kaggle/working/_model.py\", line 371, in forward\n[rank1]:     x= self.backbone(x)\n[rank1]:        ^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/peft_model.py\", line 2721, in forward\n[rank1]:     return self.base_model(\n[rank1]:            ^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1739, in _wrapped_call_impl\n[rank1]:     return self._call_impl(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py\", line 1750, in _call_impl\n[rank1]:     return forward_call(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]:   File \"/usr/local/lib/python3.11/dist-packages/peft/tuners/tuners_utils.py\", line 197, in forward\n[rank1]:     return self.model.forward(*args, **kwargs)\n[rank1]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n[rank1]: TypeError: FeatureListNet.forward() got an unexpected keyword argument 'input_ids'`"
  }
}