How do I know which ONNX transformation models are available? (Errors when loading models with CDN)

[mz-imhj]

Question

I am using a CDN to load the models, as shown in the code below.
I filtered the models in HuggingFace the way you recommend (text-generation, transformers.js) and put the id of the model I looked up. As I understand it, to change the model, I only need to change the model id.
However, I get an error for each of the below models.

Uncaught (in promise) TypeError: Cannot read properties of undefined (reading 'model')

• HuggingFaceTB/SmolLM2-135M-Instruct
• Xenova/codegen-350M-mono
  ...

Uncaught (in promise) Error: Can't create a session. ERROR_CODE: 1, ERROR_MESSAGE: Deserialize tensor model.layers.4.mlp.gate_proj.MatMul.weight_Q4 failed.Failed to load external data file ""model_q4f16.onnx_data"", error: Module.MountedFiles is not available.

• onnx-community/Phi-3.5-mini-instruct-onnx-web
  ...

I'm ultimately saying that I don't know what model will be available.
Additionally, I was wondering if there is a way to know 'in advance' which 'dtype' and 'device' can be supported.

  import { pipeline } from 'https://cdn.jsdelivr.net/npm/@huggingface/[email protected]';

    generator = await pipeline('text-generation', 'onnx-community/DeepSeek-R1-Distill-Qwen-1.5B-ONNX', {
      dtype: "auto",
      device: "auto",
    });

[benc-uk]

Duplicate of this issue #963 which I've had open since Oct last last year

AFAIK there's a bug in how transformers locates or fetches these external data files