[BUG] Opening of model from HF fails with ValueError: Missing parameters: lm_head.weight.

[workflowsguy]

Describe the bug Opening mlx-community/Qwen3-Embedding-8B-4bit-DWQ from HF fails with ValueError: Missing parameters: lm_head.weight.

The error output is:

Traceback (most recent call last):

  File "/Users/guy/Library/Application Support/CodeRunner/Debuggers/pdb.crDebugger/pdb_server.py", line 40, in main
    p._run(pdb._ScriptTarget(p.mainpyfile))
  File "/opt/local/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/pdb.py", line 1754, in _run
    self.run(target.code)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/bdb.py", line 627, in run
    exec(cmd, globals, locals)
  File "", line 1, in
  File "/Users/guy/Library/Application Support/CodeRunner/Unsaved/Untitled.py", line 5, in
    model, tokenizer = load("/Users/guy/.lmstudio/models/mlx-community/Qwen3-Embedding-8B-4bit-DWQ")
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/mlx_lm/utils.py", line 241, in load
    model, config = load_model(model_path, lazy)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/mlx_lm/utils.py", line 202, in load_model
    model.load_weights(list(weights.items()), strict=strict)
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/mlx/nn/layers/base.py", line 181, in load_weights
    raise ValueError(f"Missing parameters: {missing}.")

To Reproduce

1. Download the model from here
2. follow the instructions from the model's page

Include code snippet

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Qwen3-Embedding-8B-4bit-DWQ")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)

Expected behavior Some kind of output, no error

Desktop (please complete the following information):

• OS Version: macOS 15.5
• mlx: 0.26.1
• mlx-lm: 0.25.2

[christopherowen]

@workflowsguy https://huggingface.co/mlx-community/Qwen3-Embedding-8B-4bit-DWQ/discussions/2

mlx-lm doesn't support embedding models (at least for qwen2 and qwen3).

I have been struggling to convert nomic-embed-code (based on qwen2) to mlx.

[workflowsguy]

Thank you for pointing this out. As a novice user, I was simply following the "MLX" tag, which is obviously misleading/premature(?) here