Tencent / Tencent/AngelSlim

KeyError: 'midlayer.self_attn.qkv_proj.weight' When Loading Draft Model Qwen3-14B-EAGLE3 in SGLang

Open
#39 1 comment 0 reactions 1 assignee View on GitHub

@liusong1222 is already working on this.

Since Aug 13, 2025.

Dominant language
Python
Stars
1.7k
Forks
181
Avg merge
1d 3h
Merged PRs (30d)
2

Description

When launching an SGLang server with the Qwen3-14B model and the Qwen3-14B-EAGLE3 draft model for speculative decoding, I encounter a KeyError: 'midlayer.self_attn.qkv_proj.weight' during model weight loading. This error occurs in the qwen3.py module of SGLang.

File "/opt/miniconda/envs/sglang/lib/python3.12/site-packages/sglang/srt/model_loader/loader.py", line 437, in load_model
self.load_weights_and_postprocess(
File "/opt/miniconda/envs/sglang/lib/python3.12/site-packages/sglang/srt/model_loader/loader.py", line 445, in load_weights_and_postprocess
model.load_weights(weights)
File "/opt/miniconda/envs/sglang/lib/python3.12/site-packages/sglang/srt/models/qwen3.py", line 493, in load_weights
param = params_dict[name]
~~~~~~~~~~~^^^^^^
KeyError: 'midlayer.self_attn.qkv_proj.weight'

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.