Chranos/enginex-mlu370-vllm
1
0
Fork 0
forked from EngineX-Cambricon/enginex-mlu370-vllm
Code Pull Requests Activity

fix: handle missing tie_word_embeddings attr in MPTConfig

Browse Source 
Use getattr with default True for MPTConfig.tie_word_embeddings,
as some MPT model configs lack this attribute.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Chranos
2026-02-09 17:47:18 +08:00
parent 1d70f93cfc
commit 40dee08f7b
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm-v0.6.2/vllm/model_executor/models/mpt.py
View File

@@ -272,7 +272,7 @@ class MPTForCausalLM(nn.Module, SupportsPP):
config = vllm_config.model_config.hf_config
quant_config = vllm_config.quant_config
self.config = config
assert config.tie_word_embeddings
assert getattr(config, "tie_word_embeddings", True)
self.quant_config = quant_config
self.transformer = MPTModel(vllm_config=vllm_config,