chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

ministral3 (#14251)

Browse Source 
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Yueming Yuan <yy28@illinois.edu>
This commit is contained in:
Xinyuan Tong
2025-12-04 22:31:26 +00:00
committed by GitHub
parent c1006fd8a1
commit 6d37e70883
14 changed files with 245 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
python/sglang/srt/model_executor/model_runner.py
View File

@@ -583,8 +583,10 @@ class ModelRunner:
quantization_config := getattr(
self.model_config.hf_config, "quantization_config", None
)
) is not None and "weight_block_size" in quantization_config:
weight_block_size_n = quantization_config["weight_block_size"][0]
) is not None and (
weight_block_size := quantization_config.get("weight_block_size", None)
) is not None:
weight_block_size_n = weight_block_size[0]
if self.tp_size % self.moe_ep_size != 0:
raise ValueError(