xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-02 23:59:58 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Skip loading extra bias for Qwen2-VL GPTQ-Int8 (#8442)

Browse Source
This commit is contained in:
Jee Jee Li 2024-09-13 15:58:28 +08:00 committed by GitHub
parent cab69a15e4
commit 06311e2956
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/model_executor/models/qwen2_vl.py
View File

@ -1055,6 +1055,9 @@ class Qwen2VLForConditionalGeneration(nn.Module, SupportsMultiModal):
if weight_name not in name:
continue
name = name.replace(weight_name, param_name)
# Skip loading extra bias for GPTQ models.
if name.endswith(".bias") and name not in params_dict:
continue
param = params_dict[name]
weight_loader = param.weight_loader
weight_loader(param, loaded_weight, shard_id)
@ -1078,6 +1081,9 @@ class Qwen2VLForConditionalGeneration(nn.Module, SupportsMultiModal):
loaded_weight = loaded_weight.transpose(0, 1)
loaded_weight = loaded_weight.reshape(-1)
try:
# Skip loading extra bias for GPTQ models.
if name.endswith(".bias") and name not in params_dict:
continue
param = params_dict[name]
except KeyError:
print(params_dict.keys())