fix cuda ut fail for transformers v5.0 #1357
xin3he
merged
20 commits into
main
from
hengguo/fix_v5_cuda_ut
fix cuda ut fail for transformers v5.0
23caeec1
set default eval batch size to avoid None case
f48306fb
fix cpu llama4 issue
ba01ce87
fix typo
ff9cef42
use monkey_patch to resolve no_init_weights issue of auto_gptq
50779390
add skip for transformers issue
0b297c4f
add skip for moe and add mock_cuda_capability for fp8 model input
72e1cecb
wenhuach21
commented
on 2026-01-28
wenhuach21
commented
on 2026-01-28
update expected outputs for transformers 5.0.0 and refine model name …
1d2b0d12
Merge branch 'main' into hengguo/fix_v5_cuda_ut
f1d94e41
remove default batch size
076948bb
keep moe failure waiting for #1345
07ced708
fix
7c753c98
fix gptqmodel
f2b4af24
merge
ede38216
codescan
4c81a3df
update
5e6fada3
fix spwan error of vllm
dfd4a111
![pre-commit-ci[bot]](https://avatars.githubusercontent.com/in/68672?v=4&s=60)
[pre-commit.ci] auto fixes from pre-commit.com hooks
5a35fd16
Merge branch 'main' into hengguo/fix_v5_cuda_ut
53bf7f3f
xin3he
approved these changes
on 2026-01-29
update lm_eval requirement
112d7baa
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub