karylab_agents/vllm
1
0
Fork 0
forked from Karylab-cklius/vllm
Code Pull Requests 2 Actions 1 Packages Activity

[Bugfix] Reject non-nvfp4 dtypes when using the flashinfer_nvlink_one_sided all2all backend (#39717)

Browse Source 
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Tyler Michael Smith
2026-04-13 19:13:51 +00:00
committed by GitHub
co-authored by Claude Opus 4.6
parent 610a3efcaf
commit f72b20976c
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/model_executor/layers/fused_moe/all2all_utils.py
+8
View File
@@ -225,6 +225,14 @@ def maybe_make_prepare_finalize(
elif moe.use_fi_nvl_one_sided_kernels:
assert quant_config is not None
if quant_config.quant_dtype != "nvfp4":
raise ValueError(
"The 'flashinfer_nvlink_one_sided' all2all backend only "
"supports nvfp4 activation quantization, but got "
f"quant_dtype={quant_config.quant_dtype!r}. Use a different "
"all2all backend (e.g. 'flashinfer_nvlink_two_sided' or "
"'allgather_reducescatter') for non-nvfp4 models."
)
max_num_tokens = (
get_current_vllm_config().scheduler_config.max_num_batched_tokens
)