biondizzle a4ad5898c1 Fix post-quant hook: register on inner model, fix module refs ...

vLLM V1 calls DeepseekV4Model.forward() directly, not
DeepseekV4ForCausalLM.forward(). Hook on the outer model never fires.
Moved hook to self.model (inner) and fixed module.model.layers →
module.layers.

2026-05-18 18:15:36 +00:00

..

patches

Fix post-quant hook: register on inner model, fix module refs

2026-05-18 18:15:36 +00:00

nvfp4_cutedsl.py

HOTFIX: remove NaN checks from run() — torch.isnan().any() does CPU-GPU sync, breaks cudagraph

2026-05-17 22:28:32 +00:00