[None][feat] BREAKING: Support Inkling-NVFP4 model - #17062

Open

KleinBlueC wants to merge 52 commits into

NVIDIA:mainfrom

KleinBlueC:inkling-nvfp4-bringup

Commits on Jul 21, 2026

[None][feat] Inkling NVFP4 model bring-up on TensorRT-LLM (_torch)
KleinBlueC
committed
15fdb35

Commits on Jul 23, 2026

[None][fix] Inkling NVFP4 routed-expert activation input-scale (amax/(6*448))
KleinBlueC
committed
5dd2be7

Commits on Jul 24, 2026

[None][fix] Inkling enabled cuda-graph B2: guard tunable_allreduce during graph capture
KleinBlueC
committed
581fc44

Commits on Jul 27, 2026

[None][feat] Inkling multimodal (vision) bring-up: input processor, hMLP tower, image fusion (WIP)
KleinBlueC
committed
ff8ed9f

Commits on Jul 28, 2026

[None][fix] Resolve end_id from model config when tokenizer has no eos_token
KleinBlueC
committed
9a174ea

Commits on Jul 30, 2026

Commits on Jul 31, 2026

Commits on Aug 3, 2026

Commits on Aug 5, 2026

Commits on Aug 6, 2026