[pull] main from NVIDIA:main#342

Merged

pull[bot] merged 8 commits intoxinhe-nv:mainfrom

Aug 1, 2025

Commits on Jul 31, 2025

[feat] Auto-enable ngram with concurrency <= 32. (#6232)

SimengLiu-nv
and
mikeiovine
authored

Commits on Aug 1, 2025

fix: Fix poor generation with FP8 Gemma3 1B checkpoint (#6499)
brb-nv
authored
chore: Improve the AutoTuner log information. (#6368)
hyukn
authored
[TRTLLM-6611][feat] Add warnings and stricter validation to LoraManager adapter loading (#6453)
venkywonka
authored
Deepseek R1 FP8 Support on Blackwell (#6486)

authored
fix: remove duplicate layer multiplication in KV cache size calculation (#6481)
jaedeok-nvidia
authored
[fix] Fix missing fields in xqa kernel cache key (#6282)
lowsfer
authored
[TRTLLM-6364][infra] Validate for PR titles to ensure they follow the required format (#6278)

niukuo
and
chzblych
authored