A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 9, 14:45 UTC
main
[Bugfix] Fix Can't instantiate abstract class DeepseekV32IndexerBackend (#33052)
Chauncey committed 10 months ago
[GLM-OCR] GLM-OCR with MTP Support (#33005)
Yuxuan Zhang committed 10 months ago
[Chore] Update type annotation of `input_ids` in model forward (#33063)
Cyrus Leung committed 10 months ago
[Performance] Tune Mamba selective scan kernel for B200 (#32873)
danisereb committed 10 months ago
[Feature] Add LoRA support for Gemma3 vision components (#32764)
VihaanThat committed 10 months ago
[Misc] HF Hub LoRA Resolver (#20320)
Alex Brooks committed 10 months ago
[Model] Use mm_position to compute mrope positions for Qwen3-Omni (#33010)
Itay Etelis committed 10 months ago
[lora/moe] Improve fused MoE‑LoRA kernel indexing and memory access (#32770)
cwazai committed 10 months ago
[Doc] Further update multi-modal impl doc (#33065)
Cyrus Leung committed 10 months ago
[StepVL] add step vl offline example (#33054)
ltd0924 committed 10 months ago
[Refactor] Use data parser for matching data items to multi-modal UUIDs (#32955)
Cyrus Leung committed 10 months ago
Set splitk=1 for fused-moe-lora expand kernel (#32882)
Danielle Robinson committed 10 months ago
[Model Runner V2] Add LoRAState to consolidate lora logic (#33062)
Woosuk Kwon committed 10 months ago
[Tests] Remove Duplicates (#33032)
Robert Shaw committed 10 months ago
[StepVL] support close img patch (#32923)
ltd0924 committed 10 months ago
Lucas Wilkinson committed 10 months ago
[Model Runner V2] Minor simplification for finish_requests (#33048)
Woosuk Kwon committed 10 months ago
[Model Runner V2] Fix slot_mapping after #25954 (#33046)
Woosuk Kwon committed 10 months ago
[Bugfix][VLM] Fix transformers backend embed_multimodal for Qwen2.5-VL profiling (#32969)
Andreas Karatzas committed 10 months ago
[Model] Use mm_position to compute mrope positions for Qwen2.5-Omni (#32772)
Itay Etelis committed 10 months ago