Skip to content

High-performance, light-weight C++ LLM and VLM Inference Software for Physical AI

C++ · active 2026-01-062026-07-23 (UTC)

Partial coverage20,524 / 26,280 hourly files (78%) · 2 absent upstream · 5,753 failed, retryable2023-08-152026-08-13 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
268
Pushes
8
Pull requests
12
Issues
28
Stars
120
Forks
19

Activity over time

Daily event counts in the loaded window

Line chart, 199 days from 2026-01-06 to 2026-07-23. Pushes: 8 total, peak 2 in a day. Pull requests: 12 total, peak 3 in a day. Issues: 28 total, peak 5 in a day. Comments: 66 total, peak 8 in a day. Stars: 120 total, peak 24 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

Recent activity

Latest issues, pull requests and releases

  • Issue#109robertvandervoort2026-06-17 01:53
    CustomVoice language conditioning bug confirmed in v0.8.0 — working fix applied (ref #87 / PR #98)
  • Issue comment#83nvluxiaoz2026-05-30 01:17
    Qwen3-30B-A3B-GPTQ-Int4 export failed
  • Issue comment#93nvluxiaoz2026-05-30 01:16
    Experimental OpenAI server: SSE chunks missing required `object` field
  • Issue comment#99fans-nv2026-05-28 00:04
    [Bug] Qwen3.5-4B-AWQ uses ~13GB RAM on Orin NX 16GB — GDN mamba workspace pool prevents ASR+TTS+LLM co-residency
  • Issue comment#92nvluxiaoz2026-05-27 00:05
    Engine build error for Alpamayo-R1-10B action head
  • Issue comment#89lix199372026-05-21 02:54
    Nemotron-3-Nano-Omni-30B-A3B-Reasoning-NVFP4 (LLM part) failed to convert to TensorRT Engine
  • Issue comment#84nvluxiaoz2026-05-18 03:07
    Qwen3.5-0.8B from onnx convert to engine error
  • Issue comment#82nvluxiaoz2026-05-12 00:20
    Add ViT attention plugin FMHA cubins and packed cu_seqlens support
  • Issue#80MrWANGXUHUI2026-05-06 04:55
    Low throughput on Thor U with Qwen3-30B-A3B-GPTQ-Int4 compared to NVIDIA reported performance
  • Issue comment#45nvluxiaoz2026-04-25 14:06
    How to find the documentation for DriveOS release?
  • Pull request#73crouchggj2026-04-18 03:17
  • Issue comment#63nvamberl2026-04-13 10:22
    板端编译使用 --maxLoraRank参数报错
  • Issue comment#64YSF-A2026-04-09 02:50
    QST: can C++ Runtime build and run on x86 cpu + gpu such as L20/H20
  • Issue comment#638150347622026-04-08 05:39
    板端编译使用 --maxLoraRank参数报错
  • Issue comment#57jump-drive2026-03-21 08:32
    [Feature] XQA kernel for sm_101, nqpkv=16 (Nemotron 3 Super GQA support)
  • Issue comment#57nvluxiaoz2026-03-21 06:25
    [Feature] XQA kernel for sm_101, nqpkv=16 (Nemotron 3 Super GQA support)
  • Issue comment#57jump-drive2026-03-21 04:48
    [Feature] XQA kernel for sm_101, nqpkv=16 (Nemotron 3 Super GQA support)
  • Issue#56jump-drive2026-03-20 19:02
    [Bug] NemotronH config export fails: num_hidden_layers not found
  • Issue#54lxb3242026-03-20 03:42
    Question: Why is BF16 data type not supported?
  • Issue#54lxb3242026-03-20 03:42
    Question: Why is BF16 data type not supported?
  • Issue comment#46nvluxiaoz2026-03-20 01:33
    feat: Add multi-stage multi-arch Containerfile
  • Issue comment#52nvluxiaoz2026-03-20 01:29
    Support static CUDA runtime linking
  • Issue comment#44nvluxiaoz2026-03-18 03:30
    Support More Models like Qwen3.5-0.8b/4b/9b,Qwen3.5-35b-A3B
  • Issue comment#50nvamberl2026-03-17 09:24
    [Bug] CUDA Error 700 (Illegal Memory Access) in Myelin shapeChangeHelper During Prefill After Optimization Profile Switch# [Bug] CUDA Error 700 (Illegal Memory Access) in Myelin shapeChangeHelper During Prefill After Optimization Profile Switch
  • Issue comment#32nvluxiaoz2026-03-16 23:50
    Developer Roadmap for TensorRT Edge-LLM (H1 2026)

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 120 stars here means stars gained during the window, not the repo's star count.