All pull requests

Pull requests

• [tools][runtime] Align buddy-server Interface with llama-server

  runtime
  #994·enanandesu opened Oct 17, 2026·5 tasks done·

  ·
  1

• feat: add tile.upsample_nearest op and recognition pattern

  midend
  #993·cwrt310 opened Oct 17, 2026·1 of 5 tasks·

  ·

• Fix bn channel broadcast

  frontend
  #992·cwrt310 opened Oct 16, 2026·5 tasks·

  ·

• [agent] upstream-pr: test every torch version CI tests

  ci
  #972·zhanghb97 opened Oct 8, 2026Member·4 of 5 tasks·

  ·

• Use multiple accumulators when lowering matmul

  midendVIR
  #970·xbwcan opened Oct 8, 2026Contributor·5 tasks·

  ·

• [tools] Support FP32/FP16 packed decode with tiered KV cache

  enhancementexample
  #968·GuoningHuang opened Oct 6, 2026Collaborator·

  ·

• [Frontend] Per-token dynamic activation quantization for w8a8 MatMul

  frontend
  #945·Embers-of-the-Fire opened Sep 29, 2026·

  ·
  1

• Feat/fpga boscame add 1x1024

  AMEenhancementexamplemidend
  #944·caihanyi111 opened Sep 28, 2026Contributor·5 tasks done·

  ·

• Optimize Qwen3-VL K3 OCR with FP16 RVV, KV decode, and packed GEMV.

  enhancementmidend
  #935·caihanyi111 opened Sep 22, 2026Contributor·3 of 5 tasks·

  ·
  1

• [Model] BGE-M3: RVV pass chain + A100 hybrid offload on K3 (#888)

  enhancementexample
  #934·virtleo opened Sep 21, 2026·5 tasks done·

  ·
  1

  4

• [Gemmini] Use full scratchpad/accumulator budget for OS tiled matmul

  midend
  #933·KyungTae-Kim97 opened Sep 20, 2026·

  ·

• [Gemmini] Fix row padding of C move-out in OS tiled matmul

  midend
  #932·KyungTae-Kim97 opened Sep 20, 2026·

  ·

• Fix DeepSeek R1 decode session state reuse

  runtime
  #931·GuoningHuang opened Sep 20, 2026Collaborator·

  ·

• [#911] Measure PyTorch coverage and validate the 106-operator CPU profile

  frontend
  #928·FranklinNexus opened Sep 16, 2026·5 tasks done·

  ·

• XTAME: update xtame path design

  AMEmidend
  #927·Yranger opened Sep 16, 2026Contributor·

  ·

• [BOSCAME] Add FPGA backend support for Qwen3 W8A8.

  AMEexample
  #924·CBalaa opened Sep 15, 2026Contributor·

  ·

• [RAX] Add distributed execution and DeepSeek TP=2 integration

  exampleruntime
  #923·asdf1113 opened Sep 9, 2026Contributor·

  ·
  5

• [frontend] Preserve template output order

  frontend
  #878·asdf1113 opened Aug 27, 2026Contributor·

  ·
  2

• [Models] Add MMS-TTS audio model integration - VitsModel: 6 layers, h…

  enhancementexampleruntime
  #868·LibbyStarr10 opened Aug 9, 2026·

  ·
  3

• [Models] Add SmolVLM2-500M-Instruct VLM integration

  enhancementexampleruntime
  #867·LibbyStarr10 opened Aug 9, 2026·

  ·
  3

• [Models] Add colbertv2.0 retrieval model integration - BERT-based lat…

  enhancementexampleruntime
  #866·LibbyStarr10 opened Aug 9, 2026·

  ·
  3

• [Models] Add e5-mistral-7b-instruct embedding model - MistralModel: 3…

  enhancementexampleruntime
  #865·LibbyStarr10 opened Aug 9, 2026·

  ·
  3

• Models/Add SAM2-hiera-tiny

  enhancementexampleruntime
  #864·LibbyStarr10 opened Aug 9, 2026·

  ·
  3

• [Models] Add ChemBERTa-77M-MLM molecular model integration

  enhancementexampleruntime
  #863·LibbyStarr10 opened Aug 9, 2026·

  ·
  5

• [Models] Add MoLFormer-XL molecular transformer integration

  enhancementexampleruntime
  #862·LibbyStarr10 opened Aug 9, 2026·

  ·
  3