All pull requests

Pull requests

• [WIP]Add compressor stream functionality

  deepseek
  #1349·AndyLi429 opened Oct 20, 2026·5 tasks·

  ·

• Update glm5_1 testcases (in16k_3p1d_50ms, in64k_1p1d_20ms interleave prefill CP)

  npu
  #1348·pllimax opened Oct 20, 2026·5 tasks·

  ·

• Added seven test cases for MiMo-V2.5-Pro

  npu
  #1347·HeYao0714 opened Oct 20, 2026·5 tasks·

  ·

• Added three test cases with test parameters

  npu
  #1346·HeYao0714 opened Oct 20, 2026·5 tasks·

  ·

• Add eight test cases for Qwen3.5

  npu
  #1345·HeYao0714 opened Oct 20, 2026·5 tasks·

  ·

• [NPU][HiCache] Support DeepSeek V4 L3 with MemCache on main

  documentationhicachenpuunified-radix-cache
  #1344·cocoa-liu opened Oct 19, 2026·4 of 5 tasks·

  ·

• Add the --load-publish-endpoint parameter for testing.

  npu
  #1343·chenyang08056032 opened Oct 19, 2026·5 tasks·

  ·

• Add --fuseep-mode parameter testcase

  npu
  #1342·YZY00Raiser opened Oct 19, 2026·5 tasks done·

  ·

• Add Qwen3-VL 8B/30B accuracy cases to full-acc nightly suites

  npu
  #1340·EdwardXuy opened Oct 18, 2026·2 tasks·

  ·

• [NPU][CI] Sync patch-14 NPU test work onto testcases-261015

  apple-silicondeepseekhicacheloraMulti-modalnpuquantspeculative-decoding
  #1339·Sugar920 opened Oct 17, 2026·

  ·

• [NPU] DSV4: fix C128 page leak and C128 capacity/admission

  unified-radix-cache
  #1337·unclezhou486 opened Oct 16, 2026·5 tasks·

  ·

• [NPU] fix DSV4 C128 KV pool exhaustion and its blind diagnostics

  documentationunified-radix-cache
  #1336·unclezhou486 opened Oct 16, 2026·5 tasks·

  ·

• Add Qwen3-VL 8B/30B accuracy cases (mmlu/mmmu/aime26) with single-test wiring

  npu
  #1335·EdwardXuy opened Oct 16, 2026·2 tasks·

  ·

• Add NPU tests for mm worker pool, TP all-to-all LM head, sidecar, offload group size and MF PD router reconfig

  npurun-ci
  #1333·hhhh1252023 opened Oct 15, 2026·5 tasks·

  ·

• [NPU] Backport PR #36426 to fix ModelSlim weight mappings and Qwen vision loading on the release branch

  npu
  #1330·fishAndShrimp opened Oct 13, 2026·

  ·

• [NPU][CI] Sync 8 bf16 accuracy cases from ascend testcases_2, add debug-full-acc suites

  deepseekhicacheloraMulti-modalnpuquantspeculative-decoding
  #1329·Sugar920 opened Oct 13, 2026·

  ·

• A/B verify test_npu_vlm_input_format.py: single-test vs full-2-only full-test on B100

  npu
  #1327·EdwardXuy opened Oct 13, 2026·

  ·
  4

• [NPU] Add the --file-storage-path and --load-publish-endpoint parameter for…

  hicachenpu
  #1325·chenyang08056032 opened Oct 13, 2026·5 tasks·

  ·

• Fix hellaswag preload IndentationError causing full-test 404 (rowanz DMCA)

  npu
  #1321·EdwardXuy opened Oct 10, 2026·

  ·

• test(npu): verify reduced NPU resource allocation

  npu
  #1316·Catherinecttt opened Oct 10, 2026·5 tasks·

  ·

• [NPU][CI] Add debug full-test jobs for 397B rerun

  deepseekhicacheloraMulti-modalnpuquant
  #1315·Sugar920 opened Oct 9, 2026·

  ·

• Add Parameter

  npu
  #1314·HeYao0714 opened Oct 9, 2026·5 tasks·

  ·
  4

• [release/2026930][Backport #40516] [NPU] Fix Qwen shared-expert routing and FuseEP accuracy checks

  #1309·McZyWu opened Oct 9, 2026·

  ·

• [release/2026930][Backport #40705] [NPU] Fix MoE none EP routing and draft backend inheritance

  #1308·McZyWu opened Oct 9, 2026·

  ·

• rerun test_npu_vlm_input_format.py

  npu
  #1303·YZY00Raiser opened Oct 9, 2026·5 tasks·

  ·