All pull requests

Pull requests

• k8s: NFD v0.20.1, agent-sandbox v1.0.8, EAS patch 0003 on nri-plugins v0.15.0, kubectl v1.37.2 for the wildct and cross-host rows

  #239·88plug opened Oct 29, 2026Owner·

  ·

• Latest tooling pins (2026-10-29) and crescent amendment E4

  #238·88plug opened Oct 29, 2026Owner·

  ·

• bench: newest releases for the bench arms the ledger lets move; H7/H8 pin their own vLLM image

  #237·88plug opened Oct 29, 2026Owner·

  ·

• engines: vLLM XPU v0.32.0, llama.cpp v0.8.0 (Vulkan, plan --predict, bench --vs), Ollama v0.40.13

  #236·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H7x): amendment V1, arms B and C on EAT release-1.1.0 (vLLM CPU v0.28.0)

  #235·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H60): amendment V1, both arms on llama.cpp v0.8.0 (server-b11960)

  #234·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H58): amendment P1, the model pins

  #233·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H52b): Xeon 6741P run after host-setup --apply; amendment D3

  #232·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H51): REFUTED on the Xeon 6741P; neither CPU stack serves one graded agent within the 10 s SLO

  #231·88plug opened Oct 29, 2026Owner·

  ·

• research/model-cards: ship the upstream license texts with the Hub files

  #230·88plug opened Oct 29, 2026Owner·

  ·

• fit: the four planner fixes together, and where they meet

  #229·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H50o): INCONCLUSIVE, not deliverable; EAT's OVMS 2025.4 chart crashes loading Gemma 4

  #228·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H50): REFUTED on the Xeon 6741P; EAT release-1.1.0's stock CPU arm keeps up

  #227·88plug opened Oct 29, 2026Owner·

  ·

• bench(wildct): resume pack, bounded host calls and a host health check (amendment R1)

  #226·88plug opened Oct 29, 2026Owner·

  ·

• ci: hold.sh stops CMD's group with one TERM under GNU timeout; prefill guard test reads the pinned chart version

  #224·88plug opened Oct 29, 2026Owner·

  ·
  1

• host-setup: current kobuk-team PPA pins, gmmlib for compute plans, nightly PPA check

  #221·88plug opened Oct 29, 2026Owner·

  ·

• nightly: repin the moved digests and keep the digest PR out of .github/workflows/

  #220·88plug opened Oct 29, 2026Owner·

  ·

• docs: record the Qwen3.8-2.4T-A95B-FP8 weights before removing them from crescent

  #219·88plug opened Oct 29, 2026Owner·

  ·

• fit: size llama.cpp's layer split card by card

  #217·88plug opened Oct 29, 2026Owner·

  ·

• fit: count llama-server's host caches

  #216·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H57dp): amendment R2, the bench lock, the hybrid vendor, the quality records and the Reproduce header

  #215·88plug opened Oct 29, 2026Owner·

  ·
  2

• fit: size sliding-window KV per engine

  #214·88plug opened Oct 29, 2026Owner·

  ·

• ledger(H54b, H54c): INCONCLUSIVE, not delivered as pre-registered; H54a corrections; llm_load engine-death fix

  #213·88plug opened Oct 29, 2026Owner·

  ·

• fit: OVMS serves one GPU

  #212·88plug opened Oct 29, 2026Owner·

  ·

• docs: H52b finding, no safe host-setup path for IAA on crescent

  #210·88plug opened Oct 29, 2026Owner·

  ·