club-3090

Files

noonghunna 101077fdf2 docs: cross-link Cliff 3 ↔ LMCache — offload IS the mitigation, not a victim

Cliff 3 (CLIFFS.md) walked readers to "use llama.cpp, vLLM is hopeless
for agentic Qwen3-Next" with zero mention that LMCache is the measured
mitigation for exactly that cliff (#423: 40K prefix warm-loads 0.5-2.4s
L1 vs 34-43s cold). Native prefix cache / generic CPU-offload hit Cliff
3; LMCache's HMA hybrid recurrent-state caching is the differentiator.
Add a bounded mitigation subsection (measured for resumed prefixes,
flagged untested for growing loops), revise the recommendation to two
paths, and forward-link INTERNALS' LMCache section to Cliff 3.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EfF565T9eSLaqGzidyJ1Pm

2026-07-12 20:11:27 +00:00

agents-a1/vllm

fix(launch): runtime-agnostic GPU pinning via UUIDs — CDI/NixOS support (#610 )

2026-07-06 22:59:51 +00:00

diffusiongemma-26b-a4b/vllm

fix(launch): runtime-agnostic GPU pinning via UUIDs — CDI/NixOS support (#610 )

2026-07-06 22:59:51 +00:00

gemma-4-12b

Bump llama.cpp pin to server-cuda-b9967 across all composes (+4 think-ON) (#680 )

2026-07-12 01:00:29 +05:00

gemma-4-26b-a4b

fix(launch): runtime-agnostic GPU pinning via UUIDs — CDI/NixOS support (#610 )