4 papers
Memory-Sovereign Inference: Output-Exact Execution Beyond Full Residency
Lukas Stepanek
Storage-backed inference is easy to overclaim: process RSS excludes charged page cache, process-local device readings exclude board-wide use, and successful generation does not est…
Route-Block Membership Selects Packed-AWQ Arithmetic: A Controlled Single-Fixture Mechanism Study
Lukas Stepanek
Mixture-of-experts (MoE) inference first aligns routed tokens into padded expert blocks, then executes packed quantized matrix multiplication over those blocks. This preprocessing…
Fail-Closed Lowering of Resident KV Claims onto LLM Serving Runtimes
Lukas Stepanek
LLM serving runtimes increasingly expose KV-cache primitives that resemble future-reuse controls: retention priority, TTL-like duration, host or storage offload, block events, acti…
Resident KV Claims: A Conformance Contract for Future Reuse under Active KV Pressure
Lukas Stepanek
KV-cache reuse mechanisms increasingly expose priority, duration, offload, routing hints, scheduler modes, and event streams. These mechanisms help preserve reusable prefixes, but…