spb/localvm-research Public License

Running LLMs larger than memory on a consumer Mac — falsification-driven research: margin-gated deferred refinement, out-of-core verification on Apple Silicon. TR-01 published.

Python 63.2% JavaScript 23.5% CSS 11.8% Shell 0.9% Makefile 0.5%

Name	Last commit	Updated
…
implementation	candidate_01 scale: StreamingVerifier (per-layer...	4 h ago
analysis.md	candidate_01: W=64 lever NEGATIVE — rollback rate bounds sweep...	4 h ago
benchmark_scale.py	candidate_01 scale: StreamingVerifier (per-layer...	4 h ago
benchmark.py	candidate_01 v2: incremental q8 verify-cache (O(window) sweeps) + bf16...	4 h ago
cand01_full.log	expG@8B: favorable scaling; candidate_01 first run: bytes pass, metric...	4 h ago
cand01_v2_full.log	candidate_01 v2 results: q8-equivalent quality with q4 resident (bf16...	4 h ago
hypothesis.md	expG@8B: favorable scaling; candidate_01 first run: bytes pass, metric...	4 h ago
README.md	candidate_01: margin-gated deferred-refinement runtime + benchmark +...	5 h ago
scale32b_full.log	candidate_01 32B scale run: charter 16.A/B demonstrated in prototype...	4 h ago
scale32b_w64.log	candidate_01: W=64 lever NEGATIVE — rollback rate bounds sweep...	4 h ago