Adds SAMGO.md and weights/samgo_weights.pt (59,353,668 params, GPT-2 BPE).
FINDINGS.md: gate initialised at sigmoid(-4)=0.018 left the Hebbian kernel alive in
1 layer of 4 in every model ever trained here; opening it to 0.5 gives 4/4 and 0.054
nats, larger than the dyn12-vs-none gap. quantum_pool was flattening the measured
Born-rule distribution to exactly uniform (3.4657 = log 32) and now preserves it
(3.4598 vs true 3.4594). Retracts a 0.0645-nat wiring win that was the live corpus
rewriting itself mid-experiment; adds corpus+flag fingerprinting to prevent recurrence.
Wire converted senses, clean first birth, and paired-state benchmark
Wires opt-in browser-derived camera luminance/motion and microphone energy/spectrum into Genesis reply conditioning, associative recall, Hebbian learning, status, and the conversation ledger while keeping raw media in-browser. Adds the missing drive-relative START_GENESIS.bat and resets the public template to a true first birth with empty keys. Adds durable paired telemetry tooling, timestamp join, aligned/shuffled/shifted benchmark, and a privacy-audited machine-readable result. Documents the unpaired schedule null and the timestamp-paired NULL result: aligned beat shuffled but lost decisively to plain. Private turn text and the paired dataset are intentionally excluded.
Adds the falsifiable verification chain (CHSH S=2.7905; weight-birth matching the 32-level quantisation ceiling to 4 decimals; seed->weights determinism at delta=0.000e+00 with a 1-bit flip giving 2.26e-02; gate-matched topology at t=+80.5/+164.6; Lorenz lambda_1=0.90384 vs published 0.9056).
Adds genesis_engine/memory: working long-term recall by meaning, with the three-way failure it fixes documented, and measured proof that a generative model cannot be used as a retrieval embedder.
RETRACTS the seventh null (Section 3 CST kernel). The test harness prevented the mechanism from being learned: gate initialised in the saturated tail of a sigmoid, gradient suppressed ~56x, and zero gradient reaching the 54D projection. Re-running. No claim is made in either direction until it completes.
Fixes stale weight metadata (card described a 5,125-step model while shipping 9,425-step weights) and two author-specific paths that broke the kit on any other machine.
Cosmos: lineage-first model card, full findings + benchmarks, Cosmic Spark server
Corrects earlier claims (not GGUF; 1.8M params not 2B; architecture is a standard pre-LN decoder, not Hebbian). Adds FINDINGS.md with every measurement including six null results and the instrument failures behind them. Adds benchmarks/ with runnable reproduction scripts for CHSH (S=2.7905), Lorenz constants, gate-matched topology, and integration structure. Adds spark_serve.py to run the weights over an Ollama-compatible API.
