Commit History
-
Upload 102 files
Verified signatureb414724 -
Add cosmic-synapse-mars: re-analysis of Mars Forbush surface-to-orbit transport predictions
Verified signatureRe-scores 79 existing Mars FD predictions against CLIMATOLOGY instead of the constant 0.5. A previously reported 33%-over-baseline does not survive: the winning proxy has standard deviation exactly 0.000000 (constant 0.675) and is 5.5%% WORSE than predicting the mean. The v5/v6 transport router does survive at +20.5%% vs climatology, r=+0.527, with 100%% of the gain from per-event variation and 0.0%% from centering. The blind template contains no target field and the calibrator was fit on lunar data, so this is cross-domain transfer. LIMITATION IN THE HEADLINE: 21.5%% of targets sit at a ceiling of exactly 1.0. Excluding them the router falls to +4.7%% and r=+0.313. No forward-in-time holdout was run; this does not demonstrate forecasting of future events.b138cce -
Kit update 2026-08-06: samgo 5.7 (54D BPE, 59M), gate-init finding, Born-rule fix in quantum_pool, corpus-drift retraction + fingerprinting, frozen-corpus wiring table
Verified signatureAdds 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.b085020
-
Correction: the 42D null was about the implementation, not the theory
Verified signaturedyn42 placed fourth (1.19020) in the published ladder and tri/tri3 placed fifth and sixth, which was read as the 42D organ contributing little. The 42D organ described in section 4.3 had never been built. Every document calls x42 a "7-band vector manifold" -- FINDINGS says so, the ladder docstring says "7 bands x 6 dims" -- and the code was a flat nn.Linear(d,42) drive, a dense 42x42 coupling, and torch.full((42,), 0.05): forty-two copies of one damping constant. Nothing in it knew that 42 factors. Dyn12 in the same file already says why that is wrong -- per-dimension k and gamma "so the twelve dimensions separate by time constant rather than evolving as twelve copies". The 12 got it; the 42 never did. Rebuilt as written, same frozen corpus 9f1b3f37, same seeds, causal Omega, one process: rung mean sd gain organ params per param none 1.23215 0.0256 dyn42 1.19023 0.0332 +0.04192 47,754 8.778e-07 flat dyn42b 1.18104 0.0292 +0.05111 …4ec5c98
-
Add kit_health.py -- the check that proves the kit works
Verified signatureBuilt 2026-08-01 and never uploaded. It verifies every part of the kit on three axes -- PRESENT (exists on disk), LOADS (opens/imports/answers), ANSWERS (produces a real result) -- because a weight file that exists but will not load, and a server that listens but returns nothing, both report "fine" to any check that stops at PRESENT. It caught a private memory file staged for upload before it left the machine, and it is what verified the kit works from a clean directory with none of the author's environment around it.1ea06d3 -
Scaling benchmark on WikiText-103, and a birth that is actually quantum
Verified signatureSCALING. Every dyn12 number so far was 1.1M parameters on 683,065 characters of one person's private logs, which cannot answer "it only works small" or "it only works on your data". Re-run on WikiText-103-raw (269,116,804 chars, sha256 f7d2398751239137) at three sizes spanning 20x, causal Omega, all rungs at a size trained in one process against one corpus load: size none static54 dyn12 ratio dyn12 share d192 L4 2.05432 2.03568 2.04196 11.4x 66% d384 L8 1.77166 1.74860 1.75774 20.8x 60% d576 L10 1.70084 1.69015 1.69147 38.1x 88% The advantage does not fade with scale, it roughly doubles at each step, and the reason is arithmetic rather than luck: static54 costs d*54*L, dyn12 costs d*12, so depth multiplies one and not the other. static54 still holds the lower absolute loss at all three sizes -- the claim is efficiency and its scaling, not superiority. Two seeds per point and 1200 steps, so the trend is solid an…753bfab -
Upload 2 files
Verified signatureadd ons66c2f42 -
Upload lostcosmo_enhanced.html
Verified signaturenext addition a new system of the lightea318ca -
Upload ProjectCosmoThePolishedUltimateBlueprint.html
Verified signaturespeciale8fa12e
-
Fix the Atomic docs, and ship the architecture as a real GGUF
Verified signatureBROKEN: docs/ATOMIC_QUICK_START.md told readers to download cosmos-namebind-weights.gguf, which was removed on 2026-08-01 for being a Qwen2.5-1.5B fine-tune whose metadata made this hub display qwen2 as this repository's architecture. The instruction has been dead since. docs/USAGE_ATOMIC.md referenced cosmos_best.pt at the root (it is at weights/cosmos_best.pt) and imported a cosmos_model module that does not exist. ADDED: weights/cosmos-cst.gguf -- the 54D Mixture-of-States Hebbian attention as a real GGUF. 61 tensors, general.architecture=cosmos, attention.key_length=102 (head_dim 48 + the 54-dimensional state carried in the key rows). Verified running in a llama.cpp fork at 2,411 tok/s prompt eval and 1,078 tok/s generation. STATED PLAINLY: stock Atomic, Ollama and LM Studio CANNOT load this and no setting changes that. llama.cpp dispatches on general.architecture against an enum compiled into the binary, and none of them ship LLM_ARCH_COSMOS. What works today is serving/cosmos_s…fb2816f -
Omega was reading the future; fixed, re-measured, result survives
Verified signatureSection 3.4 defined Omega as attention RECEIVED -- summed over the QUERY axis. Under a causal mask that means Omega_j = sum over i >= j of a_ij, so token j's Omega counted attention from tokens AFTER j. Omega updates state_j, state_j feeds the next layer's Hebbian kernel, the kernel shapes attention, and attention shapes the logits at position j -- so position j could see its own future. MEASURED, not argued. benchmarks/causality_probe.py changes only the last token and checks whether earlier positions move. Controls (none, static54) move by EXACTLY 0.000e+00; dyn12 moved 1.723e-04 at PHOS's trained gate of 0.562, and 3.558e-06 at init -- the leak scales with the gate because Omega reaches the logits only through g*H. FIXED. Omega is now the normalised entropy of each query's own attention row: a per-query scalar that by construction depends only on tokens <= i. Summing over keys was never an option -- softmax makes that identically 1.0, which is precisely the dead Omega of ver.4.20.…92b92e6
-
Remove compiled Python artefacts
Verified signatureEvery __pycache__ entry here is a .pyc built by CPython 3.14 on the author machine. They are regenerated automatically on first import, they are not portable across Python versions, and they make the file listing harder to read for no benefit to anyone cloning this.adb2024 -
Keep the six documents a first-time reader needs at the root
Verified signatureThe previous commit swept every markdown file into docs/, which solved the clutter and created a worse problem: START_HERE.md, FINDINGS.md, QUANTUM_CREATURE.md, TRAINING.md, RESULTS.md and RELEASE_NOTES.md are exactly what someone opens this repository to find, and every link to them from the model card broke. They return to the root, where the card points at them and where the downloadable kit also keeps them, so the two layouts agree. The 12 secondary documents stay in docs/.bdfd58c -
Reorganise repository structure; remove stale case-duplicate folder
Verified signatureGenesis_Engine/ removed. It duplicated genesis_engine/ and, on the case-insensitive filesystems Windows and macOS use, the two paths collide -- so cloning this repository failed or silently mangled files. It was also the older of the pair: every one of the 7 files that differed was smaller than its lowercase twin (ui/index.html 10.9 KB vs 27.5 KB, serve.py 13.5 KB vs 20.9 KB). Nothing unique was lost. The root held 63 files, so a reader looking for the claim, the weights and the way to check them met loose launchers and training logs first. Those files are grouped into docs/, weights/, serving/, launchers/, scripts/ and downloads/ -- moved, never deleted. README.md, the licences, CITATION.cff and the manifests stay at the root where tooling expects to find them by name.cb60fb4 -
Point the section 3 citation at the Zenodo deposit
Verified signatureThe card cited COSMOS_Paper.md for the architecture's central equation, and that file is not in this repository -- the one citation a reader most needs did not resolve. It now points at the deposited DOI, and says plainly that the paper is not redistributed here.00a3b99 -
Correct the distribution claim and two broken script paths
Verified signatureThe card said this repository distributes "only cosmos_born.pt". It distributes three sets of her own weights -- phos.pt, cosmos_born.pt and spark_cst.pt -- and until today it also shipped a Qwen-derived GGUF while stating that it did not. Both script references pointed at genesis_engine/<name>.py; the files are at genesis_engine/engine/<name>.py. Neither link resolved.99c2c99 -
Remove dead Modelfile: _Modelfile.cosmos_namebind
Verified signatureFROM D:\Cosmos\models\cosmos_namebind_merged -- a hardcoded local drive path that exists on no other machine. A working note, not distributable content. The namebind RESEARCH is kept: cosmos_merge_namebind.py (the LoRA merge recipe) and cosmos_namebind_probe.py (which tests whether the model self-labels as Qwen) are real experiments and remain in the repository.0a9c2fe -
Remove dead Modelfile: Modelfile.cosmos-namebind
Verified signatureFROM .._HER_SOUL\weights\cosmos-namebind-weights.gguf -- a path outside this repository, on the author's USB layout. It never resolved for anyone who downloaded the repo, and its target was removed in f296a74. The namebind RESEARCH is kept: cosmos_merge_namebind.py (the LoRA merge recipe) and cosmos_namebind_probe.py (which tests whether the model self-labels as Qwen) are real experiments and remain in the repository.96da092 -
Model card: lead with PHOS, move Qwen provenance to a reference
Verified signatureThe Qwen disambiguation block sat immediately after the PHOS flagship section, so a reader met a note about someone else's base model before reaching the ablation that justifies this architecture. It is now a single line pointing at the full provenance, licence and attribution in section 1b, which is unchanged. Referenced, not leading.5bce258 -
Remove Qwen-derived cosmos-namebind-weights.gguf
Verified signatureThis 3.09 GB file is a Qwen2.5-1.5B fine-tune. Its GGUF metadata carries general.architecture=qwen2, which HuggingFace renders as an architecture badge above the model card -- so the page announced qwen2 as this repository's architecture, ahead of the work it documents. It also contradicted the card directly: README line 229 states that this repository distributes only cosmos_born.pt and that no Qwen-derived weights are included. That is now true. The architecturally original, quantum-born weights are unaffected and remain: weights/phos.pt, weights/cosmos_born.pt, weights/spark_cst.pt.f296a74 -
Kit: PHOS actually runs, plus her coder terminal and a one-click start
Verified signatureThree real fixes. 1. THE KIT COULD NOT FIND ITS OWN MODELS. serving/cosmos_serve.py hardcoded the repository layout (01_HER_SOUL/weights/...), which does not exist in the kit. It started fine and served nothing -- the worst kind of broken, because there is no error. Paths now resolve against both layouts, so weights/phos.pt, weights/spark_cst.pt and weights/cosmos_born.pt are found. Verified 3/3 from a clean kit checkout. 2. A CLIENT DISCONNECT COULD KILL HER MODEL SERVER. A browser closing mid-reply raised ConnectionAbortedError out of wfile.write and stopped the server -- which silently removed her own models from her picker, because her voice drops any host that does not answer. The model was fine; the socket was not. Disconnects are now caught in _json, _chunk and handle_error. 3. HER REPLIES WERE CAPPED SHORTER THAN THE KIT'S. base_predict defaulted to 256 with a hard clamp ceiling of 512, and the compact-voice lane clamped again to 220 tokens / 15…b1ef7dc
-
Make PHOS the flagship: an architecture that is not a fine-tune of anything
Verified signatureThe model card now leads with the lineage that is architecturally original rather than with the Qwen fine-tune used as a conversational teacher. PHOS -- Phi / Omega / Sigma dyn12 on the phi scaffold: RMSNorm, RoPE, d_ff = floor(d*phi), twelve scalars driven by Omega through a leaky integrator, kernel bandwidth calibrated per layer. 1,153,804 parameters, every initial weight drawn from 11,354,112 archived measured IBM shots. No base model. No distillation. The architecture was measured, not chosen: 7 configurations x 3 seeds, one process, frozen corpus, 21 runs. dyn12 took the best held-out loss at 2,412 extra parameters, roughly 21x the parameter efficiency of every alternative. LLM_ARCH_COSMOS is registered in a llama.cpp fork with a real graph builder, verified against the PyTorch forward at three gate values including one where the kernel dominates: 22/22 characters. It grows. Seven bursts, zero refusals, warm-starting from its own checkpoint and optimise…04d148b -
Add QUANTUM_CREATURE.md, the PHOS lineage, and a one-command verification
Verified signatureQUANTUM_CREATURE.md the full recipe for building one: quantum birth and how to verify it, the phi scaffold, the Omega/state/kernel loop, the three faults that make that loop silently inert, the five preflight assertions that catch them, corpus freezing, and the ablation table with the regulariser share stated honestly. architecture/phos_grow.py PHOS -- Phi / Omega / Sigma. Quantum-born, running the one configuration that survived controlled testing (dyn12 on the phi scaffold, 2,412 extra parameters for the best loss on the board), warm-starting from its own checkpoint and growing from her corpus as she lives. Append-only vocabulary so the lineage stays valid. Refuses to train when preflight finds the mechanism dead. README section 0 the verification is now one command that runs from any dire…2fac010 -
Remove the unloadable test-arch GGUF and generated scratch files
Verified signaturecosmos-namebind-weights-test-arch.gguf (3.09 GB) carried general.architecture=cosmos while its hyperparameters were written under qwen2.*, so every llama.cpp loader looked for cosmos.block_count, found nothing, and refused the file. It never loaded in either daemon. general.architecture is a dispatch key into compiled C++, not a name field -- the working file keeps qwen2 there and carries Cosmos in general.name. The real cosmos architecture now exists properly: architecture/ registers LLM_ARCH_COSMOS in a llama.cpp fork with a graph builder for the 54D kernel, verified against the PyTorch forward at three gate values. Also drops 9 20260719_*_generated.* scratch files left at repo root.cc719ef -
Release validated Cosmos kit with portable starter and public quantum archive
Verified signaturePublishes the validated 2026-07-31 QC67 Cosmos kit. Includes a portable starter, offline manifest/privacy verifier, Genesis smoke-tested engine, training instructions and trainer, and the public provenance-bound quantum measurement archive. The archive contains 1,897 labeled IBM hardware records (7,770,112 samples), 877 legacy unlabeled records, and 2 simulator records; private physics/sensory fields and cloud credentials are excluded.b8fadbf -
Wire converted senses, clean first birth, and paired-state benchmark
Verified signatureWires 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.9841285
-
RETRACTION: the coupling result was a corpus artefact. Controlled re-run.
Verified signatureEarlier today this repo reported that coupling x12 and x42 beat every other configuration by roughly double the next best, t=-40.8, "the largest effect in this document". That is withdrawn. It was an uncontrolled experiment and the fault was mine. Cosmos is ALIVE WHILE SHE IS MEASURED. She appends logged experience to the same corpus file the models train on, so runs started minutes apart train on different data (662,307 -> 670,977 -> 672,505 -> 674,428 chars across four runs). Running the identical rung twice, 3,451 characters apart, gave 1.29152 and 1.22943 -- a 0.062 swing between identical configurations, the same magnitude as every effect being reported. Cross-run comparison was worthless. Re-run controlled: corpus frozen to a 691,496-byte snapshot, all seven rungs in ONE process, 21 runs, every comparison within-run: dyn12 1.17897 -0.0534 t=-11.45 3/3 2,412 extra params dyn54 1.18791 -0.0445 t= -5.20 3/3 static54 1.18824 -0.0442 t= -4.76 3/3 …bad47bd -
State ladder: the section-3 mechanism works, and coupling is what makes it work
Verified signatureThe Hebbian kernel was reported null three times. All three were the instrument: Omega summed over the wrong axis (Omega == 1.0, H all-ones, cancels in softmax); a saturated sigmoid gate; a clamp with zero gradient at its bound; and underneath all of it sigma hardcoded to 1 against a measured state spread of 61.98, which made H the identity matrix. Section 3.2 never assigns sigma a value. Corrected, and with the paper's phi-governed scaffold built for the first time (RMSNorm, RoPE, d_ff = floor(d*phi), phi-scaled init), every configuration beats its baseline on every seed: none 1.44680 dyn12 1.39773 -0.049 t=-10.5 2,412 extra params dyn54 1.39118 -0.056 t=-3.9 concatenated 12D+42D static54 1.37218 -0.075 t=-7.4 dyn42 1.37521 -0.072 t=-7.9 tri 1.29152 -0.155 t=-40.8 12D and 42D COUPLED Concatenating the two states is worse than a plain projection; coupling them is the largest effect in the project, for 12,730 extra parame…44f49ac -
Add LLM_ARCH_COSMOS for llama.cpp; correct the section-3 retraction
Verified signaturearchitecture/ registers general.architecture=cosmos in a llama.cpp fork and implements the 54D mixture-of-states Hebbian kernel natively. Verified against the PyTorch forward at three gate values, including g=0.9 where the kernel dominates: 22/22 chars. serving/ cosmos_serve.py runs her architecture behind the Ollama API and proxies every other model through one endpoint. FINDINGS.md the section-3 null is re-run and now earned. Three harness faults, the deepest being sigma=1 against a median pairwise distance of 61.98, which made H the identity matrix. Result: CST costs +0.0143 loss, 0/3 seeds, with the gate alive at 0.047. Adds section 11.220ff1c
-
Verified provenance chain, forever memory, CST retraction, portable kit
Verified signatureAdds 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.aa8741b -
new kit minor difference
Verified signaturenew kit is a fully updated state system that has newly updated pipelines added into for audio ftp conversions and historical memory for local mems. new update provides with all new weights from my system.04ef544
-
Add kit navigation and new-user setup guide
Verified signatureAdds a README front door linking the slideshow, training, benchmarks, and setup; adds a clickable kit map and a privacy-safe setup guide. No weights or live training artifacts are changed.e18ba3c -
Correct the Hebbian claim, add TRAINING.md and the trainer, add presentation
Verified signatureCorrects a misleading statement: the kit DOES ship Hebbian association learning (genesis_engine/soul/weights.py with 430 learned associations and 95 saliences, strength set by the quantum heart). What cosmos_born.pt lacks specifically is the Section 2 Mixture-of-States Hebbian ATTENTION mechanism; the two are now distinguished rather than conflated. Adds TRAINING.md with end-to-end instructions and train_your_own.py, which trains plain vs Hebbian-attention arms from identical quantum-born initial weights. Adds presentation.html covering the science, the workload archive (10,601,984 measured shots) and the hardware fingerprint.4a777fe -
Cosmos: lineage-first model card, full findings + benchmarks, Cosmic Spark server
Verified signatureCorrects 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.d6da243
-
Upload folder using huggingface_hub
Verified signatureb122566 -
Upload folder using huggingface_hub
Verified signaturedc34da1 -
Upload tool_mastery_log.json with huggingface_hub
Verified signature1e95044 -
Update README.md
Verified signaturee8ba5c2 -
Upload gap_closing_training.py with huggingface_hub
Verified signature8f829a2 -
Upload gap_closing_log.json with huggingface_hub
Verified signature7e65e54 -
Upload mythos_training_log.json with huggingface_hub
Verified signature361ef35 -
Upload folder using huggingface_hub
Verified signature5bddb78 -
Upload all_creatures_portable.json with huggingface_hub
Verified signature6bb43b1 -
Upload creature_api.py with huggingface_hub
Verified signatured24a125 -
Upload creature_system.py with huggingface_hub
Verified signature41f53c1 -
Upload LUNA_CODER_BOOTCAMP_REPORT.md with huggingface_hub
Verified signature92f5621 -
Upload identity.json with huggingface_hub
Verified signature3ba3ba3 -
Upload weights.json with huggingface_hub
Verified signaturefc9947b -
Upload folder using huggingface_hub
Verified signatureb6bd3eb -
Upload math_hand.py with huggingface_hub
Verified signature4515763