Current work

Eight active questions, not fifty-two competing flagships.

TS is a verifier-first research programme exploring explicit semantic state, constraint graphs, tension signals, typed checks, adaptive compute, and deterministic state transitions.

Live

TS-Reasoner

A verifier-first operation firewall where models may propose candidates, but typed checks decide what is accepted, repaired, remembered, or rejected.

typed verificationoperation firewallreceipts
Explore TS-Reasoner
Active

TS Chat Language

A deterministic language compiler from text into dialogue acts, semantic frames, a MeaningGraph, graph-diff memory, response plans, and rendered language.

meaning graphdialogue compilerdeterministic language
Explore TS Chat Language
Active

Ten-SON-LM

A recurrent language architecture with a fixed-size semantic workspace where learned tension controls revision, sparse routing, relaxation, and readout.

semantic workspacerecurrent modelcausal tension
Explore Ten-SON-LM
Active

TensionForge

A verifier-first OpenCL training runtime for legacy AMD hardware, built around exact parity receipts before performance claims.

OpenCLlegacy AMDtraining runtime
Explore TensionForge
Experimental

TSQ

Tension-Structured Quantization: a verifier-gated runtime that keeps ordinary inference cheap and escalates precision or repair when tension justifies the cost.

adaptive precisionquantizationcompute receipts
Explore TSQ
Active

bogbin / BogOS

A deterministic verified storage and computing substrate spanning exact containers, packages, processes, filesystems, persistent workspaces, and append-only journals.

verified storageworkspace journaldeterministic runtime
Explore bogbin / BogOS
Active

TS-Benchmarks

An audit-first falsification harness for graph relaxation, hard reasoning, matched model comparisons, and compute experiments.

falsificationbaselinesreplay
Explore TS-Benchmarks
Experimental

TensionLM

The earlier model-science line testing sigmoid pairwise tension attention, telemetry, curriculum behaviour, and bounded formal-domain signals.

sigmoid attentionmodel telemetryprevious generation
Explore TensionLM

How the stack fits together

The projects are modular on purpose. Language, verification, learned mechanisms, and deterministic computing should be testable without collapsing into one giant repository.

Human interface

ts-chat-language compiles text into explicit semantic frames, MeaningGraphs, state diffs, and response plans.

Authority boundary

TS-Reasoner verifies typed support, contains failed candidates, repairs bounded errors, and controls memory writes.

Learned mechanisms

Ten-SON-LM and TensionLM test tension inside models; TSQ tests whether tension can allocate numerical compute.

Deterministic substrate

bogbin verifies state and persistence; TensionForge verifies custom training execution on legacy hardware.

Supporting nodes

These contain useful infrastructure or bounded experiments. They support the active programmes rather than starting another public storyline.

Active

TS-Core

A compact graph/tension kernel for nodes, weighted relations, activation, propagation, relaxation, contradiction handling, and structural revision.

graph runtimetensionshared kernel
Explore TS-Core
In Progress

CIG

A compact deterministic claim/evidence graph engine for provenance, contradiction, confidence, propagation, derivatives, and structural revision.

provenanceclaim graphcontradictions
Explore CIG
Published

Proof Ranker

A bounded model ladder for proof existence, path quality, local failure detection, repair, and generator/ranker/verifier loops.

proof rankingrepairlearned support
Explore Proof Ranker

Where did the rest go?

Into the lineage, not into oblivion.

Earlier GOAT, AI, OS, CIG, wave, visualisation, model, and convergence repositories are recorded by idea, outcome, and successor before their GitHub sources are removed.

Open the full lineage