Technical Whitepaper
Explore the architectures behind our work in inference, tokenization, continual learning, and governed tool use, with methods, evaluation methods, and technical scope.
ORBIT — Orthogonal Recurrence with Delta-Rule Memory
Problem. Attention's cost grows with context length, and most linear-recurrence alternatives trade away exactness for speed — the fast "parallel scan" form of the recurrence doesn't always match what a step-by-step (sequential) computation would have produced.
Approach. ORBIT uses an orthogonal (Cayley/Givens) recurrence combined with a gated delta-rule episodic memory bank, keeping state size constant per step (O(1)) regardless of sequence length.
Technical scope. ORBIT studies constant-state recurrence and episodic memory. Evaluate numerical equivalence and language-model quality independently using a fixed source revision and dataset.
// ORBIT evaluation protocol
Compare parallel and sequential recurrence.
Record numerical error, task quality,
model size, dataset, and source revision.
// UTL-C evaluation protocol
Test byte-exact round trips and Unicode
grapheme boundaries. Compare compression
on identical corpora and settings.
UTL-C — Byte-Exact Tokenization
Problem. Most production tokenizers are not byte-exact and are rarely benchmarked against the Unicode text-segmentation standard directly — grapheme-boundary errors quietly corrupt multilingual and emoji-heavy text.
Approach. A tokenizer written in C with a guarantee that decoding an encoding always reproduces the original bytes exactly, benchmarked directly against the Unicode UAX #29 conformance suite rather than an internal proxy metric.
Technical scope. UTL-C focuses on byte-exact tokenization, grapheme boundaries, and structured-token handling. Evaluate compression separately from Unicode conformance.
Cognitive Crystal — Continual Learning Without Catastrophic Forgetting
Problem. A model trained on new data typically overwrites what it knew before — "catastrophic forgetting." The usual fix is storing large numbers of raw past examples to replay, which costs memory that scales with how much you want it to remember.
Approach. Instead of replaying raw examples, distill what was learned into a small number of compact prototypes and replay those.
Technical scope. Cognitive Crystal evaluates prototype replay for continual learning on Split-MNIST. Compare forgetting, retained accuracy, memory use, and variation across repeated runs within that workload.
// Continual-learning evaluation protocol
Measure forgetting and retained accuracy
on Split-MNIST across multiple seeds.
Record prototype and replay memory separately.
SEVK-C — Treating an LLM as an Untrusted Semantic Sensor
Problem. An LLM's output is usually trusted directly as the basis for the next action an agent takes — a tool call, a decision, a write. That's the same failure mode as trusting raw sensor data without validation.
Approach. Route model output through typed evidence contracts and validators before it can influence anything, and require a "tool-proof obligation" to discharge before any tool call is allowed to fire. Technical scope: a working, exercised prototype (real session logs exist) — not yet benchmarked, so we're not publishing an accuracy number for it. What we can say is that the mechanism — typed contracts gating tool calls — is implemented and running, not just designed.
Earlier-stage and adjacent work
Nova-X CSLM
A non-Transformer architecture using energy minimization and wave-interference scoring in place of attention. The patent-pending NOVA state-space design is described separately under Patents & IP. The architecture is implemented in a working engine; we are not yet publishing latency or model-size numbers for it while our internal benchmark methodology is being validated.
Architecture Bench
A set of smaller internal model variants exploring different efficiency tradeoffs. This is source-and-design-stage work — no benchmarks are published for it yet, so there's nothing to report here beyond the fact that it's ongoing.
This page will be updated as each project's benchmark methodology and results mature — we'd rather publish four honestly-scoped results than fourteen unverified ones.
DH-ADOM
A hierarchical operating model for agent-led engineering: clear domain ownership, bounded delegation and evidence-backed handoffs.
Explore the paper, diagrams and source →