deterministic-evaluation

FreeCite: A Judge-Free Benchmark for Granular Citation Evaluation in Large Language Models

Research benchmark for evidence-grounded OS-agent collaboration, continuous state diagnosis, scoped memory reuse, and stale-state rejection.

Canonical specification and sealed execution traces for VERITAS Ω-CODE v2.0 — deterministic software verification protocol.

As of 07/09/2026 the only open-source PT-PT benchmark that avoids LLM judges and targets email-style structured generation. Deterministic CPU-friendly benchmark for European Portuguese (PT-PT) email LLMs. No LLM judges.

AI-assisted policy formalization and governance platform for source-traceable rule extraction, human review, immutable versioning, deterministic evaluation, quality analysis, and regression testing.

Public, fully local PoCs for counterfactually auditable lifecycle certification: exact paired replay, drift monitoring, post-drift replanning, and bridge-aware ledger control on synthetic tasks.

HCAC is a completion contract for agents and tools. It defines when something is truly “done” — not just structurally, but behaviorally.

Hybrid legal RAG and LLM extraction with JSON Schema rule compilation, three-valued logic, Census/TIGER geocoding, and deterministic evaluation.

Ultra pipeline framework for Gravity Binary, including deterministic evaluation logic, capsule workflows, and EC/Ultra integration artifacts.

Deterministic multi-lane evaluation domain for SnorkelAI Terminus3. Cockpit, lane orchestration, agent benchmarking, reasoning-drift detection.

SnorkelAI Terminus2 task suite with deterministic evaluators, Dockerized environments, and reproducible agent-benchmarking primitives.

Core deterministic architecture for multi-lane ML evaluation systems. Foundational logic, capsule interfaces, reproducible execution patterns.

Deterministic offline ComtradeBench judge for evaluating agent robustness under pagination, retries, duplicates, page drift, and totals traps.