Synthetic TypeSafe offload benchmark: quality, tokens, latency, dated evidence, and a portable Codex skill.
Evaluation & benchmarks — page 14
Evaluation harnesses, benchmarks and graders. Jev scores or chooses between candidate outputs with an explicit rubric.
320 active projects match this task, ordered by traction.
Benchmark-style hallucination scoring for Claude Code sessions: atomic claims, fresh-context grader, SimpleQA / AA-Omniscience scoring, per-session tracker.
Fluency First digital workbook MVP: student practice, progress tracking, weakness analytics, teacher dashboard, and leaderboard.
AI Eval Platform — an evaluation and quality-engineering layer for LLM, RAG and agentic systems, with production observability integrations.
Model-judged permission gate for OMP: classifies bash commands and spawn-bearing eval payloads before they run, prompts on risk, fails closed
Minimal local-LLM server benchmark.
No project description available.
0
Classify, route, and rank text in Python with self-hosted models. Typed decisions and evaluation.