Vibgrate Graph: over 24% fewer tokens, same results Your AI re-reads your whole repo every question. Vibgrate Graph maps it once — over 24% fewer tokens, same results. Free — see the benchmark

Rlhf

1 item tagged with "rlhf"

Filter by type:

Benchmarks1

Benchmark

RewardBench

A benchmark for reward models and LLM judges that measures how well they prefer better responses over worse ones across chat, reasoning, safety, and refusal cases.