← dashboard📚rag-bench sweeps
0sweeps

Benchmarks

Retrieval-augmented Q&A sweeps from bin/rag-bench. Each row is one model run across a question set, with hit/miss recorded per question. Click a row for per-Q detail; click any tag-comparison card to drill into an ablation.

qset
model
0 sweeps·0/0 hits·0 tok in·0s wall

Sweeps

0

Newest first by default. Click any column header to sort. Click a row for the per-question hit/miss table and (for gridown-50) per-domain accuracy.

when model qset tag acc n wall μ tokens in/out tools
no sweeps match the current filter