Multi-image benchmarks + multimodal docs (5/5) - #324

Open

jason718 wants to merge 7 commits into

mainfrom

jasonr/vision-split-5-multiimage-docs

Commits on Aug 25, 2026

Add multimodal inference: image requests, HF VLM path, OLMo-core VLM provider

jason718
and
Ziqi Gao (Roy)
committed
92def74
Address review: 8B/O-7B presets, simpler timeout override
jason718
committed
84550ed

Commits on Aug 28, 2026

Address review: shim idempotence, logprob boundaries, memory, plumbing
jason718
committed
c6d6a99
Add the vision task framework and the captioning benchmarks

jason718
and
Ziqi Gao (Roy)
committed
94a37f0
Add the pointing and counting benchmarks

jason718
and
Ziqi Gao (Roy)
committed
70a2143
Add the image-QA benchmarks

jason718
and
Ziqi Gao (Roy)
committed
cad212e
Add the multi-image benchmarks and multimodal docs

jason718
and
Ziqi Gao (Roy)
committed
3c26121