per-sample media budgets in processor; per-sample MRoPE position ids for batched vision inputs

per-sample media budgets in processor; per-sample MRoPE position ids for batched vision inputs

sync offline system prompts with training data: no_thinking -> no system message, deep_thinking -> <think>/<answer> tags

Unify modeling with the 0708 standard (transformers 4.57 compatible)

Unify MOSS-VL processor and inference preprocessing defaults

Fix query RoPE when reusing cached vision keys

Ascend NPU support: device-aware cache release, GIL-friendly idle wait, <=8-D vision processing, BICUBIC declaration (#3)

Upload README.md with huggingface_hub

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Upload benchmark-streaming.png

Delete assets/benchmark-streaming.png

Update README

Upload MOSS-VL-Realtime-0708 release

initial commit