Akasei 9bee8e1844 refactor(gradio): replace CPU offload with direct GPU unload/lazy-load ...

Instead of .to('cpu') / .to('cuda'), models are now fully del'd from
GPU (no CPU intermediate) and reloaded on demand:

- _unload_i23d_worker(): del + gc.collect() + empty_cache()
- _ensure_i23d_worker(): lazy reload from pretrained if None
- _unload_tex_pipeline(): del + gc.collect() + empty_cache()
- _ensure_tex_pipeline(): lazy load from tex_conf if None

generation_all() flow in low_vram_mode:
  shape gen → _unload_i23d_worker → _ensure_tex_pipeline →
  texture gen → _unload_tex_pipeline
  (shape model reloads on next _gen_shape call via _ensure_i23d_worker)

Startup: tex_pipeline NOT loaded in low_vram_mode (only tex_conf stored),
reducing startup VRAM from ~13.5GB to ~7.25GB.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

2026-03-16 21:15:56 +08:00

36 KiB

Raw Blame History

View Raw