Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-01 10:01:01 +02:00
Code Issues Packages Projects Releases Wiki Activity
11,820 Commits 3 Branches 34 Tags
fab224c4dffca2fb3bda7c22e849c19dea9b0263
Commit Graph

5 Commits

Author SHA1 Message Date
CalamitousFelicitousness 7714f71994 feat(vqa): un/load support and extract detection 
Make external VQA handlers (moondream3, joytag, joycaption, deepseek)
compatible with VQA load/unload mechanism for consistent model lifecycle.

- Added vqa_detection.py, add shared detection helpers
- Add load and unload functions to all external handlers
- Replace device_map="auto" with sd_models.move_model in joycaption
- Update dispatcher and moondream handlers to use shared helpers
 2025-12-05 23:52:02 +00:00
CalamitousFelicitousness eb832a4850 fix(vqa): respect offload setting in JoyCaption, add max_pixels 
Two fixes for the JoyCaption handler:
- Only offload model if shared.opts.interrogate_offload is True
- Add max_pixels=1024*1024 to AutoProcessor for consistent image handling
 2025-12-02 21:46:09 +00:00
Vladimir Mandic 1b4e1ff0ef enable quants for vlm-captioning 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
 2025-06-29 11:48:05 -04:00
Vladimir Mandic 5b486a6ef1 sdnq add xyz grid support, improve offloading compatibility 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
 2025-06-25 15:32:37 -04:00
Vladimir Mandic abd439d97c add joycaption 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
 2025-02-15 10:10:04 -05:00