Commit Graph
17 Commits
Author SHA1 Message Date
Mikei386 6255f0b8ea Add conditional external vision projector discovery to profiles 2026-09-28 22:12:50 +02:00
Mikei386 06c6292cf3 Expose model capability hints and configurable vision projector placement 2026-09-28 22:03:39 +02:00
Mikei386 37a757fd9f Refine dual GPU allocation in individual layer steps 2026-09-28 21:53:53 +02:00
Mikei386 a77266c7bd Add bounded GPU-priority automatic model testing and profile export 2026-09-28 21:29:48 +02:00
Mikei386 f99cd26b1d Avoid unnecessary CPU offload and expose full GPU profile mode 2026-09-28 21:18:54 +02:00
Mikei386 13ed059aaf Add configurable MTP and shared-weight draft memory fitting 2026-09-28 20:59:43 +02:00
Mikei386 7ada7f6f20 Add internal chat testing with memory diagnostics and cancellation 2026-09-28 20:43:40 +02:00
Mikei386 ff5d36252a Add owned OpenAI endpoint and coordinated native model switching 2026-09-28 20:24:12 +02:00
Mikei386 8d03a9a979 Add GPU distribution and sampling controls to chat profiles 2026-09-28 19:51:49 +02:00
Mikei386 c584bbcaa6 Add isolated ComfyUI image playground with real profile-based generation 2026-09-28 18:00:15 +02:00
Mikei386 413906f619 Classify library components consistently and hide standalone profile action 2026-09-28 17:41:06 +02:00
Mikei386 057b9df644 Group runtime settings and add profile component downloads and assignments 2026-09-28 17:34:03 +02:00
Mikei386 458047a1d3 Persist library profiles and download history with weight capacity checks 2026-09-28 17:26:21 +02:00
Mikei386 f326de18d1 Improve model discovery, variant selection and download visibility 2026-09-28 17:04:42 +02:00
Mikei386 ba524b0a70 Enable llama.cpp builds, release management and reference model auto-fit 2026-09-28 16:45:24 +02:00
Mikei386 5fb732469a Enable public model catalogue and isolated file downloads; define native target 2026-09-28 15:50:31 +02:00
Mikei386 e1a54e3485 Add visual model catalog, editable profile drafts and llama.cpp build settings 2026-09-28 15:24:55 +02:00