mrleo1nid/swarm-assistent
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Harden Exact reconcile and soften post-Generate warm.

Browse Source 
Overwrite foreign steps/cfg in sparse patches when the user did not ask for params, and skip the cold-load path when Ollama already has the chat model resident.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Leonid Pershin
2026-08-23 06:58:53 +03:00
co-authored by Cursor
parent 5f33130381
commit 7aadb40a11
7 changed files with 117 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+1 -3
View File
@@ -4,9 +4,7 @@ SwarmUI extension for **collaborative Krea 2** prompting via **Ollama**: chat +
**Turn model:** one user message is one *turn*. A turn may fan out into nested LLM *hops* — Krea prompt prep, empty-patch retry, vision, auto-critique. Hops share one `HOP_BUDGET`, never re-read the user's text (their prompt is client-authored), and pass the busy gate that blocks new user sends. What a reply does to generation state is decided once, in `resolveTurnIntent`: the model's `actions:["generate"]` / `look_at` win, RU intent heuristics only back it up when the model forgets, and an explicit «запомни, не генерируй» vetoes both.
**Version 0.14.1** — On Generate, force Exact turbo/raw params when live Swarm still has foreign leftovers (e.g. steps 20 / cfg 7). `detectKreaProfileName` no longer invents turbo for unlabeled ckpts (realismByStableYogi → raw). Live context injects `krea_profile` + `recommended_params`. RU «шаг/шагами» counts as param intent. Soft sparse-prompt exception when session ≠ Exact profile.
**Version 0.14.1** — Generate always merges Exact turbo|raw `steps`/`cfg`/`sigma_shift` (client-authoritative; sparse LLM omit is safe). Park/warm skip no-op Ollama round-trips when the chat model is already (un)loaded via `/api/ps`. Builds on 0.14.0.
**Version 0.14.1** — Force Exact turbo/raw numbers on Generate when live Swarm still has foreign leftovers (e.g. steps 20 / cfg 7), including when the profile *label* is turbo/raw. detectKreaProfileName no longer invents turbo for unlabeled ckpts (realismByStableYogi → raw). Live context injects krea_profile + ecommended_params. RU «шаг/шагами» counts as param intent. Soft sparse-prompt exception when session ≠ Exact profile. Post-Generate warm still runs with park off, but skips the cold path when Ollama already has the chat model resident.
**Version 0.14.0** — **Чат = сессия генерации**: у каждого чата свои params/LoRA/checkpoint/кадр/refs; модель шлёт sparse-дельту + `generate`/`look_at`/`ask`; без вкладки Карточки и Civitai/wanted hops. **Сжатие контекста**: rolling-саммари той же Ollama-моделью, чип бюджета `N / num_ctx`, авто перед отправкой, `/compress`.