Update Ollama model configurations and documentation for clarity and consistency

- Revised model descriptions in `ollama-models.example.yaml` to emphasize uncensored and abliterated requirements, enhancing user understanding.
- Updated documentation in `llm.md` to reflect changes in model tags and their meanings, ensuring accurate guidance for users.
- Modified CLI help messages to clarify the nature of presets, reinforcing that all options are abliterate models with Russian support.
- Enhanced the `llm_runtime.py` file to align preset labels with the updated model descriptions, improving consistency across the codebase.

docs/llm.md

@@ -83,13 +83,14 @@ $env:OLLAMA_HOST = "http://127.0.0.1:17811"
 
 | # | ключ | tag / смысл |
 | --- | --- | --- |
-| 1 | **recommended** | `huihui_ai/qwen2.5-vl-abliterated:7b` — vision+RU, ~6 GB |
-| 2 | light | VL 3B abliterate (~3 GB) |
-| 3 | text | text-only `qwen2.5-abliterate:7b` |
-| 4 | stock | `qwen2.5:7b` |
-| 5 | alt | другой text abliterate 7B |
-| 6 | empty | только runtime |
-| — | keep | не трогать yaml (если уже спросили повторно) |
+| 1 | **recommended** | `huihui_ai/qwen2.5-vl-abliterated:7b` — RU + vision, ~6 GB |
+| 2 | light | `…:3b` — то же, мало VRAM (~3 GB) |
+| 3 | text | `huihui_ai/qwen2.5-abliterate:7b` — RU, без vision (~5 GB) |
+| 4 | big | `huihui_ai/qwen2.5-vl-abliterated:32b` — RU + vision (~21 GB) |
+| 5 | empty | только runtime |
+| — | keep | не трогать yaml |
+
+Все пресеты (кроме `empty`) — **abliterated** Qwen2.5 с нормальным русским. Официальные censored-теги (`qwen2.5vl:…`) в меню нет.
 
 `default: true` в yaml — preferred в логе; pull идёт по всему списку.
 

ollama-models.example.yaml

@@ -1,15 +1,13 @@
 # Copy to ollama-models.yaml (gitignored). Used when LLM_RUNTIME=ollama.
-# name = exact tag for `ollama pull` (Ollama library / community).
-# Empty models: [] → runtime only, no pull.
-# Purpose: SwarmUI prompt help — vision + RU/EN on the same GPU as diffusion.
+# name = exact tag for `ollama pull`.
+# Requirement: uncensored (abliterated) + Russian. Empty models: [] → no pull.
 
 models:
-  # Recommended (~6GB): vision + Russian/English, low refusal
+  # Recommended (~6GB): vision + RU/EN, abliterated
   - name: huihui_ai/qwen2.5-vl-abliterated:7b
     default: true
 
-  # Alternatives (uncomment / use presets on setup|up):
-  # light  — huihui_ai/qwen2.5-vl-abliterated:3b   (~3GB)
-  # stock  — qwen2.5vl:7b                          (official library)
-  # text   — huihui_ai/qwen2.5-abliterate:7b       (no vision, ~5GB)
-  # big    — qwen2.5vl:32b                         (~21GB; llm-only)
+  # Presets (setup / up --ollama-preset …):
+  # light — huihui_ai/qwen2.5-vl-abliterated:3b   (~3GB)
+  # text  — huihui_ai/qwen2.5-abliterate:7b       (~5GB, no vision)
+  # big   — huihui_ai/qwen2.5-vl-abliterated:32b  (~21GB)

src/gpu_rent/cli.py

@@ -398,7 +398,7 @@ def open(
 def setup(
     llm: Optional[str] = typer.Option(None, "--llm", help="none|ollama"),
     ollama_preset: Optional[str] = typer.Option(
-        None, "--ollama-preset", help="recommended|light|stock|text|big|empty"
+        None, "--ollama-preset", help="recommended|light|text|big|empty (все — abliterate + RU)"
     ),
     watchdog: Optional[bool] = typer.Option(
         None, "--watchdog/--no-watchdog", help="Поставить local-watchdog"

src/gpu_rent/llm_runtime.py

@@ -16,22 +16,20 @@ from gpu_rent.paths import (
 VALID_RUNTIMES = frozenset({"none", "ollama"})
 
 OLLAMA_PRESETS: dict[str, list[str]] = {
-    # Ollama library / community tags only (no GGUF). Tuned for SwarmUI prompt help:
-    # vision + RU/EN, share VRAM with diffusion on typical 24 GiB.
-    "recommended": ["huihui_ai/qwen2.5-vl-abliterated:7b"],  # ~6 GB, low refusal
-    "light": ["huihui_ai/qwen2.5-vl-abliterated:3b"],  # ~3 GB, tight VRAM
-    "stock": ["qwen2.5vl:7b"],  # official library vision
+    # Requirement: uncensored (abliterated) + solid Russian. Qwen2.5 family.
+    # Vision tags preferred for SwarmUI prompt help with images.
+    "recommended": ["huihui_ai/qwen2.5-vl-abliterated:7b"],  # ~6 GB
+    "light": ["huihui_ai/qwen2.5-vl-abliterated:3b"],  # ~3 GB
     "text": ["huihui_ai/qwen2.5-abliterate:7b"],  # ~5 GB, no vision
-    "big": ["qwen2.5vl:32b"],  # ~21 GB — llm-only or ≥40 GiB free
+    "big": ["huihui_ai/qwen2.5-vl-abliterated:32b"],  # ~21 GB
     "empty": [],
 }
 
 OLLAMA_PRESET_LABELS: dict[str, str] = {
-    "recommended": "Qwen2.5-VL 7B abliterate (картинки+RU, ~6GB)",
-    "light": "Qwen2.5-VL 3B abliterate (мало VRAM, ~3GB)",
-    "stock": "официальный qwen2.5vl:7b (library, больше отказов)",
-    "text": "Qwen2.5 7B abliterate text-only (~5GB)",
-    "big": "qwen2.5vl:32b (~21GB; llm-only / большой GPU)",
+    "recommended": "VL 7B abliterate — RU + картинки (~6GB)",
+    "light": "VL 3B abliterate — RU + картинки, мало VRAM (~3GB)",
+    "text": "7B abliterate text — RU, без vision (~5GB)",
+    "big": "VL 32B abliterate — RU + картинки (~21GB)",
     "empty": "только runtime, без pull",
     "keep": "не менять ollama-models.yaml",
 }

tests/test_llm_runtime.py

@@ -55,3 +55,7 @@ def test_write_preset(tmp_path: Path):
     write_ollama_models_preset(path, "recommended")
     entries = parse_ollama_models(path)
     assert entries[0].name == "huihui_ai/qwen2.5-vl-abliterated:7b"
+    write_ollama_models_preset(path, "big")
+    assert parse_ollama_models(path)[0].name == "huihui_ai/qwen2.5-vl-abliterated:32b"
+    write_ollama_models_preset(path, "text")
+    assert parse_ollama_models(path)[0].name == "huihui_ai/qwen2.5-abliterate:7b"