Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-19 01:04:32 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix(caption): remove dead min_length param, split Florence/PromptGen prompts, fix gaze detection

Browse Source 
- Remove caption_openclip_min_length from settings, API models, endpoints, and UI
  (clip_interrogator library has no min_length support; parameter was never functional)
- Split vlm_prompts_florence into base Florence prompts and PromptGen-only prompts
  (GENERATE_TAGS, Analyze, Mixed Caption require MiaoshouAI PromptGen fine-tune)
- Add 'promptgen' category to /vqa/prompts API endpoint
- Fix gaze detection: move DETECT_GAZE check before generic 'detect ' prefix
  to prevent "Detect Gaze" matching as detect target="Gaze"
- Update test suite: remove min_length tests, fix min_flavors to use mode='best',
  add acceptance-only notes, fix thinking trace detection, improve bracket/OCR tests,
  split Florence/PromptGen test coverage
This commit is contained in:
CalamitousFelicitousness
2026-01-29 01:21:28 +00:00
parent fba942b25e
commit bf7a72f12e
6 changed files with 214 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/api/models.py
-2
View File
@@ -378,7 +378,6 @@ class ReqCaption(BaseModel):
mode: str = Field(default="best", title="Mode", description="Caption mode. Fast: Quick caption with minimal flavor terms. Classic: Standard captioning with balanced quality and speed. Best: Most thorough analysis, slowest but highest quality. Negative: Generate terms to use as negative prompt.")
analyze: bool = Field(default=False, title="Analyze", description="If True, returns detailed image analysis breakdown (medium, artist, movement, trending, flavor) in addition to caption.")
# Advanced settings (optional per-request overrides)
min_length: Optional[int] = Field(default=None, title="Min Length", description="Minimum number of tokens in the generated caption.")
max_length: Optional[int] = Field(default=None, title="Max Length", description="Maximum number of tokens in the generated caption.")
chunk_size: Optional[int] = Field(default=None, title="Chunk Size", description="Batch size for processing description candidates (flavors). Higher values speed up captioning but increase VRAM usage.")
min_flavors: Optional[int] = Field(default=None, title="Min Flavors", description="Minimum number of descriptive tags (flavors) to keep in the final prompt.")
@@ -486,7 +485,6 @@ class ReqCaptionOpenCLIP(BaseModel):
blip_model: str = Field(default="blip-large", title="Caption Model", description="BLIP model used to generate the initial image caption.")
mode: str = Field(default="best", title="Mode", description="Caption mode: 'best' (highest quality), 'fast' (quick), 'classic' (traditional), 'caption' (BLIP only), 'negative' (for negative prompts).")
analyze: bool = Field(default=False, title="Analyze", description="If True, returns detailed breakdown (medium, artist, movement, trending, flavor).")
min_length: Optional[int] = Field(default=None, title="Min Length", description="Minimum tokens in generated caption.")
max_length: Optional[int] = Field(default=None, title="Max Length", description="Maximum tokens in generated caption.")
chunk_size: Optional[int] = Field(default=None, title="Chunk Size", description="Batch size for processing flavors.")
min_flavors: Optional[int] = Field(default=None, title="Min Flavors", description="Minimum descriptive tags to keep.")