Module modality_hints 

Modality/capability hints derived from a Hugging Face model’s config.json: a first guess (architecture list, a vision block, or a few speech config keys) at what a downloaded model can do, before full identification. A diffusers model_index.json is hinted through the pipeline-family registry (from_model_index).

Structs

Hint

A best-effort guess at a model’s shape from its config metadata.

Enums

SentenceTransformersLayout

What a sentence-transformers layout in a snapshot is.

Functions

audio_hint

A speech-to-text (transcription) model.

embedding_hint

An embedding model.

from_config

The hint for a parsed config.json value: a vision-language model (a vision_config block plus a matching architecture), else the first architecture that matches an [architecture_hint] rule, else a speech model recognized by its config keys, else None.

from_config_json

The hint for a Hugging Face config.json file, or None if it is unreadable, unparseable, or matches no rule.

from_model_index

The hint for a diffusers model_index.json: its pipeline family’s modality and capabilities as a job. An unknown or absent _class_name yields an empty job hint (still a job — a diffusers bundle is never a streaming model).

gguf_hint

The default hint for a bare GGUF weight (text chat/completion).

reranker_hint

A reranker: text in, a relevance score out. It chats with no one and embeds nothing, so it claims no capability of its own; the runtime that can serve it supplies what it does.

sentence_transformers_layout

The sentence-transformers layout in dir, if there is one. Both kinds ship a config_sentence_transformers.json; only its model_type tells a cross-encoder from an embedder, and one read as the other is served nonsense.

speech_hint

A text-to-speech model.

text_hint

A text chat/completion model.

vision_chat_hint

A vision-capable chat model.

whisper_bin_hint

The default hint for a whisper .bin (transcription).