LLM inference in C/C++
C++ ggml
/**
* HuggingFace Hub enums.
*
* Values mirror the strings used by the HF REST API
* (https://huggingface.co/docs/huggingface_hub/package_reference/hf_api)
* so they can be sent and compared directly.
*/
/** Sort field for /api/models search queries. */
export enum HfModelSort {
CREATED_AT = 'createdAt',
DOWNLOADS = 'downloads',
LAST_MODIFIED = 'lastModified',
LIKES = 'likes',
TRENDING_SCORE = 'trendingScore'
}
/**
* Where the sidecar token (`mtp` / `dflash` / `mmproj` / ...) sits in the
* filename.
* - `prefix` sidecar file that lives next to the main weights, e.g. `mtp-Q4_0.gguf`
* - `suffix` embedded draft baked into the main weights, e.g. `Hy3-IQ1_M-mtp.gguf`
* - `infix` standalone sidecar named between head and quant, e.g. `model-mtp-Q8_0.gguf`
*/
export enum SidecarForm {
INFIX = 'infix',
PREFIX = 'prefix',
SUFFIX = 'suffix'
}
/** Entry type in a model repository file tree (`/tree` responses). */
export enum HfEntryType {
DIRECTORY = 'directory',
FILE = 'file'
}