tokenize-it / src /tokenizer /types.ts
Orion-zhen's picture
:boom: use web stack
21ab33c unverified
Raw History Blame Contribute Delete
1.84 kB
export type TokenizerFormat =
| "tokenizers"
| "sentencepiece"
| "tiktoken"
| "tekken";
export interface ModelReference {
repoId: string;
revision: string;
}
export interface HubSibling {
rfilename: string;
blobId?: string;
size?: number;
lfs?: {
sha256: string;
size: number;
pointerSize: number;
};
}
export interface FileHash {
algorithm: "git-sha1" | "sha256";
value: string;
}
export interface HubFile {
filename: string;
hash: FileHash;
}
export interface RemoteFile extends HubFile {
url: string;
}
export interface HubModelInfo {
id?: string;
modelId?: string;
sha: string;
private?: boolean;
gated?: boolean | "auto" | "manual";
siblings: HubSibling[];
config?: {
model_type?: string;
tokenizer_config?: Record<string, unknown>;
};
}
export interface ResolvedTokenizer {
repoId: string;
requestedRevision: string;
commitSha: string;
format: TokenizerFormat;
file: HubFile;
availableFiles: HubFile[];
tokenizerConfig?: Record<string, unknown>;
}
export interface LoadProgress {
phase: "resolving" | "downloading" | "initializing";
loaded: number;
total: number | null;
cached?: boolean;
message: string;
}
export type SpecialTokenMap = Record<string, string | string[]>;
export interface TokenizerMetadata {
repoId: string;
requestedRevision: string;
commitSha: string;
format: TokenizerFormat;
filename: string;
tokenizerType: string;
vocabSize: number | null;
specialTokens: SpecialTokenMap;
fileSize: number;
loadTimeMs: number;
sourceUrl: string;
}
export interface TokenPiece {
index: number;
id: number;
text: string;
displayText: string;
special: boolean;
}
export interface TokenizationResult {
tokens: TokenPiece[];
tokenCount: number;
characterCount: number;
elapsedMs: number;
}