@blank-utils/llm 0.1.0 → 0.2.0

package/dist/types.d.ts

@@ -0,0 +1,178 @@
+/**
+ * Local LLM - Browser-based LLM inference library
+ * Shared TypeScript types and interfaces
+ */
+/**
+ * Supported backend engines
+ */
+export type Backend = 'webllm' | 'transformers' | 'auto';
+/**
+ * Device to run inference on
+ */
+export type Device = 'webgpu' | 'wasm' | 'auto';
+/**
+ * Quantization options for model loading
+ */
+export type Quantization = 'q4' | 'q8' | 'fp16' | 'fp32';
+/**
+ * Configuration for creating an LLM instance
+ */
+export interface LLMConfig {
+    /**
+     * Model identifier. For WebLLM, use MLC model IDs.
+     * For Transformers.js, use HuggingFace model IDs.
+     * @default 'Phi-3-mini-4k-instruct-q4f16_1-MLC' for WebLLM
+     */
+    model?: string;
+    /**
+     * Which backend to use
+     * @default 'auto' - Will prefer WebLLM if WebGPU available
+     */
+    backend?: Backend;
+    /**
+     * Device to run on
+     * @default 'auto' - Prefers WebGPU, falls back to WASM
+     */
+    device?: Device;
+    /**
+     * Quantization level (transformers.js only)
+     * @default 'q4'
+     */
+    quantization?: Quantization;
+    /**
+     * System prompt to use for all conversations
+     */
+    systemPrompt?: string;
+    /**
+     * Callback for model loading progress
+     */
+    onLoadProgress?: LoadProgressCallback;
+}
+/**
+ * Role in a chat conversation
+ */
+export type MessageRole = 'system' | 'user' | 'assistant';
+/**
+ * A single message in a chat conversation
+ */
+export interface ChatMessage {
+    role: MessageRole;
+    content: string;
+}
+/**
+ * Callback for streaming token output
+ */
+export type StreamCallback = (token: string, fullText: string) => void;
+/**
+ * Loading progress information
+ */
+export interface LoadProgress {
+    /** Progress percentage (0-100) */
+    progress: number;
+    /** Current status text */
+    status: string;
+    /** Bytes loaded (if available) */
+    loaded?: number;
+    /** Total bytes (if available) */
+    total?: number;
+}
+/**
+ * Callback for model loading progress
+ */
+export type LoadProgressCallback = (progress: LoadProgress) => void;
+/**
+ * Options for text generation
+ */
+export interface GenerateOptions {
+    /**
+     * Temperature for sampling (0-2)
+     * @default 0.7
+     */
+    temperature?: number;
+    /**
+     * Maximum tokens to generate
+     * @default 512
+     */
+    maxTokens?: number;
+    /**
+     * Top-p sampling
+     * @default 0.95
+     */
+    topP?: number;
+    /**
+     * Stop sequences
+     */
+    stopSequences?: string[];
+}
+/**
+ * Unified interface for LLM backends
+ */
+export interface LLMProvider {
+    /**
+     * Backend identifier
+     */
+    readonly backend: Backend;
+    /**
+     * Whether the model is loaded and ready
+     */
+    readonly isReady: boolean;
+    /**
+     * Current model ID
+     */
+    readonly modelId: string | null;
+    /**
+     * Load a model
+     */
+    load(modelId: string, onProgress?: LoadProgressCallback): Promise<void>;
+    /**
+     * Generate a response (non-streaming)
+     */
+    chat(messages: ChatMessage[], options?: GenerateOptions): Promise<string>;
+    /**
+     * Generate a response with streaming
+     */
+    stream(messages: ChatMessage[], onToken: StreamCallback, options?: GenerateOptions): Promise<string>;
+    /**
+     * Unload the model and free resources
+     */
+    unload(): Promise<void>;
+}
+/**
+ * Options for attaching to an input element
+ */
+export interface AttachOptions {
+    /**
+     * Trigger generation on Enter key
+     * @default true
+     */
+    triggerOnEnter?: boolean;
+    /**
+     * Clear input after sending
+     * @default true
+     */
+    clearOnSend?: boolean;
+    /**
+     * Show loading indicator
+     * @default true
+     */
+    showLoading?: boolean;
+    /**
+     * Custom loading text
+     * @default 'Thinking...'
+     */
+    loadingText?: string;
+}
+/**
+ * Browser capability information
+ */
+export interface BrowserCapabilities {
+    /** WebGPU is available */
+    webgpu: boolean;
+    /** WebAssembly is available */
+    wasm: boolean;
+    /** Recommended backend based on capabilities */
+    recommendedBackend: Backend;
+    /** Recommended device based on capabilities */
+    recommendedDevice: Device;
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAMH;;GAEG;AACH,MAAM,MAAM,OAAO,GAAG,QAAQ,GAAG,cAAc,GAAG,MAAM,CAAC;AAEzD;;GAEG;AACH,MAAM,MAAM,MAAM,GAAG,QAAQ,GAAG,MAAM,GAAG,MAAM,CAAC;AAEhD;;GAEG;AACH,MAAM,MAAM,YAAY,GAAG,IAAI,GAAG,IAAI,GAAG,MAAM,GAAG,MAAM,CAAC;AAEzD;;GAEG;AACH,MAAM,WAAW,SAAS;IACxB;;;;OAIG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;IAEf;;;OAGG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAElB;;;OAGG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;;OAGG;IACH,YAAY,CAAC,EAAE,YAAY,CAAC;IAE5B;;OAEG;IACH,YAAY,CAAC,EAAE,MAAM,CAAC;IAEtB;;OAEG;IACH,cAAc,CAAC,EAAE,oBAAoB,CAAC;CACvC;AAMD;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,QAAQ,GAAG,MAAM,GAAG,WAAW,CAAC;AAE1D;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,IAAI,EAAE,WAAW,CAAC;IAClB,OAAO,EAAE,MAAM,CAAC;CACjB;AAMD;;GAEG;AACH,MAAM,MAAM,cAAc,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,KAAK,IAAI,CAAC;AAEvE;;GAEG;AACH,MAAM,WAAW,YAAY;IAC3B,kCAAkC;IAClC,QAAQ,EAAE,MAAM,CAAC;IACjB,0BAA0B;IAC1B,MAAM,EAAE,MAAM,CAAC;IACf,kCAAkC;IAClC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,iCAAiC;IACjC,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED;;GAEG;AACH,MAAM,MAAM,oBAAoB,GAAG,CAAC,QAAQ,EAAE,YAAY,KAAK,IAAI,CAAC;AAMpE;;GAEG;AACH,MAAM,WAAW,eAAe;IAC9B;;;OAGG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;OAGG;IACH,SAAS,CAAC,EAAE,MAAM,CAAC;IAEnB;;;OAGG;IACH,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd;;OAEG;IACH,aAAa,CAAC,EAAE,MAAM,EAAE,CAAC;CAC1B;AAMD;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B;;OAEG;IACH,QAAQ,CAAC,OAAO,EAAE,OAAO,CAAC;IAE1B;;OAEG;IACH,QAAQ,CAAC,OAAO,EAAE,OAAO,CAAC;IAE1B;;OAEG;IACH,QAAQ,CAAC,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IAEhC;;OAEG;IACH,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,UAAU,CAAC,EAAE,oBAAoB,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC;IAExE;;OAEG;IACH,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,EAAE,OAAO,CAAC,EAAE,eAAe,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;IAE1E;;OAEG;IACH,MAAM,CACJ,QAAQ,EAAE,WAAW,EAAE,EACvB,OAAO,EAAE,cAAc,EACvB,OAAO,CAAC,EAAE,eAAe,GACxB,OAAO,CAAC,MAAM,CAAC,CAAC;IAEnB;;OAEG;IACH,MAAM,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;CACzB;AAMD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B;;;OAGG;IACH,cAAc,CAAC,EAAE,OAAO,CAAC;IAEzB;;;OAGG;IACH,WAAW,CAAC,EAAE,OAAO,CAAC;IAEtB;;;OAGG;IACH,WAAW,CAAC,EAAE,OAAO,CAAC;IAEtB;;;OAGG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB;AAMD;;GAEG;AACH,MAAM,WAAW,mBAAmB;IAClC,0BAA0B;IAC1B,MAAM,EAAE,OAAO,CAAC;IAChB,+BAA+B;IAC/B,IAAI,EAAE,OAAO,CAAC;IACd,gDAAgD;IAChD,kBAAkB,EAAE,OAAO,CAAC;IAC5B,+CAA+C;IAC/C,iBAAiB,EAAE,MAAM,CAAC;CAC3B"}

package/package.json

@@ -1,32 +1,46 @@
 {
   "name": "@blank-utils/llm",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Run LLMs directly in your browser with WebGPU acceleration. Supports React hooks and eager background loading.",
-  "module": "./dist/index.js",
+  "type": "module",
   "main": "./dist/index.js",
+  "module": "./dist/index.js",
   "types": "./dist/index.d.ts",
-  "type": "module",
+  "sideEffects": false,
   "exports": {
     ".": {
+      "types": "./dist/index.d.ts",
       "import": "./dist/index.js",
-      "types": "./dist/index.d.ts"
+      "default": "./dist/index.js"
     },
     "./react": {
+      "types": "./dist/react/index.d.ts",
       "import": "./dist/react/index.js",
-      "types": "./dist/react/index.d.ts"
+      "default": "./dist/react/index.js"
+    },
+    "./package.json": "./package.json"
+  },
+  "typesVersions": {
+    "*": {
+      ".": ["./dist/index.d.ts"],
+      "react": ["./dist/react/index.d.ts"]
     }
   },
   "files": [
     "dist",
-    "README.md"
+    "README.md",
+    "LICENSE"
   ],
   "scripts": {
     "dev": "bun --watch ./src/index.ts",
-    "build": "bun build ./src/index.ts ./src/react/index.tsx --outdir ./dist --format esm --sourcemap --splitting --external react",
-    "build:types": "tsc --declaration --emitDeclarationOnly --outDir dist",
-    "prepublishOnly": "pnpm build",
+    "clean": "rm -rf dist",
+    "build:js": "bun build ./src/index.ts ./src/react/index.tsx --outdir ./dist --format esm --sourcemap --splitting --external react --external react-dom",
+    "build:types": "tsc -p tsconfig.build.json",
+    "build": "bun run clean && bun run build:js && bun run build:types",
+    "prepublishOnly": "bun run build",
     "demo": "bunx serve . -p 3000",
-    "typecheck": "tsc --noEmit"
+    "typecheck": "tsc --noEmit",
+    "test": "bun test"
   },
   "keywords": [
     "llm",
@@ -40,30 +54,34 @@
     "react",
     "hooks",
     "streaming",
-    "chat"
+    "chat",
+    "gpt",
+    "huggingface"
   ],
   "author": "blank",
   "license": "MIT",
   "repository": {
     "type": "git",
-    "url": "https://github.com/blank-utils/llm"
+    "url": "git+https://github.com/kiritocode1/local-llm.git"
   },
-  "homepage": "https://github.com/blank-utils/llm#readme",
+  "homepage": "https://github.com/kiritocode1/local-llm#readme",
   "bugs": {
-    "url": "https://github.com/blank-utils/llm/issues"
+    "url": "https://github.com/kiritocode1/local-llm/issues"
   },
   "publishConfig": {
     "access": "public"
   },
+  "engines": {
+    "node": ">=18"
+  },
   "devDependencies": {
     "@types/bun": "latest",
-    "@types/react": "^18.3.28",
-    "@types/react-dom": "^19.2.3",
-    "@webgpu/types": "^0.1.69"
+    "@types/react": "^18.3.0 || ^19.0.0",
+    "@webgpu/types": "^0.1.69",
+    "typescript": "^5.0.0"
   },
   "peerDependencies": {
-    "react": "^18 || ^19",
-    "typescript": "^5"
+    "react": ">=18.0.0"
   },
   "peerDependenciesMeta": {
     "react": {
@@ -75,4 +93,3 @@
     "@mlc-ai/web-llm": "^0.2.80"
   }
 }
-