llama-cpp-pro 0.2.2

package/dist/esm/workers/worker.protocol.d.ts

@@ -0,0 +1,160 @@
+export type WorkerRequest = {
+    id: string;
+    type: 'INIT';
+} | {
+    id: string;
+    type: 'LOAD_MODEL';
+    modelId: string;
+    opts?: Record<string, unknown>;
+} | {
+    id: string;
+    type: 'UNLOAD_MODEL';
+    modelId: string;
+} | {
+    id: string;
+    type: 'GENERATE';
+    modelId: string;
+    req: {
+        prompt?: string;
+        messages?: Array<{
+            role: string;
+            content: string;
+        }>;
+        max_tokens?: number;
+        temperature?: number;
+        stream?: boolean;
+    };
+} | {
+    id: string;
+    type: 'EMBED';
+    modelId: string;
+    input: string | string[];
+} | {
+    id: string;
+    type: 'TOKENIZE';
+    modelId: string;
+    text: string;
+} | {
+    id: string;
+    type: 'DETOKENIZE';
+    modelId: string;
+    tokens: number[];
+} | {
+    id: string;
+    type: 'CONVERT_GRAMMAR';
+    schemaJson: string;
+} | {
+    id: string;
+    type: 'RERANK';
+    modelId: string;
+    query: string;
+    documents: string[];
+} | {
+    id: string;
+    type: 'BENCH';
+    modelId: string;
+    pp: number;
+    tg: number;
+    pl: number;
+    nr: number;
+} | {
+    id: string;
+    type: 'SAVE_SESSION';
+    modelId: string;
+    filepath: string;
+    tokenSize: number;
+} | {
+    id: string;
+    type: 'LOAD_SESSION';
+    modelId: string;
+    filepath: string;
+} | {
+    id: string;
+    type: 'APPLY_LORA';
+    modelId: string;
+    loraAdapters: Array<{
+        path: string;
+        scaled?: number;
+    }>;
+} | {
+    id: string;
+    type: 'REMOVE_LORA';
+    modelId: string;
+} | {
+    id: string;
+    type: 'GET_LORA';
+    modelId: string;
+} | {
+    id: string;
+    type: 'INIT_MULTIMODAL';
+    modelId: string;
+    path: string;
+    useGpu?: boolean;
+} | {
+    id: string;
+    type: 'MULTIMODAL_STATUS';
+    modelId: string;
+} | {
+    id: string;
+    type: 'RELEASE_MULTIMODAL';
+    modelId: string;
+} | {
+    id: string;
+    type: 'INIT_VOCODER';
+    modelId: string;
+    path: string;
+    nBatch?: number;
+} | {
+    id: string;
+    type: 'VOCODER_ENABLED';
+    modelId: string;
+} | {
+    id: string;
+    type: 'RELEASE_VOCODER';
+    modelId: string;
+} | {
+    id: string;
+    type: 'FORMATTED_AUDIO';
+    modelId: string;
+    speakerJson: string;
+    textToSpeak: string;
+} | {
+    id: string;
+    type: 'AUDIO_GUIDE_TOKENS';
+    modelId: string;
+    textToSpeak: string;
+} | {
+    id: string;
+    type: 'DECODE_AUDIO_TOKENS';
+    modelId: string;
+    tokens: number[];
+} | {
+    id: string;
+    type: 'HEALTH';
+} | {
+    id: string;
+    type: 'MEMORY';
+};
+export type WorkerEvent = {
+    id: string;
+    type: 'TOKEN';
+    modelId: string;
+    token: string;
+    index: number;
+} | {
+    id: string;
+    type: 'PROGRESS';
+    modelId: string;
+    downloaded: number;
+    total: number;
+} | {
+    id: string;
+    type: 'RESULT';
+    payload: unknown;
+} | {
+    id: string;
+    type: 'ERROR';
+    code: string;
+    message: string;
+    meta?: Record<string, unknown>;
+};

package/dist/esm/workers/worker.protocol.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=worker.protocol.js.map

package/dist/esm/workers/worker.protocol.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"worker.protocol.js","sourceRoot":"","sources":["../../../src/workers/worker.protocol.ts"],"names":[],"mappings":"","sourcesContent":["// Fix #9: LOAD_MODEL no longer carries an ArrayBuffer from the main thread.\n// The worker reads the model from OPFS directly, halving peak memory usage.\nexport type WorkerRequest =\n  | { id: string; type: 'INIT' }\n  | {\n      id: string;\n      type: 'LOAD_MODEL';\n      modelId: string;\n      // modelBuffer removed — worker fetches from OPFS directly (#9)\n      opts?: Record<string, unknown>;\n    }\n  | { id: string; type: 'UNLOAD_MODEL'; modelId: string }\n  | {\n      id: string;\n      type: 'GENERATE';\n      modelId: string;\n      req: {\n        prompt?: string;\n        messages?: Array<{ role: string; content: string }>;\n        max_tokens?: number;\n        temperature?: number;\n        stream?: boolean;\n      };\n    }\n  | { id: string; type: 'EMBED'; modelId: string; input: string | string[] }\n  | { id: string; type: 'TOKENIZE'; modelId: string; text: string }\n  | { id: string; type: 'DETOKENIZE'; modelId: string; tokens: number[] }\n  | { id: string; type: 'CONVERT_GRAMMAR'; schemaJson: string }\n  | {\n      id: string;\n      type: 'RERANK';\n      modelId: string;\n      query: string;\n      documents: string[];\n    }\n  | {\n      id: string;\n      type: 'BENCH';\n      modelId: string;\n      pp: number;\n      tg: number;\n      pl: number;\n      nr: number;\n    }\n  | { id: string; type: 'SAVE_SESSION'; modelId: string; filepath: string; tokenSize: number }\n  | { id: string; type: 'LOAD_SESSION'; modelId: string; filepath: string }\n  | { id: string; type: 'APPLY_LORA'; modelId: string; loraAdapters: Array<{ path: string; scaled?: number }> }\n  | { id: string; type: 'REMOVE_LORA'; modelId: string }\n  | { id: string; type: 'GET_LORA'; modelId: string }\n  | { id: string; type: 'INIT_MULTIMODAL'; modelId: string; path: string; useGpu?: boolean }\n  | { id: string; type: 'MULTIMODAL_STATUS'; modelId: string }\n  | { id: string; type: 'RELEASE_MULTIMODAL'; modelId: string }\n  | { id: string; type: 'INIT_VOCODER'; modelId: string; path: string; nBatch?: number }\n  | { id: string; type: 'VOCODER_ENABLED'; modelId: string }\n  | { id: string; type: 'RELEASE_VOCODER'; modelId: string }\n  | {\n      id: string;\n      type: 'FORMATTED_AUDIO';\n      modelId: string;\n      speakerJson: string;\n      textToSpeak: string;\n    }\n  | { id: string; type: 'AUDIO_GUIDE_TOKENS'; modelId: string; textToSpeak: string }\n  | { id: string; type: 'DECODE_AUDIO_TOKENS'; modelId: string; tokens: number[] }\n  | { id: string; type: 'HEALTH' }\n  | { id: string; type: 'MEMORY' };\n\nexport type WorkerEvent =\n  | { id: string; type: 'TOKEN'; modelId: string; token: string; index: number }\n  // Fix #6: progress events for OPFS model downloads\n  | { id: string; type: 'PROGRESS'; modelId: string; downloaded: number; total: number }\n  | { id: string; type: 'RESULT'; payload: unknown }\n  | { id: string; type: 'ERROR'; code: string; message: string; meta?: Record<string, unknown> };\n"]}