update: sync latest llama.cpp branch

example/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llama-node/examples",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "description": "",
   "main": "index.js",
   "type": "module",
@@ -19,9 +19,9 @@
     "langchain": "^0.0.56"
   },
   "dependencies": {
-    "@llama-node/core": "0.1.5",
-    "@llama-node/llama-cpp": "0.1.5",
-    "@llama-node/rwkv-cpp": "0.1.5",
-    "llama-node": "0.1.5"
+    "@llama-node/core": "0.1.6",
+    "@llama-node/llama-cpp": "0.1.6",
+    "@llama-node/rwkv-cpp": "0.1.6",
+    "llama-node": "0.1.6"
   }
 }

package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama-node",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "description": "Node.js Library for Large Language Model LLaMA/RWKV",
   "keywords": [
     "llama",
@@ -58,10 +58,10 @@
   "packageManager": "[email protected]",
   "homepage": "https://llama-node.vercel.app/",
   "devDependencies": {
-    "@llama-node/cli": "0.1.5",
-    "@llama-node/core": "0.1.5",
-    "@llama-node/llama-cpp": "0.1.5",
-    "@llama-node/rwkv-cpp": "0.1.5",
+    "@llama-node/cli": "0.1.6",
+    "@llama-node/core": "0.1.6",
+    "@llama-node/llama-cpp": "0.1.6",
+    "@llama-node/rwkv-cpp": "0.1.6",
     "@napi-rs/cli": "^2.15.2",
     "@tensorflow/tfjs-node": "^4.2.0",
     "@types/node": "^18.15.5",
@@ -80,17 +80,17 @@
     "vitest": "^0.30.1"
   },
   "dependencies": {
-    "@llama-node/cli": "0.1.5"
+    "@llama-node/cli": "0.1.6"
   },
   "optionalDependencies": {
-    "@llama-node/core": "0.1.5",
-    "@llama-node/llama-cpp": "0.1.5",
-    "@llama-node/rwkv-cpp": "0.1.5"
+    "@llama-node/core": "0.1.6",
+    "@llama-node/llama-cpp": "0.1.6",
+    "@llama-node/rwkv-cpp": "0.1.6"
   },
   "peerDependencies": {
-    "@llama-node/cli": "0.1.5",
-    "@llama-node/core": "0.1.5",
-    "@llama-node/llama-cpp": "0.1.5",
-    "@llama-node/rwkv-cpp": "0.1.5"
+    "@llama-node/cli": "0.1.6",
+    "@llama-node/core": "0.1.6",
+    "@llama-node/llama-cpp": "0.1.6",
+    "@llama-node/rwkv-cpp": "0.1.6"
   }
 }

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llama-node/cli",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "description": "",
   "main": "index.js",
   "type": "commonjs",
@@ -12,7 +12,7 @@
     "test": "tsx src/index.ts"
   },
   "dependencies": {
-    "@llama-node/core": "0.1.5",
+    "@llama-node/core": "0.1.6",
     "yargs": "^17.7.1"
   },
   "author": "",

packages/cli/src/index.ts

@@ -93,54 +93,9 @@ class InferenceCommand implements yargs.CommandModule {
     }
 }
 
-class ConvertCommand implements yargs.CommandModule<any, any> {
-    command = "convert";
-    describe = "Convert llama pth to ggml, not ready yet";
-    builder(args: yargs.Argv) {
-        return args
-            .help("help")
-            .example(
-                "llama convert --dir ./model --type q4_0",
-                "Convert pth to q4_0 ggml model"
-            )
-            .options({
-                dir: {
-                    describe: "The directory of model and tokenizer directory",
-                    type: "string",
-                    demandOption: true,
-                },
-            })
-            .options({
-                type: {
-                    describe: "The type of model",
-                    type: "string",
-                    choices: convertType,
-                    demandOption: true,
-                },
-            });
-    }
-    async handler(args: yargs.ArgumentsCamelCase) {
-        const dir = args.dir as string;
-        const type = args.type as ConvertType;
-
-        const absolute = path.isAbsolute(dir)
-            ? dir
-            : path.join(process.cwd(), dir);
-        if (!existsSync(absolute)) {
-            console.error(`Directory ${absolute} does not exist`);
-            return;
-        } else {
-            const elementType = convertType.findIndex((t) => t === type);
-            await convert(absolute, elementType);
-            console.log("Convert successfully");
-        }
-    }
-}
-
 (yargs as yargs.Argv<any | CLIInferenceArguments>)
     .scriptName("llama")
     .usage("$0 <cmd> [args]")
-    .command(new ConvertCommand())
     .command(new InferenceCommand())
     .demandCommand(1, "You need at least one command before moving on")
     .strict()

packages/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llama-node/core",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "main": "index.js",
   "types": "index.d.ts",
   "napi": {

packages/docs/docs/start.md

@@ -122,6 +122,7 @@ const config = {
     useMlock: false,
     embedding: false,
     useMmap: true,
+    nGpuLayers: 0
 };
 
 const template = `How are you?`;

packages/llama-cpp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llama-node/llama-cpp",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "main": "index.js",
   "types": "index.d.ts",
   "napi": {

packages/rwkv-cpp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llama-node/rwkv-cpp",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "main": "index.js",
   "types": "index.d.ts",
   "napi": {