FCllama

Flutter binding of llama.cpp , which use platform channel .

llama.cpp: Inference of LLaMA model in pure C/C++

Installation

Flutter

flutter pub add fcllama

iOS

Please run pod install or pod update in your iOS project.

Android

You need install cmake 3.31.0、android sdk 35 and ndk 28.0.12674087. No additional operation required .

OpenHarmonyOS/HarmonyOS

This is the fastest and recommended way to add HLlama to your project.

ohpm install hllama

Or, you can add it to your project manually.

• Add the following lines to oh-package.json5 on your app module.

"dependencies": {
  "hllama": "^0.0.2",
}

• Then run

  ohpm install

How to use

Flutter

1. Initializing Llama

import 'package:fcllama/fllama.dart';

FCllama.instance()?.initContext("model path",emitLoadProgress: true)
        .then((context) {
  modelContextId = context?["contextId"].toString() ?? "";
  if (modelContextId.isNotEmpty) {
    // you can get modelContextId，if modelContextId > 0 is success.
  }
});

2. Bench model on device

import 'package:fcllama/fllama.dart';

FCllama.instance()?.bench(double.parse(modelContextId),pp:8,tg:4,pl:2,nr: 1).then((res){
  Get.log("[FCllama] Bench Res $res");
});

3. Tokenize and Detokenize

import 'package:fcllama/fllama.dart';

FCllama.instance()?.tokenize(double.parse(modelContextId), text: "What can you do?").then((res){
  Get.log("[FCllama] Tokenize Res $res");
  FCllama.instance()?.detokenize(double.parse(modelContextId), tokens: res?['tokens']).then((res){
    Get.log("[FCllama] Detokenize Res $res");
  });
});

4. Streaming monitoring

import 'package:fcllama/fllama.dart';

FCllama.instance()?.onTokenStream?.listen((data) {
  if(data['function']=="loadProgress"){
    Get.log("[FCllama] loadProgress=${data['result']}");
  }else if(data['function']=="completion"){
    Get.log("[FCllama] completion=${data['result']}");
    final tempRes = data["result"]["token"];
    // tempRes is ans
  }
});

5. Release this or Stop one

import 'package:fcllama/fllama.dart';

FCllama.instance()?.stopCompletion(contextId: double.parse(modelContextId)); // stop one completion
FCllama.instance()?.releaseContext(double.parse(modelContextId)); // release one
FCllama.instance()?.releaseAllContexts(); // release all

OpenHarmonyOS/HarmonyOS

You can see this file

Support System

Obtain the model

You can search HuggingFace for available models (Keyword: GGUF).

For get a GGUF model or quantize manually, see Prepare and Quantize section in llama.cpp.

NOTE

iOS:

• The Extended Virtual Addressing capability is recommended to enable on iOS project.
• Metal:
  • We have tested to know some devices is not able to use Metal ('params.n_gpu_layers > 0') due to llama.cpp used SIMD-scoped operation, you can check if your device is supported in Metal feature set tables, Apple7 GPU will be the minimum requirement.
  • It's also not supported in iOS simulator due to this limitation, we used constant buffers more than 14.

Android:

• Currently only supported arm64-v8a / x86_64 / armeabi-v7a platform, this means you can't initialize a context on another platforms. The 64-bit platform are recommended because it can allocate more memory for the model.
• No integrated any GPU backend yet.

License

MIT