Java Bindings for llama.cpp

The main goal of llama.cpp is to run the LLaMA model using 4-bit integer quantization on a MacBook. This repository provides Java bindings for the C++ library.

You are welcome to contribute

Quick Start

Access this library via Maven:

<dependency>
    <groupId>de.kherud</groupId>
    <artifactId>llama</artifactId>
    <version>2.3.1</version>
</dependency>

There are multiple examples. Make sure to set model.home and model.name to run them:

mvn exec:java -Dexec.mainClass="examples.MainExample" -Dmodel.home="/path/to/models" -Dmodel.name="codellama-13b.Q5_K_M.gguf"

Note: if your model is in the models directory, then you can ommit the -Dmodel.home property.

You can also run some integration tests, which will automatically download a model to the models directory:

mvn verify

No Setup required

We support CPU inference for the following platforms out of the box:

Linux x86-64, aarch64
MacOS x86-64, aarch64 (M1)
Windows x86-64, x64, arm (32 bit)

If any of these match your platform, you can include the Maven dependency and get started.

Setup required

If none of the above listed platforms matches yours, currently you have to compile the library yourself (also if you want GPU acceleration, see below). More support is planned soon.

This requires:

Git
A C++11 conforming compiler
The cmake build system
Java, Maven, and setting JAVA_HOME

Make sure everything works by running

g++ -v  # depending on your compiler
java -version
mvn -v
echo $JAVA_HOME # for linux/macos
echo %JAVA_HOME% # for windows

Then, run the following commands in the directory of this repository (java-llama.cpp):

mvn compile
mkdir build
cd build
cmake .. # add any other arguments for your backend
cmake --build . --config Release

All required files will be put in a resources directory matching your platform, which will appear in the cmake output. For example something like:

--  Installing files to /java-llama.cpp/src/main/resources/de/kherud/llama/Linux/x86_64

This includes:

Linux: libllama.so, libjllama.so
MacOS: libllama.dylib, libjllama.dylib, ggml-metal.metal
Windows: llama.dll, jllama.dll

If you then compile your own JAR from this directory, you are ready to go. Otherwise, if you still want to use the library as a Maven dependency, see below how to set the necessary paths in order for Java to find your compiled libraries.

Custom llama.cpp Setup (GPU acceleration)

This repository provides default support for CPU based inference. You can compile llama.cpp any way you want, however. In order to use your self-compiled library, set either of the JVM options:

de.kherud.llama.lib.path, for example -Dde.kherud.llama.lib.path=/directory/containing/lib
java.library.path, for example -Djava.library.path=/directory/containing/lib

This repository uses System#mapLibraryName to determine the name of the shared library for you platform. If for any reason your library has a different name, you can set it with

de.kherud.llama.lib.name, for example -Dde.kherud.llama.lib.name=myname.so

For compiling llama.cpp, refer to the official readme for details. The library can be built with the llama.cpp project:

mkdir build
cd build
cmake .. -DBUILD_SHARED_LIBS=ON  # add any other arguments for your backend
cmake --build . --config Release

Look for the shared library in build.

Important

If you are running MacOS with Metal, you have to put the file ggml-metal.metal from build/bin in the same directory as the shared library.

Documentation

Example

This is a short example on how to use this library:

public class Example {

	public static void main(String... args) throws IOException {
        LlamaModel.setLogger((level, message) -> System.out.print(message));
        ModelParameters modelParams = new ModelParameters()
                .setNGpuLayers(43);
        InferenceParameters inferParams = new InferenceParameters()
                .setTemperature(0.7f)
                .setPenalizeNl(true)
                .setMirostat(InferenceParameters.MiroStat.V2)
                .setAntiPrompt("\n");

        String modelPath = "/run/media/konstantin/Seagate/models/llama2/llama-2-13b-chat/ggml-model-q4_0.gguf";
        String system = "This is a conversation between User and Llama, a friendly chatbot.\n" +
                "Llama is helpful, kind, honest, good at writing, and never fails to answer any " +
                "requests immediately and with precision.\n";
        BufferedReader reader = new BufferedReader(new InputStreamReader(System.in, StandardCharsets.UTF_8));
        try (LlamaModel model = new LlamaModel(modelPath, modelParams)) {
            System.out.print(system);
            String prompt = system;
            while (true) {
                prompt += "\nUser: ";
                System.out.print("\nUser: ");
                String input = reader.readLine();
                prompt += input;
                System.out.print("Llama: ");
                prompt += "\nLlama: ";
                for (String output : model.generate(prompt, inferParams)) {
                    System.out.print(output);
                    prompt += output;
                }
            }
        }
    }
}

Also have a look at the other examples.

Inference

There are multiple inference tasks. In general, LlamaModel is stateless, i.e., you have to append the output of the model to your prompt in order to extend the context. If there is repeated content, however, the library will internally cache this, to improve performance.

try (LlamaModel model = new LlamaModel("/path/to/gguf-model")) {
    // Stream a response and access more information about each output.
    for (String output : model.generate("Tell me a joke.")) {
        System.out.print(output);
    }
    // Calculate a whole response before returning it.
    String response = model.complete("Tell me another one");
    // Returns the hidden representation of the context + prompt.
    float[] embedding = model.embed("Embed this");
}

Note

Since llama.cpp allocates memory that can't be garbage collected by the JVM, LlamaModel is implemented as an AutoClosable. If you use the objects with try-with blocks like the examples, the memory will be automatically freed when the model is no longer needed. This isn't strictly required, but avoids memory leaks if you use different models throughout the lifecycle of your application.

Infilling

You can simply pass prefix and suffix to generate() or complete().

Model/Inference Configuration

There are two sets of parameters you can configure, ModelParameters and InferenceParameters. Both provide builder classes to ease configuration. All non-specified options have sensible defaults.

ModelParameters modelParams = new ModelParameters()
                            .setLoraAdapter("/path/to/lora/adapter")
                            .setLoraBase("/path/to/lora/base");
InferenceParameters inferParams = new InferenceParameters()
		.setGrammar(new File("/path/to/grammar.gbnf"))
        .setTemperature(0.8);
LlamaModel model = new LlamaModel("/path/to/model.bin", modelParams);
model.generate(prompt, inferParams)

Logging

Both Java and C++ logging can be configured via the static method LlamaModel.setLogger:

// The method accepts a BiConsumer<LogLevel, String>.
LlamaModel.setLogger((level, message) -> System.out.println(level.name() + ": " + message));
// To completely silence any output, pass a no-op.
LlamaModel.setLogger((level, message) -> {});

// Similarly, a progress callback can be set (only the C++ side will call this).
// I think this is only used to report progress loading the model with a value of 0-1.
// It is thus state specific and can be done via the parameters.
new ModelParameters()
        .setProgressCallback(progress -> System.out.println("progress: " + progress));

cestella/java-llama.cpp