llmack

GoLang Fullstack LLM Framework

Overview

The GoLang Fullstack LLM Framework is a comprehensive toolkit designed to facilitate the development of applications leveraging Large Language Models (LLMs). This framework integrates various components such as LLM interaction, prompt management, Retrieval-Augmented Generation (RAG), speech processing, programmatic control, optimization, and an engine to orchestrate these components seamlessly.

Features

• LLM Integration: Interact with various Large Language Models (e.g., GPT, Claude) for text generation, summarization, and more.
• Prompt Management: Easily manage and optimize prompts for different tasks and models.
• Retrieval-Augmented Generation (RAG): Combine retrieval-based methods with generative models to enhance response quality.
• Speech Processing: Convert speech to text and vice versa, enabling voice-based interactions.
• Programmatic Control: Define and execute complex workflows programmatically.
• Optimizer: Fine-tune model parameters and prompts for better performance.
• Engine: A core engine to manage and orchestrate all components efficiently.

Usage

To use the GoLang Fullstack LLM Framework, ensure you have Go installed on your system. Then, run the following command:

go get github.com/showntop/llmack

LLM Integration

package main

import (
	"context"
	"fmt"
	"os"

	"github.com/showntop/llmack/llm"
	"github.com/showntop/llmack/llm/zhipu"
)

func main() {
	ctx := context.Background()
	llm.WithSingleConfig(map[string]any{
		"api_key": os.Getenv("zhipu_api_key"),
	})

	resp, err := llm.NewInstance(zhipu.Name).Invoke(ctx, []llm.Message{
		llm.UserPromptMessage("hello"),
	}, []llm.PromptMessageTool{}, llm.WithModel("GLM-4-Flash"))
	if err != nil {
		panic(err)
	}
	fmt.Println(resp.Result())
}

More examples can be found in the example directory.

Contributing

We welcome contributions! Please see our CONTRIBUTING.md for more details on how to get started.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Thanks to the Go community for their excellent tools and libraries.
• Special thanks to the developers of the various LLMs and related technologies that make this framework possible.

Contact

For any questions or feedback, please reach out to us at rongtaoxiao@gmail.com.

This README provides a basic overview of the GoLang Fullstack LLM Framework. For more detailed documentation, please refer to the individual package documentation within the repository.