Mistral-Nemo-Japanese-Instruct-2408 Open-source Model - A Practical Choice for Focused Japanese Text Generation

Mistral Nemo Japanese Instruct 2408

Developed by cyberagent

This is a Japanese continuous pre-trained model based on Mistral-Nemo-Instruct-2407, focusing on Japanese text generation tasks.

Large Language Model

Safetensors

Supports Multiple LanguagesOpen Source License:Apache-2.0 #Japanese instruction fine-tuning #Dialogue generation #ChatML format

Downloads 1,898

Release Time : 8/30/2024

Model Overview

This model is a variant of Mistral-Nemo-Instruct-2407 that has undergone continuous pre-training in Japanese and is suitable for various Japanese text generation tasks, such as dialogue generation and question-answering.

Model Features

Japanese optimization

Specifically optimized for Japanese through continuous pre-training, enhancing Japanese text generation capabilities.

Dialogue-friendly

Supports the ChatML format, suitable for building dialogue systems.

Efficient inference

Supports streaming output, suitable for real-time interaction scenarios.

Model Capabilities

Japanese text generation

Dialogue system construction

Question-answering system

Content creation

Use Cases

Dialogue system

AI assistant

Build a Japanese AI assistant to provide a natural and smooth dialogue experience.

Capable of generating smooth responses that conform to Japanese language habits.

Content creation

Japanese article generation

Generate articles or paragraphs that conform to Japanese expression habits.

The generated content is natural and smooth, conforming to Japanese grammar rules.

🚀 Mistral-Nemo-Japanese-Instruct-2408

This is a Japanese continually pre-trained model. It aims to enhance text - generation capabilities in Japanese, leveraging the foundation of the Mistral architecture.

🚀 Quick Start

Make sure to update your transformers installation via pip install --upgrade transformers.

💻 Usage Examples

Basic Usage

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer

model = AutoModelForCausalLM.from_pretrained("cyberagent/Mistral-Nemo-Japanese-Instruct-2408", device_map="auto", torch_dtype="auto")
tokenizer = AutoTokenizer.from_pretrained("cyberagent/Mistral-Nemo-Japanese-Instruct-2408")
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)

messages = [
    {"role": "system", "content": "あなたは親切なAIアシスタントです。"},
    {"role": "user", "content": "AIによって私たちの暮らしはどのように変わりますか？"}
]

input_ids = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
output_ids = model.generate(input_ids,
                            max_new_tokens=1024,
                            temperature=0.5,
                            streamer=streamer)

📚 Documentation

Prompt Format

ChatML Format

<s><|im_start|>system
あなたは親切なAIアシスタントです。<|im_end|>
<|im_start|>user
AIによって私たちの暮らしはどのように変わりますか？<|im_end|>
<|im_start|>assistant

📄 License

This model is licensed under the Apache - 2.0 license.

👨‍💻 Author

Ryosuke Ishigami

📖 How to cite

@misc{cyberagent-mistral-nemo-japanese-instruct-2408,
      title={Mistral-Nemo-Japanese-Instruct-2408},
      url={https://huggingface.co/cyberagent/Mistral-Nemo-Japanese-Instruct-2408},
      author={Ryosuke Ishigami},
      year={2024},
}

📦 Model Information

Property	Details
Pipeline Tag	Text Generation
Supported Languages	Japanese, English
Tags	Japanese, Mistral
Inference	False
Base Model	mistralai/Mistral-Nemo-Base-2407

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご