Llm Jp 3 8x13b Instruct3

🚀 llm-jp-3-8x13b-instruct3

LLM-jp-3 is a series of large language models developed by the Research and Development Center for Large Language Models at the National Institute of Informatics. This repository offers the llm-jp-3-8x13b-instruct3 model. For an overview of LLM-jp-3 models with different parameter sizes, refer to:

• LLM-jp-3 Pre-trained Models
• LLM-jp-3 Fine-tuned Models

Checkpoints format: Hugging Face Transformers

🚀 Quick Start

✨ Features

• Developed by the Research and Development Center for Large Language Models at the National Institute of Informatics.
• Supports multiple programming languages including C, C++, C#, Go, Java, JavaScript, Lua, PHP, Python, Ruby, Rust, Scala, and TypeScript.
• Suitable for text generation tasks.

📦 Installation

Required Libraries and Their Versions

• torch>=2.3.0
• transformers>=4.40.1
• tokenizers>=0.19.1
• accelerate>=0.29.3
• flash-attn>=2.5.8

💻 Usage Examples

Basic Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("llm-jp/llm-jp-3-8x13b-instruct3")
model = AutoModelForCausalLM.from_pretrained("llm-jp/llm-jp-3-8x13b-instruct3", device_map="auto", torch_dtype=torch.bfloat16)
chat = [
    {"role": "system", "content": "以下は、タスクを説明する指示です。要求を適切に満たす応答を書きなさい。"},
    {"role": "user", "content": "自然言語処理とは何か"},
]
tokenized_input = tokenizer.apply_chat_template(chat, add_generation_prompt=True, tokenize=True, return_tensors="pt").to(model.device)
with torch.no_grad():
    output = model.generate(
        tokenized_input,
        max_new_tokens=100,
        do_sample=True,
        top_p=0.95,
        temperature=0.7,
        repetition_penalty=1.05,
    )[0]
print(tokenizer.decode(output))

📚 Documentation

Model Details

If you want to learn more about the pretraining of the LLM-jp-3 MoE series, refer to this blog post.

Tokenizer

The tokenizer of this model is based on huggingface/tokenizers Unigram byte-fallback model. The vocabulary entries were converted from llm-jp-tokenizer v3.0. For details on the vocabulary construction procedure (the pure SentencePiece training does not reproduce our vocabulary), refer to README.md of llm-jp-tokenizer.

Datasets

Pre-training

The models were pre-trained using a combination of the following datasets:

Post-training

We fine-tuned the pre-trained checkpoint with supervised fine-tuning and further aligned it with Direct Preference Optimization.

Supervised Fine-tuning

The datasets used for supervised fine-tuning are as follows:

Direct Preference Optimization

The datasets used for supervised fine-tuning are as follows:

Evaluation

llm-jp-eval (v1.4.1)

We evaluated the models using 100 examples from the dev split. Note that we skipped the CG (Code Generation) task.

Japanese MT Bench

We evaluated the models using gpt-4o-2024-08-06. The scores represent the average values obtained from five rounds of inference and evaluation. For more details, refer to the codes.

AnswerCarefully-Eval

AnswerCarefully-Eval assesses the safety of Japanese language model outputs using the LLM-as-a-Judge approach, based on the test set from llm-jp/AnswerCarefully. We evaluated the models using gpt-4-0613. The scores represent the average values obtained from five rounds of inference and evaluation.

🔧 Technical Details

The models released here are in the early stages of our research and development and have not been tuned to ensure outputs align with human intent and safety considerations.

📄 License

Apache License, Version 2.0

How to cite

If you find our work helpful, please feel free to cite the paper.

@inproceedings{
    nakamura2025dropupcycling,
    title={Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization},
    author={Taishi Nakamura and Takuya Akiba and Kazuki Fujii and Yusuke Oda and Rio Yokota and Jun Suzuki},
    booktitle={The Thirteenth International Conference on Learning Representations},
    year={2025},
    url={https://openreview.net/forum?id=gx1wHnf5Vp}
}

Model Card Authors

The names are listed in alphabetical order.

Hirokazu Kiyomaru, Takashi Kodama and Taishi Nakamura.

Send Questions to

llm-jp(at)nii.ac.jp