mkshing's picture
  - ja
  - japanese-stablelm
  - causal-lm
pipeline_tag: text-generation
license: apache-2.0
  Name: text
  Email: text
  Country: text
  Organization or Affiliation: text
  I allow Stability AI to contact me about information related to its models and research: checkbox

Japanese StableLM-3B-4E1T Instruct

Model Description

This is a 3B-parameter decoder-only Japanese language model fine-tuned on instruction-following datasets, built on top of the base model Japanese StableLM-3B-4E1T Base.

If you are in search of a larger model, please check Japanese Stable LM Instruct Gamma 7B.


import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("stabilityai/japanese-stablelm-3b-4e1t-instruct")
model = AutoModelForCausalLM.from_pretrained(

if torch.cuda.is_available():
    model ="cuda")

def build_prompt(user_query, inputs="", sep="\n\n### "):
    sys_msg = "ไปฅไธ‹ใฏใ€ใ‚ฟใ‚นใ‚ฏใ‚’่ชฌๆ˜Žใ™ใ‚‹ๆŒ‡็คบใจใ€ๆ–‡่„ˆใฎใ‚ใ‚‹ๅ…ฅๅŠ›ใฎ็ต„ใฟๅˆใ‚ใ›ใงใ™ใ€‚่ฆๆฑ‚ใ‚’้ฉๅˆ‡ใซๆบ€ใŸใ™ๅฟœ็ญ”ใ‚’ๆ›ธใใชใ•ใ„ใ€‚"
    p = sys_msg
    roles = ["ๆŒ‡็คบ", "ๅฟœ็ญ”"]
    msgs = [": \n" + user_query, ": \n"]
    if inputs:
        roles.insert(1, "ๅ…ฅๅŠ›")
        msgs.insert(1, ": \n" + inputs)
    for role, msg in zip(roles, msgs):
        p += sep + role + msg
    return p

# Infer with prompt without any additional input
user_inputs = {
    "user_query": "ไธŽใˆใ‚‰ใ‚ŒใŸใ“ใจใ‚ใ–ใฎๆ„ๅ‘ณใ‚’ๅฐๅญฆ็”Ÿใงใ‚‚ๅˆ†ใ‹ใ‚‹ใ‚ˆใ†ใซๆ•™ใˆใฆใใ ใ•ใ„ใ€‚",
    "inputs": "ๆƒ…ใ‘ใฏไบบใฎใŸใ‚ใชใ‚‰ใš"
prompt = build_prompt(**user_inputs)

input_ids = tokenizer.encode(

tokens = model.generate(,

out = tokenizer.decode(tokens[0][input_ids.shape[1]:], skip_special_tokens=True).strip()

Model Details

  • Developed by: Stability AI
  • Model type: Japanese StableLM-3B-4E1T Instruct model is an auto-regressive language model based on the transformer decoder architecture.
  • Language(s): Japanese
  • License: This model is licensed under Apache License, Version 2.0.
  • Contact: For questions and comments about the model, please join Stable Community Japan. For future announcements / information about Stability AI models, research, and events, please follow

Model Architecture

The model is a decoder-only transformer similar to the LLaMA (Touvron et al., 2023) architecture with the following modifications:

Parameters Hidden Size Layers Heads Sequence Length
2,795,443,200 2560 32 32 4096

Training Datasets

Use and Limitations

Intended Use

The model is intended to be used by all individuals as a foundational model for application-specific fine-tuning without strict limitations on commercial use.

Limitations and bias

The pre-training dataset may have contained offensive or inappropriate content even after applying data cleansing filters which can be reflected in the model-generated text. We recommend users exercise reasonable caution when using these models in production systems. Do not use the model for any applications that may cause harm or distress to individuals or groups.


The fine-tuning was carried out by Fujiki Nakamura. Other aspects, including data preparation and evaluation, were handled by the Language Team of Stability AI Japan, notably Meng Lee, Makoto Shing, Paul McCann, Naoki Orii, and Takuya Akiba.


We are grateful for the contributions of the EleutherAI Polyglot-JA team in helping us to collect a large amount of pre-training data in Japanese. Polyglot-JA members includes Hyunwoong Ko (Project Lead), Fujiki Nakamura (originally started this project when he commited to the Polyglot team), Yunho Mo, Minji Jung, KeunSeok Im, and Su-Kyeong Jang.

We are also appreciative of AI Novelist/Sta (Bit192, Inc.) and the numerous contributors from Stable Community Japan for assisting us in gathering a large amount of high-quality Japanese textual data for model training.