metadata

license: apache-2.0
tags:
  - japanese-stablelm
  - causal-lm
pipeline_tag: text-generation
datasets:
  - wikipedia
  - mc4
  - cc100
  - oscar-corpus/OSCAR-2301
  - oscar-corpus/OSCAR-2201
  - cerebras/SlimPajama-627B
language:
  - ja

Japanese StableLM-3B-4E1T Base

Model Description

This is a 3B-parameter decoder-only language model with a focus on maximizing Japanese language modeling performance and Japanese downstream task performance. We conducted continued pretraining using Japanese data on the English language model, StableLM-3B-4E1T, to transfer the model's knowledge and capabilities to Japanese.

If you are looking for an instruction-following model, check Japanese StableLM-3B-4E1T Instruct.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("stabilityai/japanese-stablelm-3b-4e1t-base")
model = AutoModelForCausalLM.from_pretrained(
  "stabilityai/japanese-stablelm-3b-4e1t-base",
  trust_remote_code=True,
  torch_dtype="auto",
)
model.cuda()
inputs = tokenizer("AI で科学研究を加速するには、", return_tensors="pt").to("cuda")
tokens = model.generate(
  **inputs,
  max_new_tokens=64,
  temperature=0.75,
  top_p=0.95,
  do_sample=True,
)
print(tokenizer.decode(tokens[0], skip_special_tokens=True))

Model Details

Developed by: Stability AI
Model type: Japanese StableLM-3B-4E1T Base model is an auto-regressive language models 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 email lm@stability.ai

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

Position Embeddings: Rotary Position Embeddings (Su et al., 2021) applied to the first 25% of head embedding dimensions for improved throughput following Black et al. (2022).
Normalization: LayerNorm (Ba et al., 2016) with learned bias terms as opposed to RMSNorm (Zhang & Sennrich, 2019).
Tokenizer: GPT-NeoX (Black et al., 2022).

Training Dataset

Around 100B tokens from a mixture of the following corpora were used for the continued pretraining.

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.

Acknowledgements

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.