# Japanese optimization

LLM-jp-3.1-1.8b is a large language model developed by the National Institute of Informatics in Japan. Based on the LLM-jp-3 series, it incorporates instruction pre-training to enhance the instruction-following ability.

Stockmark-2-100B is a 100-billion parameter large language model focused on Japanese capabilities, pre-trained on 1.5 trillion tokens of multilingual data and enhanced with Japanese synthetic data for improved instruction following.

A Japanese-optimized model fine-tuned based on Meta AI's Llama 3.2, supporting multilingual text generation

A Japanese-optimized large language model built on Meta Llama 3, enhancing Japanese capabilities through continuous pre-training and improving instruction-following abilities through supervised fine-tuning.

A Japanese-optimized large language model fine-tuned based on Meta-Llama-3-8B-Instruct, excelling in multiple Japanese benchmark tests

A large Japanese - enhanced language model built on Meta Llama 3, which improves Japanese processing capabilities through continuous pre - training and instruction fine - tuning.

Fugaku-LLM is a domestically developed model pre-trained from scratch entirely on the supercomputer 'Fugaku', primarily using Japanese training data, excelling in Japanese language processing

Fugaku-LLM is a domestically produced large language model in Japan, pre-trained from scratch using the supercomputer 'Fugaku.' It boasts high transparency and security, with particularly outstanding performance in Japanese.

Japanese-enhanced large language model continuously pre-trained based on Mistral-7B-v0.1

This is a CTranslate2 converted version of the OpenAI Whisper large-v3 model, specifically fine-tuned for Japanese with an additional 4,000 training steps, supporting multilingual speech recognition.

A Japanese text embedding model created by merging E5 Mistral 7B Instruct and Japanese StableLM Base Gamma 7B, focusing on sentence similarity tasks

A Japanese-optimized model based on Llama 2, designed specifically to enhance the Japanese interaction experience.

ELYZA-japanese-Llama-2-13b is a model based on Llama 2 with additional pre-training to enhance Japanese language capabilities.

CyberAgentLM2 is a decoder-only language model pre-trained on 1.3 trillion tokens of publicly available Japanese and English datasets.

A large language model that conducts continuous pre-training on a Japanese-English mixed dataset based on llama2-7b, with significantly improved performance on Japanese tasks.

A language model based on Llama2 architecture with extended Japanese capabilities through additional pre-training

A language model based on the Llama2 architecture with extended Japanese capabilities through additional pre-training