Release

HOME - Release

Models / Corpus / Tools

We have released the models and tools developed by LLM-jp.

Please cite or reference the resources on this page if you use them in your research or software development.
Users of this model are requested to use it appropriately and at their own responsibility, in accordance with the “Guideline for Ensuring the Appropriateness of Research & Development and Utilization of Artificial Intelligence-Related Technology” (Decision of the Artificial Intelligence Strategic Headquarters, December 19, 2025).

Open platforms

Models

Fine-tuned / Pre-trained / Multimodal

Corpora

Pre-training Corpora / Data / Construction Scripts

Tools

Tools Evaluation & Training Utilities

Fine-tuned Models

llm-jp-4-8b-thinking

llm-jp-4-32b-a3b-thinking

Pre-trained Models

llm-jp-4-8b-base

llm-jp-4-32b-a3b-base

Multi-model Models

LLM-jp-3-VILA-14B

Voice Models

LLM-jp-Moshi-v1

Corpora for Pre-training

LLM-jp-corpus v4.1

LLM-jp-midtraining-corpus v1

LLM-jp-midtraining-corpus v2

Evaluation and fine-tuning datasets

llm-jp-4-thinking-sft-data

llm-jp-4-8b-thinking-dpo-data

llm-jp-4-32b-a3b-
thinking-dpo-data

ac-self-inst

aya-ja-evol-inst

AnswerCarefully

magpie-sft-v1.0

databricks-dolly-15k-ja

oasst1-21k-ja

oasst2-33k-ja

hh-rlhf-12k-ja

Tools

Pre-training Corpus Building Scripts v2.0

Pre-training Corpus Building Scripts v1.0

Tokenizer

［Evaluation Tools］
 llm-jp-eval

［Evaluation Tools］
  llm-jp-judge

［Fine-tuning Script］
  trl-based　SFT

［Fine-tuning Script］
  trl-based　DPO

［Fine-tuning Script］
  Nemo-Aligner based
*Supports both SFT and DPO

Leaderboards (Weights & Biases)

llm-jp-eval

Open LLM leaderboard
（English benchmark)
evaluation results of Japanese LLM

VicunaQA Japanese
benchmark

All Models

Fine-tuned Models

LLM-jp4

llm-jp-4-8b-thinking

llm-jp-4-32b-a3b-thinking

LLM-jp-3.1

LLM-jp-3.1-8x13b-instruct4

LLM-jp-3.1-13b-instruct4

LLM-jp-3.1-1.8b-instruct4

LLM-jp-3

LLM-jp-3-8x13b-instruct3

LLM-jp-3-172B-instruct3

LLM-jp-3-172B-beta2-instruct2

(Access requires approval. Redistribution and certain uses are restricted.)

LLM-jp-3-172B-beta1-instruct

(Ditto)

LLM-jp-3-13B-instruct3

LLM-jp-3-8×1.8b-instruct3

LLM-jp-3-7.2B-instruct3

LLM-jp-3-3.7B-instruct3

LLM-jp-3-1.8B-instruct3

LLM-jp-3-980M-instruct3

LLM-jp-3-440M-instruct3

LLM-jp-3-150M-instruct3

LLM-jp-3-8x13b-instruct2

LLM-jp-3-172B-instruct2

LLM-jp-3-13B-instruct2

LLM-jp-3-8×1.8b-instruct2

LLM-jp-3-7.2B-instruct2

LLM-jp-3-3.7B-instruct2

LLM-jp-3-1.8B-instruct2

LLM-jp-3-980M-instruct2

LLM-jp-3-440M-instruct2

LLM-jp-3-150M-instruct2

LLM-jp-3-13B-instruct

LLM-jp-3-7.2B-instruct

LLM-jp-3-3.7B-instruct

LLM-jp-3-1.8B-instruct

13B v2.0

LLM-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

LLM-jp-13b-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

LLM-jp-13b-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

13B v1.1

LLM-jp-13b-dpo-lora-hh_rlhf_ja-v1.1

LLM-jp-13b-instruct-full-dolly_en-dolly_ja-ichikara_003_001-oasst_en-oasst_ja-v1.1

LLM-jp-13b-instruct-lora-dolly_en-dolly_ja-ichikara_003_001-oasst_en-oasst_ja-v1.1

13B v1.0

LLM-jp-13B-instruct-full-jaster-v1.0

LLM-jp-13B-instruct-full-jaster-dolly-oasst-v1.0

LLM-jp-13B-instruct-full-dolly-oasst-v1.0

LLM-jp-13B-instruct-lora-jaster-v1.0

LLM-jp-13B-instruct-lora-jaster-dolly-oasst-v1.0

LLM-jp-13B-instruct-lora-dolly-oasst-v1.0

Pre-trained Models

LLM-jp-4

llm-jp-4-8b-base

llm-jp-4-32b-a3b-base

LLM-jp-3.1

LLM-jp-3.1-8x13b

LLM-jp-3.1-13b

LLM-jp-3.1-1.8b

LLM-jp-3

LLM-jp-3-8x13b

LLM-jp-3-172B

(Access requires approval. Redistribution and certain uses are restricted.)

LLM-jp-3-172B-beta2

(Ditto)

LLM-jp-3-172B-beta1

(Ditto)

LLM-jp-3-13B

LLM-jp-3-8×1.8b

LLM-jp-3-7.2B

LLM-jp-3-3.7B

LLM-jp-3-1.8B

LLM-jp-3-980M

LLM-jp-3-440M

LLM-jp-3-150M

13B v2.0

LLM-jp-13B-v2.0

13B v1.1

LLM-jp-13B-v1.0

13B v1.0

LLM-jp-1.3B-v1.0

Multi-model Models

LLM-jp-3-VILA-14B

Voice Models

LLM-jp-Moshi-v1

Others

Release

Models / Corpus / Tools ​

llm-jp-4-8b-thinking

llm-jp-4-32b-a3b-thinking

LLM-jp-3.1-8x13b-instruct4

LLM-jp-3.1-13b-instruct4

LLM-jp-3.1-1.8b-instruct4

LLM-jp-3-8x13b-instruct3

LLM-jp-3-172B-instruct3

LLM-jp-3-172B-beta2-instruct2

LLM-jp-3-172B-beta1-instruct

LLM-jp-3-13B-instruct3

LLM-jp-3-8×1.8b-instruct3

LLM-jp-3-7.2B-instruct3

LLM-jp-3-3.7B-instruct3

LLM-jp-3-1.8B-instruct3

LLM-jp-3-980M-instruct3

LLM-jp-3-440M-instruct3

LLM-jp-3-150M-instruct3

LLM-jp-3-8x13b-instruct2

LLM-jp-3-172B-instruct2

LLM-jp-3-13B-instruct2

LLM-jp-3-8×1.8b-instruct2

LLM-jp-3-7.2B-instruct2

LLM-jp-3-3.7B-instruct2

LLM-jp-3-1.8B-instruct2

LLM-jp-3-980M-instruct2

LLM-jp-3-440M-instruct2

LLM-jp-3-150M-instruct2

LLM-jp-3-13B-instruct

LLM-jp-3-7.2B-instruct

LLM-jp-3-3.7B-instruct

LLM-jp-3-1.8B-instruct

LLM-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

LLM-jp-13b-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

LLM-jp-13b-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0

LLM-jp-13b-dpo-lora-hh_rlhf_ja-v1.1

LLM-jp-13b-instruct-full-dolly_en-dolly_ja-ichikara_003_001-oasst_en-oasst_ja-v1.1

LLM-jp-13b-instruct-lora-dolly_en-dolly_ja-ichikara_003_001-oasst_en-oasst_ja-v1.1

LLM-jp-13B-instruct-full-jaster-v1.0

LLM-jp-13B-instruct-full-jaster-dolly-oasst-v1.0

LLM-jp-13B-instruct-full-dolly-oasst-v1.0

LLM-jp-13B-instruct-lora-jaster-v1.0

LLM-jp-13B-instruct-lora-jaster-dolly-oasst-v1.0

LLM-jp-13B-instruct-lora-dolly-oasst-v1.0

llm-jp-4-8b-base

llm-jp-4-32b-a3b-base

LLM-jp-3.1-8x13b

LLM-jp-3.1-13b

LLM-jp-3.1-1.8b

LLM-jp-3-8x13b

LLM-jp-3-172B

LLM-jp-3-172B-beta2

LLM-jp-3-172B-beta1

LLM-jp-3-13B

LLM-jp-3-8×1.8b

LLM-jp-3-7.2B

LLM-jp-3-3.7B

LLM-jp-3-1.8B

LLM-jp-3-980M

LLM-jp-3-440M

LLM-jp-3-150M

LLM-jp-13B-v2.0

LLM-jp-13B-v1.0

LLM-jp-1.3B-v1.0

LLM-jp-3-VILA-14B

LLM-jp-Moshi-v1

LLM-jp Modern BERT

Models / Corpus / Tools