diff --git a/README.md b/README.md
index e45005d..6b74460 100644
--- a/README.md
+++ b/README.md
@@ -37,17 +37,17 @@
| | アーキテクチャ | 入出力で扱える
トークン数 | 学習テキスト | 開発元 | ライセンス / 利用規約 |
|:---|:---:|:---:|:---:|:---:|:---:|
| [Sarashina2-8x70B](https://www.sbintuitions.co.jp/news/press/20241108_01/) | Mixtral
([8x70b (**465b**)](https://huggingface.co/sbintuitions/sarashina2-8x70b)) | 8,192 | Sarashina2 (70B) に対して Sparse Upcycling で学習 | SB Intuitions | Sarashina Model NonCommercial License |
-| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(計 **2.1T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: 合成データ | 大規模言語モデル研究開発センター (LLMC) | 事前学習済みモデル: LLM-jp-3 172B Terms of Use
事後学習済みモデル: llm-jp-3-172b-instruct3利用許諾契約 |
-| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(計 **1.4T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | 大規模言語モデル研究開発センター (LLMC) | LLM-jp-3 172B beta2 Terms of Use |
-| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(計 **0.7T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | 大規模言語モデル研究開発センター (LLMC) | LLM-jp-3 172B beta1 Terms of Use |
-| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(alpha1: 計 **0.7T** トークン, alpha2: 計 **1.4T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | 大規模言語モデル研究開発センター (LLMC) | Apache 2.0 |
+| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(計 **2.1T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: 合成データ | 大規模言語モデル研究開発センター | 事前学習済みモデル: LLM-jp-3 172B Terms of Use
事後学習済みモデル: llm-jp-3-172b-instruct3利用許諾契約 |
+| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(計 **1.4T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | 大規模言語モデル研究開発センター | LLM-jp-3 172B beta2 Terms of Use |
+| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(計 **0.7T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | 大規模言語モデル研究開発センター | LLM-jp-3 172B beta1 Terms of Use |
+| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)の一部
(alpha1: 計 **0.7T** トークン, alpha2: 計 **1.4T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | 大規模言語モデル研究開発センター | Apache 2.0 |
| [Stockmark-100b](https://stockmark.co.jp/news/20240516) | Llama
([**100b**](https://huggingface.co/stockmark/stockmark-100b), [**100b**-instruct-v0.1](https://huggingface.co/stockmark/stockmark-100b-instruct-v0.1)) | 4,096 | 事前学習: RedPajama, 日本語 Wikipedia, Japanese mC4, Japanese CommonCrawl, 日本語特許, Stockmark Web Corpus
(計 **910B** トークン)
Instruction Tuning (LoRA): [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/) | ストックマーク | MIT |
-| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | 事前学習: Japanese CommonCrawl, RefinedWeb, 独自のデータセット
(計: **2.0T** トークン) | Preferred Elements | PLaMo Non-Commercial License |
+| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | 事前学習: Japanese CommonCrawl, RefinedWeb, 独自のデータセット
(計: **2.0T** トークン) | Preferred Elements (Preferred Networks) | PLaMo Non-Commercial License |
| [Sarashina2](https://www.sbintuitions.co.jp/news/press/20240614_01/) | Llama
([**7b**](https://huggingface.co/sbintuitions/sarashina2-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina2-13b), [**70b**](https://huggingface.co/sbintuitions/sarashina2-70b)) | 7b, 13b: 4,096
70b: 8,192 | 事前学習: Japanese Common Crawl, SlimPajama, StarCoder
(計 **2.1T** トークン) | SB Intuitions | MIT |
| [Sarashina1](https://www.sbintuitions.co.jp/news/press/20240614_01/) | GPT-NeoX
([**7b**](https://huggingface.co/sbintuitions/sarashina1-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina1-13b), [**65b**](https://huggingface.co/sbintuitions/sarashina1-65b)) | 2,048 | 事前学習: Japanese Common Crawl
(計 **1T** トークン) | SB Intuitions | MIT |
| [Tanuki-8×8B](https://weblab.t.u-tokyo.ac.jp/2024-08-30/) | Tanuki (MoE) (**47b**)
([v1.0](https://huggingface.co/weblab-GENIAC/Tanuki-8x8B-dpo-v1.0), [v1.0-AWQ](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-AWQ), [v1.0-GPTQ-4bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-4bit), [v1.0-GPTQ-8bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-8bit), [v1.0-GGUF](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GGUF)) | 4,096 | 事前学習: 様々な Web 上のデータ, 合成データ（計 **1.7T** トークン）
SFT, DPO: 様々な合成データ [^19] | 松尾研LLM開発プロジェクト | Apache 2.0 |
| [CyberAgentLM3 (CALM3)](https://www.cyberagent.co.jp/news/detail/id=30463) | Llama
([**22b**-chat](https://huggingface.co/cyberagent/calm3-22b-chat)) | **16,384** | 不明
(計 **2.0T** トークン) | サイバーエージェント | Apache 2.0 |
-| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(計 **2.1T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | 大規模言語モデル研究開発センター (LLMC) | Apache 2.0 |
+| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | 事前学習: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(計 **2.1T** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | 大規模言語モデル研究開発センター | Apache 2.0 |
| [llm-jp-3-3.7b-instruct-EZO](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common) | Llama
([**3.7b**-instruct-EZO-Common](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common), [**3.7b**-instruct-EZO-Humanities](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Humanities)) | 4,096 | LLM-jp-3 (3.7B) に対して追加学習 | Axcxept | Apache 2.0 |
| [LLM-jp-13B v2.0](https://www.nii.ac.jp/news/release/2024/0430.html) | Llama
([**13b**-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-v2.0), [**13b**-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0)) | 4,096 | 事前学習: [llm-jp-corpus-v2](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v2)
(計 **260B** トークン)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2 | LLM-jp | Apache 2.0 |
| [Fugaku-LLM](https://pr.fujitsu.com/jp/news/2024/05/10.html) | GPT
([**13B**](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B), [**13B**-instruct](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct), [**13B**-instruct-gguf](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct-gguf)) | 2,048 | 事前学習: 独自
Instruction Tuning: OASST1, Dolly Dataset, GSM8K | 東工大, 東北大, 富士通, 理研, 名大, サイバーエージェント, Kotoba Technologies | Fugaku-LLM Terms of Use |
@@ -330,7 +330,7 @@
| | アーキテクチャ | 学習画像/テキスト | 開発元 | ライセンス / 利用規約 |
|:---|:---:|:---:|:---:|:---:|
| [llava-calm2-siglip](https://www.cyberagent.co.jp/news/detail/id=30344)
([llava-calm2-siglip](https://huggingface.co/cyberagent/llava-calm2-siglip)) | LLaVA-1.5 | MS-COCO と VisualGenome から生成された対話データ | サイバーエージェント | Apache 2.0 |
-| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | 大規模言語モデル研究開発センター (LLMC) | Apache 2.0 & OpenAI Terms of Use |
+| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | 大規模言語モデル研究開発センター | Apache 2.0 & OpenAI Terms of Use |
| [Heron](https://github.com/turingmotors/heron/blob/main/docs/README_JP.md)
([blip-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v0), [blip-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1), [blip-ja-stablelm-base-7b-v1-llava-620k](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1-llava-620k), [git-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v0), [git-ELYZA-fast-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ELYZA-fast-7b-v0), [git-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v1)) | BLIP-2 または GIT | v1: LLaVA-Instruct-150K-JA または LLaVA-Instruct-620K-JA
v0: LLaVA-Instruct-150K-JA, Japanese STAIR Captions, Japanese Visual Genome VQA dataset | Turing | CC BY-NC 4.0 |
| [Japanese Stable VLM](https://ja.stability.ai/blog/japanese-stable-vlm)
([japanese-stable-vlm](https://huggingface.co/stabilityai/japanese-stable-vlm)) | LLaVA-1.5 | Japanese CC12M, STAIR Captions, Japanese Visual Genome VQA dataset | Stability AI | STABILITY AI JAPANESE STABLE VLM COMMUNITY LICENSE |
| [Japanese InstructBLIP Alpha](https://ja.stability.ai/blog/japanese-instructblip-alpha)
([japanese-instructblip-alpha](https://huggingface.co/stabilityai/japanese-instructblip-alpha)) | InstructBLIP | Japanese CC12M, STAIR Captions, Japanese Visual Genome VQA dataset | Stability AI | JAPANESE STABLELM RESEARCH LICENSE |
@@ -442,8 +442,9 @@
| | 説明 | 開発元 |
|:---|:---|:---:|
| [Japanese MT-bench](https://github.com/Stability-AI/FastChat/tree/jp-stable/fastchat/llm_judge) | マルチターン会話能力を問う [MT-bench](https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge) の日本語版。Writing, Roleplay, Reasoning, Math, Coding, Extraction, STEM, Humanities の 8 つのカテゴリから 10 問ずつ、計 80 問が収録されている。なお、日本語版作成の際には、日本の文化に合うように質問内容に一部修正が加えられている。
GPT-4 による 10 段階の絶対評価を行うスクリプトも含まれている。 | Stability AI |
-| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | 日本の地理、歴史、政治、社会に関する[40問の自由質問](https://huggingface.co/datasets/yuzuai/rakuda-questions)に対してモデルに出力を行わせる。GPT-4 が同じ質問に対する2つのモデルの出力を比べ、どちらの答えが優れているかを判断することにより、モデルのランク付けを行う。 | YuzuAI |
| [ELYZA-tasks-100](https://huggingface.co/datasets/elyza/ELYZA-tasks-100) | 複雑な指示・タスクを含む100件の日本語データで、全てのデータに対して評価観点がアノテーションされている。
要約を修正し修正箇所を説明するタスク、具体的なエピソードから抽象的な教訓を述べるタスク、ユーザーの意図を汲み役に立つAIアシスタントとして振る舞うタスク、場合分けを必要とする複雑な算数のタスク、未知の言語からパターンを抽出し日本語訳する高度な推論を必要とするタスク、複数の指示を踏まえた上でyoutubeの対話を生成するタスク、架空の生き物や熟語に関する生成・大喜利などの想像力が求められるタスクなどが含まれている。 | ELYZA |
+| [Preferred Generation Benchmark
(pfgen-bench)](https://github.com/pfnet-research/pfgen-bench) | 50 問の日本語圏特有の常識問題をもとに、LLMの日本語生成能力を Fluency(流暢さ)、Truthfulness(真実性)、Helpfulness(有用性)の3つの評価軸から計測するベンチマーク。n-gram やルールベースでの指標の計算を行うことにより、LLM-as-a-Judge を行わずに評価を実施しているのが特徴である。 | Preferred Elements (Preferred Networks) |
+| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | 日本の地理、歴史、政治、社会に関する[40問の自由質問](https://huggingface.co/datasets/yuzuai/rakuda-questions)に対してモデルに出力を行わせる。GPT-4 が同じ質問に対する2つのモデルの出力を比べ、どちらの答えが優れているかを判断することにより、モデルのランク付けを行う。 | YuzuAI |
| [Japanese Vicuna QA Benchmark](https://github.com/ku-nlp/ja-vicuna-qa-benchmark) | MT-Bench の前身である [vicuna-blog-eval](https://github.com/lm-sys/vicuna-blog-eval) の日本語版。一般、知識、ロールプレイ、常識、フェルミ推定、反実仮想、コーディング、数学、ライティングに関する 80 問の質問を収録している。また、GPT-4 による自動評価（勝率計算）のスクリプトも含まれている。リーダーボードは[こちら](http://wandb.me/llm-jp-vicunaleaderboard) | 京大 言語メディア研究室 |
| [Tengu-Bench](https://huggingface.co/datasets/lightblue/tengu_bench) | 様々なカテゴリから成る 120 問の自由質問が収録されている。質問のカテゴリは以下の通り: 表の読み取り、論理パズル、アイデア生成、Function calling、長い文書要約（千トークン以上）、会話要約、長い文書のClosed QA（千トークン以上）、敬語、プロジェクト作成、数学、翻訳、抽出、倫理的制御、コスト見積、日本、雑談、ダジャレ、フォーマット、建設、ビジネス、法律判断、政治、架空の質問 | Lightblue |
| [Shaberi](https://github.com/lightblue-tech/japanese_llm_eval) | [Japanese MT-bench](#jp-mt-bench)、[Rakuda Benchmark](#rakuda-benchmark)、[ELYZA-tasks-100](#elyza-tasks)、[Tengu-Bench](#tengu-bench) の評価をまとめて行うことができるフレームワーク。なお、Shisa.AI による[フォーク](https://github.com/shisa-ai/shaberi)も存在する | Lightblue |
@@ -500,12 +501,14 @@
| | 説明 | 開発元 |
|:---|:---|:---:|
| [JMMMU](https://mmmu-japanese-benchmark.github.io/JMMMU/) | [MMMU ベンチマーク](https://mmmu-benchmark.github.io/)の日本語版として構築されたベンチマーク。720 件の MMMU の翻訳版の問題と 600 件の日本文化特有の新規の問題から構成される。 | 東大 相澤研 |
+| [JDocQA](https://github.com/mizuumi/JDocQA) | 日本語ドキュメント（パンフレット、スライド、レポート、Web サイト）をもとに構築された、合計 11,600 件の質問から構成される質問応答データセット。解答不能問題を含め、様々な質問形式の質問が収録されている。 | NAIST 渡辺研 |
| [Heron VLM リーダーボード powered by nejumi@WandB](https://api.wandb.ai/links/vision-language-leaderboard/h2lxge4n) | [Japanese-Heron-Bench](#japanese-heron-bench) と [LLaVA-Bench-In-the-Wild (Japanese)](#llava-bench-in-the-wild) の評価結果をまとめている。 | Turing, Weights & Biases |
| [Japanese-Heron-Bench](https://huggingface.co/datasets/turing-motors/Japanese-Heron-Bench) | 21 枚の画像に対して計 102 問の質問が割り当てられている。日本に関する知識を要求する画像・質問になっているのが特徴である。 | Turing |
| [JA-VLM-Bench-In-the-Wild](https://huggingface.co/datasets/SakanaAI/JA-VLM-Bench-In-the-Wild) | Sakana AI が EvoVLM-JP-v1-7B の評価のために独自に用意したデータセット。42 枚の画像に対して計 50 問の質問が割り当てられている。日本に関する知識を要求する画像・質問になっているのが特徴である。 | Sakana AI |
| [JA-Multi-Image-VQA](https://huggingface.co/datasets/SakanaAI/JA-Multi-Image-VQA) | 複数の画像に対する日本語での質疑応答能力を評価するデータセット。 | Sakana AI |
| [LLaVA-Bench-In-the-Wild (Japanese)](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-in-the-wild) | [LLaVA-Bench-In-the-Wild](https://huggingface.co/datasets/liuhaotian/llava-bench-in-the-wild) を DeepL で日本語に訳したもの。24 枚の画像に対して計 60 問の質問が割り当てられている。 | Turing |
| [LLaVA-Bench (COCO) Japanese](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-ja) | LLaVA の評価に使われた LLaVA-Bench (COCO) データセットを DeepL で日本語に訳したもの。30 枚の画像に対して各 3 種類の質問が割り当てられている。 | Turing |
+| [Japanese Visual Genome VQA dataset](https://github.com/yahoojapan/ja-vg-vqa) | [Visual Genome dataset](https://homes.cs.washington.edu/~ranjay/visualgenome/index.html) の画像をもとにアノテーションされた質問応答データセット。このデータセットの 500 件を切り出した [JA-VG-VQA-500](https://huggingface.co/datasets/SakanaAI/JA-VG-VQA-500) が VLM の評価ベンチマークとして用いられることもある。 | ヤフー |
## 各モデル・アーキテクチャの原論文
diff --git a/en/README.md b/en/README.md
index 789d551..7c2a907 100644
--- a/en/README.md
+++ b/en/README.md
@@ -36,17 +36,17 @@ Please point out any errors on the [issues page](https://github.com/llm-jp/aweso
| | Architecture | Max Context Length | Training Data | Developer | License / Terms of Use |
|:---|:---:|:---:|:---:|:---:|:---:|
| [Sarashina2-8x70B](https://www.sbintuitions.co.jp/news/press/20241108_01/) | Mixtral
([8x70b (**465b**)](https://huggingface.co/sbintuitions/sarashina2-8x70b)) | 8,192 | Sparse Upcycling on Sarashina2 (70B) | SB Intuitions | Sarashina Model NonCommercial License |
-| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: synthetic data | Research and Development Center for Large Language Models (LLMC) | Pre-trained model: LLM-jp-3 172B Terms of Use
Post-trained model: llm-jp-3-172b-instruct3 Terms of Use |
-| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models (LLMC) | LLM-jp-3 172B beta2 Terms of Use |
-| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/en/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**0.7T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models (LLMC) | LLM-jp-3 172B beta1 Terms of Use |
-| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(alpha1: **0.7T** tokens, alpha2: **1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 |
+| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: synthetic data | Research and Development Center for Large Language Models | Pre-trained model: LLM-jp-3 172B Terms of Use
Post-trained model: llm-jp-3-172b-instruct3 Terms of Use |
+| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models | LLM-jp-3 172B beta2 Terms of Use |
+| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/en/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**0.7T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models | LLM-jp-3 172B beta1 Terms of Use |
+| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(alpha1: **0.7T** tokens, alpha2: **1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models | Apache 2.0 |
| [Stockmark-100b](https://huggingface.co/stockmark/stockmark-100b) | Llama
([**100b**](https://huggingface.co/stockmark/stockmark-100b), [**100b**-instruct-v0.1](https://huggingface.co/stockmark/stockmark-100b-instruct-v0.1)) | 4,096 | Pre-training: RedPajama, Japanese Wikipedia, Japanese mC4, Japanese CommonCrawl, Japanese Patent, Stockmark Web Corpus
(**910B** tokens)
Instruction Tuning (LoRA): [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/) | Stockmark | MIT |
-| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | Pre-training: Japanese CommonCrawl, RefinedWeb, undisclosed
(**2.0T** tokens) | Preferred Elements | PLaMo Non-Commercial License |
+| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | Pre-training: Japanese CommonCrawl, RefinedWeb, undisclosed
(**2.0T** tokens) | Preferred Elements (Preferred Networks) | PLaMo Non-Commercial License |
| [Sarashina2](https://www.sbintuitions.co.jp/news/press/20240614_01/) | Llama
([**7b**](https://huggingface.co/sbintuitions/sarashina2-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina2-13b), [**70b**](https://huggingface.co/sbintuitions/sarashina2-70b)) | 7b, 13b: 4,096
70b: 8,192 | Pre-training: Japanese Common Crawl, SlimPajama, StarCoder
(**2.1T** tokens) | SB Intuitions | MIT |
| [Sarashina1](https://www.sbintuitions.co.jp/news/press/20240614_01/) | GPT-NeoX
([**7b**](https://huggingface.co/sbintuitions/sarashina1-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina1-13b), [**65b**](https://huggingface.co/sbintuitions/sarashina1-65b)) | 2,048 | Pre-training: Japanese Common Crawl
(**1T** tokens) | SB Intuitions | MIT |
| [Tanuki-8×8B](https://weblab.t.u-tokyo.ac.jp/2024-08-30/) | Tanuki (MoE) (**47b**)
([v1.0](https://huggingface.co/weblab-GENIAC/Tanuki-8x8B-dpo-v1.0), [v1.0-AWQ](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-AWQ), [v1.0-GPTQ-4bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-4bit), [v1.0-GPTQ-8bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-8bit), [v1.0-GGUF](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GGUF)) | 4,096 | Pre-training: various Web & synthetic datasets（**1.7T** tokens）
SFT, DPO: various synthetic datasets [^19] | Matsuo Lab LLM Development Project | Apache 2.0 |
| [CyberAgentLM3 (CALM3)](https://huggingface.co/cyberagent/calm3-22b-chat) | Llama
([**22b**-chat](https://huggingface.co/cyberagent/calm3-22b-chat)) | **16,384** | undisclosed
(**2.0T** tokens) | CyberAgent | Apache 2.0 |
-| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 |
+| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models | Apache 2.0 |
| [llm-jp-3-3.7b-instruct-EZO](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common) | Llama
([**3.7b**-instruct-EZO-Common](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common), [**3.7b**-instruct-EZO-Humanities](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Humanities)) | 4,096 | additionally trained on LLM-jp-3 (3.7B) | Axcxept | Apache 2.0 |
| [LLM-jp-13B v2.0](https://huggingface.co/llm-jp/llm-jp-13b-v2.0) | Llama
([**13b**-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-v2.0), [**13b**-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0)) | 4,096 | Pre-training: [llm-jp-corpus-v2](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v2)
(**260B** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2 | LLM-jp | Apache 2.0 |
| [Fugaku-LLM](https://www.fujitsu.com/global/about/resources/news/press-releases/2024/0510-01.html) | GPT
([**13B**](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B), [**13B**-instruct](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct), [**13B**-instruct-gguf](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct-gguf)) | 2,048 | Pre-training: undisclosed dataset
Instruction Tuning: OASST1, Dolly Dataset, GSM8K | Titech, Tohoku Univ., Fujitsu, RIKEN, Nagoya Univ., CyberAgent, Kotoba Technologies | Fugaku-LLM Terms of Use |
@@ -328,7 +328,7 @@ Please point out any errors on the [issues page](https://github.com/llm-jp/aweso
| | Architecture | Training Data | Developer | License / Terms of Use |
|:---|:---:|:---:|:---:|:---:|
| [llava-calm2-siglip](https://www.cyberagent.co.jp/news/detail/id=30344)
([llava-calm2-siglip](https://huggingface.co/cyberagent/llava-calm2-siglip)) | LLaVA-1.5 | coversational data generated from MS-COCO and VisualGenome | CyberAgent | Apache 2.0 |
-| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/en/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 & OpenAI Terms of Use |
+| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/en/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | Research and Development Center for Large Language Models | Apache 2.0 & OpenAI Terms of Use |
| [Heron](https://github.com/turingmotors/heron)
([blip-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v0), [blip-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1), [blip-ja-stablelm-base-7b-v1-llava-620k](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1-llava-620k), [git-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v0), [git-ELYZA-fast-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ELYZA-fast-7b-v0), [git-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v1)) | BLIP-2 / GIT | v1: LLaVA-Instruct-150K-JA or LLaVA-Instruct-620K-JA
v0: LLaVA-Instruct-150K-JA, Japanese STAIR Captions, Japanese Visual Genome VQA dataset | Turing | CC BY-NC 4.0 |
| [Japanese Stable VLM](https://ja.stability.ai/blog/japanese-stable-vlm)
([japanese-stable-vlm](https://huggingface.co/stabilityai/japanese-stable-vlm)) | LLaVA-1.5 | Japanese CC12M, STAIR Captions, Japanese Visual Genome VQA dataset | Stability AI | STABILITY AI JAPANESE STABLE VLM COMMUNITY LICENSE |
| [Japanese InstructBLIP Alpha](https://ja.stability.ai/blog/japanese-instructblip-alpha)
([japanese-instructblip-alpha](https://huggingface.co/stabilityai/japanese-instructblip-alpha)) | InstructBLIP | Japanese CC12M, STAIR Captions, Japanese Visual Genome VQA dataset | Stability AI | JAPANESE STABLELM RESEARCH LICENSE |
@@ -440,8 +440,9 @@ Please point out any errors on the [issues page](https://github.com/llm-jp/aweso
| | Description | Developer |
|:---|:---|:---:|
| [Japanese MT-bench](https://github.com/Stability-AI/FastChat/tree/jp-stable/fastchat/llm_judge) | The Japanese version of [MT-bench](https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge) asks about multi-turn conversational ability. It includes 80 questions, 10 each, from 8 categories: Writing, Roleplay, Reasoning, Math, Coding, Extraction, STEM, Humanities. Some questions have been modified to fit with Japanese culture during the production of the Japanese version. It also includes a script that performs a 10-level absolute evaluation by GPT-4. | Stability AI |
-| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | Ranking based on model answers to [40 open-ended questions](https://huggingface.co/datasets/yuzuai/rakuda-questions) on Japanese geography, history, politics, and society. Uses GPT-4 to judge model outputs pairwise, and then ranks models by fitting a Maximum Likelihood Elo/Bradley-Terry model to GPT-4's preferences. | YuzuAI |
| [ELYZA-tasks-100](https://huggingface.co/datasets/elyza/ELYZA-tasks-100) | Ranking based on model responses to [100 complex and diverse tasks](https://huggingface.co/datasets/elyza/ELYZA-tasks-100), including tasks testing summarization, correction, abstraction, induction, and other skills. Uses humans to score the model responses and then ranks models based on their mean scores. | ELYZA |
+| [Preferred Generation Benchmark
(pfgen-bench)](https://github.com/pfnet-research/pfgen-bench) | A benchmark to measure the Japanese language generation ability of LLMs based on 50 common sense questions unique to the Japanese context. It evaluates along three axes: Fluency, Truthfulness, and Helpfulness. The evaluation is conducted without using LLM-as-a-Judge by calculating n-gram or rule-based metrics. | Preferred Elements (Preferred Networks) |
+| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | Ranking based on model answers to [40 open-ended questions](https://huggingface.co/datasets/yuzuai/rakuda-questions) on Japanese geography, history, politics, and society. Uses GPT-4 to judge model outputs pairwise, and then ranks models by fitting a Maximum Likelihood Elo/Bradley-Terry model to GPT-4's preferences. | YuzuAI |
| [Japanese Vicuna QA Benchmark](https://github.com/ku-nlp/ja-vicuna-qa-benchmark) | This is the Japanese version of [vicuna-blog-eval](https://github.com/lm-sys/vicuna-blog-eval), which is the predecessor of MT-Bench. It includes 80 questions on general knowledge, role-playing, common sense, Fermi estimation, counterfactual thinking, coding, mathematics, and writing. It also includes a script for automatic evaluation by GPT-4 (win-rate calculation). The leaderboard can be found [here](http://wandb.me/llm-jp-vicunaleaderboard). | Kyoto University Language Media Processing Lab |
| [Tengu-Bench](https://huggingface.co/datasets/lightblue/tengu_bench) | Includes 120 free-form questions from various categories. Categories of questions: table interpretation, logic puzzles, idea generation, function calling, long document summarization (over a thousand tokens), conversation summarization, long document closed QA (over a thousand tokens), honorifics, project creation, math, translation, extraction, ethical control, cost estimation, Japan, chit-chat, puns, formatting, construction, business, legal judgment, politics, hypothetical questions. | Lightblue |
| [Shaberi](https://github.com/lightblue-tech/japanese_llm_eval) | A framework that can collectively evaluate the [Japanese MT-bench](#jp-mt-bench), [Rakuda Benchmark](#rakuda-benchmark), [ELYZA-tasks-100](#elyza-tasks), and [Tengu-Bench](#tengu-bench). There is also a [fork](https://github.com/shisa-ai/shaberi) by Shisa.AI. | Lightblue |
@@ -498,12 +499,14 @@ Please point out any errors on the [issues page](https://github.com/llm-jp/aweso
| | Description | Developer |
|:---|:---|:---:|
| [JMMMU](https://mmmu-japanese-benchmark.github.io/JMMMU/) | A benchmark constructed as the Japanese version of [MMMU Benchmark](https://mmmu-benchmark.github.io/). It consists of 720 translated MMMU problems and 600 new problems unique to Japanese culture. | University of Tokyo Aizawa Lab |
+| [JDocQA](https://github.com/mizuumi/JDocQA) | A question-answer dataset based on Japanese documents (pamphlets, slides, reports, websites), consisting of a total of 11,600 questions. It includes various question formats, including unanswerable questions. | NAIST Watanabe Lab |
| [Heron VLM Leaderboard powered by Nejumi/WandB](https://wandb.ai/vision-language-leaderboard/heron-leaderboard/reports/Heron-VLM-Leaderboard-powered-by-Nejumi-WandB--Vmlldzo4MjY3OTc5) | Summarizes the evaluation results of [Japanese-Heron-Bench](#japanese-heron-bench) and [LLaVA-Bench-In-the-Wild (Japanese)](#llava-bench-in-the-wild). | Turing, Weights & Biases |
| [Japanese-Heron-Bench](https://huggingface.co/datasets/turing-motors/Japanese-Heron-Bench) | 21 images are assigned a total of 102 questions. It is characterized by image-question pairs that require knowledge related to Japan. | Turing |
| [JA-VLM-Bench-In-the-Wild](https://huggingface.co/datasets/SakanaAI/JA-VLM-Bench-In-the-Wild) | A dataset independently prepared by Sakana AI to evaluate EvoVLM-JP-v1-7B. It consists of 50 questions assigned to 42 images. It is characterized by images and questions that require knowledge about Japan. | Sakana AI |
| [JA-Multi-Image-VQA](https://huggingface.co/datasets/SakanaAI/JA-Multi-Image-VQA) | A dataset for evaluating the question-answering ability in Japanese for multiple images. | Sakana AI |
| [LLaVA-Bench-In-the-Wild (Japanese)](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-in-the-wild) | This is the Japanese version of [LLaVA-Bench-In-the-Wild](https://huggingface.co/datasets/liuhaotian/llava-bench-in-the-wild), translated using DeepL. It consists of 60 questions assigned to 24 images. | Turing |
| [LLaVA-Bench (COCO) Japanese](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-ja) | This is the Japanese version, translated by DeepL, of the LLaVA-Bench (COCO) dataset used to evaluate LLaVA. It consists of 30 images, each with 3 types of questions assigned to them. | Turing |
+| [Japanese Visual Genome VQA dataset](https://github.com/yahoojapan/ja-vg-vqa) | A question-and-answer dataset annotated based on images from the [Visual Genome dataset](https://homes.cs.washington.edu/~ranjay/visualgenome/index.html). A subset of this dataset, [JA-VG-VQA-500](https://huggingface.co/datasets/SakanaAI/JA-VG-VQA-500), consisting of 500 questions, is sometimes used as a benchmark for evaluating VLMs. | Yahoo |
## References for Models and Architectures
diff --git a/fr/README.md b/fr/README.md
index 583c8cc..dbed508 100644
--- a/fr/README.md
+++ b/fr/README.md
@@ -36,17 +36,17 @@ N'hésitez pas à signaler les erreurs sur la page [issues](https://github.com/l
| | Architecture | Longueur Maximale du Contexte | Données d'entraînement | Développeur | Licence / Conditions d'utilisation |
|:---|:---:|:---:|:---:|:---:|:---:|
| [Sarashina2-8x70B](https://www.sbintuitions.co.jp/news/press/20241108_01/) | Mixtral
([8x70b (**465b**)](https://huggingface.co/sbintuitions/sarashina2-8x70b)) | 8,192 | Sparse Upcycling on Sarashina2 (70B) | SB Intuitions | Sarashina Model NonCommercial License |
-| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: synthetic data | Research and Development Center for Large Language Models (LLMC) | Pre-trained model: LLM-jp-3 172B Terms of Use
Post-trained model: llm-jp-3-172b-instruct3 Terms of Use |
-| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models (LLMC) | LLM-jp-3 172B beta2 Terms of Use |
-| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/en/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**0.7T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models (LLMC) | LLM-jp-3 172B beta1 Terms of Use |
-| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(alpha1: **0.7T** tokens, alpha2: **1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 |
+| [LLM-jp-3 172B](https://www.nii.ac.jp/news/release/2024/1224.html) | Llama
([**172b**](https://huggingface.co/llm-jp/llm-jp-3-172b), [**172b**-instruct3](https://huggingface.co/llm-jp/llm-jp-3-172b-instruct3)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k
DPO: synthetic data | Research and Development Center for Large Language Models | Pre-trained model: LLM-jp-3 172B Terms of Use
Post-trained model: llm-jp-3-172b-instruct3 Terms of Use |
+| [LLM-jp-3 172B beta2](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-beta2/) | Llama
([**172b**-beta2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2), [**172b**-beta2-instruct2](https://huggingface.co/llm-jp/llm-jp-3-172b-beta2-instruct2)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), [magpie-sft-v1.0](https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0), Daring-Anteater, FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft-ja, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models | LLM-jp-3 172B beta2 Terms of Use |
+| [LLM-jp-3 172B beta1](https://www.nii.ac.jp/en/news/release/2024/0917.html) | Llama
([**172b**-beta1](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1), [**172b**-beta1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-beta1-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**0.7T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models | LLM-jp-3 172B beta1 Terms of Use |
+| [LLM-jp-3 172B alpha](https://llmc.nii.ac.jp/en/topics/llm-jp-3-172b-alpha1-alpha2/) | Llama
([**172b**-alpha1](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1), [**172b**-alpha1-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha1-instruct), [**172b**-alpha2](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2), [**172b**-alpha2-instruct](https://huggingface.co/llm-jp/llm-jp-3-172b-alpha2-instruct)) | 4,096 | Pre-training: part of [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(alpha1: **0.7T** tokens, alpha2: **1.4T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2, Aya Dataset, ichikara-instruction-format, Daring-Anteater, FLAN | Research and Development Center for Large Language Models | Apache 2.0 |
| [Stockmark-100b](https://huggingface.co/stockmark/stockmark-100b) | Llama
([**100b**](https://huggingface.co/stockmark/stockmark-100b), [**100b**-instruct-v0.1](https://huggingface.co/stockmark/stockmark-100b-instruct-v0.1)) | 4,096 | Pre-training: RedPajama, Wikipedia en japonais, Japanese mC4, Japanese CommonCrawl, Japanese Patent, Stockmark Web Corpus
(**910B** tokens)
Instruction Tuning (LoRA): [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/) | Stockmark | MIT |
-| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | Pre-training: Japanese CommonCrawl, RefinedWeb, undisclosed
(**2.0T** tokens) | Preferred Elements | PLaMo Non-Commercial License |
+| [PLaMo-100B-Pretrained](https://www.preferred.jp/ja/news/pr20241015/) | Llama[^22]
([**100b**](https://huggingface.co/pfnet/plamo-100b)) | 4,096 | Pre-training: Japanese CommonCrawl, RefinedWeb, undisclosed
(**2.0T** tokens) | Preferred Elements (Preferred Networks) | PLaMo Non-Commercial License |
| [Sarashina2](https://www.sbintuitions.co.jp/news/press/20240614_01/) | Llama
([**7b**](https://huggingface.co/sbintuitions/sarashina2-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina2-13b), [**70b**](https://huggingface.co/sbintuitions/sarashina2-70b)) | 7b, 13b: 4,096
70b: 8,192 | Pre-training: Japanese Common Crawl, SlimPajama, StarCoder
(**2.1T** tokens) | SB Intuitions | MIT |
| [Sarashina1](https://www.sbintuitions.co.jp/news/press/20240614_01/) | GPT-NeoX
([**7b**](https://huggingface.co/sbintuitions/sarashina1-7b), [**13b**](https://huggingface.co/sbintuitions/sarashina1-13b), [**65b**](https://huggingface.co/sbintuitions/sarashina1-65b)) | 2,048 | Pre-training: Japanese Common Crawl
(**1T** tokens) | SB Intuitions | MIT |
| [Tanuki-8×8B](https://weblab.t.u-tokyo.ac.jp/2024-08-30/) | Tanuki (MoE) (**47b**)
([v1.0](https://huggingface.co/weblab-GENIAC/Tanuki-8x8B-dpo-v1.0), [v1.0-AWQ](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-AWQ), [v1.0-GPTQ-4bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-4bit), [v1.0-GPTQ-8bit](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GPTQ-8bit), [v1.0-GGUF](https://huggingface.co/team-hatakeyama-phase2/Tanuki-8x8B-dpo-v1.0-GGUF)) | 4,096 | Pre-training: various Web & synthetic datasets（**1.7T** tokens）
SFT, DPO: various synthetic datasets [^19] | Matsuo Lab LLM Development Project | Apache 2.0 |
| [CyberAgentLM3 (CALM3)](https://huggingface.co/cyberagent/calm3-22b-chat) | Llama
([**22b**-chat](https://huggingface.co/cyberagent/calm3-22b-chat)) | **16,384** | undisclosed
(**2.0T** tokens) | CyberAgent | Apache 2.0 |
-| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 |
+| [LLM-jp-3 13B](https://llmc.nii.ac.jp/topics/post-707/) | Llama
([**1.8b**](https://huggingface.co/llm-jp/llm-jp-3-1.8b), [**1.8b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-1.8b-instruct), [**3.7b**](https://huggingface.co/llm-jp/llm-jp-3-3.7b), [**3.7b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-3.7b-instruct), [**13b**](https://huggingface.co/llm-jp/llm-jp-3-13b), [**13b**-instruct](https://huggingface.co/llm-jp/llm-jp-3-13b-instruct)) | 4,096 | Pre-training: [llm-jp-corpus-v3](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v3)
(**2.1T** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), FLAN, ichikara-instruction-format, AutoMultiTurnByCalm3-22B, ramdom-to-fixed-multiturn-Calm3, wizardlm8x22b-logical-math-coding-sft_additional-ja, Synthetic-JP-EN-Coding-Dataset-567k | Research and Development Center for Large Language Models | Apache 2.0 |
| [llm-jp-3-3.7b-instruct-EZO](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common) | Llama
([**3.7b**-instruct-EZO-Common](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Common), [**3.7b**-instruct-EZO-Humanities](https://huggingface.co/AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Humanities)) | 4,096 | additionally trained on LLM-jp-3 (3.7B) | Axcxept | Apache 2.0 |
| [LLM-jp-13B v2.0](https://huggingface.co/llm-jp/llm-jp-13b-v2.0) | Llama
([**13b**-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-v2.0), [**13b**-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001-dolly-ichikara_004_001_single-oasst-oasst2-v2.0), [**13b**-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0](https://huggingface.co/llm-jp/llm-jp-13b-instruct-full-ac_001_16x-dolly-ichikara_004_001_single-oasst-oasst2-v2.0)) | 4,096 | Pre-training: [llm-jp-corpus-v2](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-corpus-v2)
(**260B** tokens)
Instruction Tuning: [ichikara-instruction](https://liat-aip.sakura.ne.jp/wp/llm%E3%81%AE%E3%81%9F%E3%82%81%E3%81%AE%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%82%A4%E3%83%B3%E3%82%B9%E3%83%88%E3%83%A9%E3%82%AF%E3%82%B7%E3%83%A7%E3%83%B3%E3%83%87%E3%83%BC%E3%82%BF%E4%BD%9C%E6%88%90/), [answer-carefully](https://liat-aip.sakura.ne.jp/wp/answercarefully-dataset/), Dolly Dataset, OASST1, OASST2 | LLM-jp | Apache 2.0 |
| [Fugaku-LLM](https://www.fujitsu.com/global/about/resources/news/press-releases/2024/0510-01.html) | GPT
([**13B**](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B), [**13B**-instruct](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct), [**13B**-instruct-gguf](https://huggingface.co/Fugaku-LLM/Fugaku-LLM-13B-instruct-gguf)) | 2,048 | Pre-training: undisclosed dataset
Instruction Tuning: OASST1, Dolly Dataset, GSM8K | Titech, Tohoku Univ., Fujitsu, RIKEN, Nagoya Univ., CyberAgent, Kotoba Technologies | Fugaku-LLM Terms of Use |
@@ -328,7 +328,7 @@ N'hésitez pas à signaler les erreurs sur la page [issues](https://github.com/l
| | Architecture | Données d'entraînement | Développeur | License / Terms of Use |
|:---|:---:|:---:|:---:|:---:|
| [llava-calm2-siglip](https://www.cyberagent.co.jp/news/detail/id=30344)
([llava-calm2-siglip](https://huggingface.co/cyberagent/llava-calm2-siglip)) | LLaVA-1.5 | coversational data generated from MS-COCO and VisualGenome | CyberAgent | Apache 2.0 |
-| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/en/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | Research and Development Center for Large Language Models (LLMC) | Apache 2.0 & OpenAI Terms of Use |
+| [LLM-jp-3 VILA 14B](https://llmc.nii.ac.jp/en/topics/llm-jp-3-vila-14b/)
([14b](https://huggingface.co/llm-jp/llm-jp-3-vila-14b)) | LLaVA-1.5 | [Japanese image text pairs](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-image-text-pairs), LLaVA-Pretrain, [Japanese interleaved data](https://gitlab.llm-jp.nii.ac.jp/datasets/llm-jp-japanese-interleaved-data), coyo (subset), mmc4-core (subset), [llava-instruct-ja](https://huggingface.co/datasets/llm-jp/llava-instruct-ja), [japanese-photos-conv](https://huggingface.co/datasets/llm-jp/japanese-photos-conversation), ja-vg-vqa, synthdog-ja, LLaVA-1.5 instruction data (subset) | Research and Development Center for Large Language Models | Apache 2.0 & OpenAI Terms of Use |
[Heron](https://github.com/turingmotors/heron)
([blip-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v0), [blip-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1), [blip-ja-stablelm-base-7b-v1-llava-620k](https://huggingface.co/turing-motors/heron-chat-blip-ja-stablelm-base-7b-v1-llava-620k), [git-ja-stablelm-base-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v0), [git-ELYZA-fast-7b-v0](https://huggingface.co/turing-motors/heron-chat-git-ELYZA-fast-7b-v0), [git-ja-stablelm-base-7b-v1](https://huggingface.co/turing-motors/heron-chat-git-ja-stablelm-base-7b-v1)) | BLIP-2 / GIT | v1: LLaVA-Instruct-150K-JA or LLaVA-Instruct-620K-JA
v0: LLaVA-Instruct-150K-JA, Japanese STAIR Captions, Japanese Visual Genome VQA dataset | Turing | CC BY-NC 4.0 |
| [Japanese Stable VLM](https://ja.stability.ai/blog/japanese-stable-vlm)
([japanese-stable-vlm](https://huggingface.co/stabilityai/japanese-stable-vlm)) | LLaVA-1.5 | Japanese CC12M, STAIR Captions, jeu de données Japanese Visual Genome VQA | Stability AI | STABILITY AI JAPANESE STABLE VLM COMMUNITY LICENSE |
| [Japanese InstructBLIP Alpha](https://ja.stability.ai/blog/japanese-instructblip-alpha)
([japanese-instructblip-alpha](https://huggingface.co/stabilityai/japanese-instructblip-alpha)) | InstructBLIP | Japanese CC12M, STAIR Captions, jeu de données Japanese Visual Genome VQA | Stability AI | JAPANESE STABLELM RESEARCH LICENSE |
@@ -440,8 +440,9 @@ N'hésitez pas à signaler les erreurs sur la page [issues](https://github.com/l
| | Description | Développeur |
|:---|:---|:---:|
| [Japanese MT-bench](https://github.com/Stability-AI/FastChat/tree/jp-stable/fastchat/llm_judge) | Version japonaise du [MT-bench](https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge) qui interroge sur la capacité à converser en plusieurs tournures. Il inclut 80 questions, 10 de chacune des 8 catégories : écriture, jeu de rôle, raisonnement, maths, codage, extraction, STEM, sciences humaines. Certaines questions ont été modifiées pour s'adapter à la culture japonaise lors de la création de la version japonaise. Il comprend également un script qui réalise une évaluation absolue en 10 niveaux par GPT-4. | Stability AI |
-| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | Classement basé sur les réponses des modèles avec [40 questions ouvertes](https://huggingface.co/datasets/yuzuai/rakuda-questions) la géographie, l'histoire, la politique, et la société japonaise. Utilise GPT-4 pour évaluer les résultats du modèle par paires, puis classe les modèles en ajustant le maximum de vraisemblance sur le modèle de probabilité d'Elo/Bradley-Terry avec les préférences de GPT-4. | YuzuAI |
| [ELYZA-tasks-100](https://huggingface.co/datasets/elyza/ELYZA-tasks-100) | Classement basé sur les réponses des modèles avec [100 tâches complexes et diverses](https://huggingface.co/datasets/elyza/ELYZA-tasks-100), y compris les tâches testant la synthèse, la correction, l'abstraction, l'induction et d'autres compétences. Utilise des humains pour noter les réponses du modèle, puis classe les modèles en fonction de leurs scores moyens. | ELYZA |
+| [Preferred Generation Benchmark
(pfgen-bench)](https://github.com/pfnet-research/pfgen-bench) | Un banc d'essai pour mesurer la capacité des LLMs à générer du texte en japonais basé sur 50 questions de bon sens uniques au contexte japonais. Il évalue selon trois axes : fluidité, véracité et utilité. L'évaluation est réalisée sans utiliser LLM-as-a-Judge en calculant des métriques basées sur des n-grammes ou des règles. | Preferred Elements (Preferred Networks) |
+| [Rakuda Benchmark](https://github.com/yuzu-ai/japanese-llm-ranking) | Classement basé sur les réponses des modèles avec [40 questions ouvertes](https://huggingface.co/datasets/yuzuai/rakuda-questions) la géographie, l'histoire, la politique, et la société japonaise. Utilise GPT-4 pour évaluer les résultats du modèle par paires, puis classe les modèles en ajustant le maximum de vraisemblance sur le modèle de probabilité d'Elo/Bradley-Terry avec les préférences de GPT-4. | YuzuAI |
| [Japanese Vicuna QA Benchmark](https://github.com/ku-nlp/ja-vicuna-qa-benchmark) | Il s'agit de la version japonaise de [vicuna-blog-eval](https://github.com/lm-sys/vicuna-blog-eval), qui est le précurseur de MT-Bench. Il comprend 80 questions sur la connaissance générale, le jeu de rôle, le bon sens, l'estimation de Fermi, la pensée contrefactuelle, le codage, les mathématiques, et l'écriture. Il comprend également un script pour une évaluation automatique par GPT-4 (calcul du taux de victoire). Le tableau de classement peut être trouvé [ici](http://wandb.me/llm-jp-vicunaleaderboard). | Université de Kyoto Laboratoire de traitement des langues et des médias |
| [Tengu-Bench](https://huggingface.co/datasets/lightblue/tengu_bench) | Comprend 120 questions ouvertes de diverses catégories. Catégories de questions : interprétation des tableaux, puzzles logiques, génération d'idées, appel de fonctions, résumé de longs documents (plus de mille jetons), résumé de conversations, questions fermées sur des longs documents (plus de mille jetons), honorifiques, création de projet, mathématiques, traduction, extraction, contrôle éthique, estimation des coûts, Japon, bavardage, calembours, formatage, construction, affaires, jugement juridique, politique, questions hypothétiques. | Lightblue |
| [Shaberi](https://github.com/lightblue-tech/japanese_llm_eval) | Un cadre qui peut évaluer collectivement le [Japanese MT-bench](#jp-mt-bench), le [Rakuda Benchmark](#rakuda-benchmark), le [ELYZA-tasks-100](#elyza-tasks), et le [Tengu-Bench](#tengu-bench). Il existe également un [fork](https://github.com/shisa-ai/shaberi) de Shisa.AI. | Lightblue |
@@ -498,12 +499,14 @@ N'hésitez pas à signaler les erreurs sur la page [issues](https://github.com/l
| | Description | Développeur |
|:---|:---|:---:|
| [JMMMU](https://mmmu-japanese-benchmark.github.io/JMMMU/) | Un benchmark construit comme la version japonaise du [MMMU Benchmark](https://mmmu-benchmark.github.io/). Il se compose de 720 problèmes traduits du MMMU et de 600 nouveaux problèmes uniques à la culture japonaise. | University of Tokyo Aizawa Lab |
+| [JDocQA](https://github.com/mizuumi/JDocQA) | Un jeu de données de questions-réponses basé sur des documents japonais (brochures, diapositives, rapports, sites web), comprenant un total de 11 600 questions. Il inclut divers formats de questions, y compris des questions non répondables. | NAIST Watanabe Lab |
| [Heron VLM Leaderboard powered by Nejumi/WandB](https://wandb.ai/vision-language-leaderboard/heron-leaderboard/reports/Heron-VLM-Leaderboard-powered-by-Nejumi-WandB--Vmlldzo4MjY3OTc5) | Résume les résultats d'évaluation de [Japanese-Heron-Bench](#japanese-heron-bench) et [LLaVA-Bench-In-the-Wild (Japanese)](#llava-bench-in-the-wild). | Turing, Weights & Biases |
| [Japanese-Heron-Bench](https://huggingface.co/datasets/turing-motors/Japanese-Heron-Bench) | 21 images se voient attribuer un total de 102 questions. Il est caractérisé par des paires image-question qui nécessitent une connaissance liée au Japon. | Turing |
| [JA-VLM-Bench-In-the-Wild](https://huggingface.co/datasets/SakanaAI/JA-VLM-Bench-In-the-Wild) | Un jeu de données préparé indépendamment par Sakana AI pour évaluer EvoVLM-JP-v1-7B. Il se compose de 50 questions attribuées à 42 images. Il se caractérise par des images et des questions qui exigent une connaissance du Japon. | Sakana AI |
| [JA-Multi-Image-VQA](https://huggingface.co/datasets/SakanaAI/JA-Multi-Image-VQA) | Un jeu de données pour évaluer la capacité de question-réponse en japonais pour plusieurs images. | Sakana AI |
| [LLaVA-Bench-In-the-Wild (Japanese)](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-in-the-wild) | Ceci est la version japonaise de [LLaVA-Bench-In-the-Wild](https://huggingface.co/datasets/liuhaotian/llava-bench-in-the-wild), traduite à l'aide de DeepL. Il se compose de 60 questions attribuées à 24 images. | Turing |
| [LLaVA-Bench (COCO) Japonais](https://github.com/turingmotors/heron/tree/main/playground/data/llava-bench-ja) | Il s'agit de la version japonaise, traduite par DeepL, du jeu de données LLaVA-Bench (COCO) utilisé pour évaluer LLaVA. Il se compose de 30 images, chacune avec 3 types de questions qui leur sont attribuées. | Turing |
+| [Japanese Visual Genome VQA dataset](https://github.com/yahoojapan/ja-vg-vqa) | Un jeu de données de questions-réponses annotées basé sur des images du [Visual Genome dataset](https://homes.cs.washington.edu/~ranjay/visualgenome/index.html). Un sous-ensemble de ce jeu de données, [JA-VG-VQA-500](https://huggingface.co/datasets/SakanaAI/JA-VG-VQA-500), composé de 500 questions, est parfois utilisé comme benchmark pour évaluer les VLMs. | Yahoo |
## Références pour les modèles et les architectures