datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
visual_haystacks
Visual Haystacks Dataset Card
Dataset details
Dataset type: Visual Haystacks (VHs) is a benchmark dataset specifically designed to evaluate the Large Multimodal Model's (LMM's) capability to handle long-context visual information. It can also be viewed as the first vision-centric Needle-In-A-Haystack (NIAH) benchmark dataset. Please also download COCO-2017's training set validation set.
Data Preparation and Benchmarking
Download the VQA questions:huggingface-cli… See the full description on the dataset page: https://huggingface.co/datasets/tsunghanwu/visual_haystacks.0717-calm3-22b-random-genre-inst-sft-tsub
自動生成Q&A
ランダムなジャンルについて､OpenCalm3-22bで生成したQ&Aです｡
一部の計算には東京工業大学のスーパーコンピュータTSUBAME4.0を利用しました｡
データ
jsonlファイルが数十GB程度あります
datasetsライブラリからでは､はじめの数GB程度しか読み込めない可能性があります｡git lfsなどでダウンロードする必要がありそうです｡
クリーニングはしていません｡おかしなinstructionが一定数､含まれます
Tsundere-AI
Tsundere AI Dataset
A dataset of ~300 conversations between a user and a tsundere AI assistant in chatML format. Built as part of a personal local AI project and cleaned for public release.
What's in it:
Daily life interactions, morning check-ins, schedule reminders, food, anime, exercise
Emotional moments, deflected warmth, flustered responses, buried affection,
Flustered and tsun moments, kind and caring moments, serious chats
Technical conversations, coding help, debugging… See the full description on the dataset page: https://huggingface.co/datasets/arskaz/Tsundere-AI.0723-calm3-22b-random-genre-inst-sft-multiturn-clean-tsub
自動生成Q&A
ランダムなジャンルについて､OpenCalm3-22bで生成したQ&Aです｡
一部の計算には東京工業大学のスーパーコンピュータTSUBAME4.0を利用しました｡
データ
クリーニングはしていません｡おかしなtextが一定数､含まれます
soulchat-qa0719-calm3-22b-random-genre-inst-sft-multiturn-tsub
自動生成Q&A
ランダムなジャンルについて､OpenCalm3-22bで生成したQ&Aです｡
一部の計算には東京工業大学のスーパーコンピュータTSUBAME4.0を利用しました｡
データ
jsonlファイルが数十GB程度あります
datasetsライブラリからでは､はじめの数GB程度しか読み込めない可能性があります｡git lfsなどでダウンロードする必要がありそうです｡
クリーニングはしていません｡おかしなinstructionが一定数､含まれます
Q2がQ1,A1を参照しない仕様で質疑を生成したため、ややチグハグな質疑応答になっています。
0802ramdom-to-fixed-multiturn-Calm3-pretrain-tsubopencpop0722-calm3-22b-random-genre-inst-sft-multiturn-tsub
自動生成したテキスト
Calm3で自動生成したテキストです｡
一部の計算には東京工業大学のスーパーコンピュータTSUBAME4.0を利用しました｡
visual_haystacks_v0
Visual Haystacks Dataset Card
Dataset details
Dataset type: Visual Haystacks (VHs) is a benchmark dataset specifically designed to evaluate the Large Multimodal Model's (LMM's) capability to handle long-context visual information. It can also be viewed as the first visual-centric Needle-In-A-Haystack (NIAH) benchmark dataset. Please also download COCO-2017's training set validation set.
Data Preparation and Benchmarking
Download the VQA questions:huggingface-cli… See the full description on the dataset page: https://huggingface.co/datasets/tsunghanwu/visual_haystacks_v0.Tsunami-th__Tsunami-0.5-7B-Instruct-details
Dataset Card for Evaluation run of Tsunami-th/Tsunami-0.5-7B-Instruct
Dataset automatically created during the evaluation run of model Tsunami-th/Tsunami-0.5-7B-Instruct
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Tsunami-th__Tsunami-0.5-7B-Instruct-details.Tsunami-th__Tsunami-0.5x-7B-Instruct-details
Dataset Card for Evaluation run of Tsunami-th/Tsunami-0.5x-7B-Instruct
Dataset automatically created during the evaluation run of model Tsunami-th/Tsunami-0.5x-7B-Instruct
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Tsunami-th__Tsunami-0.5x-7B-Instruct-details.Tsunami-th__Tsunami-1.0-7B-Instruct-details
Dataset Card for Evaluation run of Tsunami-th/Tsunami-1.0-7B-Instruct
Dataset automatically created during the evaluation run of model Tsunami-th/Tsunami-1.0-7B-Instruct
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Tsunami-th__Tsunami-1.0-7B-Instruct-details.0717-calm3-22b-random-genre-inst-sft-tsub-part
自動生成したテキスト
Calm3で自動生成したテキストです｡
一部の計算には東京工業大学のスーパーコンピュータTSUBAME4.0を利用しました｡
Tsunami-th__Tsunami-1.0-14B-Instruct-details
Dataset Card for Evaluation run of Tsunami-th/Tsunami-1.0-14B-Instruct
Dataset automatically created during the evaluation run of model Tsunami-th/Tsunami-1.0-14B-Instruct
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Tsunami-th__Tsunami-1.0-14B-Instruct-details.Supichi__BBAI_525_Tsu_gZ_Xia0-details
Dataset Card for Evaluation run of Supichi/BBAI_525_Tsu_gZ_Xia0
Dataset automatically created during the evaluation run of model Supichi/BBAI_525_Tsu_gZ_Xia0
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Supichi__BBAI_525_Tsu_gZ_Xia0-details.Supichi__BBAI_275_Tsunami_gZ-details
Dataset Card for Evaluation run of Supichi/BBAI_275_Tsunami_gZ
Dataset automatically created during the evaluation run of model Supichi/BBAI_275_Tsunami_gZ
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Supichi__BBAI_275_Tsunami_gZ-details.0804ramdom-to-fixed-multiturn-Calm3-pretrain-tsubSTATUS_Trainrandom-genre-inst-sft-multiturn-clean-tsub-Oumuamua-7b-instruct-v2
nitky/Oumuamua-7b-instruct-v2で生成したマルチターンデータです｡
ライセンスは一応､当該モデル(とそのマージ元)に従い､apache-2.0としています｡ただし､当該モデルは[GPT-4の出力を学習したモデル])(https://huggingface.co/prometheus-eval/prometheus-7b-v2.0)からマージで生成されている点に､ご留意ください｡
MLDeuggingMWPES-300k