Name: minhnguyent546/bge-m3-data
Creator: minhnguyent546
Published: 2026-03-20 11:45:50
License: 暂无描述

下载链接：

https://hf-mirror.com/datasets/minhnguyent546/bge-m3-data

下载链接

链接失效反馈

官方服务：

资源简介：

--- dataset_info: - config_name: cmed_qav2 features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 1506313240 num_examples: 50000 download_size: 1505466417 dataset_size: 1506313240 - config_name: coliee features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 5782070 num_examples: 463 download_size: 5749474 dataset_size: 5782070 - config_name: dureader features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 splits: - name: train num_bytes: 1264068629 num_examples: 80416 download_size: 1264993588 dataset_size: 1264068629 - config_name: hotpot_qa features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 750866579 num_examples: 84516 download_size: 748474450 dataset_size: 750866579 - config_name: law_gpt features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 17139477 num_examples: 500 download_size: 17189443 dataset_size: 17139477 - config_name: lecard_v2 features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 181165743 num_examples: 591 download_size: 177095238 dataset_size: 181165743 - config_name: miracl features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 - name: lang dtype: string splits: - name: train num_bytes: 4032715429 num_examples: 40203 download_size: 4033530249 dataset_size: 4032715429 - config_name: mldr features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: lang dtype: string splits: - name: train num_bytes: 16660652256 num_examples: 41434 download_size: 16659754101 dataset_size: 16660652256 - config_name: mmarco_zh features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 splits: - name: train num_bytes: 514566583 num_examples: 100000 download_size: 513336485 dataset_size: 514566583 - config_name: mr_tydi features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 - name: lang dtype: string splits: - name: train num_bytes: 4723736283 num_examples: 48729 download_size: 4724577711 dataset_size: 4723736283 - config_name: ms_marco features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 splits: - name: train num_bytes: 37640620073 num_examples: 485905 download_size: 37668509799 dataset_size: 37640620073 - config_name: nli_for_simcse features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 131931702 num_examples: 274951 download_size: 120681894 dataset_size: 131931702 - config_name: nli_zh features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 41325460 num_examples: 64707 download_size: 39935814 dataset_size: 41325460 - config_name: nq features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 splits: - name: train num_bytes: 3740787647 num_examples: 58568 download_size: 3745268049 dataset_size: 3740787647 - config_name: pubmed_qa features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 4107058 num_examples: 500 download_size: 4115137 dataset_size: 4107058 - config_name: squad features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 527377581 num_examples: 87599 download_size: 525769380 dataset_size: 527377581 - config_name: t2_ranking features: - name: query dtype: string - name: pos list: string - name: neg list: string - name: pos_scores list: float64 - name: neg_scores list: float64 splits: - name: train num_bytes: 1691555426 num_examples: 90467 download_size: 1695297120 dataset_size: 1691555426 - config_name: trivia_qa features: - name: query dtype: string - name: pos list: string - name: neg list: string splits: - name: train num_bytes: 3842237250 num_examples: 60315 download_size: 3846967700 dataset_size: 3842237250 configs: - config_name: cmed_qav2 data_files: - split: train path: cmed_qav2/train-* - config_name: coliee data_files: - split: train path: coliee/train-* - config_name: dureader data_files: - split: train path: dureader/train-* - config_name: hotpot_qa data_files: - split: train path: hotpot_qa/train-* - config_name: law_gpt data_files: - split: train path: law_gpt/train-* - config_name: lecard_v2 data_files: - split: train path: lecard_v2/train-* - config_name: miracl data_files: - split: train path: miracl/train-* - config_name: mldr data_files: - split: train path: mldr/train-* - config_name: mmarco_zh data_files: - split: train path: mmarco_zh/train-* - config_name: mr_tydi data_files: - split: train path: mr_tydi/train-* - config_name: ms_marco data_files: - split: train path: ms_marco/train-* default: true - config_name: nli_for_simcse data_files: - split: train path: nli_for_simcse/train-* - config_name: nli_zh data_files: - split: train path: nli_zh/train-* - config_name: nq data_files: - split: train path: nq/train-* - config_name: pubmed_qa data_files: - split: train path: pubmed_qa/train-* - config_name: squad data_files: - split: train path: squad/train-* - config_name: t2_ranking data_files: - split: train path: t2_ranking/train-* - config_name: trivia_qa data_files: - split: train path: trivia_qa/train-* --- ## Dataset Summary This is the reformatted version (to make it loadable with `datasets`) of the fine-tuning data for BGE-M3 model. The original dataset can be found at [Shitao/bge-m3-data](https://huggingface.co/datasets/Shitao/bge-m3-data). | No. | Dataset | Language | # Rows | | :---: | --- | :---: | ---: | | 1 | MS MARCO | English | 485,905 | | 2 | NQ | English | 58,568 | | 3 | HotpotQA | English | 84,516 | | 4 | TriviaQA | English | 60,315 | | 5 | SQuAD | English | 87,599 | | 6 | COLIEE | English | 463 | | 7 | PubMedQA | English | 500 | | 8 | NLI from SimCSE | English | 274,951 | | 9 | DuReader | Chinese | 80,416 | | 10 | mMARCO-zh | Chinese | 100,000 | | 11 | T2Ranking | Chinese | 90,467 | | 12 | Law-GPT | Chinese | 500 | | 13 | cMedQAv2 | Chinese | 50,000 | | 14 | NLI-zh | Chinese | 64,707 | | 15 | LeCaRDv2 | Chinese | 591 | | 16 | Mr.TyDi | 11 languages | 48,729 | | 17 | MIRACL | 16 languages | 40,203 | | 18 | MLDR | 13 languages | 41,434 |

应用场景：