multinews_dense_mean

This is a copy of the [Multi-News](https://huggingface.co/datasets/multi_news) dataset, except the input source documents of its `train`, `validation` and `test` splits have been replaced by a __dense__ retriever. The retrieval pipeline used:

- __query__: The `summary` field of each example
- __corpus__: The union of all documents in the `train`, `validation` and `test` splits
- __retriever__: [`facebook/contriever-msmarco`](https://huggingface.co/facebook/contriever-msmarco) via [PyTerrier](https://pyterrier.readthedocs.io/en/latest/) with default settings
- __top-k strategy__: `"max"`, i.e. the number of documents retrieved, `k`, is set as the maximum number of documents seen across examples in this dataset, in this case `k==3`

Retrieval results on the `train` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.8661 | 0.6867 | 0.5936 | 0.6917 |

Retrieval results on the `validation` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.8626 | 0.6859 | 0.5874 | 0.6925 |

Retrieval results on the `test` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.8625 | 0.6927 | 0.5938 | 0.6993 |

本数据集为[Multi-News](https://huggingface.co/datasets/multi_news)数据集的复刻版本，仅将其`train`（训练集划分）、`validation`（验证集划分）与`test`（测试集划分）的输入源文档替换为**密集检索器（dense retriever）**获取的内容。所采用的检索流水线如下：
- **查询项（query）**：每个样本的`summary`（摘要）字段
- **语料库（corpus）**：`train`、`validation`和`test`划分中所有文档的并集
- **检索器（retriever）**：基于[PyTerrier](https://pyterrier.readthedocs.io/en/latest/)默认配置，使用[`facebook/contriever-msmarco`](https://huggingface.co/facebook/contriever-msmarco)检索模型
- **Top-k检索策略（top-k strategy）**：采用`"max"`模式，即检索文档数`k`设置为该数据集所有样本中所需的最大文档数，本次场景下`k=3`

训练集检索结果：
| 召回率@100（Recall@100） | 平均精确率（Rprec） | Top-k精确率（Precision@k） | Top-k召回率（Recall@k） |
| ----------- | ----------- | ----------- | ----------- |
| 0.8661 | 0.6867 | 0.5936 | 0.6917 |

验证集检索结果：
| 召回率@100（Recall@100） | 平均精确率（Rprec） | Top-k精确率（Precision@k） | Top-k召回率（Recall@k） |
| ----------- | ----------- | ----------- | ----------- |
| 0.8626 | 0.6859 | 0.5874 | 0.6925 |

测试集检索结果：
| 召回率@100（Recall@100） | 平均精确率（Rprec） | Top-k精确率（Precision@k） | Top-k召回率（Recall@k） |
| ----------- | ----------- | ----------- | ----------- |
| 0.8625 | 0.6927 | 0.5938 | 0.6993 |