Datasets:

neuclir
/

bench

license: cc-by-4.0
annotations_creators:
  - NIST
task_categories:
  - text-retrieval
  - text-ranking
language:
  - en
  - zh
  - fa
  - ru
multilinguality:
  - multilingual
pretty_name: NeuCLIRBench
size_categories:
  - n<1K
task_ids:
  - document-retrieval
configs:
  - config_name: queries
    default: true
    data_files:
      - split: eng
        path: data/news.eng.tsv
      - split: fas
        path: data/news.fas.tsv
      - split: rus
        path: data/news.rus.tsv
      - split: zho
        path: data/news.zho.tsv
    format: csv
    sep: "\t"
    header: null
    names:
      - id
      - query
    dataset_info:
      features:
        - name: id
          dtype: string
        - name: query
          dtype: string
  - config_name: qrels
    default: false
    data_files:
      - split: mlir
        path: data/qrels.mlir.gains.txt
      - split: fas
        path: data/qrels.fas.gains.txt
      - split: rus
        path: data/qrels.rus.gains.txt
      - split: zho
        path: data/qrels.zho.gains.txt
    format: csv
    sep: ' '
    header: null
    names:
      - id
      - ignore
      - docid
      - relevance
    dataset_info:
      features:
        - name: id
          dtype: string
        - name: ignore
          dtype: string
        - name: docid
          dtype: string
        - name: relevance
          dtype: int

NeuCLIRBench Topics and Queries

NeuCLIRBench is an evaluation benchmark for monolingual, cross-language, and multilingual adhoc retrieval.

The document collection can be found at neuclir/neuclir1.

Supporting Tasks and Corresponding Data

NeuCLIRBench supports three types of tasks: monolingual, cross-language, and multilingual adhoc retrieval. The following specifies the documents, queries, and qrels (labels) that should be used for each task.

Please report nDCG@20 for all tasks.

We use : to indicate different subset under the dataset.

Monolingual Retrieval (`mono`)

Language	Documents	Queries	Qrels
English	All splits under `neuclir/neuclir1:mt_docs`	`eng` split of `neuclir/bench:queries`	`mlir` split of `neuclir/bench:qrels`
Persian	`fas` split of `neuclir/neuclir1:default`	`fas` split of `neuclir/bench:queries`	`fas` split of `neuclir/bench:qrels`
Russian	`rus` split of `neuclir/neuclir1:default`	`rus` split of `neuclir/bench:queries`	`rus` split of `neuclir/bench:qrels`
Chinese	`zho` split of `neuclir/neuclir1:default`	`zho` split of `neuclir/bench:queries`	`zho` split of `neuclir/bench:qrels`

Cross-Language Retrieval (`clir`)

Language	Documents	Queries	Qrels
Persian	`fas` split of `neuclir/neuclir1:default`	`eng` split of `neuclir/bench:queries`	`fas` split of `neuclir/bench:qrels`
Russian	`rus` split of `neuclir/neuclir1:default`	`eng` split of `neuclir/bench:queries`	`rus` split of `neuclir/bench:qrels`
Chinese	`zho` split of `neuclir/neuclir1:default`	`eng` split of `neuclir/bench:queries`	`zho` split of `neuclir/bench:qrels`

Multilingual Retrieval (`mlir`)

Language	Documents	Queries	Qrels
English	All splits under `neuclir/neuclir1:default`	`eng` split of `neuclir/bench:queries`	`mlir` split of `neuclir/bench:qrels`

Baseline Retrieval Results and Run Files

We also provide all reported baseline retrieval results in the NeuCLIRBench paper. Please refer to the paper for the detailed descriptions of each model.

Run Names

Please refer to the ./runs directory in this dataset to find all the runs. Files follow the naming scheme of {run_handle}_{task:mono/clir/mlir}_{lang}.trec. Please refer to the task section for the details.

Run Handle	Model Type	Model Name
bm25	Lexical	BM25
bm25dt	Lexical	BM25 w/ DT
bm25qt	Lexical	BM25 w/ QT
milco	Bi-Encoder	MILCO
plaidx	Bi-Encoder	PLAID-X
qwen8b	Bi-Encoder	Qwen3 8B Embed
qwen4b	Bi-Encoder	Qwen3 4B Embed
qwen600m	Bi-Encoder	Qwen3 0.6B Embed
arctic	Bi-Encoder	Arctic-Embed Large v2
splade	Bi-Encoder	SPLADEv3
fusion3	Bi-Encoder	Fusion
repllama	Bi-Encoder	RepLlama
me5large	Bi-Encoder	e5 Large
jinav3	Bi-Encoder	JinaV3
bgem3sparse	Bi-Encoder	BGE-M3 Sparse
mt5	Pointwise Reranker	Mono-mT5XXL
qwen3-0.6b-rerank	Pointwise Reranker	Qwen3 0.6B Rerank
qwen3-4b-rerank	Pointwise Reranker	Qwen3 4B Rerank
qwen3-8b-rerank	Pointwise Reranker	Qwen3 8B Rerank
jina-rerank	Pointwise Reranker	Jina Reranker
searcher-rerank	Pointwise Reranker	SEARCHER Reranker
rank1	Pointwise Reranker	Rank1
qwq	Listwise Reranker	Rank-K (QwQ)
rankzephyr	Listwise Reranker	RankZephyr 7B
firstqwen	Listwise Reranker	FIRST Qwen3 8B
rq32b	Listwise Reranker	RankQwen-32B

Citation

@article{neuclirbench,
      title={NeuCLIRBench: A Modern Evaluation Collection for Monolingual, Cross-Language, and Multilingual Information Retrieval}, 
      author={Dawn Lawrie and James Mayfield and Eugene Yang and Andrew Yates and Sean MacAvaney and Ronak Pradeep and Scott Miller and Paul McNamee and Luca Soldani},
      year={2025},
      eprint={2511.14758},
      archivePrefix={arXiv},
      primaryClass={cs.IR},
      url={https://arxiv.org/abs/2511.14758}, 
}