Papers with MS MARCO
CALOR-QUEST : generating a training corpus for Machine Reading Comprehension models from shallow semantic annotations (D19-58)
Copied to clipboard
Frederic Bechet, Cindy Aloui, Delphine Charlet, Geraldine Damnati, Johannes Heinecke, Alexis Nasr, Frederic Herledan
| Challenge: | Recent large corpora of triplets have opened the door to supervised machine learning approaches for Question-Answering. |
| Approach: | They propose to generate questions from the semantic Frame analysis of large corpora using a CALOR-QUEST resource in French and use it to improve machine reading comprehension. |
| Outcome: | The proposed method generates questions from the semantic Frame analysis of large corpora and then tests them on the CALOR-QUEST resource in French. |
Extremely efficient online query encoding for dense retrieval (2024.findings-naacl)
Copied to clipboard
| Challenge: | Existing dense retrieval systems use the same model architecture for encoding both passages and queries, even though queries are much shorter and simpler than passages. |
| Approach: | They propose a small efficient RNN query encoder that can reduce latency by 12 with only a minor decrease in quality. |
| Outcome: | The proposed solution reduces latency by up to 12 while achieving 35.5 MRR@10 score. |
Exp4Fuse: A Rank Fusion Framework for Enhanced Sparse Retrieval using Large Language Model-based Query Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown potential in generating hypothetical documents for query expansion, but their efficacy is dependent on the quality of the generated documents. |
| Approach: | They propose a fusion ranking framework that uses zero-shot LLM-based query expansion to enhance sparse retrieval. |
| Outcome: | The proposed framework surpasses existing LLM-based query expansion methods on three MS MARCO-related datasets and seven low-resource datasets. |
Salient Phrase Aware Dense Retrieval: Can a Dense Retriever Imitate a Sparse One? (2022.findings-emnlp)
Copied to clipboard
Xilun Chen, Kushal Lakhotia, Barlas Oguz, Anchit Gupta, Patrick Lewis, Stan Peshterliev, Yashar Mehdad, Sonal Gupta, Wen-tau Yih
| Challenge: | Existing sparse retrievers lack the ability to match salient phrases and rare entities in the query. |
| Approach: | They introduce a dense Lexical Model that can be trained to imitate a sparse one. |
| Outcome: | The proposed model outperforms sparse retrievers on a range of tasks including five question answering datasets and the MS MARCO passage retrieval. |
MixLoRA-DSI: Dynamically Expandable Mixture-of-LoRA Experts for Rehearsal-Free Generative Retrieval over Dynamic Corpora (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to update model-based indexes with new documents are expensive and require expensive retraining. |
| Approach: | They propose a framework that combines an expandable mixture of Low-Rank Adaptation experts with a layer-wise out-of-distribution-driven expansion strategy. |
| Outcome: | Experiments on NQ320k and MS MARCO Passage show that the proposed framework outperforms full-model update baselines with minimal parameter overhead and substantially lower training costs. |
NEAT-IR: Neural Explainable Analysis Tool for Information Retrieval (2026.acl-srw)
Copied to clipboard
| Challenge: | Neural IR models achieve strong performance but remain difficult to interpret. |
| Approach: | They propose a black-box analysis framework that explains ColBERT’s ranking behavior using 26 classical IR features. |
| Outcome: | The proposed framework preserves ColBERT’s rankings nearly perfectly, yet only explain R2 0.28 of score variance. |
Certified Error Control of Candidate Set Pruning for Two-Stage Relevance Ranking (2022.emnlp-main)
Copied to clipboard
| Challenge: | In information retrieval, candidate set pruning is used to speed up two-stage relevance ranking but lacks accurate error control and empirical guarantees. |
| Approach: | They propose a method that guarantees the test error after pruning is controlled under a user-specified threshold with high probability. |
| Outcome: | The proposed method reduces the average set size from 1000 to 27, increasing reranking speed by about 37 times while keeping MRR@10 greater than a pre-specified value of 0.38 with about 90% empirical coverage. |
RetroMAE: Pre-Training Retrieval-oriented Language Models Via Masked Auto-Encoder (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for dense retrieval are not effective, but there are still challenges. |
| Approach: | They propose a retrieval oriented pre-training paradigm based on Masked Auto-Encoder (MAE) where the sentence embedding is generated from the encoder’s masked input and the original sentence is recovered based upon the sentence embedded and decoded input via mangled language modeling. |
| Outcome: | The proposed model significantly improves the SOTA performance on a wide range of NLP benchmarks, like BEIR and MS MARCO. |
IndicIRSuite: Multilingual Dataset and Neural Information Models for Indian Languages (2024.acl-short)
Copied to clipboard
| Challenge: | IndicIRSuite is the first attempt at building large-scale Neural Information Retrieval resources for a large number of Indian languages. |
| Approach: | They introduce Neural Information Retrieval resources for 11 widely spoken Indian Languages from two major Indian language families. |
| Outcome: | Experiments show that Indic-ColBERT improves on INDIC-MARCO datasets for 11 languages, and that it can be used to improve IR for Indian languages. |
Leveraging Structured Metadata for Improving Question Answering on the Web (2020.aacl-main)
Copied to clipboard
| Challenge: | Using metadata information from web pages can improve the performance of answer passage selection/reranking models. |
| Approach: | They propose a neural passage selection model that leverages metadata information with a fine-grained encoding strategy to learn the representation for metadata predicates in a hierarchical way. |
| Outcome: | The proposed model outperforms baseline models on the MS MARCO and Recipe-MARCO datasets and shows that it is more accurate than baseline models. |
Self-Training for Jointly Learning to Ask and Answer Questions (N18-1)
Copied to clipboard
| Challenge: | Existing methods for question answering and question generation are hard to obtain in many domains. |
| Approach: | They propose a method for jointly learning to ask and answer questions . they leverage unlabeled text along with labeled question answer pairs for learning . |
| Outcome: | The proposed method improves on four benchmark datasets on question answering and question generation tasks. |
Read before Generate! Faithful Long Form Question Answering with Machine Reading (2022.findings-acl)
Copied to clipboard
| Challenge: | Long-form question answering (LFQA) generates a paragraph-length answer for a given question. |
| Approach: | They propose a framework that jointly models answer generation and machine reading. |
| Outcome: | The proposed model generates a more factually accurate answer from millions of documents retrieved from a large dataset. |
Document Ranking with a Pretrained Sequence-to-Sequence Model (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Experimental results on the MS MARCO passage ranking task show that our ranking approach is superior to strong encoder-only models. |
| Approach: | They propose to use a pretrained sequence-to-sequence model to generate relevance labels as "target tokens" they also show how the underlying logits of these target tokens can be interpreted as relevance probabilities for ranking. |
| Outcome: | The proposed model outperforms existing models in a data-poor setting and significantly outperformed an encoder-only model on the MS MARCO passage ranking task. |
Ultra-High Dimensional Sparse Representations with Binarization for Efficient Text Retrieval (2021.emnlp-main)
Copied to clipboard
| Challenge: | Recent approaches to information retrieval (IR) and natural language processing (NLP) use contextual language models, which can improve both synonymy and polysemy problems associated with words. |
| Approach: | They propose an ultra-high dimensional representation scheme equipped with directly controllable sparsity and a bucketing method where embeddings from multiple layers of BERT are selected/merged to represent diverse linguistic aspects. |
| Outcome: | The proposed representation scheme outperforms sparse models with MS MARCO and TREC CAR, and shows that it is highly efficient for storage and search. |
A Self-supervised Joint Training Framework for Document Reranking (2022.findings-naacl)
Copied to clipboard
| Challenge: | Pretrained language models have been successfully applied to a wide range of tasks . however, the pretraining tasks were based on the context of documents . |
| Approach: | They propose a self-supervised joint training framework with a method called Masked Query Prediction to establish semantic relations between given queries and positive documents. |
| Outcome: | The proposed framework outperforms existing models on document reranking tasks without further pre-training . it uses a self-supervised method to establish semantic relations between given queries and positive documents. |
How Does Generative Retrieval Scale to Millions of Passages? (2023.emnlp-main)
Copied to clipboard
Ronak Pradeep, Kai Hui, Jai Gupta, Adam Lelkes, Honglei Zhuang, Jimmy Lin, Donald Metzler, Vinh Tran
| Challenge: | generative retrieval is a new paradigm for information retrieval, enabling a sequence-to-sequence model with a single Transformer . generative encoders have been used on small corpora, but only on large ones . |
| Approach: | They propose to encode an entire document corpus within a single Transformer . they find generative retrieval is competitive with state-of-the-art dual encoders on small corpora . |
| Outcome: | The proposed approach is competitive with state-of-the-art dual encoders on small corpora, the study finds . the proposed approach only evaluates on document corporales on the order of 100K in size . |
Utility-Focused LLM Annotation for Retrieval and Retrieval-Augmented Generation (2025.emnlp-main)
Copied to clipboard
Hengran Zhang, Minghao Tang, Keping Bi, Jiafeng Guo, Shihao Liu, Daiting Shi, Dawei Yin, Xueqi Cheng
| Challenge: | Existing studies on large language models for document utility annotations have shown that they improve retrieval performance and RAG outcomes compared to models trained on human annotations. |
| Approach: | They propose a model that maximizes their summed marginal likelihood to annotate document utility on multiple positive samples per query. |
| Outcome: | The proposed model maximizes the marginal likelihood of multiple positive samples per query. |
FAA: Fine-grained Attention Alignment for Cascade Document Ranking (2023.acl-long)
Copied to clipboard
| Challenge: | Contemporary document ranking methods focus on transforming documents into passages to handle long inputs, but intensive query-irrelevant content may lead to harmful distraction and high query latency. |
| Approach: | They propose a fine-grained attention alignment approach to jointly optimize a cascade document ranking model. |
| Outcome: | Experiments on MS MARCO and TREC DL show that the proposed method is effective in document ranking tasks. |
Addressing Semantic Drift in Generative Question Answering with Auxiliary Extraction (2021.acl-short)
Copied to clipboard
| Challenge: | Recent work focuses on question answering based on machine reading comprehension . current approaches treat QA as extracting a consecutive piece of text to a given question. |
| Approach: | They propose a generative QA model that incorporates an extractive mechanism into a model. |
| Outcome: | The proposed model improves quality and semantic accuracy over baseline models. |
Learning to Generate Questions by Learning to Recover Answer-containing Sentences (2021.findings-acl)
Copied to clipboard
| Challenge: | Recent research has focused on synthetically generating a question from a given context and an annotated answer by training an additional generative model. |
| Approach: | They propose a method that learns to generate contextually rich questions by recovering answer-containing sentences. |
| Outcome: | The proposed approach improves the quality and accuracy of existing models and achieves comparable results to the state-of-the-art on MS MARCO and NewsQA. |
RetroMAE-2: Duplex Masked Auto-Encoder For Pre-Training Retrieval-Oriented Language Models (2023.acl-long)
Copied to clipboard
| Challenge: | Existing methods for retrieval-oriented language models focus on contextualized embedding of the [CLS] token, but recent study shows that ordinary tokens besides [CLL] may provide extra information, which help to produce a better representation effect. |
| Approach: | They propose a method where all contextualized embeddings of pre-trained model can be jointly pre-trained for retrieval tasks. |
| Outcome: | The proposed method improves the quality of representation where all contextualized embeddings of the pre-trained model can be leveraged. |
Stochastic Answer Networks for Machine Reading Comprehension (P18-1)
Copied to clipboard
| Challenge: | Several recent MRC models employ multi-step reasoning . we show that the use of a stochastic prediction dropout improves robustness . |
| Approach: | They propose a stochastic answer network that simulates multi-step reasoning in machine reading comprehension. |
| Outcome: | The proposed model improves robustness and results competitive with state-of-the-art models on the Stanford Question Answering Dataset and Microsoft MAchine Reading COmprehension Dataset. |
Multi-style Generative Reading Comprehension (P19-1)
Copied to clipboard
Kyosuke Nishida, Itsumi Saito, Kosuke Nishida, Kazutoshi Shinoda, Atsushi Otsuka, Hisako Asano, Junji Tomita
| Challenge: | Current studies on generative reading comprehension (RC) focus on extracting an answer span from textual evidence and natural language generation (NLG). |
| Approach: | They propose a multi-style abstractive summarization model for question answering called Masque. |
| Outcome: | The proposed model achieves state-of-the-art performance on the Q&A and Q& A + NLG tasks of MS MARCO and NarrativeQA. |
Corpus Poisoning via Approximate Greedy Gradient Descent (2025.findings-acl)
Copied to clipboard
| Challenge: | Dense retrievers have been shown to be vulnerable to corpus poisoning attacks . a malicious user injects a small fraction of adversarial passages into the retrieval corpus to trick the system into returning these passages among the top-ranked results for a broad set of user queries. |
| Approach: | They propose an attack on dense retrieval systems based on the widely used HotFlip method for efficiently generating adversarial passages. |
| Outcome: | The proposed method achieves a high success rate on several datasets and using several retrievers, and can generalize to unseen queries and new domains. |
Dealing with Typos for BERT-based Passage Retrieval and Ranking (2021.emnlp-main)
Copied to clipboard
| Challenge: | Current approaches to passage retrieval and ranking rely on pre-trained deep language models that model the semantic matching between queries and passages. |
| Approach: | They propose a typos-aware training framework for DR and BERT to address this issue. |
| Outcome: | The proposed models respond and adapt to keyword typos occurring in queries, and significantly improve their retrieval and ranking effectiveness. |
Enhancing Document Ranking with Task-adaptive Training and Segmented Token Recovery Mechanism (2021.emnlp-main)
Copied to clipboard
| Challenge: | Document Retrieval (DR) requires the machine to retrieve and rank documents according to their relevance with the query. |
| Approach: | They propose a ranking model DR-BERT which improves the Document Retrieval task by a task-adaptive training process and a Segmented Token Recovery Mechanism. |
| Outcome: | The proposed ranking model keeps in the top three on the MS MARCO leaderboard since 2020. |
CoRT: Complementary Rankings from Transformers (2021.naacl-main)
Copied to clipboard
| Challenge: | Recent approaches to information retrieval mitigate computational costs by using a multi-stage ranking pipeline. |
| Approach: | They propose a ranking model that leverages contextual representations from pre-trained language models to complement term-based ranking functions while causing no significant delay at query time. |
| Outcome: | The proposed model significantly increases candidate recall by complementing BM25 with missing candidates while causing no significant delay at query time. |
TOME: A Two-stage Approach for Model-based Retrieval (2023.acl-long)
Copied to clipboard
| Challenge: | Recent research has focused on model-based retrieval, which discards the index in the traditional retrieval model and memorizes the candidate corpora using model parameters. |
| Approach: | They propose a model-based retrieval approach that discards the index in the traditional retrieval model and memorizes the candidate corpora using model parameters. |
| Outcome: | The proposed approach eliminates the index in the traditional retrieval model and memorizes the candidate corpora using model parameters. |
Reinforcement Learning with Supervised Alignment (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Supervised fine-tuning (SFT) is a widely used method for adapting Large Language Models to specific tasks. |
| Approach: | They propose a method that uses supervised fine-tuning to train a reward model for reinforcement learning. |
| Outcome: | The proposed method outperforms existing methods on in-domain benchmarks but surpasses them 50 times on out-of-domain and cross-task evaluations. |
Inference Scaling for Bridging Retrieval and Augmented Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing work observed the generator bias, such that improving the retrieval results may negatively affect the outcome. |
| Approach: | They propose to use inference scaling to aggregate inference calls from the permuted order of retrieved contexts to create a new ranking. |
| Outcome: | The proposed approach improves ROUGE-L on MS MARCO and EM on HotpotQA benchmarks by 7 points. |
Towards Imperceptible Document Manipulations against Neural Ranking Models (2023.findings-acl)
Copied to clipboard
| Challenge: | Current approaches to detect vulnerabilities in neural ranking models often introduce noticeable errors and require a well-imitated surrogate NRM to guarantee the attack effect. |
| Approach: | They propose a framework called Imperceptible DocumEnt Manipulation to produce adversarial documents that are less noticeable to both algorithms and humans. |
| Outcome: | The proposed framework outperforms strong baselines while maintaining fluency and correctness of the target documents. |
GLEN: Generative Retrieval via Lexical Index Learning (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for document retrieval bypass auxiliary index structures and can be optimized through end-to-end learning. |
| Approach: | They propose a method to generate a relevant document's identifier using an index learning strategy. |
| Outcome: | The proposed method achieves state-of-the-art or competitive performance on benchmark datasets. |
LLM Prompt Duel Optimizer: Efficient Label-Free Prompt Optimization (2026.findings-acl)
Copied to clipboard
Yuanchen Wu, Saurabh Verma, Justin Lee, Fangzhou Xiong, Poppy Zhang, Amel Awadelkarim, Xu Chen, Yubai Yuan, Shawndra Hill
| Challenge: | Large language models (LLMs) are highly sensitive to prompts, but most automatic prompt optimization methods assume access to ground-truth references that are costly to obtain. |
| Approach: | They propose a sample-efficient framework for label-free prompt optimization based on pairwise preference feedback from an LLM judge. |
| Outcome: | Experiments on BIG-bench Hard and MS MARCO show that the proposed framework identifies stronger prompts than label-free baselines while offering favorable quality–cost trade-offs. |
Multi-stage Training with Improved Negative Contrast for Neural Passage Retrieval (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing neural firststage retrieval models overcome lexical gap issue by projecting query and document to a shared dense space. |
| Approach: | They propose a multi-stage framework for neural passage retrieval using synthetic data, negative sampling, and fusion techniques. |
| Outcome: | The proposed framework improves retrieval accuracy and enhances the negative contrast in both stages. |
Synthesize, Prompt and Transfer: Zero-shot Conversational Question Generation with Pre-trained Language Model (2023.acl-long)
Copied to clipboard
| Challenge: | Existing research on QG focuses on generating single-turn questions, which are formalized as independent interactions. |
| Approach: | They propose a multi-stage knowledge transfer framework to leverage knowledge from single-turn question generation instances. |
| Outcome: | The proposed framework achieves 14.81 BLEU-4 (88.2% absolute improvement compared to T5) in CoQA with knowledge transferred from three single-turn datasets. |
DiffusionRet: Diffusion-Enhanced Generative Retriever using Constrained Decoding (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Generative retrieval methods have suffered from the lack of the intermediate reasoning step . generative retrieval uses sequence-to-sequence diffusion models to map a query to relevant docids . |
| Approach: | They propose a novel method that uses query as an intermediate step before retrieval . they propose to use sequence-to-sequence diffusion models to map a query to relevant docids . |
| Outcome: | Experiments show that proposed method outperforms existing methods on MARCO and Natural Questions datasets. |
Exploring Dual Encoder Architectures for Question Answering (2022.emnlp-main)
Copied to clipboard
| Challenge: | Dual encoders have been used for question-answering and information retrieval tasks with good results. |
| Approach: | They propose to use two different versions of dual encoders for QA retrieval tasks . they propose to share parameters in projection layers between two encoder towers . |
| Outcome: | The proposed architectures outperform SDE and ADE on QA retrieval tasks. |
CITADEL: Conditional Token Interaction via Dynamic Lexical Routing for Efficient and Effective Multi-Vector Retrieval (2023.acl-long)
Copied to clipboard
Minghan Li, Sheng-Chieh Lin, Barlas Oguz, Asish Ghoshal, Jimmy Lin, Yashar Mehdad, Wen-tau Yih, Xilun Chen
| Challenge: | Existing multi-vector retrieval methods are slower and require more space to store indices compared to their single-vektor counterparts. |
| Approach: | They propose a multi-vector retrieval method that uses dynamic lexical routing to route different token vectors to the predicted lexicals. |
| Outcome: | The proposed method achieves state-of-the-art performance on several benchmark datasets while being nearly 40 times faster than the current state-out-of the-art method. |
Fusion-in-T5: Unifying Variant Signals for Simple and Effective Document Ranking with Attention Fusion (2024.lrec-main)
Copied to clipboard
| Challenge: | Current document ranking pipelines involve multiple ranking layers to integrate different information step-by-step. |
| Approach: | They propose a novel re-ranker Fusion-in-T5 which integrates text matching information, ranking features, and global document information into one single unified model via templated-based input and global attention. |
| Outcome: | The proposed model significantly improves ranking performance over complex cascade pipelines. |
Cross-lingual and Cross-domain Evaluation of Machine Reading Comprehension with Squad and CALOR-Quest Corpora (2020.lrec-1)
Copied to clipboard
| Challenge: | a recent study has shown that language mismatch and domain mismatch can affect performance of a machine reading task . a factor between language mismatched and domain-mismatched has the strongest influence on performance . |
| Approach: | They compare the cross-language and cross-domain capabilities of BERT on a machine reading comprehension task on two corpora: SQuAD and a new French Machine Reading dataset. |
| Outcome: | The proposed model matches human performance on a machine reading comprehension task with BERT on Chinese and French documents with interesting results. |
CODER: An efficient framework for improving retrieval through COntextual Document Embedding Reranking (2022.emnlp-main)
Copied to clipboard
| Challenge: | Contextual document embedding reranking is an efficient and efficient retrieval framework. |
| Approach: | They propose a highly efficient retrieval framework that uses contextual document embedding reranking to incorporate ranking context into training. |
| Outcome: | The proposed framework reduces the computational overhead of a first-stage method and can be used as stand-alone retrieval models. |
Moving Beyond Downstream Task Accuracy for Information Retrieval Benchmarking (2023.findings-acl)
Copied to clipboard
Keshav Santhanam, Jon Saad-Falcon, Martin Franz, Omar Khattab, Avi Sil, Radu Florian, Md Arafat Sultan, Salim Roukos, Matei Zaharia, Christopher Potts
| Challenge: | Neural information retrieval (IR) systems have progressed rapidly in recent years . many IR benchmarks focus on downstream task accuracy, concealing costs incurred . |
| Approach: | They propose to include efficiency considerations on IR benchmarks to help drive progress . eral et al. propose to incorporate query latency and cost budgets into evaluation . |
| Outcome: | a new study shows that the best IR system varies according to how efficiency considerations are chosen and weighed . the proposed benchmarks would allow for more thorough exploration of possible system designs . |
HYRR: Hybrid Infused Reranking for Passage Retrieval (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing passage retrieval systems typically adopt a two-stage retrieve-then-rerank pipeline. |
| Approach: | They propose a framework for training robust reranking models using hybrid retrievers . they propose HYRR framework that allows users to select training data using hybrids . |
| Outcome: | The proposed framework is robust to different first-stage retrieval settings. |
SamToNe: Improving Contrastive Loss for Dual Encoder Retrieval Models with Same Tower Negatives (2023.findings-acl)
Copied to clipboard
| Challenge: | Dual encoders have been used for retrieval tasks and representation learning with good results. |
| Approach: | They propose an improved contrastive learning objective by adding queries or documents from the same encoder towers to the negatives. |
| Outcome: | The proposed model improves retrieval quality for both symmetric and asymmetric dual encoders by adding queries or documents from the same encoder towers to the negatives. |
Stop Hardening Everything: A Training-Free Neuron-Level Defense for Neural Ranking Models (2026.acl-long)
Copied to clipboard
| Challenge: | Existing defenses for neural ranking models are data-centric and require retraining and adversarial data generation. |
| Approach: | They propose a model-centric defense that addresses vulnerability at its architectural source without costly retraining or adversarial data generation. |
| Outcome: | The proposed approach outperforms state-of-the-art models on MS MARCO and TREC 19 while maintaining strong performance on clean data. |
Threshold-driven Pruning with Segmented Maximum Term Weights for Approximate Cluster-based Sparse Retrieval (2024.emnlp-main)
Copied to clipboard
| Challenge: | Using rank score thresholding, sparse retrieval skips the index at cluster and document levels. |
| Approach: | They propose a pruning control scheme with a probabilistic guarantee on rank-safeness competitiveness. |
| Outcome: | The proposed pruning control scheme improves accuracy and safeness while delivering low latency on single-threaded CPU. |
Beyond Contrastive Learning: Synthetic Data Enables List-wise Training with Multiple Levels of Relevance (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a new approach to training with binary relevance labels uses synthetic data . contrastive learning with binary correlations leaves out subtle nuances useful for ranking . |
| Approach: | They propose to use waterstein distance as a loss function for training transformer-based retrievers with graduated relevance labels instead of real documents. |
| Outcome: | The proposed method outperforms conventional training with InfoNCE by a large margin on MARCO and BEIR benchmarks without using real documents. |