Internship: Retrieval Augmented Generation

We are looking for students who are interested in working on improving retrieval augmented generation (RAG) in multilingual or multi-domain scenarios.

Based on a simple idea of augmenting user requests with relevant passages retrieved from the Internet or a given datastore, RAG has recently emerged as a promising solution for improving LLM factuality and grounded attribution. Despite high attention this research topic has received in recent years, the vast majority of works only focus on Wikipedia-based English settings in their experiments, trained models, or collected datasets. At the same time, initial efforts were made to evaluate or extend RAG to multi-domain [1, 2] or multilingual [3-6] settings. 

The topic of this internship will be related to continuing improving advanced RAG pipelines so that they better support queries and contexts in non-English or from various domains.

Internship supervisors: Nadezhda Chirkova (https://nadiinchi.github.io/), Thibault Formal (https://scholar.google.fr/citations?user=mhVuc98AAAAJ), Vassilina Nikoulina (https://scholar.google.fr/citations?user=IVJ4wN4AAAAJ)

The intern will be working with the team of the researchers with background in RAG [5-6], natural language generation [7-9] and information retrieval [10-12].

- PhD or last year MSc student in NLP-related domains
- Solid deep learning and NLP background
- Experience with Pytorch toolkit

[1] RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation. Dongyu Ru et al., NeurIPS 2024 (https://arxiv.org/abs/2408.08067)

[2] RAFT: Adapting Language Model to Domain Specific RAG. Tianjun Zhang et al., COLM 2024 (https://arxiv.org/abs/2403.10131)

[3] MIRAGE-Bench: Automatic Multilingual Benchmark Arena for Retrieval-Augmented Generation Systems. Nandan Thakur et al., 2024 (https://arxiv.org/abs/2410.13716)

[4] Not All Languages are Equal: Insights into Multilingual Retrieval-Augmented Generation. Suhang Wu et al., 2024 (https://arxiv.org/abs/2410.21970)

[5] BERGEN: A Benchmarking Library for Retrieval-Augmented Generation. David Rau, Hervé Déjean, Nadezhda Chirkova, Thibault Formal, Shuai Wang, Vassilina Nikoulina, Stéphane Clinchant. Findings of EMNLP 2024 (https://arxiv.org/abs/2407.01102)

[6] Retrieval-augmented generation in multilingual settings. Nadezhda Chirkova, David Rau, Hervé Déjean, Thibault Formal, Stéphane Clinchant, Vassilina Nikoulina. Knowledgeable LLMs workshop @ ACL 2024 (https://arxiv.org/abs/2407.01463)

[7] Key ingredients for effective zero-shot cross-lingual knowledge transfer in generative tasks. Nadezhda Chirkova and Vasilina Nikoulina, NAACL 2024. (https://arxiv.org/abs/2402.12279)

[8] Zero-shot cross-lingual transfer in instruction tuning of large language models. Nadezhda Chirkova and Vasilina Nikoulina, INLG 2024. (https://arxiv.org/abs/2402.14778)

[9] BLOOM+1: Adding Language Support to BLOOM for Zero-Shot Prompting. Zheng Xin Yong et al., ACL 2023. (https://arxiv.org/abs/2212.09535)

[10] SPLADE: Sparse Lexical and Expansion Model for First Stage Ranking. Thibault Formal, Benjamin Piwowarski, Stéphane Clinchant. SIGIR 2021 (https://arxiv.org/abs/2107.05720)

[11] MS-Shift: An Analysis of MS MARCO Distribution Shifts on Neural Retrieval. Simon Lupart, Thibault Formal, Stéphane Clinchant. ECIR 2023 (​​https://arxiv.org/abs/2205.02870)

[12] Splate: Sparse late interaction retrieval. Thibault Formal, Stéphane Clinchant, Hervé Déjean, Carlos Lassance. SIGIR 2024 (https://arxiv.org/abs/2404.13950) 

Please note that applicants must be registered students at a university or other academic institution and that this establishment will need to sign an 'Internship Convention' with NAVER LABS Europe before the student is accepted.

You can apply for this position online. Don't forget to upload your CV and cover letter before you submit. Incomplete applications will not be accepted.

NAVER is the #1 Internet portal in Korea with activities that span a wide range of businesses including search, commerce, content, financial and cloud platforms.

NAVER LABS, co-located in Korea and France, is the organization dedicated to preparing NAVER’s future. NAVER LABS Europe is located in a spectacular setting in Grenoble, in the heart of the French Alps. Scientists at NAVER LABS Europe are empowered to pursue long-term research problems that, if successful, can have significant impact and transform NAVER. We take our ideas as far as research can to create the best technology of its kind. Active participation in the academic community and collaborations with world-class public research groups are, among others, important tools to achieve these goals. Teamwork, focus and persistence are important values for us.

NAVER LABS Europe is an equal opportunity employer.