mlsanityData to 14 Oct 2026

NLLB-CLIP -- train performant multilingual image retrieval model on a budget

arXiv 2309.01859 ↗First posted 4 Sep 2023cs.CVcited by 24

Not on the map. Why?

Abstract (arXiv)

Today, the exponential rise of large models developed by academic and industrial institutions with the help of massive computing resources raises the question of whether someone without access to such resources can make a valuable scientific contribution. To explore this, we tried to solve the challenging task of multilingual image retrieval having a limited budget of 1,000. As a result, we present NLLB-CLIP - CLIP model with a text encoder from the NLLB model. To train the model, we used an automatically created dataset of 106,246 good-quality images with captions in 201 languages derived from the LAION COCO dataset. We trained multiple models using image and text encoders of various sizes and kept different parts of the model frozen during the training. We thoroughly analyzed the trained models using existing evaluation datasets and newly created XTD200 and Flickr30k-200 datasets. We show that NLLB-CLIP is comparable in quality to state-of-the-art models and significantly outperforms them on low-resource languages.

Key facts. Cited by 24 papers in the map (data through 14 Oct 2026). Builds on 25 of its 31 references that are in the map. First posted on arXiv 4 Sep 2023.

Builds on

25 of 31 references.

References · 25

  1. 2017Attention Is All You Needcited by 60,361
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  5. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  6. 2022No Language Left Behind: Scaling Human-Centered Machine Translationcited by 1,312
  7. 2020The State and Fate of Linguistic Diversity and Inclusion in the NLP Worldcited by 798
  8. 2022PaLI: A Jointly-Scaled Multilingual Language-Image Modelcited by 766
  9. 2021LiT: Zero-Shot Transfer with Locked-image text Tuningcited by 577
  10. 2023Symbolic Discovery of Optimization Algorithmscited by 434
  11. 20218-bit Optimizers via Block-wise Quantizationcited by 403
  12. 2020Participatory Research for Low-resourced Machine Translation: A Case Study in African Languagescited by 159
13 more
  1. 2022CLIP is Also an Efficient Segmenter: A Text-Driven Approach for Weakly Supervised Semantic Segmentationcited by 118
  2. 2022Chinese CLIP: Contrastive Vision-Language Pretraining in Chinesecited by 109
  3. 2022Crossmodal-3600: A Massively Multilingual Multimodal Evaluation Datasetcited by 100
  4. 2023Exploring the Impact of Instruction Data Scaling on Large Language Models: An Empirical Study on Real-World Use Casescited by 86
  5. 2022AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilitiescited by 85
  6. 2018COCO-CN for Cross-Lingual Image Tagging, Captioning and Retrievalcited by 80
  7. 2022CLIP model is an Efficient Continual Learnercited by 65
  8. 2017STAIR Captions: Constructing a Large-Scale Japanese Image Caption Datasetcited by 47
  9. 2020Towards Zero-shot Cross-lingual Image Retrievalcited by 25
  10. 2022Adapting CLIP For Phrase Localization Without Further Trainingcited by 21
  11. 2021Contrastive Language-Image Pre-training for the Italian Languagecited by 19
  12. 2015Bridge Correlational Neural Networks for Multilingual Multimodal Representation Learningcited by 18
  13. 2019Impact of Training Dataset Size on Neural Answer Selection Modelscited by 7

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 11; 2025: 8; 2026: 5.

24 papers.

Sort:

By area:

  1. 2024NaturalBench: Evaluating Vision-Language Models on Natural Adversarial Samplescited by 97
  2. 2025Meta CLIP 2: A Worldwide Scaling Recipecited by 64
  3. 2024Two Effects, One Trigger: On the Modality Gap, Object Bias, and Information Imbalance in Contrastive Vision-Language Modelscited by 57
  4. 2024jina-clip-v2: Multilingual Multimodal Embeddings for Text and Imagescited by 37
  5. 2025X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIPcited by 21
  6. 2024Investigating and Mitigating Object Hallucinations in Pretrained Vision-Language (CLIP) Modelscited by 15
  7. 2024Multilingual Diversity Improves Vision-Language Representationscited by 14
  8. 2025SCAM: A Real-World Typographic Robustness Evaluation for Multimodal Foundation Modelscited by 13
  9. 2024Harnessing Frozen Unimodal Encoders for Flexible Multimodal Alignmentcited by 12
  10. 2024FLAME: Frozen Large Language Models Enable Data-Efficient Language-Image Pre-trainingcited by 12
  11. 2024mOSCAR: A Large-scale Multilingual and Multimodal Document-level Corpuscited by 9
  12. 2024Navigating Text-to-Image Generative Bias across Indic Languagescited by 5
12 more
  1. 2024Towards Zero-Shot Multimodal Machine Translationcited by 3
  2. 2024Uplifting Lower-Income Data: Strategies for Socioeconomic Perspective Shifts in Large Multi-modal Modelscited by 2
  3. 2025Breaking Language Barriers or Reinforcing Bias? A Study of Gender and Racial Disparities in Multilingual Contrastive Vision Language Modelscited by 2
  4. 2025Synthetic Document Question Answering in Hungariancited by 1
  5. 2025Multilingual Vision-Language Models, A Surveycited by 1
  6. 2025Mitigating Cross-modal Representation Bias for Multicultural Image-to-Recipe Retrievalcited by 1
  7. 2025uCLIP: Parameter-Efficient Multilingual Extension of Vision-Language Models with Unpaired Datacited by 1
  8. 2026Model Specific Task Similarity for Vision Language Model Selection via Layer Conductancecited by 0
  9. 2026M-MiniGPT4: Multilingual VLLM Alignment via Translated Datacited by 0
  10. 2026What Matters for Grocery Product Retrieval with Open Source Vision Language Modelscited by 0
  11. 2026Scaling Vision-Language Models Is Not Enough to Mitigate Biascited by 0
  12. 2026Where Do Multilingual Vision-Language Encoders Fail on Low-Resource Languages?cited by 0