mlsanityData to 14 Oct 2026

Task-Conditional Flow Matching for Balanced Multilingual Text Embedding Adaptation

arXiv 2608.05785 ↗First posted 6 Aug 2026cs.CL cs.AIcited by 0

On the map: Diffusion models → Discrete and text diffusion

Abstract (arXiv)

Multilingual text embedding models are commonly adapted using a single training objective across diverse tasks, despite different tasks requiring fundamentally different optimization strategies. We introduce Task-Conditional Flow Matching (TCFM), a multilingual embedding adaptation framework that selectively applies Flow Matching to translation tasks while optimizing retrieval, classification, and pair-classification tasks with objectives better aligned to their learning dynamics. TCFM further combines teacher-guided representation preservation with a three-stage curriculum to enable stable adaptation. Evaluated on the Indic Massive Text Embedding Benchmark, TCFM establishes a new state-of-the-art, consistently improving embedding quality across a diverse set of multilingual tasks and generalizing across embedding model families. We will publicly release the codebase and datasets upon acceptance of the paper.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Discrete and text diffusion, one of 14 directions in Diffusion models. Builds on 21 of its 27 references that are in the map. First posted on arXiv 6 Aug 2026.

Builds on

21 of 27 references.

References · 21

  1. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  2. 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
  3. 2016Layer Normalizationcited by 5,669
  4. 2019Unsupervised Cross-lingual Representation Learning at Scalecited by 3,935
  5. 2021SimCSE: Simple Contrastive Learning of Sentence Embeddingscited by 2,173
  6. 2025Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Modelscited by 1,261
  7. 2018Experience Replay for Continual Learningcited by 1,095
  8. 2019How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddingscited by 784
  9. 2020Language-agnostic BERT Sentence Embeddingcited by 627
  10. 2018Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyondcited by 434
  11. 2020Understanding the Behaviour of Contrastive Losscited by 398
  12. 2022One Embedder, Any Task: Instruction-Finetuned Text Embeddingscited by 333
9 more
  1. 2024mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrievalcited by 236
  2. 2021Muppet: Massive Multi-task Representations with Pre-Finetuningcited by 192
  3. 2022MASSIVE: A 1M-Example Multilingual Natural Language Understanding Dataset with 51 Typologically-Diverse Languagescited by 146
  4. 2025EmbeddingGemma: Powerful and Lightweight Text Representationscited by 139
  5. 2021Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languagescited by 136
  6. 2020Balancing Training for Multilingual Neural Machine Translationcited by 72
  7. 2022Exploring the Impact of Negative Samples of Contrastive Learning: A Case Study of Sentence Embeddingcited by 25
  8. 2023Bhasha-Abhijnaanam: Native-script and romanized Language Identification for 22 Indic languagescited by 13
  9. 2025IndicRAGSuite: Large-Scale Datasets and a Benchmark for Indian Language RAG Systemscited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026jina-embeddings-v5-text: Task-Targeted Embedding Distillation
  2. 2025Benchmarking Cross-Lingual Semantic Alignment in Multilingual Embeddings
  3. 2025Improving Multilingual Language Models by Aligning Representations through Steering
  4. 2026Less is More: Adapting Text Embeddings for Low-Resource Languages with Small Scale Noisy Synthetic Data
  5. 2025Middle-Layer Representation Alignment for Cross-Lingual Transfer in Fine-Tuned LLMs
  6. 2025KaLM-Embedding: Superior Training Data Brings A Stronger Embedding Model
  7. 2026F2LLM-v2: Inclusive, Performant, and Efficient Embeddings for a Multilingual World
  8. 2024Enhancing Cross-Language Code Translation via Task-Specific Embedding Alignment in Retrieval-Augmented Generation