TWN attaches separate reasoning and embedding LoRA adapters to a frozen backbone with gradient detachment and a self-supervised gate that decides per input whether to generate CoT, achieving SOTA on MMEB-V2 with 3-5% added parameters and up to 50% fewer reasoning tokens.
arXiv preprint arXiv:2510.05014 , year=
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 2
baseline 1
citation-polarity summary
representative citing papers
CodeMMR creates a unified embedding space for text, code, and images, outperforming baselines by 10 nDCG@10 points and boosting RAG code generation quality.
Bottleneck Tokens paired with a masked generative objective achieve state-of-the-art unified multimodal retrieval performance among 2B-scale models on the MMEB-V2 benchmark with 78 datasets.
PLUME uses latent-state autoregressive rollouts and a progressive training curriculum to deliver efficient reasoning for universal multimodal embeddings without generating explicit rationales.
Text-only contrastive fine-tuning of an MLLM with hard negatives produces embeddings that handle temporal, negation, and multimodal nuances in video retrieval and achieves SOTA performance.
TTE-Flash trains latent think tokens with CoT generation loss and embedding tokens with contrastive loss to deliver high-performance multimodal representations without generating explicit reasoning at inference time.
CausalEmbed uses auto-regressive generation with iterative margin loss to produce multi-vector embeddings that reduce visual token counts 30-155x while retaining competitive performance on VDR benchmarks.