The paper uncovers a systematic mismatch in modern text embeddings: while sentiment can be extracted with high accuracy from frozen encoders, the embeddings’ cosine similarity metric barely reflects polarity differences. Across eleven models from 2021–2026, a linear probe achieves up to 91.3% accuracy, yet cosine similarity margins are only 0.003–0.064, leading to poor retrieval performance (e.g., Qwen3-Embedding-4B’s 64.3 MAP). Introducing a lightweight supervised‑contrastive head (Sentiment‑Injected Sentence Embeddings) dramatically improves probe accuracy, retrieval MAP, and polarity margins, but at the cost of downstream semantic similarity.
"whyItMatters":"The findings reveal that widely used retrieval and clustering systems may ignore valuable sentiment information encoded in embeddings, impacting their effectiveness."
By Mahdi Naser Moghadasi
arXiv:2610.34901v1 Announce Type: cross
Abstract: Training-free few-shot adaptation of vision-language models typically constructs a classifier from CLIP's final projected image representation, leavi...
By Pavodi Maniamfu, Vogel Kiketa, Zezina Kuzulu Kuetu, Jakin Mukendi Kabongo, David Mayoya Kutangila, Pescie Honore Kinkani Mampuya, L'Or Wolo Sieka
The paper introduces Spectral Transductive Refinement (STR), a training‑free method that refines class prototypes at test time using the geometry of a joint k‑nearest‑neighbour graph and a normalized‑Laplacian spectral coordinate system. STR operates solely on frozen visual embeddings, iteratively updating pseudo‑labelled queries to improve one‑shot and few‑shot classification under domain shift. Experiments on ResNet‑18 and ResNet‑10 backbones show STR outperforms single‑prototype baselines and rivals meta‑trained cross‑domain few‑shot methods, achieving the best 1‑shot average across eight target domains.
By Fahim Rahman, S. M. Tanjeeb Meheran Rohan, Md. Taimum Ibne Sayed, Asaduzzaman Herok, Md. Bakhtiar Hasan
ORCA (Orthogonal Residual Compositional Alignment) is a method that improves text-to-image diffusion models by aligning the diffusion transformer’s latent space with a low‑rank target derived from a frozen visual encoder. It introduces an auxiliary loss that uses a predictor with an orthogonal basis parameterised by a learned residual between T5 and CLIP embeddings, providing a prompt‑dependent signal for selecting the visual readout subspace. Experiments on three diffusion‑transformer backbones show that ORCA improves FID and GenEval scores, especially on attribute binding, spatial relations, and multi‑object prompts, without adding inference‑time cost.
By Arshia Hemmat, Amirhossein Vahidi, Amitis Shidani, Mohammad Vali Sanian, Hesam Asadollahzadeh, Aryan Yazdan Parast, Mohammad Lotfollahi
CoVA introduces a lightweight framework that composes VAE features online to provide stage‑wise guidance for flow‑matching training, improving semantic discrimination and spatial structure without extra inference overhead. By jointly training a small latent composer with class supervision, masked reconstruction, and flip consistency, CoVA aligns intermediate SiT features early and later uses detached composer posteriors for sample‑dependent soft‑label conditioning. On ImageNet 256×256, CoVA cuts SiT‑XL/2 FID from 17.2 to 7.4 in 400K steps and achieves an FID of 2.03 in only 400 epochs with classifier‑free guidance, adding just 8.5 M training‑only parameters.
By Zhaowei Chen, Yikai Ge, Linze Li, Yao Tang, Xiang Li
Omni-Embed-Mini is a 0.9B‑parameter model that embeds text, speech, audio, images, video, and visually‑rich documents into a single shared cosine space without updating any text‑side parameters. It uses a dense cascaded caption as a teacher signal, allowing the teacher and student to share identical backbone weights and requiring only lightweight projectors and phased LoRA adapters for alignment. The model achieves strong text retrieval performance (49.57 nDCG@10 on MTEB‑v2 BEIR‑8) while extending to five additional modalities and is significantly smaller than other open omni‑modal embedders.
By Mohammed Irfan Kurpath, Jaseel Muhammad Kaithakkodan, Sahal Shaji Mullappilly, Ivan Laptev, Hisham Cholakkal
arXiv:2606. 00129v1 Announce Type: cross Abstract: Large language models (LLMs) have emerged as powerful representation learners whose internal features increasingly align with human cognition.
By Yousef A. Radwan, Xuhui Liu, Kilichbek Haydarov, Yuqian Fu, Mohamed Elhoseiny
The paper investigates how the modality gap— the separation between image and text representations in contrastive vision‑language models—affects different downstream tasks. By showing that a single dominant direction accounts for most of the image‑text mean separation, the authors explain why reducing or removing this gap can improve zero‑shot classification, degrade retrieval, or restore performance depending on the task. The study provides a geometric framework that clarifies when and why gap interventions should be applied in vision‑language systems.
By Aditya Sharma, Divya Saxena
SCOUT is a frozen‑encoder approach for sim‑to‑real text‑based person retrieval that predicts cross‑modal embeddings instead of fine‑tuning cross‑encoders. It uses a trainable predictor to map patch tokens from a frozen video encoder (V‑JEPA) into the embedding space of a frozen text encoder (EmbeddingGemma), guided by a bidirectional InfoNCE objective. The method achieves state‑of‑the‑art results on the AI City Challenge 2026 Track 4, with a full retrieve‑fuse‑rerank pipeline reaching 84.25 mAP@10 and a single frozen model alone scoring 60.63, while training costs are modest (≈95 GPU‑hours).
By Abdarahmane Traor\'e, Andy Couturier, \'Eric Hervet
arXiv:2610.02718v1 Announce Type: cross
Abstract: Vision-language models such as CLIP exhibit strong semantic generalization, but remain limited in fine-grained visual perception. A recent work named...
By Peilin Yang, Xiaoyu Liu, Jian Sun, Qinghua Tao
arXiv:2606. 03879v1 Announce Type: cross Abstract: As foundation models scale toward fusing more heterogeneous visual streams, understanding how diverse encoders interact under joint training becomes a prerequisite for principled design.
By Wei Ding, Yudong Zhang, Ruobing Xie, Xingwu Sun, Jiansheng Chen, Yu Wang
The paper revisits the use of contrastive loss for post‑training CLIP models, showing that the reported catastrophic forgetting is mainly due to an overly large temperature parameter in the InfoNCE objective. By setting a smaller temperature, contrastive post‑training can actually improve CLIP. The authors introduce ComCLIP, a lightweight single‑epoch recipe that freezes the text encoder, trains the vision encoder with a properly tempered contrastive loss, an MSE anchoring loss, and a relational distillation loss, achieving comparable zero‑shot classification to CLIP‑Refine while significantly enhancing visual feature transferability and maintaining downstream compatibility with LLaVA.
By Zidan Wang, Yaqian Li, Xiaokai Zhang, Kaiwen Long, Kun He, Hanpeng Liu