arXiv:2608. 06516v1 Announce Type: cross Abstract: Lightweight connectors make frozen multimodal encoders composable at the representation level.
arXiv:2608. 19492v1 Announce Type: new Abstract: World models increasingly treat compact multimodal representations as interfaces between perception and physical interaction, yet existing probes do not establish whether different sensors carry the same executable meaning or whether that meaning survives a new action composition.
Hyper-LLaVA introduces a hyperbolic uncertainty‑aware, modality‑balanced routing mechanism for multimodal continual instruction tuning. It improves intra‑modality task matching by evaluating sample‑to‑task distribution similarity in hyperbolic space and mitigates unreliable modality influence by quantifying task‑matching ambiguity to adaptively balance across modalities. The approach achieves significant performance gains over existing methods.
arXiv:2610.20532v1 Announce Type: cross Abstract: A multimodal agent must decide whether visual evidence is worth acquiring, which candidate or resolution to query, and when further interaction can n...
ICE (Interaction-aware Clifford Encoder) is a multimodal graph foundation model that uses a node-indexed Clifford latent field to encode topology, text, and images into explicit Cl(3) addresses. Edge-aware geometric products transform these directions into scalar, bivector, and trivector relations over observed neighborhoods, preserving entity semantics while enabling higher-order transport and direct field access. Across eleven graphs and multiple node‑classification, link‑prediction, and few‑shot tasks, ICE outperforms all 30 reported supervised and few‑shot comparisons, with core removals and mechanism controls demonstrating the importance of its higher‑order structure and semantic protection.
arXiv:2609.17094v1 Announce Type: new Abstract: Multimodal representation learning seeks shared representations for cross-modal retrieval and knowledge transfer. Hub-based binding reduces pairwise su...
Multimodal representation learning seeks shared representations for cross-modal retrieval and knowledge transfer. Hub-based binding reduces pairwise supervision costs, but separate hub connections can...
Enterprises deploy systems, not checkpoints. Usable capability depends jointly on weights, serving route, precision, output contract, and harness, yet all 18 audited benchmarks score advertised model...
arXiv:2609.06668v1 Announce Type: new Abstract: Multimodal graphs couple node attributes in different modalities, such as text and images, with relational structure, enabling topological structure an...
arXiv:2606. 07316v2 Announce Type: replace-cross Abstract: Can a committee of LLM agents reach agreement that is certifiable at the level of meaning, not only at the level of a label?
The paper evaluates Certo, a small non‑generative decision model that scores candidate actions based on text. It compares a joint scorer that processes state, rules, and candidates together with a cacheable encoder that pre‑encodes candidates to reduce cost. Experiments show the cacheable approach loses rule sensitivity, while targeted counterfactual supervision can recover performance on synthetic tasks; however, on real rules the joint scorer still outperforms the cacheable version, and cross‑domain mixtures do not improve accuracy.
arXiv:2606. 02581v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) faces a fundamental three-way tension: deeper retrieval improves factual grounding but inflates token costs and end-to-end latency.
arXiv:2609.10494v1 Announce Type: new Abstract: Enterprises deploy systems, not checkpoints. Usable capability depends jointly on weights, serving route, precision, output contract, and harness, yet...