collaborators

8 papers

cs.LG2026

Optimize Cheap, Deploy Strong: Cost-Aware Cross-Tier Transfer for Evolutionary Optimization

Tal Oved, Roi Pony, Oshri Naparstek +1

Evolutionary optimization of LLM prompts and agentic programs (e.g., GEPA) is dominated by fitness evaluation: scoring each candidate runs an answering LLM over a validation set, s…

cs.IR2026

FLASH-MAXSIM: IO-Aware Fused Kernels for Late-Interaction Retrieval

Roi Pony, Daniel Ezer, Adi Raz Goldfarb +3

Late-interaction retrieval (ColBERT, ColPali) scores a query against a document via the MaxSim operator. The standard PyTorch implementation materialises the full query-token x doc…

cs.IR2026

Col-Bandit: Query-Time Top- Estimation for Late-Interaction Retrieval

Roi Pony, Adi Raz Goldfarb, Oshri Naparstek +3

Multi-vector late-interaction retrievers such as ColBERT achieve state-of-the-art quality, but their query-time cost is dominated by exhaustively computing token-level MaxSim inter…

cs.CV2026

Is the Modality Gap a Bug or a Feature? A Robustness Perspective

Rhea Chowers, Oshri Naparstek, Udi Barzelay +1

Many modern multi-modal models (e.g. CLIP) seek an embedding space in which the two modalities are aligned. Somewhat surprisingly, almost all existing models show a strong modality…

cs.CL2026

Projected Autoregression: Autoregressive Language Generation in Continuous State Space

Oshri Naparstek

Standard autoregressive language models generate text by repeatedly selecting a discrete next token, coupling prediction with irreversible commitment at every step. We show that to…

cs.LG2025

Complexity as Advantage: A Regret-Based Perspective on Emergent Structure

Oshri Naparstek

We introduce Complexity as Advantage (CAA), a framework that defines the complexity of a system relative to a family of observers. Instead of measuring complexity as an intrinsic p…