Pavlo Molchanov

ProfBench: Multi-Domain Rubrics requiring Professional Knowledge to Answer and Judge

Fast-dLLM v2: Efficient Block-Diffusion LLM

3D Aware Region Prompted Vision Language Model

DLER: Doing Length pEnalty Right — Incentivizing More Intelligence per Token via Reinforcement Learning

Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed

ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestration

OmniVinci: Enhancing Architecture and Data for Omni-Modal Understanding LLM

BroRL: Scaling Reinforcement Learning via Broadened Exploration

Compact GSPN: Scaling Spatial Propagation to Vision Foundation Models

AgentHard: Hardening LLM-Agent Evaluation with a Taxonomy of Artifacts and Automated Cleaning

Efficient Hybrid Language Model Compression through Group-Aware SSM Pruning

Hymba: A Hybrid-head Architecture for Small Language Models

Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Models

Scaling RL to Long Videos

GSPN-2: Efficient Parallel Sequence Modeling

LongMamba: Enhancing Mamba's Long-Context Capabilities via Training-Free Receptive Field Enlargement

LongVILA: Scaling Long-Context Visual Language Models for Long Videos

FeatSharp: Your Vision Model Features, Sharper

LLaMaFlex: Many-in-one LLMs via Generalized Pruning and Weight Sharing

Puzzle: Distillation-Based NAS for Inference-Optimized LLMs

Minifinetuning: Low-Data Generation Domain Adaptation through Corrective Self-Distillation

PHI-S: Distribution Balancing for Agglomerative Models

LLM Pruning and Distillation in Practice

ZoomVLM: A Tuning-Free Framework for Efficient Video Understanding via Adaptive Zooming in Vision-Language Models

LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models

X-VILA: Cross-Modality Alignment for Large Language Models

VILA^2: VLM Augmented VLM with Self-Improvement

UNAST: Unified framework for Neural Architecture Search for Transformers