A paper list of some recent works about Token Compress for Vit and VLM

[CVPR 2026 Highlight] A Frame is Worth One Token: Efficient Generative World Modeling with Delta Tokens

[CVPR 2026 Highlight] ReAG: Reasoning-Augmented Generation for Knowledge-based Visual Question Answering

Official repo for "Let ViT Speak: Generative Language-Image Pre-training"

timm, evolved

[CVPR 2026] Elucidating the SNR-t Bias of Diffusion Probabilistic Models

[CVPR 2026] - IsoCLIP: Decomposing CLIP Projectors for Efficient Intra-modal Alignment

🔥 LeetCode for PyTorch — practice implementing softmax, attention, GPT-2 and more from scratch with instant auto-grading. Jupyter-based, self-hosted or try online.

Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders [Technical Report]

[CVPR26 Findings] PEPR: Privileged Event-based Predictive Regularization for Domain Generalization

This is a research project to efficiently compress diffusion models, accepted at AAAI 2026.

[ICML'26] Code and website for Self-Flow: Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis

[WACV 2026] Official implementation of the paper: “CountingDINO: A Training-free Pipeline for Exemplar-based Class-Agnostic Counting”

[CVPR'26 Demo] Mobile-O: Unified Multimodal Understanding and Generation on Mobile Device

[ICLR 2026] - Spectral Concept Selection and Cross-modal Representation Learning for Generalized Category Discovery

Code for NeurIPS 2025 paper - Covariances for Free: Exploiting Mean Distributions for Training-free Federated Learning

Revisiting Multi-Task Visual Representation Learning

[ICML 2025] No Task Left Behind: Isotropic Model Merging with Common and Task-Specific Subspaces (official repository)

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

This repository contains code for the paper "Why Diffusion Models Don't Memorize: The Role of Implicit Dynamical Regularization in Training" by T. Bonnaire, R. Urfin, G. Biroli and M. Mézard.

Seeing Beyond Words: Self-Supervised Visual Learning for Multimodal Large Language Models

Official implementation of the paper: "FlowEdit: Inversion-Free Text-Based Editing Using Pre-Trained Flow Models"

Official inference repo for FLUX.2 models

[AAAI2026] Mitigating Negative Flips via Margin Preserving Training

[NeurIPS 2025] Official implementation of "Instance-Level Composed Image Retrieval".

[ICCV 2025] What Changed? Detecting and Evaluating Instruction-Guided Image Edits with Multimodal Large Language Models

[TMLR] Public code repo for paper "A Single Transformer for Scalable Vision-Language Modeling"