Beyond Layers: Position-Resolved Gradient Conflict and Position-Aware Modulation for Unified Multimodal Models
cs.CV, cs.LG
Submitted: 2026-09-29
Updated: 2026-09-29
Terminology
Sources
- Understanding intermediate layers using linear classifier probes
- LoRA Learns Less and Forgets Less
- MaskGIT: Masked Generative Image Transformer
- Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling
- Just Pick a Sign: Optimizing Deep Multitask Models with Gradient Sign Dropout
- SemHiTok: A Unified Image Tokenizer via Semantic-Guided Hierarchical Codebook for Multimodal Understanding and Generation
- Emerging Properties in Unified Multimodal Pretraining
- Adapting Auxiliary Losses Using Gradient Similarity
- Taming Transformers for High-Resolution Image Synthesis
- MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models
- Making LLaMA SEE and Draw with SEED Tokenizer
- GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment
- WinTok: A Win-Win Hybrid Tokenizer via Decomposing Visual Understanding and Generation with Transferable Tokens
- In Defense of the Unitary Scalarization for Deep Multi-Task Learning
- Playground v2.5: Three Insights towards Enhancing Aesthetic Quality in Text-to-Image Generation
- EvoTok: A Unified Image Tokenizer via Residual Latent Evolution for Visual Understanding and Generation
- Mixture-of-Transformers: A Sparse and Scalable Architecture for Multi-Modal Foundation Models
- Rho-1: Not All Tokens Are What You Need
- Conflict-Averse Gradient Descent for Multi-task Learning
- Second-Order Multi-Level Variance Correction for Modality Competition in Multimodal Models
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models