ICLR2025

VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language …
PDF Cite Code Project Model Demo
VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
LongVILA: Scaling Long-Context Visual Language Models for Long Videos
Long-context capability is critical for multi-modal foundation models, especially for long video understanding. We introduce LongVILA, …
PDF Cite Code Project Model
LongVILA: Scaling Long-Context Visual Language Models for Long Videos
SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
Diffusion models have been proven highly effective at generating high-quality images. However, as these models grow larger, they …
PDF Cite Code Project Video Demo Blog
SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
COAT: Compressing Optimizer states and Activation for Memory-Efficient FP8 Training
FP8 training has emerged as a promising method for improving training efficiency. Existing frameworks accelerate training by applying …
PDF Cite Code Project Demo
COAT: Compressing Optimizer states and Activation for Memory-Efficient FP8 Training
Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models
We present Deep Compression Autoencoder (DC-AE), a new family of autoencoder models for accelerating high-resolution diffusion models. …
PDF Cite Code Project
Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models
DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
Deploying long-context large language models (LLMs) is essential but poses significant computational and memory challenges. Caching all …
PDF Cite Code Project
DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
HART: Efficient Visual Generation with Hybrid Autoregressive Transformer
We introduce Hybrid Autoregressive Transformer (HART), an autoregressive (AR) visual generation model capable of directly generating …
PDF Cite Code Project Demo
HART: Efficient Visual Generation with Hybrid Autoregressive Transformer
SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer
We introduce Sana, a text-to-image framework that can efficiently generate images up to 4096x4096 resolution. Sana can synthesize …
PDF Cite Code Project MIT Project Demo
SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer