IEEE

A Novel Hybrid Architecture With Fast Lightweight Encoder and Transformer Under Attention Fusion for the Enhancement of Sand Dust and Haze Image Restoration

Abstract: Outdoor weather conditions such as haze, fog, sand dust, and low light significantly degrade image quality, causing color distortions, low contrast, and poor visibility. In spite of the ...
IEEE

U-MusT: A Unified Framework for Cross-modal Translation of Score Images, Symbolic Music, and Performance Audio

Abstract: Music exists in various modalities, such as score images, symbolic scores, MIDI, and audio. Translations between such modalities are established as core tasks of music information retrieval, ...
GitHub

Add HumanV: decoder-only causal LM #42982

Dictionary containing the configuration parameters for the RoPE embeddings. Must include `rope_theta`. Dictionary containing the configuration parameters for the RoPE embeddings. attention_bias ...
marktechpost

Google Introduces T5Gemma 2: Encoder Decoder Models with Multimodal Inputs via SigLIP and 128K Context

T5Gemma 2 follows the same adaptation idea introduced in T5Gemma, initialize an encoder-decoder model from a decoder-only checkpoint, then adapt with UL2. In the above figure the research team show ...
GitHub

Zero-Training Context Extension for Transformer Encoders via Nonlinear Absolute Positional Embeddings Interpolation

Official implementation of "Zero-Training Context Extension for Transformer Encoders via Nonlinear Absolute Positional Embeddings Interpolation". Paper preprint is coming soon. This implementation ...