We introduce generative infinite-vocabulary transformers (GIVT) which generate vector sequences with real-valued entries, instead of discrete tokens from a finite vocabulary. To …
Q Liu, X Dong, J Xiao, N Chen, H Hu, J Zhu… - arXiv preprint arXiv …, 2024 - arxiv.org
Vector quantization, renowned for its unparalleled feature compression capabilities, has been a prominent topic in signal processing and machine learning research for several …
Codebook collapse is a common problem in training deep generative models with discrete representation spaces like Vector Quantized Variational Autoencoders (VQ-VAEs). We …
D Schmidt, M Jiang - arXiv preprint arXiv:2312.10812, 2023 - arxiv.org
Pre-training large models on vast amounts of web data has proven to be an effective approach for obtaining powerful, general models in several domains, including language …
Similar to natural language models, pre-trained genome language models are proposed to capture the underlying intricacies within genomes with unsupervised sequence modeling …
Y Xu, X Feng, F Qin, R Ge, Y Peng, C Wang - arXiv preprint arXiv …, 2024 - arxiv.org
Implicit neural representations (INR) excel in encoding videos within neural networks, showcasing promise in computer vision tasks like video compression and denoising. INR …
J Birk, A Hallin, G Kasieczka - arXiv preprint arXiv:2403.05618, 2024 - arxiv.org
Foundation models are multi-dataset and multi-task machine learning methods that once pre- trained can be fine-tuned for a large variety of downstream applications. The successful …
The Large Language models (LLMs) have demonstrated supreme capabilities in text understanding and generation, but cannot be directly applied to cross-modal tasks without …