The article introduces Wan2.1, a suite of open video foundation models excelling in various tasks like Text-to-Video and Image-to-Video generation. It highlights key features such as SOTA performance on consumer-grade GPUs, support for multiple tasks, and efficient video VAEs. The I2V-14B model, capable of generating 720P videos, is noted for its superior performance across benchmarks.
This article discusses the use of variational autoencoders (VAEs) to generate synthetic data as a solution to the impending data scarcity for training large language models. It explores how synthetic data can address issues like imbalanced datasets, particularly using the UCI Adult dataset, by generating synthetic samples to balance the dataset and improve classification accuracy.
Generate realistic sequential data with this easy-to-train model. This article explores using Variational Autoencoders (VAEs) to model and generate time series data. It details the specific architecture choices, like 1D convolutional layers and a seasonally dependent prior, used to capture the periodic and sequential patterns in temperature data.
