🖼️00My high-res image-to-video kept OOMing — turns out I was decoding outside no_gradDEV Community: pytorch·shinji shimizu·3 days ago#EF5LhD5G#dev#decode#fullscreen#no_grad#peak#article+2 more🧰Tag tools✨Add tagA 96GB GPU couldn't run 1024x768 I2V (83.5 GiB peak). The 54 GiB wasn't the model — it was an autograd graph from consuming a lazy VAE decode iterator outside no_grad. One-line fix, peak drops to 29.5 GiB, flat up to 2048x1536.15s0Read later0Read More
