What services do you suggest to use for training in case I do not have 2 RTX8000 GPUs?

gitbooo / CrossViVit

This repository contains code for the paper "Improving day-ahead Solar Irradiance Time Series Forecasting by Leveraging Spatio-Temporal Context"

https://arxiv.org/abs/2306.01112

MIT License

68 stars 5 forks source link

What services do you suggest to use for training in case I do not have 2 RTX8000 GPUs? #3

Closed deep404 closed 12 months ago

deep404 commented 1 year ago

I tried to minimize the batch to 1 and train locally where I have a 3070 8Gb, but without any success

jaggbow commented 12 months ago

I am afraid that 8GB might be a bit too low. For reference, we used two RTX8000 GPUs which have 48GB of VRAM each to run a batch size of 10. You can find the checkpoints for the model for inference in our release.