Setting num_images to less than the number of GPUs on the server returns more than num_images images (same number of images as the number of GPUs on the server) #59

omduggineni · 2022-06-30T17:24:04Z

This appears to be an issue in https://github.com/jina-ai/dalle-flow/blob/main/executors/dalle/dm_helper.py - images are always generated in batches equal to the number of GPUs on the server. Perhaps try utilizing multiple GPUs across multiple requests rather than across individual images in the same request?

omduggineni linked a pull request Jun 30, 2022 that will close this issue

(closes issue #59) hack: the server should always return num_images images #60

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Setting num_images to less than the number of GPUs on the server returns more than num_images images (same number of images as the number of GPUs on the server) #59

Setting num_images to less than the number of GPUs on the server returns more than num_images images (same number of images as the number of GPUs on the server) #59

Setting num_images to less than the number of GPUs on the server returns more than num_images images (same number of images as the number of GPUs on the server) #59

Setting num_images to less than the number of GPUs on the server returns more than num_images images (same number of images as the number of GPUs on the server) #59

Comments