411–415 of 415 posts
Earlier quoted context omitted.
There's a partial fix for the seed issue on Reddit.
I can't seem to find it, do you have a link?
How long does it take to generate a single image? Is it in the 30 min type range or a few mins? It's hypothetically "possible" to run e.g. OPT175B on a consumer GPU via Huggingface Accelerate, but in practice it takes like 30 mins to generate a single token.
Earlier quoted context omitted.
It isn't required for some (most?) users. Weirdly sometimes pip is picking up the wheel for `onnx`, sometimes it isn't, and we can't figure out why. Any Python packaging experts know what's going on? all macOS 12, arm64, Python 3.10. Can't think it wouldn't resolve the wheel. But yes, good idea to move up. I'll stick it next to the `pip install`.
Please consider also adding a small note to help those few that get stuck with this bug: RuntimeError: expected scalar type BFloat16 but found Float The solution is easy: append the execution command with `--precision full`
Thank you!