jwfromm opened a new pull request, #15811: URL: https://github.com/apache/tvm/pull/15811
It turns out that using datatypes besides `float32` when calculating timestep embeddings leads to substantial errors. The diffusers library just hard casts to float32 to avoid this and we should too. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
