Egil Martinsson
Egil Martinsson
Thank you! 🙏 Agree & happy to update `requirements.txt`, but preferably in another PR? If we can both dodge the great debates of `conda` vs `pip` _and_ "what goes into...
(Merged in #92596 without diff since that PR removed `future` from both README instruction and requirements.txt)
Thank you! I merged in https://github.com/pytorch/pytorch/pull/92738 with no diff since that removed the install instruction on `cffi`. This PR has the same effect since it's not in the `requirements.txt`
Only failing [unit test](https://github.com/pytorch/pytorch/actions/runs/3988450270/jobs/6840489776) is unrelated! Anything blocking merge? 🙏 @ZainRizvi @malfet
Thank you! I think it should work now 🙏
Hi! Sorry I thought I answered this issue but must have forgot to hit send! I think its because in the real world, the further from an event (here a...
Hi there, That's a correct observation. Initialization is the absolutely most important reason for exploding gradient. If it's far away initially it'll take a huge gradient-step into one direction leading...
@gabrielgonzaga yes in those months I think some high-frequency committer churned or something but yes it suddenly exploded lol. It'll be addressed in https://github.com/ragulpr/wtte-rnn/issues/41 until then just find the right...