Franz Srambical
Franz Srambical
I am indeed using the HEAD version.
I'm going to try to turn MaxText into encoder-decoder anyway, so native support is of course also appreciated :)
https://github.com/p-doom/maxtext/tree/colab_temp We finally came around to implement encoder-decoder models in our maxtext fork. The synthetic data pipeline seems to work. Will add support for the real data pipeline later today.
okay I was a bit too fast, still have to fix a few things.
lol my pr getting scooped haha; if you fix the `libcusparse` typo, i can fully close my pr (#553).
Yes, here: https://github.com/google/maxtext/blob/ef9a6e9b96aef039797ecfcd20dd9ec554f8e387/setup.sh#L106-L108
I'll try to do it if I find time but I am super busy these days and it's been some time since I used maxtext (and that was on a...
But I think the error message to be expected is clear: We don't have access to that link, thus we need to just use pypi instead
fwiw that worked for me (iirc)
For what it's worth, the history bug seems to be fixed for me as well when setting fission.bfcacheInParent to false.