Colab error and how many recycles possible?

[avilella]

Hi, I got this error from attempting to run the colab:

I also wanted to ask how many recycles would be possible to run with two sequences, one 150aa long and the other 250aa long before running out of memory in a Colab (not the pro one).

Thanks

[ZiyaoLi]

would you please show me the outputs in the first block of the notebook, to check if Unicore is appropriately installed? Seems to me that the installation of unicore is failed. For recycling there is no upper bound, because it doesn’t cost more memory but more time.

[avilella]

It seems to download it correctly, but not install correctly:

[ZiyaoLi]

thx for the feedback.

colab (again) updated their python version from 3.8 to 3.9.16, which made the installation failed.

Will PR a fix, but for quicker you may change the cp38s in the unicore url to cp39s.

[avilella]

There were 4 occurrences of the token 'cp38', which I changed to 'cp39', then started a multimer execution with 3 proteins, max_recycling_iters: 9, num_ensembles: 2 , manual_seed: 42, times: 3. It started the prediction step, so I think it's working now.

…

On Tue, Mar 21, 2023 at 6:28 AM Ziyao Li ***@***.***> wrote: thx for the feedback. colab (again) updated their python version from 3.8 to 3.9.16, which made the installation failed. Will PR a fix, but for quicker you may change the cp38s in the unicore url to cp39s. — Reply to this email directly, view it on GitHub <#107 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABGSN3XO6TMGTG2NU43J2LW5FC7VANCNFSM6AAAAAAVWOMRKE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ZiyaoLi]

should be fixed in #109 .

[burunduk57]

I just checked, and it works as is, without any editing. Thank you very much!

[busrasavas]

Hi, I got the same error about the unicore installation. Checked the unicore url, it seems updated. I attached the outputs of the related cells, I'd appreciate any help. Thanks in advance!