I am using prodigy version 1.11.13, and its en_core_web_trf model doesn't have tok2vec component because of it, i cant train it on my custom training dataset however i tried to add tok2vec manually through tok2vec documentation and its started training on cpu but whenever i try to run it on GPU it throughs error...need help
Welcome to the forum @shubhams9502 
This issue was fixed in Prodigy 1.15.1 and then for textcat in 1.15.7 - any chance you could upgrade your Prodigy version?
Thank you for your prompt response.
Given that we have purchased the product, we expect to have full access to all its features without encountering any bugs. Additionally, we are currently unable to update beyond version 1.12.
Could you please assist us in resolving these issues so that we can fully utilize the product as intended?
Given that we have purchased the product, we expect to have full access to all its features without encountering any bugs. Additionally, we are currently unable to update beyond version 1.12.
Prodigy is sold as a once-off purchase with a lifetime license. We additionally offer 12 months of free updates.
This is an extremely unusual and generous structure. Version 1.12 was released in July 2023, so if you don't have access to it your license purchase must have been before July 2022.
The reason why we offer the lifetime license is we think it's a shame to lock people out of their previous projects. Often projects can sit idle a long time, and it's reasonable to want to go back and reproduce their work. We think this shouldn't require a subscription, so we offer a license that makes this possible.
However if you're actively using your version that you purchased two years ago, we can't support that without an update. I don't think this is unreasonable.
I understand the structure and appreciate the generous offer of a lifetime license. However, I’m currently in the process of training a model and am encountering a significant bug that I cannot resolve with version 1.12. Given the circumstances, I’m in need of immediate support to move forward.
Could you please provide guidance or assistance to help address this issue? Your support in resolving the bug or finding a workaround would be greatly appreciated.
My understanding is that the bug is already resolved in the latest version. If your problem is that you don't have access to that version because it's outside of your maintenance window, then the only solution we can offer is for you to purchase the maintenance extension.
This is after all what the maintenance window means: that you'll have access to new fixes and improvements. Other users are paying the license fees in order to support continued development of the product, and it wouldn't be fair to give the same for free to other users.
1 Like