# What are task prefixes for all Conditional Generation Models?

**URL:** <https://discuss.huggingface.co/t/what-are-task-prefixes-for-all-conditional-generation-models/8811>\
**Category:** Beginners\
**Created:** [July 29, 2021, 5:13pm UTC](https://discuss.huggingface.co/t/what-are-task-prefixes-for-all-conditional-generation-models/8811 "2021-07-29T17:13:35Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![prikmm](https://avatars.discourse-cdn.com/v4/letter/p/a698b9/32.png) [@prikmm](https://discuss.huggingface.co/u/prikmm)\
**Post date:** [July 29, 2021, 5:13pm UTC](https://discuss.huggingface.co/t/what-are-task-prefixes-for-all-conditional-generation-models/8811/1 "2021-07-29T17:13:35Z")

</div>

I am currently fiddling around with different transformer architectures for summarization task, and I got curious as to what are task prefixes of pretrained Conditional Generation Models. I know T5’s prefixes and found this [website](https://simpletransformers.ai/docs/t5-specifics/) pertaining to the same. But, I have not been able to find anything for Bart, Pegasus, ProphetNet etc. I know that if we are training the model from scratch, then we can have our own task prefixes. But, would like to know of pretrained models present on huggingface