# PEGASUS decoder with input as PEGASUS's full autoencoder output

**URL:** <https://discuss.huggingface.co/t/pegasus-decoder-with-input-as-pegasuss-full-autoencoder-output/17034>\
**Category:** Beginners\
**Created:** [April 20, 2022, 11:17pm UTC](https://discuss.huggingface.co/t/pegasus-decoder-with-input-as-pegasuss-full-autoencoder-output/17034 "2022-04-20T23:17:09Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Dusan](https://avatars.discourse-cdn.com/v4/letter/d/9fc348/32.png) [@Dusan](https://discuss.huggingface.co/u/Dusan)\
**Post date:** [April 20, 2022, 11:17pm UTC](https://discuss.huggingface.co/t/pegasus-decoder-with-input-as-pegasuss-full-autoencoder-output/17034/1 "2022-04-20T23:17:09Z")

</div>

Hello, I have fine-tuned my Pegasus model (for text summarization) to generate short summaries on the input data I have given it. My question is, can I somehow use this short summary output from Pegasus autoencoder as an input to the decoder of Pegasus and make it generate some larger output (with beam search I assume)?