README.md

This is a pre-train baseline model for summarization. Input is to concatenate all articles in one cluster.
the example.json is the example result.

pipeline:
input -> sum tokenizer -> perceiver -> sum model -> summary