If you need more data for this, I've gotchu

[Hellisotherpeople]

The DebateSum Dataset was made from a huge corpus of Policy Debate evidence. Much of that evidence is BS post modernist philosophy.

https://huggingface.co/datasets/Hellisotherpeople/DebateSum and https://paperswithcode.com/paper/debatesum-a-large-scale-argument-mining-and

[Kahsolt]

Hello again! Thanks for the materials~

This repo is suspended even deprecated, and could not possibly serve as a running application. De facto, I created this repo just to make fun of my friends reading and rewriting semi-academical philosophy essays. As you know could see, train on their writings and generate absolutely insane nonsense. 🤣

Still, textgenrnn is a simple RNN-based model 5 years ago, therefore its performance is far behind the modern large-scale pretrained SOTA transformers like GPT-2. Anyone wanna a helpful text-generation application should do model refinement from those. Of course, a powerful GPU and mountainous electricity bills are inevitable. :(

[Kahsolt]

Recently I just start to explorer things on hugging face, really nice platform but the documentation is somewhat too brief, ambiguous and hard for me to understand. One day I get to know how to play with it, I will try refining DebateSum, many many thanks for the dataset~