---
license: mit
language: en
tags:
  - text-classification
  - emotion
widget:
  - text: "You love hurting me, huh?"
  - text: "I know good movies, this ain't one"
  - text: "It was fun, but I'm not going to miss you"
  - text: "My flight is delayed.. amazing."
  - text: "What is happening to me??"
  - text: "This is the shit!"
datasets: 
  - enryu43/twitter100m_tweets
---

Clone of [Uberduck/torchmoji](https://huggingface.co/Uberduck/torchmoji)

The conversion of the original model to Torch was done by 🤗: [https://github.com/huggingface/torchMoji](https://github.com/huggingface/torchMoji)

Not really a Bert model. Or I just don't know how to set Interference API correctly, as it gives wrong predictions when compared to the 🤗 Space.

Paper: [https://arxiv.org/abs/1708.00524](https://arxiv.org/abs/1708.00524)

Dataset:
* Mainly millions of tweets created around and before **-2017** which included use of emojis