CamemBERT-base for sentiment analysis on tweets
This is a Camembert-base model trained on a corpus of 50K french tweets.
- Git Repo containing the dataset and the code (scraping & training) : Git
The model can predict which of the 25 emojis it has been trained with suits the best on a given sentence / tweet. The 25 emojis are the 25 most frequent in the dataset.
We've succeeded to obtain a 32% accuracy on a small amount of tweets.
Note: We've also decided to keep the emojis in their demojized versions because some emojis could be seen as two (ex : 👍🏿)
Loading the model
from transformers import AutoModelForSequenceClassification, AutoTokenizer, TFAutoModelForSequenceClassification
MODEL = f"Jessy3ric/camembert-twitter-emoji"
tokenizer = AutoTokenizer.from_pretrained(MODEL)
model = AutoModelForSequenceClassification.from_pretrained(MODEL)
model.save_pretrained(MODEL)
- Downloads last month
- 7
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.