This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-2022-154m on the SemEval 2018 - Task 1 Affect in Tweets (subtask: El-reg / text regression).

Warning: Hosted inference API produces inaccurate values

Model Trained Using AutoTrain

Problem type: Single Column Regression
Model ID: 72772139027
CO2 Emissions (in grams): 0.0259

Validation Metrics

Loss: 0.011
MSE: 0.011
MAE: 0.079
R2: 0.726
RMSE: 0.103
Explained Variance: 0.727

Usage

You can use cURL to access this model:

$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I'm feeling down"}' https://api-inference.huggingface.co/models/garrettbaber/twitter-roberta-base-sadness-intensity

Or Python API:

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("garrettbaber/twitter-roberta-base-sadness-intensity")

tokenizer = AutoTokenizer.from_pretrained("garrettbaber/twitter-roberta-base-sadness-intensity")

inputs = tokenizer("I'm feeling down", return_tensors="pt")

outputs = model(**inputs)