The base model I used was cardiffnlp/twitter-roberta-base-sentiment-latest. I used Twitter financial news' comments and headlines, with sentiment ranging from 1 to 10 and positive, negative, or neutral to describe it. I then fine-tuned the model and tested it from more Twitter financial news data for accuracy.

Developed by: Atoma Media
Model type: Classification
Language(s) (NLP): English
License: Apache-2.0
Finetuned from model: cardiffnlp/twitter-roberta-base-sentiment-latest
Source Repository: https://github.com/sn82978/financial-news-headers-sentiment-analysis

How to Get Started with the Model

python

from transformers import pipeline

pipe = pipeline("text-classification", model="snoneeightfive/financial-news-headers-sentiment-analysis")
pipe("Defense stocks are steadily rising ") # Your financial headline

[{'label': 'positive', 'score': 0.6553508639335632}] # Example output

Use a pipeline as a high-level helper

Evaluation

Accuracy: 80%

Testing Data, Factors & Metrics

Testing Data

Financial headlines from Twittter.

Model Card Authors

Shreya Nakum

Model Card Contact

snakum@uci.edu