metadata
license: mit
tags:
- sentiment
- sentiment-analysis
- financial
- fine-tuned
- fine-tuned-bert
- bert-uncased
Model Overview:
This NLP model is fine-tuned with a focus on analyzing sentiment in financial text and news headlines. It was trained using the bert-base-uncased model on the financial_phrasebank and auditor_sentiment datasets.
Accuracies:
financial_phrasebank accuracy: 0.993
auditor_senitment accuracy: 0.974\
Training Hyperparameters:
Learning Rate: 2e-05
Train Batch Size: 16
Eval Batch Size: 16
Random Seed: 42
Optimizer: AdamW-betas(0.9, 0.999)
Learning Rate Scheduler: Linear
Number of Epochs: 6
Number of Warmup Steps: 0.2 * Number of Training Steps\
How To Use:
from transformers import pipeline
pipe = pipeline("sentiment-analysis", model="mstafam/fine-tuned-bert-financial-sentimental-analysis")
text = "Example company has seen a 5% increase in revenue this quarter."
print(pipe(text))
[{'label': 'Positive', 'score': 0.9993795156478882}]