hanifsajid/deed-v01

XLM-RoBERTa Large Fine-Tuned for DEED Multi-Class Classification

Model Description

This model is fine-tuned from xlm-roberta-large for multi-class classification of Democratic Erosion Event Dataset (DEED) events.
It predicts one of four classes:

• 0: Destabilizing Event
• 1: Precursor
• 2: Resistance
• 3: Symptom

Intended Use

• Classify DEED-related textual data into four event types.
• Academic research or internal NLP pipelines.
• Not intended for critical real-world decision-making without further validation.

Metrics

Label	Precision	Recall	F1-Score	Support
Destabilizing Event (0)	0.598	0.392	0.473	125
Precursor (1)	0.659	0.716	0.686	675
Resistance (2)	0.846	0.845	0.846	645
Symptom (3)	0.800	0.786	0.793	805
Accuracy	0.760	0.760	0.760	2250
Macro Avg	0.726	0.685	0.700	2250
Weighted Avg	0.760	0.760	0.758	2250

Usage

from transformers import pipeline

classifier = pipeline("text-classification", model="hanifsajid/deed-v01")
result = classifier("Example text here")
print(result)