lzw1008/ConspEmoLLM-v2
ConspEmoLLM-v2 by lzw1008 is a 7 billion parameter causal language model specifically fine-tuned for classifying text related to COVID-19 conspiracy theories and misinformation. This model excels at identifying whether text is unrelated, related but not supporting, or actively supporting COVID-19 conspiracy narratives. With a 4096-token context length, it provides a specialized solution for content moderation and analysis in this domain.
Loading preview...
ConspEmoLLM-v2: Specialized COVID-19 Conspiracy Classification
ConspEmoLLM-v2 is a 7 billion parameter causal language model developed by lzw1008, uniquely fine-tuned for the nuanced task of classifying text concerning COVID-19 conspiracy theories and misinformation. This model is designed to help identify and categorize content based on its relationship to such narratives.
Key Capabilities
- Precise Classification: Distinguishes between three distinct categories for COVID-19 related text:
- 0. Unrelated: Text that has no connection to COVID-19 conspiracy theories.
- 1. Related (but not supporting): Text that mentions or discusses COVID-19 conspiracy theories without endorsing them.
- 2. Conspiracy (related and supporting): Text that actively supports or promotes COVID-19 conspiracy theories.
- Specialized Fine-tuning: Optimized specifically for this domain, offering a targeted solution for content analysis.
- Standard LLM Integration: Compatible with the Hugging Face Transformers library for easy deployment and use in Python projects.
Good For
- Content Moderation: Automatically flagging or categorizing online content for review.
- Misinformation Research: Analyzing trends and prevalence of COVID-19 conspiracy theories in textual data.
- Social Media Monitoring: Identifying and tracking the spread of specific narratives.