TSOTSALLM: A LLaMA 2 Fine-Tune for Table Annotation

yvelos/Tsotsallm is a 7 billion parameter large language model built upon the LLaMA 2 architecture. Its core purpose is to facilitate the automatic annotation of TSOTSATable data, addressing tasks such as Cell Entity Annotation (CEA), Column Type Annotation (CTA), Column Property Annotation (CPA), and Information Type Detection (ITD).

Key Capabilities

• Automated Table Annotation: Designed to process and annotate structured table data efficiently.
• Task-Specific Fine-Tuning: Optimized for specific table-related annotation tasks (CEA, CTA, CPA, ITD).
• Composition Table Generation: After initial annotation, the model can generate comprehensive composition tables.

Good For

• Structured Data Processing: Ideal for applications requiring automated understanding and annotation of tabular information.
• Information Extraction from Tables: Useful for extracting specific entities, types, and properties from tables.
• Research in Table Understanding: Provides a specialized tool for researchers working on table-based NLP tasks.