FiLLM-POSDEPSUM: Filipino-Optimized NLP Model

FiLLM-POSDEPSUM is an 8.5 billion parameter model, part of the FiLLM (Filipino-optimized Large Language Model) family, developed by Isaiah Job Cuenca Enriquez, Carlos Jude Maminta, and Deandre Nigel Corpuz Nuñez. It is fine-tuned from the SeaLLM-7B 2.5 model, which itself is based on Gemma 7B, utilizing Low-Rank Adaptation (LoRA) for memory-efficient optimization. This model is specifically designed to address key Natural Language Processing (NLP) tasks in the Filipino language.

Key Capabilities

Part-of-Speech (POS) Tagging: Identifies and labels the grammatical category of words in Filipino text.
Dependency Parsing: Analyzes the grammatical structure of sentences by showing relationships between words.
Text Summarization: Generates concise summaries of Filipino text.

Training and Evaluation

The model was trained and evaluated using diverse Filipino datasets to ensure robust performance across its specialized tasks. A related model, FiLLM-NER, handles Named Entity Recognition and is available separately. Users should be aware of potential hallucinations, especially if prompts do not end with a period. For more detailed information, refer to the associated research paper: FiLLM - A Filipino-optimized Large Language Model based on Southeast Asia Large Language Model (SEALLM).

Good For

Applications requiring grammatical analysis of Filipino text.
Automated summarization of documents or articles in Filipino.
Developing tools for Filipino language education or research.