Overview

This model, Deign86/deped-math-qwen2.5-7b-deped-math-merged, is a specialized 7.6 billion parameter language model based on Qwen/Qwen2.5-7B-Instruct. It has been fine-tuned by Deign86 specifically for DepEd Grade 11-12 mathematics tutoring, with its LoRA adapter merged into the base model weights for direct inference without PEFT adapters. The training was conducted up to step 700 on a dedicated dataset to enhance its mathematical instructional capabilities.

Key Capabilities

• Provides DepEd-aligned tutoring support for Grade 11-12 mathematics.
• Generates step-by-step instructional responses for math problems.
• Offers a standalone fp16 Transformers model for easy integration.

Intended Use Cases

• Prototyping and evaluation of AI-powered math tutoring systems.
• Serving as a classroom assistant for Grade 11-12 math educators.
• Developing tools for instructional support in Philippine educational settings.

Limitations

It is important to note that this model is not intended to replace teacher assessment or curriculum validation. It may occasionally produce incorrect or incomplete reasoning, especially for edge cases, and requires human review of its outputs in real classroom environments.