openthaigpt/openthaigpt1.5-7b-instruct

OpenThaiGPT 1.5-7b-instruct Overview

OpenThaiGPT 1.5-7b-instruct is a 7-billion-parameter instruction-tuned large language model specifically designed for the Thai language. Developed by the OpenThaiGPT team and built upon the Qwen v2.5 architecture, this model was fine-tuned using over 2,000,000 Thai instruction pairs, enabling it to handle Thai-specific domain questions effectively.

Key Capabilities & Features

• Leading Thai Language Performance: Achieves the highest average scores across various Thai language exams compared to other open-source Thai LLMs, as demonstrated on the OpenThaiGPT Eval and scb10x/thai_exam benchmarks.
• Extensive Context Handling: Processes up to 131,072 tokens for input and generates up to 8,192 tokens, facilitating detailed and complex interactions. It also supports YaRN for enhanced length extrapolation beyond 32,768 tokens.
• Multi-turn Conversation: Capable of engaging in extended dialogues, maintaining context across multiple exchanges.
• Retrieval Augmented Generation (RAG) Compatibility: Designed to work with RAG systems for improved response generation by integrating external knowledge.
• Tool Calling Support: Enables the model to efficiently call various functions, such as external APIs for real-time data retrieval (e.g., weather information), through intelligent responses.

Ideal Use Cases

• Thai-centric Chatbots: Developing conversational AI agents that require deep understanding and generation of Thai language.
• Information Retrieval in Thai: Applications needing to answer questions or summarize content from Thai documents.
• Automated Thai Customer Support: Implementing AI for handling customer inquiries in Thai, leveraging multi-turn and RAG capabilities.
• Tool-Enhanced Thai Applications: Integrating external tools and APIs to provide dynamic and data-driven responses in Thai.