openthaigpt/openthaigpt1.5-72b-instruct
OpenThaiGPT 1.5 72B Instruct is a 72-billion-parameter Thai language chat model developed by the OpenThaiGPT team, based on Qwen v2.5. Fine-tuned on over 2,000,000 Thai instruction pairs, it excels at answering Thai-specific domain questions and achieves high scores on various Thai language exams. The model supports a large context window of up to 131,072 tokens and features multi-turn conversation, RAG compatibility, and tool calling.
Loading preview...
OpenThaiGPT 1.5 72B Instruct Overview
OpenThaiGPT 1.5 72B Instruct is an advanced 72-billion-parameter large language model developed by the OpenThaiGPT team, built upon the Qwen v2.5 architecture. Released on September 30, 2024, this model is specifically optimized for the Thai language, having been fine-tuned on over 2,000,000 Thai instruction pairs.
Key Capabilities
- Exceptional 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 benchmark (76.73% Micro Average) and the scb10x/thai_exam (64.07%).
- Extensive Context Handling: Processes up to 131,072 tokens for input and generates up to 8,192 tokens, enabling detailed and complex interactions.
- Multi-turn Conversation: Supports extended dialogues for more natural and continuous interactions.
- RAG Compatibility: Designed for seamless integration with Retrieval Augmented Generation systems to enhance response accuracy and relevance.
- Tool Calling Support: Enables the model to efficiently call various functions, facilitating interactions with external APIs for real-time data retrieval or other functionalities.
Good For
- Applications requiring high-accuracy Thai language understanding and generation.
- Chatbots and conversational AI systems for Thai users.
- Tasks involving complex Thai-specific domain questions.
- Developers needing a powerful Thai LLM with large context capabilities and tool-use functionality.