openthaigpt/openthaigpt1.5-7b-instruct

OpenThaiGPT 1.5 7B Instruct: Thai-Centric Chat Model

OpenThaiGPT 1.5 7B Instruct is a 7-billion-parameter Thai language chat model, fine-tuned by OpenThaiGPT on over 2,000,000 Thai instruction pairs. Released on September 30, 2024, and built upon Qwen v2.5, this model is optimized for Thai-specific domain questions and general Thai chat.

Key Capabilities

• State-of-the-art Thai Language Performance: Achieves the highest average scores across various Thai language exams compared to other open-source Thai LLMs, including a micro average of 65.78% on the OpenThaiGPT Eval benchmark.
• Extensive Context Handling: Processes up to 131,072 tokens of input and generates up to 8,192 tokens, enabling detailed and complex interactions.
• Multi-turn Conversation Support: Facilitates extended and coherent dialogues.
• RAG Compatibility: Designed for Retrieval Augmented Generation to enhance response accuracy and relevance.
• Tool Calling Support: Enables efficient function calls for external APIs (e.g., weather data) through intelligent responses.

Good for

• Thai Language Applications: Ideal for chatbots, customer service, and content generation requiring deep understanding and generation in Thai.
• Limited GPU Environments: The 7B parameter size makes it suitable for deployment on systems with less powerful GPUs, such as an Nvidia RTX 4060 8GB for 4-bit quantized versions.
• Research and Commercial Use: Licensed under the Qwen license, allowing both research and commercial applications, with specific terms for large user bases.