KISTI-KONI/KONI-Llama3.1-8B-Instruct-20241024

KISTI-KONI/KONI-Llama3.1-8B-Instruct-20241024 is an 8 billion parameter instruction-tuned large language model developed by the Korea Institute of Science and Technology Information (KISTI). Built upon a merged base of Meta-Llama-3-8B and KISTI-KONI/KONI-Llama3.1-8B-20240824, it is specifically designed and optimized for tasks within science and technology domains. This model leverages Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) on specialized datasets, making it highly effective for scientific reasoning, mathematical problems, and technical writing with a 32768 token context length.