Synthetic prompt-response datasets generated using non-reasoning models, primarily meta-llama/Llama-3.1-405B-Instruct