Gemini-3.1-pro ?
Hi,
Can you specify the name of the datasets and their origin ?
Are the datasets listed below included ?
Roman1111111/gemini-3-flash-120000x-high-reasoning
Roman1111111/gemini-3-pro-10000x-hard-high-reasoning
Roman1111111/gemini-3.1-pro-hard-high-reasoning
Thank you in advance.
Hi,
Thank you for your question.
We primarily used a proprietary dataset developed internally, which will be published in the near future. This dataset was constructed through a combination of synthetic generation and curation processes. Additionally, a small portion of the data may include questions derived or adapted from publicly available open source datasets to ensure coverage and diversity.
Regarding the specific datasets you mentioned (Roman1111111/gemini-3-flash-120000x-high-reasoning, Roman1111111/gemini-3-pro-10000x-hard-high-reasoning, Roman1111111/gemini-3.1-pro-hard-high-reasoning), they are not explicitly included as standalone datasets in our training corpus. However, as noted, some questions may overlap with or be inspired by content that exists in open source resources.
