caspar/MiniMind-Mixed-SFT-Training-v1.0 download history
caspar/MiniMind-Mixed-SFT-Training-v1.0 is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 425 times (40 in the last 7 days), and 425 times in total. It ranks #51,704 among datasets by monthly downloads.
MiniMind Mixed SFT Training — v1.0 Exact retained conversations, labels and windows for the completed Dense/MoE mixed-SFT experiment of 2026-10-24/25. Both models used this same release. This does not merge older UltraChat-only or GSM8K-only experiments. Training source Retained
Models trained on MiniMind-Mixed-SFT-Training-v1.0
5 models list it as training data.
- caspar/MiniMind-MoE-198M-A64M-GSM8K-GRPO-Skywork-v1.0 531 downloads in 30 days
- caspar/MiniMind-Dense-64M-GSM8K-GRPO-Skywork-v1.0 519 downloads in 30 days
- caspar/MiniMind-MoE-198M-A64M-GSM8K-GRPO-Rule-v1.0 513 downloads in 30 days
- caspar/MiniMind-Dense-64M-GSM8K-GRPO-Rule-v1.0 478 downloads in 30 days
- caspar/MiniMind-Dense-64M-GSM8K-SoftDistill-MoE-v1.0 450 downloads in 30 days
Open caspar/MiniMind-Mixed-SFT-Training-v1.0 on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.