MiniMax-M1, first open-weight, large-scale hybrid-attention reasoning model

3 points| hallh | 8 months ago |github.com