T2MLR pretraining checkpoints: 3 scales x 2 token budgets, trained from scratch on FineWeb-Edu.
-
JupiterZhu/T2MLR_135M_lstart13_lend18_10B_FineWebEdu
Text Generation • 0.2B • Updated • 181 -
JupiterZhu/T2MLR_135M_lstart13_lend18_50B_FineWebEdu
Text Generation • 0.2B • Updated • 185 -
JupiterZhu/T2MLR_362M_lstart9_lend24_10B_FineWebEdu
Text Generation • 0.4B • Updated • 186 -
JupiterZhu/T2MLR_362M_lstart9_lend24_50B_FineWebEdu
Text Generation • 0.4B • Updated • 184