Google's Recipes Telugu BERT models pre-trained from scratch with Google's original BERT recipe (Devlin 2018, Turc 2019): 1M steps, batch 256, 128 then 512 tokens. bammera-pothana/BERT-tiny Fill-Mask • 4.62M • Updated 1 day ago • 44 bammera-pothana/BERT-mini Fill-Mask • 11.6M • Updated about 2 hours ago
Google's Recipes Telugu BERT models pre-trained from scratch with Google's original BERT recipe (Devlin 2018, Turc 2019): 1M steps, batch 256, 128 then 512 tokens. bammera-pothana/BERT-tiny Fill-Mask • 4.62M • Updated 1 day ago • 44 bammera-pothana/BERT-mini Fill-Mask • 11.6M • Updated about 2 hours ago
