Hey guys, I just released a 435M parameter Kannada language model, fully pre-trained from scratch called Kalike. It means learning in Kannada. Trained on an H100 GPU for about 9 hours on around 8 GB of clean Kannada data…
Hey guys, I just released a 435M parameter Kannada language model, fully pre-trained from scratch called Kalike. It means learning in Kannada. Trained on an H100 GPU for about 9 hours on around 8 GB of clean Kannada dataset
讨论这个产品