Carlos García

cgarciams

CarlosJGarcia

AI & ML interests

Building a GPT-2 medium size (approx. 400 M parameters) model from scratch, using PyTorch, the OpenWebText dataset, tiktoken, AdamW optimizer and FlashAttention. Just for fun.

Recent Activity

published a dataset about 16 hours ago

cgarciams/hle-text

View all activity

Organizations

models 0

None public yet

datasets 1

cgarciams/hle-text

Updated about 16 hours ago • 1