GulkoA/TinyStories-tokenized-Llama-3.2-1024-context

TinyStories dataset tokenized with Llama-3.2 Useful for accelerated training and testing of sparse autoencoders Context window: 1024, not shuffled