tdooms/ts-large

TS Large

This is the medium version of the bilinear transformers trained on TinyStories. The primary purpose of this model is interpretability, most design choices were made with that in mind.

The code to run this custom model can be found here, along with many utility functions for weight-based interpretability.

Model Details

• 82 million parameters
• 8 layers
• 12 attention heads
• model dimension 768
• bilinear MLP with expansion factor 4
• context length of 256
• trained for 1 epoch (~500M tokens)
• rotary positional embedding
• custom tinystories tokenizer