Explainers

Teaching you stuff in blog form so I don’t have to become a professor.

2024

• But really, what is amortisation? 10 Dec
• Does DeBERTa have infinite context length, and how large is the receptive field of a token? 22 Sep
• How does HuggingFace's from_pretrained() know which weights in a checkpoint go where? 31 Aug
• Bits-per-character and its relation to perplexity 29 Jul