Recommended Papers
Rough draft. The section and its sub-areas are being populated incrementally. Most per-paper pages currently exist as stubs; substantive write-ups will follow.
Most of the field's best ideas live in a small number of papers. This section is my growing ML reading list, organised by topic, with notes for each. NLP has its own top-level section; its paper collection is linked below.
Related reading collections
-
NLP
The chronological arc of language modelling: from Shannon's entropy estimates of English, through the neural-LM era, attention, the Transformer, BERT, GPT-2/3, and the Kaplan/Chinchilla scaling laws.
More sub-areas (systems, alignment, scaling, data, vision, etc.) will appear as the section grows.