Transformers know more than they can tell: Learning the Collatz sequence (arxiv.org) | HN Screenshots

YHacker Newsscreenshots

Screenshot of arxiv.org: Transformers know more than they can tell: Learning the Collatz sequence

First screen, captured 10 months ago.

Looks similar

Screenshot of arxiv.org: Different language models learn similar number representations

Different language models learn similar number representations (arxiv.org)

92 points by Anon84 6 months ago | 38 comments

Screenshot of arxiv.org: Transformers without normalization

Transformers without normalization (arxiv.org)

33 points by kaycebasques 1 year ago | 3 comments

Screenshot of arxiv.org: Transformers Can Do Arithmetic with the Right Embeddings

Transformers Can Do Arithmetic with the Right Embeddings (arxiv.org)

186 points by byt3h3ad 2 years ago | 191 comments

Screenshot of arxiv.org: Transformer Layers as Painters

Transformer Layers as Painters (arxiv.org)

67 points by fzliu 2 years ago | 9 comments

Screenshot of kipp.ly: Transformer Inference Arithmetic (2022)

Transformer Inference Arithmetic (2022)(kipp.ly)

36 points by lawrencechen 3 years ago | 4 comments

Screenshot of arxiv.org: A Mathematical Perspective on Transformers

A Mathematical Perspective on Transformers (arxiv.org)

63 points by georgehill 2 years ago | 4 comments