Author

Aur Shalev Merin

h-index 1 View on Semantic Scholar

We have 2 of 5 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Oct 2026

How Much Gradient Does Truncated Backprop Lose in Looped Transformers?

This work measures how much of the weight gradient this throws away, exactly and for every k from one backward pass, on nine looped language models, and reports three findings that show a trained model's value overstates what truncation removes later.

Aur Shalev Merin · 0 citations

Preprint Oct 2026

Present but Hard to Read: Probing Latent Reasoning with Minimal Pairs

A protocol, including a synthetic null for generic mixing, is given for testing computed values against copies and mixtures of the inputs, and in pre-registered studies of two-step problems it finds a small signal for the unwritten intermediate a+b just before Ouro writes the answer's last digit.

Aur Shalev Merin · 1 citation