pith. sign in

If all entriesbpjk are positive, choose logits (W ⊤ j,:)k = logbpjk (up to an additive constant), such that softmax(W ⊤ j,:) =bpj

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

fields

cs.CY 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

LLM Biases

cs.CY · 2026-04-07 · unverdicted · novelty 4.0

Transformer attention in generative recommenders can create positional, popularity, latent-driver, and synthetic-data biases that concentrate exposure and reduce diversity.

citing papers explorer

Showing 1 of 1 citing paper.

  • LLM Biases cs.CY · 2026-04-07 · unverdicted · none · ref 5

    Transformer attention in generative recommenders can create positional, popularity, latent-driver, and synthetic-data biases that concentrate exposure and reduce diversity.