pith. sign in

Qwen3: The next generation of qwen large language models, 2026

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

1 Pith paper citing it

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

Robust Reasoning Benchmark

cs.LG · 2026-03-26 · unverdicted · novelty 7.0 · 2 refs

The Robust Reasoning Benchmark shows frontier LLMs are mostly resilient to textual perturbations on AIME problems while open-weight models suffer up to 54% accuracy drops and exhibit accuracy decay on later problems due to attention dilution during chain-of-thought.

citing papers explorer

Showing 1 of 1 citing paper.

  • Robust Reasoning Benchmark cs.LG · 2026-03-26 · unverdicted · none · ref 42 · 2 links

    The Robust Reasoning Benchmark shows frontier LLMs are mostly resilient to textual perturbations on AIME problems while open-weight models suffer up to 54% accuracy drops and exhibit accuracy decay on later problems due to attention dilution during chain-of-thought.