pith. sign in

The MaskLLM† rows are 2:4 semi-structured at 50% density and are reproduced from (Hourri et al., 2025)

Each table covers one model at 50%, 60% unstructured sparsity across the six zero-shot tasks (MMLU, PIQA, ARC-E, ARC-C, Winogrande, OBQA) plus WikiText2 perplexity at sequence length4096 · 2025 · arXiv 8923.8071

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

1 Pith paper citing it
read on arXiv browse 1 citing papers

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

LEAP: Learnable End-to-End Adaptive Pruning of Large Language Models

cs.LG · 2026-05-17 · unverdicted · novelty 7.0

LEAP replaces intractable categorical mask parameterization with a differentiable per-weight Bernoulli relaxation, delivering +2.59 average zero-shot accuracy gain over the best layer-wise baseline across 0.5B-8B LLMs at 50-60% sparsity.

citing papers explorer

Showing 1 of 1 citing paper.

  • LEAP: Learnable End-to-End Adaptive Pruning of Large Language Models cs.LG · 2026-05-17 · unverdicted · none · ref 13

    LEAP replaces intractable categorical mask parameterization with a differentiable per-weight Bernoulli relaxation, delivering +2.59 average zero-shot accuracy gain over the best layer-wise baseline across 0.5B-8B LLMs at 50-60% sparsity.