MagR: Weight Magnitude Reduction for Enhancing Post-Training Quantization

Feb-16-2026, 22:09:19 GMT–Neural Information Processing Systems

For example, we achieve a Wikitext2 perplexity of 5.95 on the LLaMA2-70B

large language model, machine learning, quantization, (19 more...)

Neural Information Processing Systems

Feb-16-2026, 22:09:19 GMT

Conferences PDF

Country:
- Europe > United Kingdom > England > Cambridgeshire > Cambridge (0.04)

Genre:
- Research Report > Experimental Study (0.93)

Industry:
- Information Technology (0.46)

Technology:
- Information Technology > Artificial Intelligence
  - Machine Learning > Neural Networks
    - Deep Learning (0.91)
  - Natural Language > Large Language Model (0.93)
  - Representation & Reasoning (1.00)

Duplicate Docs Excel Report

Title
MagR: Weight Magnitude Reduction for Enhancing Post-Training Quantization

Similar Docs Excel Report more

Title	Similarity	Source
None found