undefined

Markus Nagel

Research scientist at Qualcomm AI Research, focusing on machine learning efficiency and inference efficiency techniques like quantization and pruning. Presented research at NeurIPS 2023.

Best podcasts with Markus Nagel

Ranked by the Snipd community
undefined
6 snips
Dec 26, 2023 • 47min

Quantizing Transformers by Helping Attention Heads Do Nothing with Markus Nagel - #663

In this discussion, Markus Nagel, a research scientist at Qualcomm AI Research, shares insights from his recent papers at NeurIPS 2023, focusing on machine learning efficiency. He tackles the challenges of quantizing transformers, particularly in minimizing outlier issues in attention mechanisms. The conversation explores the pros and cons of pruning versus quantization for model weight compression and dives into innovative methods for multitask and multidomain learning. Additionally, the use of geometric algebra in enhancing algorithms for robotics is highlighted.