Mark Kurtz - Profile and Journalist Details

Beats

Content

Total articles 8

• 2:4 Sparse Llama: Smaller Models for Efficient GPU Inference

  By Eldar Kurtic, Alexandre Marques, Mark Kurtz, Dan Alistarh| neuralmagic.com@ Nov 25, 2024 Sparse Foundation Model: The first sparse, highly accurate foundation model built on top of Meta’s Llama 3.1 8B with 98% recovery on Open LLM Leaderboard v1 and full recovery across fine-tuning tasks, including math, coding, and chat. Hardware-Accelerated Sparsity: Features a 2:4 sparsity pattern designed for NVIDIA Ampere GPUs and newer, delivering up to 30% higher throughput and 1.8x lower latency from sparsity alone with vLLM.

  By Eldar Kurtic, Alexandre Marques, Mark Kurtz, Dan Alistarh · neuralmagic.com

  Dec. 02, 2024

• 2022 State of Competitive ML

  By Mark Kurtz| Medium@ Percent of winning solutions utilizing PyTorch vs. TensorFlow from 2022 State of Competitive Machine Learning The 2022 State of Competitive Machine Learning was released recently, and in it, the authors highlight the core trends across the past year. In their words, "2022 was a big year for competitive machine learning, with a total prize pool of more than $5m across all platforms".

  By Mark Kurtz · Medium

  Mar. 11, 2023

• 2:4 Sparse Llama: Smaller Models for Efficient GPU Inference

  By Eldar Kurtic, Alexandre Marques, Mark Kurtz, Dan Alistarh| neuralmagic.com@ Nov 25, 2024 Sparse Foundation Model: The first sparse, highly accurate foundation model built on top of Meta’s Llama 3.1 8B with 98% recovery on Open LLM Leaderboard v1 and full recovery across fine-tuning tasks, including math, coding, and chat. Hardware-Accelerated Sparsity: Features a 2:4 sparsity pattern designed for NVIDIA Ampere GPUs and newer, delivering up to 30% higher throughput and 1.8x lower latency from sparsity alone with vLLM.

  By Eldar Kurtic, Alexandre Marques, Mark Kurtz, Dan Alistarh · neuralmagic.com

  Dec. 02, 2024

As seen in

Company Info

WCAV-TV (Charlottesville, VA)