#MixedPrecision
🚀 Do check out our latest work on Modular Mixed-precision Quantization!

🌐 Page: saqibjaved1.github.io/MQAT_/

#Quantization #Mixedprecision #DeepLearning #Robotics #AI #6Dpose #edgeAI
Modular Quantization-Aware Training for 6D Object Pose Estimation

Saqib Javed, Chengkun Li, Andrew Lawrence Price, Yinlin Hu, Mathieu Salzmann

Action editor: Aaron Klein

https://openreview.net/forum?id=lIy0TEUou7

#quantization #pose #quantized
December 19, 2024 at 2:39 PM
Mixed-precision numerics in scientific applications: survey and perspectives

#GPU #MixedPrecision #Review

hgpu.org?p=30704
Mixed-precision numerics in scientific applications: survey and perspectives
The explosive demand for artificial intelligence (AI) workloads has led to a significant increase in silicon area dedicated to lower-precision computations on recent high-performance computing hard…
hgpu.org
March 25, 2026 at 10:09 PM
⚡ Mixed precision = lower energy, same accuracy?
Learn how to safely reduce precision in HPC codes with VerifiCarlo.

🗓️ May 27 | 15:00 CET
💻 Online
Join CEEC & make your simulations more efficient and sustainable.

👉 ceec-coe.eu/event/enabli...
#HPC #Exascale #MixedPrecision
May 6, 2026 at 12:31 PM
🆕 Mixed Precision Training NaN Loss: 4 Root Causes & Fixes

#DeepLearning #AI #MixedPrecision #PyTorch
https://tildalice.io/mixed-precision-training-nan-loss-pytorch-fixes/
July 26, 2026 at 3:04 PM
Don't forget to register for our webinar next week and learn how to increase your energey efficiency without sacrificing accuracy!

Make the most of #MixedPrecision using VerifiCarlo. We can show you how!

🗓️ May 27 | 15:00–16:00 CET | Online

🔗 Register 👉 ceec-coe.eu/event/enabli...
May 21, 2026 at 12:17 PM
Want to improve your HPC code’s energy efficiency?

We've just published our last webinar on our experience using VerifiCarlo, an open-source tool to detect numerical bugs & analyze precision-performance trade-offs.

📺 Video & slides: ceec-coe.eu/2025/03/06/e...
#HPC #CFD #MixedPrecision
Enabling mixed-precision with VerifiCarlo: Sharing CEEC experience – CEEC CoE
ceec-coe.eu
March 6, 2025 at 2:08 PM
HPL-MxP benchmark uses lower‑precision LU factorization with GMRES‑based iterative refinement, achieving throughput comparable to Exascale machines on supercomputers. Submitted 23 Sep 2025. https://getnews.me/hpl-mxp-mixed-precision-lu-factorization-and-iterative-refinement/ #hplmxp #mixedprecision
September 26, 2025 at 3:18 PM
RateQuant just uncovered a mixed‑precision KV‑cache quirk—β decay rates are all over the place, ranging 3.6‑5.3. Could this bite your LLM performance? Dive into the details and see what to watch out for. #MixedPrecision #KVCaches #BetaDecay

🔗 aidailypost.com/news/ratequa...
May 11, 2026 at 3:51 PM