I looked into this lately to check our options and came to the same conclusion. Yes, I can run Kimi K2.5 on a cluster of 10 DGX Spark for around 40kโฌ but it's unusably slow. I think your numbers are still too low. If you want to run Kimi K2.5 with full precision and context length you need 67xH100
06.03.2026 06:15
๐ 0
๐ 0
๐ฌ 0
๐ 0
VoxelPrompt: A Vision-Language Agent for Grounded Medical Image Analysis
We present VoxelPrompt, an agent-driven vision-language framework that tackles diverse radiological tasks through joint modeling of natural language, image volumes, and analytical metrics. VoxelPrompt...
Recently thought about a model which takes a delineation instruction / segmentation protocol as input and provides the respective segmentation(s). Today I read VoxelPrompt. Super interesting work by Andrew Hoopes, Victor Ion Butoi, John Guttag and @adriandalca.bsky.social.
03.03.2025 23:22
๐ 2
๐ 2
๐ฌ 0
๐ 0
Our work focuses on using ML for MRI segmentation to quantify brain regions, tissues, and lesions. This supports radiologists and neurologists in diagnosing and monitoring neurodegenerative diseases.
10.12.2024 05:20
๐ 2
๐ 0
๐ฌ 0
๐ 0
Thanks for the list. Would like to be added as well.
08.12.2024 11:34
๐ 0
๐ 0
๐ฌ 2
๐ 0
I understand why it's not atm. but looking forward to a future where quantification is the routine.
05.12.2024 19:25
๐ 1
๐ 0
๐ฌ 0
๐ 0
For a high quality volumetry we need 3D scans.
05.12.2024 17:29
๐ 1
๐ 0
๐ฌ 1
๐ 0
PyInstaller Manual โ PyInstaller 6.11.1 documentation
Would that allow uv to offer building an OS independent executable like PyInstaller does?
pyinstaller.org
04.12.2024 06:43
๐ 2
๐ 0
๐ฌ 1
๐ 0