uMdZHVXdOL / README.md
DineshAI's picture
Correct training hardware metadata to Google Colab G4
23e273d verified
|
Raw
History Blame Contribute Delete
652 Bytes
---
title: "Repro - FiBeR DP-Noise Attenuation (uMdZHVXdOL)"
emoji: ๐Ÿ”‡
colorFrom: yellow
colorTo: red
sdk: static
pinned: false
tags:
- trackio
- trackio-logbook
- open-experiment
- icml2026-repro
- paper-uMdZHVXdOL
- google-colab-g4
- gpu-rtx-pro-6000-blackwell
---
# Repro - FiBeR DP-Noise Attenuation (uMdZHVXdOL)
An open experiment logbook, published with [Trackio](https://github.com/gradio-app/trackio).
The terminal C2/C3/C5/C6 training campaign ran on a **Google Colab G4 GPU**:
**NVIDIA RTX PRO 6000 Blackwell Server Edition**, 101,974,081,536 bytes
(approximately 95 GiB) GPU memory. Only the separate analytic C1 check used CPU.