kalyanv

6 ms·

1. 1.▲
   by kalyanv 6y ago

   "But it is only a start. Haibe-Kains points out that code alone is often not enough to rerun an experiment. Building AI models involves making many small changes—adding parameters here, adjusting values there. Any one of these can make
2. 2.▲
   The replicability crisis in ML and AI(technologyreview.com)

   3 points by kalyanv 6y ago|1 comments
3. 3.▲
   by kalyanv 6y ago

   Yes, I think it's more about Synthetic Data Generation than versioning. Much closer to what SDV [0] does. [0] https://github.com/sdv-dev/SDV
4. 4.▲
   by kalyanv 9y ago

   Hi, I'm one of the authors of this work. We're very proud that this has attracted so much attention on Hacker News. I'm happy to answer a few questions. We had two requirements for the synthetic data: From the paper, “This s