ML Training Data Limits
Comments discuss the scarcity, quality, and size limitations of training data as the primary bottleneck for machine learning models, often more critical than compute power.
Activity Over Time
Top Contributors
Keywords
Sample Comments
Because you need more training data for better results and they are running out of new training data.
You can't possibly get enough training data for this.
I'd imagine training data would be the limiting factor.
Because they wouldn't have enough good quality training data then probably.
I could be wrong but I think part of the issue is this needs some large files for the trained dataset?
Disclaimer: Not an ML expertI suspect that's a function of the size and quality of their training set?
Smaller % of training data doesn't necessarily mean lower quality.
Computer power is not stagnating, but the availability of training data is. It's not like there's a second stackoverflow or reddit to scrape.
The problem is the amount of pictures you'd need. It's much easier to use available datasets if you know how to preprocess the data.
Where "enough training examples" has proven to be the real difficult problem.