Back to DrivenData | Blog

One-shot learning

You may not update your model weights after February 15

It is mean then we cannot using the one-shot learning?
Example: unsupervised fine-tuning the global model for current date

@FBykov Thank you for your question. That is correct - any methods that would require ongoing updates to model weights (e.g. unsupervised fine-tuning throughout the evaluation period) are not permitted since, as you mention, model weights must be frozen by February 15.