3 ms·
Most engineers and researchers at big tech companies wouldn't intentionally do that. The bigger problem is that public evals leak into the training data. You ca
by scarmig 3y ago
Most engineers and researchers at big tech companies wouldn't intentionally do that. The bigger problem is that public evals leak into the training data. You can try to cleanse your training data, but at some point it's inevitable.
- Racing0461 3y agoYeah, i not saying it was intentional (misleading shareholders would be the worse crime here). Having these things in the training data without knowing due to how vast the dataset is is the issue.
- FergusArgyll 3y ago> We filter our evaluation sets from our training corpus. Page 5 of the report (they mention it again a little later) https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf https://storage.googleapis.com/deepmind-media/gemini/gemini_...