3 ms·
I genuinely don't see how "The information we learn only comes from aggregated data" relates to training LLMs, which need raw data, not aggregated data, as thei
by simonw 3y ago
I genuinely don't see how "The information we learn only comes from aggregated data" relates to training LLMs, which need raw data, not aggregated data, as their input.
Maybe we have different definitions of the term "aggregated"?
This suggests to me that GitHub need to extend that text to explain what they mean by "aggregated".
- atq2119 3y agoThe LLM itself is a form of aggregated data.
- simonw 3y agoSure, but the raw training data isn't. I think GitHub need to clarify this themselves.
- atq2119 3y agoSure. All aggregated data is ultimately derived from raw, unaggregated data. One can make the argument that training an LLM is "just" an unusually complicated form of aggregation. Whether that would hold up is another question. But yeah, I agree with the conclusion that they need to clarify this.
- hedora 3y agoTraining the LLM is a form of “learning”, and putting all the data in an input training set is a form of aggregation. The clause seems to mean “we can do whatever we want with your data as long as we violate many people’s privacy at scale at the same time”.
- mtlmtlmtlmtl 3y agoBut then the training can surreptitiously be called data aggregation. We don't look at the raw data, we aggregate it and then query the aggregation. Definitely needs clarification, though somehow I suspect this is all by design.
- waynecochran 3y agoEnter lawyers.