How do you weight the value of semantic layers when all of the labs are chasing shell usage benchmarks like TerminalBench?
Aside from the enterprise stuff like consistent metrics I’m not convinced semantic layers improve agent performance. Case in point is Snowflake Analyst has been routing 90%+ of queries to traditional SQL as opposed to their own semantic SQL dialect.
A semantic layer is a concept from 2018 in the BI world. Metadata is one thing but semantic layer implies an abstraction from physical data with lossy translations.
Great question. Semantic layer has been around before I was even in the industry. Probably late 90's. Innovation on this layer has gone backwards actually. Simpler and less capable (Looker vs Strategy for example).
It does however improve on agent performance in one important way: rejects plausible queries that are invalid. Agents writing sql are more likely to be inventive in unexpected ways. Business users won't know the difference. It is definitely not useful for developers/data engineers anymore. But quite a good tool as part of analytics harness for business users.
The one issue with Snowflake semantic dialect is the lack of expressiveness. Agents have to route to sql for many common scenarios. What do you think?
I understand why this external view exists, as dissemination is inherently part of the scientific mission; but if you look more carefully, you will see that it is simultaneously science promotion, and many of the best participants do not shamelessly promote themselves.
I did a PhD, so not external. Sure one doesn't have to be maximally cynical but a lot of it is self promotion.
And in context: the same can be said about Kaggle, about Youtubers, about music creators etc. Every endeavor is a mix of "pure" promotion of the art and of shameless self promotion and status games. The common factor is humans. My point was, academia is not more pure than the Kaggle guys who fish for industry jobs.
I cannot speak for the intended purpose of ArXiv by its creators but I can tell you that, in the conference circles, its main intended use was flag planting: people were afraid that their competitors would tweet some results while their (earlier) paper was under anonymous review, and so researchers started putting their stuff on ArXiv first to ensure no one would "steal" their claim of being there first.
I don't think it has much to do with tweets, but the flag planting is correct. You have to timestamp your idea because otherwise if you're too slow, the community doesn't much care about "pretty much same idea but now done properly and thoroughly" and will just elevate the faster one.
If the researchers didn't care about promoting themselves they would not care whether a competitor published a result first or they did. Either way science is being pushed forward.
Based on the info if you click into them, likely no. I would have expected them to be incidental materials from tunneling, but reading the description that's not the case.
Hello all, using an ideas first approach and an AI agent with access to a novel backtesting language, this blog post details the hunt for institutional buyers building positions in NVDA either for themselves or on behalf of their clients. Happy whale hunting.
Aside from the enterprise stuff like consistent metrics I’m not convinced semantic layers improve agent performance. Case in point is Snowflake Analyst has been routing 90%+ of queries to traditional SQL as opposed to their own semantic SQL dialect.
A semantic layer is a concept from 2018 in the BI world. Metadata is one thing but semantic layer implies an abstraction from physical data with lossy translations.