3 ms·
Precisely this. This doesn't solve anything and appears to be absurd techno-solutionism at this point. Every single problem does not need to be solved using an
by rvz 3y ago
Precisely this. This doesn't solve anything and appears to be absurd techno-solutionism at this point.
Every single problem does not need to be solved using an LLM, which this paper tells us the amount of desperation in this hype cycle.
Those reading this paper and giving it credibility have fallen for this nonsense and are probably gullible enough to believe in this non use-case.
- Philpax 3y agoDid you read the paper? What exactly do you think it does?
- danielbln 3y ago> Every single problem does not need to be solved using an LLM How do you know if a problem is suitable for transformers/LLMs unless you try? We have this great generalized architecture, I would hope people throw everything at it and see what sticks, because what's the downside? Less focus on bespoke predictive models? Oh no.