Deep Learning Scaling is Predictable, Empirically · HackerPapers discussions

Deep Learning Scaling is Predictable, Empirically — mentioned by gdiamos

Sat, 01 Mar 2025 05:27:47 GMT

Scaling laws just say that accuracy improves predictably - https://arxiv.org/pdf/1712.00409 That’s what we put in our 2017 paper. It does mean that there is a simple way to keep building smarter LLMs. I have never seen a clear definition of what AGI is and what it means to achieve it

Paper and discussion history. A mention is not an endorsement.

Deep Learning Scaling is Predictable, Empirically — HN submission

Wed, 27 Nov 2024 22:28:18 GMT

Deep Learning Scaling Is Predictable, Empirically

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Deep Learning Scaling is Predictable, Empirically — mentioned by dpflan

Thu, 18 Apr 2024 21:03:31 GMT

The comments here are trending towards "stop cramming 'AI' into everything". I am curious how the end-user consumer (versus 'AI' for enterprise/business) differ in experience and use. We are in the beginning of this AI-fication, and it seems deep learning models are doing really well and that DL can predictably scale [1.]; therefore, do we have to wait a bit for really life-changing AI for the end-user consumer? I can see AI in enterprise/business being extremely useful in different industries, but at the same time, is the current 'AI' actually good/useful for the end-user consumer? [1.]…

Paper and discussion history. A mention is not an endorsement.

Deep Learning Scaling is Predictable, Empirically — mentioned by gdiamos

Sat, 28 Oct 2023 01:17:44 GMT

We showed this in our original scaling law paper: https://arxiv.org/pdf/1712.00409.pdf The slope of the power law is determined by the problem and dataset. Compute, parameter count, and data move you along the curve. Change in architecture/bias is a constant offset. So architecture can give an advantage, but that advantage can be overcome by scale.

Paper and discussion history. A mention is not an endorsement.

Deep Learning Scaling is Predictable, Empirically — HN submission

Fri, 26 Jan 2018 17:12:39 GMT

Deep Learning Scaling Is Predictable, Empirically

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.