Although the article focused primarily on AlphaFold, many other ML approaches are making impactful contributions in the general scientific field. One example is the diffusion model and its use of stochastic differential equations (SDEs).
I personally dislike these kinds of comments. The idea that androgens aid in recovery may be known in theory, but this study provides empirical evidence to support that claim. So yes, there is scientific value in it.
Distribution shift in the real world data will always be inherent to any data driven methods. Unless there are major advances in continual learning for DL models, they will always struggle with distribution shift degradation.
Similarly, humans are also prone to the distribution shift unless we get updated information on a specific topic. The key differences are that we are great at continual learning and we are much better at learning abstraction
A 17 million parameter model (~Resnet50) takes more than 50s proof time. Is this on top of the inference time?
I can see some niche applications for this system, but I am very skeptical it's ability to handle larger models (100M+) and the ability to and it's scalability when there are increased demand.
It's not impossible to see that LLMs can make finite step "reasoning" between input and output, as each block of the transformer can model probabilistic causality. Transformers are sometimes considered a fully connected graph neural network, which can be used for modeling causal graphs. One additional supporting evidence on the finite step reasoning hypothesis is that "train of thought" improves the performance of LLMs[1], meaning that letting the model explain itself explicitly reduces the amount of implicit "reasoning" steps that need to happen within the model.
Additionally, while we don't know how us humans functionally reason, it's believed that the predictive nature of our brain is central to our reasoning abilities. Maybe in some way, the autoregressive nature of LLMs is similar to our predictive brain.
For a collection of emergent properties in LLMs, I recommend this paper [2]
Apple Watch has so many health features, but the main hinderance for me is its relatively short battery life - I often forget to put it back on after charging it.
Has anyone tried having two apple watches so one can be worn while the other charges?
[edit] Charging it at night is kind of missing the point, as I would like to track both my daily activities and my sleep
Microsoft has an initiative called AI4Sciencie (https://www.microsoft.com/en-us/research/lab/microsoft-resea...) which published a fair amount of SDE/diffusion-based method to solve scientific problems