RSchaeffer

6 ms·

1.
Turning Down the Heat: A Critical Analysis of Min-P Sampling in Language Models(twitter.com)
1 points by RSchaeffer 1y ago|1 comments
2.
by RSchaeffer 1y ago
We examine min-p sampling (ICLR 2025 oral) & find significant problems in all 4 lines of evidence: human eval, NLP evals, LLM-as-judge evals, community adoption claims
3.
How Do Large Language Monkeys Get Their Power (Laws)?(arxiv.org)
2 points by RSchaeffer 1y ago|1 comments
4.
by RSchaeffer 1y ago
Best of N was shown to exhibit power (polynomial) law scaling (left), but maths suggest one should expect exponential scaling (center). We show how to resolve this "paradox", then use our insights to design methods for predicting
5.
No Free Lunch from Deep Learning in Neuroscience(twitter.com)
1 points by RSchaeffer 4y ago|0 comments
6.
No Free Lunch from Deep Learning in Neuroscience(twitter.com)
1 points by RSchaeffer 4y ago|0 comments
7.
by RSchaeffer 4y ago
Thanks for taking the time to explain :)
8.
by RSchaeffer 4y ago
Thanks for the advice and links! Do you know of a Render tutorial that involves getting Flask and ReactJS services to communicate with one another? Your 2nd and 3rd links demonstrate each independently. I don't know whether the same ch
9.
Ask HN: How to develop React+Flask locally in a way that can easily be deployed?
9 points by RSchaeffer 4y ago|9 comments
10.
by RSchaeffer 4y ago
Can you share more about the habits you built to break that cycle of scrolling for an hour in the morning?
11.
Experience replay in machines and mammals(rylanschaeffer.github.io)
14 points by RSchaeffer 5y ago|1 comments
12.
Bay Area housing is fucked in more ways than you might think(rylanschaeffer.github.io)
34 points by RSchaeffer 5y ago|1 comments
13.
by RSchaeffer 5y ago
Quitting smoking isn't a one-time event. Anyone with an addiction will tell you that it's a lifelong struggle.
14.
by RSchaeffer 6y ago
Why make people search instead of quoting the relevant section? "The human fasting mimicking diet (FMD) program is a plant-based diet program designed to attain fasting-like effects while providing micronutrient nourishment (vitamins,
15.
by RSchaeffer 6y ago
Then you're being far too generous with your interpretation
16.
by RSchaeffer 6y ago
Does anyone know how frequently these are offered?
17.
by RSchaeffer 7y ago
People frequently recommend Strang's teaching as an amazing pedagogical approach for engineers and applied mathematicians, but I find I'm frustrated every time I read his books or listen to his lectures. They don't work well
18.
by RSchaeffer 7y ago
But how did their model compare against others? The article only mentions how their interpretable model compared against their own ML attempts
19.
by RSchaeffer 7y ago
I just finished this fantastic class taught by Cengiz Pehlevan ( https://pehlevan.seas.harvard.edu/ ), so I thought I might share the lectures and exercises with HN.
20.
Harvard Applied Math 226 Neural Computation(github.com)
2 points by RSchaeffer 7y ago|1 comments
21.
Fundamental bounds on learning performance in neural circuits [pdf](rylanschaeffer.github.io)
1 points by RSchaeffer 7y ago|0 comments
22.
by RSchaeffer 7y ago
Why is a VAE not a generative model?
23.
by RSchaeffer 7y ago
Are you hiring? This looks interesting
24.
by RSchaeffer 8y ago
This is going to sound cynical, but I recently invested a week in rllib for a project before discovering that much of the under-the-hood implementation was horribly confusing, poorly documented and missing critical functionality (for instan
25.
Cryptographic preregistration: from Newton to fMRI(medium.com)
2 points by RSchaeffer 8y ago|0 comments
26.
by RSchaeffer 8y ago
I think you misread my question. I'm asking what math students will learn, not what math students should already know.
27.
by RSchaeffer 8y ago
Hey! Thanks for these great courses and materials! How much additional math (beyond high school and introductory college courses) do these courses teach? For example, if I were to take both courses, would I be able to understand the papers
28.
by RSchaeffer 8y ago
I'm curious to know how your paper differs from Learning and Querying Fast Generative Models for Reinforcement Learning. It seems relevant, but you don't mention it iirc.
29.
by RSchaeffer 9y ago
IIRC, his "Bayesian interpretation of the noise" actually shows that dropout performs approximate integration over model parameters. As he says, dropout doesn't work because of the noise but despite the noise. https:&#x2
30.
by RSchaeffer 9y ago
Go watch Yatin Gal's talk on dropout in neural networks. He shows pretty convincingly that the belief that dropout reduces network overfitting by introducing noise is wrong.

More ›