TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Numpyro: Probabilistic programming with NumPy powered by Jax

115 点作者 lnyan6 个月前

7 条评论

ghgr6 个月前
If you are struggling to understand the README, I highly recommend the book <i>Statistical Rethinking: A Bayesian Course with Examples in R and Stan</i> by Richard McElreath. Although the examples are in R, the same concepts apply to Pyro (and NumPyro)<p>[1] <a href="https:&#x2F;&#x2F;www.goodreads.com&#x2F;book&#x2F;show&#x2F;26619686-statistical-rethinking" rel="nofollow">https:&#x2F;&#x2F;www.goodreads.com&#x2F;book&#x2F;show&#x2F;26619686-statistical-ret...</a>
评论 #42157459 未加载
评论 #42161237 未加载
评论 #42160751 未加载
radarsat16 个月前
In some downstream applications such as filtering data (say, good&#x2F;bad), I am training simple NN classifiers based on relatively small datasets. So, my personal confidence in the classifier is not so high, I&#x27;d like to reject things that are &quot;definitely bad&quot; and keep anything that may be good. Even more, I&#x27;d like to put aside &quot;maybe good&quot; data for human verification and keep &quot;definitely good&quot; data.<p>In other words, I think I have a practical use case for calibrated confidence scores, which I definitely don&#x27;t get from my NN classifiers. They are right a certain percentage of the time, which is great, but when they are wrong sometimes they still have high confidence scores. So it&#x27;s hard to make a firm decision based on the result without manually reviewing everything.<p>So my question is: is this an appropriate use case for PyRo? Will training my NN classifiers blindly converted to probabilistic classifiers and sampled appropriately give me actually reliable and useful confidence scores for this purpose? Is that the intended usage for this stuff?
评论 #42164146 未加载
diab0lic6 个月前
For those with more experience how does (Num)Pyro compare with PyMC? I haven’t had the good fortune of working with any of these libraries since before Pyro (and presumably numpyro), and with PyMC3 back when it used Theano under the hood.<p>Are the two libraries in competition? Or complimentary? I’ve been playing with PyMC for a personal project and am curious what I might gain from investigating (Num)Pyro?
评论 #42160314 未加载
评论 #42158049 未加载
评论 #42158146 未加载
评论 #42158456 未加载
评论 #42158588 未加载
评论 #42162773 未加载
snthpy6 个月前
Related question: are there there any algorithms &#x2F; optimizations for probabilistic programming in an online context?<p>What i mean is that I have a model and I&#x27;ve run my inference on my historical data. Now I have new observations streaming in and I want to update my inference in an efficient manner. Basically I&#x27;d like something like a Kalman Filter for general probabilistic models.
评论 #42163091 未加载
评论 #42163150 未加载
评论 #42162860 未加载
评论 #42162407 未加载
whimsicalism6 个月前
What is probabilistic programming actually useful for? Interpretable inference or something like that?<p>It seems like if raw prediction of a data distribution is what you are interested in, explicitly specified statistical models are probably less useful? At least if you have lots of data and can tolerate a model with lots of &#x27;variance&#x27;
评论 #42172055 未加载
评论 #42161350 未加载
uptownfunk6 个月前
What is this used for?
评论 #42160872 未加载
评论 #42160541 未加载
svilen_dobrev6 个月前
side note: the Pyro used here - or actually pyro-ppl - is not to be confused with Pyro as of python-remote-objects..