jkdufair

From personal experience: "yes, no question", and "maybe though it'll take a long time and a lot of compute and training data". The day level 3 automation (basically what Tesla's FSD is, even though it's classified as level 2) comes to some other car, I'll strongly consider switching.

Maybe a stock crash and total failure of Tesla will be the best thing that happens. The US government (assuming us citizens can somehow continue to claw it back) or some benevolent entity can buy the ML models for cheap and open source them. It's really their only truly valuable, resilient asset, IMO.

This was my initial reaction as well. I have been thoroughly invested in the craft of software engineering for 3+ decades. From the blog posts mentioned in the all the way back to Byte magazine at the library as a 10 year old. But those days appear to be perhaps behind us now, as suggested in the article. I am grieving it, despite the fact I have embraced LLMs for creating software.

But I refuse to use LLMs for things like making music. I like your term "purely human". There's no place for fancy statistical engines in human art. Art belongs to human communities and human culture. Which is where, apparently, I will continue to find my refuge if not my living.

I believe efcore is really well designed and handles the ORM tradeoffs in a very usable and mostly efficient way. And someone would have to pry LINQ out of my cold, dead hands. SQL is fine and I'm glad I know it. But I thank god I almost never have to use it.

It's a single piezo pickup per string

I think the parent was referring to the Variax, which models guitars, not amps. That line has been discontinued, sadly for me who plays one. I hadn't thought of using AI to reverse engineer the modeling, implemented in firmware in the guitar. That could be interesting.

If claude says "load bearing" once more, I think I'll vomit.

I really enjoy and appreciate how @yaxu does work in both textiles and live coding. I appreciate the abstractions he explores and I think it directly benefits the Tidal/Strudel projects.

Getting easier to avoid buying their oil. One of the reasons I switched to an EV.

Pretty sure most corn is grown for animal feed. We could help by eating less - or no - meat.

Whether or not someone is allowed to distribute creative works without the permission of the author, the point of copyright, is and will contribute to be independent of the technology used to do the distribution.

Making a copy in RAM, cache etc does not involve distributing it to other people.

I don’t disagree with the ruling. IA is not lending their copy that they purchased. They are lending a copy of their copy. While that copy is lent out, they are still free to, I.e. read the copy in their possession.

One of the main reasons I choose to shop at Aldi is that when I am in the store, I feel like the customer vs. feeling like the product at Kroger. Shopping at Kroger (or Walmart, etc.) feels like I am being assaulted.

Heard they have outstanding mountain bike trails and other good outdoor activity options.

If you are an Emacs user, check out Smudge. https://github.com/danielfm/smudge

It gives you just such a list.

LINQPad

RadPower bikes have hit the sweet spot between price and quality.

A 5 mile commute on an eBike is trivial.

Add button to GitLab files/diffs to open them in your local editor. Currently supports VSCode and Emacs.

Plotting assassinations of governors.

It is indeed surprising to me that people can tolerate the low bitrate sound. I would consider subscribing but I can’t stand to actually listen. Do subscribers get a higher bitrate? My impression is that it’s a limitation of the satellite signal bandwidth and the number of channels they cram into it.

As a die hard Emacser, I’ve always been drawn to the idea of plain text productivity. What keeps me using apps like Todoist and Calendar.app is I am unaware of a way to have location-based reminders, collaborative to-do lists, and shared calendars. Is anyone doing this with plain text?

That sounds like recipe for near-zero participation. I think both the intent and execution are spot-on here for GitLab. They are offering a very straightforward option for using GitLab, telemetry-free.

I would have generally agreed with this sentiment until we took a leap of faith and chose Entity Framework 4-5 years ago. That, coupled with the power of LINQ has allowed us to do things we simply wouldn’t be able to do otherwise. Pass IQueryables around without realizing them right away. This leads to functional, composable queries. And the queries we generate would simply not be possible to be written by hand. If we need to write SQL, we can and do. But with good design and good indexes, we rarely, if ever, do. I was skeptical, but EF is a tool we have not regretted.

It's an internal grant and I'm not sure I'd be allowed to share it. We are adding AES to our peer-review app. Currently as an additional "grader" to the peer reviews since that's what the PI requested. Since the tool allows unlimited submissions until the review date, I hope to add it as a "pre-flight" estimate to give students a chance to get a rough prediction of the score they will receive and a metric they can use as they revise until the due date.

I'm not aware of any meta-analyses myself. I have been keeping up with the ASAP competition and various attempts to improve on the initial systems for a number of years. The two papers I believe are having the most success are [1] and [2]. [3] seems promising for balancing the opposing forces of high accuracy for true positives and the risk of false positives via adversarially crafted inputs.

I'm also vaguely aware of research happening around extracting features from neural nets. I'd love to be able to help students understand why the system is predicting a particular score.

[1] https://www.aclweb.org/anthology/D16-1193 [2] https://arxiv.org/pdf/1606.04289.pdf [3] https://arxiv.org/pdf/1804.06898.pdf

I have a grant where were are doing just that. Implementing more or less SOTA research using fairly vanilla LSTM networks from 2-3 years ago (primarily Taghipour & Ng) to provide low stakes feedback to students on their essays in one of our teaching tools at Purdue. It’s based on research using the Kaggle ASAP database and we have found it to be pretty accurate across a variety of domains in early testing. Though some essay prompts seem to do better with CNNs vs. RNNs. I doubt many of the systems in TFA are based on LSTMs or neural nets at all. They are probably doing regression on hand-crafted features.

Alas there are some pretty generous tax breaks on gifts.

If I were the interviewee and discovered it was a poor fit, I’d consider the lack of an offer or the opportunity to refuse one a win

Emacs has multiple cursors: https://github.com/magnars/multiple-cursors.el

I used VSCode for a few years, but never found a way to work with code in semantic units vs. as text. And I like that no matter what language I'm working in, the experience is the same. VSCode plugin editors may or may not (read: wouldn't have) agreed on a set of conventions for working with code independent of the programming language.

This is a library we built to simplify access to RESTful APIs in applications using Redux. After a few apps built using thunks for RESTful access, we wanted something that was a bit more opinionated and required less ceremony. There is also an accompanying library, (studiokit-auth-js) [https://github.com/purdue-tlt/studiokit-auth-js], that sits on top of this library and handles authentication and bearer token management. We've used it in production at Purdue for about 18 months now. As the original architect of this library, I thought I'd post it in the event it's useful to others.