Table of contents & the books description seem a little gloomy to anyone else?
GNU Radio, filters, AM/FM, IQ demod ... I remember working through all these topics on GNU Radio Tutorials wiki [0] but I don't know if the book offers anything more of value?
Also, if the authors focus on GNU Radio as their software stack why would they not include a chapter on creating your own Python Blocks which is the biggest upside (imo) to GNU Radio. I love SDRs and think anyone interested in electrical engineering should play around with them. I dont know if I'd recommend this book based off what the sample chapter 4 provided.
I think sending people directly to GNU Radio is a bit of a risk. Sending folks that just learned how to spell SDR deep into the bowels of DSP is a bit steep of a learning curve that many might equate with a brick wall.
A little over ten years ago (!) I got started with a windows box, sdrsharp and a cheap RTL-based SDR. Just cruising around the spectrum, clicking on signals that were interesting, cobbling together decoding pipelines and getting real results was a way better way for me. Getting started with software that works and interesting use cases you can get into with cheap hardware got me hooked and THEN I had something that I was genuinely craving an understanding of to drive me into GNU Radio.
Agreed, GNU Radio exists in this weird no-mans land where it works on low-level concepts wired up with high-level guis. Do not want. I want to playing with CubicSDR to zoom around the spectrum, then if I I need to do anything fancy with the data, switch over to signal processing and start writing code.
I think this is the proper way to use llms for tasks that require high fidelity. currently im working on binary analysis using llms for natural language and letting ghidra/codeql do the symbolic work. scalability is a massive issue, perhaps the biggest besides fidelity.
its interesting to see many people come to the same neuro-symbolic conclusion around the same time.
last years falcon (crowdstrike specific conference) they for the first time every showed live the interviews of 3 north koreans trying to get a job in software engineering positions at some forture 500 companies. i was baffled at every 'security' question to validate the person is actually in the US gets glossed over like: "my ID is at my home right now, and im in my office so i don't have that with me".
I mean you see that here on HN right? People claiming that any arbitrary question is something they have no idea about, like the color of their front door.
I’m not sure I know what you mean—I’m not sure I’d want to discuss the specifics of my living environment here though. Would you have any examples handy?
If your resume says you live in NYC for example, and I do something like "Man, I went to NYC once and got stuck in traffic on that stupid highway that goes up and down the coast of Brooklyn, what was the name of that thing?" and they respond with I-278, that would raise red flags. I have never heard of anyone calling the I-278 anything but the BQE.
It's just like the bar scene in Inglorious Bastards, with the fingers. There are so many obvious tells you can have people divulge if they aren't actually telling the truth.
> If your resume says you live in NYC for example, and I do something like "Man, I went to NYC once and got stuck in traffic on that stupid highway that goes up and down the coast of Brooklyn, what was the name of that thing?" and they respond with I-278, that would raise red flags. I have never heard of anyone calling the I-278 anything but the BQE.
A counterstory: When my former boss started at the company, for the first years [!] he only "knew" very specific places (office, appartment, and one or two places associated with intensely practiced hobbies of him) in the city where the company is located, and basically lived inside the bubbles associated with these places and their surroundings.
Thus, to me it is very plausible that even if you lived in a city for many years, it is very easy to live in very isolated bubbles, and have barely any contact to people and their habits outside these bubbles.
I had a candidate who said he lived in San Francisco, so I asked him what neighborhood, and he responded "Uh, by the Golden Gate Bridge." Cool.
Later I looked more closely at the resume and saw some more red flags, like, he had a degree from "CA State University" -- like, which of the 23 CSUs bro?
We did have a couple fake people make it to the final round, the last one was cheating and still bombing -- I sent a picture to the guy who did the second-round interview like "is this the Jason Smith you interviewed?" and he said "Lol, no"
> "Man, I went to NYC once and got stuck in traffic on that stupid highway that goes up and down the coast of Brooklyn, what was the name of that thing?"
I lived in NYC for a year and I have no clue. My answer would be probably something along the line of "Haha! Yeah. Traffic is terrible in the city... or so do my friends with cars say. I for one take the subway everywhere, so no clue what you are talking about. But sounds like a pain! Hope you were not delayed too long."
> It's just like the bar scene in Inglorious Bastards, with the fingers.
The problem is that's a work of fiction. These shibboleth tests work great in fiction where the author has full control over the whole universe. Work less well in reality where "universal" signals turn out to be a lot less universal. You will have a ton of false positives and a ton of false negatives.
I could answer questions about the lines I used, yes. Doesn’t mean that I studied the subway maps.
But my point is that the “everyone in X calls Y Z” kind of trivia is not reliable way to say if someone is in X. For example because not everyone in X is native to X. Also because many would use the proper and official name of the landmark in an interview setting.
imo its a little more difficult to publish at these conferences using: "take any architecture and any corresponding benchmark, tweak the architecture slightly and publish a paper". at t2-t3 conferences ... sure.
One of the biggest problems i have with running basic text editors like vim/nvim is the investment time to spin up a fully loaded workable development env; esp since i've never done it before. basic vim with some modifications in .vimrc is all i have and i know some of my colleagues are also this way!
nowadays though i really want to use LLMs to write code for me instead of switching contexts on different platforms. can i ask what you use for LLM stuff on nvim? how do you like it compared to running bare bones vim and switching platforms?
It works with vanilla vim (and MacVim) these days too. "Vim 9.0.0185 or newer" is mentioned in the Getting Started section.
I was about to install it a couple of years ago, but then started thinking about the privacy threat model.
I realized that having a "copilot" in my everyday editor (not just for public open source coding!) is never gonna fly. I may end up accidentally uploading any file I open to a 3rd party for tab completion and "AI stuff". Even if I can configure it to hopefully ignore some directories, too risky for me. With a separate editor just for coding (Zed in my case), the risk of accidentally opening and uploading a wrong file would be much lower for me, as I'll keep using vim without any AI for everything other than OSS coding.
Edit: I'm sure there's an option only manually load the copilot plugin when you explicitly want it, but it still makes me too uneasy.
i know europe is making it easier for grants/proposals to go through and lessening the overall time it takes for grants to be accepted, for people in academia leaving the US
Vouch for this approach ... not just in ml/dl but as a general way of learning things in life.
resources like this are useful, in an academic setting. lets not forget we all forget 50% or more of what we learn about 20 minutes afterwards unless we consistently remind ourselves of what we learn.
unless others prove me wrong using personal anecdotes.
FPGAs are widely available. They are not the cheapest hardware, but they allow you to have something similar to custom silicon, but quickly and in quantities starting from 1.
Modern CPU architecture for example the Versal architecture from AMD has seamless integration with compute accelerators namely FPGA, GPU, DSP, etc [1].
They even has built-in ADC/DAC for communicatiobmn, sensing and data acquisition (DAQ) [2].
On top of that they have native support for dataflow library and framework including linear algebra that is more suitable for data-intensive in which AI is only one of the intended applications [3].
I was recently asked by my colleague why we still need CPU, since GPU is very dominant now and it seems it's all that we need. I just pointed out GPU is only one of the many accelerators available to the CPU, but a very good one that happened to be very useful and biased towards fancy killer applications namely graphic (game), blockchain (bitcoin) and AI (LLM/ChatGPT).
GNU Radio, filters, AM/FM, IQ demod ... I remember working through all these topics on GNU Radio Tutorials wiki [0] but I don't know if the book offers anything more of value?
Also, if the authors focus on GNU Radio as their software stack why would they not include a chapter on creating your own Python Blocks which is the biggest upside (imo) to GNU Radio. I love SDRs and think anyone interested in electrical engineering should play around with them. I dont know if I'd recommend this book based off what the sample chapter 4 provided.
[0] https://wiki.gnuradio.org/index.php/Tutorials