My GitHub (Trash code I wrote during PhD)

#phdlife #github #researchcode

A brief browse through my public GitHub and musings about my old code.

Links:

If you want to support me, the best thing to do is to share out the content :)

If you want to support me financially (completely optional and voluntary, but a lot of people have asked for this):
Bitcoin (BTC): bc1q49lsw3q325tr58ygf8sudx2dqfguclvngvy2cq
Ethereum (ETH): 0x7ad3513E3B8f66799f507Aa7874b1B0eBC7F85e2
Litecoin (LTC): LQW2TRyKYetVC8WjFkhpPhtpbDM4Vw7r9m
Monero (XMR): 4ACL8AGrEo5hAir8A9CeVrW8pEauWvnp1WnSDZxW7tziCDLhZAGsgzhRQABDnFy8yuM9fWJDviJPHKRjV4FWt19CJZN9D4n

[aidegrod]

I love videos like these lately, they reveal more of a person, and make anyone more friendly to you. TY for this. I hope to get a Ph.D. in 2 years, by the end of my postgraduate studies, your analyzes of articles and links to them help a lot to understand all this mess with DL lately. And also helps with new studies. Feel same with old github projects now.

[ulf]

You wrote unit tests? That's already much much better than 99% of PhD spaghetti code repos, I'm usually tasked to deploy ...

[CristianGarcia]

TLDR; Yannic reimplementing MLOps stuff... we've all been there 😅

[glennkroegel]

You can die happy knowing that all of this will live on forever in the github arctic vault.

[JohnSmith-bmzg]

Doing my phd now. I can also confirm the code is utter filth.

[socratic-programmer]

It's not what most people watch your channel for, but I got a proper laugh out of this. That '__lambdas__' part at the end had me cackling. Congrats again on the PhD!

[lucasbeyer]

It felt like you went through my early PhD folders. Seriously, like 70% overlap!

[hlen]

There have been many times where the other PhDs canceled my work and essentially told me to fuck off when I told them I was using the server...
Giving them your code 1.5 years ago would save us so much work...

[michaelwangCH]

That is a process of learning, therefore github is optimal for student, coders and developers. Most projects and code are garbage.
That is similar to scientific publications and few papers will do right. That few will bring the scientific discovery one or even two steps forward. Like I said before, most publications are garbage. There are got published, because they have to keep their status in scientific community present and fresh. So they can claim position of higher salaries. The corporations are doing this for marketing and status reason to show their stakeholders of sophistications and financial strength. That is wast of talent and human resources. Specially, PhDs work for hedgfunds and investments.

[jakubkahoun]

My first github upload was lightsaber filter, that would put lightsaber(s) in your hand(s). (webcam input)

[alpers.]

I love having a lot of garbage repos in my github

[qq]

I wrote the equivalent of half of your rat tool for the same reason (along with the same hyperparameter naming). I think it's been the most qualitative code I've written so far in my PhD lol

[thongnguyen]

The man is proud of his code and it shows.

[XX-vujo]

Why do we as coders always upload our code and say its garbage? For me, its because I don’t want them to expect too much and criticize me too much. Probably also I have impostor’s syndrome.

This shit exempts the TechLead of course. He is the best.

[Kram]

Congrats on the PhD!
Still a fun collection of tools

[michaelwangCH]

Yannic, can you show us how can we visualize the high dimensioanl manifold of underlying pretrained-NN, graphically.
Tim, explained the problem in very understandable way, but we can not see it in graphical plot - is this even possible to visualize the very high dim. parameter space in very lower dimensions, without lose information?

Yannic, would you show us how you do it and make a tutorial video. Thanks.

[adithyavenkateswaran]

Some of the tools look pretty awesome (the GPU one especially). Could you (please) link your github page to the video description (to make it easier for lazy asses like me xD)?

[herp_derpingson]

gpustat - Look at the server > test > main script. Its funny :)
.
2:07 10/10 NO
.
8:35 I remember using the VS Code extension for latex for one of my Math assignment in 2016. It is also a "continuous compiler" with live reload and you can see the output live on the side of the screen.
.
Asking the big questions here. How did you manage to see anything in Github dark mode with the glasses on?

[atharvaparulekar]

I work in the industry and most of the code I write is garbage :p

[twobob]

I tried your video generatorwith Stylegan :) Funny