Matthew Skala (@mattskala) header
Matthew Skala (@mattskala) profile picture

Anchorite, apostate, asteroid.

Toronto, Ontario
Joined July 2009
Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)Photo by Matthew Skala (@mattskala)
I've launched my latest project, a video service called Matthew Explains focused on machine learning and computer science. (link in thread for algorithm reasons)
1
1
7
498
Incumbents call for regulation when it can convert a temporary advantage into a durable barrier to entry.
6
2
1
27
1,346
From today's free video on Matthew Explains.
1
3
298
Today's free video on Matthew Explains is about the Well-Actually test. Solving millenia-old questions of epistemology with large language models. Is there anything they can't do? matthewexplains.com/11188177…
2
62
Twitter DMs appear not to work anymore. I can't even read my old ones. Reminder that if you want to contact me, you should be using email.
2
99
Today's free video on Matthew Explains covers the introduction of Qwen3, a family of open-weights language models originating with AliBaba. Notable features of this family are the introduction of selectable "thinking mode," and less US politics than some other models contain.
1
1
63
But the training and alignment aren't open like the weights; and am I just not recognizing the political slant after all? matthewexplains.com/11185795…
1
55
Today's free video on Matthew Explains looks at the "goldfish" loss function, used in the recent Apertus language models. It's supposed to prevent copyright infringement by making the model unable to reproduce word-for-word quotations. Does that work? matthewexplains.com/11182442…
1
2
142
My latest video just went live, and as well as being free, it covers some points I know are of interest to people here regarding the distinctions among training, fine-tuning, and prompting.  I'm hoping it gets a lot of views. matthewexplains.com/11179488…
1
2
190
"offers him the opportunity to receive further blows"
1
2
189
The most charitable interpretation I can place on this quote from Google Gemma3-27B-IT is that because an LLM is capable of insulting people and not capable of actual violence, its trainers prioritize telling it "insults are bad!" Just hope they remember to update that later.
2
160
I envision Matthew Explains as being similar in nature to a graduate seminar. Most videos focus on recent or important scientific papers in the relevant disciplines, with each video presenting the content of one paper as well as my own thoughts about it. matthewexplains.com
110
Generate and read: Oh No They Didn't. video.northcoastsynthesis.co… It's thought you can get more truthful answers, and fewer hallucinations, from an LLM if you give it relevant Wikipedia documents in the prompt; but what if you generate pretend Wikipedia articles to use instead?
119
Dog-whistle GANs video.northcoastsynthesis.co… Free video from my new project, Matthew Explains. I go back to the original Generative Adversarial Nets paper from 2014, review how those work, and discuss the implications for watermarking and rock paper scissors lizard Spock.
128
Hey @canadapostcorp your online shipping tool for business doesn't work now - blank screen instead of "dashboard," blank screens instead of most other login-required pages after I log in and navigate to them. "Support" tool just sends me to the help file, which is unhelpful.
1
113