iiinova — where your digital self keeps evolving

Chris McMaster @rheumai.com · 24/03/2025

LLMs become much less useful the moment you dismantle bureaucracy.

Chris McMaster @rheumai.com · 17/03/2025

The modern internet penalises anyone who thinks for more than 3 seconds before forming a strong opinion. You’ll be happy to know that I only thought for 2 seconds before typing this.

030

Chris McMaster @rheumai.com · 28/02/2025

The major barrier here is that the massive budgets get spent on AI researcher/engineer compensation and GPUs, leaving very little to pay for the best domain experts. I think this is a massive miscalculation. The first lab to realise this will quickly become the market leader.

010

Chris McMaster @rheumai.com · 28/02/2025

It should therefore come as no surprise that new models like grok 3 and GPT 4.5 feel like small incremental improvements. The focus now should be on improving post-training data. Frankly, the quality kinda sucks even at the best labs.

120

Chris McMaster @rheumai.com · 28/02/2025

The idea that scaling up LLMs on human data will produce superhuman performance is magical thinking. The highest attainable performance in any given domain is simply the best human performance. Yes, maybe some insights span across domains, but I don’t think there’s actually evidence for that.

120

Chris McMaster @rheumai.com · 27/02/2025

The people who laugh about hacky “vibe coding” with LLMs are the same people who think that grok 3 is better than a doctor. Absolutely nuts.

020

Chris McMaster @rheumai.com · 24/02/2025

“Below is the updated code” followed by absolutely no code is such an o3-mini thing to do that at this point I don’t really understand why this model exists.

010

Chris McMaster @rheumai.com · 31/01/2025

After much time spent looking at reasoning traces from DeepSeek R1 for medical cases, I have to conclude that there isn’t a strong correlation between good reasoning and a good answer.

040

Chris McMaster @rheumai.com · 21/01/2025

The really interesting thing is that they’re not all made equal. Llama 8b distilled can solve medical cases that the Qwen-based R1 (and even R1 itself) cannot. World knowledge still matters for solving real problems and no open models beat Llama in that regard.

000

Chris McMaster @rheumai.com · 21/01/2025

Say the benchmark has 100 questions, generate 64 responses per question and then pass@1 is total number correct / 6400.

170

Chris McMaster @rheumai.com · 20/01/2025

These R1 distilled models are absolutely amazing on a single turn, but truly horrible on multi-turn conversations.

110

Chris McMaster @rheumai.com · 15/01/2025

"Mildly elevated rheumatoid factor has a very low positive predictive value that is completely overwhelmed in magnitude by the negative predictive value of not having any signs or symptoms of rheumatoid arthritis."

010

Chris McMaster @rheumai.com · 13/01/2025

My 2 year old has 3 adjectives for the size of things. In increasing order: small, mummy, big.

020

Chris McMaster @rheumai.com · 21/12/2024

5.7B tokens to solve 100 tasks??? I don’t understand why we’re thinking of this as being incredibly smart, when what this suggests is that it’s incredibly dumb.

010

Reposted by Chris McMaster

Leandro von Werra @lvwerra.bsky.social · 19/12/2024

Releasing Jupyter Agents - LLMs running data analysis directly in a notebook! The agent can load data, execute code, plot results and following your guidance and ideas! A very natural way to collaborate with an LLM over data and it's just scratching the surface of what's possible soon!

1144

Chris McMaster @rheumai.com · 17/12/2024

What percentage of “rhupus” is just misdiagnosed Sjögren?

131

Chris McMaster @rheumai.com · 17/12/2024

“ILD, hyperglobulinemia & Lab abnormalities” sounds like SjD to me!

000

Chris McMaster @rheumai.com · 17/12/2024

I just follow everyone and then spend all my time on the quiet posters feed (except when I want to come judge the yappers)

010

Chris McMaster @rheumai.com · 17/12/2024

o1 is equal parts brilliant and boring. Very, very boring.

010

Chris McMaster @rheumai.com · 16/12/2024

Llamafile is a cheat code.

000

Reposted by Chris McMaster

Kip Williams sort of @kipwilliams.bsky.social · 14/12/2024

"Zuckerberg's eyes brimmed with tears, and his heart felt full. He truly loved Big Brother!"

1144

Chris McMaster @rheumai.com · 15/12/2024

You better believe it. “California boomer” is the vibe I am getting from this guy.

000

Chris McMaster @rheumai.com · 15/12/2024

Prop stethoscope checks out, though.

000

Chris McMaster @rheumai.com · 15/12/2024

Sora’s idea of a hand exam. This 60 year-old rheumatologist is giving me very 2nd year medical student vibes with this bizarre technique. No synovitis was detected this day. #rheumsky

120

Chris McMaster @rheumai.com · 14/12/2024

Oh, and if you follow him then you’ll end up on a list, which will lead to you seeing less of these people. I genuinely have no interest in seeing his posts (I find him uniquely annoying), but it’s probably worth it.

1160

Chris McMaster @rheumai.com · 13/12/2024

Ultimately, if performance is anything like previous iterations of Phi, it will greatly underwhelm outside of benchmarks. So the license has no meaning to me.

140

Reposted by Chris McMaster

ACR Journals @acr-journals.bsky.social · 11/12/2024

Four-year follow-up of patients with hand osteoarthritis found that the course of pain was variable. Changes in pain were associated with BMI, employment status, mental wellbeing, illness perceptions, and coping styles In Arthritis Care & Research doi.org/10.1002/acr....

doi.org

Course of hand pain over four years in patients with hand osteoarthritis depends on BMI, employment and mental wellbeing: The Hand OSTeoArthritis in Secondary care cohort study

Objective We aimed to characterize hand osteoarthritis (OA) patients with deteriorating or improving hand pain, and to investigate patients achieving good clinical outcome after four years. Method...

0114

Reposted by Chris McMaster

Will Oremus @willoremus.com · 10/12/2024

Vlado Chernozemski, who assassinated King Alexander I of Yugoslavia, belonged in his youth to a club that was devoted to a game of regicide. In the game, known as "chess," contestants subtly maneuver into a position that allows them to topple the opponent's monarch.

3850241410

Reposted by Chris McMaster

Michael Harriot @michaelharriot.bsky.social · 02/12/2024

For five years, I have been covering what is unquestionably the biggest criminal justice scandal in American history. Today it ended when the most corrupt cop in history took his life. But it's not over. A thread.

8329341396

Chris McMaster @rheumai.com · 09/12/2024

It’s the PRs I’m waiting for.

000

Chris McMaster @rheumai.com · 09/12/2024

But does he have a TPOT anon?

110

Chris McMaster @rheumai.com · 09/12/2024

What if the embargo agreement with MKBHD was unique and they put in the wrong time zone (or none at all)? It seems possible that this is an unforced error from OpenAI that MKBHD cleverly exploited.

020

Chris McMaster @rheumai.com · 09/12/2024

Well, depression is a hypercortisolemic state, so almost certainly!

000

Chris McMaster @rheumai.com · 08/12/2024

Oh, you’re a domain expert who can contribute high-level insights from primary sources? Be prepared to have your contributions removed by some fedora-wearing know-nothing. But don’t worry, our algorithm will now let you know of your egregious violation before you even finish it.

030

Chris McMaster @rheumai.com · 08/12/2024

I love the fact that Wikipedia’s awful contributor guidelines have lead them to work on algorithms that force contributors to adhere to the guidelines, rather than just making better guidelines www.mediawiki.org/wiki/Edit_ch...

mediawiki.org

Edit check - MediaWiki

100

Chris McMaster @rheumai.com · 08/12/2024

Finally have COVID. 5 year streak comes to an end 😢

020

Chris McMaster @rheumai.com · 08/12/2024

Sort of guy to order an iced coffee and put it in the microwave.

030

Chris McMaster @rheumai.com · 08/12/2024

I think anthropic and pretty much anyone can beat o1 with better training data. I give o1 a complex problem, o1 thinks for 4 seconds, outputs a bad answer. Seriously, who trained this thing?

120

Reposted by Chris McMaster

Mark Lindsay @markelindsay.bsky.social · 07/12/2024

We need some normies here, too many word smiths, phrasing police, you can’t use that word posters, ie annoying left people

1121

Chris McMaster @rheumai.com · 07/12/2024

*this has become a cricket account until further notice

010

Chris McMaster @rheumai.com · 07/12/2024

Reports of Australian cricket’s death are greatly exaggerated*

120

Chris McMaster @rheumai.com · 07/12/2024

btw “alliterate” is a portmanteau of “alliterate” and “populate” 🙃

010

Chris McMaster @rheumai.com · 07/12/2024

“Prompting”, “priming” and “pre-filling” are the variants I get. I really think it should have a name that makes it easily distinguished from all the other p-words that alliterate the LLM literature.

110