dofm

30 ms·

391.
by dofm 2mo ago
It also has the broader public health benefit of reducing the preventative use of that class of antibiotics, which lowers the risk of antibiotic resistance generally. (In an ideal world we'd never use them preventatively.) Could be a v
392.
by dofm 2mo ago
Absolutely yes, because the most severe risks are associated with the bacteria reaching the heart and into the nervous system, which is a relatively slow process. There's a window of a couple of weeks. If you could know as soon as you
393.
by dofm 2mo ago
Yeah. I think we simply did not see it enough and to be honest far too few doctors ask the "have you travelled" question about unusual disease. It is better now, but I think the "could it be?" thought doesn't occur
394.
by dofm 2mo ago
Wow, this seems a pretty big deal. Here in the UK, Lyme traditionally been a marginal problem (tick bites aren't) but we now have such rapid climate change, along with rapidly increasing wild deer and boar populations, that some areas
395.
by dofm 2mo ago
I really like the clean, neutral, not over-keen way Gemma writes, and I guess my unwieldy thesis is that the way it writes is in part a consequence of the culture of the team being international. Then again I quite like the way Muse Glimmer
396.
by dofm 2mo ago
Oh, that's cool, I saw those and I did wonder! Thank you.
397.
by dofm 2mo ago
Gemma 4 26B does really well at this specific task (and a general MySQL-related puzzle I test on). I rather like it and now they have fixed tool calling, I would use it. I think maybe it has been trained well with "consumer" progr
398.
by dofm 2mo ago
So far what I am seeing in my seemingly simple "Wordpress last-login plugin" test is that in xhigh reasoning mode (the default, seemingly) it overthinks so badly that it writes terrible bushy code with edge cases caused by going
399.
by dofm 2mo ago
Have you tried turning down the new Qwen's reasoning effort level from xhigh, which it defaults at? LM Studio isn't exposing a dropdown for this, at least with the unsloth build. Unsloth Studio / Desktop does.
400.
by dofm 2mo ago
It looks like the caveman speak is a facet of the 'xhigh' reasoning effort. Set (bodged!) to medium it is not caveman.
401.
by dofm 2mo ago
I think so too — it is something to test, for sure. ETA: a bit of testing before I climb the wooden hill to Bedfordshire. LM Studio doesn't seem to display the little dropdown to set reasoning effort, so I bodged the chat template on l
402.
by dofm 2mo ago
I would characterise it as obsessive, not loopy. It's definitely burning through a lot of tokens to ruminate about aspects of tasks that earlier models get done better seemingly on memory. I still need to understand that. Setting a r
403.
by dofm 2mo ago
I strongly suspect that many of these accounts you think might be bots from the labs are just people who only have one interest. One thing I have noted a lot more of is that there is comparative fanboying going on. Like "this model has
404.
by dofm 2mo ago
I was quite impressed by Muse Glimmer, and while I am sure people will observe that it is less good on benchmarks, my first experiences with this new 27B have been somewhat exasperating, whereas testing Muse Glimmer was rather fun. I have n
405.
by dofm 2mo ago
FWIW, Farage owes _most_ of his cultural visibility to the BBC's tendency to interview him, feature him on Question Time. He's been on that show nearly forty times! Binface and Farage will have got more or less equal time on the B
406.
by dofm 2mo ago
Notably, Binface got about 1500 more actual votes than that petition would need signatures, on a hot August day. The petition is published for six weeks! Farage must be very uncomfortable now.
407.
by dofm 2mo ago
Ahhh that is the one I said I couldn't remember the name of. Can't believe I did -- you still catch that installed, I think OctoPrint/OctoPi uses it?
408.
by dofm 2mo ago
Annnnd the code of my WP code test is not better. It is bushy, overcomplicated, and has gone around the houses to do stuff it would not need to do if it hadn't overthought. Oh dear. I need to try to understand what is going on here.
409.
by dofm 2mo ago
Also radically better on an M1 Max. I get well up into the 60s t/s with the A3B, stuck at 9.5 or so with this new 27B, though perhaps an MLX build will help.
410.
by dofm 2mo ago
Hah! Not usually, to be honest. I may have strayed into Gish galloping with the first one, though. But the second one, it actually needs to be broken down that way, because it's sort of complex point-by-point stuff. And I was thinking
411.
by dofm 2mo ago
There's a real change (compared to 3.6) in the way it writes in thinking — it drops words like "to" and "we" in "We need to", talks generally in note form, drops the/and all over the place, avoids &qu
412.
by dofm 2mo ago
I don't think the committee will be swayed by this stunt at all, especially since Farage has essentially admitted that he started all this to make them look silly. And they are highly likely to open another investigation into the Posh
413.
by dofm 2mo ago
(ETA: I see you edited your post to clarify — I will leave mine because there is significant nuance about the constituency) Not really. Clacton itself certainly wasn't; it has its problems for sure but other seaside towns do, and the c
414.
by dofm 2mo ago
Yes. Unless you're a male resident of Uxbridge of drinking age.
415.
by dofm 2mo ago
It didn't even unify the side that won, considering there were _two_ leave campaigns.
416.
by dofm 2mo ago
More votes in an absolute sense but nowhere near as large as a proportion — he got 24000 votes when he stood for London Mayor but that is a tiny, tiny proportion of the vote — the two leading candidates between them got two million. So he h
417.
by dofm 2mo ago
And this time he got enough votes that if they were to be signatures in any recall petition if triggered by the standards investigation, Farage will lose his seat and have to fight it again if he wants it.
418.
by dofm 2mo ago
Hahahaha — doesn't that story just tell you who he is, deep down. A shallow, transactional grifter.
419.
by dofm 2mo ago
This is facetious and silly. He was always going to win this election. Nobody practically-minded ever thought Binface winning was anything more than technically possible. What is your understanding about why the election happened?
420.
by dofm 2mo ago
OK so: - it won't stop the investigation, which has restarted - it won't stop the investigation triggering a recall petition if the findings are serious enough - the recall petition needs 10% of the electorate to sign it to trigge

More ›