Search results

I’ve been looking more at some of Google’s documents and models and I wonder if tuning a model or using attributed question answering may be options for ensuring more accuracy and easing manual verification of data. A couple of links which may be of interest...

@forestglip Thanks for sharing your progress, comparisons and results.I’m not surprised your brain is a little fried, mine has been just from following! There’s some really interesting ideas here.

Interesting, are you using sonnet or opus? If you don’t mind rate limits maybe it’s worth seeing what Gemini comes up with, I’ve been doing some simple experiments (unrelated to this) using it. While Gemini 1.5 Flash may not be powerful enough the limits are low, 1.5 Pro would take too long, but...

I guess that is what you’re hoping to minimise with RAG? Given forestglip’s original plan of doing this whole process manually, perhaps having a smaller set of data to go through that needs some manual checking to check for hallucinations would still be less work than going through everything...

I seem to remember Deepmind showing a demo (using Gemini I think) to help compile meta-analyses and then searching for updated papers to update the relevant datasets. How much this was staged and how much work is involved versus the old fashioned way I’m not sure.

After a bit of fiddling with the unintuitive OneDrive interface I have downloaded all the Agenda and Presentation documents. I can post them here or share them somewhere if it helps people? If you try to do the natural thing and select/click the icon of a document you’re prompted to login (I...

I had an email but there’s not a lot more info

I hope @MrMagoo gets any relevant help they need and finds the replies here useful. I also wanted to say although the occasional voice issues I have are minor I have always found them (like so many symptoms) somewhat interesting. So I’ve found the thread extremely informative, reassuring in a...

That’s a really interesting idea and I wonder if it could also be integrated with as a sort of quality mark process for researchers. I was thinking about the understandable concerns people have and mention of blacklists, while I can see the attraction they also come with risks, not least in how...

I think Nellie is talking about tools like Microsoft Teams or Slack which include a lot of collaboration features. They’re perhaps better thought of as a crossover of forums and chat clients but with document and many other collaborative features. Teams didn’t exist when I was working but Slack...

I’ve just completed the shorter form for people who may be interested in providing feedback via focus groups or forums and that has an edit your responses button at the end so it looks like you can? It looks like the URLs to edit are unique and include the token to access your data, so...

Many thanks to all involved. Especially Trish. It all feels very positive.

The point @Hoopoe makes is really important. People that get ME are rare, so if not this then those running the projects need to be those open enough to really listen. Worst is probably people who think they understand but don’t or are unwilling to listen or see/do things differently. I saw...

Some thoughts which echo what some others have said. I can’t attend meetings. In person is a complete impossibility. And the often offered ‘easy option’ of phone or video calls still completely rules me out. I can sometimes listen to a call for 10-15 mins but anything interactive is much less...

I have no idea I’m afraid other than to say ‘me too’. Not as bad as you’ve had it but my voice often goes quiet or hoarse/raspy as you describe when I am bad. And returns to ‘normal’ when I’m less bad. As with many symptoms I just put it in the large box of ‘one of those things’. I’d be...

I’ve processed the text of the paper to remove figures, references, URLs and formatting to make it better for text to speech engines and attached it to his post. I’ve also uploaded the audio to soundcloud. It’s not great quality but is hopefully easier than reading for some.

from the paper Another thought/question I have is how much these fundings correlate with results from the Raman spectroscopy study from a while back (https://onlinelibrary.wiley.com/doi/10.1002/advs.202302146) which I seem to remember showed changes around lipid metabolism as well as other...

There looks to be more detailed info in the paper although I’ve only had a quick look.

I may be wrong and the naming gets confusing but this looks like the larger UK Biobank rather than the smaller ME focussed UK ME CFS Biobank.

Another question I have…How feasible or indeed useful would it be to reproduce this with UK ME CFS Biobank samples? Or other international biobanks?