Yeah I’d really like the idea of talking to a smart home but can’t justify the power consumption of leaving such a big box running 24/7.
I also don’t think we need models this large for that purpose. It just needs to understand the semantics of home automation. Being able to compose limericks is a parlour trick that’s not worth the overhead.
Reinforcement learning involves a reward signal (e.g. a score in a game) which I don’t think is present here.
Diffusion models, as you’ve also mentioned, seem a better metaphor. These try to generate a structured image (e.g. matching a prompt) from noise. Perhaps your visual cortex is just trying to make sense of random sensory input while your eyes are closed.
It’s also interesting to think about dreaming in terms of the more general set of representation learning techniques. As I understand it you’re trying to process the day’s experiences and reflect on past memories - sifting through and deciding what to retain and what to forget - essentially mental filing.
You may be interested in Deep dream. This is a program that runs an e.g. convolutional neural network in reverse. Instead of adjusting it’s belief about whether an image should be classified as a dog or not it adjusts a given input image so that it looks more and more like a dog. The results are pretty psychedelic!
This is basically lucid dreaming.
If you make notes of what you remember of your dreams when you first wake up then after a few days you’ll become conscious while dreaming. Then you can basically decide what to dream.
It’s kind of fun flying around but I stopped doing it as I didn’t find it particularly restful.
This reminds me of the book “Only you can save mankind” by Terry Pratchett. The aliens surrender once they realise the player is apparently immortal!
When you use the KDE Desktop Environment it will also show notifications from your phone and sync media controls (e.g. stop playback when you get a phone call). It’s great!
So does e.g. rm $HOPEFULLY_SAFE but we don’t blame rm.
Possibly because it’s very NVIDIA focused?
I’m excited about the prospect of small models e.g. fine-tuning against MCP calls to give APIs language interfaces.
The yellow Lupulin powder is hidden inside the cones. You can only really see it when you peel the layers back.
You picked them with bare hands?! I wear gloves as the little hairs on the leaves and bines scratch my skin to shreds.
I can see why the “you’re spending too long on twitter” stuff seems a bit creepy but something like this would really help me remember what I was doing to prepare timesheets/ invoices! I currently have to wade through chat logs/ git/ file modification dates to try and figure it out which is a pain.
This looks great - thanks a lot!
Can you really isolate a single cell without a laboratory?! What sort of equipment? This sounds more involved than just innoculating some wort with a spoonful of yeast sediment…
I tried to look at yeast under a 40x-640x microscope but just saw a brown blur! I presumed you would need expensive stains/ growth media to experiment properly.
Could you elaborate on what you mean by “1-cell” steps?
I didn’t realise yeasts have “killing features” either… I thought the whole idea with fermentation was to create the right conditions for your favoured micro-organism to out compete the rest!
You can leave soil out on a tarp in the sun but as Gunther says you don’t want to sterilise it completely. It’s the bacteria that break down the organic matter into nutrients the plants can absorb.
If you’re worried about the blight returning and don’t want to replace the soil I’d think about switching crop for next year. Maybe some nitrogen fixing beans instead?
Idk isn’t it just that the name is misleading?!
Brilliant. This looks great. Thanks for taking the time for a thorough write up!
Great idea! We’re overrun with Lemon Balm!
How did you extract the flavour? Did you add it to the boil or make a vodka tincture or something?
In fact I’d really be up for trying this out if you’d care to share your recipe?
It looks like a Windows XP activation key!
Salt, Fat, Acid, Heat by Samin Nosrat does this. It begins by introducing the eponymous 4 elements at length then provides a decision tree to help you work through the recipes at the back according to the gaps in your experience.
It’s not specifically focused on Indian food but does contrast flavour profiles from different countries. For British Indian Restaurant Curry I’d recommend Dan Toombs “The Curry Guy”. It’s not written like a textbook but I’ve learned loads of tricks from it (e.g. base curry sauce/ freezing fried paneer for later, yoghurt marinades etc).
230B parameters is “reasonably small” now?!