Newtra

The website does a bad job explaining what its current state actually is. Here’s the GitHub repo’s explanation:

Memory Cache is a project that allows you to save a webpage while you’re browsing in Firefox as a PDF, and save it to a synchronized folder that can be used in conjunction with privateGPT to augment a local language model.

So it’s just a way to get data from browser into privateGPT, which is:

PrivateGPT is a production-ready AI project that allows you to ask questions about your documents using the power of Large Language Models (LLMs), even in scenarios without an Internet connection. The project provides an API offering all the primitives required to build private, context-aware AI applications.

So basically something you can ask questions like “how much butter is needed for that recipe I saw last week?” and “what are the big trends across the news sites I’ve looked at recently?”. But eventually it’ll automatically summarize and data mine everything you look at to help you learn/explore.

Neat.

I agree that older commercialized battery types aren’t so interesting, but my point was about all the battery types that haven’t had enough R&D yet to be commercially mass-produced.

Power grids don’t care much about density - they can build batteries where land is cheap, and for fire control they need to artificially space out higher-density batteries anyway. There are heaps of known chemistries that might be cheaper per unit stored (molten salt batteries, flow batteries, and solid state batteries based on cheaper metals), but many only make sense for energy grid applications because they’re too big/heavy for anything portable.

I’m saying it’s nuts that lithium ion is being used for cases where energy density isn’t important. It’s a bit like using bottled water on a farm because you don’t want to pay to get the nearby river water tested. It’s great that sodium ion could bring new economics to grid energy storage, but weird that the only reason it got developed in the first place was for a completely different industry.

Arg, my first reply was completely off topic. I thought this was a reply in another post.

Yeah, it’s great that VRE is cheaper now, but we shouldn’t celebrate companies/countries for just taking the cheapest option. It drowns out the legitimate celebration of the companies/countries that are actually making hard decisions by funding renewables when they aren’t the cheapest option, investing in long-term R&D, taxing carbon, fixing bureaucratically-entrenched perverse incentive structures, etc.

deleted by creator

This is awesome news. Not because of the car, but because it builds the supply lines for an alternative battery chemistry.

People have been using lithium-ion batteries for home and grid storage, which is nuts if you compare it to other battery types. Lithium is expensive and polluting and only makes sense if you’re limited by weight & space. Cheaper batteries, even if they’re bigger/heavier, will do wonders to the economics of sustainable electricity production.

An increasing proportion of renewables doesn’t necessarily mean a decrease in overall carbon output. Our per-person electricity consumption keeps rising. AI, electric cars, crypto, air conditioning to mitigate climate change, etc. all demand more electricity each year as they become more popular.

Wins don’t come from new growth being sustainable. We need to be actively shutting down the existing unsustainable energy production. It doesn’t matter whether this is done by replacing it with renewables, or by reducing our consumption with e.g. efficiency standards for AI and cars.

China is also still building new coal plants at a truly alarming rate.

Don’t let heavy carbon emitters steer the narrative this way. Building renewables is just the cheapest way to keep expanding your energy grid at there moment, but if you’re not actively taking power plants offline to reduce carbon emissions, you’re not actually getting greener.

EDIT: LMAO I’m being mass-downvoted. This is why it’s important to think critically about every headline about China - there is an army of propagandists trying to make sure you only see the good stuff.

But the comments below say they’re not able to access the new page, even with the direct URL… It seems certain tiers of customers can’t opt out. Possibly they can’t be included in the first place (e.g. EU users), but it’s a pretty big screw up to hide one’s status on such an important privacy setting.

Honestly, I don’t think that there’s room for a competitor until a whole new paradigm is found. PyTorch’s community is the biggest and still growing. With their recent focus on compilation, not only are TF and Jax losing any chance at having an advantage, but the barrier to entry for new competitors is becoming much higher. Compilation takes a LOT of development time to implement, and it’s hard to ignore 50-200% performance boosts.

Community size tends to ultimately drive open source software adoption. You can see the same with the web frameworks - in the end, most people didn’t learn React because it was the best available library, they learned it because the massive community had published so many tutorials and driven so many job adverts that it was a no-brainer to choose it over Angular, Vue, etc. Only the paradigm-shift libraries like Svelte and Htmx have had a chance at chipping away at React’s dominance.

The easiest way to get the basics is to search for articles, online courses, and youtube videos about the specific modules you’re interested in. Papers are written for people who are already deep in the field. You’ll get there, but they’re not the most efficient way to get up to speed. I have no experience with textbooks.

It helps to think of PyTorch as just a fancy math library. It has some well-documented frameworky structure (nn.Module) and a few differentiation engines, but all the deep learning-specific classes/functions (Conv2d, BatchNorm1d, ReLU, etc.) are just optimized math under the hood.

You can see the math by looking for projects that reimplement everything in numpy, e.g. picoGPT or ConvNet in NumPy.

If you can’t get your head around the tensor operations, I suggest searching for “explainers”. Basically for every impactful module there will be a bunch of “(module) Explained” articles or videos out there, e.g. Grouped Convolution, What are Residual Connections. There are also ones for entire models, e.g. The Illustrated Transformer. Once you start googling specific modules’ explainers, you’ll find people who have made mountains of them - I suggest going through their guides and learning everything that seems relevant to what you’re working on.

If you’re not getting an explanation of something, just google and find another one. People have done an incredible job of making this information freely accessible in many different formats. I basically learned my way from webdev to an AI career with a couple years of casually watching YouTube videos.

I’m glad to hear I’m not missing out on anything. (It’s still not out in Europe.)

Keto-ade can be great for taking the edge off keto flu. It’s basically a home-made concentrated sports drink - artificial sweetener, potassium & magnesium salts and a lot of water. All the fluid & electrolytes help your kidneys filter out the metabolized fat that would otherwise linger in your bloodstream.

Some minor/hard-to-notice health-related things can dramatically reduce alcohol tolerance and/or give “hangovers” shortly after starting a session.

For me, inflammation is a big cause. I have (barely noticeable) cat allergies, and (obvious but hard to avoid) food intolerances & gut issues. If I don’t stay on top of avoiding triggers, my alcohol tolerance goes from multiple G&Ts giving a nice buzz, to 1-2 sips of G&T giving dizziness and headaches. Electrolyte imbalance can also cause it. I’ve found I have to add magnesium and potassium salt to my diet, or else I generally feel tired more, and my alcohol tolerance plummets. Once you start controlling these factors, you’ll start getting clear feedback from your body when you have too much or too little salt, in the form of water and food tasting different and general feelings of tension or tiredness.

My advice: try antihistamines, easily-digestible meals, and/or sports drinks for a few days before you drink. If those help your tolerance, you probably have some health stuff going on - figure it out and you’ll probably find a way to generally feel better.

They’ve had days to prepare this response. They didn’t rescind or explain the one thing that people universally hated, which means they’re just stalling and trying to save their reputation without actually changing trajectory.

We’ve seen this corporate bullshit so much in recent years. No more “benefit of the doubt”.

ooo, I love this. It reminds me of how nice C#'s LINQ is…

“Pipeline style” DB queries have some interesting advantages as well:

• It’s straightforward to write efficient queries for DBs that don’t include a query optimizer stares at Datomic
• You can split the pipeline into server-side and client-side steps when working with less capable DBs stares at most of NoSQL
• It would be much easier to transition from a pipeline API to a non-text-based API so that our ORMs/query builders can directly talk to DBs without the overhead of generating and parsing SQL.

I still use Google for ~95% of my queries because I like real sources, comprehensive documentation, and not having to read a wall of text when a one-line answer would have sufficed.

ChatGPT is a good replacement for Quora/Stack Exchange for explaining general knowledge stuff like other languages’ grammar and simple science, as well as finding authors/books/movies from descriptions when you’ve forgotten their names.

Bard is… kinda dumb. I gave it a few chances, but it was nothing compared to ChatGPT’s free tier.