- TiramisuBased on a classic tiramisu recipe. I adjusted portion size to 1/4 of the recipe in the link. Then I halved sugar because @jameshoffmann recommended that (and I’m happy with that decision, I liked the resulting sweetness level). Then I used 12 biscuits instead of 14 because that many were in the package. I thinkContinue reading “Tiramisu”
- My washer-dryer kinda sucksAnd they all might suck. About 18 months ago I bought a washer-dryer. At first I liked a lot. Now I think it suck. And I think the most important problem it has might affect other washer-dryers too. Basically the problem is that: So those can be worked with if: In addition to the limitationsContinue reading “My washer-dryer kinda sucks”
- Numbers on tuna, mercury, and pregnancy found on the internetI kind of knew that: But some time ago I did a quick search for how much tuna it’s safe to eat when pregnant. None wants to give direct answer. At least not that I could find. I could find a few numbers that one can do a little bit of very simple math withContinue reading “Numbers on tuna, mercury, and pregnancy found on the internet”
- English muffinsAdapted from this recipe. Scaled, some notes on how I like to make it. Mix everything together, put in a container, put in fridge overnight (I like taste better this way than quick rising at room temperature). They come out looking rustic, tasting delicious. It’s a bunch of wall time (overnight) but little work timeContinue reading “English muffins”
- Shallow pattern matching – 14 found article on predicting coinsI found an article I was looking for A reply to Francois Chollet on intelligence explosion . I think I mentioned in one of the previous notes. I couldn’t find it before. The part I was remembering is about predicting series of coin flips, it compares Laplace rule of succession, max entropy, and being ableContinue reading “Shallow pattern matching – 14 found article on predicting coins”
- Setting up TF and datasets on GCE VMI wanted to set up VM on GCE with: To see how much time some things take. This took a bit longer than I expected so I thought I’ll write down what I did. Step 1: Creating VM. When creating VM remember to configure disk size. I think default is 10GB and that’s too smallContinue reading “Setting up TF and datasets on GCE VM”
- Note on why roads have infinite demand at particular level of trafficInspired by Housing and Transit Roundup #3. I’ve been thinking about the bit at the end: I’ve read the Wikipedia article on the phenomenon and the piece in Wired and seen the study results, and they imply very strange things. If a 10% increase in capacity within a few years creates a 10% increase inContinue reading “Note on why roads have infinite demand at particular level of traffic”
- Why floating point numbers in neural nets?I was thinking about implementing simple logic functions as neural nets (as an exercise to get a better feel for how they work as “universal approximators”). And I noticed that using floating-point numbers makes things more complicated. Well I started with logical functions but I think it’s easier to show the problem with something likeContinue reading “Why floating point numbers in neural nets?”
- Why NNsSo I’ve been thinking about why we use neural nets. And I think it’s a way to encode computations, that has the following advantages: Let’s compare to a python program. [A1] I can change any weight or bias to almost any value and I’ll get a valid NN as a result (it’s probably a goodContinue reading “Why NNs”
- Note on neural network trainingThis is something I thought when I was thinking about choosing a topic for my masters thesis. I think I’ve had this thought a few times already so I think I should write it down. There were a few things that made me think about how having good examples / evaluating outputs is really aContinue reading “Note on neural network training”
- Note on VNM theoremI realized that I didn’t quite know what’s up with utility. And it bothered me so I checked a bit. I remembered someone mentioning VNM theorem and saying that violating its assumptions will result in silly / exploitable behavior. But then I’ve read someone claiming that (financial) markets can’t be represented as having an utilityContinue reading “Note on VNM theorem”
- Visualizing a specific pivotal actThis one might be worth cleaning up and turning into a nice, readable blog post. Or it might not, I’m not yet sure, see epistemic status later. For now its dumping all the thoughts that are attached to the topic in last couple days. Historia Civilis video on battle of Actium came out a fewContinue reading “Visualizing a specific pivotal act”
- Quick note on equityI’m confused about equity. Yesterday post from Zvi reminded me about my confusion. Some feel like they’re fitting together on this topic. Also I don’t want to think to much about this right now and I feel like it’l be easier for me to stop thinking about this if I write things down. So I’mContinue reading “Quick note on equity”
- Couple of links on pseudotheorem themeCurrently I’m sick so I can’t read those with as much attention as I’d like. I found a couple of posts worth reading carefully. My impression is that in them someone else wrote down significant amount of thinking I kind of would have to do to move forward with ideas described in the previous post:
- PseudotheoremsSo today on a walk I had a thought related to Agent Foundations (Specifically Logical Uncertainty). This is very vague but I want to make a note so that I don’t forget. Not sure if this is any good. So I think that a lot of thinking in rigorous domains (math, engineering) is that weContinue reading “Pseudotheorems”
- Note on Pascal muggingInspired by this comment So I think this approach is wrong. You see someone on the street. What do you think is probability that if you give them 1$ they’ll give you {big number}$ tomorrow? Probably the probability is >0 (because you don’t want to go to Bayes Hell(can’t find link)) but not enough toContinue reading “Note on Pascal mugging”
- Note on deep understanding (part 13)Today on a walk I was thinking about this some more. And I want to quickly write down my thoughts. So as a part of my job I sometimes conduct job interviews for software engineers. And last week I conducted one. I’m not supposed to write questions I ask in public places so I’m goingContinue reading “Note on deep understanding (part 13)”
- Quick note on speedI’ve read Some reasons to work on productivity and velocity some time in the last 10 days and it’s been on my mind since then. It makes some very good points. It matters how much time and effort something takes. Being able to do things faster and more easily is great. On the other handContinue reading “Quick note on speed”
- Waffle recipe that works for meFrom here but scaled to 1 egg (because that’s more than enough for us) and in g (so I don’t have to calculate each time): Mix & put in waffle maker.
- Checking math on example of exponential growth of policy count(NOTE: I’m not exactly sure what was the claim I probably should go back and check, it was kind of part of conversation and it’s bad form to nitpick what someone says without bothering to check what exactly they said; this is a draft mostly for my self so I’m not doing that but ifContinue reading “Checking math on example of exponential growth of policy count”
- Fire alarms appear to have good cost / benefit ratioI watched recent video about fire alarms and I started wondering if installing a fire alarm makes sense. I think they’re required in USA but they’re not required over here. So I started to wonder if getting some makes sense. I don’t have time to do a detailed check (also I don’t think it makesContinue reading “Fire alarms appear to have good cost / benefit ratio”
- Quick note on shallow / deep understandingFor some time I wanted to write what I think about shallow pattern matching / deep understanding but things are busy. And it looks like I’ll be very busy for at least 2 months more. And I intended this blog as a place for unpolished drafts anyways so her e goes. I think depth ofContinue reading “Quick note on shallow / deep understanding”
- Shallow pattern matching 12 – gamingI got annoyed with computer gaming some time around 2013/2014 so I’m worried about quality of this example but.. Well I sometimes still play Civ 2. Side note: I find this very suspicious that all examples I think off are something that’s not a very good example; mostly because I don’t actually feel like IContinue reading “Shallow pattern matching 12 – gaming”
- Shallow Pattern Matching 11 – my current thinking about thisMy current thinking on what is a shallow pattern vs deep one is something like “how robust is it”. Cooking: following recipe very exactly -> learning techniques -> understanding gears generating techniques (an LW post I remember about this; I might want to reread). When you follow recipe to the letter and you miss orContinue reading “Shallow Pattern Matching 11 – my current thinking about this”
- Note about greedy doctor problemYesterday I was thinking about the greedy doctor problem / ELK. I thought how are we dealing with this situation with real doctors? So first thing apparently not very well. Since more medicine on margin doesn’t result in better health (finding citation would be good here). Second when we can easily check things work wellContinue reading “Note about greedy doctor problem”
- People totally not talking about regime change in RussiaThis sound’s really weird to me: Sergei Markov, former member of Russia’s lower house the Duma and former member of President Putin’s party United Russia, has been speaking to the BBC Radio 4’s Today programme.Asked whether it is wrong for a country to pursue regime change in a sovereign independent country, such as Ukraine, heContinue reading “People totally not talking about regime change in Russia”
- Cost Disease Post IdeaI had a thought about cost disease. I was thinking about The Debtors’ Revolt post from Compass Rose blog. Specifically about the pharmacy example. Specifically about business model described there (it looks like the post was published about 2021-12-31 and I think I’ve read it close to that so my recollection of that…). So asContinue reading “Cost Disease Post Idea”
- Shallow pattern matching 10 – inspired by ELK prize resultsPart of notes I write as I think about shallow pattern matching. This is inspired by me reading ELK prize results post on LW. I don’t mean this as criticism of the work there. I don’t follow the work closely and I think that if any ideas there turn out to work then that’s great. OrContinue reading “Shallow pattern matching 10 – inspired by ELK prize results”
- Shallow pattern matching 9 – babiesPart of notes I write as I think about shallow pattern matching. I’ve been thinking a bit more about how babies learn language / does it resemble how language models learn. There are some significant differences – generally babies get reward for different things than trying to complete sentences in away they heard others speak. AndContinue reading “Shallow pattern matching 9 – babies”
- Shallow pattern matching 8 – Gwern on improving language modelsPart of notes I write as I think about shallow pattern matching. In the previous post I mentioned I wanted to add a note about something Gwern wrote about language models getting better. I found the thing I was thinking of it’s quite long section on The Scaling Hypothesis page so I’ll cut the quote asContinue reading “Shallow pattern matching 8 – Gwern on improving language models”
- Shallow pattern matching 7 – two of examples from PutanumonitPart of notes I write as I think about shallow pattern matching. Two more things catched my attention in Artificial Wordcells: 1) Steps in which a baby learns language (note: I didn’t check if it’s accurate). It doesn’t feel like good example of shallow pattern matching -> deep pattern matching transition. But it might beContinue reading “Shallow pattern matching 7 – two of examples from Putanumonit”
- Shallow pattern matching 6 – no neuron labeled tomatoPart of notes I write as I think about shallow pattern matching. I’m reading Artificial Wordcells on Putanumonit and I had some thoughts about shallow / deep patterns: not a single neuron in our brain is labeled “tomato detector” or even “redness detector” or even “vision” Is it true? If so then that looks likeContinue reading “Shallow pattern matching 6 – no neuron labeled tomato”
- Shallow pattern matching 5 – d&d examplePart of notes I write as I think about shallow pattern matching. Another example: my D&D. I kind sucked in playing D&D (3.5). I did read a bunch of char op guides, even make some builds which had pretty good stats. But in practice I sucked at playing those characters. Part of it was I thinkContinue reading “Shallow pattern matching 5 – d&d example”
- Shallow pattern matching 4 – astronomy examplePart of notes I write as I think about shallow pattern matching. The first example of shallow pattern-matching vs deeper understanding I thought of was from astronomy (I don’t know that much about astronomy so I keep looking for better examples). What I wrote down: Something like Feynman example of Mayan(?) astronomers finding a pattern inContinue reading “Shallow pattern matching 4 – astronomy example”
- Shallow Pattern Matching 3 – More from LWA new post came out on Less Wrong in the meantime: Ngo and Yudkovsky on scientific reasoning and pivotal acts. They talk quite a bit about shallow thoughts. I’m still reading through but I found this snippet interesting: … it continues to seem to me that science-and-engineering on the level of “invent nanotech” still seemsContinue reading “Shallow Pattern Matching 3 – More from LW”
- Shallow Pattern Matching 2 – Questions going inPart of notes I write as I think about shallow pattern matching. When I started writing about shallow pattern matching questions I wrote down were: What is a shallow pattern-memoriser? How it’s different from other things? How to recognize one (probably let’s ignore things pretending to be shallow pattern-memorizers for now)? Why do people thinkContinue reading “Shallow Pattern Matching 2 – Questions going in”
- Shallow pattern matching 1 – IntroI want to understand what people mena when they write that something is doing just shallow pattern matching. For example here: A first disagreement is related to Eliezer’s characterisation of GPT-3 as a shallow pattern-memoriser. And then they talk as if they have shard understanding what is shallow pattern matching and that it’s what GTP-3Continue reading “Shallow pattern matching 1 – Intro”
- Note on “Trauma”A friend recently shared a post from https://musingsandroughdrafts.wordpress.com/ on Facebook and I’m reading through the blog. Post My current high-level strategic picture of the world reminded me about something. Specifically the “Clearing “Trauma”??” section. The thing is that I’m less optimistic than some people about “post scarcity” thing. I think that even with what would lookContinue reading “Note on “Trauma””
- Note on putting together Ikea Leirsund Bed BaseThis week I was (again) putting together a Leirsund bed base and I made a mistake. It’s the same mistake I made last time I was putting one together so I thought I should make a note about it. When you’re putting together the frame don’t tighten screws all the way. Leave yourself a fewContinue reading “Note on putting together Ikea Leirsund Bed Base”
- Thinking About MovingI’m thinking about moving. Two top locations I’m thinking about are: New York and San Francisco Bay Area. A bunch of factors I consider important: Weather. Bay Area has much better weather. But NYC has much better weather than the place I’m right now. Things to do. NYC was much more interesting when I wasContinue reading “Thinking About Moving”
- Taking care of children is hardI just want to write down what’s the hardest thing in taking care of my baby. We need to take care of her all the time. 24 hours a day, 7 days a … you know all the time. Not actually all the time. Right now she’s asleep and I can write a quick blogContinue reading “Taking care of children is hard”
- Day care bitesAfter about two months of mostly skipping daycare (she was teething and mildly sick most of that time) my 1 year old went beck to daycare. I was very happy about that. But then on the first day she came back with a nasty bite mark. It came as an unpleasant surprise because staff didn’tContinue reading “Day care bites”
-
Subscribe
Subscribed
Already have a WordPress.com account? Log in now.
cashoyat 