0xSasuke

@namubois nice, ordering now. know how long till shipped to the US?

@0xSasuke Yup you can chose the language on the shop in my bio :)

quelqu’un voulait une fresque d’harley personnalisée, avec un cadre gravé et le texte en vieux norrois… 😮‍💨 le résultat est magnifique!!

@hugochimaru how was it?

je l’ai commandé…

@karpathy Can you post a "tree" of your folder structure? Would be interesting to see. I do something similar but I just don't store the raw stuff.

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

@teortaxesTex You dislike people who care about AI safety?

Apparently you get bullied and rejected from Anthropic for being insufficiently woke on risks of open source AI. Experience with Anth HR interview. 1point3acres, translated from Chinese. People were asking sometimes why I dislike Dario Amodei. I hope this clarifies matters.

@kechogarcia Most of my code is written by AI

Tomorrow is the day %100 of code will be written by AI. Looking forward to not have to code again.

@SouthernYankee @JRLV_Inc @jackfriks whats the model

@JRLV_Inc @jackfriks Jack—I have that breville, have had it 10 years, it’s a TANK. Worth every penny.

what’s a material thing under $1000 you’ve bought that actually changed your life?

@neilcybart The culture there is terrible. Worked there before as a SWE.

This is a wacky way to run a company. Crazy spending followed by layoffs followed by crazy spending followed by layoffs. No way that leads to great company culture.

@xbxnxdxcxtx @xai recruiter posting this btw

@a__tomala That is not how it works lol. It’s 25% and that means you aren’t getting paid that

An early Anthropic perk is that you can 3x the amount of stock you get by donating it to charity. I know some people used this to accumulate a ton of stock into Donor Advised Funds. Once that becomes liquid, there will be a ton of EA-aligned donations

@Rizwan_Ali @Taran Wdym? 100m sold at 350? Is it gonna do a downrohndv

@Taran Just saw someone selling for 350B 100M size :/

Hearing anthropic secondaries are now 650b lol

@J3K420 @TylerAlterman I have that filter on from 7pm+ on every device

@TylerAlterman The night filters on a pc make it so noticeable that something is happening physically with the screens. Turning it on there's an instant "relief" feeling that I imagine is even better with no screen at all lol

I have bad news you already know. For the past 2 yrs I’ve been testing whether screens an hour before bed impact my sleep. They do. The effect is not small, it’s very noticeable, and even calming activity like baths or reading doesn’t make it go away. Screens are now banned in the hour before bed

@rohanpaul_ai Link to interview?

Dario Amodei's new interview: He was asked "If you were in the room with the President right now, what would you say to him?" He says “Disagreeing with the government is the most American thing in the world.” He looks visibly strained.

@shoe0nhead @Swalas420 @sbodrojan I find it largely because they’ve never experienced it. Visiting Taiwan and other East Asian countries exposed me to “nice things” that I wish we had in America.

idk I think it is imperative to treat your relative discomfort w strangers pissing in public as basically irrelevant compared to the bloodthirsty indignities forced on the unhoused and disenfranchised in America.

@lauriewired How did you figure out what you should improve on specifically? Did you just find "things" that made you go "oh... I do that?"

literally talk about anything, just fill up the time. I learned the technique from a professional twitch streamer years ago; it was their way of training themselves to not have dead air. It outshines any other possible practice method I’ve tried. I'm pretty introverted irl.

no one’s gonna believe me but becoming a good speaker is really easy just record yourself for 10 minutes every day, first thing in the morning. don’t send it to anyone, just force yourself to watch it later. you’ll notice every possible flaw you can imagine.

@CoachCloud_ @RapidRL what about via usb?

@RapidRL Nah fr tho I have the mv7+ and it’s great super easy setup and if you have an XLR interface 1000/10 and its affordable

Whats a good quality microphone looking for a new one had a at2020 and quadcast I need a better one any recommendations.

@luffyscenes do i watch one pace or one piece

So much about this scene… forever top 5 for me

@Pepox_D_Vill @ApothecaryAnime If i liked frieren would I like apothecary diaries?

@ApothecaryAnime What they have in common is that they are the best in the current anime industry.

@animeposts I saw the first 2, what’s the last one?

@MGLKO_777 @kayn3_ Yea I just finished S1 it’s 10/10 for me! I’ll wait for S2 to finish then.

@0xSasuke @kayn3_ Your choice Both the manga and anime are great U can read it after S2 finishes

Despite so many anime trying over the last 2 years, it took Frieren S2 to finally dethrone S1 on Mal. Ik this might be temporary due to premiere hype but still kinda crazy