dimava

6 ms·

1.
by dimava 1mo ago
There still may be Opus->Fable type bump when they finish training larger models. Remember, Grok 4.6 is still Opus-sized and they have larger models in training.
2.
by dimava 2mo ago
"Claude did implement multithreading in JS but we don't want to mantain a fork of WebKit" - jared apparently
3.
I Think AGI Just Happened(danielmiessler.com)
4 points by dimava 2mo ago|4 comments
4.
by dimava 3mo ago
Just @AGENTS.md should be enough, as @s is CLAUDE.md are inlined (and !`ls` are executed)
5.
by dimava 3mo ago
The ones who accept rewrites are expected to support it indefinitely. That's why Jared didn't consider using it for Bun even if it's undeniably better, they don't have human resources to support it
6.
by dimava 3mo ago
1. Owning your own company stock is not having money, it's merely ability to control the company. Elon tries to be BDFL of his companies, so selling the stock is not really an option. 2. Stock valuation and money are not the same thing
7.
by dimava 4mo ago
Since I know russian well, here's a proper translation for y'all FUNC FACT (N); NAMES: P; (* variable names *) 1 -> P; FOR I FROM 1 TO N :: P * I -> P DONE
8.
by dimava 4mo ago
If it's purely bot traffic, then Anubis could help You could have seen it on some websites already https://anubis.techaro.lol/
9.
by dimava 6mo ago
The amount of compute you get is not documented anywhere. Cursor docs do say "You get $70 of api priced tokens on $60 tier" Anthropic limits were 8.33x with free cached tokens, but those numbers themselves were a leak rather the
10.
by dimava 6mo ago
Cursor may be losing money only on $200 sub people who do over $200 of usage (it grants $400) Everyone else pays them at API prices
11.
by dimava 6mo ago
And then you have a const CatDog = { bark(){}, meow(){} } and const TreeCat = { bark: "oak", meow(){} } and your code stops working To make types discriminatable you need either type A = { bark: fn, meow?:
12.
by dimava 7mo ago
Just to reiterate ANTHROPIC IS GIVING EVERY DECENTLY LARGE MAINTAINER $1000 WORTH OF INFERENCE (~x8 that in API prices) They likely made a marketing budget for this of $1M or so Other OSS stuff like Copilot or JetBrains costs to providers m
13.
by dimava 7mo ago
Also check the https://botblock.ai/ , AI extension to detect AI replies on twitter
14.
by dimava 8mo ago
Open source models costs are determined only by electricity usage, as anyone can rent a GPU qnd host them Closed source models cost x10 more just because they can A simple example is Claude Opus, which costs ~1/10 if not less in Claude
15.
by dimava 9mo ago
Tokens will cost same on Mac and on API because electricity is not free And you can only generate like $20 of tokens a month Cloud tokens made on TPU will always be cheaper and waaay faster then anything you can make at home
16.
by dimava 9mo ago
When other models would grep, then read results, then use search, then read results, then read 100 lines from a file, then read results, Composer 1 is trained to grep AND search AND read in one round trip It may read 15 files, and then make
17.
by dimava 9mo ago
Just ask LLM to write one on top of OpenRouter, AI SDK and Bun To take your .md input file and save outputs as md files (or whatever you need) Take https://github.com/T3-Content/auto-draftify as example
18.
by dimava 9mo ago
$30 in API pricing > I was running this against my $20/month ChatGPT Plus account
19.
by dimava 11mo ago
refined title: ArXiv CS requires peer review for surveys amid flood of AI-written ones - nothing happened to preprints - "summarization" articles always required it, they are just pointing at it out loud
20.
by dimava 1y ago
DeepSeek on GPUs is like 5x cheaper then GPT And TPUs are like 5x cheaper then GPUs, per token Inference is very much profitable
21.
by dimava 1y ago
Except 12:01 is in 24-hour clock which doesn't have 12:00 problem in the first place
22.
by dimava 1y ago
Can you make a variant for relative passing time? You probably barely remember anything up to around 10, and then each doubling of age adds one logarithmical unit So 10 is 1, 20 is 2, 40 is 3 and 80 is 4 (or maybe 0, 1 and 2?) 20 is alrea
23.
by dimava 1y ago
Aka VSCode DevContainer? Could work I think (be wary of sending .env to the web though)
24.
by dimava 1y ago
Edit: it got fixed, thanks to the author I think with majority of TypeScript projects using Prettier, 2 is more likely to be the default[0] The linked page literally says to ignore it [1] > STOP READING IMMEDIATELY THIS PAGE PROBABLY DO
25.
by dimava 1y ago
They had tools to prevent this, but the tools were still in development They have since unrolled prod/dev db split and dB snapshots And restored that user's db from backup ofc
26.
by dimava 1y ago
There's an ESLint rule with autofixes for that (annotating type imports) One for extension should exist somewhere too
27.
by dimava 1y ago
Specifically for Linus Torvalds, the author or Linux He has a very distinctive style and large amount of training data from all the reviews and emails he made while collaborating on Linux And as he manages a huge project that's in deve
28.
by dimava 1y ago
I'm 50% sure adding a "Install offline app" button that installs PWA would be enough to make them believe otherwise Add a "Check for updates" button that redirects to chamgelog page in browser for extra points
29.
by dimava 1y ago
You should make your app a PWA [1] PWAs are basically "installable" pages which open in their own browser window and generally can work offline [1] https://developer.mozilla.org/en-US/docs/Web/Progre
30.
by dimava 1y ago
Habr is a great example why you should force your site to be 'out of politics' if you want senseful discussions Otherwise half of threads will be about Nazis by Godwin's law [0] Like this one ¯\_(ツ)_/¯ [0] https:&#x2F

More ›