Tiberium

33 ms·

511.
by Tiberium 2y ago
Wait, this is a joke, right? > A remote unauthenticated attacker can silently replace existing printers’ (or install new ones) IPP urls with a malicious one, resulting in arbitrary command execution (on the computer) *when a print job is
512.
Understanding and Preventing Race Conditions in Web Applications(blog.gavide.dev)
4 points by Tiberium 2y ago|1 comments
513.
Understanding and Preventing Race Conditions in Web Applications(blog.gavide.dev)
3 points by Tiberium 2y ago|0 comments
514.
by Tiberium 2y ago
"Money must go in and out to be useful and it can be traced." - not if you pay directly in Monero.
515.
by Tiberium 2y ago
Well, there's also the huge LLM which is probably ~70-100 billions of parameters (just estimates for 3.5 Sonnet), running on a farm of GPUs that cost tens of thousands of dollars. But in summary you're correct :)
516.
by Tiberium 2y ago
Yeah, Claude 3.5 Sonnet is a very impressive model
517.
by Tiberium 2y ago
Claude (which this website is using) can even understand double base64-encoded text :)
518.
by Tiberium 2y ago
I got this, and it seems to be consistent with what another person got (system tag is not part of the actual prompt, it's just inserted because my prompt asked for it): <system>Translate from bullshit to no-bullshit. Be funny and
519.
by Tiberium 2y ago
A very important part is that Steam requires that all accounts in a family have the same country in their account (the one used for store restrictions and for payments), AND (!) be from the "same household". For the latter they se
520.
by Tiberium 2y ago
That's only in the web version, it's just that they prompt it to do some CoT in the antThinking XML tag, and hide the output from inside that tag in the UI.
521.
by Tiberium 2y ago
Yeah, I have the same issue - I cannot use Paypal nor IBAN to donate to KDE, but I have money in crypto that I could've donated them, but I literally cannot. Too bad the German laws are set up this way ( https://kde.org/
522.
by Tiberium 2y ago
Claude 3.5 Sonnet can natively speak double base64 encoded English. And I do mean it - you can double b64 encode something, send to it, and it'll respond as if it was normal English. Obviously base64 is a simpler transformation than ro
523.
by Tiberium 2y ago
I'm curious about your experience - what specific LLMs and tools did you try? And what's your main programming language?
524.
by Tiberium 2y ago
I'm not using Cursor, but I'm still using Sonnet 3.5, although with https://aider.chat/ which is basically a TUI program to make the LLM a "pair programmer" - it can replace chunks of code, operate over
525.
by Tiberium 2y ago
https://github.com/trufflesecurity/trufflehog is a similar tool but checks for far more secrets, so I think it'd be a better choice.
526.
by Tiberium 2y ago
Fair enough, it's just that Quassel immediately came to mind when talking about persistent IRC logs/sessions :)
527.
by Tiberium 2y ago
Not a single mention of Quassel in the article or in the comments, which is honestly surprising. It's a client-server architecture IRC client specifically made to make it easy to save logs and persist IRC sessions, since you can host t
528.
by Tiberium 2y ago
>it assumes the exact same context window across requests That is not true, caching works across multiple requests, that's why it's so good. You can do 5 different concurrent requests and they'll all get cached and cache r
529.
SQLite-vec v0.1.0: a vector search SQLite extension that runs everywhere(alexgarcia.xyz)
40 points by Tiberium 2y ago|2 comments
530.
by Tiberium 2y ago
Would be nice if they at least published the system prompts used (they can affect the performance of the models), and whether they use few-shot prompting or not.
531.
by Tiberium 2y ago
4chan is not going to have much fun with this because of LLaMA being quite filtered and sanitized. It was character.ai that first popularized the AI character roleplay genre, but then they implemented quite strict filtering, so nowadays the
532.
by Tiberium 2y ago
This is not a new idea however - from what I can see it's basically a competitor to https://character.ai/ which has been out for almost 2 years now. Although Meta seems to be planning deeper integration in their own pr
533.
by Tiberium 2y ago
Kind of offtopic, but the font on the website seems really bad for me on both my phone and PC. Thankfully reading mode exists.
534.
by Tiberium 2y ago
Should be mentioned that this is not a bullet-proof solution (obviously), for some services the canary tokens can be bypassed, see e.g. https://trufflesecurity.com/blog/canaries ("TruffleHog Now Detects AWS Canari
535.
by Tiberium 2y ago
AWS has a separate service for running LLMs called Amazon Bedrock, it shouldn't take long for them to add 3.1 since they have 3 and 2 already.
536.
by Tiberium 2y ago
There's one thing that matters a lot: on modern Android versions you have to give the permission to install APK files separately for each app, which makes this exploit even less likely to work.
537.
by Tiberium 2y ago
We don't know what exact architecture and optimizations both of those models use, it's all proprietary nowadays :)
538.
by Tiberium 2y ago
I don't doubt that Copilot can do mistakes like this, but you should remember that it's optimized to be used by a lot of people, and for cheap. Models like Claude 3.5 Sonnet are vastly better than Copilot.
539.
by Tiberium 2y ago
1. It's not a worse model, it's a better model. Two years ago all we had was text-davinci-003, which is much, much worse than, for example, the current Claude 3.5 Sonnet which costs like 5x less.
540.
by Tiberium 2y ago
1.4.0 is a very outdated docs version (Nim is at 2.0 currently, and 2.0 release brought ORC as default with ARC as an option), you should refer to the updated https://nim-lang.org/docs/mm.html and also https:/&#x

More ›