Tiberium

35 ms·

631.
by Tiberium 3y ago
It's funnily the opposite - ChatGPT (the web frontend)-only, and only for some users for now.
632.
by Tiberium 3y ago
forsen forsen I'm your number one fan
633.
by Tiberium 3y ago
Because this website bypasses paywalls and loginwalls - it has accounts for a lot of the social media websites, so you don't see a generic "Login to see this page" but the actual contents.
634.
by Tiberium 3y ago
I'm not sure if you're joking or not, because OpenAI actually used LSAT as one of the examples where GPT-4 is much better - https://openai.com/research/gpt-4 https://files.catbox.moe/l5n5gd.jp
635.
by Tiberium 3y ago
It's miles ahead of 3.5, the difference is really really big. Mainly because GPT-4 is just so much better at "understanding" and logic, it follows everything you ask for or say very, very closely.
636.
by Tiberium 3y ago
It doesn't get synced anywhere - "cross-browser" in this case means that the extension works in multiple different browsers, but you still have to manually export/import sessions between them.
637.
by Tiberium 3y ago
The link you provided is completely different - it's free disposable real Linux servers (more specifically it's probably just containers) that you can easily ssh into, or even connect from the browser.
638.
by Tiberium 3y ago
There are no lower-level prompts than the ones described in the link. If you're asking about how the model sees the context - the messages are formatted using ChatML [1] which is a format with tokens to denote messages with their roles
639.
by Tiberium 3y ago
If anyone is interested in the contents, it's anti-Semitism (related to the current world events) added in the Ukrainian translation https://github.com/canonical/ubuntu-desktop-provision/commit...
640.
by Tiberium 3y ago
Maybe this is why :) https://i.imgur.com/odYgMpd.png
641.
by Tiberium 3y ago
Sorry, but your reply is very misleading - you're linking the blog post of the release of the original DALL-E through OpenAI's own website (which was in 2022 - centuries ago by the current AI trends). The one integrated in Paint i
642.
by Tiberium 3y ago
To join the other commenters, in Nim you use backticks to allow using Nim keywords - this is important for e.g. nice deserialization where some JSON object has a field like `type` which is a keyword in Nim. https://nim-lang.org&#
643.
by Tiberium 3y ago
Yes, you can - you must fill the "Request" form for Anthropic models, but it gets instantly approved (it doesn't look like it gets checked at all) and then you can just click "Edit" and enable access to Claude (whic
644.
by Tiberium 3y ago
No, it's not because of that, it's because you need to request access to the models per each region. So to get access to Claude 2 in another region you have to change the region in the AWS Console when you have Bedrock open, and t
645.
by Tiberium 3y ago
There is pricing for Claude 2 there, because "Claude" in the "Anthropic" section refers to both Claude 2 and Claude 1.3 (don't ask me why). See https://i.imgur.com/ev5O22G.png . And yeah, it's t
646.
by Tiberium 3y ago
If you're talking inference speed, it's probably just because simple words are a single token, but complex words or code are much more tokens so it'll output them "slower" (while in reality each token will take the
647.
by Tiberium 3y ago
Godot packs the whole engine runtime when packing the game no matter the platform, but in this case it's also probably because the internet connection on the server is slow - https://lfgslots.com/classicvideopoker/
648.
by Tiberium 3y ago
That Google incident sounds interesting, do you have links to the technical explanations? From a quick search all I could find are news reports.
649.
by Tiberium 3y ago
I don't know Next.js, but was that feature introduced later than 2021? I think both GPT-3.5 Turbo and GPT-4 largely share their datasets, and it has the data cutoff at roughly September 2021 (with a small amount of newer knowledge). Th
650.
by Tiberium 3y ago
From the article: > Plus and Enterprise users will get to experience voice and images in the next two weeks. We’re excited to roll out these capabilities to other groups of users, including developers, soon after.
651.
by Tiberium 3y ago
I'm sorry but I'll have to disagree with you - Claude 2 is worse than GPT-4 in almost all tasks I've tried it with, the two things that it's better at than GPT-4 are the context window and the dataset being from early 20
652.
by Tiberium 3y ago
Looks like it's just creating a VPN by using GitHub Actions, not sure if that's even allowed usage: >HOLD UP You need to fork or generate this repo first, so you have your own copy. Then open an issue and you will be guided thr
653.
by Tiberium 3y ago
That's been widely known for quite some time, there's no engine, it just happens "magically" when a LLM gets big enough (and they sure have a lot of that stuff in their training datasets). GPT-4 is generally very good at
654.
by Tiberium 3y ago
A bit off-topic, but regarding "Chinese ‘internet-of-things’ search engine FOFA indexes favicons": A lot of other internet host search engines do that, including the well-known Shodan and the newer (but, IMO better) Censys.
655.
by Tiberium 3y ago
There's actually a somewhat curious detail about this - Unity doesn't allow you to remove the Unity logo when using the free (Personal IIRC) license for your game. This has led to people sometimes associating Unity with bad games,
656.
by Tiberium 3y ago
It does give more space in the sense that the model generates more tokens of steps/etc that it can then base its actual answer on, rather than being forced into generating the answer right away.
657.
by Tiberium 3y ago
Just a random question, but would most of your service not become obsolete when OpenAI releases their advertised image understanding capabilities for GPT-4? EDIT: And what about privacy? What if some of my screenshots potentially contain NS
658.
by Tiberium 3y ago
You're probably aware, but OpenAI has different speeds for different kinds of customers. The most prioritized ones are accounts that have gpt-4-32k. An easy to see this difference is comparing a trial account's Turbo speed to a pa
659.
by Tiberium 3y ago
This argument is not really valid anymore, especially in a paper that talks about LLMs.
660.
by Tiberium 3y ago
Found one problem - when I get a really big json (e.g. https://a.4cdn.org/g/catalog.json , sorry for a 4chan example), I can't pass it into the code block because I get "{ "error": "limits.read.

More ›