Tiberium

28 ms·

571.
by Tiberium 2y ago
Maybe I'm missing something, but isn't this also directly applicable to services like Reddit, especially after their API changes?
572.
by Tiberium 2y ago
There are hundreds of NSFW finetuned models on HuggingFace and whole ERP communities built around them. So there are models that can do precisely that :) And yeah, all big models can write those things too, the best currently is Claude 3 Op
573.
by Tiberium 2y ago
It's been used in this meaning on 4chan for at least half a year, maybe more.
574.
by Tiberium 2y ago
The problem is that they sold the game in a lot of countries where you couldn't register a PSN account, including a few countries in the EU. In specific cases like Ukraine you have to own the console to be able to register a PSN accoun
575.
by Tiberium 2y ago
Am I missing something, or is there no repo to try it out?
576.
by Tiberium 2y ago
That's just not true, GPT-4 is pretty bad for languages that are relatively niche, e.g. for Nim, Zig, Crystal as some random examples (and those do have some popularity at least).
577.
by Tiberium 2y ago
It does run recent apps as long as they don't use the new Wayland protocols. I decided against that link because there are a lot of files before the actual README.
578.
12to11 – run Wayland applications on an X server(sourceforge.net)
93 points by Tiberium 2y ago|96 comments
579.
by Tiberium 3y ago
My current title does state that it is just a proposal, but fair enough, I'll update the title to the original one.
580.
by Tiberium 3y ago
I'm sorry if the title might seem misleading, but I've based it on the actual PR [1], which says "GitHub does not allow any projects that are designed for, encourage, promote, support, or suggest in any way the use of synthet
581.
A policy proposal on our approach to deepfake tools and responsible AI(github.blog)
5 points by Tiberium 3y ago|5 comments
582.
OpenAI Batch API: bulk requests with results in 24 hours, 50% discount(twitter.com)
2 points by Tiberium 3y ago|0 comments
583.
by Tiberium 3y ago
Not .safetensors though
584.
by Tiberium 3y ago
FYI, it's quite easy to change "roles" that LLMs take (via the system prompt) so that they'll appear much more like a human in text. That's how people roleplay with models such as GPT-3.5/4 and Claude 2/3.
585.
Using LLMs to reverse JavaScript variable name minification (2023)(thejunkland.com)
15 points by Tiberium 3y ago|0 comments
586.
by Tiberium 3y ago
Big LLMs like GPT-4 (and even GPT 3.5 Turbo) can be directly used to beautify obfuscated/minified JS, see e.g. https://thejunkland.com/blog/using-llms-to-reverse-javascrip... and https://news.ycombinato
587.
by Tiberium 3y ago
Interesting to note that there's an alternative specifically for running Android games on Windows - Google Play Games [1] which is Google's official offering, and it does not rely on WSA tech so that it works on Windows 10 as well
588.
Supermium – Chromium fork for Win 2003/XP and newer(win32subsystem.live)
221 points by Tiberium 3y ago|192 comments
589.
by Tiberium 3y ago
For SDXL this meant that there were almost no NSFW (porn and similar) images included in the dataset, so the community had to fine-tune the model themselves to make it generate those.
590.
by Tiberium 3y ago
Actually, there have been new model releases after LLaMA 2. For example, for small models Mistral 7B is simply unbeatable, with a lot of good fine-tunes available for it. Usually people compare models with all the different benchmarks, but
591.
by Tiberium 3y ago
This is a legit issue, although they claimed to have mostly fixed it: https://twitter.com/sama/status/1754172149378810118 (by Sam Altman) > gpt-4 had a slow start on its new year's resolutions but should n
592.
by Tiberium 3y ago
I think the issue was exclusive to ChatGPT (a web frontend for their models), issues with ChatGPT don't usually affect the API.
593.
by Tiberium 3y ago
Original: If anyone's curious about the (probable) non-humorous explanation: I believe this is because they set the frequency/presence penalty too high for the requests made by ChatGPT to the backend models. If you try to raise th
594.
by Tiberium 3y ago
Is there a repo, especially for training? I'd like to see how SigLIP performs on a dataset of only anime images.
595.
by Tiberium 3y ago
I'm not the only one ;)
596.
by Tiberium 3y ago
It was directly mentioned in the post: > I attempted to withdraw the bitcoin balance into my own wallet, and hit the limit that Mt Gox has, preventing you from withdrawing more than $1000 USD worth of bitcoins (at the current market valu
597.
by Tiberium 3y ago
For some more context see my submission from a few days ago - https://news.ycombinator.com/item?id=39175611 , although it's admittedly not mistral-medium, but llama2 trained on the same dataset (since the outputs do mat
598.
by Tiberium 3y ago
There's no conclusive evidence that this is the exact mistral-medium leak, however a lot of people tried it and confirmed that outputs are very close to the ones that mistral-medium gives through the API, sometimes an exact match. You
599.
Miqu 70B – possible leak of the mistral-medium LLM(huggingface.co)
16 points by Tiberium 3y ago|1 comments
600.
by Tiberium 3y ago
OpenRouter is generally a good option (already mentioned), the best part is that you have a unified API for all LLMs, and the pricing is the same as with the providers themselves. Although for OpenAI/Anthropic models they were forced (

More ›