nacs

11 ms·

121.
by nacs 3y ago
So 3 backordered products launching in the next 2 months? They're going to have a busy Christmas holiday..
122.
by nacs 3y ago
Not sure how the iPad fits into that. It’s pretty heavily developed and they’ve even spun off a dedicated iPadOS to take advantage of the additional tablet size/hardware features.
123.
by nacs 3y ago
The CLI will also prompt you to sign into huggingface in a browser and accept the model terms before it downloads.
124.
by nacs 3y ago
Outlines looks like an actively maintained alternative to Guidance. Do you have any plans to support llama.cpp?
125.
by nacs 3y ago
I think you mean September 2021 because that's what it has been on for a long time (and that's what mine is currently saying).
126.
by nacs 3y ago
Cloud bandwidth is definitely not cheap. If anything, it's where they rip you off the most. You can get "baremetal"/dedicated servers from places like Hetzner and OVH that give unmetered gigabit connections for like $50.
127.
by nacs 3y ago
The hardware may have cost more in 2021 when this article was written.
128.
by nacs 3y ago
1) That's not Dreams, that's original Procreate's animation tools 2) The video is cutting out the drawing of other frames after the first of each layer.
129.
by nacs 3y ago
I don't think that's Dreams.. That's just the basic animation flow in Procreate (frame by frame).
130.
by nacs 3y ago
Another great thing about it -- NO subscriptions or IAPs! They've done some huge updates over the last 5 years and not even charged for a new version.
131.
by nacs 3y ago
Yes it's inference only (and usually pretty slow at that).
132.
by nacs 3y ago
Yeah "on crack" usually means they're not right in the head. In this context though, OP is referring to the drug where the Github engineers are being quick/energetic about fixing the issue because they're "boos
133.
by nacs 3y ago
Yes, 7B is perfectly usable on low-end hardware if you're using it for instruction tuning/chat. But for code completion in an IDE where it has to react as you type, every 100 millisecond delay in response time is noticable. Even w
134.
by nacs 3y ago
The number of users is almost irrelevant in the context of Nostr because many clients generate a new ID. These aren't email-verified/phone-verified users -- this is literally just a new key/ID being generated. Some web apps a
135.
by nacs 3y ago
Yes, AutoGPTQ supports this (8, 4, 3, and 2 bit quantization/"compression" of weights + inference). GPTQ has also been merged into Transformers library recently ( https://huggingface.co/blog/gptq-integrat
136.
by nacs 3y ago
How is Facebook behind? They've released the most powerful open source LLM models so far (Llama 1 + 2) and are a serious threat to the Openai closed-source monopoly.
137.
by nacs 3y ago
I have Dark Reader and had no issues. There's a global setting in Dark Reader to "Detect dark theme" pages -- when that is enabled, this game is automatically detected as using a dark theme so the Dark Reader plugin is disabl
138.
by nacs 3y ago
Absolutely. For QLORA / 4bit / GPTQ finetuning, you can train a 7B easily on an RTX 3060 (12GB VRAM). If you have a 24GB VRAM GPU like a RTX 3090/4090, you can Qlora finetune a 13B or even a 30B model (in a few hours).
139.
by nacs 3y ago
A ton of people have moved to Threads from Meta which is the most direct competition. Then there is of course Mastodon and the upcoming Bluesky.
140.
by nacs 3y ago
Instagram (the web version at least) is built in React.
141.
by nacs 3y ago
Scale isn't an issue for Meta. While Bluesky has been having problems with their (invite-only) spike in signups from Twitter refugees, Threads has been chugging along with no issue. Also, Zuckerberg made a post saying they got somethin
142.
by nacs 3y ago
Meta already has a massive follower graph from using Instagram for signin. When I logged in to Threads, it added my entire follower list of hundreds from Instagram. Also, Instagram already links Instagram accounts with Facebook accounts for
143.
by nacs 3y ago
Meta mentioned ActivityPub/Fediverse support but I'll believe it when I see it. Maybe it's the cynic in me but I think Meta will just implement a tiny portion of the spec that benefits them (importing posts from Mastodon?) an
144.
by nacs 3y ago
At the rate that Musk is destroying Twitter, all Meta has to do is drag the court case out a year or 2 and Twitter won't exist to see the end of it.
145.
by nacs 3y ago
GPTQ support would be amazing (AutoGPTQ is an easy way to integrate GPTQ support - it's basically just importing autogptq and switching out 1 line in the model loading code).
146.
by nacs 3y ago
They have a section about it in the README: https://github.com/bentoml/OpenLLM#-telemetry
147.
by nacs 3y ago
Online version using the emulator is here: https://grimacesbirthday.com/
148.
by nacs 3y ago
Yes but have you seen their pricing? 20 million a year will be the API cost for the top (freemium) iOS app. https://arstechnica.com/gadgets/2023/05/reddits-api-pricing-... They're basically using it to k
149.
by nacs 3y ago
Sounds to me like these mega companies are expecting favorable regulation changes to give them more “freedom” (to reject valid claims). Either that or they’re wanting a tax-payer funded bailout.
150.
by nacs 3y ago
This allows outbound network access, allows program execution (within the container) and more. You might want to restrict some of these things before Amazon shuts your account down for abuse requests. You're basically handing everyone

More ›