nodja

14 ms·

121.
by nodja 3y ago
> I'd love to read a less technical writeup explaining the challenges faced and why it took so long to figure out spelling. Scrolling through the paper is a bit overwhelming and it goes beyond my current understanding of the topic.
122.
by nodja 3y ago
> The goal will be to run them out of money. They're already making $30k/month on patreon and you can easily foresee that this legal action will cause a call to arms for even greater funding, if that was indeed their goal they
123.
by nodja 3y ago
This is also true for 2D diffusion models[1]. I suppose they need to understand how 3d works for stuff like lighting/shadows/object occlusion, etc. [1] https://dreamfusion3d.github.io/
124.
by nodja 3y ago
People have been saying that about VR headsets in general, but when I got mine and beat HL:Alyx--which was admittedly pretty fun--the only games I came back to were beat saber and that bow and arrow game in "the lab", I gave a try
125.
by nodja 3y ago
If your web app serves dynamic routes (i.e. client only) this helps with SEO because those routes are now directly visible through most crawlers.
126.
by nodja 3y ago
TL;DR: Android only needs winusb to work, before Windows 8 (2012) windows only used Vendor ID and Product ID to decide what driver to load, so you had to install a new .inf driver file for every new device. Newer versions ask the device for
127.
by nodja 3y ago
https://github.com/bluesky-social/social-app https://github.com/bluesky-social/atproto
128.
by nodja 3y ago
> ultimately dlss/frame gen is a hack that no one on a high end card would ever use. The irony of those technologies is that they shine best on high end cards. If you're on a low end card and say pushing the limit of a game and
129.
by nodja 3y ago
They're better than most consumers x86 CPUs but worse than using a GPU. Where they shine is when the ML model can't fit the GPU's VRAM since you have better options for ram size with macs.
130.
by nodja 3y ago
If anyone like me was curious as to where the Portuguese got the sweet oranges from: It was from China through the newly found maritime trade routes.
131.
by nodja 3y ago
It's pretty much the same architecture since GPT2, just a bunch of self-attention transformer blocks. The reason these have been better is because we have more GPU, more data, and have scaled the attention calculations to be linear ins
132.
by nodja 3y ago
The owner of the space.
133.
by nodja 3y ago
Yes, an A100 costs $4.13/hour https://huggingface.co/pricing#spaces
134.
by nodja 3y ago
This is correct, also the steam keys are free to generate. I.e. when you buy a steam key from sites like greenmangaming, valve/steam gets $0. Steam could easily not provide the keys service or charge their cut as keys are claimed, but
135.
by nodja 3y ago
oh wow this model is kinda amazing, it passes my "creative" tests that only chatgpt 3.5 did decently well with, I've recently been disillusioned that open source has been moving the wrong way due to the focus on benchmarks, b
136.
by nodja 3y ago
The table is from benchmarks ran on a T4 GPU, that's roughly a GTX 1080 levels of performance. More recent GPUs are vastly faster (5-10x).
137.
by nodja 3y ago
> Is there any reason to believe that the interaction of multiple agents (using the same model) will yield some emergent property that is beyond the capabilities of the agent model? If you write a short story it's often better to sp
138.
by nodja 3y ago
> Middle of video ads should be content creator controlled. They are, sort of. As a content creator you pick the position for the ads, but it doesn't mean they'll pop up there for everyone, there's only a chance they'
139.
by nodja 3y ago
I have a tiny channel that posts long (3-4 hours) gameplay videos and premium is 80% of my revenue. I do have mid-rolls and unskippable ads turned off, but in my experience those would only increase my ad revenue by 10-20%, i.e. premium wou
140.
by nodja 3y ago
> Though as it says in the article, Godot does lack proper support for consoles. None of the open source engines do, or ever will as publishing to consoles requires that you sign an NDA. If you're a godot dev and want to publish on
141.
by nodja 3y ago
https://github.com/abetlen/llama-cpp-python has a web server mode that replicates openai's API iirc and the readme shows it has docker builds already.
142.
by nodja 3y ago
ruDALL-E is not DALL-E, it's a replication attempt made on a Russian image-text dataset and is not affiliated with openai. The original DALL-E was trained on a private dataset and wouldn't have watermarks.
143.
Watching Neural Networks Learn [video](youtube.com)
5 points by nodja 3y ago|0 comments
144.
by nodja 3y ago
Here's my attempt at an explanation without jargon, you can just read the last paragraph, the first 4 are just context. These image models are trained on 1000 steps of noise, where at 0 no noise is added to the training image and at 10
145.
by nodja 3y ago
Note that Paella is a bit old in image model terms (Nov 2022) and modern stable diffusion tools have access to optimized workflows. My 3060 can generate a 256x256 8 step image in 0.5 seconds, no A100 needed. A 3090 is double the performance
146.
by nodja 3y ago
Yes, but premium revenue is decided by watch time, long videos get more revenue usually, with videos at around 90min length being 50/50 payout from ads and watch time. Assuming ads only at the beginning and end of videos and the video
147.
by nodja 4y ago
Naive question: would this also mean you can also get more memory for cheaper at the cost of space? i.e. instead of using expensive high density 2GB chips you could use 2x1GB or something?
148.
by nodja 4y ago
Similar experience, mine would turn background noises when no one's talking into random japanese words repeated. I was using the large model. I ended up fixing it by using the medium.en model and setting condition_on_previous_text to
149.
by nodja 4y ago
I haven't used tiktok in a month, but right before I left lots of creators were complaining that their videos weren't getting as much views anymore, specially some that mentioned political stuff and the obvious thirst traps. That
150.
by nodja 4y ago
Not one with 360 view if that's what you're thinking. See it as one of those hard halloween masks, from the front they look like the shape of a person's face, but as soon as you look at it from the side/back you just see

More ›