b1t (@bitpeks) header
b1t (@bitpeks) profile picture

likes are bookmarks

Metaverse
Joined August 2021
10 must-know Japanese concepts that will improve your life:
1,026
20,820
3,084
70,388
Impressive, @karpathy's repo "minGPT" just reached 10k stars on Github! minGPT is a minimal PyTorch re-implementation of the OpenAI GPT training. "GPT is not a complicated model and this implementation is appropriately about 300 lines of code" 🛠️ Repo: github.com/karpathy/minGPT
21
349
15
2,124
281,854
20 life lessons every father must teach his son: THREAD
1,047
10,522
677
40,076
9,642,056
The OpenAI cookbook is the top trending repo on Github this month! If you haven’t checked it out, it’s an incredible resource for getting started on the API — maintained by our team (and others!) github.com/openai/openai-coo…
54
601
31
3,720
700,325
Side-by-side comparison: @OpenAI's ChatGPT vs. @AnthropicAI's Claude Each model is asked to compare itself to the machine from Stanisław Lem's "The Cyberiad" (1965) that can create any object whose name begins with "n":
28
129
32
842
240,871
🏺Ancient Greek vase types illustrated by Drawing Ancient History / Instagram. instagram.com/p/CnIEYvkLatE/…
16
665
27
4,166
236,615
ok stable diffusion is wild
21
21
3
227
24,010
As promised, sharing the Founding Artists 🎨 concept. Founding Artists is a feature and concept that has never been done in the NFT world.
95
72
23
424
108,779
1/14 🎲How does Bitcoin Core generate entropy for private keys? Mixing entropy from 4 main sources: - low-level processor instructions - entropy from the OS - dynamic events - static events 🧵Thread👇
12
105
25
331
What happened to art?
949
7,814
1,797
49,171
5,638,680
High resolution infinite zoom experiments with #StableDiffusion2. Managed to generate these hallucinations at 2048x2048 pixel resolution. #StableDiffusion #AIArt
24
113
11
770
114,965
The first rule of Twitter: don’t take it seriously. In fact, that’s the first rule of life too.
123
395
29
4,381
1,614,260
OpenAI Whisper (speech2text) is trained on 680K hours of audio. VALL-E is only trained on 60K hours. This means the potential of text2speech models is way under-tapped. OpenAI can just scale up VALL-E 10x on their private training set and nail it. It's such a low-hanging fruit.
DALL-E generates pixels from text. Now meet its cousin, VALL-E, that generates audio from text @MSFTResearch! VALL-E’s resemblance to DALL-E v1 and Parti @GoogleAI is striking. Image and audio are both continuous signals, but they can be quantized into discrete tokens. 1/🧵
12
111
6
706
157,448
How did a 26 year old Japanese man become a billionaire? He created Anycolor, a VTuber company! VTubers are virtual YouTubers & they are HUGE in Asia Right now, VTubers are just people that "wear" a digital skin & often use a voice changer Soon, AI will operate VTubers
15
32
11
186
56,389
thinking about the egyptian hieroglyphic "pot with legs"
118
1,828
213
26,662
1,284,545
BitTorrent’s killer apps: Sharing music, videos, software, ebooks, scientific papers, and now … weight parameters for large neural network models trained on gigantic datasets.
6
27
2
373
88,188
GPT presents: the fake it till you make it, 'make things up as you go'-driven development methodology TL;DR: just pretend some code repo exists, interact with it, and GPT will happily fill in the blanks. To start, let's clone an imaginary vet app repo and start its API server:
3
54
13
264
76,889
1. As we approach the end of Season 1 of The Memes by @punk6529, I want to share my thoughts. A 🧵...
26
75
30
279
96,834
Why do elite colleges pump students into the same five industries: law, medicine, tech, investing, and management consulting? For answers, let’s turn to the philosopher René Girard.
292
3,062
382
17,171
5,370,793
How to make new habits stick. The Two-Day Rule:
202
7,816
482
62,184
10,246,637