dimmuborgir

5 ms·

1.
Interwhen: A Generalizable Framework for Verifiable Reasoning(arxiv.org)
2 points by dimmuborgir 6mo ago|0 comments
2.
by dimmuborgir 11mo ago
Nano Banana for me. After the initial wow phase it's meh now. Randomly refuses to adhere to the prompt. Randomly makes unexpected changes. Randomly triggers censorship filter. Randomly returns the image as is without making any changes
3.
by dimmuborgir 1y ago
From the paper: "A single run of the DGM on SWE-bench...takes about 2 weeks and incurs significant API costs." ($22,000)
4.
by dimmuborgir 2y ago
Yup, they are called GCCs (Global Capability Centers).
5.
The Premise Trap(world.hey.com)
3 points by dimmuborgir 2y ago|0 comments
6.
by dimmuborgir 2y ago
The result announcement blog post sounded too hypey and gave the impression that he changed his tone but for the last one year he has been consistently saying that LLM+"discreet program search" is the way to go. All the top scorin
7.
by dimmuborgir 3y ago
Dark mode finally!!! This was the only bummer for me all these years. Not anymore!
8.
Getting from Generative AI to Trustworthy AI: What LLMs Might Learn from Cyc(arxiv.org)
1 points by dimmuborgir 3y ago|0 comments
9.
From word models to world models(arxiv.org)
100 points by dimmuborgir 3y ago|113 comments
10.
by dimmuborgir 3y ago
We can extend this phenomenon to all of knowledge. You ask ChatGPT something and it gives you some very generic 10-point blogspam-esque answer. You immediately think ChatGPT is intelligent and that it has understood your particular question
11.
by dimmuborgir 4y ago
His book "The Algebraic Mind" goes into great detail about connectionism (neural networks), symbolic systems, limits of connectionism and proposals to integrate neural networks with symbolic systems (hybrid systems). The "dee
12.
by dimmuborgir 4y ago
I think her target audience is extremely broad. From teenagers to middle-aged people. Plus her latest album was marketed aggressively.
13.
by dimmuborgir 4y ago
Those models are not trained on short loops. They are trained on whole songs just like image generation models are trained on whole images. And yet they struggle to repeat sections, modulate to a different key, create bridges, intros and ou
14.
by dimmuborgir 4y ago
AI is bad at music also. Even the state of the art transformer models can't produce more than a few seconds of coherent melodic phrases.
15.
by dimmuborgir 4y ago
This one suffers from the same problem that previous audio generation methods had. It correctly mimics the piano timbre but there is no global structure in the generated melodic lines. There is some style imitation but no melodic coherence.
16.
by dimmuborgir 4y ago
Lots of weird artifacts which are very hard to fix. The argument that users can now generate professional grade art by bypassing artists entirely feels so strange. I have access to Dall-E. To generate images without artifacts, you have to d
17.
by dimmuborgir 5y ago
JSPatcher is similar to Max/MSP. Even the UI design is a blatant ripoff. There are already full blown online DAWs like BandLab, Soundation, Amped Studio etc.
18.
by dimmuborgir 6y ago
Previous discussion: https://news.ycombinator.com/item?id=22448933
19.
by dimmuborgir 6y ago
Interesting. Is there anywhere I can listen to the music generated by your algorithm?
20.
by dimmuborgir 6y ago
Audio Signal Processing for Music Applications Professor Xavier Serra[1] is a highly respected veteran in the field. [1] https://en.m.wikipedia.org/wiki/Xavier_Serra
21.
by dimmuborgir 6y ago
Care to elaborate?
22.
by dimmuborgir 6y ago
Magenta Studio https://magenta.tensorflow.org/studio
23.
by dimmuborgir 6y ago
Who gets to decide what is the "point" of music? Music is a twenty billion dollar industry. An AI system that can spit out highly "realistic" and "pleasing" music can change the music industry as we know it.
24.
by dimmuborgir 6y ago
This might be onto something! Just listen to this from 30s: https://soundcloud.com/openai_audio/pop-rock-in-the-6355437/... Such coherent and pleasing melodic phrases in the style of Avril Lavigne. I thought it co
25.
by dimmuborgir 7y ago
Judging by the four provided melodies: 1) the notes have very little rhythmic variation. 2) the melodies don't seem to have any concept of metre or metric accent.
26.
by dimmuborgir 7y ago
I was going to type Jukedeck. Thought I'd visit the site first. Offline. Apparently TikTok has bought it!
27.
by dimmuborgir 7y ago
Thanks a lot.
28.
by dimmuborgir 7y ago
How does this compare to Code Maps in Visual Studio? Can anyone who has used both comment?
29.
by dimmuborgir 8y ago
Apart from the opposition parties, I could not find a single reputed source that says India's GDP numbers are fudged.
30.
by dimmuborgir 8y ago
Germany and audio software. Name a better duo.

More ›