mpaepper

8 ms·

31.
by mpaepper 1y ago
Are languages other than English also supported? Or is this for English only?
32.
by mpaepper 1y ago
Nope... It IS using a vector database / faiss embedings for the semantic search. Check the indexer: https://github.com/Olow304/memvid/blob/main/memvid/index.py
33.
by mpaepper 1y ago
Yes, that's exactly what I thought as well. An attempt to get more share in the developer tooling space for the long term.
34.
by mpaepper 1y ago
Moin Tom, Thanks for putting up with the work - let's go!
35.
How to run an open-source local voice assistant which sees your screen(paepper.com)
2 points by mpaepper 1y ago|0 comments
36.
by mpaepper 1y ago
It's open-source and everything is running locally on your computer, so you can even use it offline :)
37.
Show HN: Speak, capture, type - my voice assistant works anywhere on your system(github.com)
2 points by mpaepper 1y ago|1 comments
38.
by mpaepper 2y ago
Seems rushed out in response to Manus? The example just calls a single LLM what about that is agentic?
39.
Open-source super fast voice coding (locally running speech-to-text)(github.com)
3 points by mpaepper 2y ago|0 comments
40.
FlashAttention – optimizing GPU memory for more scalable transformers(paepper.com)
1 points by mpaepper 2y ago|0 comments
41.
Show HN: Vibevoice – super fast speech-to-text in any app on your system(github.com)
1 points by mpaepper 2y ago|1 comments
42.
by mpaepper 2y ago
On X it was mentioned that the seller received part of the compensation in OpenAI stock.
43.
Small 1.6B parameter VLM with great vision capabilities(moondream.ai)
3 points by mpaepper 2y ago|0 comments
44.
by mpaepper 2y ago
And they work well outside in the sun?
45.
Ask HN: Best Monitor for Terrace Use?
1 points by mpaepper 2y ago|2 comments
46.
by mpaepper 2y ago
What kind of thermal camera do you recommend and what price range?
47.
by mpaepper 2y ago
Interesting analysis showing that the supposed emergent abilities with sharp changes in metrics when scaling LLMs are actually rather metric dependent and when you change to an actual linear metric, you also see linear progress.
48.
LLMs having emergent abilities is an illusion driven by metrics selection [pdf](proceedings.neurips.cc)
7 points by mpaepper 2y ago|1 comments
49.
Stable Diffusion 3 Turbo(arxiv.org)
1 points by mpaepper 2y ago|0 comments
50.
Google's generative AI model for text2game(sites.google.com)
8 points by mpaepper 3y ago|0 comments
51.
First Apple Vision Pro Kernel Exploit(twitter.com)
43 points by mpaepper 3y ago|6 comments
52.
LiquidAI out of stealth – new generation of AI models from first principles(liquid.ai)
1 points by mpaepper 3y ago|1 comments
53.
by mpaepper 3y ago
Can you apply the sliders also to existing images that were not generated with SD?
54.
by mpaepper 3y ago
This sounds a bit like it. Haven't tried it yet, though. https://github.com/phil294/vimium-everywhere
55.
by mpaepper 3y ago
I use xmonad as a tiling wm, which I guess is quite similar to i3, but what I mean is that you can control all apps with vim like experience a la vimium. For example controlling gimp with vim key bindings. Does i3wm support that? It sounds
56.
by mpaepper 3y ago
Wow, does something like this exist for Ubuntu as well?
57.
by mpaepper 3y ago
I agree it's not so easy to work with Langchain. That's wy I built the simple LLM agents repo which you can understand in 5 minutes: https://github.com/mpaepper/llm_agents
58.
US knew about Ukrainian attack on nord stream pipelines(washingtonpost.com)
6 points by mpaepper 3y ago|1 comments
59.
Lawsuit against OpenAI for being a scheme of Y Combinator(twitter.com)
4 points by mpaepper 3y ago|2 comments
60.
by mpaepper 3y ago
In case you want a much simpler implementation of the ReAct agent with tools, it's here with ~200 loc: https://github.com/mpaepper/llm_agents

More ›