crakenzak

18 ms·

61.
by crakenzak 3y ago
The 128k context window GPT-4 Turbo model looks unreal. Seems like Anthropic's day of reckoning is here?
62.
by crakenzak 3y ago
Does training an LLM on a variety of different languages improve its reasoning and understanding in other languages? For example, if GPT-5 (or whatever next gen model) was trained on a staggering amount of high quality Norwegian tokens that
63.
by crakenzak 3y ago
> Tech is an interest-rate-sensitive industry. why? funding?
64.
by crakenzak 3y ago
> If it's 2023 and on, I dust off my 90's "everything on the World Wide Web is wrong" glasses. because misinformation written by humans didn’t exist before LLMs?
65.
by crakenzak 3y ago
whisper from OpenAI works great for me.
66.
Quick Tip: Enable Touch ID for Sudo (2020)(sixcolors.com)
2 points by crakenzak 3y ago|0 comments
67.
by crakenzak 3y ago
Paper: https://arxiv.org/abs/2307.07635 Github: https://github.com/facebookresearch/co-tracker Demo: https://huggingface.co/spaces/facebook/cotracker
68.
Meta AI releases CoTracker, a model for tracking any points (pixels) on a video(co-tracker.github.io)
345 points by crakenzak 3y ago|84 comments
69.
by crakenzak 3y ago
code: https://github.com/facebookresearch/seamless_communication paper: https://ai.meta.com/research/publications/seamless-m4t/ demo: https://seamless.metademolab.com/
70.
by crakenzak 3y ago
code: https://github.com/facebookresearch/seamless_communication paper: https://ai.meta.com/research/publications/seamless-m4t/ demo: https://seamless.metademolab.com/
71.
by crakenzak 3y ago
Very interesting, so you’re basically saying that a proper technique could potentially predict a papers chance of being successfully replicated? If so, could be great for the field!
72.
by crakenzak 3y ago
So exciting! Just FYI, the wikipedia page for LK-99 has a very useful tracking grid of replication attempts with sources: https://en.wikipedia.org/wiki/LK-99
73.
by crakenzak 3y ago
Just FYI, the wikipedia page for LK-99 has a very useful tracking grid of replication attempts with sources: https://en.wikipedia.org/wiki/LK-99
74.
Perfectionism vs. Pragmatism in Product Engineering(warp.dev)
1 points by crakenzak 3y ago|0 comments
75.
Apple unveils watchOS 10 for Apple Watch(9to5mac.com)
3 points by crakenzak 3y ago|0 comments
76.
by crakenzak 3y ago
> the competition are light-years ahead in terms of brand recognition. Who are you talking about?? Not a single brand is as synonymous with VR as Meta & Oculus Quest. In fact, whenever I speak with any non tech focused people about V
77.
by crakenzak 3y ago
Link to paper: https://cdn.openai.com/improving-mathematical-reasoning-with...
78.
by crakenzak 3y ago
Paper: https://cdn.openai.com/improving-mathematical-reasoning-with...
79.
by crakenzak 3y ago
> you can use google what?! do we look like uncivilized coders to you? /s
80.
by crakenzak 3y ago
Paper: https://arxiv.org/abs/2305.07185 Wow, seems like Meta AI is so ahead of the curve compared to even Google and OpenAI recently especially with their open sourcing pushes. Great for the research community as a who
81.
by crakenzak 3y ago
probably because they were spending all those years putting annoying ads in the OS and start menu.
82.
by crakenzak 3y ago
> We have added native support for additional archive formats, including tar, 7-zip, rar, gz and many others using the libarchive open-source project. You now can get improved performance of archive functionality during compression on Wi
83.
by crakenzak 3y ago
Can someone give a TL;DR?
84.
by crakenzak 3y ago
according to their blog post[1], MMS achieves ~half the error rate on words, while supporting 11x more languages. pretty impressive. [1] https://ai.facebook.com/blog/multilingual-model-speech-recog...
85.
by crakenzak 3y ago
Doesn't seem like there's a demo set up by them yet, but you can just download the model weights and run the inference yourself and compare it to OpenAI Whisper or anything you have access to.
86.
by crakenzak 3y ago
Code: https://github.com/facebookresearch/fairseq/tree/main/exampl... Blog Post: https://ai.facebook.com/blog/multilingual-model-speech-recog... Paper: https://research.f
87.
Meta AI announces Massive Multilingual Speech code, models for 1000+ languages(github.com)
705 points by crakenzak 3y ago|222 comments
88.
by crakenzak 3y ago
Wow this is incredible, thanks for sharing! I was quoted $800/mo after insurance for a moderately low dose of Semaglutide (Ozempic)
89.
by crakenzak 3y ago
Now if only it didn’t cost so damn much every month even under insurance =(
90.
by crakenzak 3y ago
None

More ›