lordswork

7 ms·

31.
by lordswork 2y ago
When it comes to how "open" LLMs are, there are basically three categories: - Open Source: The architecture of the model is available, so you can bring your own data and compute to train a similar model. - Open Weights: The traine
32.
by lordswork 2y ago
Emissions are the bigger issue. Unfortunately, some sort of mining is still required for emission and emission-free sources of energy.
33.
by lordswork 2y ago
We need a progressive carbon tax. They should be covering their emissions plus that of many others. Even they are in favor of this.
34.
by lordswork 2y ago
Character.ai does invent everything from scratch (for better and worse).
35.
by lordswork 2y ago
What are your predictions on how H1B visa holders will be impacted if Trump wins the election?
36.
by lordswork 2y ago
But ultimately, it's an unsolved problem in the field. Every single LLM has been jailbroken.
37.
by lordswork 2y ago
Counter-nit: I found the examples within a few seconds.
38.
by lordswork 2y ago
The media hates Google.
39.
by lordswork 2y ago
Some interesting context on this work: 2 researchers were bullied to the point of leaving Google for Anthropic by a senior researcher (who has now been terminated himself): https://www.wired.com/story/google-brain-ai-re
40.
by lordswork 2y ago
There seems to be several components involved: tree of thoughts, MCTS, RL, high quality chain of thought data, possibly multiple models.
41.
by lordswork 2y ago
It's much more than CoT. I suspect it will take other labs some time to replicate.
42.
by lordswork 2y ago
Then why hasn't it been done?
43.
by lordswork 2y ago
It's not just CoT
44.
by lordswork 2y ago
To be fair, o1 is a major breakthrough in the field. If other AI labs can't crack scaling useful inference compute, OpenAI will maintain a big lead.
45.
by lordswork 2y ago
That sounds like a great broader vision, but let's also celebrate the significant step in that direction that this work presents. This appears to be very useful as is.
46.
by lordswork 2y ago
Sergey is still down in the trenches quite a bit. He regularly pair-programs with folks on the Gemini team and ran an AI paper review for a while.
47.
by lordswork 2y ago
The background loads for me but then nothing (on mobile Android)
48.
by lordswork 2y ago
Timing is everything. Maybe now is the right time for a music discovery app to take off.
49.
by lordswork 2y ago
There's just so many new attack vectors with advanced AI, many of which we haven't imagined yet. It's a new world, and we have a lot to learn. Anyone claiming 100% certainty about how safe or unsafe it will be is either delus
50.
by lordswork 2y ago
BMGF promotes vaccines to decrease global mortality rates. It's not surprising a bunch of anti-vax groups are unhappy with them. The scientific method will always prevail.
51.
by lordswork 2y ago
The lack of laws around this right now are so damaging. A negligent homeowner can unknowingly be growing a nursery of invasive plants that are putting out tens of thousands of seeds to the surrounding area every year, not to mention direct
52.
by lordswork 2y ago
It's already here. There are communities forming around generating passive income from mass producing AI videos as tiktoks and shorts.
53.
by lordswork 2y ago
It is pretty amazing. I've been using it every day. I do wish you could easily upload an entire repo into it though.
54.
by lordswork 2y ago
I am also wondering how to use it..
55.
by lordswork 2y ago
The regions where the plant is considered native, how aggressive it spreads (there is some metric that captures this IIRC), and regions where it is labeled as invasive. The PictureThis app shows this info, and it's incredibly useful.
56.
by lordswork 2y ago
What prevents an AI agent from becoming the product manager as well, and communicating with you (the customer) to clarify requirements?
57.
by lordswork 2y ago
>Will you just be able ask the AI to refactor it all and clean it up? Then it wouldn't be a problem I presume. For smaller contexts, LLMs tend to be really good at reviewing, suggesting changes, and refactoring. I haven't seen
58.
by lordswork 2y ago
If we have fixed resources, how should they be allocated? I'd argue for every million dollars towards reaching net zero emissions, we should maybe spend 1 dollar towards carbon capture, where we scale that investment up as we approach
59.
by lordswork 2y ago
We need both, but reaching zero emissions is far more urgent and cost effective (by many orders of magnitude) right now.
60.
by lordswork 2y ago
I got "an error has occurred"

More ›