thorum

7 ms·

91.
Training a Flappy Bird Diffusion World Model to Run in a Web Browser(njkumar.com)
3 points by thorum 1y ago|0 comments
92.
by thorum 1y ago
It implies that training on synthetic data will always shift the model’s behavior in unpredictable ways. When the base model is different you don’t get the same correlations, but you get something, likely reinforced with each synthetic trai
93.
by thorum 1y ago
I recommend litellm if you’re writing Python code, since it handles provider differences for you through a common interface: https://docs.litellm.ai/
94.
by thorum 1y ago
Beyond a basic understanding of how LLMs work, I find most LLM news fits into one of these categories: - Someone made a slightly different tool for using LLMs (may or may not be useful depending on whether existing tools meet your needs) -
95.
by thorum 1y ago
I understand the idea. My position is that this is a largely speculative claim from people who have not spent much time seriously applying agents for spreadsheet or video editing work (since those agents didn’t even exist until now). “Getti
96.
by thorum 1y ago
People say this, but in my experience it’s not true. 1) The cognitive burden is much lower when the AI can correctly do 90% of the work. Yes, the remaining 10% still takes effort, but your mind has more space for it. 2) For experts who have
97.
by thorum 1y ago
> The leaked responses show clear signs of being real conversations: they start with contextually appropriate replies, sometimes reference the original user question, appear in various languages, and maintain coherent conversational flow
98.
by thorum 1y ago
The interesting part of context engineering (the actual engineering part) is figuring out how to gather the information the LLM needs to do a task correctly from your system. For example, the secret sauce of GitHub Copilot is how it decides
99.
by thorum 1y ago
It’s very, very hard to remove things from the training data and be sure there is zero leakage. Another idea would be to use, for example, a 2024 state of the art model to try to predict discoveries or events from 2025.
100.
by thorum 1y ago
Interesting project! I’m a little surprised that Claude is willing to call these functions. The demo screenshot is downloading a public domain work, I wonder if it would also happily go along with requests for Harry Potter or other copyrigh
101.
by thorum 1y ago
Some ideas are too complex to explain accurately in simple terms. You can give someone a simple explanation of quantum chromodynamics and have them walk away feeling like they learned something, but only by glossing over or misrepresenting
102.
by thorum 1y ago
I’ve built several apps on yjs and highly recommend it. My only complaint is that storing user data as a CRDT isn’t great for being able to inspect or query the user data server-side (or outside the application). You have to load all the us
103.
by thorum 1y ago
This GitHub issue says 6-7 GB VRAM: https://github.com/resemble-ai/chatterbox/issues/44 But if the model is any good someone will probably find a way to optimize it to run on even less. Edit: Got it running o
104.
by thorum 1y ago
We’ll know AGI has arrived when it can figure out Python dependency conflicts
105.
by thorum 1y ago
Try AI Studio if you haven’t already: https://aistudio.google.com/
106.
by thorum 1y ago
Constraints make the specific goal (moon landing) harder, but force technological development. If landing on the moon had been 'easy' with existing tech and not required that massive investment of resources, progress on everything
107.
by thorum 1y ago
With Jules, I almost always end up making significant changes before approving the PR. So “successful merge” is not great indicator of how well the model did in my case. I’ve merged PRs that were initially terrible after going in and fixing
108.
by thorum 1y ago
> To actually get to the bottom of things: I think most normal folks are concerned more about getting by and making decent money in “the age of AI” than they are about being brilliant whizkid prodigies coming up with original ideas. A lo
109.
by thorum 1y ago
Thanks for the recommendations and I agree completely. There's some hope, we can get better at this even in small ways by carving out pockets of time away from phones and notifications and other external distractions and inputs.
110.
by thorum 1y ago
If you're interested in preserving your ability to think for yourself in the age of AI, I recommend Henrik Karlsson's blog Escaping Flatland. While not directly about AI, his articles "Cultivating a state of mind where new id
111.
by thorum 1y ago
Makes sense, I could see the human touch on the article too, so I figured it was something like that.
112.
by thorum 1y ago
Humorous that this article has a strong AI writing smell - the author should publish the prompts they used!
113.
by thorum 1y ago
Good article. Agree that general unreliability will continue to be an issue since it's fundamental to how LLMs work. However, it would surprise me if there was still a significant gap between single-turn and multi-turn performance in 1
114.
by thorum 1y ago
Good points. I suspect that o3 is able to reason more deeply about different paths through a codebase than earlier models, though, which might make it better at this kind of work in particular.
115.
by thorum 1y ago
Sorry, I added links! Just a week ago someone built a system that used o3 to find novel zero days in the Linux kernel’s SMB implementation.
116.
by thorum 1y ago
My takeaway - from this article, from Google’s AlphaEvolve [1], and the recent announcement about o3 finding a zero day in the Linux kernel [2] - is that Gemini Pro 2.5 and o3 in particular have reached a new level of capability where these
117.
by thorum 1y ago
Apple makes webapps/PWAs hard because they want you to make a native app instead.
118.
by thorum 1y ago
It works on iOS 18. Safari’s audio support was pretty rough on earlier versions, especially <17, but I didn’t need to support those for my project.
119.
by thorum 1y ago
The solution I found after approximately two months of struggling with this problem: you have to generate an audio file that is a few seconds of silence, play it on a loop, and play it at the same time as the actual audio file you want to p
120.
by thorum 1y ago
Google’s ability to offer inference for free is a massive competitive advantage vs everyone else: > Is Jules free of charge? > Yes, for now, Jules is free of charge. Jules is in beta and available without payment while we learn from u

More ›