j-wang

7 ms·

1.
by j-wang 1y ago
I was about to say exactly this—it's not really that different from managing a bunch of junior programmers. You outline, they implement, and then you need to review certain things carefully to make sure they didn't do crazy things
2.
Fine-tuning LLMs is a waste of time(codinginterviewsmadesimple.substack.com)
193 points by j-wang 1y ago|88 comments
3.
by j-wang 1y ago
"But this logic breaks down for advanced models, and badly so. At high performance, fine-tuning isn’t merely adding new data — it’s overwriting existing knowledge. Every neuron updated risks losing information that’s already intricatel
4.
by j-wang 2y ago
I miss the entire Maker Movement moment. It isn’t actually super complex to make a simple circuit board that does some basic things for you. It’s cheaper to just outsource making the board, but I even made my own simple boards (etched using
5.
by j-wang 2y ago
How AI agents of the current era (with LLMs) differs from the history of reinforcement learning-based agents
6.
The AI Agent Spectrum(interconnects.ai)
1 points by j-wang 2y ago|1 comments
7.
Is OpenAI Overpriced?(financetldr.com)
2 points by j-wang 2y ago|0 comments
8.
by j-wang 2y ago
ML researchers in statistics departments write stuff in R, which makes everyone scream. ML researchers absolutely do. My point in the article was basically the class was "indoctrinating" (too strong, but you get the point) the fut
9.
by j-wang 2y ago
> I dont understand this - arent almost all ML NN models built in pytorch, and arent these compiled / jit'd into a lower level format - and can we not have various backends/drivers for that, such as CUDA / ROCM /
10.
CUDA Is Still a Giant Moat for Nvidia(weightythoughts.com)
84 points by j-wang 2y ago|111 comments
11.
Who's Winning the AI War?(weightythoughts.com)
1 points by j-wang 3y ago|0 comments
12.
Compute Is Overrated as AI's Bottleneck(weightythoughts.com)
1 points by j-wang 3y ago|0 comments
13.
by j-wang 3y ago
Potentially—I suppose free can lead to more justifiable laziness in finding different resources for different things. The argument I generally hear from certain microeconomists is that they still expect there to be value in niches, given Go
14.
by j-wang 3y ago
That's an interesting question since advertising certainly follows search dominance, but it doesn't necessarily follow the other way around. Google figured out how to monetize its dominance with advertising before it had the behem
15.
by j-wang 3y ago
Search engines have always been a bit weird. Do they have network effects? Why is it actually winner-take-all? I've had spirited conversations with a lot of different people and academics/microeconomists on the topic and I don&#x2
16.
by j-wang 3y ago
I think various of those aspects you call out here, I do as well. The specificity of the application is fairly key, whether it comes through proprietary data or application-specific stuff or simply business-lock-in. Interesting hill analogy
17.
by j-wang 3y ago
Good point—that is the point. When there's a hype cycle, people often check their normal business sense at the door in terms of customers, value generation, and defensibility. Are there any of these? No, but it's crypto. Or now AI
18.
Most AI startups are doomed(weightythoughts.com)
177 points by j-wang 3y ago|132 comments