dworks

9 ms·

91.
by dworks 10mo ago
FWIW, Qwen Z-Image is much better than Seedream and people (redditors) are saying its better than Nano Banana in their first trials. Its also 7B I think, and open.
92.
Meta-Coding(dilemmaworks.substack.com)
1 points by dworks 10mo ago|0 comments
93.
by dworks 11mo ago
Software is 10x more valuable than inference tokens because tokens do nothing for the user, just like a database request by itself does nothing. Software is what makes inference valuable because it builds a workflow that transforms tokens a
94.
by dworks 1y ago
Well, how are they going to "fix it" when their system is an "end to end neural net"? It's a black box. There's no code to update to tell it to behave differently during red lights.
95.
Why Super Apps took off in China(dilemmaworks.substack.com)
2 points by dworks 1y ago|0 comments
96.
by dworks 1y ago
We didn't support the time dimensions, but I think it could be done without too much issue. You suggest displaying search results in a chat interface but that doesn't work because there are easily hundreds of hotel results for mos
97.
by dworks 1y ago
I don't think such massive amounts of text should be parsed at runtime. Hotels can have 100s or 1000s of reviews. We batch created attributes for hotels based on reviews, and when a semantic search was run, those attributes were matche
98.
by dworks 1y ago
> Find me hotels in Capetown that have a pool by the beach .Should cost between 200 dollars to 800 dollars a night I built this 18 months ago at an OTA platform. We parse the query and identify which terms are locations, which are hotel
99.
by dworks 1y ago
People naturally make maps for their own needs and purposes. The article seems to think there's only one map but there's many. China has it's own version with itself in the center: https://studycli.org/chinese
100.
by dworks 1y ago
It definitely has resulted in mass prosperity.
101.
by dworks 1y ago
The US cannot sanction China's shipping of goods.
102.
by dworks 1y ago
They made Japan agree to doubling the exchange rate of the Yen, which crashed their exports among other things.
103.
by dworks 1y ago
If you read the Wikipedia article more carefully you would have understood that the loose monetary policy was an effect of the Plaza Accord, hence why I mentioned it.
104.
by dworks 1y ago
look up the plaza accord. It led to a bubble that eventually burst and an uncompetitive export industry as the JPY doubled.
105.
Alibaba enters robotaxi market with investment in Ant Group-backed Hello(scmp.com)
1 points by dworks 1y ago|0 comments
106.
by dworks 1y ago
Most chipmakers in China are making or have made their new generation of products CUDA-compatible.
107.
by dworks 1y ago
"Your margin is my opportunity" as someone said. Certainly Google must have plans to sell its chips externally with this much up for grabs?
108.
China bans its biggest tech companies from acquiring Nvidia chips(tomshardware.com)
11 points by dworks 1y ago|0 comments
109.
Alibaba lands China Unicom as flagship client for its AI chips(cryptopolitan.com)
4 points by dworks 1y ago|0 comments
110.
Alibaba's new AI chip: Key specifications comparable to H20(news.futunn.com)
297 points by dworks 1y ago|308 comments
111.
Qwen 3 now supports ARM and MLX(alizila.com)
55 points by dworks 1y ago|8 comments
112.
by dworks 1y ago
OpenAI will never be profitable selling raw tokens. They need the application layer that allows them to sell additional functionality and decouple the cost of a plan from the cost of tokens. See Lovable, they abstract away tokens as "c
113.
by dworks 1y ago
youre hand wavibng all the other million use cases where returning false information isnt OK.
114.
by dworks 1y ago
the return may still not reflect the sandbox reality.
115.
by dworks 1y ago
its non-deterministic. it gives different answers each time you ask, potentially, and small differences in your prompt yields different completions. it doesnt actually understand your prompt, you know.
116.
by dworks 1y ago
to be fair?
117.
by dworks 1y ago
get a second agent to validate the return from the first agent. but it might get it wrong because reasons, so you need a third agent just to make sure. and then a fourth. and so on. this is obviously not a working direction.
118.
by dworks 1y ago
that's because its "answers" are actually "completions". cant escape that fact - LLMs will always "hallucinate".
119.
by dworks 1y ago
We're going to see increasingly more of these, and it's going to cause a big scandal at one point, that pops the current AI bubble. It's really obvious that you can't use non-deterministic systems this way but companies
120.
My Two and a Half Years at TikTok E-Commerce in the US: Hope to Disillusion(dilemmaworks.substack.com)
7 points by dworks 1y ago|3 comments

More ›