crazylogger

7 ms·

61.
by crazylogger 1y ago
This is like asking me "how much of your software is built by the compiler?" -> the answer is 100%. Ask "how much did you build then?" -> also 100%. The compiler and I operate on different layers.
62.
by crazylogger 1y ago
This is exactly how human society scaled from the cavemen era to today. We didn't need to make our brains bigger in order to get to the modern industrial age - increasingly sophisticated tool use and organization was all we did. It onl
63.
by crazylogger 1y ago
I think next year's AI benchmarks are going to be like this project: https://www.anthropic.com/research/project-vend-1 Give the AI tools and let it do real stuff in the world: "FounderBench": Ask the A
64.
by crazylogger 1y ago
Me and my LLM buddy together understand exactly how computers work!
65.
by crazylogger 1y ago
Humans are checked against various rules and laws (often carried out by other humans.) So this is how it's going to be implemented in an "AI organization" as well. Nothing strange about this really. LLM is easier to work wi
66.
by crazylogger 1y ago
I can see this makes sense for simple { user_query -> search -> llm_answer } usage, where tool use is only a means to retrieve background info. For complex real-world agent flows though, tool use is often the only thing that the LLM i
67.
by crazylogger 1y ago
US I-94: https://i94.cbp.dhs.gov/search/history-search Knowing the passport number + name + birthday gives you access to someone's US travel history.
68.
by crazylogger 1y ago
Ask the AI to document each module in a 100-line markdown. These should be very high level, don't contain any detail, but just include pointers to relevant files for AI to find out by itself. With a doc as the starting point, AI will h
69.
by crazylogger 1y ago
I imagine Iran will just pick a 1000-meter mountain to dig under then?
70.
by crazylogger 1y ago
Sub-agent is another LLM loop that you simply import and provide as a tool to your orchestrator LLM. For example in Claude Code, sub-agent is a tool called "Task(<description>)" made available to the main LLM (the one that y
71.
by crazylogger 1y ago
It’s not a clear line though. Compilers have been writing programs for us. The plaintext programming language code that we talk about is but a spec for the actual program. From this perspective, English-as-spec is a natural progression in t
72.
by crazylogger 1y ago
What you described can be (and is being) achieved by agentic systems like Claude Code. When you give it a task, it knows to learn best practices on the web, find out what other devs are doing in your codebase, and it adapts. And it condense
73.
by crazylogger 1y ago
I don't think it will settle things even if we did manage to train an 1800 LLM with sufficient size. LLMs are blank slates (like an uncultured primitive human being - albeit LLM comes with knowledge built-in, but builtin knowledge is m
74.
by crazylogger 1y ago
What are you basing this on? Is there an “inventiveness test” that humans can pass but LLMs don’t? I’m not aware of any.
75.
by crazylogger 1y ago
Nothing fundamentally prevents an LLM from achieving this. You can ask an LLM to produce a PR, another LLM to review a PR, and another LLM to critique the review, then another LLM to question the original issue's validity, and so on...
76.
by crazylogger 1y ago
Surely at that point the CEO would be AI as well. The owner (human) would say "build a company, make me a billion dollars" and that would be the only valuable input needed from him/her. Everything else would be derived &
77.
by crazylogger 2y ago
This is trivially bypassed by OpenAI asking the user to take control of their computer (or a sandboxed browser within it,) then for all intents and purposes it’s the user themselves accessing your site (with some productivity/accessibi
78.
by crazylogger 2y ago
Yes, this is the point made by the root-level comment: the US would have to build Chinese style Great Firewall to achieve its goal if Bytedance didn't willingly take down their sites. In my understanding this law in question requires U
79.
by crazylogger 2y ago
You raise a good point. Let me step back and say that previously it's mostly authoritarian countries doing it and this type of censorship is generally unacceptable in Western societies, that the US is fundamentally different from China
80.
by crazylogger 2y ago
IMO Pandora's box is now open, and it's a matter of time before we see countries around the world offer the same divest-or-ban choice to Meta, Google, etc. The question is who first and how quickly it will happen.
81.
by crazylogger 2y ago
The funny thing here is that complying means they never do business in the US anyway. So potentially it is beneficial for Bytedance to still serve US viewers but keep no physical presence in the US, and deal only with non-American adverti
82.
Test-driven development with an LLM for fun and profit(blog.yfzhou.fyi)
219 points by crazylogger 2y ago|89 comments
83.
by crazylogger 2y ago
Intelligence has not been LLM's major limiting factor since GPT4. The original GPT4 reports in late-2022 & 2023 already established that it's well beyond an average human in professional fields: https://www.microsof
84.
by crazylogger 2y ago
When doing X becomes cheaper with the invention of new tools, X is now more profitable and humanity tends to do much more of it. Nearly all code was machine-generated after the invention of compilers. Did the compiler destroy programming? A
85.
by crazylogger 2y ago
For one thing, Chinese government does have an incentive to enforce good encryption so that foreign adversaries cannot snoop in on important Chinese communications. Only the Chinese government has access via Tencent’s backend.
86.
by crazylogger 3y ago
Great stuff! I wonder if it's possible to make it available as a web app? It would be even cooler if no app install is required, just send the link to a receiver via text message.
87.
by crazylogger 3y ago
How are you running the model? I believe it's a bug from a rushed instruct fine-tuning or in the chat template. The base model can't possibly be this bad. https://github.com/ollama/ollama/issues/2650
88.
by crazylogger 3y ago
I think with macOS’s design of the dock (resizable, unlike on Windows where the task bar always takes up the whole monitor width,) having one or more windows “full screen” on your desktop feels wrong. There is always some some space left un
89.
by crazylogger 4y ago
How is using this tool different from manually concatenating (website name + username + secret key), e.g. “reddit$myname$123456”? Since Reddit would hash it anyway, this plaintext string should be as secure as a Pashword.
90.
by crazylogger 4y ago
Or it could simply be that users of public transport tend to be poorer, and therefore live farther from work for lower rent.

More ›