bayes-song

4 ms·

1.
Show HN: Unlock Claude Sonnet 5's original reasoning(thinking-signature-demo-829446634001.asia-east1.run.app)
2 points by bayes-song 3mo ago|0 comments
2.
by bayes-song 3mo ago
The new Codex app displays Chat or Code sessions is an absolute disaster. I can't tell which project is which at all.
3.
Can we reconstruct a tokenizer using only two oracles from the chat API?(old.reddit.com)
1 points by bayes-song 3mo ago|1 comments
4.
by bayes-song 3mo ago
From the chat APIs alone, we can extract two useful oracles: Token length oracle: Given any string s, return len(tokenize(s)). Prefix token oracle: Given a string s and integer n, return the string decoded from the first n tokens of tokeniz
5.
by bayes-song 3mo ago
I saw the earlier HN discussion about Claude’s encrypted reasoning blobs and found it interesting. After some experiments, I think I may have found a way to get the original reasoning from signature. I built a verification test. The idea is
6.
I think I can get the original reasoning of Claude models. Is this real?(thinking-signature-demo-5g65bijswq-de.a.run.app)
3 points by bayes-song 3mo ago|2 comments
7.
Show HN: I made an AI that reviews iPhone apps – 1h of autonomous GUI work(understudy-ai.github.io)
3 points by bayes-song 6mo ago|0 comments
8.
by bayes-song 7mo ago
In Chinese internet slang there's actually a joke called the 确实型人格, someone who just replies "true" or "your are correct" to everything.
9.
by bayes-song 7mo ago
Fair enough.
10.
by bayes-song 7mo ago
not true
11.
by bayes-song 7mo ago
Thanks! And good luck with your project as well. One of the motivations for open-sourcing this is exactly to see it grow beyond macOS. I personally don’t have much development experience on Windows or Linux, so it’s great to see people pick
12.
by bayes-song 7mo ago
thx
13.
by bayes-song 7mo ago
I really like the Claude Chrome extension, but unfortunately it has too many limitations. Not only is it restricted to Chrome, but even within Chrome some websites especially financial ones are blocked.
14.
by bayes-song 7mo ago
sad…
15.
by bayes-song 7mo ago
That’s true. The demo I showed was somewhat cherry-picked, and agentic systems themselves inherently introduce uncertainty. To address this, a possible approach was proposed earlier in this thread: currently, after /teach is completed,
16.
by bayes-song 7mo ago
In the demo, I used GPT-5.4:medium accessed through the Codex subscription.
17.
by bayes-song 7mo ago
That sounds like a good idea. During the use of a skill, if the agent finds something unclear, it could proactively ask the user for clarification and update the skill accordingly. This seems like a very worthwhile direction to explore. In
18.
by bayes-song 7mo ago
Fair point that Linux is underserved. My own view is that the bigger long-term opportunity is actually Windows, simply because more desktop software and more professional workflows still live there. macOS-first here is mostly an implementat
19.
by bayes-song 7mo ago
That’s exactly the hard part, and I agree it matters more than the happy path. A few concrete things we do today: 1. It’s fully agentic rather than a fixed replay script. The model is prompted to treat GUI as one route among several, to pre
20.
Show HN: Understudy – Teach a desktop agent by demonstrating a task once(github.com)
120 points by bayes-song 7mo ago|41 comments
21.
by bayes-song 3y ago
This is a good question. Personally, I think it is a better way for most companies to directly use the API of OpenAI, which is currently the best in terms of ease of use and ability in the general field. I think there are only a few types t
22.
by bayes-song 3y ago
Thank you for your suggestion, this will indeed be more intuitive, I will add relevant results as soon as possible.
23.
by bayes-song 3y ago
Check out this model trained using the Open-Llama project at http://home.ustc.edu.cn/~sl9292 . This model is trained primarily on English and Chinese, but also has capabilities in other languages like Japanese and Korean. N
24.
Open-Llama: Complete training pipeline for building large language models(github.com)
141 points by bayes-song 3y ago|12 comments
25.
Open-Llama: A Open Source Project for Training Language Models(github.com)
4 points by bayes-song 3y ago|1 comments
26.
by bayes-song 3y ago
Exciting news! Check out this model trained using the Open-Llama project at http://home.ustc.edu.cn/~sl9292 . This model is trained primarily on English and Chinese, but also has capabilities in other languages like Japanes