unclecode.bsky.social
Author of Crawl4AI (#1 GitHub Trending, 16K+ stars). Founder of Kidocode, SE Asia's largest tech & biz school. Based in Singapore, running an AI research lab on synthetic data. AI researcher/consultant, systems engineer, musician, coffee enthusiast.
17 posts
17 followers
9 following
Conversation Starter
comment in response to
post
Oh yeah, such a good idea, I even didn’t think of it!! Definitely consider that. Btw thx for kind words.
comment in response to
post
I have to write a short essay on this to convey my message!
comment in response to
post
In my view, automate coding tools often fail with complex tasks because of a lack of linguistic awareness. Both the developers building these systems and the users relying on them miss the importance of language as the core mechanism driving these models, limiting their effectiveness.
comment in response to
post
Language models don’t truly “understand” function calls, they interpret language. We guide them by embedding functions as part of a linguistic framework. To build better AI tools, developers need to understand linguistics and the philosophy of language, an area often overlooked.
comment in response to
post
This definitely outshines OpenAI GPTs! You can wrap any tools into a simple server app, pass it to Claude, and that’s it. Claude can use them all! It’s far more flexible and, with good design, allows for scalability. To me, this is one of the best moves in increasing LLM utility in recent years.
comment in response to
post
Amazing, was thinking of it today, thx for sharing the link
comment in response to
post
Nostalgic! HSB2 dataset and descriptive stats… 😍 such a good time, for me was like ~15 years ago 😄 but hsb2, I think was around 2015… not sure, hv to define the confidence interval 😁
comment in response to
post
We occasionally chat on X, but you don't following me. Hopefully, we can connect here. I focus on synthetic data generation, started with Crawl4AI, an open-source crawler and data extraction tool, and planning to continue with a synthetic-data-on-demand pipeline.
comment in response to
post
How can I dm you? Like to join to community.
comment in response to
post
I hope this new platform will be different, where posts actually reach people who care. I still love X for the friends and connections I’ve made, but this issue is big for me. Let’s see how it goes here, hoping for better!
5/5 🤞
comment in response to
post
This has left me replying to others’ posts from people I like just to stay visible and get exposure. But that’s not the only thing I want to do, I want to share my own ideas. It feels disconnected, like I can’t reach the audience that matters.
4/5 🧵
comment in response to
post
My first post about my repo got 90K views and great engagement. Since then, nothing. People say you need to “volume post” (spam posts all day hoping one breaks through). What the hell!! Why should I hack algorithms just to reach relevant people? Is this designed to force us to pay to boost?
3/5 🧵
comment in response to
post
No matter what I post, even random words, the X algorithm caps my impressions at 200. Meanwhile, low-quality posts easily get thousands of views. When only 200 people see my posts, how can I reach those who actually need my content? No way to accept only 200 in world are relevant to my posts!
2/5 🧵
comment in response to
post
Done and I connect my main reporr github.com/unclecode/cr...