@ForBo7_
Account based inHong Kong
About this account
- Account based in
- Hong Kong
- Connected via
- Web
Account-level information from X, not a live location or the device used for a specific post.
「Open to Projects」 • Dabbler • Learner • Explorer • https://nitter.cf/t.co/rMAvL5souA student • https://nitter.cf/t.co/7FBysT47M6 solver • Dabbling in Embodied AI • 自学中文 // Teaching myself Chinese
Hong Kong/Shenzhen/GBA
Joined September 2022
- Tweets1.5K
- Following1K
- Followers261
- Likes388
Pinned Tweet
Doing lesson 15 of the @fastdotai course; deducing how to rearrange convolutions as a matrix product
Salman // 萨尔曼 retweeted
Always before, in mathematics, solving a problem was nearly entirely aligned with finding useful reusable teachable ideas that pushed the field forwards.
Now that's not always true any more.
I think that's a really important and interesting insight! :)
an idiot in motion beats a genius at rest
I can’t stress enough how little an idea matters compared to the agency of the people executing the idea.
I have had the privilege of knowing and sometimes even working with some of the most successful people (by various metrics).
The difference between mediocre and excellent work and outcomes is predominantly one of agency.
In practice this means: they dont wait for things to happen to them they go out and make things happen for them.
They don’t wait for someone else to do something, for someone to teach them, for someone to give them the path, etc. They just go out and find a way to do it.
I think the single biggest superpower these people have is the realization/belief that the world around them is completely mutable. Most everything that happens is because a person made it happen.
I used to tell people to look around the room you’re sitting in. Look at everything. Every noun. It almost all exists because a person willed it into existence. Nothing is stopping you from doing the same.
I see people online all the time dismissing someone else’s success because “I had that idea first” or whatever. I mean… yeah? If so then the difference is… you. So a bit of a self own whenever I hear that.
Number one tip: act with agency.
Salman // 萨尔曼 retweeted
I love Astra BUT
I genuinely get ick whenever people use these models to tokenmaxx on some task they could solve with like 5k tokens using existing CV stack like @huggingface transformers or @roboflow under 200 LoC and low latency (real time)
I used GPT-6 Astra Ultra to track a tennis ball. The results are impressive until I tell you the following.
Ball annotation, using 3 parallel agents took 8m 35s and 17,611 tokens.
Complete workflow: 11m 49s and 30,481 tokens.
The complete workflow also processed 313,914 uncached input tokens and 7,528,320 cached input tokens.
That's a total of 7.87 million total, mostly context reused across frame-inspection calls.
Salman // 萨尔曼 retweeted
I’m seeing lots of tweet replies asking basically “but how do I get two models to talk to each other? Without me manually copy/pasting results between apps?”
And I’ve heard this same question repeatedly from friends who use AIs mainly through GUI apps.
Yes, there are various apps starting to be released for just this problem. But it’s worth knowing that in a pinch you can already do all this yourself, on the command line, just by using tmux.
Here’s how.
On the command line start a named tmux session with the command “tmux new -s chat”.
Create a second window within that session. (The default keystroke for this is doing “ctrl-b c”. You can switch between windows by doing “ctrl-b n” for the next window.)
Launch the codex CLI in one window, and the claude CLI in the other.
Then you can literally say to the AI in window 1, “please read the analysis of the AI in window 2 of the tmux session ‘chat’. Wdyt?” And vice versa. Now you have cross-harness review.
Or you can instruct one AI to delegate and manage the other AI over multiple turns, or monitor it over time, or whatever. This works because each AI can use the tmux CLI to read and write to the other’s window.
This does not on its own solve coordination problems like deciding who is in charge, or who talks first, or who waits for whom, but it establishes the basic communication primitives of “read the other AI’s output” and “write to the other AI”. You can build more complex workflows on top of that mostly by prompting.
The fundamental reason this all works is that the Unix-era command line environment is more interoperable than the modern GUI environment, so a decades-old tool like tmux is still one of the best ways to compose modern AIs. Obviously the command line tools are a PITA in other ways. But it’s great that they are so flexible, they already exist, and they work right now, so you don’t need to wait for the vendors to get their apps to play nicely with each other (which might never happen) or for someone else to catch up and write a new app just to connect things.
Buying a book is a poor investment unless you make it yours, literally. The book is a part of you and you are a part of the book.
You not need to only read between the lines, but also write between the lines, literally.
Speed reading does improve comprehension, but reading quicker only means the material itself isn't worth spending time on.
Reading is as activate as a ball game where the author is the thrower and the reader is the catcher.
Reading is as complex as snowboarding, if not more. There are various practices that comprise snowboarding, and they are easier practiced first separately.
This inspectional reading is skimming/prereading. The next step is coarse reading
You read start from end without stopping, even if you don't understand. If you stop, you miss the crucial points which are often easy to understand.
However, most books aren't worth coarse reading
After these 4 steps, you can now decide if the material is worth reading more closely.
5. Read the summaries of those chapters you deem have relation to the material's theme
6. Randomly flip around the material and read some portions. Also read the last 2-3 pages.
1. Look at the title page and quickly read the preface
2. Read the table of contents and read the summary for each chapter
2. Check the index and then read the material where important vocabulary appears
4. Look at the material's intro, blurb, and any author/publisher comments
Inspectional reading allows you to answer
1. Is this material worth reading?
2. What information/viewpoints are present in the material?
There are 2 steps to inspectional reading. Seasoned readers perform them simultaneously.
There are 4 layers to reading:
1. Foundational reading
2. Inspectional reading
3. Analytical reading
4. Thematic reading
Most readers remain at level 1. Reading is not a passive activity, and requires agency and activeness.
Read my full post here
snqv.dev/posts/2026-09-03_ho…
How to Inspectionally Read
I thought I knew how to read. Turns out, I didn't quite.
In this thread, I highlight a few takeaways from what I learned from part 1 of Adler's and Doren's 1972 "How to Read a Book"
To complement this, I also created close reading notebooks for @3blue1brown's essence of linear algebra series.
Close reading is where you read *out* of the text. With an LLM, you can potentially stay in a flow state for longer–you ask right there, with all context.
Created close reading notebooks for all lessons in
@math_rachel's fastai computational linear algebra course
You want to read out of the text (not into the text) when you close read. Use a LLM, and you're in a flow state for longer. You can ask right there, with all context.
Essence of linear algebra repo. The repo has been primarily designed for use with solve.it.com, though it should be adaptable for other LLM tools
github.com/ForBo7/essence-of…
SolveIt is currently in limited access. If you want to try it out, you can use this link: solve.it.com/signup