mohsen1

9 ms·

151.
by mohsen1 7mo ago
Uses public dataset to evaluate which is not meant for evaluation. Writes super specific prompt[1] and claims eye catching results. This is the state of "AI" these days I guess... [1] https://github.com/symbolica-a
152.
by mohsen1 7mo ago
If it was not spinning so many Python processes and not overwhelming the system with those (friends found out this is consuming too much CPU from the fan noise!) it would have been much more successful. So similar to xz attack it does a lot
153.
by mohsen1 7mo ago
I highly recommend adding `/simplify` to your workflow. It walks back over-engineerings quite often for me.
154.
by mohsen1 7mo ago
playwright can do all of that too. I'm confused why this is necessary. If coding agents are given the Playwright access they can do it better actually because using Chrome Developer Tools Protocol they can interact with the browser and
155.
by mohsen1 7mo ago
`^` is the symbol for the Control key not `⌘`
156.
by mohsen1 7mo ago
Paper says adding period at the end changes this behavior
157.
by mohsen1 7mo ago
Why this person is so mean to someone who gifted Deno and Node to the JS ecosystem? It's not fair. They are trying to build a company on top of open source.
158.
by mohsen1 7mo ago
I never claimed to beat xz as a general-purpose compressor. .tar.xz is the dominant format for Linux source tarballs and distro packages. So optimizing for ELF + x86_64 is optimizing for a very real and common case, not some toy benchmark.
159.
by mohsen1 7mo ago
When there is a solid test harness, AI Coding can do magic! It was able to beat XZ on its own game by a good margin: https://github.com/mohsen1/fesh
160.
by mohsen1 7mo ago
I haven’t really been opening an IDE for a while now, and that does seem to be the direction I’m heading. I even deleted VS Code from my machine. On the rare occasions I need to edit a file by hand, I just use Vim. Different developers are
161.
by mohsen1 7mo ago
Cursor Composer 1 was Qwen and this is Kimi. IDE is based on VSCode. The entire company is build on packaging open source and reselling it. Ollama is also doing this. There is so much money to be made repackaging open source these days. So
162.
by mohsen1 7mo ago
For the Chinese customer it is not a headline, it is an actual price tag
163.
by mohsen1 7mo ago
making that work cross platform is pure pain
164.
by mohsen1 7mo ago
Yarn, pnpm, webpack all have solutions for this. Great to see this becoming a standard. I have a project that is severely handicapped due to FS. Running 13k tests takes 40 minutes where a virtual file system that Node would just work with i
165.
by mohsen1 7mo ago
As a black box it works. It produces smaller binaries. when extracted matching bit-by-bit to the original file. I tested across 100 packages. better efficiency across the board. But I don't know if I (or anyone) want to maintain softwa
166.
by mohsen1 7mo ago
Are you an expert in this field? I'm curious if the AI generated code here is actually good.
167.
by mohsen1 7mo ago
Hers is one: https://github.com/mohsen1/fesh beats the best compression out there by 6% on average. Yet nobody will care because it was not hand written
168.
by mohsen1 7mo ago
Neighbor?!
169.
by mohsen1 7mo ago
in Iran there's a funny situation where you're connected but not to the internet. if that makes sense.
170.
by mohsen1 7mo ago
This is a life saver in Iran right now. Maybe only 0.01% have access to internet using Starlink
171.
Ceno, browse the web without internet access(ceno.app)
134 points by mohsen1 7mo ago|40 comments
172.
by mohsen1 7mo ago
It's possible that they use the camera to measure room brightness
173.
by mohsen1 7mo ago
They missed some other type of windows like Activity Monitor graphs. Those are even sharper corners!
174.
by mohsen1 7mo ago
Just wanted to see my Claude account usages https://github.com/mohsen1/claude-usage Interesting time to be alive. there are software similar to this but not _exactly_ what I wanted so I made it in a few hours!
175.
by mohsen1 7mo ago
when asking your AI to write hacker news comments, at least tell it to format it for hacker news comment format!!
176.
by mohsen1 7mo ago
Because it's a diffusion model that hardly able to redo what it just did.
177.
by mohsen1 7mo ago
I really want someone to build this: Using a webcam, monitor finger movements and find mistakes (using some sort of AI video analysis) to help user figure out how to improve. It's a hard thing to build but if you build it there is goin
178.
by mohsen1 7mo ago
Wanted to see if AI could figure out how to compress executable binaries better than existing generic tools without me actually knowing much about compression engineering or ELF internals. The result is an experiment called fesh. It works s
179.
by mohsen1 7mo ago
it is basically a “performance regression detector” It looks at a time series of benchmark results (e.g. test runtime, latency, memory usage across commits) and tries to answer one question: did something actually change, or is this just no
180.
by mohsen1 8mo ago
It was Christmas gift. so maybe 2020... not super positive about this

More ›