171 karma · joined May 23, 2011
Qwen3-4B-Instruct-2507-4bit Qwen3.5-35B-A3B-4bit Qwen3.5-9B-MLX-4bit Qwen3-Reranker-0.6B-4bit Qwen3-Coder-30B-A3B-Instruct-4bit qwen2.5:0.5b
and the results are what i kinda expected to begin with, this adds next to nothing? - also the repo was pivoted from a playwright sub assembly to this not long ago - so my conclusion - THIS MIGHT be worth some watching if you have a model where no-one!, has optimized it at all - and where it does not use anything native to your platform.
results (averages)
VIA rapid-mlx :8902 (MLX) Decode: 175 tok/s TTFT: 64 ms prefill (~760 tok cold): 594 ms
Magnitude 0.2.1 (GGUF/llama.cpp+Rust) Decode: 161 tok/s TTFT: 111 ms prefill (~760 tok cold): 669 ms
I wish it was different but estetics, color sensitivity and for a lack of better word “taste” is a skill that needs training - the kerning alone on those make me sad - but the dirty secret is that meticulous design and craft is a dying breed - why spend days on designing whitespace when you can get something the client likes in 10 minutes from a prompt?
For the type interested - get this: https://amzn.eu/d/09z7Mut8
You’d change seperators and position if you truly wanted to do this right
Its basic date wrangling and tbh i see nothing malfeasant here
Its basically yyyy/mm/dd yy/M/d mm-dd-yyyy stuff but suuuuper lazily done
using modern terminal tooling and better integrated pipeline / parallel tools average discovery and file based operations can be significantly improved and speeded up.
let me know what you think and suggestions welcome... totally vibecoded from local setup so ymmv
Love japanese and japan but their work culture is horrific - Japanese are inefficient and the veneer of looking to work "hard" is more important than the hard work itself. People often stay until ridiculously late just to show they "put in the effort" which is more important than outcome.
Then again that happens in many other countries as well ...
Happy new years and thanks
Cancer metabolism isn’t a 2-bit meme. Tumors adapt. If they couldn’t, they wouldn’t metastasize.
Works well for me
Lit.dev
Even if it is widely dangerous to do so (most american cities i've ever visited)
You can hem and haw - but its pretty bang on
When you then add finding parking at the ends of your trip to it it is crazyly more efficient timewise.
Now even copenhagen denmark has rain causing many more to take a car or public transport (that works).
But it is very clear that the time argument is simply not true.
Now you can argue convenience at the start of the trip vs agony in the end (finding that parking space)
Or for "need to lug an ikea sofa across time"
Or even for "my kids and familiy needs to go as well"
That's super fine, and all true - but 70-80% of ALL trips in cars are by 1 person sitting in 1 car. So moving just 10% of car users to alternate means free up a tremendous amount of space in the city.
I love my car, my bikes and my public transports and each does something nice for me - but seriously do you think cities like l.a. are even livable on a human scale - people don't even walk if the distance is over 1000meters.
I certainly agree with the idea of "uhm lets try to plan for otherthings than cars going forward"
My only success has been internal reviews of wcag compliance and the threat of fines if found lacking.
But for non digital design its embarrassingly hard
Now you CAN so it so that is not the case, but tbh i have never seen that in the wild -