m1keil

I actually do like it a lot.

Small bug report: on mobile layout, the like/dislike button is getting a bit clipped and the watch later button seems a bit off too.

Here's a delicious meal we prepared for you. We don't recommend you going to the kitchen and trying to prepare it all over again.

..but you can!

Nothing controversial here.

The update experience in Ghostty is great. I think it takes under 5 seconds from the moment you click accept update and until the new updated terminal is running.

I never had an iPod and only borrowed it a few times from friends. It is quite amazing how they manage to nail down this interface which still feels good even in the day and age of touch screens.

My 2021 car has sort of similar scrolling wheel to control the infotainment (which has no touch) and it's nowhere near this level of polish.

I assume this type of application is the new JavaScript frameworks for the near future.

Have you considered there are other countries in the world?

Have you considered that not all places have the emotional baggage like your Atlanta?

Agree. I understand the rational but having a quick way to "deobfescate" the brand and model would help.

For now..

Waiting for QuackDB the OSS spin off any second now

huh.. I'm a bit of local LLM noob so I wasn't familiar with mlx_vlm.

I gave it a shot now:

mlx_vlm.generate --model mlx-community/Qwen3.8-27B-4bit --prompt 'give me fizz buzz in rust' --enable-thinking --draft-kind mtp --draft-model mlx-community/Qwen3.8-27B-MTP-4bit --verbose

==========

Prompt: 58 tokens, 90.717 tokens-per-sec Generation: 145 tokens, 36.392 tokens-per-sec Peak memory: 17.419 GB Speculative decoding: 2.79 accepted tokens/round (1.79 accepted drafts/round, 89.4% of drafted, avg draft 2.00) over 52 rounds

Which is very close to ollama, thank you!

I'm not sure if I can get rid of the drafter model, if I understand correctly, the Qwen model already includes a built in draft headers, but just having --draft-kind mtp results in about 17 t/s.

I have a 36gb M3 Max. I tested it across quite a few different options: llama.cpp, oLMX, ollama with different options.

So far ollama managed to be the most performant of them all. I will get 30 to 40 tokes/sec with it when using the -mlx version of Qwen3.8.

Whatever the sauce the ollama folks baked into the mlx + MTP mix is currently working the best out of the box.

Can be easily the price you will pay across Australia, depending on the time of day.

The problem is it's not the engineers that overthink it. The requirement for soc2 usually comes with the first "serious" customer. It is usually a big blocker on some fat contract and now the business makes it your problem for the next 6 months.

So what do you do? You engage and some 3rd party 1800-need-soc2 clowns which will hold your hand and implement all the cookie cutter solutions they know will make auditor happy (oh and btw, they know the auditor personally).