2damoon

I'm a seasoned engineering graduate with a deep passionate interest for AI and IT infrastructure. I'm the maintainer of afm (maclocal-api) and Vesta AI Explorer

[2damoon @scouzi]

5 inference backend in 1 single app. Access to all models and methods. For MacOS. Control Qwen Thinking and chose LLM or VLM.

https://t.co/AOSbxCh5PT

https://t.co/Z5NnsZPW08

#llm #apple #LocalAI #Qwen3

[2damoon @scouzi]

@atanu__07 Let's connect and build it! https://t.co/E7EYX5UUZL

[2damoon @scouzi]

@ZunnuMetaX Yes I do!

[2damoon @scouzi]

@ChainChaserVN Let's connect and build! https://t.co/E7EYX5UUZL

[2damoon @scouzi]

@james84_ Let's connect and build! https://t.co/E7EYX5UUZL

[2damoon @scouzi]

@james84_ Let's connect and build. https://t.co/E7EYX5UUZL

[2damoon @scouzi]

@kevinxu It makes IT management more complicated than it should be.

[2damoon @scouzi]

@icanvardar The trend is modular monolith now. Microservices introduce inter-service API latency, operational complexity, increased Cloud costs and wasted CPU cycles due to overhread burden of distributed systems.

[2damoon @scouzi]

@ivanfioravanti Priced an 128 GB MacBook pro 2TB yesterday. Max Cpu/gpu with nano texture. Went up about $1300 CAD from what I remember.

[2damoon @scouzi]

If there is such a demand, why not sell them at a premium to the local Chinese AI providers? It does not make a lot of sense.

[Greg Kamradt @GregKamradt]

obvious in retrospect but I had no idea there was a black market for tokens

[2damoon @scouzi]

@FirstSquawk Qwen must be getting too much attention

[2damoon @scouzi]

@danielnewmanUV It will make more net profit than the stock price a year ago. You were buying a foward PE of 1 without knowing it.

[2damoon @scouzi]

@vulcanbench I like this repo. I'm applying against local opsen source models on Mac.

[2damoon @scouzi]

@DudeWhoInvests caught it perfectly overnight

[2damoon @scouzi]

@coryalthoff https://t.co/E7EYX5UUZL

[2damoon @scouzi]

I recommend VulcanBench. Awesome.

Love @vulcanbench . Thank you! I'm finding latent bugs in my mlx inference engine.

https://t.co/E7EYX5UUZL

[VulcanBench @vulcanbench]

Okay, it's official, VulcanBench v0.1 is here.

Free and open source llm benchmarking tool, pre-loaded with 52 tests so you can really go deep.

The goal is simple: completely transparent benchmarking, every detail from the code that runs the tests, to the tests themselves, is available, for free, for everyone to see.

Please test, test, and test some more, and give feedback, or even better, submit PRs to make it better.

Live long and Benchmark 🖖

[2damoon @scouzi]

@charliebilello The craziest roundtrip ever is $CVNA

[2damoon @scouzi]

@LuxAlgo Qwen getting too much love

[2damoon @scouzi]

@Mr_Derivatives Why didn't you buy it at $100 a year ago?

[2damoon @scouzi]

Be honest:

How many of you local AI enthusiasts actually look at model output quality vs pure tokens/sec?

I'll go first.

When benchmarking, I rarely look the at the output. But I decided to put more emphasis on this. In the end, that's what really matters!

[2damoon @scouzi]

@TeksEdge That's probably why $BABA stock is tanking lately.