Mehran Kiani

Today I started an open source ML project on top of 4diac forte. See description for more info. @AZoitl github.com/MehranKianiHub…

Started a coaching session on "marketing and brand fundamentals"

@GithubProjects Hii :) github.com/MehranKianiHub

-▬▬.◙.▬▬‐ ▂▄▄▓▄▄▂ ◢◤ █▀▀████▄▄▄◢◤ █▄ █ █▄ ███▀▀▀▀▀▀╬ ◥█████◤ ══╩══╩══ ╬═╬ ╬═╬ Just dropped down to say ╬═╬ Share Your GitHub Profile ╬═╬ And Let's Connect ╬═╬ ╬═╬ ☻/ ╬═╬/▌ ╬═╬/ \

Gonna try @opencode today

Please take a look at my new open source project here:

Please take a look at my new open source project here:

WhirDetective: An Open Source Machine Learning Component for Industrial Bearing Fault Diagnostics github.com/MehranKianiHub…

In 5 years from now, you won’t care how long it took, you’ll just be glad you started.

I started this open source ML project recently. I’ll become super happy to hear your suggestions 🙏

I started this open source ML project recently. I’ll become super happy to hear your suggestions 🙏

I really enjoyed reading this article 👌

Learn effectively @DrJustinSung youtu.be/6U8zNlqCa6M?si…

Happy to see that I'm the basis for such an interesting work! ☺️

@4diac You’re an amazing project @4diac

@AZoitl Thanks! 😊 I will submit a news post there.

@Mehran__Kiani This is very cool. Feel free to also submit a guest news post for our news blog.

Today I started an open source ML project on top of 4diac forte. See description for more info. @AZoitl github.com/MehranKianiHub…

@jehniiee Yes I will :)

@0xKusha what do you use?

Be honest, are you still using vscode ?

@Mkpareek19_ it's d but still struggling with a 🤣🤣

What's your X growth goal for this month? a) First 100 followers b) First 500 followers c) First 1,000 followers d) First 2,000+ followers

@kargarisaac @Meta How do you compare it with Chinese open source models like deepseek?

. @Meta's next model Avocado apparently beats top open source models at the pretraining stage alone, with 10x efficiency over Maverick and 100x over Behemoth. The numbers sound impressive but all of this is coming from an internal memo with zero public benchmarks and no model size disclosed. We have heard "most capable model ever" before from Meta and then Llama 4 happened. What actually matters to me as someone who runs models locally and builds agent systems: give me a good sub 100B model that is strong at tool calling. That is what practitioners need. Not another frontier model I can only access through an API. And this is the part that annoys me. There are rumors that Avocado might not be open source at all. Meta was the company that made open source LLMs a real thing. Llama 3 was genuinely good and I think many of us built serious workflows around it. If they pivot to closed source now, it does not matter how efficient the model is if I cannot run it on my own hardware and finetune it for my specific use cases. I am not against closed models in general. If the pricing is competitive with what Chinese labs are offering and the quality is there, I will try it. But Meta going proprietary would leave a real gap in the open source ecosystem that someone else will have to fill. Llama 3 was great. Llama 4 was a disappointment. I hope Avocado is Meta getting back on track, but I want to see actual benchmarks and a clear answer on open source before getting excited.

Watch this video 10 times (Nvidia CEO, Jensen Huang)