Back

Another open-source model has arrived, and it’s even better than DeepSeek-V3. The Allen Institute for AI just introduced Tülu 3 (405B) 🐫, a post-training model that is a fine-tune of Llama 3.1 405B, which outperforms DeepSeek V3.

"Just fine-tuned LLaMA 3.2 using Apple's MLX framework and it was a breeze! The speed and simplicity were unmatched. Here's the LoRA command I used to kick off training: ``` python lora.py \ --train \ --model 'mistralai/Mistral-7B-Instruct-v0.2' \ -

Latest update to my Jarvis Today I have added some features to my jarvis to make it more special and more sophisticated Now my jarvis can do the linkdin post which is done with the help of making automation. Added few agents which means like thi

If I were to offer 1 GB of data at ₹5, with no expiry, and similar plans for higher data amounts at a hotspot location, would this model appeal to consumers? Is there a market need for such a service, where data affordability and flexibility are prio

Random Thought : I was wondering, why ChatGPT weren't build on the Increment Learning modle.. Because I might destroy it's algo.... Let me expain.. In the world of machine learning, training models can be approached in two main ways, Batch Lea

Entrepreneur myths nobody tells you: More sales ≠ less stress Funding won’t fix a broken model Hustle culture is a trap Most of your friends won’t get it Success is lonely sometimes, and that’s fine Learn the game. Play it smart.

I’ve always found naming things harder than building them. So I’m creating a tool that lets you generate names for anything — but with a twist. What sets it apart is the deep level of customizability and control. You can fine-tune everything — the to

I'm transitioning from full stack development to data engineering due to the needs of my upcoming project. I would like to learn more about the field and discover the best ways to master it. Can anyone guide me with a good roadmap, courses, and mat