I’ll do more tests on this deepshit thing but it feels like a chinese propaganda, unless they’ve discovered new laws of thermodynamics that we didn’t know about.
Discussion
But muh AGI!!!
Propaganda to kill propoganda 🤣🤣🤣🤣
The big impact of deepseek isn't their web application, its the fact that they open sourced everything (architecture, weights, training process). Why pay $200/mo for openAI when you can just download, finetune, and self-host your own version of deepseek?
Where and how they trained the model that’s what matters! I can give you my model and you can run it on your local
too but it doesn’t make it good
Right which is why the paper that came out at the same time describes all their training methods. They basically handed the whole world a "here's what you can do if you want to be as advanced as openAI's o1-pro models". There is no moat anymore for proprietary AI models or services.
But it all boils down to the hardware. You need h100s or similar high performance gpus to train and deploy models at that scale, which is still a significant barrier for most companies regardless of their methodologies or frameworks. It’s not the lack of knowledge for the most part but the hardware
True, and there are still lots of rumors going around about the actual training hardware for deepseek, but I still think open sourcing a SOTA model is a huge step forward. One of the things the community has been best at is reducing the compute cost for various models.
AI model wars are coming...
Fuck Nvidia
Fuck OpenAI
Deepseek looks to be open source