I'm amazed by this. Llama 3 on Groq makes GPT-4 look like a grandpa: I asked both models to list all the prime numbers from 1 to 1000. Llama 3 hit over 830 tokens per second(!) Plus, it generated the entire sequence again whilst GPT-4 was still inferencing. In an arena as competitive as LLMs, speeds like this really set you apart and open up hundreds of use cases. You must give it a go. If you enjoy insights like this, follow me Alex Banks for more on AI.
Groq is insanely fast, but I wonder chatGPT being slow because of too many users connected to the server at a time. It is a reason ?
While it’s not search, Google trained us to expect near instantaneous responses. Speed will become a competitive advantage I suspect.
That's impressive Llama 3 seems like a speed demon in generating prime numbers. The competition in LLMs must be intense Will definitely follow Alex Banks for more AI insights. Alex Banks
Impressive speed and efficiency with Llama 3, truly groundbreaking Alex Banks
It's like lightning, and groq is awesome.
Insanely powerful!! AI is improving like crazy.
That's impressive speed from Llama 3 compared to GPT-4! AI advancements are fascinating. Alex Banks
I agree that Groq is mind-blowing. But is this comparison fair? ChatGPT started the analyzing mode, which typically means that it starts writing down the content in Python code. So even though it is hidden to the user (you can click on the "Analyzing" widget to open the code being generated, where you see that it is actually still okay-ish in speed), it still generates the prime numbers. Again, I totally agree, Groq is way faster with Llama and all, but i think in this comparison, it's a bit "unfair" since ChatGPT is generating the prime numbers twice: once in the Python code, and once in the answer.
This is not what an LLM is supposed to do. Neither it is possible to infer prime numbers by learning any initial segment thereof.
Head of Product/Chief Technology Officer @ Concord Technologies | Delivering Healthcare Solutions
4moWhat’s wrong with being a grandpa?