Spaces:
Running
Interesting use case for using the crew platform
Hi,
I took the liberty to look under the hood and it seems you used crew.ai to orchestrate multiple agents to get this going - quite cool.
I am curious about the use of sambanova api. Was there anything in particular you used there that you couldn't have done with another hosted LLM provider?
Thanks.
@llmchamp this example, leveraging crew.ai, was written by a sambanova employee as a demonstration of how you can use crew,ai with sambanova.api. We welcome you to leverage our api for your projects as well .
-Coby
sambanova customer support
@llmchamp Thanks for your comments and trying this app. Sambanova api help this agentic ai use case because the api delivers much faster throughput (tokens/sec) than what other GPU based AaPIS do. While the same logic could easily be re-wired to other OpenAI API compatible providers (Like Together.ai, fireworks.ai)... it would take much longer time to plan the itinerary and get a result. As an example, with the Llama 3.3 70B model, the app usually completes building a travel itinerary in 30 seconds. The same on a GPU based provider would take over a minute. You are welcome to download the source code and try it out, or use Sambanova APIs for any other projects. Please keep me updated on how it goes for you.
Thanks for the quick explanation - and it all makes sense now.
And thanks for clarifying that sambanova is different (and efficient) from the other providers as they use their custom chip for inference.
https://sambanova.ai/blog/sn40l-chip-best-inference-solution