Hugging Face转发了
?? Cerebras' crazy inference speed is now available for everyone on?Hugging Face. Generating thousands of samples in under 5 minutes is now a reality. Here I generate 100 rows in just 10 seconds with Llama70B using Cerebras Systems as Inference Provider. Do the test yourself here: https://lnkd.in/d6aS9NGY
Daniel Vila Suero - is there a list with supported models or do I have to watch for it above the model card? At Cerebras‘ Playground there are only a few supported models…
Is this an example for generating synth data sets?
The speed of AI solution processing is truly transformative. ?? With technological advances like Cerebras, AI systems are becoming faster and more efficient. ?? At qantum.one, we add to this efficiency by ensuring these systems are reliable and work as expected through our Bionic Testing approach. These are exciting times for AI! ?? #AI #InferenceSpeed #Testing #Cerebras #BionicTesting #QaAutomation #qantumone
I really like that there are so many options for Inference providers! It would be helpful to see metrics such as tokens/sec or cost/token next to them.
Cerebras is excited to partner with ??. Unleash the ideas.
What’s the cost of running on it though?
Building @ Hugging Face ??
1 天前Learn about Inference Providers: https://huggingface.co/blog/inference-providers