DeepSpeed-MII
DeepSpeed-MII copied to clipboard
Issues with llama 2 model example
I'm running examples with NousResearch/Llama-2-7b-chat-hf HuggingFace repo and output of the model is gibberish and broken. What can be an issue? Thanks!
Hi @ttim I have confirmed that I can replicate this error. I will share an update when we have a solution. Thanks