Hi, thanks for the great work. I am using the code to finetune the Phi3 with LoRA. I have saved intermediate checkpoints. The weights are like this, which are not the same as the provided model. Is it because the new version of transformers? Is it okay to use for inference?
Hi, thanks for the great work. I am using the code to finetune the Phi3 with LoRA. I have saved intermediate checkpoints. The weights are like this, which are not the same as the provided model. Is it because the new version of transformers? Is it okay to use for inference?
Thank you in advance:)