Merge checkpoint adaptor weights with model

[sidracha]

My fine tuning was interrupted on a 4-bit quantized model. How can I merge the weights of a checkpoint with the base model to use for inference? I couldn't find a way to do it anywhere else