huluhuluu/llama-3.2-3b-instruct-eagle3-sharegpt-sw512-epoch9-step231810
Different architecture vs meta-llama/Llama-3.2-3B-Instruct. Different architecture: only 1 of 254/15 tensors match by name and shape.
Different architecture vs meta-llama/Llama-3.2-3B-Instruct. Different architecture: only 1 of 254/15 tensors match by name and shape.