huluhuluu/qwen3-1.7b-eagle3-sharegpt-sw512-epoch9-step463620
Different architecture vs Qwen/Qwen3-1.7B. Different architecture: only 1 of 311/15 tensors match by name and shape.
Different architecture vs Qwen/Qwen3-1.7B. Different architecture: only 1 of 311/15 tensors match by name and shape.