yujiepan/ernie-4.5-moe-tiny-random
Different architecture vs baidu/ERNIE-4.5-21B-A3B-Thinking. Different architecture: only 1 of 5492/214 tensors match by name and shape.
Different architecture vs baidu/ERNIE-4.5-21B-A3B-Thinking. Different architecture: only 1 of 5492/214 tensors match by name and shape.