kgrabko/JiRackTernary_405b
Different architecture vs meta-llama/Llama-3.1-405B. Different architecture: only 255 of 1137/2901 tensors match by name and shape.
Different architecture vs meta-llama/Llama-3.1-405B. Different architecture: only 255 of 1137/2901 tensors match by name and shape.