Class GGMLFileType

utils/convert.py:177–192

class GGMLFileType(enum.IntEnum):
    AllF32     = 0
    MostlyF16  = 1  # except 1d tensors
    MostlyI2   = 2  # except 1d tensors
    MostlyQ8_0 = 7  # except 1d tensors

    def type_for_tensor(self, name: str, tensor: LazyTensor) -> DataType:
        dt = GGML_FILE_TYPE_TO_DATA_TYPE.get(self)
        if dt is None:
            raise ValueError(self)
        # Convert all 1D tensors to F32.  Most of the codebase that takes in 1D tensors only handles F32 tensors, and most of the outputs tensors are F32.
        #  Also The 1d tensors aren't much of a performance/size issue.  So instead of having to have separate F32 and F16 implementations of both, just convert everything to F32 for now.
        dt = dt if len(tensor.shape) > 1 else DT_F32
        if name == "token_embd.weight" or name == "output.weight":
            dt = DT_F32
        return dt


GGML_FILE_TYPE_TO_DATA_TYPE: dict[GGMLFileType, DataType] = {