Module custom_float::google
source · Type Aliases§
- bfloat8 (brain float), with 8-bits using the exponent of FP16
- bfloat16 (brain float), with 16-bits using the exponent of FP32
- bfloat32 (brain float), with 32-bits using the exponent of FP64
- bfloat64 (brain float), with 64-bits using the exponent of FP128
- bfloat128 (brain float), with 128-bits using the exponent of FP256