You have it backwards. An int8_t takes only one byte. A float is always a multi-byte type.
int8_t
float