Shorter or more descriptive?

[estarq]

Which one should I commit?

for i in range(0, len(data), 2):
  decimal_number = int(data[i:i + 2], 16)
  binary_number = bin(decimal_number)[2:]
  byte = binary_number.zfill(8)
  self.rom[int(first_byte_address, 16) + i // 2] = byte

or

for i in range(0, len(data), 2):
  self.rom[int(first_byte_address, 16) + i // 2] = f'{int(data[i:i + 2], 16):08b}'

[Gribouillis]

The second one looks better, but here is another possibility

base = int(first_byte_address, 16)
for k, s in enumerate((data[i:i+2] for i in range(0, len(data), 2)), base):
    rom[k] = f'{int(s, 16):08b}'

[DeaD_EyE]

I would use binacsii.unhexlify to convert the str or bytes hex representation into bytes.

Then using the fact that bytes yields integers if you iterate over it.
Same with index access.

import binascii


def to_bin(data: str) -> list[str]:

    data = binascii.unhexlify(data)

    def transformer():
        for char in data:
            yield f"{char:08b}"

    return list(transformer())

I made a closure, because it looks nicer.
The difference is here, that noting is mutated.
You get a new list.