No. Comparing the emissions of different models trained in different countries is what would actually be misleading. In this case it would falsely suggest that training a BLOOM model emits less CO2 than training a LLaMA, while this is the opposite.

→ View original post on X — @guillaumelample,