But, the numbers are real. Training Stable Diffusion in a cleaner region could’ve saved over 15,000 kg CO₂e and around $150k.
Where we train models matters more than ever, not just for the planet, but for your bottom line too.
I want to explore how we can shift certain compute to the lowest CO2 regions, saving money and CO2 along the way.
---
Would love to hear your thoughts, especially if you've made region-level decisions for training infrastructure. I know it’s rare to find devs with hands-on experience here, but if you're one of them, your insights would be gold.
But, the numbers are real. Training Stable Diffusion in a cleaner region could’ve saved over 15,000 kg CO₂e and around $150k.
Where we train models matters more than ever, not just for the planet, but for your bottom line too.
I want to explore how we can shift certain compute to the lowest CO2 regions, saving money and CO2 along the way.
---
Would love to hear your thoughts, especially if you've made region-level decisions for training infrastructure. I know it’s rare to find devs with hands-on experience here, but if you're one of them, your insights would be gold.