Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

>We are giving the enjoyable parts of life to a computer. And we are left with the drudgery.

Yesterday I asked a local llm to write a python script to have a several multimodal llms rank 50,000 images generated by a stable diffusion model. I then used those images to train a new checkpoint for the model and can now repeat the process ad infinitum.

In the olden days of 2020 I would have had to hire 5000 people each working for a day to do the same.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: