-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
Hello everyone, I'm looking into putting RLHF into improving generative quality of the Stable Diffusion model. I'm aware that there are many publicly available reward models (such as link) for training LLMs with human preferences, but I don't know if there is any such model/dataset for images. Though I see some interests in this direction (e.g. pickapic and Aligning Text-to-Image Models using Human Feedback) Pickapic has released some data on huggingface but hasn't received an update since January.
Perhaps you could use an "aesthetic-evaluator": https://github.com/christophschuhmann/improved-aesthetic-predictor (probably not SoTA but it works)