🤗 Serve any model with Inference Endpoints + Custom Handlers
•
3
distilabel
, so we implemented PrometheusEval
.PrometheusEval
running their 7B variant with vLLM in a single L40 on top of
HuggingFaceH4/instruction-dataset, we got the 327 existing prompt-completion pairs evaluated and pushed to the Hub in less than 2 minutes!