It is difficult to recommend an ideal web tool to assess the performance of solution without knowing anything about your ML model, parameters, hypothesis and etc. If you are not using any well known/commonly used benchmark datasets, it is not likely you can conduct a reliable comparison experiment.
this may give you a better idea how to benchmark your ML solution : https://www.slideshare.net/papisdotio/machine-learning-services-benchmark-ins-almeida-papis-connect
It is difficult to recommend an ideal web tool to assess the performance of solution without knowing anything about your ML model, parameters, hypothesis and etc. If you are not using any well known/commonly used benchmark datasets, it is not likely you can conduct a reliable comparison experiment.
this may give you a better idea how to benchmark your ML solution : https://www.slideshare.net/papisdotio/machine-learning-services-benchmark-ins-almeida-papis-connect
A standard data set is always contextual based on the problem. One best place that I discover while my research work is Kaggle and it is open data source : https://www.kaggle.com/datasets