Batch Transform Issue

We are working on a use case where in we use PipelineModel and Batch Transform API.In the Pipeline model,we have two models:The first model being SKLearn column transformer model and the second model being xgboost model.The issue is when we are performing batch transform we get this error for the xgboost container : "Request Entity Too Large" . The catch is we can set Max Payload but that would be set for the first model container(Sklearn container in our case) in our Pipeline model .The output of the first model clearly exceeds the payload size for xgboost container, but there is no way for use to handle that.

Start with a minimal PipelineModel containing an SKLearn column transformer and an XGBoost model, then run a Batch Transform request that reproduces the “Request Entity Too Large” error. Investigate how Max Payload is applied across the pipeline and determine whether per-container payload handling is supported; done means identifying a verified configuration or clearly documenting the limitation.

Written by the indexing model from the issue text.