Can i run LLM with pyspark

I have been trying to run LLM with pyspark as we have a dataset of 1.2 Million texts, but have been failing due to pipeline and base types errors

LLM being used:

Error:
bases and types should match