The community members are discussing issues with pickling in Llama Index, specifically the error "cannot pickle 'builtins.CoreBPE' object". One community member shared a link to an issue thread on GitHub, and another community member noted that pickling may not work due to the tiktoken tokenizer being attached to many objects. The community members are exploring potential workarounds, such as removing the tiktoken tokenizer from the node parser and prompt helper, but a simple solution has not been found yet.
