When I run the "Single Agent React Summary" code and set the parameters n and best_of in the llama.yaml file to values greater than 1, I encounter the following error:
ValueError: All stacks are empty, so the only token accepted is EOS(128009), but got 1641. This error is likely due to the previous token not being accepted by the grammar.
An error occurred while running the episode: All stacks are empty, so the only token accepted is EOS(128009), but got 1641. This error is likely due to the previous token not being accepted by the grammar.
How can I resolve this?
When I run the "Single Agent React Summary" code and set the parameters n and best_of in the llama.yaml file to values greater than 1, I encounter the following error:
ValueError: All stacks are empty, so the only token accepted is EOS(128009), but got 1641. This error is likely due to the previous token not being accepted by the grammar.
An error occurred while running the episode: All stacks are empty, so the only token accepted is EOS(128009), but got 1641. This error is likely due to the previous token not being accepted by the grammar.
How can I resolve this?