This is the PR with the changes in case people missed it:
https://github.com/mieciu/tau2-bench/pull/1/files
That seems so strongly directed, that it feels like an attempt to reproduce a classic chat bot.
Can one customer get the model to return the bill details for another customer?
Thanks! I also updated the post with the link on the website.
That seems so strongly directed, that it feels like an attempt to reproduce a classic chat bot.