Finally, 'The Judge' is here! Evaluating agents is no small feat. Can't wait to dive into the golden dataset and your strategies. Off to check the notebook now!
Very interesting, thank you! Could you please also make a video about the ideal workflow of building large datasets from production usage, in order to iterate after release + build datasets for fine-tuning models and prompts optimization (when using dspy). I saw that Langgraph studio can add examples to datasets easily, but it's not super clear how to do that in practice. I love the tooling around Langgraph, but this part around maintaining and improving an agent after release lacks tutorials
honestly it's better to use the development server with the web UI. I have mac os but the setup for the desktop app with docker is a nightmare if you don't start with one of their template
watched the whole thing :) Big like for the Agent example. Thanks Dave
Finally !!!
A well defined and detailed professional code for customer service agent
Awesome! Thanks!
Need more videos (and text tutorials) like this, in customer needs 🔥
Finally, 'The Judge' is here! Evaluating agents is no small feat. Can't wait to dive into the golden dataset and your strategies. Off to check the notebook now!
Very interesting, thank you!
Could you please also make a video about the ideal workflow of building large datasets from production usage, in order to iterate after release + build datasets for fine-tuning models and prompts optimization (when using dspy). I saw that Langgraph studio can add examples to datasets easily, but it's not super clear how to do that in practice.
I love the tooling around Langgraph, but this part around maintaining and improving an agent after release lacks tutorials
That's nice, but when will LangGraph Studio be available for anything besides Mac OS?
you can use docker, through the langgraph cli
honestly it's better to use the development server with the web UI. I have mac os but the setup for the desktop app with docker is a nightmare if you don't start with one of their template