create new notebook example #74
Conversation
uses custom chain
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
| "cell_type": "code", | ||
| "source": [ | ||
| "for i, prompt in enumerate(inputs):\n", | ||
| " if prompt[\"role\"] == \"system\":\n", |
There was a problem hiding this comment.
Can you fix this? And include the entire history below?
| " tool_output = get_delivery_date(order_id)\n", | ||
| " wf.add_tool(\n", | ||
| " input=response.choices[0].message.tool_calls[0].function.arguments,\n", | ||
| " output=f\"Your delivery date is {tool_output}.\",\n", |
There was a problem hiding this comment.
I would log the 'output' of the tool as-is (without putting into a "Your delivery date is..." string).
There was a problem hiding this comment.
Then separately, if this ""Your delivery date is {tool_output}."" is what you're showing to the user, log that as a step
| " )\n", | ||
| "\n", | ||
| " # On LLM steps, also run Protect\n", | ||
| " if output_message.content is not None:\n", |
There was a problem hiding this comment.
There's an if output_message.content is not None already. Combine the two? Or re-arrange the code?
| " )\n", | ||
| "\n", | ||
| " # Conclude the workflow\n", | ||
| " wf.conclude(output={\"output\": output_message.content})\n", |
There was a problem hiding this comment.
You're not using the output of the tool, or the output of protect here.
| "\n", | ||
| " # On LLM steps, also run Protect\n", | ||
| " if output_message.content is not None:\n", | ||
| " response_protect = protect_galileo(prompt[\"content\"], output_message.content)\n", |
There was a problem hiding this comment.
Do something with the response? e.g. if protect triggers, alter the response?
| "source": [ | ||
| "# Conversational Flow\n", | ||
| "inputs = [\n", | ||
| "{\"role\": \"system\", \"content\": \"You are a helpful customer support assistant. Use the supplied tools to assist the user.\"},\n", |
There was a problem hiding this comment.
I would not include the system message as an input here. it's not an input from 'the user'. it's an instruction / your prompt template.
uses custom chain, per feedback add as an example and link out instead of in the docs https://github.com/rungalileo/docs-mintlify/pull/86