Spaces:
Sleeping
Sleeping
pwilczewski
commited on
Commit
·
13ab127
1
Parent(s):
3d1690c
updating prompts
Browse files
app.py
CHANGED
@@ -28,7 +28,8 @@ system_prompt = """You are working with a pandas dataframe in Python. The name o
|
|
28 |
</df>
|
29 |
|
30 |
You are not meant to use only these rows to answer questions - they are meant as a way of telling you about the shape and schema of the dataframe.
|
31 |
-
You also do not have use only the information here to answer questions - you can run intermediate queries to do exporatory data analysis to give you more information as needed.
|
|
|
32 |
system_prompt = system_prompt.format(dhead=df.head().to_markdown())
|
33 |
|
34 |
# The agent state is the input to each node in the graph
|
@@ -59,7 +60,7 @@ def create_agent(llm: ChatOpenAI, tools: list, task: str):
|
|
59 |
# AIMessage will have all kinds of metadata, so treat it all as HumanMessage I suppose?
|
60 |
def agent_node(state: AgentState, agent, name):
|
61 |
result = agent.invoke(state)
|
62 |
-
return {"messages": [HumanMessage(content=result["output"], name=name)]}
|
63 |
# return {"messages": [result]}
|
64 |
|
65 |
# I need to write the message to state here? or is that handled automatically?
|
|
|
28 |
</df>
|
29 |
|
30 |
You are not meant to use only these rows to answer questions - they are meant as a way of telling you about the shape and schema of the dataframe.
|
31 |
+
You also do not have use only the information here to answer questions - you can run intermediate queries to do exporatory data analysis to give you more information as needed.
|
32 |
+
Do not ask if the user wants additional information or analysis."""
|
33 |
system_prompt = system_prompt.format(dhead=df.head().to_markdown())
|
34 |
|
35 |
# The agent state is the input to each node in the graph
|
|
|
60 |
# AIMessage will have all kinds of metadata, so treat it all as HumanMessage I suppose?
|
61 |
def agent_node(state: AgentState, agent, name):
|
62 |
result = agent.invoke(state)
|
63 |
+
return {"messages": [HumanMessage(content=name + ": " + result["output"], name=name)]}
|
64 |
# return {"messages": [result]}
|
65 |
|
66 |
# I need to write the message to state here? or is that handled automatically?
|