dosubot[bot]
commented
10 months ago 🤖
Hello @dianz17,
I'm here to assist you with your issue. Let's try to address the problems you're encountering.
Firstly, the error "Missing some input keys: {'input'}" suggests that the necessary 'input' key is not being provided in the dictionary when calling the agent. When calling the agent or agent executor, you need to pass a dictionary with an 'input' key that contains the text you want the agent to process. Here's an example:
result = agent_executor({"input": "hi, im bob"})Secondly, the error "Agent.plan() got multiple values for argument 'intermediate_steps'" indicates that the 'intermediate_steps' argument is being passed multiple times when calling the agent executor. The 'intermediate_steps' argument should only be passed once when creating the agent executor. Here's an example:
from langchain.agents import AgentExecutor
agent_executor = AgentExecutor(agent=agent, tools=tools, memory=memory, verbose=True,
return_intermediate_steps=True)After creating the agent executor, you can call it without passing the 'intermediate_steps' argument:
result = agent_executor({"input": "hi, im bob"})Regarding the use of ConversationBufferWindowMemory, it's used to keep a list of the interactions of the conversation over time. It only uses the last K interactions, which can be useful for keeping a sliding window of the most recent interactions. Here's an example of how to use it:
from langchain.memory import ConversationBufferWindowMemory
memory = ConversationBufferWindowMemory(k=5)
memory.save_context({"input": "hi"}, {"output": "whats up"})
memory.save_context({"input": "not much you"}, {"output": "not much"})
memory.load_memory_variables({})This will return the most recent interaction:
{'history': 'Human: not much you\nAI: not much'}You can also get the history as a list of messages by setting return_messages=True when initializing ConversationBufferWindowMemory.
If you're still encountering errors, please provide more details about your code and how you're using the agent and agent executor. I hope this helps! If you have any more questions, feel free to ask.
Best, Dosu
Sources
- Issue with input variables in conversational agents with memory
- docs/extras/use_cases/question_answering/how_to/conversational_retrieval_agents.ipynb
- docs/docs/use_cases/question_answering/conversational_retrieval_agents.ipynb
- docs/docs/modules/memory/types/buffer_window.mdx
- libs/langchain/langchain/memory/buffer_window.py
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.
Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.
If you want to continue the conversation, start your reply with @dosu-bot.
dianz17
failable
Issue you'd like to raise.
Im trying to use an agent as a router for two retrievals. I also want it to have memory so im using the conversational agent with two RetrievalQA chains and ConversationBufferWindowMemory.
Heres the code after creating the retrievals and chatTemplates:
Now, I've been trying two things. The first one is to call my agent with the memory directly. I tried it with and without using .run():
but with some questions, not all, i get the following error:
I also tried calling the agent like this:
and got the same error.
I tried to use an agent executor, moving the memory from the agent to the executor but I get the following error:
Im not sure what im doing wrong. @dosu-bot
Suggestion: