langchain
langchain copied to clipboard
community: add `tools` support to ChatTongyi(can be used with `create_openai_tools_agent`)
Description
Tongyi now supports "tools"(Official Docs), and the data structure is compatible with OpenAI tools, therefore it's possible now to use create_openai_tools_agent
along with ChatTongyi.
Ex:
from langchain_community.chat_models.tongyi import ChatTongyi
from langchain.agents import AgentExecutor, create_openai_tools_agent
from langchain import hub
from your_tools import your_tool, another_tool
prompt = hub.pull('hwchase17/openai-tools-agent')
tools = [your_tool, another_tool]
llm = ChatTongyi(temperature=0)
agent = create_openai_tools_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent = agent, tools = tools, verbose = True, handle_parsing_errors=True)
response = agent_executor.invoke(
{"input": "Your question."}
)
print(response)
I tried the above example in my project, it works fine.
I have run the original integrations tests and they all passed.
To actually test the tools feature itself, I would need to import create_openai_tools_agent
from langchain.agents
, which is not a dependency of the community
package. So I think I should leave the original test as it is.
Dependencies
No dependency is required for this change
The latest updates on your projects. Learn more about Vercel for Git ↗︎
1 Ignored Deployment
Name | Status | Preview | Comments | Updated (UTC) |
---|---|---|---|---|
langchain | ⬜️ Ignored (Inspect) | Visit Preview | Apr 22, 2024 1:55am |
@eyurtsev Would you mind take a look at this ?