[OpenFunctions] Unnecessarily and incorrectly invoking functions about gorilla HOT 1 OPEN

Hey Keith! Thank you very much for trying our Gorilla OpenFunctions model.

We have read your inputs & outputs. Here are some corrections and suggestions to have it output better result:

1. The messages variable defines the natural language prompt that you come up with. For example, if you're using the get_current_weather function, a prompt like "What's the weather like in SF?" will make the most sense for the model to assemble the function call. Here's an example setup:

messages = [{"role": "user", "content": "What's the weather like in SF?"}]
functions = [
    {
        "name": "get_current_weather",
        "description": "Get the current weather in a given location",
        "parameters": {
            "type": "object",
            "properties": {
                "location": {
                    "type": "string",
                    "description": "The city and state, e.g. San Francisco, CA. State must be included",
                },
                "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]},
            },
            "required": ["location"],
        },
    }
]

2. If you are experimenting with normal function calling, you should stick with v0 as it provides more stable experience. V1 is specifically designed for parallel function calling and we are making effort to actively improve that model.
3. The motivation behind OpenFunctions is making function calls based on function documentation and natural language instruction. As this model is specifically designed for function calling, we don’t know how well it might behave for general purpose chatting :)
4. In case you are not aware, we have a sample Jupyter notebook https://colab.research.google.com/drive/16M5J2H9F8YQora_W2PDnp120slZH-Mqd#scrollTo=eBd_fso7qFPX where we provided several examples and instructions to follow.
   Again, thank you for using our model and let me know here if you need help!

from gorilla.