A ChatGPTBot() would be created for the models in the litellm support list, while in the following else block a LiteLLMChatGPTBot() is going to be created for model that's not in its support list.
Actually it causes problem when the model is not supported in the given version of litellm. For instance, If you set model to gpt-4-turbo-preview in config and it's going to completely break the bot, since it's going to create a LiteLLMChatGPTBot() which doesn't support that model.
It probably makes sense to change the line to elif model not in litellm.open_ai_chat_completion_models
BTW, the current litellm dep is really out of date and might need either upgrade or deprecation
Repro steps
Set config model to gpt-4-turbo-preview which should be supported by ChatGPTBot
Start the bot and see it breaks from LiteLLMChatGPTBot
Search for answers in existing issues
Python version
python 3.10
Issue description
I am not sure if it's a bug or intended change, but this line doesn't look right: https://github.com/iuiaoin/wechat-gptbot/blob/main/bot/bot.py#L18
A
ChatGPTBot()
would be created for the models in thelitellm
support list, while in the followingelse
block aLiteLLMChatGPTBot()
is going to be created for model that's not in its support list.Actually it causes problem when the model is not supported in the given version of litellm. For instance, If you set model to
gpt-4-turbo-preview
in config and it's going to completely break the bot, since it's going to create aLiteLLMChatGPTBot()
which doesn't support that model.It probably makes sense to change the line to
elif model not in litellm.open_ai_chat_completion_models
BTW, the current litellm dep is really out of date and might need either upgrade or deprecation
Repro steps
gpt-4-turbo-preview
which should be supported by ChatGPTBotRelevant log output
No response