Using Open Source Models

SalesGPT allows you to use open-source models instead of paid models like OpenAI's GPT. To do this, follow these steps:

Change the GPT_MODEL Environment Variable: Update the GPT_MODEL environment variable in your .env file to the desired open-source model. For example:
```
GPT_MODEL=your_open_source_model
```
Update the _streaming_generator Function: If you are using an open-source model, you may need to add an api_base argument to the _streaming_generator function in the agents.py script. This argument should be set to the inference URL of the model. If the model is deployed locally, use a localhost URL; otherwise, use any other working URL depending on the deployment.

Here is an example of how to modify the _streaming_generator function:
```
return self.sales_conversation_utterance_chain.llm.completion_with_retry(
    messages=messages,
    stop="<END_OF_TURN>",
    stream=True,
    model=self.model_name,
    api_base="your_inference_url"  # Add this line for open-source models
)
```

By following these steps, you can configure SalesGPT to use an open-source model for generating responses.

For more information on possible model deployments, you can visit: LiteLLM Providers Documentation