[Question]: LocalAIChat How to set num_ctx in ollama

[dawn9551]

Describe your problem

ragflow configures local AI, and when in conversation, it ultimately calls the OpenAI() interface. I tracked the entire process and found that the num_ctx parameter was not set, causing ollama to be unable to handle long contexts

`class LocalAIChat(OllamaChat):

def __init__(self, key, model_name, base_url):
    if not base_url:
        raise ValueError("Local llm url cannot be None")
    if base_url.split("/")[-1] != "v1":
        base_url = os.path.join(base_url, "v1")
    self.client = OpenAI(api_key="empty", base_url=base_url)
    self.model_name = model_name.split("___")[0]`

[kamui-fin]

I'm having the same issue here. Unless I create a new modelfile with num_ctx set to a much higher number, it gives me this error in ollama:

source=prompt.go:77 msg="truncating input messages which exceed context length" truncated=2