Agent.load unable to maintain slots

Hello,

i’m using rasa 1.4.5 , updating model on runtime on every message to handle multi-agent concept. i have used Agent.load(model_path, action_endpoint) to update model. but it is unable to maintain SLOTS on every hit. creating empty slots on every load model.

INPUT

                # action_endpoint
                action_endpoint = EndpointConfig(url=action_server)

                # load_model
                global agent
                agent = Agent.load(model_path, action_endpoint=action_endpoint)
                out = await parse(text)

Handling Message

async def parse(text: Text):
    global agent
    data = await agent.handle_text(text)
    return data

Output

first_hit

slot_one: test_one
slot_two: None

second_hit

slot_one: None
slot_two: test_two

can you guys please help me

What do you mean by that?

For me it seems like something is wrong with the tracker store or with your stories :thinking:

i’m using custom slots and custom actions

when i hit api from front end agent.load loads latest model of the selected bot, and it removes stored slots. i want to store slots

loaded all models on starting of rasa run, it’s working