Streaming Output Issues from Llama 2 deployed in Sagemaker Studio

I am unable to parse -the streaming output from the Llama2-7b-chat deployed on Sagemaker Jumpstart.

I have written a custom class to parse each event from the stream. However, the entire output is returned at once, instead of bytes array. The code snippet is as shown below:

client = boto3.client("sagemaker-runtime")
request_body = {
        "inputs":  
          [
            [
             {"role": "system", "content": "You are a bot designed to answer queries from user"},
             {"role": "user", "content": "What are the must-visit attractions in Florida?"},
            ]   
          ],
        "parameters":{
            "max_new_tokens":512,
            "return_full_text": False
        },
    }

response = client.invoke_endpoint_with_response_stream(
                EndpointName=endpoint_name, 
                Body=json.dumps(request_body), 
                ContentType="application/json",
                CustomAttributes="accept_eula=true"
        )
       
        event_stream = response['Body']
        for event in event_stream:
            resp= event['PayloadPart']["Bytes"]
           data_str = resp.decode('utf-8')[1:-1].replace("}","")
           content_start = data_str.find('"content":')
            if content_start != -1:
                content = data_str[content_start + len('"content":'):].strip()
                print(content)
            else:
                print(data_str)