Conversation Evals
Query Resolution
Conversation Evals
Query Resolution
Evaluates the ability of the LLM to resolve the user’s query.
A good AI assistant should be able to effectively address the user’s query. Query resolution evaluates the ability of the AI assistant to resolve the user’s query effectively.
How to use it?
from uptrain import EvalLLM, QueryResolution
OPENAI_API_KEY = "sk-********************" # Insert your OpenAI key here
data = [{
'conversation' : [
{"role": "patient", "content": "Help"},
{"role": "nurse", "content": "what do you need"},
{"role": "patient", "content": "Having chest pain"},
{"role": "nurse", "content": "Sorry, I am not sure what that means"},
{"role": "patient", "content": "You don't understand. Do something! I am having severe pain in my chest"}
]
}]
eval_llm = EvalLLM(openai_api_key=OPENAI_API_KEY)
res = eval_llm.evaluate(
data=data,
checks=[QueryResolution(user_persona="patient", llm_persona="nurse")],
)
By default, we are using GPT 3.5 Turbo. If you want to use a different model, check out this tutorial.
Sample Response:
{
"score_query_resolution": 0.0,
"explanation_query_resolution": "The AI assistant failed to effectively address the patient's urgent concern of chest pain, which could indicate a potential medical emergency. The responses provided by the AI assistant did not offer appropriate assistance or guidance in such a critical situation."
}
A higher query resolution score reflects that the LLM effectively addresses the user’s query.
The nurse in the conversation was not able to address the patient’s query, which was about chest pain, indicating a potential medical emergency.
Resulting in a low query resolution score.
How it works?
We evaluate query resolution by determining which of the following cases apply for the given task data:
- The given responses effectively resolve the user’s query.
- The given responses partially resolve the user’s query.
- The given responses do not resolve the user’s query.
Was this page helpful?