Hallucinating tools response
Hello.
I have an assistant that is made to get today's time.
I created a tool with the following spec:
{
"type": "function",
"messages": [
{
"type": "request-start",
"content": "How do you feel?"
},
{
"type": "request-complete",
"content": "So today's time is is"
},
{
"type": "request-failed",
"content": "I couldn't get the current time information right now."
},
{
"type": "request-response-delayed",
"content": "It appears there is some delay in communication with the time API.",
"timingMilliseconds": 2000
}
],
"function": {
"name": "GetTodaysTime",
"description": "Use this tool to get today's time"
},
"async": true,
"server": {
"url": "https://hook.us1.make.com/btjnlkr8x1na3mqfafy8mvtxmrbw09er"
}
}
I made a make scenario and I make sure the response returned looks like the following:
{"results":[{"result":"12:01","toolCallId":"call_ZSYgW6OKknzU664Hg3UjX4es"}]}
Always I made sure to add "Tool Calls" in the ClientMessages and ServerMessages section as per the documentation.
The assistant still hallucinats the response and will say 10AM instead of 12:10 (It's 12:10 for me now)
I have an assistant that is made to get today's time.
I created a tool with the following spec:
{
"type": "function",
"messages": [
{
"type": "request-start",
"content": "How do you feel?"
},
{
"type": "request-complete",
"content": "So today's time is is"
},
{
"type": "request-failed",
"content": "I couldn't get the current time information right now."
},
{
"type": "request-response-delayed",
"content": "It appears there is some delay in communication with the time API.",
"timingMilliseconds": 2000
}
],
"function": {
"name": "GetTodaysTime",
"description": "Use this tool to get today's time"
},
"async": true,
"server": {
"url": "https://hook.us1.make.com/btjnlkr8x1na3mqfafy8mvtxmrbw09er"
}
}
I made a make scenario and I make sure the response returned looks like the following:
{"results":[{"result":"12:01","toolCallId":"call_ZSYgW6OKknzU664Hg3UjX4es"}]}
Always I made sure to add "Tool Calls" in the ClientMessages and ServerMessages section as per the documentation.
The assistant still hallucinats the response and will say 10AM instead of 12:10 (It's 12:10 for me now)
