How to send system input in api as we do it webui

[bhanuash7]

{
"model": "string",
"messages": [
{
"role": "user",
"content": "string"
}
],
"tools": [],
"do_sample": true,
"temperature": 0,
"top_p": 0,
"n": 1,
"max_tokens": 0,
"stream": false
}

the body that im using

{
"model": "convo",
"messages": [
{
"role": "user",
"content": "hello"
},
{
"role": "assistant",
"content": "Hi there! How can I assist you today?"
},
{
"role": "user",
"content": "I need help in planning trip\n"
},
{
"role": "assistant",
"content": "Of course! When and where would you like to go?"
},
{
"role": "user",
"content": "on 7 august and goa"
}
],
"tools": [],
"do_sample": true,
"temperature": 0.1,
"top_p": 0.1,
"n": 1,
"max_tokens": 512,
"stream": false
}

[bhanuash7]

and the system message that i want to pass

"The itinerary includes:\nDay 1 - Vienna, Meals: Dinner, Description: Welcome to Vienna\nThe Tour Manager / Station Manager will meet you outside the airport. Proceed for dinner and then to the hotel and check\u2013in.\n\u00a0 Note: Joining direct passenger / NRI can check-in on your own at the hotel. The standard check-in time is 1500 HRS. Packed dinner will be provided if flight arrival is late or if there is a long queue at the immigration, GeoLocation: (Lat: 48.208174, Lon: 16.373819)."

[hiyouga]

add a system message whose role is system in the front of the message list

[hiyouga]

"messages": [
{
"role": "system",
"content": "system instructions"
},
{
"role": "user",
"content": "hello"
}
]

[bhanuash7]

"dataset_name": {
"file_name": "data.json",
"columns": {
"prompt": "instruction",
"query": "input", <---------------------
"response": "output",
"system": "system",
"history": "history"
}
}

[
{
"instruction": "human instruction (required)",
"input": "human input (optional)", <---------------------
"output": "model response (required)",
"system": "system prompt (optional)",
"history": [
["human instruction in the first round (optional)", "model response in the first round (optional)"],
["human instruction in the second round (optional)", "model response in the second round (optional)"]
]
}
]

the input that we are using for training how should i pass that information through api?

[hiyouga]

https://github.com/hiyouga/LLaMA-Factory/tree/main/data#supervised-fine-tuning-dataset

In supervised fine-tuning, the instruction column will be concatenated with the input column and used as the human prompt, then the human prompt would be instruction\ninput.

[bhanuash7]

{
"model": "string",
"messages": [
{
"role": "user",
"content": "instruction\ninput" <--------------
}
],
"tools": [],
"do_sample": true,
"temperature": 0,
"top_p": 0,
"n": 1,
"max_tokens": 0,
"stream": false
}

should i do this?

[hiyouga]

yep