File size: 1,771 Bytes
93adfea |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
from llamafactory.data import Role
from llamafactory.data.converter import get_dataset_converter
from llamafactory.data.parser import DatasetAttr
from llamafactory.hparams import DataArguments
def test_alpaca_converter():
dataset_attr = DatasetAttr("hf_hub", "llamafactory/tiny-supervised-dataset")
data_args = DataArguments()
example = {
"instruction": "Solve the math problem.",
"input": "3 + 4",
"output": "The answer is 7.",
}
dataset_converter = get_dataset_converter("alpaca", dataset_attr, data_args)
assert dataset_converter(example) == {
"_prompt": [{"role": Role.USER.value, "content": "Solve the math problem.\n3 + 4"}],
"_response": [{"role": Role.ASSISTANT.value, "content": "The answer is 7."}],
"_system": "",
"_tools": "",
"_images": None,
"_videos": None,
"_audios": None,
}
def test_sharegpt_converter():
dataset_attr = DatasetAttr("hf_hub", "llamafactory/tiny-supervised-dataset")
data_args = DataArguments()
example = {
"conversations": [
{"from": "system", "value": "You are a helpful assistant."},
{"from": "human", "value": "Solve the math problem.\n3 + 4"},
{"from": "gpt", "value": "The answer is 7."},
]
}
dataset_converter = get_dataset_converter("sharegpt", dataset_attr, data_args)
assert dataset_converter(example) == {
"_prompt": [{"role": Role.USER.value, "content": "Solve the math problem.\n3 + 4"}],
"_response": [{"role": Role.ASSISTANT.value, "content": "The answer is 7."}],
"_system": "You are a helpful assistant.",
"_tools": "",
"_images": None,
"_videos": None,
"_audios": None,
}
|