mirror of
https://github.com/hiyouga/LLaMA-Factory.git
synced 2025-08-02 03:32:50 +08:00
71 lines
2.0 KiB
Python
71 lines
2.0 KiB
Python
# Copyright 2024 the LlamaFactory team.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
import os
|
|
|
|
import pytest
|
|
|
|
from llamafactory.train.tuner import export_model, run_exp
|
|
|
|
|
|
DEMO_DATA = os.environ.get("DEMO_DATA", "llamafactory/demo_data")
|
|
|
|
TINY_LLAMA = os.environ.get("TINY_LLAMA", "llamafactory/tiny-random-Llama-3")
|
|
|
|
TINY_LLAMA_ADAPTER = os.environ.get("TINY_LLAMA_ADAPTER", "llamafactory/tiny-random-Llama-3-lora")
|
|
|
|
TRAIN_ARGS = {
|
|
"model_name_or_path": TINY_LLAMA,
|
|
"do_train": True,
|
|
"finetuning_type": "lora",
|
|
"dataset_dir": "REMOTE:" + DEMO_DATA,
|
|
"template": "llama3",
|
|
"cutoff_len": 1,
|
|
"overwrite_cache": True,
|
|
"overwrite_output_dir": True,
|
|
"per_device_train_batch_size": 1,
|
|
"max_steps": 1,
|
|
"fp16": True,
|
|
}
|
|
|
|
INFER_ARGS = {
|
|
"model_name_or_path": TINY_LLAMA,
|
|
"adapter_name_or_path": TINY_LLAMA_ADAPTER,
|
|
"finetuning_type": "lora",
|
|
"template": "llama3",
|
|
"infer_dtype": "float16",
|
|
"export_dir": "llama3_export",
|
|
}
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"stage,dataset",
|
|
[
|
|
("pt", "c4_demo"),
|
|
("sft", "alpaca_en_demo"),
|
|
("rm", "dpo_en_demo"),
|
|
("dpo", "dpo_en_demo"),
|
|
("kto", "kto_en_demo"),
|
|
],
|
|
)
|
|
def test_train(stage: str, dataset: str):
|
|
output_dir = "train_{}".format(stage)
|
|
run_exp({"stage": stage, "dataset": dataset, "output_dir": output_dir, **TRAIN_ARGS})
|
|
assert os.path.exists(output_dir)
|
|
|
|
|
|
def test_export():
|
|
export_model(INFER_ARGS)
|
|
assert os.path.exists("llama3_export")
|