mirror of
https://github.com/hiyouga/LLaMA-Factory.git
synced 2026-08-17 05:25:44 +08:00
90 lines
3.1 KiB
Python
90 lines
3.1 KiB
Python
# Copyright 2025 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 sys
|
|
from pathlib import Path
|
|
from unittest.mock import patch
|
|
|
|
from llamafactory.v1.config.arg_parser import get_args
|
|
|
|
|
|
def test_get_args_from_yaml(tmp_path: Path):
|
|
config_yaml = """
|
|
### model
|
|
model: llamafactory/tiny-random-qwen3
|
|
trust_remote_code: true
|
|
model_class: llm
|
|
kernel_config:
|
|
name: auto, flash-linear-attention
|
|
include_kernels: chunk_gated_delta_rule, fused_recurrent_gated_delta_rule
|
|
chunk_size: 32
|
|
peft_config:
|
|
name: lora
|
|
r: 8
|
|
quant_config: null
|
|
|
|
### data
|
|
train_dataset: llamafactory/v1-sft-demo
|
|
|
|
### training
|
|
output_dir: outputs/test_run
|
|
micro_batch_size: 1
|
|
global_batch_size: 1
|
|
cutoff_len: 2048
|
|
learning_rate: 1.0e-4
|
|
bf16: false
|
|
dist_config: null
|
|
|
|
### sample
|
|
sample_backend: hf
|
|
max_new_tokens: 128
|
|
"""
|
|
|
|
config_file = tmp_path / "config.yaml"
|
|
config_file.write_text(config_yaml, encoding="utf-8")
|
|
|
|
test_argv = ["test_args_parser.py", str(config_file)]
|
|
|
|
with patch.object(sys, "argv", test_argv):
|
|
model_args, data_args, training_args, sample_args = get_args()
|
|
assert data_args.train_dataset == "llamafactory/v1-sft-demo"
|
|
assert model_args.model == "llamafactory/tiny-random-qwen3"
|
|
assert model_args.kernel_config.name == "auto, flash-linear-attention"
|
|
assert model_args.kernel_config.get("include_kernels") == (
|
|
"chunk_gated_delta_rule, fused_recurrent_gated_delta_rule"
|
|
)
|
|
assert model_args.kernel_config.get("chunk_size") == 32
|
|
assert model_args.peft_config.name == "lora"
|
|
assert model_args.peft_config.get("r") == 8
|
|
assert training_args.output_dir == "outputs/test_run"
|
|
assert training_args.micro_batch_size == 1
|
|
assert training_args.global_batch_size == 1
|
|
assert training_args.learning_rate == 1.0e-4
|
|
assert training_args.bf16 is False
|
|
assert training_args.dist_config is None
|
|
assert sample_args.sample_backend == "hf"
|
|
|
|
|
|
def test_qwen35_fsdpturbo_example_uses_v1_arguments():
|
|
config_file = (
|
|
Path(__file__).parents[2] / "examples" / "v1" / "train_full" / "train_full_qwen3_moe_fsdpturbo_ep_fsdp.yaml"
|
|
)
|
|
|
|
with patch.object(sys, "argv", ["test_args_parser.py", str(config_file)]):
|
|
model_args, _, training_args, _ = get_args()
|
|
|
|
assert model_args.model == "Qwen/Qwen3.5-35B-A3B"
|
|
assert model_args.custom_chat_template is None
|
|
assert training_args.dist_config.name == "fsdpturbo"
|