Spaces:
Build error
Build error
File size: 3,065 Bytes
b6af722 |
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 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 |
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# 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 argparse
import importlib
import os
from loguru import logger as logging
from cosmos_predict1.utils.config import Config, pretty_print_overrides
from cosmos_predict1.utils.config_helper import get_config_module, override
from cosmos_predict1.utils.lazy_config import instantiate
from cosmos_predict1.utils.lazy_config.lazy import LazyConfig
@logging.catch(reraise=True)
def launch(config: Config, args: argparse.Namespace) -> None:
# Check that the config is valid
config.validate()
# Freeze the config so developers don't change it during training.
config.freeze() # type: ignore
trainer = config.trainer.type(config)
# Create the model
model = instantiate(config.model)
model.on_model_init_end()
dataloader_train = instantiate(config.dataloader_train)
dataloader_val = instantiate(config.dataloader_val)
# Start training
trainer.train(
model,
dataloader_train,
dataloader_val,
)
if __name__ == "__main__":
# Usage: torchrun --nproc_per_node=1 -m scripts.train --config=projects/tutorials/mnist/config.py
# Get the config file from the input arguments.
parser = argparse.ArgumentParser(description="Training")
parser.add_argument("--config", help="Path to the config file", required=True)
parser.add_argument(
"opts",
help="""
Modify config options at the end of the command. For Yacs configs, use
space-separated "PATH.KEY VALUE" pairs.
For python-based LazyConfig, use "path.key=value".
""".strip(),
default=None,
nargs=argparse.REMAINDER,
)
parser.add_argument(
"--dryrun",
action="store_true",
help="Do a dry run without training. Useful for debugging the config.",
)
args = parser.parse_args()
config_module = get_config_module(args.config)
config = importlib.import_module(config_module).make_config()
config = override(config, args.opts)
if args.dryrun:
logging.info(
"Config:\n" + config.pretty_print(use_color=True) + "\n" + pretty_print_overrides(args.opts, use_color=True)
)
os.makedirs(config.job.path_local, exist_ok=True)
LazyConfig.save_yaml(config, f"{config.job.path_local}/config.yaml")
print(f"{config.job.path_local}/config.yaml")
else:
# Launch the training job.
launch(config, args)
|