Hey ScaryDeer25 , we just released clearml==1.11.1rc2 which should solve the compatibility issues for lightning >= 2.0. Can you install it and check whether it solves your problem?
Hi ScaryDeer25 , is thie from a task run by an agent?
The issue may be related to the fact that right now we have some edge cases when working with lightning >= 2.0, we should have better support in the upcoming release
It seems the task initialization process is successful, can you share a snippet of the code, how click is used and what are you doing in the main part of the code (including Task.init() etc.)?
Hi 🙂
Thanks for your reply, I will find it out
Yeah sure 🙂
Will share it, let me get it ^^
Here are the code.
When using ClearML Agent it does not work but if we copy paste the command line specified in the log as Entry point it works
"entry_point = -m emotions_classification.__main__ lightning train -c configs/logging.yml -c configs/lightning_module/multiclass-transformer.yml -c configs/data_module/fr-emotions-data-module.yml -c configs/trainer/debug.yml -c configs/common/camembert-base.yml trainer.logger.name=debug
working_dir = .
Can you please share the entire task log? Also, how are you running the agent?
Yeah, this is the ouput of the run from the agent
Hello 🙂
This is the snipper using Click
@define()
class ClearMLConfig:
task_name: str = "First experiment"
task_type: Task.TaskTypes = Task.TaskTypes.training
reuse_last_task_id: bool = True
output_uri: Optional[str] = None
clearml_config = ClearMLConfig()
task = Task.init(project_name='Illuin Emotions',
task_name=clearml_config.task_name,
reuse_last_task_id=clearml_config.reuse_last_task_id,
task_type=clearml_config.task_type,
output_uri=clearml_config.output_uri)
@click.group()
def cli() -> None:
if os.path.isfile(".env"):
logger.info("Loading .env file")
load_dotenv(".env")
cli.add_command(lightning_cli, name="lightning")
cli.add_command(huggingface_cli, name="huggingface")
And to log it we use:Task.current_task().connect(attrs.asdict(config))