
Reputation
Badges 1
69 × Eureka!logger = task.get_logger()
train(model_dir=trained_model_dst, pipeline_config_path=pipeline_config_path, save_checkpoints_steps=args.checkpoints)
logger.report_scalar(title='evaluate', series='score', value=5, iteration=task.get_last_iteration())
Hi CostlyOstrich36 CrookedWalrus33 AgitatedDove14
When I init an agent and run a task it works, but when I run the same task again it does not map the keys..
SweetBadger76
There's something I can do to help you check? (task id, project name etc.)
CostlyOstrich36 More precisely, My function only calculates the accuracy as I defined it.
I passing the accuracy to logger.report_scalar by
logger.report_scalar(title='evaluate', series='score', value=my_acc, iteration=task.get_last_iteration())
I have nowhere else to bring iteration number ..
CostlyOstrich36 When I do
logger = task.get_logger()
logger.report_scalar(title='evaluate', series='score', value=5, iteration=task.get_last_iteration())
train(model_dir=trained_model_dst, pipeline_config_path=pipeline_config_path, save_checkpoints_steps=args.checkpoints)
It only captures the first iteration...
CostlyOstrich36 tensorflow reporting it, ClearML capture it, and I get it with that function.
SuccessfulKoala55
Hi,
Using task.upload_artifact
pipe = PipelineController( name='grab2train_tf', project='main pipeline', version='0.0.1', abort_on_failure=True, # any failed step will cause the pipeline to immediately abort, stop all running steps, and mark the pipeline as failed. )
I'm missing something? CostlyOstrich36
CostlyOstrich36 thanks.
Maybe I should pass something in extra_docker_arguments in the config file?
Now, in other task I want to upload the artifact by using get_local_copy()dataset_upload_task = Task.get_task(task_id=args['dataset_task_id']) local_json = dataset_upload_task.artifacts['dataset'].get_local_copy()
But the path that I got contain '' and '/' (i.e combination of unix and win foramt)
MelancholyElk85 So I need to change the paths each time?
Or do I have to dive into the code in train function and write the code there?
ExasperatedCrab78
CostlyOstrich36 Not sure I understood, the current iterations come from the function
task.get_last_iteration() ...
I just added -p port_num:port_num to extra_docker_arguments
I know, but I run a scheduler on the script that downloads a dataset, and if there is no new dataset to download, I try to figure out what it will do
SuccessfulKoala55
The pipline demo is still stack on running,
The first step still on pending, and belong to services queue
No, Its stuck here:
Collecting botocore<1.23.0,>=1.22.9
Using cached botocore-1.22.12-py3-none-any.whl (8.1 MB)