Hi All! I’M Still New To This And Would Be Very Grateful If You Could Answer My Question Or Point Me To A Topic Where This Is Discussed. I’M Working On My First Pet Project, And I Managed To Create A Dataset And Train A Model. However, After That The Mode

Answered

Hi all! I’m still new to this and would be very grateful if you could answer my question or point me to a topic where this is discussed. I’m working on my first pet project, and I managed to create a dataset and train a model. However, after that the model training task includes a reference to the local file path, but there is no download arrow. At first, I see the model file in my local folder, but then it disappears, and I cannot create predictions on "new" data in the inference task (get_local_copy() returns None). Am I using the wrong approach to storing my model? 😢

model_path = "model.cbm"
task.upload_artifact(name="trained_model", artifact_object=model_path)
model.save_model(model_path)
output_model = OutputModel(task=task, framework="CatBoost")
output_model.update_weights(model_path)

  				
Posted 
	13 days ago

					More
				  		
  Report
		
					TenseSwan1
				
					0
					 × 1

Votes Newest

Answers 2

Thank you, @<1523701070390366208:profile|CostlyOstrich36> ! I’ve gone through the documentation on output_uri and noticed that it defaults to None in my code. When I use ClearML with the website app.clearml and my notebook as an agent (the pipeline is started with pipe.start(queue='default') ), all logs appear in the UI console, but not in my terminal. Does this mean it’s impossible to store the model in this setup?
In the UI, I see:
"Snapshot stored. Created by task ID: 72420a1ec3d54c37b4ed479cce9047cc"
but there’s no download arrow for the model file.
In contrast, in the ClearML example pipeline, the download arrow is present, and "snapshot" isn’t mentioned... 😢

  				
Posted 
	11 days ago

					More
				  		
  Report
		
					TenseSwan1
				
					0
					 × 1

Hi @<1775332371021697024:profile|TenseSwan1> , I think what you're looking for is the output_uri parameter in Task.init

  				
Posted 
	12 days ago

					More
				  		
  Report
		
					CostlyOstrich36
				
					0

Write your answer

101 Views

2 Answers

13 days ago

11 days ago