Disable caching for steps in pipeline (#86)

Signed-off-by: Daniel Ryszka <dryszka@redhat.com>
Signed-off-by: Witold Nowogorski <wnowogor@redhat.com>
Co-authored-by: Witold Nowogorski <wnowogor@redhat.com>

Author: 2 people

pipelines/training/automl/autogluon_tabular_training_pipeline/pipeline.py

@@ -154,6 +154,7 @@ def autogluon_tabular_training_pipeline(
         label_column=label_column,
         task_type=task_type,
     )
+    data_loader_task.set_caching_options(False)
     data_loader_task.set_cpu_request("2").set_memory_request("8Gi")
 
     use_secret_as_env(
@@ -179,6 +180,7 @@ def autogluon_tabular_training_pipeline(
         top_n=top_n,
         workspace_path=dsl.WORKSPACE_PATH_PLACEHOLDER,
     )
+    selection_task.set_caching_options(False)
     selection_task.set_cpu_request("2").set_memory_request("8Gi")
 
     # Stage 2: Model Refitting
@@ -197,13 +199,15 @@ def autogluon_tabular_training_pipeline(
             sample_row=data_loader_task.outputs["sample_row"],
             extra_train_data_path=data_loader_task.outputs["extra_train_data_path"],
         )
+        refit_full_task.set_caching_options(False)
         refit_full_task.set_cpu_request("2").set_memory_request("8Gi")
 
     # Generate leaderboard
     leaderboard_evaluation_task = leaderboard_evaluation(
         models=dsl.Collected(refit_full_task.outputs["model_artifact"]),
         eval_metric=selection_task.outputs["eval_metric"],
     )
+    leaderboard_evaluation_task.set_caching_options(False)
     leaderboard_evaluation_task.set_cpu_request("1").set_memory_request("4Gi")
 
 

pipelines/training/autorag/documents_rag_optimization_pipeline/pipeline.py

@@ -82,16 +82,25 @@ def documents_rag_optimization_pipeline(
         test_data_path=test_data_key,
     )
 
+    test_data_loader_task.set_caching_options(False)
+    test_data_loader_task.set_cpu_request("2").set_memory_request("8Gi")
+
     documents_discovery_task = documents_discovery(
         input_data_bucket_name=input_data_bucket_name,
         input_data_path=input_data_key,
         test_data=test_data_loader_task.outputs["test_data"],
     )
 
+    documents_discovery_task.set_caching_options(False)
+    documents_discovery_task.set_cpu_request("2").set_memory_request("8Gi")
+
     text_extraction_task = text_extraction(
         documents_descriptor=documents_discovery_task.outputs["discovered_documents"],
     )
 
+    text_extraction_task.set_caching_options(False)
+    text_extraction_task.set_cpu_request("2").set_memory_request("8Gi")
+
     for task, secret_name in zip(
         [test_data_loader_task, documents_discovery_task, text_extraction_task],
         [test_data_secret_name, input_data_secret_name, input_data_secret_name],
@@ -114,6 +123,9 @@ def documents_rag_optimization_pipeline(
         generation_models=generation_models,
     )
 
+    mps_task.set_caching_options(False)
+    mps_task.set_cpu_request("2").set_memory_request("8Gi")
+
     hpo_task = rag_templates_optimization(
         extracted_text=text_extraction_task.outputs["extracted_text"],
         test_data=test_data_loader_task.outputs["test_data"],
@@ -127,6 +139,10 @@ def documents_rag_optimization_pipeline(
         input_data_key=input_data_key,
     )
 
+    hpo_task.set_caching_options(False)
+    hpo_task.set_cpu_request("2").set_memory_request("8Gi")
+
+
     use_secret_as_env(
         mps_task,
         llama_stack_secret_name,
@@ -145,6 +161,8 @@ def documents_rag_optimization_pipeline(
     )
 
     leaderboard_evaluation_task = leaderboard_evaluation(rag_patterns=hpo_task.outputs["rag_patterns"])
+    leaderboard_evaluation_task.set_caching_options(False)
+    leaderboard_evaluation_task.set_cpu_request("1").set_memory_request("4Gi")
 
 
 if __name__ == "__main__":