Fix Windows CI OOM (#1878)

* set fixture scope to "function"

* run FARMReader without multiprocessing

* dispose off ray after tests

* run most expensive tasks first in test files

* run expensive tests first

* run garbage collector between tests

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

test/conftest.py

@@ -2,6 +2,7 @@ import subprocess
 import time
 from subprocess import run
 from sys import platform
+import gc
 
 import numpy as np
 import psutil
@@ -117,6 +118,15 @@ def pytest_collection_modifyitems(config,items):
                 item.add_marker(skip_docstore)
 
 
+@pytest.fixture(scope="function", autouse=True)
+def gc_cleanup(request):
+    """
+    Run garbage collector between tests in order to reduce memory footprint for CI.
+    """
+    yield
+    gc.collect()
+
+
 @pytest.fixture(scope="session")
 def elasticsearch_fixture():
     # test if a ES cluster is already running. If not, download and start an ES instance locally.
@@ -247,7 +257,7 @@ def xpdf_fixture():
             )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def rag_generator():
     return RAGenerator(
         model_name_or_path="facebook/rag-token-nq",
@@ -256,17 +266,17 @@ def rag_generator():
     )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def question_generator():
     return QuestionGenerator(model_name_or_path="valhalla/t5-small-e2e-qg")
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def eli5_generator():
     return Seq2SeqGenerator(model_name_or_path="yjernite/bart_eli5", max_length=20)
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def summarizer():
     return TransformersSummarizer(
         model_name_or_path="google/pegasus-xsum",
@@ -274,21 +284,21 @@ def summarizer():
     )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def en_to_de_translator():
     return TransformersTranslator(
         model_name_or_path="Helsinki-NLP/opus-mt-en-de",
     )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def de_to_en_translator():
     return TransformersTranslator(
         model_name_or_path="Helsinki-NLP/opus-mt-de-en",
     )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def test_docs_xs():
     return [
         # current "dict" format for a document
@@ -300,7 +310,7 @@ def test_docs_xs():
     ]
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def reader_without_normalized_scores():
     return FARMReader(
         model_name_or_path="distilbert-base-uncased-distilled-squad",
@@ -311,7 +321,7 @@ def reader_without_normalized_scores():
     )
 
 
-@pytest.fixture(params=["farm", "transformers"], scope="module")
+@pytest.fixture(params=["farm", "transformers"], scope="function")
 def reader(request):
     if request.param == "farm":
         return FARMReader(
@@ -328,32 +338,32 @@ def reader(request):
         )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def table_reader():
     return TableReader(model_name_or_path="google/tapas-base-finetuned-wtq")
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def ranker_two_logits():
     return SentenceTransformersRanker(
         model_name_or_path="deepset/gbert-base-germandpr-reranking",
     )
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def ranker():
     return SentenceTransformersRanker(
         model_name_or_path="cross-encoder/ms-marco-MiniLM-L-12-v2",
     )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def document_classifier():
     return TransformersDocumentClassifier(
         model_name_or_path="bhadresh-savani/distilbert-base-uncased-emotion",
         use_gpu=False
     )
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def zero_shot_document_classifier():
     return TransformersDocumentClassifier(
         model_name_or_path="cross-encoder/nli-distilroberta-base",
@@ -362,7 +372,7 @@ def zero_shot_document_classifier():
         labels=["negative", "positive"]
     )
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def batched_document_classifier():
     return TransformersDocumentClassifier(
         model_name_or_path="bhadresh-savani/distilbert-base-uncased-emotion",
@@ -370,7 +380,7 @@ def batched_document_classifier():
         batch_size=16
     )
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def indexing_document_classifier():
     return TransformersDocumentClassifier(
         model_name_or_path="bhadresh-savani/distilbert-base-uncased-emotion",
@@ -381,7 +391,7 @@ def indexing_document_classifier():
 
 # TODO Fix bug in test_no_answer_output when using
 # @pytest.fixture(params=["farm", "transformers"])
-@pytest.fixture(params=["farm"], scope="module")
+@pytest.fixture(params=["farm"], scope="function")
 def no_answer_reader(request):
     if request.param == "farm":
         return FARMReader(
@@ -401,14 +411,14 @@ def no_answer_reader(request):
         )
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def prediction(reader, test_docs_xs):
     docs = [Document.from_dict(d) if isinstance(d, dict) else d for d in test_docs_xs]
     prediction = reader.predict(query="Who lives in Berlin?", documents=docs, top_k=5)
     return prediction
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def no_answer_prediction(no_answer_reader, test_docs_xs):
     docs = [Document.from_dict(d) if isinstance(d, dict) else d for d in test_docs_xs]
     prediction = no_answer_reader.predict(query="What is the meaning of life?", documents=docs, top_k=5)
@@ -543,7 +553,7 @@ def get_document_store(document_store_type, embedding_dim=768, embedding_field="
     return document_store
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def adaptive_model_qa(num_processes):
     """
     PyTest Fixture for a Question Answering Inferencer based on PyTorch.
@@ -571,7 +581,7 @@ def adaptive_model_qa(num_processes):
     assert len(children) == 0
 
 
-@pytest.fixture(scope="module")
+@pytest.fixture(scope="function")
 def bert_base_squad2(request):
     model = QAInferencer.load(
             "deepset/minilm-uncased-squad2",

test/samples/pipeline/test_pipeline.yaml

@@ -6,6 +6,7 @@ components:
     params:
       no_ans_boost: -10
       model_name_or_path: deepset/roberta-base-squad2
+      num_processes: 0
   - name: ESRetriever
     type: ElasticsearchRetriever
     params:

test/samples/pipeline/test_pipeline_tfidfretriever.yaml

@@ -6,6 +6,7 @@ components:
     params:
       no_ans_boost: -10
       model_name_or_path: deepset/minilm-uncased-squad2
+      num_processes: 0
   - name: Retriever
     type: TfidfRetriever
     params:

test/test_distillation.py

@@ -2,8 +2,8 @@ from haystack.nodes import FARMReader
 import torch
 
 def test_distillation():
-    student = FARMReader(model_name_or_path="prajjwal1/bert-tiny")
-    teacher = FARMReader(model_name_or_path="prajjwal1/bert-small")
+    student = FARMReader(model_name_or_path="prajjwal1/bert-tiny", num_processes=0)
+    teacher = FARMReader(model_name_or_path="prajjwal1/bert-small", num_processes=0)
 
     # create a checkpoint of weights before distillation
     student_weights = []

test/test_eval.py

@@ -1,16 +1,74 @@
 import pytest
 from haystack.document_stores.base import BaseDocumentStore
+from haystack.document_stores.memory import InMemoryDocumentStore
+from haystack.document_stores.elasticsearch import ElasticsearchDocumentStore
+from haystack.nodes.answer_generator.transformers import RAGenerator, RAGeneratorType
+from haystack.nodes.retriever.dense import EmbeddingRetriever
 from haystack.nodes.preprocessor import PreProcessor
 from haystack.nodes.evaluator import EvalAnswers, EvalDocuments
 from haystack.nodes.query_classifier.transformers import TransformersQueryClassifier
 from haystack.nodes.retriever.dense import DensePassageRetriever
 from haystack.nodes.retriever.sparse import ElasticsearchRetriever
 from haystack.pipelines.base import Pipeline
-from haystack.pipelines import ExtractiveQAPipeline
+from haystack.pipelines import ExtractiveQAPipeline, GenerativeQAPipeline, SearchSummarizationPipeline
 from haystack.pipelines.standard_pipelines import DocumentSearchPipeline, FAQPipeline, RetrieverQuestionGenerationPipeline, TranslationWrapperPipeline
+from haystack.nodes.summarizer.transformers import TransformersSummarizer
 from haystack.schema import Answer, Document, EvaluationResult, Label, MultiLabel, Span
 
 
+@pytest.mark.parametrize("document_store_with_docs", ["memory"], indirect=True)
+@pytest.mark.parametrize("retriever_with_docs", ["embedding"], indirect=True)
+def test_generativeqa_calculate_metrics(document_store_with_docs: InMemoryDocumentStore, rag_generator, retriever_with_docs):
+    document_store_with_docs.update_embeddings(retriever=retriever_with_docs)
+    pipeline = GenerativeQAPipeline(generator=rag_generator, retriever=retriever_with_docs)
+    eval_result: EvaluationResult = pipeline.eval(
+        labels=EVAL_LABELS,
+        params={"Retriever": {"top_k": 5}}
+    )
+
+    metrics = eval_result.calculate_metrics()
+
+    assert "Retriever" in eval_result
+    assert "Generator" in eval_result
+    assert len(eval_result) == 2
+
+    assert metrics["Retriever"]["mrr"] == 0.5
+    assert metrics["Retriever"]["map"] == 0.5
+    assert metrics["Retriever"]["recall_multi_hit"] == 0.5
+    assert metrics["Retriever"]["recall_single_hit"] == 0.5
+    assert metrics["Retriever"]["precision"] == 1.0/6
+    assert metrics["Generator"]["exact_match"] == 0.0
+    assert metrics["Generator"]["f1"] == 1.0/3
+
+
+@pytest.mark.parametrize("document_store_with_docs", ["memory"], indirect=True)
+@pytest.mark.parametrize("retriever_with_docs", ["embedding"], indirect=True)
+def test_summarizer_calculate_metrics(document_store_with_docs: ElasticsearchDocumentStore, summarizer, retriever_with_docs):
+    document_store_with_docs.update_embeddings(retriever=retriever_with_docs)
+    pipeline = SearchSummarizationPipeline(retriever=retriever_with_docs, summarizer=summarizer, return_in_answer_format=True)
+    eval_result: EvaluationResult = pipeline.eval(
+        labels=EVAL_LABELS,
+        params={"Retriever": {"top_k": 5}}
+    )
+
+    metrics = eval_result.calculate_metrics()
+
+    assert "Retriever" in eval_result
+    assert "Summarizer" in eval_result
+    assert len(eval_result) == 2
+
+    assert metrics["Retriever"]["mrr"] == 0.5
+    assert metrics["Retriever"]["map"] == 0.5
+    assert metrics["Retriever"]["recall_multi_hit"] == 0.5
+    assert metrics["Retriever"]["recall_single_hit"] == 0.5
+    assert metrics["Retriever"]["precision"] == 1.0/6
+    assert metrics["Summarizer"]["mrr"] == 0.5
+    assert metrics["Summarizer"]["map"] == 0.5
+    assert metrics["Summarizer"]["recall_multi_hit"] == 0.5
+    assert metrics["Summarizer"]["recall_single_hit"] == 0.5
+    assert metrics["Summarizer"]["precision"] == 1.0/6
+
+
 @pytest.mark.parametrize("document_store", ["elasticsearch", "faiss", "memory", "milvus"], indirect=True)
 @pytest.mark.parametrize("batch_size", [None, 20])
 def test_add_eval_data(document_store, batch_size):

test/test_eval_embedding_retriever.py

@@ -1,72 +0,0 @@
-import pytest
-from haystack.document_stores.elasticsearch import ElasticsearchDocumentStore
-from haystack.nodes.retriever.dense import EmbeddingRetriever
-from haystack.document_stores.memory import InMemoryDocumentStore
-from haystack.nodes.summarizer.transformers import TransformersSummarizer
-from haystack.pipelines import GenerativeQAPipeline, SearchSummarizationPipeline
-from haystack.schema import EvaluationResult
-from test_eval import EVAL_LABELS
-
-
-# had to be separated from other eval tests to work around OOM in Windows CI
-
-@pytest.mark.parametrize("document_store_with_docs", ["memory"], indirect=True)
-@pytest.mark.parametrize("retriever_with_docs", ["embedding"], indirect=True)
-def test_generativeqa_calculate_metrics(document_store_with_docs: InMemoryDocumentStore, rag_generator, retriever_with_docs):
-    document_store_with_docs.update_embeddings(retriever=retriever_with_docs)
-    pipeline = GenerativeQAPipeline(generator=rag_generator, retriever=retriever_with_docs)
-    eval_result: EvaluationResult = pipeline.eval(
-        labels=EVAL_LABELS,
-        params={"Retriever": {"top_k": 5}}
-    )
-
-    metrics = eval_result.calculate_metrics()
-
-    assert "Retriever" in eval_result
-    assert "Generator" in eval_result
-    assert len(eval_result) == 2
-
-    assert metrics["Retriever"]["mrr"] == 0.5
-    assert metrics["Retriever"]["map"] == 0.5
-    assert metrics["Retriever"]["recall_multi_hit"] == 0.5
-    assert metrics["Retriever"]["recall_single_hit"] == 0.5
-    assert metrics["Retriever"]["precision"] == 1.0/6
-    assert metrics["Generator"]["exact_match"] == 0.0
-    assert metrics["Generator"]["f1"] == 1.0/3
-
-
-@pytest.mark.parametrize("document_store_with_docs", ["memory"], indirect=True)
-def test_summarizer_calculate_metrics(document_store_with_docs: ElasticsearchDocumentStore):
-    summarizer = TransformersSummarizer(
-        model_name_or_path="sshleifer/distill-pegasus-xsum-16-4",
-        use_gpu=False
-    )
-    document_store_with_docs.embedding_dim = 384
-    retriever = EmbeddingRetriever(
-            document_store=document_store_with_docs,
-            embedding_model="sentence-transformers/all-MiniLM-L6-v2",
-            use_gpu=False
-        )
-    document_store_with_docs.update_embeddings(retriever=retriever)
-    pipeline = SearchSummarizationPipeline(retriever=retriever, summarizer=summarizer, return_in_answer_format=True)
-    eval_result: EvaluationResult = pipeline.eval(
-        labels=EVAL_LABELS,
-        params={"Retriever": {"top_k": 5}}
-    )
-
-    metrics = eval_result.calculate_metrics()
-
-    assert "Retriever" in eval_result
-    assert "Summarizer" in eval_result
-    assert len(eval_result) == 2
-
-    assert metrics["Retriever"]["mrr"] == 0.5
-    assert metrics["Retriever"]["map"] == 0.5
-    assert metrics["Retriever"]["recall_multi_hit"] == 0.5
-    assert metrics["Retriever"]["recall_single_hit"] == 0.5
-    assert metrics["Retriever"]["precision"] == 1.0/6
-    assert metrics["Summarizer"]["mrr"] == 0.0
-    assert metrics["Summarizer"]["map"] == 0.0
-    assert metrics["Summarizer"]["recall_multi_hit"] == 0.0
-    assert metrics["Summarizer"]["recall_single_hit"] == 0.0
-    assert metrics["Summarizer"]["precision"] == 0.0

test/test_extractor.py

@@ -12,7 +12,7 @@ def test_extractor(document_store_with_docs):
     
     es_retriever = ElasticsearchRetriever(document_store=document_store_with_docs)
     ner = EntityExtractor()
-    reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2")
+    reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2", num_processes=0)
 
     pipeline = Pipeline()
     pipeline.add_node(component=es_retriever, name="ESRetriever", inputs=["Query"])
@@ -36,7 +36,7 @@ def test_extractor_output_simplifier(document_store_with_docs):
     
     es_retriever = ElasticsearchRetriever(document_store=document_store_with_docs)
     ner = EntityExtractor()
-    reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2")
+    reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2", num_processes=0)
 
     pipeline = Pipeline()
     pipeline.add_node(component=es_retriever, name="ESRetriever", inputs=["Query"])

test/test_generator.py

@@ -12,6 +12,35 @@ from haystack.pipelines import TranslationWrapperPipeline, GenerativeQAPipeline
 from conftest import DOCS_WITH_EMBEDDINGS
 
 
+# Keeping few (retriever,document_store) combination to reduce test time
+@pytest.mark.slow
+@pytest.mark.generator
+@pytest.mark.parametrize(
+    "retriever,document_store",
+    [("embedding", "memory")],
+    indirect=True,
+)
+def test_generator_pipeline_with_translator(
+    document_store,
+    retriever,
+    rag_generator,
+    en_to_de_translator,
+    de_to_en_translator
+):
+    document_store.write_documents(DOCS_WITH_EMBEDDINGS)
+    query = "Was ist die Hauptstadt der Bundesrepublik Deutschland?"
+    base_pipeline = GenerativeQAPipeline(retriever=retriever, generator=rag_generator)
+    pipeline = TranslationWrapperPipeline(
+        input_translator=de_to_en_translator,
+        output_translator=en_to_de_translator,
+        pipeline=base_pipeline
+    )
+    output = pipeline.run(query=query, params={"Generator": {"top_k": 2}, "Retriever": {"top_k": 1}})
+    answers = output["answers"]
+    assert len(answers) == 2
+    assert "berlin" in answers[0].answer
+
+
 @pytest.mark.slow
 @pytest.mark.generator
 def test_rag_token_generator(rag_generator):
@@ -105,32 +134,3 @@ def test_lfqa_pipeline_invalid_converter(document_store, retriever):
     with pytest.raises(Exception) as exception_info:
         output = pipeline.run(query=query, params={"top_k": 1})
     assert ("does not have a valid __call__ method signature" in str(exception_info.value))
-
-
-# Keeping few (retriever,document_store) combination to reduce test time
-@pytest.mark.slow
-@pytest.mark.generator
-@pytest.mark.parametrize(
-    "retriever,document_store",
-    [("embedding", "memory")],
-    indirect=True,
-)
-def test_generator_pipeline_with_translator(
-    document_store,
-    retriever,
-    rag_generator,
-    en_to_de_translator,
-    de_to_en_translator
-):
-    document_store.write_documents(DOCS_WITH_EMBEDDINGS)
-    query = "Was ist die Hauptstadt der Bundesrepublik Deutschland?"
-    base_pipeline = GenerativeQAPipeline(retriever=retriever, generator=rag_generator)
-    pipeline = TranslationWrapperPipeline(
-        input_translator=de_to_en_translator,
-        output_translator=en_to_de_translator,
-        pipeline=base_pipeline
-    )
-    output = pipeline.run(query=query, params={"Generator": {"top_k": 2}, "Retriever": {"top_k": 1}})
-    answers = output["answers"]
-    assert len(answers) == 2
-    assert "berlin" in answers[0].answer

test/test_pipeline.py

@@ -65,6 +65,7 @@ def test_load_and_save_yaml(document_store, tmp_path):
           params:
             model_name_or_path: deepset/roberta-base-squad2
             no_ans_boost: -10
+            num_processes: 0
           type: FARMReader
         pipelines:
         - name: query
@@ -127,6 +128,7 @@ def test_load_and_save_yaml_prebuilt_pipelines(document_store, tmp_path):
           params:
             model_name_or_path: deepset/roberta-base-squad2
             no_ans_boost: -10
+            num_processes: 0
           type: FARMReader
         pipelines:
         - name: query

test/test_pipeline_debug_and_validation.py

@@ -23,7 +23,7 @@ def test_node_names_validation(document_store_with_docs, tmp_path):
         name="Retriever", 
         inputs=["Query"])
     pipeline.add_node(
-        component=FARMReader(model_name_or_path="deepset/minilm-uncased-squad2"), 
+        component=FARMReader(model_name_or_path="deepset/minilm-uncased-squad2", num_processes=0), 
         name="Reader", 
         inputs=["Retriever"])
 
@@ -50,7 +50,7 @@ def test_node_names_validation(document_store_with_docs, tmp_path):
 def test_debug_attributes_global(document_store_with_docs, tmp_path):
 
     es_retriever = ElasticsearchRetriever(document_store=document_store_with_docs)
-    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2")
+    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2", num_processes=0)
 
     pipeline = Pipeline()
     pipeline.add_node(component=es_retriever, name="ESRetriever", inputs=["Query"])
@@ -81,7 +81,7 @@ def test_debug_attributes_global(document_store_with_docs, tmp_path):
 def test_debug_attributes_per_node(document_store_with_docs, tmp_path):
 
     es_retriever = ElasticsearchRetriever(document_store=document_store_with_docs)
-    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2")
+    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2", num_processes=0)
 
     pipeline = Pipeline()
     pipeline.add_node(component=es_retriever, name="ESRetriever", inputs=["Query"])
@@ -111,7 +111,7 @@ def test_debug_attributes_per_node(document_store_with_docs, tmp_path):
 def test_global_debug_attributes_override_node_ones(document_store_with_docs, tmp_path):
 
     es_retriever = ElasticsearchRetriever(document_store=document_store_with_docs)
-    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2")
+    reader = FARMReader(model_name_or_path="deepset/minilm-uncased-squad2", num_processes=0)
 
     pipeline = Pipeline()
     pipeline.add_node(component=es_retriever, name="ESRetriever", inputs=["Query"])

test/test_ray.py

@@ -17,3 +17,13 @@ def test_load_pipeline(document_store_with_docs):
     assert ray.serve.get_deployment(name="Reader").num_replicas == 1
     assert prediction["query"] == "Who lives in Berlin?"
     assert prediction["answers"][0].answer == "Carla"
+
+
+@pytest.fixture(scope="function", autouse=True)
+def shutdown_ray():
+    yield
+    try:
+        import ray
+        ray.shutdown()
+    except:
+        pass

test/test_reader.py

@@ -56,7 +56,7 @@ def test_prediction_attributes(prediction):
 def test_model_download_options():
     # download disabled and model is not cached locally
     with pytest.raises(OSError):
-        impossible_reader = FARMReader("mfeb/albert-xxlarge-v2-squad2", local_files_only=True)
+        impossible_reader = FARMReader("mfeb/albert-xxlarge-v2-squad2", local_files_only=True, num_processes=0)
 
 def test_answer_attributes(prediction):
     # TODO Transformers answer also has meta key