ML pipelines: RunInference - OSS Image Object detection, OSS Image Captioning, OSS Image Classification

.github/workflows/beam_Inference_Python_Benchmarks_Dataflow.yml

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-name: Inference Python Benchmarks Dataflow
+name: Inference Python Benchmarks Dataflow (1 part)
 
 on:
   schedule:
@@ -95,24 +95,6 @@ jobs:
       # The env variables are created and populated in the test-arguments-action as "<github.job>_test_arguments_<argument_file_paths_index>"
       - name: get current time
         run: echo "NOW_UTC=$(date '+%m%d%H%M%S' --utc)" >> $GITHUB_ENV
-      - name: Build VLLM Development Image
-        id: build_vllm_image
-        uses: ./.github/actions/build-push-docker-action
-        with:
-          dockerfile_path: 'sdks/python/apache_beam/ml/inference/test_resources/vllm.dockerfile'
-          image_name: 'us-docker.pkg.dev/apache-beam-testing/beam-temp/beam-vllm-gpu-base'
-          image_tag: ${{ github.sha }}
-      - name: Run VLLM Gemma Batch Test
-        uses: ./.github/actions/gradle-command-self-hosted-action
-        timeout-minutes: 180
-        with:
-          gradle-command: :sdks:python:apache_beam:testing:load_tests:run
-          arguments: |
-            -PloadTest.mainClass=apache_beam.testing.benchmarks.inference.vllm_gemma_benchmarks \
-            -Prunner=DataflowRunner \
-            -PsdkLocationOverride=false \
-            -PpythonVersion=3.10 \
-            -PloadTest.requirementsTxtFile=apache_beam/ml/inference/vllm_tests_requirements.txt '-PloadTest.args=${{ env.beam_Inference_Python_Benchmarks_Dataflow_test_arguments_8 }} --mode=batch --job_name=benchmark-tests-vllm-with-gemma-2b-it-batch-${{env.NOW_UTC}} --sdk_container_image=${{ steps.build_vllm_image.outputs.image_url }}'
       - name: run Pytorch Sentiment Streaming using Hugging Face distilbert-base-uncased model
         uses: ./.github/actions/gradle-command-self-hosted-action
         timeout-minutes: 180
@@ -189,4 +171,22 @@ jobs:
             -Prunner=DataflowRunner \
             -PpythonVersion=3.10 \
             -PloadTest.requirementsTxtFile=apache_beam/ml/inference/torch_tests_requirements.txt \
-            '-PloadTest.args=${{ env.beam_Inference_Python_Benchmarks_Dataflow_test_arguments_5 }} --job_name=benchmark-tests-pytorch-imagenet-python-gpu-${{env.NOW_UTC}} --output=gs://temp-storage-for-end-to-end-tests/torch/result_resnet152_gpu-${{env.NOW_UTC}}.txt'
+            '-PloadTest.args=${{ env.beam_Inference_Python_Benchmarks_Dataflow_test_arguments_5 }} --job_name=benchmark-tests-pytorch-imagenet-python-gpu-${{env.NOW_UTC}} --output=gs://temp-storage-for-end-to-end-tests/torch/result_resnet152_gpu-${{env.NOW_UTC}}.txt' \
+      - name: Build VLLM Development Image
+        id: build_vllm_image
+        uses: ./.github/actions/build-push-docker-action
+        with:
+          dockerfile_path: 'sdks/python/apache_beam/ml/inference/test_resources/vllm.dockerfile'
+          image_name: 'us-docker.pkg.dev/apache-beam-testing/beam-temp/beam-vllm-gpu-base'
+          image_tag: ${{ github.sha }}
+      - name: Run VLLM Gemma Batch Test
+        uses: ./.github/actions/gradle-command-self-hosted-action
+        timeout-minutes: 180
+        with:
+          gradle-command: :sdks:python:apache_beam:testing:load_tests:run
+          arguments: |
+            -PloadTest.mainClass=apache_beam.testing.benchmarks.inference.vllm_gemma_benchmarks \
+            -Prunner=DataflowRunner \
+            -PsdkLocationOverride=false \
+            -PpythonVersion=3.10 \
+            -PloadTest.requirementsTxtFile=apache_beam/ml/inference/vllm_tests_requirements.txt '-PloadTest.args=${{ env.beam_Inference_Python_Benchmarks_Dataflow_test_arguments_8 }} --mode=batch --job_name=benchmark-tests-vllm-with-gemma-2b-it-batch-${{env.NOW_UTC}} --sdk_container_image=${{ steps.build_vllm_image.outputs.image_url }}'

.github/workflows/load-tests-pipeline-options/beam_Inference_Python_Benchmarks_Dataflow_2_Pytorch_Image_Captioning.txt

@@ -0,0 +1,37 @@
+#  distributed with this work for additional information
+#  regarding copyright ownership.  The ASF licenses this file
+#  to you under the Apache License, Version 2.0 (the
+#  "License"); you may not use this file except in compliance
+#  with the License.  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+--region=us-central1
+--worker_machine_type=n1-standard-4
+--num_workers=50
+--disk_size_gb=50
+--autoscaling_algorithm=NONE
+--staging_location=gs://temp-storage-for-perf-tests/loadtests
+--temp_location=gs://temp-storage-for-perf-tests/loadtests
+--requirements_file=apache_beam/ml/inference/pytorch_image_captioning_requirements.txt
+--publish_to_big_query=true
+--metrics_dataset=beam_run_inference
+--input_options={}
+--influx_measurement=result_torch_inference_image_captioning
+--input=gs://apache-beam-ml/testing/inputs/openimage_50k_benchmark.txt
+--blip_model_name=Salesforce/blip-image-captioning-base
+--blip_batch_size=4
+--num_captions=5
+--max_new_tokens=30
+--num_beams=5
+--clip_model_name=openai/clip-vit-base-patch32
+--clip_batch_size=8
+--clip_score_normalize=false
+--runner=DataflowRunner
+--experiments=use_runner_v2

.github/workflows/load-tests-pipeline-options/beam_Inference_Python_Benchmarks_Dataflow_2_Pytorch_Image_Classification_Rightfit.txt

@@ -0,0 +1,42 @@
+#  Licensed to the Apache Software Foundation (ASF) under one
+#  or more contributor license agreements.  See the NOTICE file
+#  distributed with this work for additional information
+#  regarding copyright ownership.  The ASF licenses this file
+#  to you under the Apache License, Version 2.0 (the
+#  "License"); you may not use this file except in compliance
+#  with the License.  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+--region=us-central1
+--machine_type=n1-standard-4
+--num_workers=50
+--disk_size_gb=50
+--autoscaling_algorithm=NONE
+--staging_location=gs://temp-storage-for-perf-tests/loadtests
+--temp_location=gs://temp-storage-for-perf-tests/loadtests
+--requirements_file=apache_beam/ml/inference/pytorch_rightfit_requirements.txt
+--publish_to_big_query=true
+--metrics_dataset=beam_run_inference
+--influx_measurement=torch_inference_imagenet_stream_rightfit
+--pretrained_model_name=efficientnet_b0
+--input_file=gs://apache-beam-ml/testing/inputs/openimage_50k_benchmark.txt
+--runner=DataflowRunner
+--mode=streaming
+--input_mode=gcs_uris
+--input_options={}
+--pubsub_topic=projects/apache-beam-testing/topics/images_topic
+--pubsub_subscription=projects/apache-beam-testing/subscriptions/images_subscription
+--model_state_dict_path=gs://apache-beam-ml/models/efficientnet_b0_state_dict.pth
+--rate_limit=250
+--image_size=224
+--top_k=5
+--inference_batch_size=auto
+--window_sec=60
+--trigger_proc_time_sec=30

.github/workflows/load-tests-pipeline-options/beam_Inference_Python_Benchmarks_Dataflow_2_Pytorch_Image_Object_Detection.txt

@@ -0,0 +1,37 @@
+#  Licensed to the Apache Software Foundation (ASF) under one
+#  or more contributor license agreements.  See the NOTICE file
+#  distributed with this work for additional information
+#  regarding copyright ownership.  The ASF licenses this file
+#  to you under the Apache License, Version 2.0 (the
+#  "License"); you may not use this file except in compliance
+#  with the License.  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+--region=us-central1
+--worker_machine_type=n1-standard-4
+--num_workers=50
+--disk_size_gb=50
+--autoscaling_algorithm=NONE
+--staging_location=gs://temp-storage-for-perf-tests/loadtests
+--temp_location=gs://temp-storage-for-perf-tests/loadtests
+--requirements_file=apache_beam/ml/inference/pytorch_image_object_detection_requirements.txt
+--publish_to_big_query=true
+--metrics_dataset=beam_run_inference
+--input_options={}
+--influx_measurement=result_torch_inference_image_object_detection_batch
+--pretrained_model_name=fasterrcnn_resnet50_fpn
+--inference_batch_size=8
+--resize_shorter_side=800
+--score_threshold=0.5
+--max_detections=50
+--input=gs://apache-beam-ml/testing/inputs/openimage_50k_benchmark.txt
+--model_state_dict_path=gs://apache-beam-ml/models/torchvision.detection.fasterrcnn_resnet50_fpn.pth
+--runner=DataflowRunner
+--experiments=use_runner_v2

.test-infra/tools/refresh_looker_metrics.py

@@ -43,6 +43,24 @@
     ("82", ["263", "264", "265", "266", "267"]),  # PyTorch Sentiment Streaming DistilBERT base uncased
     ("85", ["268", "269", "270", "271", "272"]),  # PyTorch Sentiment Batch DistilBERT base uncased
     ("86", ["284", "285", "286", "287", "288"]),  # VLLM Batch Gemma
+
+    # PyTorch Image Classification EfficientNet-B0 Streaming (Right-fit) CPU
+    ("92", ["289", "290", "291", "292", "293"]),
+    #TODO: PyTorch Image Classification EfficientNet-B0 Streaming (Right-fit) GPU
+    #TODO: PyTorch Image Classification EfficientNet-B0 Streaming (Right-fit Exactly-once) CPU
+    #TODO: PyTorch Image Classification EfficientNet-B0 Streaming (Right-fit Exactly-once) GPU
+
+    # PyTorch Image Object Detection Faster R-CNN ResNet-50 Batch CPU
+    ("93", ["294", "295", "296", "298", "299"]),
+    #TODO: PyTorch Image Object Detection Faster R-CNN ResNet-50 Batch GPU
+    #TODO: PyTorch Image Object Detection Faster R-CNN ResNet-50 Streaming CPU
+    #TODO: PyTorch Image Object Detection Faster R-CNN ResNet-50 Streaming GPU
+
+    # PyTorch Image Captioning BLIP + CLIP Batch CPU
+    ("94", ["297", "300", "301", "302", "303"]),
+    #TODO: PyTorch Image Captioning BLIP + CLIP Batch GPU
+    #TODO: PyTorch Image Captioning BLIP + CLIP Streaming CPU
+    #TODO: PyTorch Image Captioning BLIP + CLIP Streaming GPU
 ]