Google_Cloud_for_Machine_Learning_-_Sungmin_Han__Cloud_Next_.pdf

Transcript

1. MLOps lead, Riiid Google Developers Expert for ML Sungmin Han Google Cloud for Machine Learning
2. 01 02 Cloud TPU GPUs in Google Cloud
3. 03 VertexAI
4. GPUs in Google Cloud Training resources
5. 2018. Oct BERT 340M parameters 2019. Oct T5 770M parameters 2019. Feb GPT-2 1.5B parameters 2020. Jun GPT-3 175B parameters 2022. Apr DALL-E 2 3.5B parameters 2021. Jan DALL-E 12B parameters And more..
7. Compute Engine In preview GPU on Google Cloud
8. NVIDIA A100 NVIDIA T4 NVIDIA V100 NVIDIA P100 NVIDIA P4 NVIDIA K80 HBM2 40GB, 80GB GDDR6 16GB HBM2 16GB HBB2 8GB GDDR5 8GB GDDR5 12GB
9. A100 Multi-region support
10. https://cloud.google.com/blog/products/compute/a2-vms-with-nvidia-a100-gpus-are-ga
11. https://developer.nvidia.com/blog/introducing-hgx-a100-most-powerful-accelerated-server-platform-for-ai-hpc/
12. https://cloud.google.com/blog/products/compute/a2-vms-with-nvidia-a100-gpus-are-ga
13. GDPR CCPA COPPA SOC1, 2, 3 ISO/IEC 27701 ISO/IEC 27018 HIPAA General Data Protection Regulation (유럽) California Consumer Privacy Act (USA) Children's Online Privacy Protection Act General / Global System Security Certification General / Global Privacy Certification General / Global Cloud Privacy Certification Health Insurance Portability and Accountability Act
14. Cloud Billing Monitoring With BigQuery
16. 33% Cheaper by using preemptible (Spot Instance)
17. On-demand Multi-region support Dashboard support (BigQuery) 01 02 03 04 No setup required (IPMI, Firewall device..)
18. On-demand Multi-region support Dashboard support (BigQuery) 01 02 03 No setup required (IPMI, Firewall device..) 04
19. On-demand Multi-region support Dashboard support (BigQuery) 01 02 03 No setup required (IPMI, Firewall device..) 04
20. On-demand Multi-region support Dashboard support (BigQuery) 01 02 03 No setup required (IPMI, Firewall device..) 04
21. Cloud TPU Tensor Processing Unit
22. https://cloud.google.com/blog/products/ai-machine-learning/cloud-tpu-v4-mlperf-2-0-results
24. TPUv3 consists with 2 Systolic array With 128x128 ALUs. https://cloud.google.com/tpu/docs/intro-to-tpu?hl=ko
25. By replacing to TPU Saving costs 38% (compare with V100)
26. Suitable for long- term training of several weeks or longer
27. VertexAI Subheading text
28. https://codelabs.developers.google.com/vertex-pipelines-intro#1
29. NVIDIA A100 NVIDIA T4 NVIDIA V100 NVIDIA P100 NVIDIA P4 NVIDIA K80
30. https://cloud.google.com/blog/topics/developers-practitioners/pytorch-google-cloud-how-train-and-tune-pytorch-models-vertex-ai
31. VertexAI Pipeline
32. https://www.kubeflow.org/docs/components/pipelines/v1/introduction/ Kubeflow Pipeline
33. # keras model python path including run_fn(). module_file=os.path.join(MODULE_ROOT, _trainer_module_file) example_gen = tfx.components.CsvExampleGen(input_base=data_root) trainer = tfx.extensions.google_cloud_ai_platform.Trainer( module_file=module_file, examples=example_gen.outputs['examples'], train_args=tfx.proto.TrainArgs(num_steps=100), eval_args=tfx.proto.EvalArgs(num_steps=5), ) pusher = tfx.extensions.google_cloud_ai_platform.Pusher( model=trainer.outputs['model'] ) components = [example_gen, trainer, pusher] return tfx.dsl.Pipeline( pipeline_name=pipeline_name, pipeline_root=pipeline_root, components=components)
34. client_options = { 'api_endpoint': GOOGLE_CLOUD_REGION + '-aiplatform.googleapis.com' } client = aiplatform.gapic.PredictionServiceClient(client_options=client_options) instances = [{ 'culmen_length_mm':[0.71], 'culmen_depth_mm':[0.38], 'flipper_length_mm':[0.98], 'body_mass_g': [0.78], }] endpoint = client.endpoint_path( project=GOOGLE_CLOUD_PROJECT, location=GOOGLE_CLOUD_REGION, endpoint=ENDPOINT_ID, ) response = client.predict(endpoint=endpoint, instances=instances) print('species:', np.argmax(response.predictions[0]))
35. Serving (Endpoint) Registry (Model registry)
36. Thank you!