http://pipeline.ai
Applying my Netflix experience to a real-world problem in the ML and AI world, I will demonstrate a full-featured, open-source, end-to-end TensorFlow Model Training and Deployment System using the latest advancements from Kubernetes, Istio, and TensorFlow.
In addition to training and hyper-parameter tuning, our model deployment pipeline will include continuous canary deployments of our TensorFlow Models into a live, hybrid-cloud production environment.
This is the holy grail of data science - rapid and safe experiments of ML / AI models directly in production.
Following the Successful Netflix Culture that I lived and breathed (https://www.slideshare.net/reed2001/culture-1798664/2-Netflix_CultureFreedom_Responsibility2), I give Data Scientists the Freedom and Responsibility to extend their ML / AI pipelines and experiments safely into production.
Offline, batch training and validation is for the slow and weak. Online, real-time training and validation on live production data is for the fast and strong.
Learn to be fast and strong by attending this talk.
http://pipeline.ai
Building Google Cloud ML Engine From Scratch on AWS with PipelineAI - ODSC London 2017 - Oct 13, 2017
1. BUILDING GOOGLE CLOUD ML ENGINE
FROM SCRATCH WITH PIPELINE.AI
ODSC CONFERENCE
LONDON, ENGLAND
OCTOBER 13, 2017
CHRIS FREGLY,
FOUNDER @ PIPELINE.AI
2. INTRODUCTIONS: ME
§ Chris Fregly, Research Engineer @ PipelineAI
§ Formerly Netflix, Databricks, IBM Spark Center
§ Advanced Spark and TensorFlow Meetup
Please Join Our 40,000+ Members Globally!
Contact Me
chris@pipeline.ai
@cfregly
*San Francisco
*Chicago
*Austin
*Washington DC
*London
3. INTRODUCTIONS: YOU
§ Software Engineer or Data Scientist interested in optimizing
and deploying TensorFlow models to production
§ Assume you have a working knowledge of TensorFlow
4. CONTENT BREAKDOWN
§ PipelineAI Features
§ 50% Training Optimizations (GPUs, Pipeline, XLA+JIT)
§ 50% Prediction Optimizations (XLA+AOT, TF Serving)
§ Why Heavy Focus on Predicting?
§ Training: boring batch O(num_data_scientists)
§ Inference: exciting real-time O(num_users_of_app)
5. 100% OPEN SOURCE CODE
§ https://github.com/PipelineAI/pipeline/
§ Please 🌟 this GitHub Repo!
§ All slides, code, notebooks, and Docker images here:
https://github.com/PipelineAI/pipeline/tree/master/gpu
6. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
8. EXPERIMENT SAFELY IN PRODUCTION
§ Setup Experiments Directly from Jupyter Notebooks
§ Deploy to 1% Prod Traffic
§ Or Deploy in Shadow Mode
§ Tear-Down Experiments Quickly
9. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
10. MODEL + RUNTIME PACKAGING
§ Package Model + Runtime into Immutable Docker Image
§ Same Package: Local, Dev, and Prod
§ No Dependency Surprises in Production
11. OPTIMIZE MODEL + RUNTIME AS ONE
§ Tune Model Params + Runtime Configs Together
§ Generate Native CPU + GPU Code
§ Quantize Model Weights + Activations
§ Swap Runtimes: TF Serving, TensorRT, CPU, GPU, TPU
12. NVIDIA TENSORRT RUNTIME
§ Performs Post-Training Optimizations
§ GPU-Optimized Prediction Runtime
§ Alternative to TensorFlow Serving
13. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
20. CONTINUOUS MODEL TRAINING
§ Identify and Fix Borderline Predictions (50-50% Confidence)
§ Fix Along Class Boundaries
§ Retrain on New Labeled Data
§ Enables Crowd Sourcing
§ Game-ify Labeling Process
21. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
22. SETTING UP TENSORFLOW WITH GPUS
§ Very Painful!
§ Especially inside Docker
§ Use nvidia-docker
§ Especially on Kubernetes!
§ Use Kubernetes 1.7+
§ http://pipeline.ai for GitHub + DockerHub Links
23. GPU HALF-PRECISION SUPPORT
§ FP32 is “Full Precision”, FP16 is “Half Precision”
§ Supported by Pascal P100 (2016) and Volta V100 (2017)
§ Flexible FP32 GPU Cores Can Fit 2 FP16’s for 2x Throughput!
§ Half-Precision is OK for Approximate Deep Learning Use Cases
24. VOLTA V100 RECENTLY ANNOUNCED
§ 84 Streaming Multiprocessors (SM’s)
§ 5,376 GPU Cores
§ 672 Tensor Cores (ie. Google TPU)
§ Mixed FP16/FP32 Precision
§ More Shared Memory
§ New L0 Instruction Cache
§ Faster L1 Data Cache
§ V100 vs. P100 Performance
§ 12x TFLOPS @ Peak Training
§ 6x Inference Throughput
25. V100 AND CUDA 9
§ Independent Thread Scheduling - Finally!!
§ Similar to CPU fine-grained thread synchronization semantics
§ Allows GPU to yield execution of any thread
§ Still Optimized for SIMT (Same Instruction Multiple Thread)
§ SIMT units automatically scheduled together
§ Explicit Thread Synchronization
P100 V100
26. GPU CUDA PROGRAMMING
§ Barbaric, But Fun Barbaric
§ Must Know Hardware Very Well
§ Hardware Changes are Painful
§ Many Great Debuggers Exist
27. CUDA STREAMS
§ Asynchronous I/O Transfer
§ Overlap Compute and I/O
§ Keeps GPUs Saturated
§ Fundamental to Queue Framework in TensorFlow
28. TRAINING TERMINOLOGY
§ Tensors: N-Dimensional Arrays
§ ie. Scalar, Vector, Matrix
§ Operations: MatMul, Add, SummaryLog,…
§ Graph: Graph of Operations (DAG)
§ Session: Contains Graph(s)
§ Feeds: Feed inputs into Placeholder
§ Fetches: Fetch output from Operation
§ Variables: What we learn through training
§ aka “weights”, “parameters”
§ Devices: Hardware device on which we train
-TensorFlow-
Trains
Variables
-User-
Fetches
Outputs
-User-
Feeds
Inputs
-TensorFlow-
Performs
Operations
-TensorFlow-
Flows
Tensors
with tf.device(“/gpu:0,/gpu:1”)
29. TENSORFLOW MODEL
§ MetaGraph
§ Combines GraphDef and Metadata
§ GraphDef
§ Architecture of your model (nodes, edges)
§ Metadata
§ Asset: Accompanying assets to your model
§ SignatureDef: Maps external : internal tensors
§ Variables
§ Stored separately during training (checkpoint)
§ Allows training to continue from any checkpoint
§ Variables are “frozen” into Constants when deployed for inference
GraphDef
x
W
mul add
b
MetaGraph
Metadata
Assets
SignatureDef
Tags
Version
Variables:
“W” : 0.328
“b” : -1.407
32. DON’T USE FEED_DICT
§ Not Optimized for Production Pipelines
§ feed_dict Requires Python <-> C++ Serialization
§ Single-threaded, Synchronous, SLOW!
§ Can’t Retrieve Until Current Batch is Complete
§ CPUs/GPUs Not Fully Utilized!
§ Use Queue or Dataset API
33. QUEUES
§ More than just a traditional Queue
§ Perform I/O, pre-processing, cropping, shuffling
§ Pulls from HDFS, S3, Google Storage, Kafka, ...
§ Combine many small files into large TFRecord files
§ Use CPUs to free GPUs for compute
§ Uses CUDA Streams
§ Helps saturate CPUs and GPUs
34. QUEUE CAPACITY PLANNING
§ batch_size
§ # examples / batch (ie. 64 jpg)
§ Limited by GPU RAM
§ num_processing_threads
§ CPU threads pull and pre-process batches of data
§ Limited by CPU Cores
§ queue_capacity
§ Limited by CPU RAM (ie. 5 * batch_size)
35. DETECT UNDERUTILIZED CPUS, GPUS
§ Instrument training code to generate “timelines”
§ Analyze with Google Web
Tracing Framework (WTF)
§ Monitor CPU with `top`, GPU with `nvidia-smi`
http://google.github.io/tracing-framework/
from tensorflow.python.client import timeline
trace =
timeline.Timeline(step_stats=run_metadata.step_stats)
with open('timeline.json', 'w') as trace_file:
trace_file.write(
trace.generate_chrome_trace_format(show_memory=True))
36. SINGLE NODE, MULTI-GPU TRAINING
§ cpu:0
§ By default, all CPUs
§ Requires extra config to target a CPU
§ gpu:0..n
§ Each GPU has a unique id
§ TF usually prefers a single GPU
§ xla_cpu:0, xla_gpu:0..n
§ “JIT Compiler Device”
§ Hints TensorFlow to attempt JIT Compile
with tf.device(“/cpu:0”):
with tf.device(“/gpu:0”):
with tf.device(“/gpu:1”):
GPU 0 GPU 1
37. MULTI-NODE DISTRIBUTED TRAINING
§ TensorFlow Automatically Inserts Send and Receive Ops into Graph
§ Parameter Server Synchronously Aggregates Updates to Variables
§ Nodes with Multiple GPUs will Pre-Aggregate Before Sending to PS
Worker0 Worker0
Worker1
Worker0 Worker1 Worker2
gpu0 gpu1
gpu2 gpu3
gpu0 gpu1
gpu2 gpu3
gpu0 gpu1
gpu2 gpu3
gpu0
gpu1
gpu0
gpu0
38. SYNCHRONOUS VS. ASYNCHRONOUS
§ Synchronous
§ Nodes compute gradients
§ Nodes update Parameter Server (PS)
§ Nodes sync on PS for latest gradients
§ Asynchronous
§ Some nodes delay in computing gradients
§ Nodes don’t update PS
§ Nodes get stale gradients from PS
§ May not converge due to stale reads!
39. BATCH NORMALIZATION
§ Each mini-batch may have wildly different distributions
§ Normalize per batch (and layer)
§ Speeds up training!!
§ Weights are learned quicker
§ Final model is more accurate
§ Final mean and variance will be folded into Graph later
-- Pretty Much Always Use Batch Normalization! --
z = tf.matmul(a_prev, W)
a = tf.nn.relu(z)
a_mean, a_var = tf.nn.moments(a, [0])
scale = tf.Variable(tf.ones([depth/channels]))
beta = tf.Variable(tf.zeros ([depth/channels]))
bn = tf.nn.batch_normalizaton(a, a_mean, a_var,
beta, scale, 0.001)
40. OPTIMIZE GRAPH EXECUTION ORDER
§ https://github.com/yaroslavvb/stuff
Linearize to
minimize graph
memory usage
41. SEPARATE TRAINING + VALIDATION
§ Separate Training and Validation Clusters
§ Validate Upon Checkpoint
§ Avoids Resource Contention
Training
Cluster
Validation
Cluster
Parameter Server
Cluster
42. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
43. XLA FRAMEWORK
§ Accelerated Linear Algebra (XLA)
§ Goals:
§ Reduce reliance on custom operators
§ Improve execution speed
§ Improve memory usage
§ Reduce mobile footprint
§ Improve portability
§ Helps TensorFlow Stay Both Flexible and Performant
44. XLA HIGH LEVEL OPTIMIZER (HLO)
§ Compiler Intermediate Representation (IR)
§ Independent of Source and Target Language
§ Define Graphs using HLO Operations
§ XLA Step 1 Emits Target-Independent HLO
§ XLA Step 2 Emits Target-Dependent LLVM
§ LLVM Emits Native Code Specific to Target
§ Supports x86-64, ARM64 (CPU), and NVPTX (GPU)
45. JIT COMPILER
§ Just-In-Time Compiler
§ Built on XLA Framework
§ Goals:
§ Reduce memory movement – especially useful on GPUs
§ Reduce overhead of multiple function calls
§ Similar to Spark Operator Fusing in Spark 2.0
§ Unroll Loops, Fuse Operators, Fold Constants, …
§ Scope to session, device, or `with jit_scope():`
46. VISUALIZING JIT COMPILER IN ACTION
Before After
Google Web Tracing Framework:
http://google.github.io/tracing-framework/
from tensorflow.python.client import timeline
trace =
timeline.Timeline(step_stats=run_metadata.step_stats)
with open('timeline.json', 'w') as trace_file:
trace_file.write(
trace.generate_chrome_trace_format(show_memory=True))
48. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
49. AOT COMPILER
§ Standalone, Ahead-Of-Time (AOT) Compiler
§ Built on XLA framework
§ tfcompile
§ Creates executable with minimal TensorFlow Runtime needed
§ Includes only dependencies needed by subgraph computation
§ Creates functions with feeds (inputs) and fetches (outputs)
§ Packaged as cc_libary header and object files to link into your app
§ Commonly used for mobile device inference graph
§ Currently, only CPU x86-64 and ARM are supported - no GPU
50. GRAPH TRANSFORM TOOL (GTT)
§ Optimize Trained Models for Inference
§ Remove training-only Ops (checkpoint, drop out, logs)
§ Remove unreachable nodes between given feed -> fetch
§ Fuse adjacent operators to improve memory bandwidth
§ Fold final batch norm mean and variance into variables
§ Round weights/variables improves compression (ie. 70%)
§ Quantize (FP32 -> INT8) to speed up math operations
53. AFTER STRIPPING UNUSED NODES
§ Optimizations
§ strip_unused_nodes
§ Results
§ Graph much simpler
§ File size much smaller
54. AFTER REMOVING UNUSED NODES
§ Optimizations
§ strip_unused_nodes
§ remove_nodes
§ Results
§ Pesky nodes removed
§ File size a bit smaller
55. AFTER FOLDING CONSTANTS
§ Optimizations
§ strip_unused_nodes
§ remove_nodes
§ fold_constants
§ Results
§ Placeholders (feeds) -> Variables*
(*Why Variables and not Constants?)
56. AFTER FOLDING BATCH NORMS
§ Optimizations
§ strip_unused_nodes
§ remove_nodes
§ fold_constants
§ fold_batch_norms
§ Results
§ Graph remains the same
§ File size approximately the same
57. WEIGHT QUANTIZATION
§ FP16 and INT8 Are Computationally Simpler and Faster
§ Weights/Variables are Constants
§ Easy to Linearly Quantize
58. AFTER QUANTIZING WEIGHTS
§ Optimizations
§ strip_unused_nodes
§ remove_nodes
§ fold_constants
§ fold_batch_norms
§ quantize_weights
§ Results
§ Graph is same, file size is smaller, compute is faster
60. ACTIVATION QUANTIZATION
§ Activations Not Known Ahead of Time
§ Depends on input, not easy to quantize
§ Requires Additional Calibration Step
§ Use a “representative” dataset
§ Per Neural Network Layer…
§ Collect histogram of activation values
§ Generate many quantized distributions with different saturation thresholds
§ Choose threshold to minimize…
KL_divergence(ref_distribution, quant_distribution)
§ Not Much Time or Data is Required (Minutes on Commodity Hardware)
63. AGENDA
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
64. TENSORFLOW SERVING OVERVIEW
§ Inference
§ Only Forward Propagation through Network
§ Predict, Classify, Regress, …
§ Bundle
§ GraphDef, Variables, Metadata, …
§ Assets
§ ie. Map of ClassificationID -> String
§ {9283: “penguin”, 9284: “bridge”}
§ Version
§ Every Model Has a Version Number (Integer)
§ Version Policy
§ ie. Serve Only Latest (Highest), Serve Both Latest and Previous, …
65. MULTI-HEADED INFERENCE
§ Multiple “heads” (aka “responses”) from 1 model prediction
§ Optimizes bandwidth, CPU, latency, memory, coolness
§ Response includes both class and scores
§ Inputs sent only once
§ Feed scores into ensemble models
§ Use model for feature engineering
66. REQUEST BATCHING
§ max_batch_size
§ Enables throughput/latency tradeoff
§ Bounded by RAM
§ batch_timeout_micros
§ Defines batch time window, latency upper-bound
§ Bounded by RAM
§ num_batch_threads
§ Defines parallelism
§ Bounded by CPU cores
§ max_enqueued_batches
§ Defines queue upper bound, throttling
§ Bounded by RAM
Reaching either threshold
will trigger a batch
67. YOU JUST LEARNED…
§ PipelineAI Features
§ Experiment Safely in Production
§ Tune Both Model + Runtime Parameters
§ Compare Models Both Offline + Online
§ Shift Traffic (Across Clouds) to Winning Model
§ Optimize TensorFlow Training
§ GPUs + Ingestion + Training Pipeline
§ XLA JIT Compiler
§ Optimize TensorFlow Inference
§ XLA AOT Compiler + Graph Transform Tool (GTT)
§ TensorFlow Serving
68. THANKS! ANY QUESTIONS?
§ https://github.com/PipelineAI/pipeline/
§ Please 🌟 this GitHub Repo!
§ All slides, code, notebooks, and Docker images here:
https://github.com/PipelineAI/pipeline/tree/master/gpu
Contact Me
chris@pipeline.ai
@cfregly