Optimizing, Profiling, and Deploying TensorFlow AI Models with GPUs - San Francisco Python Meetup -...

HIGH PERFORMANCE TENSORFLOW IN PRODUCTION WITH GPUSSF PYTHON MEETUP NOV 8, 2017SPECIAL THANKS TO YELP!! !!CHRIS FREGLY, FOUNDER @PIPELINE.AI

INTRODUCTIONS: ME§ Chris Fregly, Founder & Engineer @ PipelineAI

§ Formerly Netflix and Databricks

§ Advanced Spark and TensorFlow MeetupPlease Join Our 50,000+ Global Members!!

Contact [email protected]

@cfregly

* San Francisco

* Chicago* Austin* Washington DC* Dusseldorf* London

INTRODUCTIONS: YOU§ Software Engineer, Data Scientist, Data Engineer, Data Analyst

§ Interested in Optimizing and Deploying TF Models to Production

§ Nice to Have a Working Knowledge of TensorFlow (Not Required)

CONTENT BREAKDOWN50% Training Optimizations (GPUs, Training Pipeline, JIT)50% Prediction Optimizations (AOT Compile, TF Serving)

Why Heavy Focus on Model Prediction vs. Just Training?

10s of Data Scientists <<< Millions of App Users

TrainingBoring & Batch

PredictionExciting & Real-Time!!

AGENDA

Part 0: Latest PipelineAI Research

Part 1: Optimize TensorFlow Model Training

Part 2: Optimize TensorFlow Model Serving

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.ml

https://github.com/rviscomi/red-dwarf

HANDS-ON EXERCISES§ Combo of Jupyter Notebooks and Command Line§ Command Line through Jupyter Terminal

§ Some Exercises Based on Experimental Features

You May See Errors. Stay Calm. You Will Be OK!!

PIPELINE.AI OVERVIEW400,000 Docker Downloads50,000 Users registered for

PipelineAI GA Release2,000 GitHub Stars

15 Enterprise Beta Users

AGENDA


§ Package, Deploy, and Tune Both Model + Runtime§ Deploy Models and Experiments Safely to Prod§ Compare Models Both Offline and Online§ Auto-Shift Traffic to Winning Model or Cloud

PACKAGE MODEL + RUNTIME AS ONE§ Package Model + Runtime into Immutable Docker Image§ Same Environment: Local, Dev, and Prod§ No Dependency Surprises in Production§ Deploy and Tune Model + Runtime Together

pipeline predict-server-build --model-type=tensorflow \--model-name=mnist \--model-tag=”c" \--model-path=./models/tensorflow/mnist

Package Model Server C Locally

pipeline predict-server-push --model-type=tensorflow \--model-name=mnist \--model-tag=”c” \

Push Image C To Docker Registry

TUNE MODEL + RUNTIME TOGETHER

§ Try Different Model Hyper-Parameters + Runtime Configs§ Even Different Runtimes: TF Serving, TensorRT§ Auto-Quantize Model Weights + Activations§ Auto-Fuse Neural Network Layers Together§ Generate Native CPU + GPU Code

pipeline predict-server-start --model-type=tensorflow \--model-name=mnist \--model-tag=”c"

Start ModelServer C Locally

LOAD TEST MODEL + RUNTIME LOCALLY

§ Perform Mini-Load Test on Local Model Server§ Provides Immediate Feedback on Prediction Performance§ Relative Performance Compared to Other Variations§ No Need to Deploy to Test or Prod for Prediction Metrics§ See Where Time is Being Spent During Prediction

pipeline predict --model-server-url=http://localhost:6969 \--model-type=tensorflow \--model-name=mnist \--model-tag=”c”--test-request-concurrency=1000

Load Test ModelServer C Locally

RUNTIME OPTION: NVIDIA TENSOR-RT

§ GPU-Optimized Prediction Runtime§ Alternative to TensorFlow Serving

§ Post-Training Model Optimizations§ Similar to TF Graph Transform Tool

§ PipelineAI Supports TensorRT!

AGENDA



DEPLOY MODELS SAFELY TO PROD§ Deploy from Jupyter Notebook in 1-Click § Deploy to 1-2% Split or Shadowed Traffic§ Tear-Down or Rollback Quickly§ Command Line Interface (CLI)

pipeline predict-cluster-start --model-type=tensorflow \--model-name=mnist \--model-tag=”b” \--traffic-split=“0.02”

Start ModelCluster B in Prod

pipeline predict-cluster-start --model-type=tensorflow \--model-name=mnist \--model-tag=”c” \--traffic-split=“0.01”

Start ModelCluster C in Prod

pipeline predict-cluster-start --model-type=tensorflow \--model-name=mnist \--model-tag=”a” \--traffic-split=“0.97”

Start ModelCluster A in Prod

Implementation Details…

No need to understand these.We do, so you don’t have to! J

DEPLOY EXPERIMENTS SAFELY TO PROD§ Create Experiments Directly from Jupyter or Command Line

§ Deploy Experiment

pipeline experiment-add --experiment-name=my_experiment \--model-type=tensorflow \--model-name=mnist \--model-tag=“a” \--traffic-split=“97%”

CLI

Drag n’ Drop

pipeline experiment-start --experiment-name=my_experiment \--traffic-shadow=“20%”

pipeline experiment-add --experiment-name=my_experiment \--model-type=tensorflow \--model-name=mnist \--model-tag=“b” \--traffic-split=“2%”

pipeline experiment-add --experiment-name=my_experiment \--model-type=tensorflow \--model-name=mnist \--model-tag=“c” \--traffic-split=“1%”

1-Click

Start Experimentwith 20% Shadowed

of Production Traffic

AGENDA



COMPARE MODELS OFFLINE & ONLINE§ Offline, Batch Metrics

§ Validation Accuracy§ Training Accuracy§ CPU/GPU Utilization

§ Actual, Live Predictions!§ Relative Prediction Precision

§ Online, Real-Time Metrics§ Response Time & Throughput§ Cost Per Prediction!

PREDICTION PROFILING AND TUNING§ Pinpoint Performance Bottlenecks

§ Fine-Grained Prediction Metrics

§ Three (3) Logic Prediction Steps1.transform_request()2.predict()3.transform_response()

VIEW REAL-TIME PREDICTION STREAM§ Visually Compare Real-Time Predictions

FeatureInputs

PredictionConfidence

CONTINUOUS MODEL TRAINING§ Identify and Fix Borderline Predictions (~50-50% Confidence)

§ Fix Along Class Boundaries

§ Retrain on New Labeled Data

§ Game-ify Labeling Process

§ Enables Crowd Sourcing

AGENDA



SHIFT TRAFFIC TO MAX(REVENUE)§ Shift Traffic to Winning Model using AI Bandit Algorithms

SHIFT TRAFFIC TO MIN(CLOUD COST)

§ Real-Time Cost Per Prediction

§ Across Clouds & On-Premise

§ Bandit-based Explore/Exploit

AGENDA




AGENDA


§ GPUs and TensorFlow§ Train, Inspect, and Debug TensorFlow Models§ TensorFlow Distributed Model Training on a Cluster§ Optimize Training with JIT XLA Compiler

EVERYBODY GETS A GPU!

SETUP ENVIRONMENT

§ Step 1: Browse to the following:http://allocator.community.pipeline.ai/allocate

§ Step 2: Browse to the following:http://<ip-address>

§ Step 3: Browse around. I will provide a Jupyter Username/Password soon.

Need Help? Use the Chat!

VERIFY SETUP

http://<ip-address>

Any username,Any password!

LET’S EXPLORE OUR ENVIRONMENT§ Navigate to the following notebook:

01_Explore_Environment

§ https://github.com/PipelineAI/pipeline/tree/master/gpu.ml/notebooks

PULSE CHECK

BREAK




SETTING UP TENSORFLOW WITH GPUS

§ Very Painful!

§ Especially inside Docker§ Use nvidia-docker

§ Especially on Kubernetes!§ Use Kubernetes 1.8+

§ http://pipeline.ai for GitHub + DockerHub Links

GPU HALF-PRECISION SUPPORT§ FP32 is “Full Precision”, FP16 is “Half Precision”§ Supported by Pascal P100 (2016) and Volta V100 (2017)§ Half-Precision is OK for Approximate Deep Learning Use Cases§ Fit Two(2) FP16’s into FP32 GPU Cores for 2x Throughput!

You Can Set TF_FP16_MATMUL_USE_FP32_COMPUTE=0

on GPU w/ Compute Capability(CC) 5.3+

VOLTA V100 (2017) VS. PASCAL P100 (2016)§ 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

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 Synchronization

P100 V100

GPU CUDA PROGRAMMING

§ Barbaric, But Fun Barbaric

§ Must Know Hardware Very Well

§ Hardware Changes are Painful

§ Use the Profilers & Debuggers

CUDA STREAMS

§ Asynchronous I/O Transfer

§ Overlap Compute and I/O

§ Keeps GPUs Saturated

§ Fundamental to Queue Framework in TensorFlow

LET’S SEE WHAT THIS THING CAN DO!§ Navigate to the following notebook:

01a_Explore_GPU01b_Explore_Numba


AGENDA



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 (GPU, CPU, TPU, ...)

-TensorFlow-Trains

Variables

-User-FetchesOutputs

-User-FeedsInputs

-TensorFlow-Performs

Operations

-TensorFlow-Flows

Tensors

with tf.device(“/cpu:0,/gpu:15”):

TENSORFLOW SESSION

Session

graph: GraphDef

Variables:“W” : 0.328“b” : -1.407

Variables are Randomly

Initialized, thenPeriodically

Checkpointed

GraphDef is Created During

Training, then Frozen for Inference

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 preparing for inference

GraphDef

x

W

mul add

b

MetaGraphMetadata

AssetsSignatureDef

TagsVersion

Variables:“W” : 0.328“b” : -1.407

BATCH NORMALIZATION (2015)

§ Each Mini-Batch May Have Wildly Different Distributions§ Normalize per Batch (and Layer)§ Faster Training, Learns Quicker§ Final Model is More Accurate§ TensorFlow is already on 2nd Generation Batch Algorithm§ First-Class Support for Fusing Batch Norm Layers§ Final mean + variance Are Folded Into Our Graph Later

-- (Almost)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)

DROPOUT (2014)§ Training Technique§ Prevents Overfitting§ Combines Exponential Set of Diff Neural Architectures

§ Inherent Ensembling, If You Will§ Expressed as Probability Percentage (ie. 50%)§ Boost Other Weights During Validation & Prediction

TrainingDropout

Validation & PredictionDropout

0%Dropout

50% Dropout

OPTIMIZE GRAPH EXECUTION ORDER

§ https://github.com/yaroslavvb/stuff

"Linearize” Causes TF to Minimize

GraphMemory Usage.

This is Useful on Single GPU with Relatively Low

RAM.

EXTEND EXISTING DATA PIPELINES§ Data Processing

§ HDFS/Hadoop§ Spark

§ Containers§ Docker

§ Schedulers§ Kubernetes§ Mesos

<dependency> <groupId>org.tensorflow</groupId> <artifactId>tensorflow-hadoop</artifactId>

</dependency>

https://github.com/tensorflow/ecosystem

DON’T USE FEED_DICT!!

§ feed_dict Requires Python <-> C++ Serialization§ Not Optimized for Production Ingestion Pipelines§ Retrieves Next Batch After Current Batch is Done§ Single-Threaded, Synchronous§ CPUs/GPUs Not Fully Utilized!§ Use Queue or Dataset API

sess.run(train_step, feed_dict={…}

QUEUES

§ More than traditional Queue§ Uses CUDA Streams§ Perform I/O, pre-processing, cropping, shuffling, …§ Pull from HDFS, S3, Google Storage, Kafka, ...§ Combine many small files into large TFRecord files§ Use CPUs to free GPUs for compute§ Helps saturate CPUs and GPUs

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)

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))

LET’S FEED DATA WITH A QUEUE§ Navigate to the following notebook:

02_Feed_Queue_HDFS


PULSE CHECK

BREAK




LET’S TRAIN A MODEL (CPU)§ Navigate to the following notebook:

03_Train_Model_CPU


LET’S TRAIN A MODEL (GPU)§ Navigate to the following notebook:

03a_Train_Model_GPU


TENSORFLOW DEBUGGER§ Step through Operations§ Inspect Inputs and Outputs§ Wrap Session in Debug Session

sess = tf.Session(config=config)sess =

tf_debug.LocalCLIDebugWrapperSession(sess)

LET’S DEBUG A MODEL§ Navigate to the following notebook:

04_Debug_Model


AGENDA



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

DISTRIBUTED, MULTI-NODE 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

SingleNode

MultipleNodes

DATA PARALLEL VS MODEL PARALLEL

§ Data Parallel (“Between-Graph Replication”)§ Send exact same model to each device§ Each device operates on partition of data

§ ie. Spark sends same function to many workers§ Each worker operates on their partition of data

§ Model Parallel (“In-Graph Replication”)§ Send different partition of model to each device§ Each device operates on all data

Very Difficult!, ButRequired for Large Models.(GPU RAM Limitation)

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!

CHIEF WORKER§ Worker Task 0 is Usually the Chief

§ Task 0 is guaranteed to exist

§ Performs Maintenance Tasks§ Writes log summaries§ Instructs PS to checkpoint vars§ Performs PS health checks§ (Re-)Initialize variables at (re-)start of training

NODE AND PROCESS FAILURES

§ Checkpoint to Persistent Storage (HDFS, S3)§ Use MonitoredTrainingSession and Hooks§ Use a Good Cluster Orchestrator (ie. Kubernetes,Mesos)§ Understand Failure Modes and Recovery States

Stateless, Not Bad: Training Continues Stateful, Bad: Training Must Stop Dios Mio! Long Night Ahead…

USE ESTIMATOR AND EXPERIMENT APIS§ Simplify Model Building § Provide Clear Path to Production§ Enable Rapid Model Experiments§ Provide Flexible Parameter Tuning§ Enable Downstream Optimizing & Serving Infra( )§ Nudge Users to Best Practices Through Opinions§ Provide Hooks/Callbacks to Override Opinions§ Unified API for Local and Distributed TensorFlow

https://arxiv.org/pdf/1708.02637.pdf

ESTIMATOR API§ “Train-to-Serve” Design§ Create Custom - or Use a Canned Estimator§ Hides Session, Graph, Layers, Iterative Loops (Train, Eval, Predict)§ Hooks for All Phases of Model Training and Evaluation

§ Load Input: input_fn()§ Train: model_fn() and train() § Evaluate: evaluate()§ Save and Export: export_savedmodel()§ Predict: predict() Uses sess.run() Slow Predictions!

Example:https://github.com/GoogleCloudPlatform/cloudml-samples/blob/master/census/customestimator/

LAYERS API§ Standalone Layer or Entire Sub-Graphs§ Functions of Tensor Inputs & Outputs§ Mix and Match with Operations§ Assumes 1st Dimension is Batch Size § Handles One (1) to Many (*) Inputs§ Special Types of Layers

§ Loss per Mini-Batch§ Accuracy and MSE Track Across Mini-Batches

CANNED ESTIMATORS

§ Commonly-Used Estimators§ Pre-Tested and Pre-Tuned§ DNNClassifer, TensorForestEstimator§ Always Use Canned Estimators If Possible§ Reduced Lines of Code, Complexity, and Bugs§ Use FeatureColumns to Define & Create Features

Custom vs. Canned@ Google, August, 2017

FEATURECOLUMN ABSTRACTION

§ Used by Canned Estimator§ Simplifies Input Ingestion§ Declarative Way to Specify Model Training Inputs§ Converts Sparse Features to Dense Tensors§ Sparse Features: Query Keyword, Url, ProductID,…§ Wide/Linear Models Use Feature-Crossing§ Deep Models Use Embeddings

SINGLE VS. MULTI-OBJECTIVES + HEADS

§ Single-Objective Estimator § Single classification prediction

§ Multi-Objective Estimator§ Two (2) classification predictions§ Or One (1) classification prediction + One(1) final layer

§ Multiple Heads Are Used to Ensemble Models§ Treats neural network as a feature engineering step!

EXPERIMENT API

§ Easier-to-Use Distributed TensorFlow§ Combines Estimator with input_fn() § Used for Training, Evaluation, & Hyper-Parameter Tuning§ Distributed Training Default to Data-Parallel & Async§ Cluster Configuration is Fixed at Start of Training Job§ No Auto-Scaling Allowed!!

ESTIMATOR & EXPERIMENT CONFIGS

§ TF_CONFIG § Special environment variable for config§ Defines ClusterSpec in JSON incl. master, workers, PS’s§ Must set ”{environment”:“cloud”} for distributed mode§ RunConfig: Defines checkpoint interval, output directory, …

§ HParams: Hyper-parameter tuning parameters and ranges§ learn_runner creates RunConfig before calling run() & tune()§ schedule is set based on {”task”:{”type”}}

§ Set to train_and_evaluate for local, single-node training

TF_CONFIG='{"environment": "cloud", "cluster":{"master":["worker0:2222”],"worker":["worker1:2222"],"ps": ["ps0:2222"]}, "task": {"type": "ps",

"index": "0"}}'

SEPARATE TRAINING + EVALUATION

§ Separate Training and Evaluation Clusters

§ Evaluate Upon Checkpoint

§ Avoid Resource Contention

§ Let Training Continue in Parallel with Evaluation

TrainingCluster

EvaluationCluster

Parameter ServerCluster

LET’S TRAIN DISTRIBUTED TENSORFLOW§ Navigate to the following notebook:

05_Train_Model_Distributed_CPUor 05a_Train_Model_Distributed_GPU


PULSE CHECK

BREAK




AGENDA



XLA FRAMEWORK§ Accelerated Linear Algebra (XLA)§ Goals:

§ Reduce reliance on custom operators§ Improve execution speed§ Improve memory usage§ Reduce mobile footprint§ Improve portability

§ Helps TF Stay Flexible and Performant

XLA HIGH LEVEL OPTIMIZER (HLO)

§ Compiler Intermediate Representation (IR)§ Independent of source and target language§ Define Graphs using HLO Language§ 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)

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():`

VISUALIZING JIT COMPILER IN ACTION

Before After

Google Web Tracing Framework:http://google.github.io/tracing-framework/

from tensorflow.python.client import timelinetrace = 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))

VISUALIZING FUSING OPERATORS

pip install graphviz

dot -Tpng \/tmp/hlo_graph_1.w5LcGs.dot \-o hlo_graph_1.png

GraphViz:http://www.graphviz.org

hlo_*.dot files generated by XLA

LET’S TRAIN WITH XLA CPU§ Navigate to the following notebook:

06_Train_Model_XLA_CPU


LET’S TRAIN WITH XLA GPU§ Navigate to the following notebook:

06a_Train_Model_XLA_GPU


AGENDA




AGENDA


§ AOT XLA Compiler and Graph Transform Tool§ Key Components of TensorFlow Serving§ Deploy Optimized TensorFlow Model§ Optimize TensorFlow Serving Runtime

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

GRAPH TRANSFORM TOOL (GTT)

§ Post-Training Optimization to Prepare 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 to improve compression (ie. 70%)§ Quantize (FP32 -> INT8) to Speed Up Math Operations

BEFORE OPTIMIZATIONS

GRAPH TRANSFORM TOOLtransform_graph \--in_graph=tensorflow_inception_graph.pb \ ß Original Graph--out_graph=optimized_inception_graph.pb \ ß Transformed Graph--inputs='Mul' \ ß Feed (Input)--outputs='softmax' \ ß Fetch (Output) --transforms=' ß List of Transforms strip_unused_nodesremove_nodes(op=Identity, op=CheckNumerics) fold_constants(ignore_errors=true) fold_batch_normsfold_old_batch_normsquantize_weightsquantize_nodes'

AFTER STRIPPING UNUSED NODES

§ Optimizations§ strip_unused_nodes

§ Results§ Graph much simpler§ File size much smaller

AFTER REMOVING UNUSED NODES

§ Optimizations§ strip_unused_nodes§ remove_nodes

§ Results§ Pesky nodes removed§ File size a bit smaller

AFTER FOLDING CONSTANTS

§ Optimizations§ strip_unused_nodes§ remove_nodes§ fold_constants

§ Results§ Placeholders (feeds) -> Variables*

(*Why Variables and not Constants?)

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

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

WEIGHT QUANTIZATION

§ FP16 and INT8 Are Smaller and Computationally Simpler§ Weights/Variables are Constants§ Easy to Linearly Quantize

LET’S OPTIMIZE FOR INFERENCE§ Navigate to the following notebook:

07_Optimize_Model**Why just CPU version? Why not GPU?


BUT WAIT, THERE’S MORE!

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)

AFTER ACTIVATION QUANTIZATION

§ Optimizations§ strip_unused_nodes§ remove_nodes§ fold_constants§ fold_batch_norms§ quantize_weights§ quantize_nodes (activations)

§ Results§ Larger graph, needs calibration!

Requires additional freeze_requantization_ranges

LET’S OPTIMIZE FOR INFERENCE§ Navigate to the following notebook:

08_Optimize_Model_Activations


FREEZING MODEL FOR DEPLOYMENT§ Optimizations

§ strip_unused_nodes§ remove_nodes§ fold_constants§ fold_batch_norms§ quantize_weights§ quantize_nodes§ freeze_graph

§ Results§ Variables -> Constants

Finally!We’re Ready to Deploy!!

AGENDA



MODEL SERVING TERMINOLOGY§ 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, …

TENSORFLOW SERVING FEATURES§ Supports Auto-Scaling§ Custom Loaders beyond File-based§ Tune for Low-latency or High-throughput§ Serve Diff Models/Versions in Same Process§ Customize Models Types beyond HashMap and TensorFlow§ Customize Version Policies for A/B and Bandit Tests§ Support Request Draining for Graceful Model Updates§ Enable Request Batching for Diff Use Cases and HW§ Supports Optimized Transport with GRPC and Protocol Buffers

PREDICTION SERVICE§ Predict (Original, Generic)

§ Input: List of Tensor§ Output: List of Tensor

§ Classify§ Input: List of tf.Example (key, value) pairs§ Output: List of (class_label: String, score: float)

§ Regress§ Input: List of tf.Example (key, value) pairs§ Output: List of (label: String, score: float)

PREDICTION INPUTS + OUTPUTS§ SignatureDef

§ Defines inputs and outputs§ Maps external (logical) to internal (physical) tensor names§ Allows internal (physical) tensor names to change

from tensorflow.python.saved_model import utilsfrom tensorflow.python.saved_model import signature_constantsfrom tensorflow.python.saved_model import signature_def_utils

graph = tf.get_default_graph()x_observed = graph.get_tensor_by_name('x_observed:0') y_pred = graph.get_tensor_by_name('add:0') inputs_map = {'inputs': x_observed} outputs_map = {'outputs': y_pred} predict_signature = signature_def_utils.predict_signature_def(inputs=inputs_map,

outputs=outputs_map)

MULTI-HEADED INFERENCE

§ Inputs Pass Through Model Once§ Model Returns Multiple Predictions or “Heads” including:

1. Human-readable prediction (ie. “penguin”, “church”,…)2. Final layer of scores (float vector)

§ Final Layer of floats Pass to the Next Model in Ensemble§ Optimizes Bandwidth, CPU/GPU, Latency, Memory§ Enables Complex Model Composing and Ensembling

BUILD YOUR OWN MODEL SERVER§ Adapt GRPC(Google) <-> HTTP (REST of the World)§ Perform Batch Inference vs. Request/Response§ Handle Requests Asynchronously§ Support Mobile, Embedded Inference§ Customize Request Batching§ Add Circuit Breakers, Fallbacks§ Control Latency Requirements§ Reduce Number of Moving Parts

#include “tensorflow_serving/model_servers/server_core.h”

class MyTensorFlowModelServer {ServerCore::Options options;

// set options (model name, path, etc)std::unique_ptr<ServerCore> core;

TF_CHECK_OK(ServerCore::Create(std::move(options), &core)

);}

Compile and Link withlibtensorflow.so

NVIDIA TENSOR-RT RUNTIME

§ Post-Training Model Optimizations§ Similar to TF Graph Transform Tool

§ GPU-Optimized Prediction Runtime§ Alternative to TensorFlow Serving

§ PipelineAI Supports TensorRT!

AGENDA



SAVED MODEL FORMAT§ Navigate to the following notebook:

09_Deploy_Optimized_Model


AGENDA



REQUEST BATCH TUNING§ 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 thresholdwill trigger a batch

ADVANCED BATCHING & SERVING TIPS§ Batch Just the GPU/TPU Portions of the Computation Graph

§ Batch Arbitrary Sub-Graphs using Batch / Unbatch Graph Ops§ Distribute Large Models Into Shards Across TensorFlow Model Servers§ Batch RNNs Used for Sequential and Time-Series Data§ Find Best Batching Strategy For Your Data Through Experimentation

§ BasicBatchScheduler: Homogeneous requests (ie Regress or Classify)§ SharedBatchScheduler: Mixed requests, multi-step, ensemble predict§ StreamingBatchScheduler: Mixed CPU/GPU/IO-bound Workloads

§ Serve Only One (1) Model Inside One (1) TensorFlow Serving Process§ Much Easier to Debug, Tune, Scale, and Manage Models in Production.

LET’S DEPLOY OPTIMIZED MODEL§ Navigate to the following notebook:

10_Optimize_Model_Server


AGENDA




THANK YOU!! QUESTIONS?§ https://github.com/PipelineAI/pipeline/



Contact [email protected]

@cfregly

Date post:	22-Jan-2018
Category:	Data & Analytics
Upload:	chris-fregly
View:	486 times
Download:	4 times

Optimizing, Profiling, and Deploying TensorFlow AI Models with GPUs - San Francisco Python Meetup -...

Data & Analytics