Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
Chris Fregly authored Nov 11, 2017
1 parent 1b0013e commit 7454fc1
Showing 1 changed file with 84 additions and 25 deletions.
109 changes: 84 additions & 25 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -37,33 +37,33 @@ Use the same Docker Image from Local Laptop to Production to avoid dependency su

More [model samples](https://github.com/PipelineAI/models) coming soon (ie. R).

![Nvidia GPU](http://pipeline.ai/img/nvidia-cuda-338x181.png) ![TensorFlow](http://pipeline.ai/img/tensorflow-logo-202x168.png)
![Nvidia GPU](http://pipeline.ai/assets/img/nvidia-cuda-338x181.png) ![TensorFlow](http://pipeline.ai/assets/img/tensorflow-logo-202x168.png)

![Spark ML](http://pipeline.ai/img/spark-logo-24x163.png) ![Scikit-Learn](http://pipeline.ai/img/scikit-logo-277x150.png)
![Spark ML](http://pipeline.ai/assets/img/spark-logo-254x163.png) ![Scikit-Learn](http://pipeline.ai/assets/img/scikit-logo-277x150.png)

![R](http://pipeline.ai/img/r-logo-280x212.png) ![PMML](http://pipeline.ai/img/pmml-logo-210x96.png)
![R](http://pipeline.ai/assets/img/r-logo-280x212.png) ![PMML](http://pipeline.ai/img/pmml-logo-210x96.png)

![Xgboost](http://pipeline.ai/img/xgboost-logo-280x120.png) ![Ensembles](http://pipeline.ai/img/ensemble-logo-285x125.png)
![Xgboost](http://pipeline.ai/assets/img/xgboost-logo-280x120.png) ![Ensembles](http://pipeline.ai/assets/img/ensemble-logo-285x125.png)

# Pre-Requisites
## Docker
* Install [Docker](https://www.docker.com/community-edition#/download)

## Python3 (Conda is Optional)
* Install [Miniconda](https://conda.io/docs/install/quick.html) with Python3 Support
## Python2 or Python3 (Conda is Optional)
* Install [Miniconda](https://conda.io/docs/install/quick.html) with Python3 (Preferred) Support

## Install PipelineCLI
_Note: This command line interface requires **Python3** and **Docker** as detailed above._
```
pip install cli-pipeline==1.3.10 --ignore-installed --no-cache -U
pip install cli-pipeline==1.3.16 --ignore-installed --no-cache -U
```

## Verify Successful PipelineCLI Installation
```
pipeline version
### EXPECTED OUTPUT ###
cli_version: 1.3.10 <-- MAKE SURE YOU ARE ON THIS VERSION OR BAD THINGS MAY HAPPEN!
cli_version: 1.3.16 <-- MAKE SURE YOU ARE ON THIS VERSION OR BAD THINGS MAY HAPPEN!
api_version: v1
capabilities_enabled: ['predict_server', 'predict', 'version']
Expand All @@ -73,6 +73,12 @@ Email `[email protected]` to enable the advanced capabilities.
```

## Review CLI Functionality
[Community Edition](http://pipeline.ai/products/)

[Standalone Edition](http://pipeline.ai/products/)

[Enterprise Edition](http://pipeline.ai/products/)

```
pipeline
Expand All @@ -89,7 +95,8 @@ Usage: pipeline <-- This List of CLI Commands
(Community) pipeline predict <-- Predict with Model Server or Cluster
(Enterprise) pipeline predict-cluster-connect <-- Create Secure Tunnel to Prediction Cluster
(Enterprise) pipeline predict-cluster-autoscale <-- Configure AutoScaling for Prediction Cluster
pipeline predict-cluster-connect <-- Create Secure Tunnel to Prediction Cluster
pipeline predict-cluster-describe <-- Describe Prediction Cluster
pipeline predict-cluster-logs <-- View Prediction Cluster Logs
pipeline predict-cluster-scale <-- Scale Prediction Cluster
Expand All @@ -113,7 +120,7 @@ Usage: pipeline <-- This List of CLI Commands
pipeline train-cluster-shell <-- Shell into Training Cluster
pipeline train-cluster-start <-- Start Training Cluster from Docker Registry
pipeline train-cluster-status <-- Status of Training Cluster
pipeline train-cluster-stop <-- Stop Traininhg Cluster
pipeline train-cluster-stop <-- Stop Training Cluster
(Standalone) pipeline train-server-build <-- Build Prediction Server
pipeline train-server-logs <-- View Prediction Server Logs
Expand All @@ -126,34 +133,79 @@ Usage: pipeline <-- This List of CLI Commands
(Community) pipeline version <-- View This CLI Version
```

# Prepare Model Samples
# Prepare Sample Models
## Clone the PipelineAI Predict Repo
```
git clone https://github.com/PipelineAI/predict
git clone https://github.com/PipelineAI/pipeline/models
```

## Change into `predict` Directory
```
cd predict
cd models
```

## Switch to Latest Branch (master)
_Note: Master may be unstable. See Releases Tab for stable releases._
```
git checkout master
```

# Train Models
## Inspect Model Directory
```
ls -l ./models/tensorflow/mnist
### EXPECTED OUTPUT ###
...
pipeline_conda_environment.yml <-- Required. Sets up the conda environment
pipeline_train.py <-- Required. `main()` is required
...
```

## Build Training Server
```
pipeline train-server-build --model-type=tensorflow --model-name=mnist --model-tag=master --model-path=./models/tensorflow/mnist
```

## Switch to Latest Release Branch (r1.3)
## Start Training UI
```
git checkout r1.3
pipeline train-server-start --model-type=tensorflow --model-name=mnist --model-tag=master
```

# Model Predictions
_Note: If you see the error below, run `docker rm -f train-tensorflow-mnist-master` first._
```
docker: Error response from daemon: Conflict. The container name "/train-tensorflow-mnist-master" is already in use by container.
```

## View Training UI (TensorFlow Models Only)
```
http://localhost:6334
```
_This UI sometimes requires a couple refreshes. We are working to stabilize the UI._

![PipelineAI Model UI](http://pipeline.ai/assets/img/pipelineai-train-compare-ui.png)

![PipelineAI Model UI](http://pipeline.ai/assets/img/pipelineai-train-compare-ui-2.png)

## Stop Training UI
```
pipeline train-server-stop --model-type=tensorflow --model-name=mnist --model-tag=master
```

# Serve Model Predictions
## Inspect Model Directory
```
ls -l ./models/tensorflow/mnist
### EXPECTED OUTPUT ###
...
pipeline_conda_environment.yml <-- Required. Sets up the conda environment
pipeline_predict.py <-- Required. `predict(request: bytes) -> bytes` is required
versions/ <-- Optional. If directory exists, we start TensorFlow Serving
...
```

## Inspect PipelineAI Predict Module `./models/tensorflow/mnist/pipeline_predict.py`
## Inspect `pipeline_predict.py`
_Note: Only the `predict()` method is required. Everything else is optional._
```
cat ./models/tensorflow/mnist/pipeline_predict.py
Expand Down Expand Up @@ -206,20 +258,20 @@ def predict(request: bytes) -> bytes: <-- Required. Cal
## Build the Model into a Runnable Docker Image
This command bundles the TensorFlow runtime with the model.
```
pipeline predict-server-build --model-type=tensorflow --model-name=mnist --model-tag="v1" --model-path=./models/tensorflow/mnist
pipeline predict-server-build --model-type=tensorflow --model-name=mnist --model-tag=v1 --model-path=./models/tensorflow/mnist
```
_`model-path` must be a relative path._

## Start the Model Server
```
pipeline predict-server-start --model-type=tensorflow --model-name=mnist --model-tag="v1" --memory-limit=2G
pipeline predict-server-start --model-type=tensorflow --model-name=mnist --model-tag=v1 --memory-limit=2G
```
_If the port is already allocated, run `docker ps`, then `docker rm -f <container-id>`._

## Monitor Runtime Logs
Wait for the model runtime to settle...
```
pipeline predict-server-logs --model-type=tensorflow --model-name=mnist --model-tag="v1"
pipeline predict-server-logs --model-type=tensorflow --model-name=mnist --model-tag=v1
### EXPECTED OUTPUT ###
...
Expand All @@ -232,17 +284,24 @@ INFO[0050] Completed initial partial maintenance sweep through 4 in-memory finge
```
_You need to `ctrl-c` out of the log viewing before proceeding._


## PipelineAI Prediction CLI
### Perform Prediction
_You may see `502 Bad Gateway` if you predict too quickly. Let the server startup completely, then predict again._

_The first call takes 10-20x longer than subsequent calls for lazy initialization and warm-up. Predict again if you see a "fallback" message._

_You may see `502 Bad Gateway` if you predict too quickly. Let the server startup completely, then predict again._

_Before proceeding, make sure you hit `ctrl-c` after viewing the logs in the previous step._
```
pipeline predict --model-type=tensorflow --model-name=mnist --model-tag="v1" --predict-server-url=http://localhost:6969 --test-request-path=./models/tensorflow/mnist/data/test_request.json
pipeline predict --model-type=tensorflow --model-name=mnist --model-tag=v1 --predict-server-url=http://localhost:6969 --test-request-path=./models/tensorflow/mnist/data/test_request.json
### IGNORE THIS ERROR. WAIT A MINUTE AND RE-RUN THE COMMAND ABOVE ###
...
'<html>\r\n<head><title>502 Bad Gateway</title></head></html>
...
### Expected Output ###
...
{"outputs": [0.0022526539396494627, 2.63791100074684e-10, 0.4638307988643646, 0.21909376978874207, 3.2985670372909226e-07, 0.29357224702835083, 0.00019597385835368186, 5.230629176367074e-05, 0.020996594801545143, 5.426473762781825e-06]}
### Formatted Output ###
Expand All @@ -262,7 +321,7 @@ Digit Confidence

### Perform 100 Predictions in Parallel (Mini Load Test)
```
pipeline predict --model-type=tensorflow --model-name=mnist --model-tag="v1" --predict-server-url=http://localhost:6969 --test-request-path=./models/tensorflow/mnist/data/test_request.json --test-request-concurrency=100
pipeline predict --model-type=tensorflow --model-name=mnist --model-tag=v1 --predict-server-url=http://localhost:6969 --test-request-path=./models/tensorflow/mnist/data/test_request.json --test-request-concurrency=100
```

## PipelineAI Prediction REST API
Expand Down Expand Up @@ -330,7 +389,7 @@ _Create additional PipelineAI Prediction widgets using [THIS](https://prometheus

## Stop Model Server
```
pipeline predict-server-stop --model-type=tensorflow --model-name=mnist --model-tag="v1"
pipeline predict-server-stop --model-type=tensorflow --model-name=mnist --model-tag=v1
```

# [PipelineAI Standalone and Enterprise Features](http://pipeline.ai/features)
Expand Down

0 comments on commit 7454fc1

Please sign in to comment.