您好,登錄后才能下訂單哦!
本篇內(nèi)容介紹了“怎么用Kubernetes和Helm進(jìn)行高效的超參數(shù)調(diào)優(yōu)”的有關(guān)知識(shí),在實(shí)際案例的操作過(guò)程中,不少人都會(huì)遇到這樣的困境,接下來(lái)就讓小編帶領(lǐng)大家學(xué)習(xí)一下如何處理這些情況吧!希望大家仔細(xì)閱讀,能夠?qū)W有所成!
在進(jìn)行Hyperparameter Sweep的時(shí)候,我們需要根據(jù)許多不同的超參數(shù)組合進(jìn)行不同的訓(xùn)練,為同一模型進(jìn)行多次訓(xùn)練需要消耗大量計(jì)算資源或者耗費(fèi)大量時(shí)間。
如果根據(jù)不同的超參數(shù)并行進(jìn)行訓(xùn)練,這需要大量計(jì)算資源。
如果在固定計(jì)算資源上順序進(jìn)行所有不同超參數(shù)組合對(duì)應(yīng)的訓(xùn)練,這需要花費(fèi)大量時(shí)間完成所有組合對(duì)應(yīng)的訓(xùn)練。
因此在落地時(shí)中,大多數(shù)人通過(guò)非常有限的幾次手動(dòng)微調(diào)他們的超參數(shù)就挑選一個(gè)相對(duì)最優(yōu)的組合。
通過(guò)Kubernetes與Helm,您可以非常輕松地探索非常大的超參數(shù)空間,同時(shí)最大化集群的利用率,從而優(yōu)化成本。
Helm使我們能夠?qū)?yīng)用程序打包到chart中并輕松地對(duì)其進(jìn)行參數(shù)化。在Hyperparameter Sweep時(shí),我們可以利用Helm chart values的配置,在template中生成對(duì)應(yīng)的TFJobs進(jìn)行訓(xùn)練部署,同時(shí)chart中還可以部署一個(gè)TensorBoard實(shí)例來(lái)監(jiān)控所有這些TFJobs,這樣我們就可以快速比較我們所有的超參數(shù)組合訓(xùn)練的結(jié)果,對(duì)那些訓(xùn)練效果不好的超參數(shù)組合,我們可以盡早刪除對(duì)應(yīng)的訓(xùn)練任務(wù),這無(wú)疑會(huì)大幅的節(jié)省集群的計(jì)算資源,從而降低成本。
我們將通過(guò)Azure/kubeflow-labs/hyperparam-sweep中的例子進(jìn)行Demo。
首先通過(guò)以下Dockerfile制作訓(xùn)練的鏡像:
FROM tensorflow/tensorflow:1.7.0-gpu COPY requirements.txt /app/requirements.txt WORKDIR /app RUN mkdir ./output RUN mkdir ./logs RUN mkdir ./checkpoints RUN pip install -r requirements.txt COPY ./* /app/ ENTRYPOINT [ "python", "/app/main.py" ]
其中main.py訓(xùn)練腳本內(nèi)容如下:
import click import tensorflow as tf import numpy as np from skimage.data import astronaut from scipy.misc import imresize, imsave, imread img = imread('./starry.jpg') img = imresize(img, (100, 100)) save_dir = 'output' epochs = 2000 def linear_layer(X, layer_size, layer_name): with tf.variable_scope(layer_name): W = tf.Variable(tf.random_uniform([X.get_shape().as_list()[1], layer_size], dtype=tf.float32), name='W') b = tf.Variable(tf.zeros([layer_size]), name='b') return tf.nn.relu(tf.matmul(X, W) + b) @click.command() @click.option("--learning-rate", default=0.01) @click.option("--hidden-layers", default=7) @click.option("--logdir") def main(learning_rate, hidden_layers, logdir='./logs/1'): X = tf.placeholder(dtype=tf.float32, shape=(None, 2), name='X') y = tf.placeholder(dtype=tf.float32, shape=(None, 3), name='y') current_input = X for layer_id in range(hidden_layers): h = linear_layer(current_input, 20, 'layer{}'.format(layer_id)) current_input = h y_pred = linear_layer(current_input, 3, 'output') #loss will be distance between predicted and true RGB loss = tf.reduce_mean(tf.reduce_sum(tf.squared_difference(y, y_pred), 1)) tf.summary.scalar('loss', loss) train_op = tf.train.AdamOptimizer(learning_rate).minimize(loss) merged_summary_op = tf.summary.merge_all() res_img = tf.cast(tf.clip_by_value(tf.reshape(y_pred, (1,) + img.shape), 0, 255), tf.uint8) img_summary = tf.summary.image('out', res_img, max_outputs=1) xs, ys = get_data(img) with tf.Session() as sess: tf.global_variables_initializer().run() train_writer = tf.summary.FileWriter(logdir + '/train', sess.graph) test_writer = tf.summary.FileWriter(logdir + '/test') batch_size = 50 for i in range(epochs): # Get a random sampling of the dataset idxs = np.random.permutation(range(len(xs))) # The number of batches we have to iterate over n_batches = len(idxs) // batch_size # Now iterate over our stochastic minibatches: for batch_i in range(n_batches): batch_idxs = idxs[batch_i * batch_size: (batch_i + 1) * batch_size] sess.run([train_op, loss, merged_summary_op], feed_dict={X: xs[batch_idxs], y: ys[batch_idxs]}) if batch_i % 100 == 0: c, summary = sess.run([loss, merged_summary_op], feed_dict={X: xs[batch_idxs], y: ys[batch_idxs]}) train_writer.add_summary(summary, (i * n_batches * batch_size) + batch_i) print("epoch {}, (l2) loss {}".format(i, c)) if i % 10 == 0: img_summary_res = sess.run(img_summary, feed_dict={X: xs, y: ys}) test_writer.add_summary(img_summary_res, i * n_batches * batch_size) def get_data(img): xs = [] ys = [] for row_i in range(img.shape[0]): for col_i in range(img.shape[1]): xs.append([row_i, col_i]) ys.append(img[row_i, col_i]) xs = (xs - np.mean(xs)) / np.std(xs) return xs, np.array(ys) if __name__ == "__main__": main()
docker build制作鏡像時(shí),會(huì)將根目錄下的starry.jpg圖片打包進(jìn)去供main.py讀取。
main.py使用基于Andrej Karpathy's Image painting demo的模型,這個(gè)模型的目標(biāo)是繪制一個(gè)盡可能接近原作的新圖片,文森特梵高的“星夜”。
在Helm chart values.yaml中配置如下:
image:ritazh / tf-paint:gpu useGPU:true hyperParamValues: learningRate: - 0.001 - 0.01 - 0.1 hiddenLayers: - 5 - 6 - 7
image: 配置訓(xùn)練任務(wù)對(duì)應(yīng)的docker image,就是前面您制作的鏡像。
useGPU: bool值,默認(rèn)true表示將使用gpu進(jìn)行訓(xùn)練,如果是false,則需要您制作鏡像時(shí)使用tensorflow/tensorflow:1.7.0
base image。
hyperParamValues: 超參數(shù)們的配置,在這里我們只配置了learningRate
, hiddenLayers
兩個(gè)超參數(shù)。
Helm chart中主要是TFJob對(duì)應(yīng)的定義、Tensorboard的Deployment及其Service的定義:
# First we copy the values of values.yaml in variable to make it easier to access them {{- $lrlist := .Values.hyperParamValues.learningRate -}} {{- $nblayerslist := .Values.hyperParamValues.hiddenLayers -}} {{- $image := .Values.image -}} {{- $useGPU := .Values.useGPU -}} {{- $chartname := .Chart.Name -}} {{- $chartversion := .Chart.Version -}} # Then we loop over every value of $lrlist (learning rate) and $nblayerslist (hidden layer depth) # This will result in create 1 TFJob for every pair of learning rate and hidden layer depth {{- range $i, $lr := $lrlist }} {{- range $j, $nblayers := $nblayerslist }} apiVersion: kubeflow.org/v1alpha1 kind: TFJob # Each one of our trainings will be a separate TFJob metadata: name: module8-tf-paint-{{ $i }}-{{ $j }} # We give a unique name to each training labels: chart: "{{ $chartname }}-{{ $chartversion | replace "+" "_" }}" spec: replicaSpecs: - template: spec: restartPolicy: OnFailure containers: - name: tensorflow image: {{ $image }} env: - name: LC_ALL value: C.UTF-8 args: # Here we pass a unique learning rate and hidden layer count to each instance. # We also put the values between quotes to avoid potential formatting issues - --learning-rate - {{ $lr | quote }} - --hidden-layers - {{ $nblayers | quote }} - --logdir - /tmp/tensorflow/tf-paint-lr{{ $lr }}-d-{{ $nblayers }} # We save the summaries in a different directory {{ if $useGPU }} # We only want to request GPUs if we asked for it in values.yaml with useGPU resources: limits: nvidia.com/gpu: 1 {{ end }} volumeMounts: - mountPath: /tmp/tensorflow subPath: module8-tf-paint # As usual we want to save everything in a separate subdirectory name: azurefile volumes: - name: azurefile persistentVolumeClaim: claimName: azurefile --- {{- end }} {{- end }} # We only want one instance running for all our jobs, and not 1 per job. apiVersion: v1 kind: Service metadata: labels: app: tensorboard name: module8-tensorboard spec: ports: - port: 80 targetPort: 6006 selector: app: tensorboard type: LoadBalancer --- apiVersion: extensions/v1beta1 kind: Deployment metadata: labels: app: tensorboard name: module8-tensorboard spec: template: metadata: labels: app: tensorboard spec: volumes: - name: azurefile persistentVolumeClaim: claimName: azurefile containers: - name: tensorboard command: - /usr/local/bin/tensorboard - --logdir=/tmp/tensorflow - --host=0.0.0.0 image: tensorflow/tensorflow ports: - containerPort: 6006 volumeMounts: - mountPath: /tmp/tensorflow subPath: module8-tf-paint name: azurefile
按照上面的超參數(shù)配置,在helm install時(shí),9個(gè)超參數(shù)組合會(huì)產(chǎn)生9個(gè)TFJob,對(duì)應(yīng)我們指定的3個(gè)learningRate和3個(gè)hiddenLayers所有組合。
main.py訓(xùn)練腳本有3個(gè)參數(shù):
argument | description | default value |
---|---|---|
--learning-rate | Learning rate value | 0.001 |
--hidden-layers | Number of hidden layers in our network. | 4 |
--log-dir | Path to save TensorFlow's summaries | None |
執(zhí)行helm install命令即可輕松完成所有不同超參數(shù)組合對(duì)應(yīng)的訓(xùn)練部署,這里我們只使用了單機(jī)訓(xùn)練,您也可以使用分布式訓(xùn)練。
helm install . NAME: telling-buffalo LAST DEPLOYED: NAMESPACE: tfworkflow STATUS: DEPLOYED RESOURCES: ==> v1/Service NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE module8-tensorboard LoadBalancer 10.0.142.217 <pending> 80:30896/TCP 1s ==> v1beta1/Deployment NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE module8-tensorboard 1 1 1 0 1s ==> v1alpha1/TFJob NAME AGE module8-tf-paint-0-0 1s module8-tf-paint-1-0 1s module8-tf-paint-1-1 1s module8-tf-paint-2-1 1s module8-tf-paint-2-2 1s module8-tf-paint-0-1 1s module8-tf-paint-0-2 1s module8-tf-paint-1-2 1s module8-tf-paint-2-0 0s ==> v1/Pod(related) NAME READY STATUS RESTARTS AGE module8-tensorboard-7ccb598cdd-6vg7h 0/1 ContainerCreating 0 1s
部署chart后,查看已創(chuàng)建的Pods,您應(yīng)該看到對(duì)應(yīng)的一些列Pods,以及監(jiān)視所有窗格的單個(gè)TensorBoard實(shí)例:
$ kubectl get pods NAME READY STATUS RESTARTS AGE module8-tensorboard-7ccb598cdd-6vg7h 1/1 Running 0 16s module8-tf-paint-0-0-master-juc5-0-hw5cm 0/1 Pending 0 4s module8-tf-paint-0-1-master-pu49-0-jp06r 1/1 Running 0 14s module8-tf-paint-0-2-master-awhs-0-gfra0 0/1 Pending 0 6s module8-tf-paint-1-0-master-5tfm-0-dhhhv 1/1 Running 0 16s module8-tf-paint-1-1-master-be91-0-zw4gk 1/1 Running 0 16s module8-tf-paint-1-2-master-r2nd-0-zhws1 0/1 Pending 0 7s module8-tf-paint-2-0-master-7w37-0-ff0w9 0/1 Pending 0 13s module8-tf-paint-2-1-master-260j-0-l4o7r 0/1 Pending 0 10s module8-tf-paint-2-2-master-jtjb-0-5l84q 0/1 Pending 0 9s
注意:由于群集中可用的GPU資源,某些pod正在等待處理。如果群集中有3個(gè)GPU,則在給定時(shí)間最多只能有3個(gè)TFJob(每個(gè)TFJob請(qǐng)求了一塊gpu)并行訓(xùn)練。
TensorBoard Service也會(huì)在Helm install執(zhí)行時(shí)自動(dòng)完成創(chuàng)建,您可以使用該Service的External-IP連接到TensorBoard。
$ kubectl get service NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE module8-tensorboard LoadBalancer 10.0.142.217 <PUBLIC IP> 80:30896/TCP 5m
通過(guò)瀏覽器訪問(wèn)TensorBoard的Public IP地址,你會(huì)看到類(lèi)似如下的頁(yè)面(TensorBoard需要一點(diǎn)時(shí)間才能顯示圖像。)
在這里我們可以看到一些超參數(shù)對(duì)應(yīng)的模型比其他模型表現(xiàn)更好。例如,所有l(wèi)earning rate為0.1對(duì)應(yīng)的模型全部產(chǎn)生全黑圖像,模型效果極差。幾分鐘后,我們可以看到兩個(gè)表現(xiàn)最好的超參數(shù)組合是:
hidden layers = 5,learning rate = 0.01
hidden layers = 7,learning rate = 0.001
此時(shí),我們可以立刻Kill掉其他表現(xiàn)差的模型訓(xùn)練,釋放寶貴的gpu資源。
“怎么用Kubernetes和Helm進(jìn)行高效的超參數(shù)調(diào)優(yōu)”的內(nèi)容就介紹到這里了,感謝大家的閱讀。如果想了解更多行業(yè)相關(guān)的知識(shí)可以關(guān)注億速云網(wǎng)站,小編將為大家輸出更多高質(zhì)量的實(shí)用文章!
免責(zé)聲明:本站發(fā)布的內(nèi)容(圖片、視頻和文字)以原創(chuàng)、轉(zhuǎn)載和分享為主,文章觀點(diǎn)不代表本網(wǎng)站立場(chǎng),如果涉及侵權(quán)請(qǐng)聯(lián)系站長(zhǎng)郵箱:is@yisu.com進(jìn)行舉報(bào),并提供相關(guān)證據(jù),一經(jīng)查實(shí),將立刻刪除涉嫌侵權(quán)內(nèi)容。