Tensorflow＆＃39; trainable = False＆＃39;被忽略了

时间：2017-05-09 09:41:10

标签： tensorflow

我想将固定权重矩阵传递给张量流中的2D卷积运算。我尝试将java.lang.NoSuchMethodError: No such DSL method '$' found among steps [AddInteractivePromotion, ArtifactoryGradleBuild, ArtifactoryMavenBuild, ConanAddRemote, ConanAddUser, InitConanClient, MavenDescriptorStep, RunConanCommand, ansiblePlaybook, archive, artifactoryDownload, artifactoryPromoteBuild, artifactoryUpload, bat, build, catchError, checkout, collectEnv, deleteDir, dir, dockerFingerprintFrom, dockerFingerprintRun, dockerPullStep, dockerPushStep, echo, emailext, emailextrecipients, envVarsForTool, error, fileExists, getArtifactoryServer, getContext, getDatabaseConnection, git, input, isUnix, library, libraryResource, load, mail, milestone, newArtifactoryServer, newBuildInfo, newGradleBuild, newMavenBuild, node, parallel, properties, publishBuildInfo, pwd, readFile, readTrusted, resolveScm, retry, script, sh, sleep, sql, stage, stash, step, svn, timeout, timestamps, tool, unarchive, unstash, validateDeclarativePipeline, waitForQualityGate, waitUntil, withContext, withCredentials, withDockerContainer, withDockerRegistry, withDockerServer, withEnv, wrap, writeFile, ws, xrayScanBuild] or symbols [all, allOf, always, ant, antFromApache, antOutcome, antTarget, any, anyOf, apiToken, architecture, archiveArtifacts, artifactManager, batchFile, booleanParam, branch, buildButton, buildDiscarder, caseInsensitive, caseSensitive, choice, choiceParam, cleanWs, clock, cloud, command, configFile, configFileProvider, cron, crumb, defaultView, demand, disableConcurrentBuilds, docker, dockerfile, downloadSettings, downstream, dumb, envVars, environment, expression, file, fileParam, filePath, fingerprint, frameOptions, freeStyle, freeStyleJob, git, github, githubPush, gradle, hyperlink, hyperlinkToModels, installSource, jdk, jdkInstaller, jgit, jgitapache, jnlp, jobName, junit, label, lastDuration, lastFailure, lastGrantedAuthorities, lastStable, lastSuccess, legacy, legacySCM, list, local, location, logRotator, loggedInUsersCanDoAnything, masterBuild, maven, maven3Mojos, mavenErrors, mavenMojos, mavenWarnings, modernSCM, msbuild, msbuildError, msbuildWarning, myView, node, nodeProperties, nonStoredPasswordParam, none, not, overrideIndexTriggers, paneStatus, parameters, password, pattern, pipeline-model, pipelineTriggers, plainText, plugin, pollSCM, projectNamingStrategy, proxy, queueItemAuthenticator, quietPeriod, remotingCLI, run, runParam, schedule, scmRetryCount, search, security, shell, skipDefaultCheckout, skipStagesAfterUnstable, slave, stackTrace, standard, status, string, stringParam, swapSpace, text, textParam, tmpSpace, toolLocation, unsecured, upstream, usernameColonPassword, usernamePassword, viewsTabBar, weather, withSonarQubeEnv, zfs, zip] or globals [Artifactory, currentBuild, docker, env, params, pipeline, scm] at org.jenkinsci.plugins.workflow.cps.DSL.invokeMethod(DSL.java:149) at org.jenkinsci.plugins.workflow.cps.CpsScript.invokeMethod(CpsScript.java:108) at groovy.lang.GroovyObject$invokeMethod.call(Unknown Source) at org.codehaus.groovy.runtime.callsite.CallSiteArray.defaultCall(CallSiteArray.java:48) at org.codehaus.groovy.runtime.callsite.AbstractCallSite.call(AbstractCallSite.java:113) at org.kohsuke.groovy.sandbox.impl.Checker$1.call(Checker.java:151) at org.kohsuke.groovy.sandbox.GroovyInterceptor.onMethodCall(GroovyInterceptor.java:21) at org.jenkinsci.plugins.scriptsecurity.sandbox.groovy.SandboxInterceptor.onMethodCall(SandboxInterceptor.java:115) at org.kohsuke.groovy.sandbox.impl.Checker$1.call(Checker.java:149) at org.kohsuke.groovy.sandbox.impl.Checker.checkedCall(Checker.java:146) at org.kohsuke.groovy.sandbox.impl.Checker.checkedCall(Checker.java:123) at org.kohsuke.groovy.sandbox.impl.Checker.checkedCall(Checker.java:123) at com.cloudbees.groovy.cps.sandbox.SandboxInvoker.methodCall(SandboxInvoker.java:16) at WorkflowScript.run(WorkflowScript:16) at ___cps.transform___(Native Method) at com.cloudbees.groovy.cps.impl.ContinuationGroup.methodCall(ContinuationGroup.java:57) at com.cloudbees.groovy.cps.impl.FunctionCallBlock$ContinuationImpl.dispatchOrArg(FunctionCallBlock.java:109) at com.cloudbees.groovy.cps.impl.FunctionCallBlock$ContinuationImpl.fixArg(FunctionCallBlock.java:82) at sun.reflect.GeneratedMethodAccessor637.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source) at java.lang.reflect.Method.invoke(Unknown Source) at com.cloudbees.groovy.cps.impl.ContinuationPtr$ContinuationImpl.receive(ContinuationPtr.java:72) at com.cloudbees.groovy.cps.impl.ClosureBlock.eval(ClosureBlock.java:46) at com.cloudbees.groovy.cps.Next.step(Next.java:74) at com.cloudbees.groovy.cps.Continuable.run0(Continuable.java:154) at org.jenkinsci.plugins.workflow.cps.SandboxContinuable.access$001(SandboxContinuable.java:18) at org.jenkinsci.plugins.workflow.cps.SandboxContinuable$1.call(SandboxContinuable.java:33) at org.jenkinsci.plugins.workflow.cps.SandboxContinuable$1.call(SandboxContinuable.java:30) at org.jenkinsci.plugins.scriptsecurity.sandbox.groovy.GroovySandbox.runInSandbox(GroovySandbox.java:108) at org.jenkinsci.plugins.workflow.cps.SandboxContinuable.run0(SandboxContinuable.java:30) at org.jenkinsci.plugins.workflow.cps.CpsThread.runNextChunk(CpsThread.java:165) at org.jenkinsci.plugins.workflow.cps.CpsThreadGroup.run(CpsThreadGroup.java:330) at org.jenkinsci.plugins.workflow.cps.CpsThreadGroup.access$100(CpsThreadGroup.java:82) at org.jenkinsci.plugins.workflow.cps.CpsThreadGroup$2.call(CpsThreadGroup.java:242) at org.jenkinsci.plugins.workflow.cps.CpsThreadGroup$2.call(CpsThreadGroup.java:230) at org.jenkinsci.plugins.workflow.cps.CpsVmExecutorService$2.call(CpsVmExecutorService.java:64) at java.util.concurrent.FutureTask.run(Unknown Source) at hudson.remoting.SingleLaneExecutorService$1.run(SingleLaneExecutorService.java:112) at jenkins.util.ContextResettingExecutorService$1.run(ContextResettingExecutorService.java:28) at java.util.concurrent.Executors$RunnableAdapter.call(Unknown Source) at java.util.concurrent.FutureTask.run(Unknown Source) at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source) at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source) at java.lang.Thread.run(Unknown Source) Finished: FAILURE 置于如下，但TF似乎忽略了该选项。

echo ${server}
print ${server}
println ${server}
println "${server}"

在训练期间不断失去精确度。 1变为0.98然后变为0.96等，0变为0.012等。

如果我trainable=False w = tf.Variable(w, trainable=False, dtype=tf.float32, name='upscaleW') data = tf.nn.conv2d_transpose(data, w, outshapeF, strides, padding="SAME", data_format=data_format, name='UpsamplingDeconv2D') 不在那里。我只能在tf.trainable_variables()中找到它们，因此它们甚至不在可训练变量列表中。我无法弄清楚如何冻结重量。

这条线路可能有问题吗？ https://github.com/tensorflow/tensorflow/blob/r1.1/tensorflow/python/ops/nn_ops.py#L1075

2 个答案:

答案 0 :(得分：1)

没关系。我的错。在我的代码中，我传递给minimize(var_list=tf.contrib.framework.get_variables())而不是get_trainable_variables，这显然会覆盖trainable=False参数。

答案 1 :(得分：0)

设置trainable=False会将变量保留在图中的GraphKeys.TRAINABLE_VARIABLES集合之外，因此在向后传播时不会对其进行训练。

应该可以。没有最近的错误。

相关问题
最新问题