Code from BasicOptimizer.scala:75 executed in 421.81 seconds (3.398 gc):
val lineSearchInstance: LineSearchStrategy = lineSearchFactory
IterativeTrainer.wrap(trainable)
.setOrientation(orientation())
.setMonitor(new TrainingMonitor() {
override def clear(): Unit = trainingMonitor.clear()
override def log(msg: String): Unit = trainingMonitor.log(msg)
override def onStepFail(currentPoint: Step): Boolean = {
BasicOptimizer.this.onStepFail(trainable, currentPoint)
}
override def onStepComplete(currentPoint: Step): Unit = {
if (0 < logEvery && (0 == currentPoint.iteration % logEvery || currentPoint.iteration < logEvery)) {
val image = currentImage
timelineAnimation += image
val caption = "Iteration " + currentPoint.iteration
out.p(caption + "\n" + out.jpg(image, caption))
}
BasicOptimizer.this.onStepComplete(trainable, currentPoint)
trainingMonitor.onStepComplete(currentPoint)
super.onStepComplete(currentPoint)
}
})
.setTimeout(trainingMinutes, TimeUnit.MINUTES)
.setMaxIterations(trainingIterations)
.setLineSearchFactory((_: CharSequence) => lineSearchInstance)
.setTerminateThreshold(java.lang.Double.NEGATIVE_INFINITY)
.runAndFree
.asInstanceOf[lang.Double]
Reset training subject: 1148611481695500
Reset training subject: 1148617837333200
Adding measurement 35165d14 to history. Total: 0
LBFGS Accumulation History: 1 points
Constructing line search parameters: GD+Trust
th(0)=1.4409939870238304;dx=-1.012605967467474E-7
New Minimum: 1.4409939870238304 > 1.440993808209896
WOLFE (weak): th(2.154434690031884)=1.440993808209896; dx=-1.011967909081769E-7 evalInputDelta=1.7881393432617188E-7
New Minimum: 1.440993808209896 > 1.4409933909773827
WOLFE (weak): th(4.308869380063768)=1.4409933909773827; dx=-1.0093833633997184E-7 evalInputDelta=5.960464477539062E-7
New Minimum: 1.4409933909773827 > 1.440992098301649
WOLFE (weak): th(12.926608140191302)=1.440992098301649; dx=-1.0108672827128985E-7 evalInputDelta=1.8887221813201904E-6
New Minimum: 1.440992098301649 > 1.4409862607717514
WOLFE (weak): th(51.70643256076521)=1.4409862607717514; dx=-1.0114317341510622E-7 evalInputDelta=7.72625207901001E-6
New Minimum: 1.4409862607717514 > 1.4409554935991764
WOLFE (weak): th(258.53216280382605)=1.4409554935991764; dx=-1.0097725901178247E-7 evalInputDelta=3.849342465400696E-5
New Minimum: 1.4409554935991764 > 1.440763469785452
WOLFE (weak): th(1551.1929768229563)=1.440763469785452; dx=-1.0106052431243507E-7 evalInputDelta=2.3051723837852478E-4
New Minimum: 1.440763469785452 > 1.4393826574087143
WOLFE (weak): th(10858.350837760694)=1.4393826574087143; dx=-1.0102038839888186E-7 evalInputDelta=0.0016113296151161194
New Minimum: 1.4393826574087143 > 1.4282382875680923
WOLFE (weak): th(86866.80670208555)=1.4282382875680923; dx=-9.965543903855631E-8 evalInputDelta=0.012755699455738068
New Minimum: 1.4282382875680923 > 1.3381439819931984
END: th(781801.26031877)=1.3381439819931984; dx=-8.646060673553015E-8 evalInputDelta=0.10285000503063202
Fitness changed from 1.4409939870238304 to 1.3381439819931984
Iteration 1 complete. Error: 1.3381439819931984 Total: 82.8321; Orientation: 0.0053; Line Search: 62.8017
<a id="p-3"></a>Iteration 1
<a id="p-2"></a>=0.3443514332175255;dx=-3.908228587088592E-8
Armijo: th(1.0965753619410196E7)=0.69981649518013; dx=-4.980920002825725E-9 evalInputDelta=-0.3554650619626045
New Minimum: 0.3443514332175255 > 0.3287089318037033
END: th(5482876.809705098)=0.3287089318037033; dx=-1.5628088235590873E-8 evalInputDelta=0.015642501413822174
Fitness changed from 0.3443514332175255 to 0.3287089318037033
Iteration 14 complete. Error: 0.3287089318037033 Total: 25.2224; Orientation: 0.1246; Line Search: 18.8284
Adding measurement 7e96fb67 to history. Total: 3
Rejected: LBFGS Orientation magnitude: 2.414e+03, gradient 1.992e-04, dot -0.713; [6b492d23-9e94-4ca3-bba6-17cd1c08f0d8 = 1.000/1.000e+00]
Orientation rejected. Popping history element from 0.3888847082853317, 0.36032911390066147, 0.3443514332175255, 0.3287089318037033
LBFGS Accumulation History: 3 points
Removed measurement 6111f669 to history. Total: 4
th(0)=0.3287089318037033;dx=-3.944573666512609E-8
Armijo: th(1.1812500000000007E7)=0.7353792861104012; dx=-2.2382118367727356E-9 evalInputDelta=-0.40667035430669785
New Minimum: 0.3287089318037033 > 0.31747470796108246
END: th(5906250.000000004)=0.31747470796108246; dx=-1.2811171402317472E-8 evalInputDelta=0.01123422384262085
Fitness changed from 0.3287089318037033 to 0.31747470796108246
Iteration 15 complete. Error: 0.31747470796108246 Total: 24.9459; Orientation: 0.1243; Line Search: 18.4681
<a id="p-15"></a>Iteration 15
<a id="p-14"></a>
Final threshold in iteration 16: 0.31747470796108246 (> -Infinity) after 421.805s (< 3600.000s)
0.31747470796108246