Code from BasicOptimizer.scala:75 executed in 3315.44 seconds (31.475 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: 857522047185400
Reset training subject: 857562695129100
Adding measurement 5c78f456 to history. Total: 0
LBFGS Accumulation History: 1 points
Constructing line search parameters: GD+Trust
th(0)=9.380591332912445;dx=-1.9340246507410842E-6
New Minimum: 9.380591332912445 > 9.380586445331573
WOLFE (weak): th(2.154434690031884)=9.380586445331573; dx=-1.9340236977875817E-6 evalInputDelta=4.887580871582031E-6
New Minimum: 9.380586445331573 > 9.3805832862854
WOLFE (weak): th(4.308869380063768)=9.3805832862854; dx=-1.934019776497128E-6 evalInputDelta=8.046627044677734E-6
New Minimum: 9.3805832862854 > 9.380566716194153
WOLFE (weak): th(12.926608140191302)=9.380566716194153; dx=-1.934011993540453E-6 evalInputDelta=2.4616718292236328E-5
New Minimum: 9.380566716194153 > 9.38049304485321
WOLFE (weak): th(51.70643256076521)=9.38049304485321; dx=-1.933988625138379E-6 evalInputDelta=9.828805923461914E-5
New Minimum: 9.38049304485321 > 9.380101501941681
WOLFE (weak): th(258.53216280382605)=9.380101501941681; dx=-1.9336490720165812E-6 evalInputDelta=4.898309707641602E-4
New Minimum: 9.380101501941681 > 9.377655446529388
WOLFE (weak): th(1551.1929768229563)=9.377655446529388; dx=-1.9317378200857723E-6 evalInputDelta=0.0029358863830566406
New Minimum: 9.377655446529388 > 9.360112488269806
WOLFE (weak): th(10858.350837760694)=9.360112488269806; dx=-1.9152650226117683E-6 evalInputDelta=0.02047884464263916
New Minimum: 9.360112488269806 > 9.220856666564941
WOLFE (weak): th(86866.80670208555)=9.220856666564941; dx=-1.7949685155335929E-6 evalInputDelta=0.15973466634750366
New Minimum: 9.220856666564941 > 8.236478805541992
END: th(781801.26031877)=8.236478805541992; dx=-1.1007233651485292E-6 evalInputDelta=1.1441125273704529
Fitness changed from 9.380591332912445 to 8.236478805541992
Iteration 1 complete. Error: 8.236478805541992 Total: 514.9163; Orientation: 0.0683; Line Search: 394.9693
<a id="p-3"></a>Iteration 1
<a id="p-2"></a>=0.9264429733157158;dx=-4.733560194380149E-9
Armijo: th(1.1310782122667398E8)=1.2660411298274994; dx=4.873420345992615E-9 evalInputDelta=-0.3395981565117836
Armijo: th(5.655391061333699E7)=1.0000419318675995; dx=2.9432166406342296E-9 evalInputDelta=-0.0735989585518837
New Minimum: 0.9264429733157158 > 0.9004241153597832
WOLF (strong): th(1.8851303537778996E7)=0.9004241153597832; dx=5.372077187154178E-10 evalInputDelta=0.026018857955932617
END: th(4712825.884444749)=0.9090898707509041; dx=-2.8215653977673667E-9 evalInputDelta=0.017353102564811707
Fitness changed from 0.9264429733157158 to 0.9004241153597832
Iteration 19 complete. Error: 0.9004241153597832 Total: 238.9206; Orientation: 1.7834; Line Search: 197.1866
Adding measurement 1b30b6f8 to history. Total: 3
Rejected: LBFGS Orientation magnitude: 5.675e+03, gradient 5.198e-05, dot -0.587; [e56eaa01-eb6f-4f0d-8f26-9a8c5ad65ae8 = 1.000/1.000e+00]
Orientation rejected. Popping history element from 0.9837072789669037, 0.9570591971278191, 0.9264429733157158, 0.9004241153597832
LBFGS Accumulation History: 3 points
Removed measurement f252762 to history. Total: 4
th(0)=0.9004241153597832;dx=-2.7023411352154687E-9
New Minimum: 0.9004241153597832 > 0.8821309506893158
END: th(1.0153475573527962E7)=0.8821309506893158; dx=-1.1270619766245575E-9 evalInputDelta=0.018293164670467377
Fitness changed from 0.9004241153597832 to 0.8821309506893158
Iteration 20 complete. Error: 0.8821309506893158 Total: 120.4166; Orientation: 1.8369; Line Search: 79.2154
<a id="p-17"></a>Iteration 20
<a id="p-16"></a>
Final threshold in iteration 21: 0.8821309506893158 (> -Infinity) after 3315.438s (< 3600.000s)
0.8821309506893158