-
Notifications
You must be signed in to change notification settings - Fork 7
/
pvlv.go
850 lines (728 loc) · 26.8 KB
/
pvlv.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
// Copyright (c) 2023, The Emergent Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
/*
pvlv: simulates the primary value, learned value model of classical conditioning and phasic dopamine
in the amygdala, ventral striatum and associated areas.
*/
package main
//go:generate core generate -add-types
import (
"fmt"
"log"
"os"
"reflect"
"strings"
"cogentcore.org/core/base/mpi"
"cogentcore.org/core/base/randx"
"cogentcore.org/core/base/reflectx"
"cogentcore.org/core/core"
"cogentcore.org/core/events"
"cogentcore.org/core/icons"
"cogentcore.org/core/math32"
"cogentcore.org/core/math32/minmax"
"cogentcore.org/core/plot/plotcore"
"cogentcore.org/core/tensor/stats/split"
"cogentcore.org/core/tensor/stats/stats"
"cogentcore.org/core/tensor/table"
"cogentcore.org/core/tree"
"github.com/emer/axon/v2/axon"
"github.com/emer/axon/v2/examples/pvlv/cond"
"github.com/emer/emergent/v2/econfig"
"github.com/emer/emergent/v2/egui"
"github.com/emer/emergent/v2/elog"
"github.com/emer/emergent/v2/emer"
"github.com/emer/emergent/v2/env"
"github.com/emer/emergent/v2/estats"
"github.com/emer/emergent/v2/etime"
"github.com/emer/emergent/v2/looper"
"github.com/emer/emergent/v2/netview"
"github.com/emer/emergent/v2/params"
"github.com/emer/emergent/v2/paths"
)
func main() {
sim := &Sim{}
sim.New()
sim.ConfigAll()
if sim.Config.GUI {
sim.RunGUI()
} else {
sim.RunNoGUI()
}
}
// see params.go for network params, config.go for Config
// Sim encapsulates the entire simulation model, and we define all the
// functionality as methods on this struct. This structure keeps all relevant
// state information organized and available without having to pass everything around
// as arguments to methods, and provides the core GUI interface (note the view tags
// for the fields which provide hints to how things should be displayed).
type Sim struct {
// simulation configuration parameters -- set by .toml config file and / or args
Config Config `new-window:"+"`
// the network -- click to view / edit parameters for layers, paths, etc
Net *axon.Network `new-window:"+" display:"no-inline"`
// all parameter management
Params emer.NetParams `display:"add-fields"`
// contains looper control loops for running sim
Loops *looper.Manager `new-window:"+" display:"no-inline"`
// contains computed statistic values
Stats estats.Stats `new-window:"+"`
// Contains all the logs and information about the logs.'
Logs elog.Logs `new-window:"+"`
// Environments
Envs env.Envs `new-window:"+" display:"no-inline"`
// axon timing parameters and state
Context axon.Context `new-window:"+"`
// netview update parameters
ViewUpdate netview.ViewUpdate `display:"add-fields"`
// manages all the gui elements
GUI egui.GUI `display:"-"`
// a list of random seeds to use for each run
RandSeeds randx.Seeds `display:"-"`
}
// New creates new blank elements and initializes defaults
func (ss *Sim) New() {
ss.Net = axon.NewNetwork("PVLV")
econfig.Config(&ss.Config, "config.toml")
ss.Params.Config(ParamSets, ss.Config.Params.Sheet, ss.Config.Params.Tag, ss.Net)
ss.Stats.Init()
ss.RandSeeds.Init(100) // max 100 runs
ss.InitRandSeed(0)
ss.Context.Defaults()
ss.Context.ThetaCycles = int32(ss.Config.Run.ThetaCycles)
}
////////////////////////////////////////////////////////////////////////////////////////////
// Configs
// ConfigAll configures all the elements using the standard functions
func (ss *Sim) ConfigAll() {
ss.ConfigEnv()
ss.ConfigNet(ss.Net)
ss.ConfigLogs()
ss.ConfigLoops()
if ss.Config.Params.SaveAll {
ss.Config.Params.SaveAll = false
ss.Net.SaveParamsSnapshot(&ss.Params.Params, &ss.Config, ss.Config.Params.Good)
os.Exit(0)
}
}
func (ss *Sim) ConfigEnv() {
// Can be called multiple times -- don't re-create
var trn *cond.CondEnv
if len(ss.Envs) == 0 {
trn = &cond.CondEnv{}
} else {
trn = ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
}
// note: names must be standard here!
trn.Name = etime.Train.String()
if ss.Config.Env.Env != nil {
params.ApplyMap(trn, ss.Config.Env.Env, ss.Config.Debug)
}
trn.Config(ss.Config.Run.NRuns, ss.Config.Env.RunName)
trn.Init(0)
ss.ConfigRubicon()
// note: names must be in place when adding
ss.Envs.Add(trn)
}
func (ss *Sim) ConfigRubicon() {
rp := &ss.Net.Rubicon
rp.SetNUSs(&ss.Context, cond.NUSs, 1) // 1=neg
rp.Defaults()
rp.USs.PVposGain = 2
rp.USs.PVnegGain = 1
rp.LHb.VSPatchGain = 4 // 4 def -- needs more for shorter trial count here
rp.LHb.VSPatchNonRewThr = 0.1 // 0.1 def
rp.USs.USnegGains[0] = 2 // big salient input!
// note: costs weights are very low by default..
rp.Urgency.U50 = 50 // no pressure during regular trials
if ss.Config.Params.Rubicon != nil {
params.ApplyMap(rp, ss.Config.Params.Rubicon, ss.Config.Debug)
}
rp.Update()
}
func (ss *Sim) ConfigNet(net *axon.Network) {
ctx := &ss.Context
net.SetMaxData(ctx, 1)
net.SetRandSeed(ss.RandSeeds[0]) // init new separate random seed, using run = 0
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
ny := ev.NYReps
nuBgY := 5
nuBgX := 5
nuCtxY := 6
nuCtxX := 6
popY := 4
popX := 4
space := float32(2)
pone2one := paths.NewPoolOneToOne()
one2one := paths.NewOneToOne()
_ = one2one
full := paths.NewFull()
_ = pone2one
stim := ev.CurStates["CS"]
ctxt := ev.CurStates["ContextIn"]
vSgpi, vSmtxGo, vSmtxNo, vSpatchD1, vSpatchD2, urgency, usPos, pvPos, usNeg, usNegP, pvNeg, pvNegP, blaPosAcq, blaPosExt, blaNegAcq, blaNegExt, blaNov, ofcPos, ofcPosCT, ofcPosPTp, ofcPosPT, ilPos, ilPosCT, ilPosPT, ilPosPTp, ilPosMD, ofcNeg, ofcNegCT, ofcNegPT, ofcNegPTp, accCost, accCostCT, accCostPT, accCostPTp, accCostMD, ilNeg, ilNegCT, ilNegPT, ilNegPTp, ilNegMD, sc := net.AddRubiconOFCus(&ss.Context, ny, popY, popX, nuBgY, nuBgX, nuCtxY, nuCtxX, space)
// note: list all above so can copy / paste and validate correct return values
_, _, _, _, _, _ = vSgpi, vSmtxGo, vSmtxNo, vSpatchD1, vSpatchD2, urgency
_, _, _, _, _, _ = usPos, pvPos, usNeg, usNegP, pvNeg, pvNegP
_, _, _, _ = ilPos, ilPosCT, ilPosPTp, ilPosMD
_, _, _ = ofcNeg, ofcNegCT, ofcNegPTp
_, _, _, _ = ilNeg, ilNegCT, ilNegPTp, ilNegMD
_, _, _, _ = accCost, accCostCT, accCostPTp, accCostMD
_, _, _, _, _ = ofcPosPT, ofcNegPT, ilPosPT, ilNegPT, accCostPT
// todo: connect more of above
time, timeP := net.AddInputPulv4D("Time", 1, cond.MaxTime, ny, 1, space)
cs, csP := net.AddInputPulv4D("CS", stim.DimSize(0), stim.DimSize(1), stim.DimSize(2), stim.DimSize(3), space)
ctxIn := net.AddLayer4D("ContextIn", axon.InputLayer, ctxt.DimSize(0), ctxt.DimSize(1), ctxt.DimSize(2), ctxt.DimSize(3))
///////////////////////////////////////////
// CS -> BLA, OFC
net.ConnectToSC1to1(cs, sc)
net.ConnectCSToBLApos(cs, blaPosAcq, blaNov)
net.ConnectToBLAAcq(cs, blaNegAcq, full)
net.ConnectLayers(cs, vSpatchD1, full, axon.ForwardPath) // these are critical for discriminating A vs. B
net.ConnectLayers(cs, vSpatchD2, full, axon.ForwardPath)
// note: context is hippocampus -- key thing is that it comes on with stim
// most of ctxIn is same as CS / CS in this case, but a few key things for extinction
// ptpred input is important for learning to make conditional on actual engagement
net.ConnectToBLAExt(ctxIn, blaPosExt, full)
net.ConnectToBLAExt(ctxIn, blaNegExt, full)
// OFCus predicts cs
net.ConnectToPFCBack(cs, csP, ofcPos, ofcPosCT, ofcPosPT, ofcPosPTp, full, "CSToPFC")
net.ConnectToPFCBack(cs, csP, ofcNeg, ofcNegCT, ofcNegPT, ofcNegPTp, full, "CSToPFC")
///////////////////////////////////////////
// OFC predicts time, effort, urgency
// todo: a more dynamic US rep is needed to drive predictions in OFC
net.ConnectToPFCBack(time, timeP, ofcPos, ofcPosCT, ofcPosPT, ofcPosPTp, full, "TimeToPFC")
net.ConnectToPFCBack(time, timeP, ilPos, ilPosCT, ilPosPT, ilPosPTp, full, "TimeToPFC")
net.ConnectToPFCBack(time, timeP, ofcNeg, ofcNegCT, ofcNegPT, ofcNegPTp, full, "TimeToPFC")
net.ConnectToPFCBack(time, timeP, accCost, accCostCT, accCostPT, accCostPTp, full, "TimeToPFC")
net.ConnectToPFCBack(time, timeP, ilNeg, ilNegCT, ilNegPT, ilNegPTp, full, "TimeToPFC")
////////////////////////////////////////////////
// position
time.PlaceRightOf(pvPos, space*2)
cs.PlaceRightOf(time, space)
ctxIn.PlaceRightOf(cs, space)
net.Build(ctx)
net.Defaults()
net.SetNThreads(ss.Config.Run.NThreads)
ss.ApplyParams()
net.InitWeights(ctx)
}
func (ss *Sim) ApplyParams() {
ss.Params.SetAll() // first hard-coded defaults
if ss.Config.Params.Network != nil {
ss.Params.SetNetworkMap(ss.Net, ss.Config.Params.Network)
}
}
////////////////////////////////////////////////////////////////////////////////
// Init, utils
// Init restarts the run, and initializes everything, including network weights
// and resets the epoch log table
func (ss *Sim) Init() {
if ss.Config.GUI {
ss.Stats.SetString("RunName", ss.Params.RunName(0)) // in case user interactively changes tag
}
ss.Loops.ResetCounters()
ss.InitRandSeed(0)
// ss.ConfigEnv() // re-config env just in case a different set of patterns was
// selected or patterns have been modified etc
ss.GUI.StopNow = false
ss.ApplyParams()
ss.Net.GPU.SyncParamsToGPU()
ss.NewRun()
ss.ViewUpdate.RecordSyns()
ss.ViewUpdate.Update()
}
// InitRandSeed initializes the random seed based on current training run number
func (ss *Sim) InitRandSeed(run int) {
ss.RandSeeds.Set(run)
ss.RandSeeds.Set(run, &ss.Net.Rand)
}
// ConfigLoops configures the control loops: Training, Testing
func (ss *Sim) ConfigLoops() {
man := looper.NewManager()
nCycles := ss.Config.Run.ThetaCycles
man.AddStack(etime.Train).
AddTime(etime.Run, ss.Config.Run.NRuns).
AddTime(etime.Condition, 1). // all these counters will be set from env
AddTime(etime.Block, 50).
AddTime(etime.Sequence, 8).
AddTime(etime.Trial, 5).
AddTime(etime.Cycle, nCycles)
axon.LooperStdPhases(man, &ss.Context, ss.Net, nCycles-50, nCycles-1) // plus phase timing
axon.LooperSimCycleAndLearn(man, ss.Net, &ss.Context, &ss.ViewUpdate) // std algo code
for m, _ := range man.Stacks {
stack := man.Stacks[m]
stack.Loops[etime.Trial].OnStart.Add("ApplyInputs", func() {
ss.ApplyInputs()
})
}
man.GetLoop(etime.Train, etime.Run).OnStart.Add("NewRun", ss.NewRun)
/////////////////////////////////////////////
// Logging
man.AddOnEndToAll("Log", ss.Log)
axon.LooperResetLogBelow(man, &ss.Logs, etime.Sequence)
man.GetLoop(etime.Train, etime.Block).OnStart.Add("ResetLogTrial", func() {
ss.Logs.ResetLog(etime.Train, etime.Trial)
})
man.GetLoop(etime.Train, etime.Sequence).OnStart.Add("ResetDebugTrial", func() {
ss.Logs.ResetLog(etime.Debug, etime.Trial)
})
////////////////////////////////////////////
// GUI
if ss.Config.GUI {
axon.LooperUpdateNetView(man, &ss.ViewUpdate, ss.Net, ss.NetViewCounters)
axon.LooperUpdatePlots(man, &ss.GUI)
}
if ss.Config.Debug {
mpi.Println(man.DocString())
}
ss.Loops = man
}
// UpdateLoopMax gets the latest loop counter Max values from env
func (ss *Sim) UpdateLoopMax() {
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
trn := ss.Loops.Stacks[etime.Train]
trn.Loops[etime.Condition].Counter.Max = ev.Condition.Max
trn.Loops[etime.Block].Counter.Max = ev.Block.Max
trn.Loops[etime.Sequence].Counter.Max = ev.Sequence.Max
trn.Loops[etime.Trial].Counter.Max = ev.Tick.Max
if ss.Config.Env.SetNBlocks {
trn.Loops[etime.Block].Counter.Max = ss.Config.Env.NBlocks
}
}
// ApplyInputs applies input patterns from given environment.
// It is good practice to have this be a separate method with appropriate
// args so that it can be used for various different contexts
// (training, testing, etc).
func (ss *Sim) ApplyInputs() {
ctx := &ss.Context
net := ss.Net
ev := ss.Envs.ByMode(ctx.Mode).(*cond.CondEnv)
ev.Step()
ss.UpdateLoopMax()
net.InitExt(ctx)
lays := net.LayersByType(axon.InputLayer, axon.TargetLayer)
for _, lnm := range lays {
ly := ss.Net.LayerByName(lnm)
pats := ev.State(ly.Name)
if !reflectx.AnyIsNil(pats) {
ly.ApplyExt(ctx, 0, pats)
}
switch lnm {
case "CS":
ly.Pool(0, 0).Inhib.Clamped.SetBool(ev.CurTick.CSOn)
}
}
ss.ApplyRubicon(ctx, &ev.CurTick)
net.ApplyExts(ctx) // now required for GPU mode
}
// ApplyRubicon applies current Rubicon values to Context.Rubicon,
// from given trial data.
func (ss *Sim) ApplyRubicon(ctx *axon.Context, seq *cond.Sequence) {
rp := &ss.Net.Rubicon
di := uint32(0) // not doing NData here -- otherwise loop over
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
rp.NewState(ctx, di, &ss.Net.Rand) // first before anything else is updated
rp.SetGoalMaintFromLayer(ctx, di, ss.Net, "ILposPT", 0.3)
rp.DecodePVEsts(ctx, di, ss.Net)
dist := math32.Abs(float32(3 - ev.Tick.Cur))
rp.SetGoalDistEst(ctx, di, dist)
rp.EffortUrgencyUpdate(ctx, di, 1)
if seq.USOn {
if seq.Valence == cond.Pos {
rp.SetUS(ctx, di, axon.Positive, seq.US, seq.USMag)
} else {
rp.SetUS(ctx, di, axon.Negative, seq.US, seq.USMag) // adds to neg us
}
}
drvs := make([]float32, cond.NUSs)
drvs[seq.US] = 1
rp.SetDrives(ctx, di, 1, drvs...)
rp.Step(ctx, di, &ss.Net.Rand)
}
// InitEnvRun intializes a new environment run, as when the RunName is changed
// or at NewRun()
func (ss *Sim) InitEnvRun() {
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
ev.RunName = ss.Config.Env.RunName
ev.Init(0)
ss.LoadCondWeights(ev.CurRun.Weights) // only if nonempty
ss.Loops.ResetCountersBelow(etime.Train, etime.Sequence)
ss.Logs.ResetLog(etime.Train, etime.Trial)
ss.Logs.ResetLog(etime.Train, etime.Sequence)
}
// LoadRunWeights loads weights specified in current run, if any
func (ss *Sim) LoadRunWeights() {
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
ss.LoadCondWeights(ev.CurRun.Weights) // only if nonempty
}
// LoadCondWeights loads weights saved after named condition, in wts/cond.wts.gz
func (ss *Sim) LoadCondWeights(cond string) {
if cond == "" {
return
}
wfn := "wts/" + cond + ".wts.gz"
err := ss.Net.OpenWeightsJSON(core.Filename(wfn))
if err != nil {
log.Println(err)
}
}
// SaveCondWeights saves weights based on current condition, in wts/cond.wts.gz
func (ss *Sim) SaveCondWeights() {
ev := ss.Envs.ByMode(etime.Train).(*cond.CondEnv)
cnm, _ := ev.CurRun.Cond(ev.Condition.Cur)
if cnm == "" {
return
}
wfn := "wts/" + cnm + ".wts.gz"
err := ss.Net.SaveWeightsJSON(core.Filename(wfn))
if err != nil {
log.Println(err)
} else {
fmt.Printf("Saved weights to: %s\n", wfn)
}
}
// NewRun intializes a new run of the model, using the TrainEnv.Run counter
// for the new run value
func (ss *Sim) NewRun() {
ctx := &ss.Context
ss.InitRandSeed(ss.Loops.GetLoop(etime.Train, etime.Run).Counter.Cur)
ss.InitEnvRun()
ctx.Reset()
ctx.Mode = etime.Train
ss.Net.InitWeights(ctx)
ss.LoadRunWeights()
ss.InitStats()
ss.StatCounters()
ss.Logs.ResetLog(etime.Train, etime.Condition)
ss.Logs.ResetLog(etime.Train, etime.Block)
ss.UpdateLoopMax()
}
////////////////////////////////////////////////////////////////////////////////////////////
// Stats
// InitStats initializes all the statistics.
// called at start of new run
func (ss *Sim) InitStats() {
ss.Stats.SetString("Debug", "") // special debug notes per trial
ss.Stats.SetString("Cond", "")
ss.Stats.SetString("TrialName", "")
ss.Stats.SetString("SeqType", "")
ss.Stats.SetString("TickType", "")
}
// StatCounters saves current counters to Stats, so they are available for logging etc
// Also saves a string rep of them for ViewUpdate.Text
func (ss *Sim) StatCounters() {
ctx := &ss.Context
mode := ctx.Mode
ss.Loops.Stacks[mode].CountersToStats(&ss.Stats)
ss.Stats.SetInt("Cycle", int(ctx.Cycle))
ev := ss.Envs.ByMode(ctx.Mode).(*cond.CondEnv)
ss.Stats.SetString("TrialName", ev.SequenceName)
ss.Stats.SetString("SeqType", ev.SequenceType)
trl := ss.Stats.Int("Trial")
ss.Stats.SetString("TickType", fmt.Sprintf("%02d_%s", trl, ev.CurTick.Type.String()))
ss.Stats.SetString("Cond", ev.CondName)
}
func (ss *Sim) NetViewCounters(tm etime.Times) {
if ss.ViewUpdate.View == nil {
return
}
ss.StatCounters()
ss.ViewUpdate.Text = ss.Stats.Print([]string{"Run", "Condition", "Block", "Sequence", "Trial", "SeqType", "TrialName", "TickType", "Cycle", "Time", "HasRew", "Gated", "GiveUp"})
}
// TrialStats computes the tick-level statistics.
// Aggregation is done directly from log data.
func (ss *Sim) TrialStats() {
ctx := &ss.Context
diu := uint32(0)
ss.Stats.SetFloat32("HasRew", axon.GlobalScalars[axon.GvHasRew), diu]
ss.Stats.SetFloat32("Gated", axon.GlobalScalars[axon.GvVSMatrixJustGated), diu]
ss.Stats.SetFloat32("Time", axon.GlobalScalars[axon.GvTime), diu]
ss.Stats.SetFloat32("GiveUp", axon.GlobalScalars[axon.GvGiveUp), diu]
ss.Stats.SetFloat32("SC", ss.Net.LayerByName("SC").Pool(0, 0).AvgMax.CaSpkD.Cycle.Max)
}
//////////////////////////////////////////////////////////////////////////////
// Logging
func (ss *Sim) ConfigLogs() {
ss.Stats.SetString("RunName", ss.Params.RunName(0)) // used for naming logs, stats, etc
ss.Logs.AddCounterItems(etime.Run, etime.Condition, etime.Block, etime.Sequence, etime.Trial, etime.Cycle)
ss.Logs.AddStatStringItem(etime.AllModes, etime.AllTimes, "Cond")
ss.Logs.AddStatStringItem(etime.AllModes, etime.AllTimes, "RunName")
ss.Logs.AddStatStringItem(etime.AllModes, etime.Trial, "TrialName")
ss.Logs.AddStatStringItem(etime.AllModes, etime.Trial, "SeqType")
ss.Logs.AddStatStringItem(etime.AllModes, etime.Trial, "TickType")
// ss.Logs.AddPerTrlMSec("PerTrlMSec", etime.Run, etime.Epoch, etime.Trial)
axon.LogAddGlobals(&ss.Logs, &ss.Context, etime.Train, etime.Run, etime.Condition, etime.Block, etime.Sequence, etime.Trial)
plots := ss.ConfigLogItems()
// layers := ss.Net.LayersByType(axon.SuperLayer, axon.CTLayer, axon.TargetLayer)
// axon.LogAddDiagnosticItems(&ss.Logs, layers, etime.Train, etime.Block, etime.Trial)
// axon.LogInputLayer(&ss.Logs, ss.Net, etime.Train)
ss.Logs.PlotItems("DA", "RewPred")
ss.Logs.CreateTables()
ss.Logs.SetContext(&ss.Stats, ss.Net)
// don't plot certain combinations we don't use
ss.Logs.NoPlot(etime.Train, etime.Epoch, etime.Cycle)
// note: Analyze not plotted by default
ss.Logs.SetMeta(etime.Train, etime.Run, "LegendCol", "RunName")
ss.Logs.SetMeta(etime.Train, etime.Trial, "LegendCol", "Sequence")
// plot selected agg data at higher levels
times := []etime.Times{etime.Block, etime.Condition, etime.Run}
for _, tm := range times {
ss.Logs.SetMeta(etime.Train, tm, "DA:On", "-")
ss.Logs.SetMeta(etime.Train, tm, "VSPatch:On", "-")
for _, pl := range plots {
ss.Logs.SetMeta(etime.Train, tm, pl+":On", "+")
}
}
}
func (ss *Sim) ConfigLogItems() []string {
ss.Logs.AddStatAggItem("SC", etime.Run, etime.Condition, etime.Block, etime.Sequence, etime.Trial)
var plots []string
points := []string{"CS", "US"}
for ci := 0; ci < 2; ci++ { // conditions
ci := ci
for _, pt := range points {
for _, st := range ss.Config.Log.AggStats {
itmName := fmt.Sprintf("C%d_%s_%s", ci, pt, st)
plots = append(plots, itmName)
statName := fmt.Sprintf("%s_%s", pt, st)
ss.Logs.AddItem(&elog.Item{
Name: itmName,
Type: reflect.Float64,
// FixMin: true,
// FixMax: true,
Range: minmax.F32{Max: 1},
Write: elog.WriteMap{
etime.Scope(etime.AllModes, etime.Block): func(ctx *elog.Context) {
ctx.SetFloat64(ctx.Stats.FloatDi(statName, ci))
}, etime.Scope(etime.AllModes, etime.Condition): func(ctx *elog.Context) {
ix := ctx.LastNRows(ctx.Mode, etime.Block, 5) // cached
ctx.SetFloat64(stats.MeanColumn(ix, ctx.Item.Name)[0])
}, etime.Scope(etime.Train, etime.Run): func(ctx *elog.Context) {
ctx.SetAgg(ctx.Mode, etime.Condition, stats.Mean)
}}})
}
}
}
// Add a special debug message -- use of etime.Debug triggers
// inclusion
ss.Logs.AddStatStringItem(etime.Debug, etime.Trial, "Debug")
return plots
}
// Log is the main logging function, handles special things for different scopes
func (ss *Sim) Log(mode etime.Modes, time etime.Times) {
if mode != etime.Analyze && mode != etime.Debug {
ss.Context.Mode = mode // Also set specifically in a Loop callback.
}
dt := ss.Logs.Table(mode, time)
if dt == nil {
return
}
row := dt.Rows
switch {
case time == etime.Cycle:
return
case mode == etime.Train && time == etime.Trial:
ss.TrialStats()
ss.StatCounters()
if ss.Config.GUI {
ss.Logs.Log(etime.Debug, etime.Trial)
ss.GUI.UpdateTableView(etime.Debug, etime.Trial)
}
case time == etime.Block:
ss.BlockStats()
}
ss.Logs.LogRow(mode, time, row) // also logs to file, etc
}
func (ss *Sim) BlockStats() {
stnm := "BlockByType"
ix := ss.Logs.IndexView(etime.Train, etime.Trial)
spl := split.GroupBy(ix, "SeqType", "TickType")
for _, ts := range ix.Table.ColumnNames {
if ts == "SeqType" || ts == "TrialName" || ts == "TickType" {
continue
}
split.AggColumn(spl, ts, stats.Mean)
}
dt := spl.AggsToTable(table.ColumnNameOnly)
for ri := 0; ri < dt.Rows; ri++ {
tt := dt.StringValue("SeqType", ri)
trl := int(dt.Float("Trial", ri))
dt.SetString("SeqType", ri, fmt.Sprintf("%s_%d", tt, trl))
}
dt.SetMetaData("DA:On", "+")
dt.SetMetaData("RewPred:On", "+")
dt.SetMetaData("DA:FixMin", "+")
dt.SetMetaData("DA:Min", "-1")
dt.SetMetaData("DA:FixMax", "-")
dt.SetMetaData("DA:Max", "1")
dt.SetMetaData("XAxisRot", "45")
ss.Logs.MiscTables[stnm] = dt
// grab selected stats at CS and US for higher level aggregation,
nrows := dt.Rows
curSeq := ""
seq := -1
for ri := 0; ri < nrows; ri++ {
st := dt.StringValue("SeqType", ri)
ui := strings.LastIndex(st, "_")
st = st[:ui]
if curSeq != st {
seq++
curSeq = st
ss.Stats.SetStringDi("SeqType", seq, curSeq)
}
tt := dt.StringValue("TickType", ri)
if strings.Contains(tt, "_CS") {
for _, st := range ss.Config.Log.AggStats {
ss.Stats.SetFloatDi("CS_"+st, seq, dt.Float(st, ri))
}
}
if strings.Contains(tt, "_US") {
for _, st := range ss.Config.Log.AggStats {
ss.Stats.SetFloatDi("US_"+st, seq, dt.Float(st, ri))
}
}
}
if ss.Config.GUI {
plt := ss.GUI.Plots[etime.ScopeKey(stnm)]
plt.SetTable(dt)
plt.GoUpdatePlot()
}
}
////////////////////////////////////////////////////////////////////////////////////////////
// GUI
// ConfigGUI configures the Cogent Core GUI interface for this simulation.
func (ss *Sim) ConfigGUI() {
title := "Axon PVLV"
ss.GUI.MakeBody(ss, "pvlv", title, `This is the PVLV test model in Axon, in the Rubicon framework. See <a href="https://github.com/emer/emergent">emergent on GitHub</a>.</p>`)
ss.GUI.CycleUpdateInterval = 10
nv := ss.GUI.AddNetView("Network")
nv.Options.MaxRecs = 400
nv.Options.Raster.Max = ss.Config.Run.ThetaCycles
nv.Options.LayerNameSize = 0.02
nv.SetNet(ss.Net)
ss.ViewUpdate.Config(nv, etime.Phase, etime.Phase)
ss.GUI.ViewUpdate = &ss.ViewUpdate
nv.SceneXYZ().Camera.Pose.Pos.Set(0, 1.4, 2.6)
nv.SceneXYZ().Camera.LookAt(math32.Vec3(0, 0, 0), math32.Vec3(0, 1, 0))
ss.GUI.AddPlots(title, &ss.Logs)
ss.GUI.AddTableView(&ss.Logs, etime.Debug, etime.Trial)
stnm := "BlockByType"
dt := ss.Logs.MiscTable(stnm)
bcp, _ := ss.GUI.Tabs.NewTab(stnm + " Plot")
plt := plotcore.NewSubPlot(bcp)
ss.GUI.Plots[etime.ScopeKey(stnm)] = plt
plt.Options.Title = stnm
plt.Options.XAxis = "SeqType"
plt.SetTable(dt)
ss.GUI.FinalizeGUI(false)
if ss.Config.Run.GPU {
// vgpu.Debug = ss.Config.Debug
ss.Net.ConfigGPUnoGUI(&ss.Context) // must happen after gui or no gui
core.TheApp.AddQuitCleanFunc(func() {
ss.Net.GPU.Destroy()
})
}
}
func (ss *Sim) MakeToolbar(p *tree.Plan) {
tree.Add(p, func(w *core.Chooser) {
w.SetStrings(cond.RunNames...)
w.SetCurrentValue(ss.Config.Env.RunName)
w.OnChange(func(e events.Event) {
ss.Config.Env.RunName = w.CurrentItem.Value.(string)
ss.InitEnvRun()
})
})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "Init", Icon: icons.Update,
Tooltip: "Initialize everything including network weights, and start over. Also applies current params.",
Active: egui.ActiveStopped,
Func: func() {
ss.Init()
ss.GUI.UpdateWindow()
},
})
ss.GUI.AddLooperCtrl(p, ss.Loops, []etime.Modes{etime.Train})
tree.Add(p, func(w *core.Separator) {})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "Save Wts", Icon: icons.Save,
Tooltip: "Save weights for the current condition name.",
Active: egui.ActiveStopped,
Func: func() {
ss.SaveCondWeights()
// ss.GUI.UpdateWindow()
},
})
////////////////////////////////////////////////
tree.Add(p, func(w *core.Separator) {})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "Reset RunLog",
Icon: icons.Reset,
Tooltip: "Reset the accumulated log of all Runs, which are tagged with the ParamSet used",
Active: egui.ActiveAlways,
Func: func() {
ss.Logs.ResetLog(etime.Train, etime.Run)
ss.GUI.UpdatePlot(etime.Train, etime.Run)
},
})
////////////////////////////////////////////////
tree.Add(p, func(w *core.Separator) {})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "New Seed",
Icon: icons.Add,
Tooltip: "Generate a new initial random seed to get different results. By default, Init re-establishes the same initial seed every time.",
Active: egui.ActiveAlways,
Func: func() {
ss.RandSeeds.NewSeeds()
},
})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "Plot Drive & Effort",
Icon: icons.PlayArrow,
Tooltip: "Opens a new window to plot Rubicon Drive and Effort dynamics.",
Active: egui.ActiveAlways,
Func: func() {
go DriveEffortGUI()
},
})
ss.GUI.AddToolbarItem(p, egui.ToolbarItem{Label: "README",
Icon: icons.FileMarkdown,
Tooltip: "Opens your browser on the README file that contains instructions for how to run this model.",
Active: egui.ActiveAlways,
Func: func() {
core.TheApp.OpenURL("https://github.com/emer/axon/blob/main/examples/pvlv/README.md")
},
})
}
func (ss *Sim) RunGUI() {
ss.Init()
ss.ConfigGUI()
ss.GUI.Body.RunMainWindow()
}
func (ss *Sim) RunNoGUI() {
if ss.Config.Params.Note != "" {
mpi.Printf("Note: %s\n", ss.Config.Params.Note)
}
if ss.Config.Log.SaveWeights {
mpi.Printf("Saving final weights per run\n")
}
runName := ss.Params.RunName(ss.Config.Run.Run)
ss.Stats.SetString("RunName", runName) // used for naming logs, stats, etc
netName := ss.Net.Name
elog.SetLogFile(&ss.Logs, ss.Config.Log.Block, etime.Train, etime.Block, "blk", netName, runName)
elog.SetLogFile(&ss.Logs, ss.Config.Log.Cond, etime.Train, etime.Condition, "cnd", netName, runName)
elog.SetLogFile(&ss.Logs, ss.Config.Log.Trial, etime.Test, etime.Trial, "trl", netName, runName)
netdata := ss.Config.Log.NetData
if netdata {
mpi.Printf("Saving NetView data from testing\n")
ss.GUI.InitNetData(ss.Net, 200)
}
ss.Init()
mpi.Printf("Running %d Runs starting at %d\n", ss.Config.Run.NRuns, ss.Config.Run.Run)
ss.Loops.GetLoop(etime.Train, etime.Run).Counter.SetCurMaxPlusN(ss.Config.Run.Run, ss.Config.Run.NRuns)
if ss.Config.Run.GPU {
ss.Net.ConfigGPUnoGUI(&ss.Context)
}
mpi.Printf("Set NThreads to: %d\n", ss.Net.NThreads)
ss.Loops.Run(etime.Train)
ss.Logs.CloseLogFiles()
if netdata {
ss.GUI.SaveNetData(ss.Stats.String("RunName"))
}
ss.Net.GPU.Destroy() // safe even if no GPU
}