nivSrSSKJr SSKJr SSKJr SSKJs Jr "SS\ 5r g!\ a SSK rNf=f)DNormalize tensors based on streaming estimates of mean and variance.)absolute_import)division)print_functionNcF\rSrSrSrS SjrSrSrSrSr Sr S r S r g ) StreamingNormalizercX lX0lX@lXPl[R "U5 [R "SS5Ul[R "[R"U5S5Ul [R "[R"U5S5Ul SSS5 g!,(df  g=f)aNormalize tensors based on streaming estimates of mean and variance. Centering the value, scaling it by the standard deviation, and clipping outlier values are optional. Args: template: Example tensor providing shape and dtype of the vaule to track. center: Python boolean indicating whether to subtract mean from values. scale: Python boolean indicating whether to scale values by stddev. clip: If and when to clip normalized values. name: Parent scope of operations provided by this class. rFN) _center_scale_clip_nametf name_scopeVariable_count zeros_like_mean_var_sum)selftemplatecenterscaleclipnames \/home/james-whalen/.local/lib/python3.13/site-packages/pybullet_envs/agents/ppo/normalize.py__init__StreamingNormalize.__init__sxLKJJ t KK5)dk;;r}}X6>djkk"--"95Adm   s A=B55 Cc^[R"TRS-5 URRTR RR:HnU(aUSnTR (aUTR S-nTR(a3U[R"TRS:U4SjU4Sj5S-nTR(a-[R"UTR*TR5nU(aUSn[R"US5sSSS5 $!,(df  g=f) zNormalize a single or batch tensor. Applies the activated transformations in the constructor using current estimates of mean and variance. Args: value: Batch or single value tensor. Returns: Normalized batch or single value tensor. z /transformN.c*>TR5S-$)Ng:0yE>)_stdrsr.StreamingNormalize.transform..IsTYY[4%7cD>[R"TR5$N)r ones_likerr$srr%r&IsdmmA\r'Nrvalue) rrrshapendimsrr r condrr clip_by_valuecheck_numerics)rr+ no_batch_dims` r transformStreamingNormalize.transform3s tzzL0 1[[&&$***:*:*@*@@l i   I&&   KK!O79\^^bd d    TZZ@ a   ug .! 2 1 1s D D:: EcZ^^[R"URS-5 TRRUR RR:XaTSm[R"T5Sn[R "URRU5/5 [R"UR[R5n[R"TUR S- S5nUR XC- -m[R"URS:U4SjU4Sj5mTUR S- TTS- -nUR[R"US5-nSSS5 [R "TW/5 UR RT5URRU54nSSS5 [R "W5 TRRS:Xa[R"T5mUR!S[R"T55sSSS5 sSSS5 $!,(df  N=f!,(df  N=f!,(df  O=fSSS5 g!,(df  g=f) zrUpdate the mean and variance estimates. Args: value: Batch or single value tensor. Returns: Summary tensor. z/updater rr!c>T$r))new_meansrr%+StreamingNormalize.update..csHr'c>TS$)Nrr6)r+srr%r8cs eAhr'Nr+)rrrr,r-rcontrol_dependenciesr assign_addcastfloat32 reduce_sumr.rassign reduce_mean_summary) rr+countstep mean_delta var_delta new_var_sumupdater7s ` @rrGStreamingNormalize.updateQs tzzI- .   djj..44 4i hhuoa e " "DKK$:$:5$A#B Cwwt{{BJJ/]]54::i+@#@!D :: 11774;;?,<>NOTZZ 22ux ?R7RS mmbmmIq&AA D  " "Hk#: ;""8,dmm.B.B;.OO < " "6 * ;;   !..'%}}WbnnU&;< + * / . D C < ; * * * / . .sVBJ-CI J!7I0J6AJ J I- )J0 I> :J J J J*c [R"URS-5 [R"URR S5UR R [R"UR 55URR [R"UR555sSSS5 $!,(df  g=f)zgReset the estimates of mean and variance. Resets the full state of this class. Returns: Operation. z/resetrN) rrrgrouprr?rrrr$s rresetStreamingNormalize.resetms tzzH, - XXdkk((+TZZ->->r}}TZZ?X-Ymm**2==+GHJ . - -s BC  Cct^[R"TRS-5 [R"TRS:U4Sj[ 5n[R"TRS:U4Sj[ 5n[R RX/5sSSS5 $!,(df  g=f)zKSummary string of mean and standard deviation. Returns: Summary tensor. z/summaryrc<>TRSTR5$)Nmean)rArr$srr%,StreamingNormalize.summary..sdmmFDJJ6Wr'r!cD>TRSTR55$)Nstddev)rAr#r$srr%rPsT]]8TYY[5Yr'N)rrrr.rstrsummarymerge)r mean_summary std_summarys` rrTStreamingNormalize.summaryyss tzzJ. /WWT[[1_.WY\]lGGDKK!O-Y[^_k ZZ  |9 : 0 / /s A:B)) B7c^[R"TRS:U4SjU4Sj5n[R"US-5$)zComputes the current estimate of the standard deviation. Note that the standard deviation is not defined until at least two samples were seen. Returns: Tensor of current variance. r!c>TR[R"TRS- [R5- $)Nr!)rrr<rr=r$srr%)StreamingNormalize._std..s+  aA%1%r'c\>[R"TR5[S5-$)Nnan)rr*rfloatr$srr%r[sbll4==.IERWL.Xr'g-C6?)rr.rsqrt)rvariances` rr#StreamingNormalize._stds;wwt{{Q)%&XZH 778d? ##r'cURRS:Xa[RR X5$[RR X5$)zCreate a scalar or histogram summary matching the rank of the tensor. Args: name: Name for the summary. tensor: Tensor to summarize. Returns: Summary tensor. r)r,r-rrTscalar histogram)rrtensors rrAStreamingNormalize._summarys@||Q ZZ  t ,, ZZ ! !$ //r')r r rrrr rN)TT normalize) __name__ __module__ __qualname____firstlineno____doc__rr2rGrKrTr#rA__static_attributes__r6r'rrrs*LB,/<=8 J ;$ 0r'r) rm __future__rrrtensorflow.compat.v1compatv1r Exception tensorflowobjectrr6r'rrvsAK&%## G0G0 s + 88