https://www.gopili.de/_next/static/chunks/398ee818.22303a0cdcd30d55.js

1!function(){try{var e="u">typeof window?window:"u">typeof global?global:"u">typeof globalThis?globalThis:"u">typeof self?self:{};e._sentryModuleMetadata=e._sentryModuleMetadata||{},e._sentryModuleMetadata[(new e.Error).stack]=function(e){for(var t=1;t<arguments.length;t++){var s=arguments[t];if(null!=s)for(var n in s)s.hasOwnProperty(n)&&(e[n]=s[n])}return e}({},e._sentryModuleMetadata[(new e.Error).stack],{"_sentryBundlerPluginAppKey:kombo-app":!0})}catch(e){}}();"use strict";(self.webpackChunk_N_E=self.webpackChunk_N_E||[]).push([[5458],{29397:(e,t,s)=>{s.d(t,{m:()=>y});var n=s(47740),i=s(11631),o=s(98112),r=s(77086),a=s(69140),l=s(9933),u=s(61952),h=s(27487),p=s(65385),d=s(48124),f=s(42648),c=s(90224);class y extends c.Wd{constructor(e){if(super({}),this.containerNodes=new Set,this.name=e.name,null==this.name){const e=this.getClassName().toLowerCase();this.name=(0,i.v)(e)}if(this.supportsMasking=!1,this.trainable_=!0,Array.isArray(e.inputs)?this.inputs=e.inputs.slice():this.inputs=[e.inputs],Array.isArray(e.outputs)?this.outputs=e.outputs.slice():this.outputs=[e.outputs],a.Am(this.inputs).length!==this.inputs.length)throw new o.Qp(`The list of inputs passed to the model is redundant. All inputs should only appear once. Found: ${this.inputs.map(e=>e.name)}`);for(const e of(a.Am(this.outputs).length!==this.outputs.length&&console.warn(`The list of outputs passed to the model is redundant. All outputs should only appear once. Found: ${this.outputs.map(e=>e.name)}`),this.inputLayers=[],this.inputLayersNodeIndices=[],this.inputLayersTensorIndices=[],this.outputLayers=[],this.outputLayersNodeIndices=[],this.outputLayersTensorIndices=[],this.layers=[],this.internalContainerRefs=[],this.outputs)){const t=e.sourceLayer,s=e.nodeIndex,n=e.tensorIndex;this.outputLayers.push(t),this.outputLayersNodeIndices.push(s),this.outputLayersTensorIndices.push(n)}for(const e of this.inputs){const t=e.sourceLayer,s=e.nodeIndex,n=e.tensorIndex;a.vA(0===s,"input layer has >1 nodes"),a.vA(0===n,"input layer has >1 tensors"),this.inputLayers.push(t),this.inputLayersNodeIndices.push(s),this.inputLayersTensorIndices.push(n)}this.inputNames=[],this.outputNames=[],this.feedInputShapes=[],this.feedInputNames=[],this.feedOutputNames=[];for(let t=0;t<this.inputLayers.length;t++){const s=this.inputLayers[t];if(!(s instanceof f.m))throw TypeError(`Input layers to a LayersModel must be InputLayer objects. Received inputs: ${e.inputs}. Input ${t} (0-based) originates from layer type ${s.getClassName()}.`);this.inputNames.push(s.name),this.feedInputShapes.push(s.batchInputShape),this.feedInputNames.push(s.name)}for(const e of this.outputLayers)this.outputNames.push(e.name);this.internalInputShapes=this.inputs.map(e=>e.shape),this.internalOutputShapes=this.outputs.map(e=>e.shape);const t={},s={},n={},r={},l={},u=[],h=(e,t,s,n,i,r)=>{(null==n||null==i||null==r)&&(n=e.sourceLayer,i=e.nodeIndex,r=e.tensorIndex);let a=n.inboundNodes[i];if(-1!==s.indexOf(a))throw new o.bu(`The tensor ${e.name} at layer "${n.name}" is part of a cycle.`);if(-1!==t.indexOf(a))return;this.containerNodes.add(y.nodeKey(n,i)),n.id in l||(l[n.id]=Object.keys(l).length),-1===s.indexOf(a)&&s.push(a);let p=a.inboundLayers.length;for(let e=0;e<p;e++)h(a.inputTensors[e],t,s,a.inboundLayers[e],a.nodeIndices[e],a.tensorIndices[e]);for(t.push(a);s.indexOf(a)>=0;)s.splice(s.indexOf(a),1);u.push(a)},p=[],d=[];for(const e of this.outputs)h(e,p,d);for(const e of u.slice().reverse()){s[e.id]=e,e.id in t||(t[e.id]=0);let i=t[e.id];i=Math.max(i,null==n[e.outboundLayer.id]?0:n[e.outboundLayer.id]),n[e.outboundLayer.id]=i,r[e.outboundLayer.id]=e.outboundLayer,t[e.id]=i;for(let n=0;n<e.inboundLayers.length;n++){const o=e.inboundLayers[n],r=e.nodeIndices[n],a=o.inboundNodes[r],l=null==t[a.id]?0:t[a.id];t[a.id]=Math.max(i+1,l),s[a.id]=a}}const m={};for(const e in t){const n=t[e];n in m||(m[n]=[]),m[n].push(s[e])}const g={};for(const e in n){const t=n[e];t in g||(g[t]=[]),g[t].push(r[e])}let b=Object.keys(g).map(e=>parseInt(e,10)).sort(a.th);for(const e of(this.layers=[],b)){const t=g[e];for(const e of(t.sort((e,t)=>

1{let s=l[e.id],n=l[t.id];return s<n?-1:+(s>n)}),t))e instanceof y&&this.internalContainerRefs.push(e),this.layers.push(e)}this.layersByDepth=g,b=Object.keys(m).map(e=>parseInt(e,10)).sort(a.th);const L=this.inputs.slice(),w=[];for(const e of b)for(const t of m[e]){const e=t.outboundLayer;if(null!=e){for(const s of t.inputTensors)if(-1===L.indexOf(s))throw new o.bu(`Graph disconnected: cannot obtain value for tensor ${s} at layer "${e.name}". The following previous layers were accessed without issue: ${w}`);for(const e of t.outputTensors)L.push(e);w.push(e.name)}}this.nodesByDepth=m;const N=this.layers.map(e=>e.name);for(const e of N){const t=N.filter(t=>t===e).length;if(1!==t)throw new o.bu(`The name "${e}" is used ${t} times in the model. All layer names should be unique. Layer names: `+JSON.stringify(N))}this.outboundNodes=[],this.inboundNodes=[],new c.bP({outboundLayer:this,inboundLayers:[],nodeIndices:[],tensorIndices:[],inputTensors:this.inputs,outputTensors:this.outputs,inputMasks:this.inputs.map(e=>null),outputMasks:this.outputs.map(e=>null),inputShapes:this.inputs.map(e=>e.shape),outputShapes:this.outputs.map(e=>e.shape)}),this.built=!0,this._refCount=1}assertNotDisposed(){if(0===this._refCount)throw Error(`Container '${this.name}' is already disposed.`)}dispose(){this.assertNotDisposed();let e={refCountAfterDispose:null,numDisposedVariables:0};if(0==--this._refCount){for(let t of this.layers)e.numDisposedVariables+=t.dispose().numDisposedVariables;for(let t of this.internalContainerRefs)e.numDisposedVariables+=t.dispose().numDisposedVariables}return e.refCountAfterDispose=this._refCount,e}get trainable(){return this.trainable_}set trainable(e){this.layers.forEach(t=>{t._trainableWeights.forEach(t=>t.trainable=e)}),this.trainable_=e}get trainableWeights(){if(this._trainableWeights.length>0)throw new o.Qp("Container instance unexpectedly contains _trainableWeights.The trainable weights of a Container are a union of the trainable weights of its consituent Layers. Its own _trainableWeights must remain an empty Array.");if(!this.trainable)return[];let e=[];for(let t of this.layers)e=e.concat(t.trainableWeights);return e}get nonTrainableWeights(){let e=[];for(let t of this.layers)e.push(...t.nonTrainableWeights);if(!this.trainable){let t=[];for(let e of this.layers)t.push(...e.trainableWeights);return t.concat(e)}return e}get weights(){return this.trainableWeights.concat(this.nonTrainableWeights)}loadWeights(e,t=!0){let s={},n=0,i=(e=>{let t=Object.keys(e);if(0===t.length)return!1;let s=t[0].split("/");return!isNaN(parseInt(s[s.length-1],10))})(e);for(let t of(i&&this.parseWeights(e),this.layers))for(let[e,r]of t.weights.entries()){let t=i?`${r.name.split("/").slice(0,-1).join("/")+"/"}${e}`:r.originalName;if(null!=s[t])throw new o.Qp(`Duplicate weight name: ${t}`);s[t]=r,n++}let r=[];for(let n in e){let i=n;if(null==s[n]){let e=n.split("/");i=e.slice(0,-2).concat([e[e.length-1]]).join("/")}if(null!=s[i])r.push([s[i],e[n]]);else if(t)throw new o.Qp(`Provided weight data has no target variable: ${n}`);delete s[i]}if(t){let e=[];for(let t in s)e.push(t);if(e.length>0)throw new o.Qp(`${e.length} of ${n} weights are not set: ${e}`)}(0,h.UM)(r)}parseWeights(e){for(let t in Object.keys(e)){let s=t.split("/"),n=["vars","layer_checkpoint_dependencies"],i=s.map(e=>e.startsWith("_")?e.slice(1):e).filter(e=>!n.includes(e)).join("/");i!==t&&(e[i]=e[t],delete e[t])}}updatedConfig(){let e=this.getConfig(),t={};return t.className=this.getClassName(),t.config=e,t.kerasVersion=`tfjs-layers ${p.r}`,t.backend="TensorFlow.js",t}toJSON(e,t=!0){let s=(0,l.M)(this.updatedConfig());return t?JSON.stringify(s):s}call(e,t){return(0,n.DZQ)(()=>{e=a.st(e);let s=new d.RW;for(let t=0;t<this.inputs.length;++t)s.add(this.inputs[t],e[t]);return(0,d.g7)(this.outputs,s,t)})}computeMask(e,t){return(0,n.DZQ)(()=>{let s;return e=a.st(e),s=null==t?a.fD(null,e.length):a.st(t),this.runInternalGraph(e,s)[1]})}computeOutputShape(e){let t=u.FS(e);if(t.length!==this.inputLayers.length)throw new o.Qp(`Invalid inputShape argument ${e}: model has ${this.inputLayers.length} tensor inputs.`);let s={};for(let e=0;e<t.length;e++){let n=this.inputLayers[e],i=t[e];s[n.name+"_0_0"]=i}let n=Object.keys(this.nodesByDepth).map(e=>parseInt(e,10)).sort(a.th);if(n.length>1)for(let e of n)for(let t of this.nodesByDepth[e]){let e=t.outboundLayer;if(-1!==this.inputLayers.map(e=>e.id).indexOf(e.id))continue;let n=[];for(let e=0;

1e<t.inboundLayers.length;e++){let i=t.inboundLayers[e],o=t.nodeIndices[e],r=t.tensorIndices[e],a=s[`${i.name}_${o}_${r}`];n.push(a)}let i=e.computeOutputShape(a.wL(n)),o=u.FS(i),r=e.inboundNodes.indexOf(t);for(let t=0;t<o.length;t++)s[`${e.name}_${r}_${t}`]=o[t]}let i=[],r=[];for(let e=0;e<this.outputLayers.length;e++){let t=this.outputLayers[e],s=this.outputLayersNodeIndices[e],n=this.outputLayersTensorIndices[e],i=`${t.name}_${s}_${n}`;r.push(i)}for(let e=0;e<r.length;e++){let t=r[e];a.vA(t in s),i.push(s[t])}return a.wL(i)}runInternalGraph(e,t){null==t&&(t=a.fD(null,e.length));let s={};for(let n=0;n<this.inputs.length;++n){let i=this.inputs[n],o=e[n],r=t[n];s[i.id]=[o,r]}for(let e of Object.keys(this.nodesByDepth).map(e=>parseInt(e,10)).sort(a.th))for(let t of this.nodesByDepth[e]){let e=t.outboundLayer,n=t.inputTensors,i=t.outputTensors,r=[];for(let e of n)e.id in s&&r.push(s[e.id]);if(r.length===n.length){let n,l,u,h,p={};if(null!=t.callArgs&&(p=t.callArgs),1===r.length){let[t,s]=r[0];null==p.mask&&(p.mask=s),u=a.st(e.call(t,p)),h=a.st(e.computeMask(t,s)),n=[t],l=[s]}else n=r.map(e=>e[0]),l=r.map(e=>e[1]),null==p.mask&&(p.mask=l),u=a.st(e.call(n,p)),h=a.st(e.computeMask(n,l));if(e.activityRegularizer)throw new o.EH("LayersModel invocation with concrete Tensor value(s) in the presence of activity regularizer(s) is not supported yet.");for(let e=0;e<i.length;++e){let t=i[e],n=u[e],o=h[e];s[t.id]=[n,o]}}}let n=[],i=[],r=[];for(let e of this.outputs){a.vA(e.id in s,`Could not compute output ${e.name} : ${e.id}`);let[t,o]=s[e.id];r.push(t.shape),n.push(t),i.push(o)}return[n,i,r]}buildNodeConversionMap(e){let t,s={};for(let e of this.layers){t=+(e instanceof y);for(let n=0;n<e.inboundNodes.length;n++){let i=y.nodeKey(e,n);this.containerNodes.has(i)&&(s[i]=t,t+=1)}}return s}getLayer(e,t){if(null!=t)return this.findLayer(t);if(null==e)throw new o.Qp("Provide either a layer name or layer index");if("number"==typeof e)return this.findLayer(e);for(let t of this.layers)if(t.name===e)return t;throw new o.Qp(`No such layer: ${e}`)}findLayer(e){if(!(this.layers.length<=e))return this.layers[e];throw new o.Qp(`Was asked to retrieve layer at index ${e}, but model only has ${this.layers.length} layer(s).`)}calculateLosses(){return(0,n.DZQ)(()=>{let e=[];for(let t of this.layers)for(let s=0;s<t.inboundNodes.length;++s){let n=y.nodeKey(t,s);this.containerNodes.has(n)&&e.push(...t.calculateLosses())}return e})}getConfig(){let e={name:this.name},t=this.buildNodeConversionMap(this.layers),s=[];for(let e of this.layers){let n=e.getClassName(),i=e.getConfig(),o=[];for(let s=0;s<e.inboundNodes.length;s++){let n=e.inboundNodes[s],i=y.nodeKey(e,s),r={};if(this.containerNodes.has(i)){if(n.callArgs)try{JSON.stringify(n.callArgs),r=n.callArgs}catch(t){console.warn(`Layer ${e.name} was passed non-serializable keyword arguments: ${n.callArgs}. They will not be included in the serialized model (and thus will be missing at deserialization time).`),r={}}if(n.inboundLayers.length>0){let e=[];for(let s=0;s<n.inboundLayers.length;s++){let i=n.inboundLayers[s],o=n.nodeIndices[s],a=n.tensorIndices[s],l=t[y.nodeKey(i,o)];null==l&&(l=0),e.push([i.name,l,a,r])}o.push(e)}}}let r={};r.name=e.name,r.className=n,r.config=i,r.inboundNodes=o,s.push(r)}e.layers=s;let n=[];for(let e=0;e<this.inputLayers.length;e++){let s=this.inputLayers[e],i=this.inputLayersNodeIndices[e],o=y.nodeKey(s,i);if(!this.containerNodes.has(o))continue;let r=t[o];null==r&&(r=0);let a=this.inputLayersTensorIndices[e];n.push([s.name,r,a])}e.inputLayers=n;let i=[];for(let e=0;e<this.outputLayers.length;e++){let s=this.outputLayers[e],n=this.outputLayersNodeIndices[e],o=y.nodeKey(s,n);if(!this.containerNodes.has(o))continue;let r=t[o];null==r&&(r=0);let a=this.outputLayersTensorIndices[e];i.push([s.name,r,a])}return e.outputLayers=i,e}static fromConfig(e,t,s={},n=!1){let i={},l={};function u(e,t){e.name in l?l[e.name].push(t):l[e.name]=[t]}let h=t.name,p=t.layers;for(let e of p)!function(e){let s=e.name,a=(0,r.i)(e,null!=t.customObjects?t.customObjects:{});

1a.setFastWeightInitDuringBuild(n),i[s]=a,e.inboundNodes.forEach(e=>{if(!(e instanceof Array))throw new o.Qp(`Corrupted configuration, expected array for nodeData: ${e}`);u(a,e)})}(e);for(;!a.ZF(l);)for(let e of p){let t=i[e.name];if(t.name in l){let e=l[t.name];for(let s of(delete l[t.name],e))!function(e,t){let s,n=[];for(let o of t){let r=o[0],a=o[1],l=o[2];if(s=null==o[3]?{}:o[3],!(r in i))return void u(e,t);let h=i[r];if(h.inboundNodes.length<=a)return void u(e,t);let p=h.inboundNodes[a];n.push(p.outputTensors[l])}n.length>0&&e.apply(a.wL(n),s)}(t,s)}}let d=[],f=[];for(let e of t.inputLayers){let t=e[0],s=e[1],n=e[2];a.vA(t in i);let o=i[t].inboundNodes[s].outputTensors;d.push(o[n])}for(let e of t.outputLayers){let t=e[0],s=e[1],n=e[2];a.vA(t in i);let o=i[t].inboundNodes[s].outputTensors;f.push(o[n])}return new e({inputs:d,outputs:f,name:h})}get stateful(){if(this._stateful)throw new o.Qp("Container instance unexpectedly has _stateful = true. The statefulness of a Container is determined by the Layers it contains. Its _stateful property must remain the default false.");for(let e of this.layers)if(e.stateful)return!0;return!1}resetStates(){(0,n.DZQ)(()=>{this.layers.forEach(e=>{e.stateful&&e.resetStates()})})}}}}]);