@genai-fi/nanogpt 0.4.5 → 0.5.1

package/dist/tfjs_backend-NucKez4s.js

@@ -0,0 +1,1010 @@
+import { o as h, i as f, E as $, ap as Te, k as _, g as Ee, aq as xe, ar as Ie, as as Le, at as Ne, au as be, av as Ce, aw as Pe, b as Q, ax as Fe, a9 as U, q as ae, p as ie, N as le, c as fe, ay as he, am as pe, az as je, t as S, y as $e, ai as Me, a4 as Be } from "./index-iNhkcAEQ.js";
+import { s as C, t as Ke, a as Ue, b as ve } from "./ops-ObfXLHYQ.js";
+import { r as Re, d as Ve } from "./dropout-kbDY39Ci.js";
+import { r as u } from "./reshape-DxTPgnwL.js";
+import { g as qe } from "./gather-Bxe1Qip8.js";
+import { s as Ge } from "./sum-B_92TaHD.js";
+import { m as A } from "./mat_mul-D0SifYfJ.js";
+import { c as M } from "./concat-Cxbo2sOz.js";
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function Je(e) {
+  const t = { x: f(e, "x", "sigmoid", "float32") };
+  return $.runKernel(Te, t);
+}
+const Ze = /* @__PURE__ */ h({ sigmoid_: Je });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function We(e, n, t) {
+  const r = f(e, "x", "clipByValue");
+  if (_(n <= t, () => `Error in clip: min (${n}) must be less than or equal to max (${t}).`), n === t)
+    return Ee(r.shape, n, r.dtype);
+  const s = { x: r }, o = { clipValueMin: n, clipValueMax: t };
+  return $.runKernel(xe, s, o);
+}
+const Ye = /* @__PURE__ */ h({ clipByValue_: We });
+function He(e) {
+  return M(
+    e,
+    0
+    /* axis */
+  );
+}
+const Qe = /* @__PURE__ */ h({ concat1d_: He });
+function Xe(e, n) {
+  return M(e, n);
+}
+const ze = /* @__PURE__ */ h({ concat2d_: Xe });
+function en(e, n) {
+  return M(e, n);
+}
+const nn = /* @__PURE__ */ h({ concat3d_: en });
+function tn(e, n) {
+  return M(e, n);
+}
+const rn = /* @__PURE__ */ h({ concat4d_: tn });
+/**
+ * @license
+ * Copyright 2020 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function sn(e) {
+  const t = { x: f(e, "x", "elu", "float32") };
+  return $.runKernel(Ie, t);
+}
+const ke = /* @__PURE__ */ h({ elu_: sn });
+/**
+ * @license
+ * Copyright 2020 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function on(e, n = 0.2) {
+  const r = { x: f(e, "x", "leakyRelu") }, s = { alpha: n };
+  return $.runKernel(Le, r, s);
+}
+const un = /* @__PURE__ */ h({ leakyRelu_: on });
+/**
+ * @license
+ * Copyright 2020 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function cn(e, n) {
+  const t = f(e, "x", "prelu"), r = f(n, "alpha", "prelu"), s = { x: t, alpha: r };
+  return $.runKernel(Ne, s);
+}
+const an = /* @__PURE__ */ h({ prelu_: cn });
+/**
+ * @license
+ * Copyright 2020 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function ln(e) {
+  const t = { x: f(e, "x", "relu") };
+  return $.runKernel(be, t);
+}
+const fn = /* @__PURE__ */ h({ relu_: ln });
+/**
+ * @license
+ * Copyright 2020 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function hn(e) {
+  const t = { x: f(e, "x", "relu6") };
+  return $.runKernel(Ce, t);
+}
+const pn = /* @__PURE__ */ h({ relu6_: hn });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function dn(e, n, t) {
+  const r = f(e, "x", "slice1d");
+  return _(r.rank === 1, () => `slice1d expects a rank-1 tensor, but got a rank-${r.rank} tensor`), C(r, [n], [t]);
+}
+const X = /* @__PURE__ */ h({ slice1d_: dn });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function mn(e, n, t) {
+  const r = f(e, "x", "slice2d");
+  return _(r.rank === 2, () => `slice2d expects a rank-2 tensor, but got a rank-${r.rank} tensor`), C(r, n, t);
+}
+const we = /* @__PURE__ */ h({ slice2d_: mn });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function gn(e, n, t) {
+  const r = f(e, "x", "slice3d");
+  return _(r.rank === 3, () => `slice3d expects a rank-3 tensor, but got a rank-${r.rank} tensor`), C(r, n, t);
+}
+const z = /* @__PURE__ */ h({ slice3d_: gn });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function $n(e, n, t) {
+  const r = f(e, "x", "slice4d");
+  return _(r.rank === 4, () => `slice4d expects a rank-4 tensor, but got a rank-${r.rank} tensor`), C(r, n, t);
+}
+const K = /* @__PURE__ */ h({ slice4d_: $n });
+/**
+ * @license
+ * Copyright 2018 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function kn(e, n = 0) {
+  const r = { x: f(e, "x", "step") }, s = { alpha: n };
+  return $.runKernel(Pe, r, s);
+}
+const wn = /* @__PURE__ */ h({ step_: kn });
+/**
+ * @license
+ * Copyright 2019 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function An(e, n, t) {
+  if (t == null || t === "linear")
+    return e;
+  if (t === "relu")
+    return Q(e, wn(n));
+  throw new Error(`Cannot compute gradient for fused activation ${t}.`);
+}
+function Sn(e, n) {
+  let t = n;
+  const r = Fe(e.shape, n.shape);
+  return r.length > 0 && (t = Ge(t, r)), u(t, e.shape);
+}
+function yn(e, n, t, r) {
+  if (n === "linear")
+    return e;
+  if (n === "relu")
+    return fn(e);
+  if (n === "elu")
+    return ke(e);
+  if (n === "relu6")
+    return pn(e);
+  if (n === "prelu")
+    return an(e, t);
+  if (n === "leakyrelu")
+    return un(e, r);
+  if (n === "sigmoid")
+    return Ze(e);
+  throw new Error(`Unknown fused activation ${n}.`);
+}
+const On = (e, n) => !(e > 0) || n === "linear";
+/**
+ * @license
+ * Copyright 2019 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+function _n({ a: e, b: n, transposeA: t = !1, transposeB: r = !1, bias: s, activation: o = "linear", preluActivationWeights: i, leakyreluAlpha: p = 0.2 }) {
+  if (On($.state.gradientDepth, o) === !1) {
+    let D = A(e, n, t, r);
+    return s != null && (D = U(D, s)), yn(D, o, i, p);
+  }
+  let c = f(e, "a", "fused matMul"), a = f(n, "b", "fused matMul");
+  [c, a] = ae(c, a);
+  const k = t ? c.shape[c.rank - 2] : c.shape[c.rank - 1], g = r ? a.shape[a.rank - 1] : a.shape[a.rank - 2], E = t ? c.shape[c.rank - 1] : c.shape[c.rank - 2], d = r ? a.shape[a.rank - 2] : a.shape[a.rank - 1], ne = c.shape.slice(0, -2), x = a.shape.slice(0, -2), te = ie(ne), re = ie(x);
+  _(k === g, () => `Error in fused matMul: inner shapes (${k}) and (${g}) of Tensors with shapes ${c.shape} and ${a.shape} and transposeA=${t} and transposeB=${r} must match.`);
+  const R = le(c.shape.slice(0, -2), a.shape.slice(0, -2)).concat([E, d]), V = t ? u(c, [te, k, E]) : u(c, [te, E, k]), q = r ? u(a, [re, d, g]) : u(a, [re, g, d]);
+  let I;
+  s != null && (I = f(s, "bias", "fused matMul"), [I] = ae(I, c), le(R, I.shape));
+  let se;
+  i != null && (se = f(i, "prelu weights", "fused matMul"));
+  const oe = (D, P) => {
+    const [y, O, T, B] = P, w = An(u(D, T.shape), T, o);
+    let L, N;
+    if (!t && !r ? (L = A(w, O, !1, !0), N = A(y, w, !0, !1)) : !t && r ? (L = A(w, O, !1, !1), N = A(w, y, !0, !1)) : t && !r ? (L = A(O, w, !1, !0), N = A(y, w, !1, !1)) : (L = A(O, w, !0, !0), N = A(w, y, !0, !0)), s != null) {
+      const De = Sn(B, w);
+      return [L, N, De];
+    } else
+      return [L, N];
+  }, ue = {
+    a: V,
+    b: q,
+    bias: I,
+    preluActivationWeights: se
+  }, ce = { transposeA: t, transposeB: r, activation: o, leakyreluAlpha: p };
+  return s == null ? fe((P, y, O) => {
+    const T = (
+      // tslint:disable-next-line: no-unnecessary-type-assertion
+      $.runKernel(he, ue, ce)
+    );
+    return O([P, y, T]), { value: u(T, R), gradFunc: oe };
+  })(V, q) : fe((P, y, O, T) => {
+    const B = (
+      // tslint:disable-next-line: no-unnecessary-type-assertion
+      $.runKernel(he, ue, ce)
+    );
+    return T([P, y, B, O]), { value: u(B, R), gradFunc: oe };
+  })(V, q, I);
+}
+const de = /* @__PURE__ */ h({ fusedMatMul_: _n });
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+const Dn = ["channelsFirst", "channelsLast"], Tn = ["nearest", "bilinear"], En = ["valid", "same", "causal"], xn = ["max", "avg"], Gn = ["sum", "mul", "concat", "ave"];
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+class Ae extends Error {
+  constructor(n) {
+    super(n), Object.setPrototypeOf(this, Ae.prototype);
+  }
+}
+class Se extends Error {
+  constructor(n) {
+    super(n), Object.setPrototypeOf(this, Se.prototype);
+  }
+}
+class l extends Error {
+  constructor(n) {
+    super(n), Object.setPrototypeOf(this, l.prototype);
+  }
+}
+class j extends Error {
+  constructor(n) {
+    super(n), Object.setPrototypeOf(this, j.prototype);
+  }
+}
+class ee extends Error {
+  constructor(n) {
+    super(n), Object.setPrototypeOf(this, ee.prototype);
+  }
+}
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+function Jn(e, n) {
+  if (Array.isArray(e)) {
+    let t = [];
+    for (let r = 0; r < n; r++)
+      t = t.concat(e);
+    return t;
+  } else {
+    const t = new Array(n);
+    return t.fill(e), t;
+  }
+}
+function me(e, n) {
+  if (!e)
+    throw new ee(n);
+}
+function Zn(e, n) {
+  let t = 0;
+  for (const r of e)
+    r === n && t++;
+  return t;
+}
+function Wn(e) {
+  return e.length === 1 ? e[0] : e;
+}
+function Yn(e) {
+  return Array.isArray(e) ? e : [e];
+}
+function Hn(e) {
+  const t = e.replace(/(.)([A-Z][a-z0-9]+)/g, "$1_$2").replace(/([a-z])([A-Z])/g, "$1_$2").toLowerCase();
+  return t[0] !== "_" ? t : "private" + t;
+}
+function Qn(e) {
+  return e.length <= 1 || e.indexOf("_") === -1 ? e : e.replace(/[_]+(\w|$)/g, (n, t) => t.toUpperCase());
+}
+let m = {};
+function Xn(e) {
+  if (e == null)
+    return null;
+  const n = {};
+  return n.className = e.getClassName(), n.config = e.getConfig(), n;
+}
+function W(e) {
+  if (!(e == null || typeof e != "object"))
+    if (Array.isArray(e))
+      e.forEach((n) => W(n));
+    else {
+      const n = Object.keys(e);
+      for (const t of n) {
+        const r = e[t];
+        r != null && typeof r == "object" && (!Array.isArray(r) && r.type === "ndarray" && typeof r.value == "number" ? e[t] = r.value : W(r));
+      }
+    }
+}
+function zn(e, n = {}, t = {}, r = "object", s = !1) {
+  if (typeof e == "string") {
+    const o = e;
+    let i;
+    if (o in t)
+      i = t[o];
+    else if (o in m)
+      i = m[o];
+    else if (i = n[o], i == null)
+      throw new l(`Unknown ${r}: ${e}. This may be due to one of the following reasons:
+1. The ${r} is defined in Python, in which case it needs to be ported to TensorFlow.js or your JavaScript code.
+2. The custom ${r} is defined in JavaScript, but is not registered properly with tf.serialization.registerClass().`);
+    return i;
+  } else {
+    const o = e;
+    if (o.className == null || o.config == null)
+      throw new l(`${r}: Improper config format: ${JSON.stringify(o)}.
+'className' and 'config' must set.`);
+    const i = o.className;
+    let p, c;
+    if (i in t ? [p, c] = t[i] : i in m ? [p, c] = m.className : i in n && ([p, c] = n[i]), p == null)
+      throw new l(`Unknown ${r}: ${i}. This may be due to one of the following reasons:
+1. The ${r} is defined in Python, in which case it needs to be ported to TensorFlow.js or your JavaScript code.
+2. The custom ${r} is defined in JavaScript, but is not registered properly with tf.serialization.registerClass().`);
+    if (c != null) {
+      const a = {};
+      for (const d of Object.keys(m))
+        a[d] = m[d];
+      for (const d of Object.keys(t))
+        a[d] = t[d];
+      const k = o.config;
+      k.customObjects = a;
+      const g = Object.assign({}, m);
+      for (const d of Object.keys(t))
+        m[d] = t[d];
+      W(o.config);
+      const E = c(p, o.config, t, s);
+      return m = Object.assign({}, g), E;
+    } else {
+      const a = Object.assign({}, m);
+      for (const g of Object.keys(t))
+        m[g] = t[g];
+      const k = new p(o.config);
+      return m = Object.assign({}, a), k;
+    }
+  }
+}
+function In(e, n) {
+  return e < n ? -1 : e > n ? 1 : 0;
+}
+function et(e, n) {
+  return -1 * In(e, n);
+}
+function nt(e) {
+  if (e == null)
+    return e;
+  const n = [];
+  for (const t of e)
+    n.indexOf(t) === -1 && n.push(t);
+  return n;
+}
+function tt(e) {
+  if (e == null)
+    throw new l(`Invalid value in obj: ${JSON.stringify(e)}`);
+  for (const n in e)
+    if (e.hasOwnProperty(n))
+      return !1;
+  return !0;
+}
+function v(e, n, t) {
+  if (t != null && e.indexOf(t) < 0)
+    throw new l(`${t} is not a valid ${n}.  Valid values are ${e} or null/undefined.`);
+}
+function rt(e, n, t = 0, r = 1 / 0) {
+  return me(t >= 0), me(r >= t), Array.isArray(e) && e.length >= t && e.length <= r && e.every((s) => typeof s === n);
+}
+function Ln(e, n) {
+  Array.isArray(e) ? (_(e.length > 0, () => `${n} is unexpectedly an empty array.`), e.forEach((t, r) => Ln(t, `element ${r + 1} of ${n}`))) : _(Number.isInteger(e) && e > 0, () => `Expected ${n} to be a positive integer, but got ${ye(e)}.`);
+}
+function ye(e) {
+  return e === null ? "null" : Array.isArray(e) ? "[" + e.map((n) => ye(n)).join(",") + "]" : typeof e == "string" ? `"${e}"` : `${e}`;
+}
+function st(e, n, t) {
+  let r = t != null ? t() : pe(), s;
+  return (...i) => {
+    const p = t != null ? t() : pe();
+    return p - r < n || (r = p, s = e(...i)), s;
+  };
+}
+function ot(e) {
+  return e === "relu" ? "relu" : e === "linear" ? "linear" : e === "elu" ? "elu" : null;
+}
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+const b = /* @__PURE__ */ new Map();
+function Nn(e) {
+  v(Dn, "DataFormat", e);
+}
+function ut(e) {
+  v(Tn, "InterpolationFormat", e);
+}
+function ct(e) {
+  v(En, "PaddingMode", e);
+}
+function at(e) {
+  v(xn, "PoolMode", e);
+}
+const F = [], ge = "/";
+function it(e, n) {
+  F.push(e);
+  try {
+    const t = n();
+    return F.pop(), t;
+  } catch (t) {
+    throw F.pop(), t;
+  }
+}
+function bn() {
+  return F.length === 0 ? "" : F.join(ge) + ge;
+}
+function lt(e) {
+  if (!Oe(e))
+    throw new Error("Not a valid tensor name: '" + e + "'");
+  return bn() + e;
+}
+function ft(e) {
+  if (!Oe(e))
+    throw new Error("Not a valid tensor name: '" + e + "'");
+  b.has(e) || b.set(e, 0);
+  const n = b.get(e);
+  if (b.set(e, b.get(e) + 1), n > 0) {
+    const t = `${e}_${n}`;
+    return b.set(t, 1), t;
+  } else
+    return e;
+}
+const Cn = new RegExp(/^[A-Za-z0-9][-A-Za-z0-9\._\/]*$/);
+function Oe(e) {
+  return !!e.match(Cn);
+}
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+function ht(e) {
+  return e === parseInt(e.toString(), 10);
+}
+function _e(e, n, t) {
+  n == null && (n = 0), t == null && (t = e.length);
+  let r = 1;
+  for (let s = n; s < t; ++s)
+    r *= e[s];
+  return r;
+}
+function pt(e) {
+  if (e.length === 0)
+    return Number.NaN;
+  let n = Number.POSITIVE_INFINITY;
+  for (let t = 0; t < e.length; t++) {
+    const r = e[t];
+    r < n && (n = r);
+  }
+  return n;
+}
+function dt(e) {
+  if (e.length === 0)
+    return Number.NaN;
+  let n = Number.NEGATIVE_INFINITY;
+  for (let t = 0; t < e.length; t++) {
+    const r = e[t];
+    r > n && (n = r);
+  }
+  return n;
+}
+function mt(e, n) {
+  if (n < e)
+    throw new l(`end (${n}) < begin (${e}) is forbidden.`);
+  const t = [];
+  for (let r = e; r < n; ++r)
+    t.push(r);
+  return t;
+}
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+let G;
+function gt() {
+  return G == null && (G = je().epsilon()), G;
+}
+function Y() {
+  return "channelsLast";
+}
+/**
+ * @license
+ * Copyright 2018 Google LLC
+ *
+ * Use of this source code is governed by an MIT-style
+ * license that can be found in the LICENSE file or at
+ * https://opensource.org/licenses/MIT.
+ * =============================================================================
+ */
+function $t(e, n) {
+  return $e(e, n);
+}
+function Pn(e, n = -1) {
+  const t = e.shape.slice();
+  return n < 0 && (n = t.length + n + 1), t.splice(n, 0, 1), u(e, t);
+}
+function kt(e, n) {
+  return S(() => {
+    if (e.shape.length !== 2)
+      throw new l(`repeat() expects a rank-2 tensor, but received a rank-${e.shape.length} tensor.`);
+    const t = Pn(e, 1);
+    return Fn(t, [1, n, 1]);
+  });
+}
+function wt(e) {
+  const n = [_e(e.shape)];
+  return u(e, n);
+}
+function At(e) {
+  if (e.rank <= 1)
+    throw new l(`batchFlatten requires a minimum rank of 2. Got rank: ${e.rank}.`);
+  const n = [e.shape[0], _e(e.shape, 1)];
+  return u(e, n);
+}
+function J(e, n, t) {
+  return S(() => {
+    switch (e.rank) {
+      case 1:
+        return X(e, n, t);
+      case 2:
+        return we(e, [n, 0], [t, e.shape[1]]);
+      case 3:
+        return z(e, [n, 0, 0], [t, e.shape[1], e.shape[2]]);
+      case 4:
+        return K(e, [n, 0, 0, 0], [t, e.shape[1], e.shape[2], e.shape[3]]);
+      case 5:
+        return C(e, [n, 0, 0, 0, 0], [
+          t,
+          e.shape[1],
+          e.shape[2],
+          e.shape[3],
+          e.shape[4]
+        ]);
+      case 6:
+        return C(e, [n, 0, 0, 0, 0, 0], [
+          t,
+          e.shape[1],
+          e.shape[2],
+          e.shape[3],
+          e.shape[4],
+          e.shape[5]
+        ]);
+      default:
+        throw new l(`sliceAlongFirstAxis() received an unsupported tensor rank: ${e.rank}`);
+    }
+  });
+}
+function Z(e, n, t) {
+  return S(() => {
+    switch (e.rank) {
+      case 1:
+        return X(e, n, t);
+      case 2:
+        return we(e, [0, n], [e.shape[0], t]);
+      case 3:
+        return z(e, [0, 0, n], [e.shape[0], e.shape[1], t]);
+      case 4:
+        return K(e, [0, 0, 0, n], [e.shape[0], e.shape[1], e.shape[2], t]);
+      default:
+        throw new l(`sliceAlongLastAxis() received an unsupported tensor rank: ${e.rank}`);
+    }
+  });
+}
+function St(e, n, t, r) {
+  return S(() => {
+    switch (e.rank) {
+      case 1:
+        return X(e, n, t);
+      case 2:
+        switch (r) {
+          case 1:
+            return J(e, n, t);
+          case 2:
+            return Z(e, n, t);
+          default:
+            throw new l(`The axis is not within the rank of the tensor ${r}`);
+        }
+      case 3:
+        switch (r) {
+          case 1:
+            return J(e, n, t);
+          case 2:
+            return z(e, [0, n, 0], [e.shape[0], t, e.shape[2]]);
+          case 3:
+            return Z(e, n, t);
+          default:
+            throw new l(`The axis is not within the rank of the tensor ${r}`);
+        }
+      case 4:
+        switch (r) {
+          case 1:
+            return J(e, n, t);
+          case 2:
+            return K(e, [0, n, 0, 0], [e.shape[0], t, e.shape[2], e.shape[3]]);
+          case 3:
+            return K(e, [0, 0, n, 0], [e.shape[0], e.shape[1], t, e.shape[3]]);
+          case 4:
+            return Z(e, n, t);
+          default:
+            throw new l(`The axis is not within the rank of the tensor ${r}`);
+        }
+      default:
+        throw new l(`sliceAlongLastAxis() received an unsupported tensor rank: ${e.rank}`);
+    }
+  });
+}
+function yt(e, n = -1) {
+  let t;
+  return n < 0 && (t = e[0].rank, t !== 0 ? n = t : n = 0), n === e[0].rank && (n = -1), M(e, n);
+}
+function Ot(e, n) {
+  switch (e.rank) {
+    case 1:
+      return Qe([e, n]);
+    case 2:
+      return ze([e, n], 0);
+    case 3:
+      return nn([e, n], 0);
+    case 4:
+      return rn([e, n], 0);
+    default:
+      throw new l(`concatAlongFirstAxis() received an unsupported tensor rank: ${e.rank}`);
+  }
+}
+function Fn(e, n) {
+  if (Array.isArray(n) || (n = [n]), e.rank !== n.length)
+    throw new l(`The length of input n (${n.length}) does not match the number of dimensions in input x (${e.rank})`);
+  return Ue(e, n);
+}
+function _t(e, n = 0, t = 1, r, s) {
+  return Re(e, n, t, r, s);
+}
+function Dt(e, n, t, r) {
+  if (e.rank < 2 || n.rank < 2)
+    throw new j(`dot requires both inputs to be rank >= 2 but got x shape = ${e.shape} and y shape = ${n.shape}`);
+  if (n.rank >= 3) {
+    const s = e.shape.slice(-1)[0], o = n.shape.slice(-2)[0];
+    if (s !== o)
+      throw new j(`If rank y >= 3, then the second last dim of y must equal the last dim of x but got x shape = ${e.shape} and  y shape = ${n.shape}`);
+  }
+  if (e.rank === 2 && n.rank === 2)
+    return de({
+      a: e,
+      b: n,
+      transposeA: !1,
+      transposeB: !1,
+      bias: r ? H(e.rank, r, Y()) : null,
+      activation: t
+    });
+  {
+    const s = e.shape.slice(), o = s.pop();
+    e = u(e, [-1, o]);
+    const i = n.shape.slice(), p = i.pop(), c = i.pop(), a = [...i, p], k = Array.from({ length: n.rank }, (ne, x) => x === 0 ? n.rank - 2 : x <= n.rank - 2 ? x - 1 : x);
+    n = u(ve(n, k), [c, -1]);
+    const g = [...s, ...a];
+    return u(de({
+      a: e,
+      b: n,
+      transposeA: !1,
+      transposeB: !1,
+      bias: r ? H(e.rank, r, Y()) : null,
+      activation: t
+    }), g);
+  }
+}
+function Tt(e, n, t) {
+  return S(() => (Array.isArray(n) ? n = Ke(n, "int32") : n = $e(n, "int32"), qe(e, n, t)));
+}
+function Et(e) {
+  return Q(e, e);
+}
+function H(e, n, t) {
+  const r = n.shape;
+  if (n.rank !== 1 && n.rank !== e)
+    throw new l(`Unexpected bias dimensions: ${n.rank}; expected it to be 1 or ${e}`);
+  if (e === 5) {
+    if (t === "channelsFirst")
+      return r.length === 1 ? u(n, [1, r[0], 1, 1, 1]) : u(n, [1, r[3], r[0], r[1], r[2]]);
+    if (t === "channelsLast")
+      return r.length === 1 ? u(n, [1, 1, 1, 1, r[0]]) : u(n, [1].concat(r));
+  } else if (e === 4) {
+    if (t === "channelsFirst")
+      return r.length === 1 ? u(n, [1, r[0], 1, 1]) : u(n, [1, r[2], r[0], r[1]]);
+    if (t === "channelsLast")
+      return r.length === 1 ? u(n, [1, 1, 1, r[0]]) : u(n, [1].concat(r));
+  } else if (e === 3) {
+    if (t === "channelsFirst")
+      return r.length === 1 ? u(n, [1, r[0], 1]) : u(n, [1, r[1], r[0]]);
+    if (t === "channelsLast")
+      return r.length === 1 ? u(n, [1, 1, r[0]]) : u(n, [1].concat(r));
+  } else if (e < 3)
+    return n;
+  throw new l(`Unsupported input rank by biasAdd: ${n.rank}`);
+}
+function xt(e, n, t) {
+  return S(() => (t == null && (t = Y()), Nn(t), U(e, H(e.rank, n, t))));
+}
+function It(e, n = 1) {
+  if (n !== 1)
+    throw new j(`Support for alpha values other than 1 (${n}) is not implemented yet.`);
+  return ke(e);
+}
+function Lt(e) {
+  return S(() => Me(e, U(Be(e), 1)));
+}
+function Nt(e, n, t, r) {
+  return S(() => Ve(e, n, t, r));
+}
+function bt(e) {
+  return S(() => {
+    const n = U(0.5, Q(0.2, e));
+    return Ye(n, 0, 1);
+  });
+}
+function Ct(e, n, t = !1) {
+  return t ? e() : n();
+}
+export {
+  Ln as $,
+  Ae as A,
+  $t as B,
+  Qn as C,
+  nt as D,
+  et as E,
+  Jn as F,
+  tt as G,
+  J as H,
+  Pn as I,
+  Tt as J,
+  mt as K,
+  Zn as L,
+  It as M,
+  j as N,
+  bt as O,
+  Lt as P,
+  un as Q,
+  Se as R,
+  an as S,
+  ke as T,
+  dt as U,
+  l as V,
+  ht as W,
+  rt as X,
+  xt as Y,
+  St as Z,
+  ut as _,
+  yn as a,
+  ot as a0,
+  ct as a1,
+  Y as a2,
+  Fn as a3,
+  pt as a4,
+  Ot as a5,
+  Ct as a6,
+  Nt as a7,
+  yt as a8,
+  At as a9,
+  kt as aa,
+  at as ab,
+  Gn as ac,
+  Sn as b,
+  v as c,
+  Dt as d,
+  Nn as e,
+  _e as f,
+  An as g,
+  zn as h,
+  Xn as i,
+  lt as j,
+  ft as k,
+  Wn as l,
+  Yn as m,
+  it as n,
+  wn as o,
+  Ze as p,
+  Ye as q,
+  _t as r,
+  On as s,
+  Hn as t,
+  gt as u,
+  fn as v,
+  st as w,
+  wt as x,
+  Et as y,
+  me as z
+};