@genai-fi/nanogpt 0.6.3 → 0.7.1

package/dist/{ops-C0sQEcPw.js → ops-aRTXR2Sr.js}

@@ -1,19 +1,22 @@
-import { o as h, q as l, E as f, aO as hn, i as p, aP as fn, h as w, aQ as bn, aR as dn, aS as gn, I as $n, ah as xn, B as j, aT as kn, aU as yn, aV as In, aW as _n, aX as En, aY as Mn, aZ as Nn, L as V, a_ as An, a$ as Tn, b0 as Sn, aK as wn, V as zn, O as Bn, Q as qn, b1 as Kn, t as Rn, b2 as J, b3 as Gn, b4 as Vn, b5 as Dn, b6 as Pn, b7 as vn, b8 as jn, s as Fn, b9 as On, ba as Wn, a as k, a7 as Z, aj as R, f as Yn, b as S, bb as Cn, H as L, d as Jn } from "./index-BAzbokzv.js";
-import { r as _ } from "./reshape-krWGKraP.js";
-import { s as un } from "./split-eiktj-6L.js";
-import { s as q } from "./sum-BE_Irnim.js";
-import { b as X } from "./broadcast_to-DARN-DBD.js";
-import { r as W } from "./range-DYsrnfiy.js";
-import { t as Ln } from "./tensor-Xyi595sG.js";
-import { s as Y } from "./stack-dfEEz2OY.js";
-import { c as Xn, z as Qn } from "./zeros-YJDE7oRb.js";
-import { n as ln } from "./norm-CzltS9Fz.js";
-import { c as Q } from "./concat-5aPGqw3Z.js";
-import { m as T } from "./mat_mul-7121rsJk.js";
-import { t as H } from "./tensor2d-CPEkynbH.js";
+import { B as m, C as u, l, E as f, aP as gn, k as S, aQ as $n, aR as kn, aS as yn, p as En, aT as w, q as rn, x as _n, b as pn, aU as In, aV as Nn, N as An, al as Tn, I as v, aW as xn, aX as wn, aY as Sn, aZ as Mn, a_ as zn, a$ as qn, b0 as Bn, X as K, b1 as Gn, b2 as Kn, b3 as Rn, aG as Dn, $ as vn, Z as Vn, _ as Pn, b4 as jn, t as Fn, b5 as L, b6 as Wn, b7 as Yn, b8 as On, b9 as Cn, ba as Ln, bb as Xn, bc as Zn, bd as Jn, a as $, o as U, n as q, h as Qn, c as A, be as Hn, M as X, d as Un } from "./index-UdZhlibC.js";
+import { n as ne, a as ee, b as se } from "./non_max_suppression_impl-CsEgBuMA.js";
+import { r as k } from "./reshape-BR0eoLYN.js";
+import { s as mn } from "./split-CRU0PjVV.js";
+import { s as E } from "./sum-NLYbiDag.js";
+import { b as Z } from "./broadcast_to-Ba9h_8DO.js";
+import { s as _ } from "./slice-BHbDHObE.js";
+import { r as W } from "./range-D9CZhVlR.js";
+import { t as te } from "./tensor-Do9PKbIE.js";
+import { s as Y } from "./stack-ikk2Y8_P.js";
+import { c as re, z as ae } from "./zeros-DnPT2nD4.js";
+import { e as oe } from "./axis_util-5DTW2tFV.js";
+import { m as J } from "./max-ByjEGoFx.js";
+import { c as Q } from "./concat-CbXTetof.js";
+import { m as N } from "./mat_mul-dwmZz69e.js";
+import { t as H } from "./tensor2d-CWHxHpLh.js";
 /**
  * @license
- * Copyright 2018 Google LLC. All Rights Reserved.
+ * 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
@@ -27,14 +30,13 @@ import { t as H } from "./tensor2d-CPEkynbH.js";
  * limitations under the License.
  * =============================================================================
  */
-function Hn(r, t, n) {
-  const e = l(r, "x", "slice", "string_or_numeric");
-  if (e.rank === 0)
-    throw new Error("Slicing scalar is not possible");
-  const s = { x: e }, a = { begin: t, size: n };
-  return f.runKernel(hn, s, a);
+function ie(s, e, n) {
+  const r = u(s, "x", "bincount"), t = u(e, "weights", "bincount");
+  l(r.dtype === "int32", () => `Error in bincount: input dtype must be int32, but got ${r.dtype}`), l(n >= 0, () => `size must be non-negative, but got ${n}.`), l(t.size === r.size || t.size === 0, () => `Error in bincount: weights must have the same size as input or0-length, but got input shape: ${r.shape}, weights shape: ${t.shape}.`);
+  const a = { x: r, weights: t }, o = { size: n };
+  return f.runKernel(gn, a, o);
 }
-const M = /* @__PURE__ */ h({ slice_: Hn });
+const ce = /* @__PURE__ */ m({ bincount_: ie });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -51,16 +53,18 @@ const M = /* @__PURE__ */ h({ slice_: Hn });
  * limitations under the License.
  * =============================================================================
  */
-function Zn(r, t, n) {
-  const e = l(r, "x", "bincount"), s = l(t, "weights", "bincount");
-  p(e.dtype === "int32", () => `Error in bincount: input dtype must be int32, but got ${e.dtype}`), p(n >= 0, () => `size must be non-negative, but got ${n}.`), p(s.size === e.size || s.size === 0, () => `Error in bincount: weights must have the same size as input or0-length, but got input shape: ${e.shape}, weights shape: ${s.shape}.`);
-  const a = { x: e, weights: s }, o = { size: n };
-  return f.runKernel(fn, a, o);
+function ue(s, e, n) {
+  const r = u(e, "a", "where"), t = u(n, "b", "where"), a = u(s, "condition", "where", "bool"), o = S(S(a.shape, r.shape), t.shape), i = Z(a, o), p = Z(r, o), c = Z(t, o), b = {
+    condition: i,
+    t: p,
+    e: c
+  };
+  return f.runKernel($n, b);
 }
-const Un = /* @__PURE__ */ h({ bincount_: Zn });
+const G = /* @__PURE__ */ m({ where_: ue });
 /**
  * @license
- * Copyright 2020 Google LLC. All Rights Reserved.
+ * Copyright 2021 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
@@ -74,18 +78,14 @@ const Un = /* @__PURE__ */ h({ bincount_: Zn });
  * limitations under the License.
  * =============================================================================
  */
-function ne(r, t, n) {
-  const e = l(t, "a", "where"), s = l(n, "b", "where"), a = l(r, "condition", "where", "bool"), o = w(w(a.shape, e.shape), s.shape), i = X(a, o), u = X(e, o), c = X(s, o), b = {
-    condition: i,
-    t: u,
-    e: c
-  };
-  return f.runKernel(bn, b);
+function le(s, ...e) {
+  const n = e.map((t, a) => u(t, `tensors${a}`, "einsum")), r = { equation: s };
+  return f.runKernel(kn, n, r);
 }
-const G = /* @__PURE__ */ h({ where_: ne });
+const D = /* @__PURE__ */ m({ einsum_: le });
 /**
  * @license
- * Copyright 2021 Google LLC. All Rights Reserved.
+ * Copyright 2020 Google Inc. All Rights Reserved.
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
  * You may obtain a copy of the License at
@@ -99,11 +99,67 @@ const G = /* @__PURE__ */ h({ where_: ne });
  * limitations under the License.
  * =============================================================================
  */
-function ee(r, ...t) {
-  const n = t.map((s, a) => l(s, `tensors${a}`, "einsum")), e = { equation: r };
-  return f.runKernel(dn, n, e);
+function pe(s, e = null, n = !1) {
+  const t = { x: u(s, "x", "min") }, a = { axis: e, keepDims: n };
+  return f.runKernel(yn, t, a);
 }
-const v = /* @__PURE__ */ h({ einsum_: ee });
+const an = /* @__PURE__ */ m({ min_: pe });
+/**
+ * @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 me(s, e = "euclidean", n = null, r = !1) {
+  s = u(s, "x", "norm");
+  const t = fn(s, e, n);
+  let a = t.shape;
+  if (r) {
+    const o = En(n, s.shape);
+    a = oe(t.shape, o);
+  }
+  return k(t, a);
+}
+function fn(s, e, n = null) {
+  if (s.rank === 0)
+    return w(s);
+  if (s.rank !== 1 && n === null)
+    return fn(k(s, [-1]), e, n);
+  if (s.rank === 1 || typeof n == "number" || Array.isArray(n) && n.length === 1) {
+    if (e === 1)
+      return E(w(s), n);
+    if (e === 1 / 0)
+      return J(w(s), n);
+    if (e === -1 / 0)
+      return an(w(s), n);
+    if (e === "euclidean" || e === 2)
+      return rn(E(_n(w(s), pn(2, "int32")), n));
+    throw new Error(`Error in norm: invalid ord value: ${e}`);
+  }
+  if (Array.isArray(n) && n.length === 2) {
+    if (e === 1)
+      return J(E(w(s), n[0]), n[1] - 1);
+    if (e === 1 / 0)
+      return J(E(w(s), n[1]), n[0]);
+    if (e === -1 / 0)
+      return an(E(w(s), n[1]), n[0]);
+    if (e === "fro" || e === "euclidean")
+      return rn(E(In(s), n));
+    throw new Error(`Error in norm: invalid ord value: ${e}`);
+  }
+  throw new Error(`Error in norm: invalid axis: ${n}`);
+}
+const hn = /* @__PURE__ */ m({ norm_: me });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -120,13 +176,13 @@ const v = /* @__PURE__ */ h({ einsum_: ee });
  * limitations under the License.
  * =============================================================================
  */
-function se(r, t = 0) {
-  const n = l(r, "x", "expandDims", "string_or_numeric");
-  p(t <= n.rank, () => "Axis must be <= rank of the tensor");
-  const e = { input: n }, s = { dim: t };
-  return f.runKernel(gn, e, s);
+function fe(s, e = 0) {
+  const n = u(s, "x", "expandDims", "string_or_numeric");
+  l(e <= n.rank, () => "Axis must be <= rank of the tensor");
+  const r = { input: n }, t = { dim: e };
+  return f.runKernel(Nn, r, t);
 }
-const K = /* @__PURE__ */ h({ expandDims_: se });
+const z = /* @__PURE__ */ m({ expandDims_: fe });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -143,13 +199,13 @@ const K = /* @__PURE__ */ h({ expandDims_: se });
  * limitations under the License.
  * =============================================================================
  */
-function te(r, t) {
-  const n = l(r, "x", "tile", "string_or_numeric");
-  p(n.rank === t.length, () => `Error in transpose: rank of input ${n.rank} must match length of reps ${t}.`);
-  const e = { x: n }, s = { reps: t };
-  return f.runKernel($n, e, s);
+function he(s, e) {
+  const n = u(s, "x", "tile", "string_or_numeric");
+  l(n.rank === e.length, () => `Error in transpose: rank of input ${n.rank} must match length of reps ${e}.`);
+  const r = { x: n }, t = { reps: e };
+  return f.runKernel(An, r, t);
 }
-const O = /* @__PURE__ */ h({ tile_: te });
+const F = /* @__PURE__ */ m({ tile_: he });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -166,20 +222,20 @@ const O = /* @__PURE__ */ h({ tile_: te });
  * limitations under the License.
  * =============================================================================
  */
-function re(r, t, n, e = "float32") {
-  t == null && (t = r);
-  const s = xn([r, t], e), a = r <= t ? r : t;
+function be(s, e, n, r = "float32") {
+  e == null && (e = s);
+  const t = Tn([s, e], r), a = s <= e ? s : e;
   for (let i = 0; i < a; ++i)
-    s.set(1, i, i);
-  const o = _(s.toTensor(), [r, t]);
+    t.set(1, i, i);
+  const o = k(t.toTensor(), [s, e]);
   if (n == null)
     return o;
   if (n.length === 1)
-    return O(K(o, 0), [n[0], 1, 1]);
+    return F(z(o, 0), [n[0], 1, 1]);
   if (n.length === 2)
-    return O(K(K(o, 0), 0), [n[0], n[1], 1, 1]);
+    return F(z(z(o, 0), 0), [n[0], n[1], 1, 1]);
   if (n.length === 3)
-    return O(K(K(K(o, 0), 0), 0), [
+    return F(z(z(z(o, 0), 0), 0), [
       n[0],
       n[1],
       n[2],
@@ -188,7 +244,7 @@ function re(r, t, n, e = "float32") {
     ]);
   throw new Error(`eye() currently supports only 1D and 2D batchShapes, but received ${n.length}D.`);
 }
-const ae = /* @__PURE__ */ h({ eye_: re });
+const de = /* @__PURE__ */ m({ eye_: be });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -205,13 +261,13 @@ const ae = /* @__PURE__ */ h({ eye_: re });
  * limitations under the License.
  * =============================================================================
  */
-function oe(r, t) {
-  let n = l(r, "a", "greater", "string_or_numeric"), e = l(t, "b", "greater", "string_or_numeric");
-  [n, e] = j(n, e), w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(kn, s);
+function ge(s, e) {
+  let n = u(s, "a", "greater", "string_or_numeric"), r = u(e, "b", "greater", "string_or_numeric");
+  [n, r] = v(n, r), S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(xn, t);
 }
-const U = /* @__PURE__ */ h({ greater_: oe });
+const nn = /* @__PURE__ */ m({ greater_: ge });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -228,13 +284,13 @@ const U = /* @__PURE__ */ h({ greater_: oe });
  * limitations under the License.
  * =============================================================================
  */
-function ie(r, t) {
-  let n = l(r, "a", "greaterEqual", "string_or_numeric"), e = l(t, "b", "greaterEqual", "string_or_numeric");
-  [n, e] = j(n, e), w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(yn, s);
+function $e(s, e) {
+  let n = u(s, "a", "greaterEqual", "string_or_numeric"), r = u(e, "b", "greaterEqual", "string_or_numeric");
+  [n, r] = v(n, r), S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(wn, t);
 }
-const ce = /* @__PURE__ */ h({ greaterEqual_: ie });
+const ke = /* @__PURE__ */ m({ greaterEqual_: $e });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -251,11 +307,11 @@ const ce = /* @__PURE__ */ h({ greaterEqual_: ie });
  * limitations under the License.
  * =============================================================================
  */
-function ue(r) {
-  const n = { input: l(r, "input", "imag") };
-  return f.runKernel(In, n);
+function ye(s) {
+  const n = { input: u(s, "input", "imag") };
+  return f.runKernel(Sn, n);
 }
-const le = /* @__PURE__ */ h({ imag_: ue });
+const Ee = /* @__PURE__ */ m({ imag_: ye });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -272,13 +328,13 @@ const le = /* @__PURE__ */ h({ imag_: ue });
  * limitations under the License.
  * =============================================================================
  */
-function pe(r, t) {
-  let n = l(r, "a", "less", "string_or_numeric"), e = l(t, "b", "less", "string_or_numeric");
-  [n, e] = j(n, e), w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(_n, s);
+function _e(s, e) {
+  let n = u(s, "a", "less", "string_or_numeric"), r = u(e, "b", "less", "string_or_numeric");
+  [n, r] = v(n, r), S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(Mn, t);
 }
-const tn = /* @__PURE__ */ h({ less_: pe });
+const on = /* @__PURE__ */ m({ less_: _e });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -295,13 +351,13 @@ const tn = /* @__PURE__ */ h({ less_: pe });
  * limitations under the License.
  * =============================================================================
  */
-function me(r, t) {
-  let n = l(r, "a", "lessEqual", "string_or_numeric"), e = l(t, "b", "lessEqual", "string_or_numeric");
-  [n, e] = j(n, e), w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(En, s);
+function Ie(s, e) {
+  let n = u(s, "a", "lessEqual", "string_or_numeric"), r = u(e, "b", "lessEqual", "string_or_numeric");
+  [n, r] = v(n, r), S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(zn, t);
 }
-const pn = /* @__PURE__ */ h({ lessEqual_: me });
+const bn = /* @__PURE__ */ m({ lessEqual_: Ie });
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -318,11 +374,11 @@ const pn = /* @__PURE__ */ h({ lessEqual_: me });
  * limitations under the License.
  * =============================================================================
  */
-function he(r) {
-  const n = { x: l(r, "x", "neg") };
-  return f.runKernel(Mn, n);
+function Ne(s) {
+  const n = { x: u(s, "x", "neg") };
+  return f.runKernel(qn, n);
 }
-const nn = /* @__PURE__ */ h({ neg_: he });
+const en = /* @__PURE__ */ m({ neg_: Ne });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -339,13 +395,13 @@ const nn = /* @__PURE__ */ h({ neg_: he });
  * limitations under the License.
  * =============================================================================
  */
-function fe(r, t) {
-  const n = l(r, "a", "logicalAnd", "bool"), e = l(t, "b", "logicalAnd", "bool");
-  w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(Nn, s);
+function Ae(s, e) {
+  const n = u(s, "a", "logicalAnd", "bool"), r = u(e, "b", "logicalAnd", "bool");
+  S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(Bn, t);
 }
-const be = /* @__PURE__ */ h({ logicalAnd_: fe });
+const Te = /* @__PURE__ */ m({ logicalAnd_: Ae });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -362,13 +418,13 @@ const be = /* @__PURE__ */ h({ logicalAnd_: fe });
  * limitations under the License.
  * =============================================================================
  */
-function de(r, t) {
-  let n = l(r, "a", "minimum"), e = l(t, "b", "minimum");
-  [n, e] = j(n, e), n.dtype === "bool" && (n = V(n, "int32"), e = V(e, "int32")), w(n.shape, e.shape);
-  const s = { a: n, b: e };
-  return f.runKernel(An, s);
+function xe(s, e) {
+  let n = u(s, "a", "minimum"), r = u(e, "b", "minimum");
+  [n, r] = v(n, r), n.dtype === "bool" && (n = K(n, "int32"), r = K(r, "int32")), S(n.shape, r.shape);
+  const t = { a: n, b: r };
+  return f.runKernel(Gn, t);
 }
-const rn = /* @__PURE__ */ h({ minimum_: de });
+const cn = /* @__PURE__ */ m({ minimum_: xe });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -385,11 +441,11 @@ const rn = /* @__PURE__ */ h({ minimum_: de });
  * limitations under the License.
  * =============================================================================
  */
-function ge(r) {
-  const n = { input: l(r, "input", "real") };
-  return f.runKernel(Tn, n);
+function we(s) {
+  const n = { input: u(s, "input", "real") };
+  return f.runKernel(Kn, n);
 }
-const $e = /* @__PURE__ */ h({ real_: ge });
+const Se = /* @__PURE__ */ m({ real_: we });
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -406,11 +462,11 @@ const $e = /* @__PURE__ */ h({ real_: ge });
  * limitations under the License.
  * =============================================================================
  */
-function xe(r) {
-  const n = { x: l(r, "x", "round") };
-  return f.runKernel(Sn, n);
+function Me(s) {
+  const n = { x: u(s, "x", "round") };
+  return f.runKernel(Rn, n);
 }
-const ke = /* @__PURE__ */ h({ round_: xe });
+const ze = /* @__PURE__ */ m({ round_: Me });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -427,11 +483,11 @@ const ke = /* @__PURE__ */ h({ round_: xe });
  * limitations under the License.
  * =============================================================================
  */
-function ye(r, t) {
-  const n = l(r, "x", "squeeze", "string_or_numeric");
-  return _(n, wn(n.shape, t).newShape);
+function qe(s, e) {
+  const n = u(s, "x", "squeeze", "string_or_numeric");
+  return k(n, Dn(n.shape, e).newShape);
 }
-const Ie = /* @__PURE__ */ h({ squeeze_: ye });
+const Be = /* @__PURE__ */ m({ squeeze_: qe });
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -448,12 +504,12 @@ const Ie = /* @__PURE__ */ h({ squeeze_: ye });
  * limitations under the License.
  * =============================================================================
  */
-function N(r, t) {
-  zn(r);
-  const n = Bn(r, t);
+function I(s, e) {
+  vn(s);
+  const n = Vn(s, e);
   if (n.length !== 1)
     throw new Error("tensor1d() requires values to be a flat/TypedArray");
-  return qn(r, null, n, t);
+  return Pn(s, null, n, e);
 }
 /**
  * @license
@@ -471,13 +527,13 @@ function N(r, t) {
  * limitations under the License.
  * =============================================================================
  */
-function _e(r, t = 0) {
-  const n = l(r, "x", "unstack", "string_or_numeric");
-  p(t >= -n.shape.length && t < n.shape.length, () => `Axis = ${t} is not in [-${n.shape.length}, ${n.shape.length})`);
-  const e = { value: n }, s = { axis: t };
-  return f.runKernel(Kn, e, s);
+function Ge(s, e = 0) {
+  const n = u(s, "x", "unstack", "string_or_numeric");
+  l(e >= -n.shape.length && e < n.shape.length, () => `Axis = ${e} is not in [-${n.shape.length}, ${n.shape.length})`);
+  const r = { value: n }, t = { axis: e };
+  return f.runKernel(jn, r, t);
 }
-const mn = /* @__PURE__ */ h({ unstack_: _e });
+const dn = /* @__PURE__ */ m({ unstack_: Ge });
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -494,19 +550,19 @@ const mn = /* @__PURE__ */ h({ unstack_: _e });
  * limitations under the License.
  * =============================================================================
  */
-function Ee(r, t, n) {
-  const e = l(r, "x", "transpose");
-  if (t == null && (t = e.shape.map((o, i) => i).reverse()), p(e.rank === t.length, () => `Error in transpose: rank of input ${e.rank} must match length of perm ${t}.`), t.forEach((o) => {
-    p(o >= 0 && o < e.rank, () => `All entries in 'perm' must be between 0 and ${e.rank - 1} but got ${t}`);
-  }), e.rank <= 1)
-    return e.clone();
-  const s = { x: e }, a = { perm: t };
-  return e.dtype === "complex64" ? Rn(() => {
-    let o = $e(e), i = le(e);
-    return o = f.runKernel(J, { x: o }, a), i = f.runKernel(J, { x: i }, a), n && (i = nn(i)), Xn(o, i);
-  }) : f.runKernel(J, s, a);
+function Ke(s, e, n) {
+  const r = u(s, "x", "transpose");
+  if (e == null && (e = r.shape.map((o, i) => i).reverse()), l(r.rank === e.length, () => `Error in transpose: rank of input ${r.rank} must match length of perm ${e}.`), e.forEach((o) => {
+    l(o >= 0 && o < r.rank, () => `All entries in 'perm' must be between 0 and ${r.rank - 1} but got ${e}`);
+  }), r.rank <= 1)
+    return r.clone();
+  const t = { x: r }, a = { perm: e };
+  return r.dtype === "complex64" ? Fn(() => {
+    let o = Se(r), i = Ee(r);
+    return o = f.runKernel(L, { x: o }, a), i = f.runKernel(L, { x: i }, a), n && (i = en(i)), re(o, i);
+  }) : f.runKernel(L, t, a);
 }
-const an = /* @__PURE__ */ h({ transpose_: Ee });
+const un = /* @__PURE__ */ m({ transpose_: Ke });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -523,13 +579,13 @@ const an = /* @__PURE__ */ h({ transpose_: Ee });
  * limitations under the License.
  * =============================================================================
  */
-function Me(r, t, n, e, s = "bilinear", a = 0) {
-  const o = l(r, "image", "cropAndResize"), i = l(t, "boxes", "cropAndResize", "float32"), u = l(n, "boxInd", "cropAndResize", "int32"), c = i.shape[0];
-  p(o.rank === 4, () => `Error in cropAndResize: image must be rank 4,but got rank ${o.rank}.`), p(i.rank === 2 && i.shape[1] === 4, () => `Error in cropAndResize: boxes must be have size [${c},4] but had shape ${i.shape}.`), p(u.rank === 1 && u.shape[0] === c, () => `Error in cropAndResize: boxInd must be have size [${c}] but had shape ${i.shape}.`), p(e.length === 2, () => `Error in cropAndResize: cropSize must be of length 2, but got length ${e.length}.`), p(e[0] >= 1 && e[1] >= 1, () => `cropSize must be atleast [1,1], but was ${e}`), p(s === "bilinear" || s === "nearest", () => `method must be bilinear or nearest, but was ${s}`);
-  const b = { image: o, boxes: i, boxInd: u }, m = { method: s, extrapolationValue: a, cropSize: e };
-  return f.runKernel(Gn, b, m);
+function Re(s, e, n, r, t = "bilinear", a = 0) {
+  const o = u(s, "image", "cropAndResize"), i = u(e, "boxes", "cropAndResize", "float32"), p = u(n, "boxInd", "cropAndResize", "int32"), c = i.shape[0];
+  l(o.rank === 4, () => `Error in cropAndResize: image must be rank 4,but got rank ${o.rank}.`), l(i.rank === 2 && i.shape[1] === 4, () => `Error in cropAndResize: boxes must be have size [${c},4] but had shape ${i.shape}.`), l(p.rank === 1 && p.shape[0] === c, () => `Error in cropAndResize: boxInd must be have size [${c}] but had shape ${i.shape}.`), l(r.length === 2, () => `Error in cropAndResize: cropSize must be of length 2, but got length ${r.length}.`), l(r[0] >= 1 && r[1] >= 1, () => `cropSize must be atleast [1,1], but was ${r}`), l(t === "bilinear" || t === "nearest", () => `method must be bilinear or nearest, but was ${t}`);
+  const b = { image: o, boxes: i, boxInd: p }, h = { method: t, extrapolationValue: a, cropSize: r };
+  return f.runKernel(Wn, b, h);
 }
-const Ne = /* @__PURE__ */ h({ cropAndResize_: Me });
+const De = /* @__PURE__ */ m({ cropAndResize_: Re });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -546,13 +602,13 @@ const Ne = /* @__PURE__ */ h({ cropAndResize_: Me });
  * limitations under the License.
  * =============================================================================
  */
-function Ae(r) {
-  const t = l(r, "image", "flipLeftRight", "float32");
-  p(t.rank === 4, () => `Error in flipLeftRight: image must be rank 4,but got rank ${t.rank}.`);
-  const n = { image: t };
-  return f.runKernel(Vn, n, {});
+function ve(s) {
+  const e = u(s, "image", "flipLeftRight", "float32");
+  l(e.rank === 4, () => `Error in flipLeftRight: image must be rank 4,but got rank ${e.rank}.`);
+  const n = { image: e };
+  return f.runKernel(Yn, n, {});
 }
-const Te = /* @__PURE__ */ h({ flipLeftRight_: Ae });
+const Ve = /* @__PURE__ */ m({ flipLeftRight_: ve });
 /**
  * @license
  * Copyright 2021 Google LLC. All Rights Reserved.
@@ -569,13 +625,13 @@ const Te = /* @__PURE__ */ h({ flipLeftRight_: Ae });
  * limitations under the License.
  * =============================================================================
  */
-function Se(r) {
-  const t = l(r, "image", "grayscaleToRGB"), n = t.rank - 1, e = t.shape[n];
-  p(t.rank >= 2, () => `Error in grayscaleToRGB: images must be at least rank 2, but got rank ${t.rank}.`), p(e === 1, () => `Error in grayscaleToRGB: last dimension of a grayscale image should be size 1, but got size ${e}.`);
-  const s = new Array(t.rank);
-  return s.fill(1, 0, n), s[n] = 3, O(t, s);
+function Pe(s) {
+  const e = u(s, "image", "grayscaleToRGB"), n = e.rank - 1, r = e.shape[n];
+  l(e.rank >= 2, () => `Error in grayscaleToRGB: images must be at least rank 2, but got rank ${e.rank}.`), l(r === 1, () => `Error in grayscaleToRGB: last dimension of a grayscale image should be size 1, but got size ${r}.`);
+  const t = new Array(e.rank);
+  return t.fill(1, 0, n), t[n] = 3, F(e, t);
 }
-const we = /* @__PURE__ */ h({ grayscaleToRGB_: Se });
+const je = /* @__PURE__ */ m({ grayscaleToRGB_: Pe });
 /**
  * @license
  * Copyright 2023 Google LLC.
@@ -592,56 +648,33 @@ const we = /* @__PURE__ */ h({ grayscaleToRGB_: Se });
  * limitations under the License.
  * =============================================================================
  */
-function ze(r) {
-  const t = l(r, "image", "RGBToGrayscale"), n = t.rank - 1, e = t.shape[n];
-  p(t.rank >= 2, () => `Error in RGBToGrayscale: images must be at least rank 2, but got rank ${t.rank}.`), p(e === 3, () => `Error in RGBToGrayscale: last dimension of an RGB image should be size 3, but got size ${e}.`);
-  const s = t.dtype, a = V(t, "float32"), o = N([0.2989, 0.587, 0.114]);
+function Fe(s) {
+  const e = u(s, "image", "RGBToGrayscale"), n = e.rank - 1, r = e.shape[n];
+  l(e.rank >= 2, () => `Error in RGBToGrayscale: images must be at least rank 2, but got rank ${e.rank}.`), l(r === 3, () => `Error in RGBToGrayscale: last dimension of an RGB image should be size 3, but got size ${r}.`);
+  const t = e.dtype, a = K(e, "float32"), o = I([0.2989, 0.587, 0.114]);
   let i;
-  switch (t.rank) {
+  switch (e.rank) {
     case 2:
-      i = v("ij,j->i", a, o);
+      i = D("ij,j->i", a, o);
       break;
     case 3:
-      i = v("ijk,k->ij", a, o);
+      i = D("ijk,k->ij", a, o);
       break;
     case 4:
-      i = v("ijkl,l->ijk", a, o);
+      i = D("ijkl,l->ijk", a, o);
       break;
     case 5:
-      i = v("ijklm,m->ijkl", a, o);
+      i = D("ijklm,m->ijkl", a, o);
       break;
     case 6:
-      i = v("ijklmn,n->ijklm", a, o);
+      i = D("ijklmn,n->ijklm", a, o);
       break;
     default:
       throw new Error("Not a valid tensor rank.");
   }
-  return i = K(i, -1), V(i, s);
-}
-const Be = /* @__PURE__ */ h({ rgbToGrayscale_: ze });
-/**
- * @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 qe(r, t, n = 0, e = 0.5) {
-  const s = l(r, "image", "rotateWithOffset", "float32");
-  p(s.rank === 4, () => `Error in rotateWithOffset: image must be rank 4,but got rank ${s.rank}.`);
-  const a = { image: s }, o = { radians: t, fillValue: n, center: e };
-  return f.runKernel(Dn, a, o);
+  return i = z(i, -1), K(i, t);
 }
-const Ke = /* @__PURE__ */ h({ rotateWithOffset_: qe });
+const We = /* @__PURE__ */ m({ rgbToGrayscale_: Fe });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -658,11 +691,13 @@ const Ke = /* @__PURE__ */ h({ rotateWithOffset_: qe });
  * limitations under the License.
  * =============================================================================
  */
-function D(r, t, n, e, s, a) {
-  e == null && (e = 0.5), s == null && (s = Number.NEGATIVE_INFINITY), a == null && (a = 0);
-  const o = r.shape[0];
-  return n = Math.min(n, o), p(0 <= e && e <= 1, () => `iouThreshold must be in [0, 1], but was '${e}'`), p(r.rank === 2, () => `boxes must be a 2D tensor, but was of rank '${r.rank}'`), p(r.shape[1] === 4, () => `boxes must have 4 columns, but 2nd dimension was ${r.shape[1]}`), p(t.rank === 1, () => "scores must be a 1D tensor"), p(t.shape[0] === o, () => `scores has incompatible shape with boxes. Expected ${o}, but was ${t.shape[0]}`), p(0 <= a && a <= 1, () => `softNmsSigma must be in [0, 1], but was '${a}'`), { maxOutputSize: n, iouThreshold: e, scoreThreshold: s, softNmsSigma: a };
+function Ye(s, e, n = 0, r = 0.5) {
+  const t = u(s, "image", "rotateWithOffset", "float32");
+  l(t.rank === 4, () => `Error in rotateWithOffset: image must be rank 4,but got rank ${t.rank}.`);
+  const a = { image: t }, o = { radians: e, fillValue: n, center: r };
+  return f.runKernel(On, a, o);
 }
+const Oe = /* @__PURE__ */ m({ rotateWithOffset_: Ye });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -679,47 +714,10 @@ function D(r, t, n, e, s, a) {
  * limitations under the License.
  * =============================================================================
  */
-function Re(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY) {
-  const a = l(r, "boxes", "nonMaxSuppression", "float32"), o = l(t, "scores", "nonMaxSuppression", "float32"), i = D(a, o, n, e, s);
-  n = i.maxOutputSize, e = i.iouThreshold, s = i.scoreThreshold;
-  const u = { maxOutputSize: n, iouThreshold: e, scoreThreshold: s };
-  return f.runKernel(Pn, { boxes: a, scores: o }, u);
-}
-const Ge = /* @__PURE__ */ h({ nonMaxSuppression_: Re });
-/**
- * @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 Ve(r, t, n) {
-  const e = De(r, t, n), s = e < 0 ? -(e + 1) : e;
-  r.splice(s, 0, t);
-}
-function De(r, t, n) {
-  return ve(r, t, n || Pe);
-}
-function Pe(r, t) {
-  return r > t ? 1 : r < t ? -1 : 0;
-}
-function ve(r, t, n) {
-  let e = 0, s = r.length, a = 0, o = !1;
-  for (; e < s; ) {
-    a = e + (s - e >>> 1);
-    const i = n(t, r[a]);
-    i > 0 ? e = a + 1 : (s = a, o = !i);
-  }
-  return o ? e : -e - 1;
+function R(s, e, n, r, t, a) {
+  r == null && (r = 0.5), t == null && (t = Number.NEGATIVE_INFINITY), a == null && (a = 0);
+  const o = s.shape[0];
+  return n = Math.min(n, o), l(0 <= r && r <= 1, () => `iouThreshold must be in [0, 1], but was '${r}'`), l(s.rank === 2, () => `boxes must be a 2D tensor, but was of rank '${s.rank}'`), l(s.shape[1] === 4, () => `boxes must have 4 columns, but 2nd dimension was ${s.shape[1]}`), l(e.rank === 1, () => "scores must be a 1D tensor"), l(e.shape[0] === o, () => `scores has incompatible shape with boxes. Expected ${o}, but was ${e.shape[0]}`), l(0 <= a && a <= 1, () => `softNmsSigma must be in [0, 1], but was '${a}'`), { maxOutputSize: n, iouThreshold: r, scoreThreshold: t, softNmsSigma: a };
 }
 /**
  * @license
@@ -737,84 +735,13 @@ function ve(r, t, n) {
  * limitations under the License.
  * =============================================================================
  */
-function je(r, t, n, e, s) {
-  return en(
-    r,
-    t,
-    n,
-    e,
-    s,
-    0
-    /* softNmsSigma */
-  );
-}
-function Fe(r, t, n, e, s, a) {
-  return en(
-    r,
-    t,
-    n,
-    e,
-    s,
-    0,
-    !1,
-    a,
-    !0
-    /* returnValidOutputs */
-  );
-}
-function Oe(r, t, n, e, s, a) {
-  return en(
-    r,
-    t,
-    n,
-    e,
-    s,
-    a,
-    !0
-    /* returnScoresTensor */
-  );
-}
-function en(r, t, n, e, s, a, o = !1, i = !1, u = !1) {
-  const c = [];
-  for (let d = 0; d < t.length; d++)
-    t[d] > s && c.push({ score: t[d], boxIndex: d, suppressBeginIndex: 0 });
-  c.sort(on);
-  const b = a > 0 ? -0.5 / a : 0, m = [], $ = [];
-  for (; m.length < n && c.length > 0; ) {
-    const d = c.pop(), { score: E, boxIndex: I, suppressBeginIndex: A } = d;
-    if (E < s)
-      break;
-    let z = !1;
-    for (let B = m.length - 1; B >= A; --B) {
-      const P = We(r, I, m[B]);
-      if (P >= e) {
-        z = !0;
-        break;
-      }
-      if (d.score = d.score * Ye(e, b, P), d.score <= s)
-        break;
-    }
-    d.suppressBeginIndex = m.length, z || (d.score === E ? (m.push(I), $.push(d.score)) : d.score > s && Ve(c, d, on));
-  }
-  const g = m.length, x = n - g;
-  i && x > 0 && (m.push(...new Array(x).fill(0)), $.push(...new Array(x).fill(0)));
-  const y = { selectedIndices: m };
-  return o && (y.selectedScores = $), u && (y.validOutputs = g), y;
-}
-function We(r, t, n) {
-  const e = r.subarray(t * 4, t * 4 + 4), s = r.subarray(n * 4, n * 4 + 4), a = Math.min(e[0], e[2]), o = Math.min(e[1], e[3]), i = Math.max(e[0], e[2]), u = Math.max(e[1], e[3]), c = Math.min(s[0], s[2]), b = Math.min(s[1], s[3]), m = Math.max(s[0], s[2]), $ = Math.max(s[1], s[3]), g = (i - a) * (u - o), x = (m - c) * ($ - b);
-  if (g <= 0 || x <= 0)
-    return 0;
-  const y = Math.max(a, c), d = Math.max(o, b), E = Math.min(i, m), I = Math.min(u, $), A = Math.max(E - y, 0) * Math.max(I - d, 0);
-  return A / (g + x - A);
-}
-function Ye(r, t, n) {
-  const e = Math.exp(t * n * n);
-  return n <= r ? e : 0;
-}
-function on(r, t) {
-  return r.score - t.score || r.score === t.score && t.boxIndex - r.boxIndex;
+function Ce(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY) {
+  const a = u(s, "boxes", "nonMaxSuppression", "float32"), o = u(e, "scores", "nonMaxSuppression", "float32"), i = R(a, o, n, r, t);
+  n = i.maxOutputSize, r = i.iouThreshold, t = i.scoreThreshold;
+  const p = { maxOutputSize: n, iouThreshold: r, scoreThreshold: t };
+  return f.runKernel(Cn, { boxes: a, scores: o }, p);
 }
+const Le = /* @__PURE__ */ m({ nonMaxSuppression_: Ce });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -831,13 +758,13 @@ function on(r, t) {
  * limitations under the License.
  * =============================================================================
  */
-async function Ce(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY) {
-  const a = l(r, "boxes", "nonMaxSuppressionAsync"), o = l(t, "scores", "nonMaxSuppressionAsync"), i = D(a, o, n, e, s);
-  n = i.maxOutputSize, e = i.iouThreshold, s = i.scoreThreshold;
-  const u = await Promise.all([a.data(), o.data()]), c = u[0], b = u[1], { selectedIndices: m } = je(c, b, n, e, s);
-  return a !== r && a.dispose(), o !== t && o.dispose(), N(m, "int32");
+async function Xe(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY) {
+  const a = u(s, "boxes", "nonMaxSuppressionAsync"), o = u(e, "scores", "nonMaxSuppressionAsync"), i = R(a, o, n, r, t);
+  n = i.maxOutputSize, r = i.iouThreshold, t = i.scoreThreshold;
+  const p = await Promise.all([a.data(), o.data()]), c = p[0], b = p[1], { selectedIndices: h } = ne(c, b, n, r, t);
+  return a !== s && a.dispose(), o !== e && o.dispose(), I(h, "int32");
 }
-const Je = Ce;
+const Ze = Xe;
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -854,13 +781,13 @@ const Je = Ce;
  * limitations under the License.
  * =============================================================================
  */
-function Le(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY, a = 0) {
-  const o = l(r, "boxes", "nonMaxSuppression"), i = l(t, "scores", "nonMaxSuppression"), u = D(o, i, n, e, s, a);
-  n = u.maxOutputSize, e = u.iouThreshold, s = u.scoreThreshold, a = u.softNmsSigma;
-  const c = { boxes: o, scores: i }, b = { maxOutputSize: n, iouThreshold: e, scoreThreshold: s, softNmsSigma: a }, m = f.runKernel(vn, c, b);
-  return { selectedIndices: m[0], selectedScores: m[1] };
+function Je(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY, a = 0) {
+  const o = u(s, "boxes", "nonMaxSuppression"), i = u(e, "scores", "nonMaxSuppression"), p = R(o, i, n, r, t, a);
+  n = p.maxOutputSize, r = p.iouThreshold, t = p.scoreThreshold, a = p.softNmsSigma;
+  const c = { boxes: o, scores: i }, b = { maxOutputSize: n, iouThreshold: r, scoreThreshold: t, softNmsSigma: a }, h = f.runKernel(Ln, c, b);
+  return { selectedIndices: h[0], selectedScores: h[1] };
 }
-const Xe = /* @__PURE__ */ h({ nonMaxSuppressionWithScore_: Le });
+const Qe = /* @__PURE__ */ m({ nonMaxSuppressionWithScore_: Je });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -877,16 +804,16 @@ const Xe = /* @__PURE__ */ h({ nonMaxSuppressionWithScore_: Le });
  * limitations under the License.
  * =============================================================================
  */
-async function Qe(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY, a = 0) {
-  const o = l(r, "boxes", "nonMaxSuppressionAsync"), i = l(t, "scores", "nonMaxSuppressionAsync"), u = D(o, i, n, e, s, a);
-  n = u.maxOutputSize, e = u.iouThreshold, s = u.scoreThreshold, a = u.softNmsSigma;
-  const c = await Promise.all([o.data(), i.data()]), b = c[0], m = c[1], { selectedIndices: $, selectedScores: g } = Oe(b, m, n, e, s, a);
-  return o !== r && o.dispose(), i !== t && i.dispose(), {
-    selectedIndices: N($, "int32"),
-    selectedScores: N(g)
+async function He(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY, a = 0) {
+  const o = u(s, "boxes", "nonMaxSuppressionAsync"), i = u(e, "scores", "nonMaxSuppressionAsync"), p = R(o, i, n, r, t, a);
+  n = p.maxOutputSize, r = p.iouThreshold, t = p.scoreThreshold, a = p.softNmsSigma;
+  const c = await Promise.all([o.data(), i.data()]), b = c[0], h = c[1], { selectedIndices: g, selectedScores: d } = ee(b, h, n, r, t, a);
+  return o !== s && o.dispose(), i !== e && i.dispose(), {
+    selectedIndices: I(g, "int32"),
+    selectedScores: I(d)
   };
 }
-const He = Qe;
+const Ue = He;
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -903,24 +830,24 @@ const He = Qe;
  * limitations under the License.
  * =============================================================================
  */
-function Ze(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY, a = !1) {
-  const o = l(r, "boxes", "nonMaxSuppression"), i = l(t, "scores", "nonMaxSuppression"), u = D(
+function ns(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY, a = !1) {
+  const o = u(s, "boxes", "nonMaxSuppression"), i = u(e, "scores", "nonMaxSuppression"), p = R(
     o,
     i,
     n,
-    e,
-    s,
+    r,
+    t,
     null
     /* softNmsSigma */
-  ), c = u.maxOutputSize, b = u.iouThreshold, m = u.scoreThreshold, $ = { boxes: o, scores: i }, g = {
+  ), c = p.maxOutputSize, b = p.iouThreshold, h = p.scoreThreshold, g = { boxes: o, scores: i }, d = {
     maxOutputSize: c,
     iouThreshold: b,
-    scoreThreshold: m,
+    scoreThreshold: h,
     padToMaxOutputSize: a
-  }, x = f.runKernel(jn, $, g);
-  return { selectedIndices: x[0], validOutputs: x[1] };
+  }, y = f.runKernel(Xn, g, d);
+  return { selectedIndices: y[0], validOutputs: y[1] };
 }
-const Ue = /* @__PURE__ */ h({ nonMaxSuppressionPadded_: Ze });
+const es = /* @__PURE__ */ m({ nonMaxSuppressionPadded_: ns });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -937,22 +864,22 @@ const Ue = /* @__PURE__ */ h({ nonMaxSuppressionPadded_: Ze });
  * limitations under the License.
  * =============================================================================
  */
-async function ns(r, t, n, e = 0.5, s = Number.NEGATIVE_INFINITY, a = !1) {
-  const o = l(r, "boxes", "nonMaxSuppressionAsync"), i = l(t, "scores", "nonMaxSuppressionAsync"), u = D(
+async function ss(s, e, n, r = 0.5, t = Number.NEGATIVE_INFINITY, a = !1) {
+  const o = u(s, "boxes", "nonMaxSuppressionAsync"), i = u(e, "scores", "nonMaxSuppressionAsync"), p = R(
     o,
     i,
     n,
-    e,
-    s,
+    r,
+    t,
     null
     /* softNmsSigma */
-  ), c = u.maxOutputSize, b = u.iouThreshold, m = u.scoreThreshold, [$, g] = await Promise.all([o.data(), i.data()]), { selectedIndices: x, validOutputs: y } = Fe($, g, c, b, m, a);
-  return o !== r && o.dispose(), i !== t && i.dispose(), {
-    selectedIndices: N(x, "int32"),
-    validOutputs: Fn(y, "int32")
+  ), c = p.maxOutputSize, b = p.iouThreshold, h = p.scoreThreshold, [g, d] = await Promise.all([o.data(), i.data()]), { selectedIndices: y, validOutputs: M } = se(g, d, c, b, h, a);
+  return o !== s && o.dispose(), i !== e && i.dispose(), {
+    selectedIndices: I(y, "int32"),
+    validOutputs: pn(M, "int32")
   };
 }
-const es = ns;
+const ts = ss;
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -969,15 +896,15 @@ const es = ns;
  * limitations under the License.
  * =============================================================================
  */
-function ss(r, t, n = !1, e = !1) {
-  const s = l(r, "images", "resizeBilinear");
-  p(s.rank === 3 || s.rank === 4, () => `Error in resizeBilinear: x must be rank 3 or 4, but got rank ${s.rank}.`), p(t.length === 2, () => `Error in resizeBilinear: new shape must 2D, but got shape ${t}.`), p(e === !1 || n === !1, () => "Error in resizeBilinear: If halfPixelCenters is true, alignCorners must be false.");
-  let a = s, o = !1;
-  s.rank === 3 && (o = !0, a = _(s, [1, s.shape[0], s.shape[1], s.shape[2]]));
-  const i = { images: a }, u = { alignCorners: n, halfPixelCenters: e, size: t }, c = f.runKernel(On, i, u);
-  return o ? _(c, [c.shape[1], c.shape[2], c.shape[3]]) : c;
+function rs(s, e, n = !1, r = !1) {
+  const t = u(s, "images", "resizeBilinear");
+  l(t.rank === 3 || t.rank === 4, () => `Error in resizeBilinear: x must be rank 3 or 4, but got rank ${t.rank}.`), l(e.length === 2, () => `Error in resizeBilinear: new shape must 2D, but got shape ${e}.`), l(r === !1 || n === !1, () => "Error in resizeBilinear: If halfPixelCenters is true, alignCorners must be false.");
+  let a = t, o = !1;
+  t.rank === 3 && (o = !0, a = k(t, [1, t.shape[0], t.shape[1], t.shape[2]]));
+  const i = { images: a }, p = { alignCorners: n, halfPixelCenters: r, size: e }, c = f.runKernel(Zn, i, p);
+  return o ? k(c, [c.shape[1], c.shape[2], c.shape[3]]) : c;
 }
-const ts = /* @__PURE__ */ h({ resizeBilinear_: ss });
+const as = /* @__PURE__ */ m({ resizeBilinear_: rs });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -994,15 +921,15 @@ const ts = /* @__PURE__ */ h({ resizeBilinear_: ss });
  * limitations under the License.
  * =============================================================================
  */
-function rs(r, t, n = !1, e = !1) {
-  const s = l(r, "images", "resizeNearestNeighbor");
-  p(s.rank === 3 || s.rank === 4, () => `Error in resizeNearestNeighbor: x must be rank 3 or 4, but got rank ${s.rank}.`), p(t.length === 2, () => `Error in resizeNearestNeighbor: new shape must 2D, but got shape ${t}.`), p(s.dtype === "float32" || s.dtype === "int32", () => "`images` must have `int32` or `float32` as dtype"), p(e === !1 || n === !1, () => "Error in resizeNearestNeighbor: If halfPixelCenters is true, alignCorners must be false.");
-  let a = s, o = !1;
-  s.rank === 3 && (o = !0, a = _(s, [1, s.shape[0], s.shape[1], s.shape[2]]));
-  const i = { images: a }, u = { alignCorners: n, halfPixelCenters: e, size: t }, c = f.runKernel(Wn, i, u);
-  return o ? _(c, [c.shape[1], c.shape[2], c.shape[3]]) : c;
+function os(s, e, n = !1, r = !1) {
+  const t = u(s, "images", "resizeNearestNeighbor");
+  l(t.rank === 3 || t.rank === 4, () => `Error in resizeNearestNeighbor: x must be rank 3 or 4, but got rank ${t.rank}.`), l(e.length === 2, () => `Error in resizeNearestNeighbor: new shape must 2D, but got shape ${e}.`), l(t.dtype === "float32" || t.dtype === "int32", () => "`images` must have `int32` or `float32` as dtype"), l(r === !1 || n === !1, () => "Error in resizeNearestNeighbor: If halfPixelCenters is true, alignCorners must be false.");
+  let a = t, o = !1;
+  t.rank === 3 && (o = !0, a = k(t, [1, t.shape[0], t.shape[1], t.shape[2]]));
+  const i = { images: a }, p = { alignCorners: n, halfPixelCenters: r, size: e }, c = f.runKernel(Jn, i, p);
+  return o ? k(c, [c.shape[1], c.shape[2], c.shape[3]]) : c;
 }
-const as = /* @__PURE__ */ h({ resizeNearestNeighbor_: rs });
+const is = /* @__PURE__ */ m({ resizeNearestNeighbor_: os });
 /**
  * @license
  * Copyright 2021 Google LLC. All Rights Reserved.
@@ -1019,38 +946,38 @@ const as = /* @__PURE__ */ h({ resizeNearestNeighbor_: rs });
  * limitations under the License.
  * =============================================================================
  */
-function os(r, t = "binary", n = !1, e = 0.5) {
-  const s = l(r, "image", "threshold"), a = 0.2989, o = 0.587, i = 0.114, u = s.shape[0] * s.shape[1];
-  let c = k(N([e]), 255), b, m, $, g;
-  if (p(s.rank === 3, () => `Error in threshold: image must be rank 3,but got rank ${s.rank}.`), p(s.shape[2] === 3 || s.shape[2] === 1, () => `Error in threshold: image color channel must be equal to 3 or 1but got ${s.shape[2]}.`), p(s.dtype === "int32" || s.dtype === "float32", () => `Error in dtype: image dtype must be int32 or float32,but got dtype ${s.dtype}.`), p(t === "otsu" || t === "binary", () => `Method must be binary or otsu, but was ${t}`), s.shape[2] === 3) {
-    [b, m, $] = un(s, [1, 1, 1], -1);
-    const d = k(b, a), E = k(m, o), I = k($, i);
-    g = Z(Z(d, E), I);
+function cs(s, e = "binary", n = !1, r = 0.5) {
+  const t = u(s, "image", "threshold"), a = 0.2989, o = 0.587, i = 0.114, p = t.shape[0] * t.shape[1];
+  let c = $(I([r]), 255), b, h, g, d;
+  if (l(t.rank === 3, () => `Error in threshold: image must be rank 3,but got rank ${t.rank}.`), l(t.shape[2] === 3 || t.shape[2] === 1, () => `Error in threshold: image color channel must be equal to 3 or 1but got ${t.shape[2]}.`), l(t.dtype === "int32" || t.dtype === "float32", () => `Error in dtype: image dtype must be int32 or float32,but got dtype ${t.dtype}.`), l(e === "otsu" || e === "binary", () => `Method must be binary or otsu, but was ${e}`), t.shape[2] === 3) {
+    [b, h, g] = mn(t, [1, 1, 1], -1);
+    const T = $(b, a), B = $(h, o), x = $(g, i);
+    d = U(U(T, B), x);
   } else
-    g = r;
-  if (t === "otsu") {
-    const d = Un(V(ke(g), "int32"), Ln([]), 256);
-    c = is(d, u);
+    d = s;
+  if (e === "otsu") {
+    const T = ce(K(ze(d), "int32"), te([]), 256);
+    c = us(T, p);
   }
-  const x = n ? pn(g, c) : U(g, c);
-  return V(k(x, 255), "int32");
+  const y = n ? bn(d, c) : nn(d, c);
+  return K($(y, 255), "int32");
 }
-function is(r, t) {
-  let n = N([-1]), e = N([0]), s = N([0]), a, o, i, u, c, b;
-  for (let m = 0; m < r.size - 1; m++) {
-    a = M(r, 0, m + 1), o = M(r, m + 1), c = R(q(a), t), b = R(q(o), t);
-    const $ = q(k(a, W(0, a.size)));
-    i = R($, q(a));
-    const g = Yn(o.shape, a.size), x = Z(W(0, o.size), g), y = k(o, x);
-    u = R(q(y), q(o));
-    const d = S(i, u), E = S(i, u), I = k(c, b);
-    s = k(k(I, d), E);
-    const A = U(s, e);
-    e = G(A, s, e), n = G(A, N([m]), n);
+function us(s, e) {
+  let n = I([-1]), r = I([0]), t = I([0]), a, o, i, p, c, b;
+  for (let h = 0; h < s.size - 1; h++) {
+    a = _(s, 0, h + 1), o = _(s, h + 1), c = q(E(a), e), b = q(E(o), e);
+    const g = E($(a, W(0, a.size)));
+    i = q(g, E(a));
+    const d = Qn(o.shape, a.size), y = U(W(0, o.size), d), M = $(o, y);
+    p = q(E(M), E(o));
+    const T = A(i, p), B = A(i, p), x = $(c, b);
+    t = $($(x, T), B);
+    const V = nn(t, r);
+    r = G(V, t, r), n = G(V, I([h]), n);
   }
   return n;
 }
-const cs = /* @__PURE__ */ h({ threshold_: os });
+const ls = /* @__PURE__ */ m({ threshold_: cs });
 /**
  * @license
  * Copyright 2021 Google LLC. All Rights Reserved.
@@ -1067,13 +994,13 @@ const cs = /* @__PURE__ */ h({ threshold_: os });
  * limitations under the License.
  * =============================================================================
  */
-function us(r, t, n = "nearest", e = "constant", s = 0, a) {
-  const o = l(r, "image", "transform", "float32"), i = l(t, "transforms", "transform", "float32");
-  p(o.rank === 4, () => `Error in transform: image must be rank 4,but got rank ${o.rank}.`), p(i.rank === 2 && (i.shape[0] === o.shape[0] || i.shape[0] === 1) && i.shape[1] === 8, () => "Error in transform: Input transform should be batch x 8 or 1 x 8"), p(a == null || a.length === 2, () => `Error in transform: outputShape must be [height, width] or null, but got ${a}.`);
-  const u = { image: o, transforms: i }, c = { interpolation: n, fillMode: e, fillValue: s, outputShape: a };
-  return f.runKernel(Cn, u, c);
+function ps(s, e, n = "nearest", r = "constant", t = 0, a) {
+  const o = u(s, "image", "transform", "float32"), i = u(e, "transforms", "transform", "float32");
+  l(o.rank === 4, () => `Error in transform: image must be rank 4,but got rank ${o.rank}.`), l(i.rank === 2 && (i.shape[0] === o.shape[0] || i.shape[0] === 1) && i.shape[1] === 8, () => "Error in transform: Input transform should be batch x 8 or 1 x 8"), l(a == null || a.length === 2, () => `Error in transform: outputShape must be [height, width] or null, but got ${a}.`);
+  const p = { image: o, transforms: i }, c = { interpolation: n, fillMode: r, fillValue: t, outputShape: a };
+  return f.runKernel(Hn, p, c);
 }
-const ls = /* @__PURE__ */ h({ transform_: us });
+const ms = /* @__PURE__ */ m({ transform_: ps });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -1090,16 +1017,16 @@ const ls = /* @__PURE__ */ h({ transform_: us });
  * limitations under the License.
  * =============================================================================
  */
-function ps(r, t, n) {
-  const e = l(r, "a", "bandPart");
-  p(e.rank >= 2, () => `bandPart(): Rank must be at least 2, got ${e.rank}.`);
-  const s = e.shape, [a, o] = e.shape.slice(-2);
-  let i, u;
-  typeof t == "number" ? (p(t % 1 === 0, () => `bandPart(): numLower must be an integer, got ${t}.`), p(t <= a, () => `bandPart(): numLower (${t}) must not be greater than the number of rows (${a}).`), i = l(t < 0 ? a : t, "numLower", "bandPart")) : (p(t.dtype === "int32", () => "bandPart(): numLower's dtype must be an int32."), i = G(tn(t, 0), a, rn(t, a))), typeof n == "number" ? (p(n % 1 === 0, () => `bandPart(): numUpper must be an integer, got ${n}.`), p(n <= o, () => `bandPart(): numUpper (${n}) must not be greater than the number of columns (${o}).`), u = l(n < 0 ? o : n, "numUpper", "bandPart")) : (p(n.dtype === "int32", () => "bandPart(): numUpper's dtype must be an int32."), u = G(tn(n, 0), o, rn(n, o)));
-  const c = _(W(0, a, 1, "int32"), [-1, 1]), b = W(0, o, 1, "int32"), m = S(c, b), $ = be(pn(m, i), ce(m, nn(u))), g = Qn([a, o], e.dtype);
-  return _(Y(mn(_(e, [-1, a, o])).map((x) => G($, x, g))), s);
+function fs(s, e, n) {
+  const r = u(s, "a", "bandPart");
+  l(r.rank >= 2, () => `bandPart(): Rank must be at least 2, got ${r.rank}.`);
+  const t = r.shape, [a, o] = r.shape.slice(-2);
+  let i, p;
+  typeof e == "number" ? (l(e % 1 === 0, () => `bandPart(): numLower must be an integer, got ${e}.`), l(e <= a, () => `bandPart(): numLower (${e}) must not be greater than the number of rows (${a}).`), i = u(e < 0 ? a : e, "numLower", "bandPart")) : (l(e.dtype === "int32", () => "bandPart(): numLower's dtype must be an int32."), i = G(on(e, 0), a, cn(e, a))), typeof n == "number" ? (l(n % 1 === 0, () => `bandPart(): numUpper must be an integer, got ${n}.`), l(n <= o, () => `bandPart(): numUpper (${n}) must not be greater than the number of columns (${o}).`), p = u(n < 0 ? o : n, "numUpper", "bandPart")) : (l(n.dtype === "int32", () => "bandPart(): numUpper's dtype must be an int32."), p = G(on(n, 0), o, cn(n, o)));
+  const c = k(W(0, a, 1, "int32"), [-1, 1]), b = W(0, o, 1, "int32"), h = A(c, b), g = Te(bn(h, i), ke(h, en(p))), d = ae([a, o], r.dtype);
+  return k(Y(dn(k(r, [-1, a, o])).map((y) => G(g, y, d))), t);
 }
-const ms = /* @__PURE__ */ h({ bandPart_: ps });
+const hs = /* @__PURE__ */ m({ bandPart_: fs });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -1116,30 +1043,30 @@ const ms = /* @__PURE__ */ h({ bandPart_: ps });
  * limitations under the License.
  * =============================================================================
  */
-function hs(r) {
-  let t;
-  if (Array.isArray(r)) {
-    t = !1, p(r != null && r.length > 0, () => "Gram-Schmidt process: input must not be null, undefined, or empty");
-    const s = r[0].shape[0];
-    for (let a = 1; a < r.length; ++a)
-      p(r[a].shape[0] === s, () => `Gram-Schmidt: Non-unique lengths found in the input vectors: (${r[a].shape[0]} vs. ${s})`);
+function bs(s) {
+  let e;
+  if (Array.isArray(s)) {
+    e = !1, l(s != null && s.length > 0, () => "Gram-Schmidt process: input must not be null, undefined, or empty");
+    const t = s[0].shape[0];
+    for (let a = 1; a < s.length; ++a)
+      l(s[a].shape[0] === t, () => `Gram-Schmidt: Non-unique lengths found in the input vectors: (${s[a].shape[0]} vs. ${t})`);
   } else
-    t = !0, r = un(r, r.shape[0], 0).map((s) => Ie(s, [0]));
-  p(r.length <= r[0].shape[0], () => `Gram-Schmidt: Number of vectors (${r.length}) exceeds number of dimensions (${r[0].shape[0]}).`);
-  const n = [], e = r;
-  for (let s = 0; s < r.length; ++s)
+    e = !0, s = mn(s, s.shape[0], 0).map((t) => Be(t, [0]));
+  l(s.length <= s[0].shape[0], () => `Gram-Schmidt: Number of vectors (${s.length}) exceeds number of dimensions (${s[0].shape[0]}).`);
+  const n = [], r = s;
+  for (let t = 0; t < s.length; ++t)
     n.push(f.tidy(() => {
-      let a = e[s];
-      if (s > 0)
-        for (let o = 0; o < s; ++o) {
-          const i = k(q(k(n[o], a)), n[o]);
-          a = S(a, i);
+      let a = r[t];
+      if (t > 0)
+        for (let o = 0; o < t; ++o) {
+          const i = $(E($(n[o], a)), n[o]);
+          a = A(a, i);
         }
-      return R(a, ln(a, "euclidean"));
+      return q(a, hn(a, "euclidean"));
     }));
-  return t ? Y(n, 0) : n;
+  return e ? Y(n, 0) : n;
 }
-const fs = /* @__PURE__ */ h({ gramSchmidt_: hs });
+const ds = /* @__PURE__ */ m({ gramSchmidt_: bs });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -1156,60 +1083,60 @@ const fs = /* @__PURE__ */ h({ gramSchmidt_: hs });
  * limitations under the License.
  * =============================================================================
  */
-function bs(r, t = !1) {
-  if (p(r.rank >= 2, () => `qr() requires input tensor to have a rank >= 2, but got rank ${r.rank}`), r.rank === 2)
-    return cn(r, t);
+function gs(s, e = !1) {
+  if (l(s.rank >= 2, () => `qr() requires input tensor to have a rank >= 2, but got rank ${s.rank}`), s.rank === 2)
+    return ln(s, e);
   {
-    const n = r.shape.slice(0, r.shape.length - 2).reduce((u, c) => u * c), e = mn(_(r, [
+    const n = s.shape.slice(0, s.shape.length - 2).reduce((p, c) => p * c), r = dn(k(s, [
       n,
-      r.shape[r.shape.length - 2],
-      r.shape[r.shape.length - 1]
-    ]), 0), s = [], a = [];
-    e.forEach((u) => {
-      const [c, b] = cn(u, t);
-      s.push(c), a.push(b);
+      s.shape[s.shape.length - 2],
+      s.shape[s.shape.length - 1]
+    ]), 0), t = [], a = [];
+    r.forEach((p) => {
+      const [c, b] = ln(p, e);
+      t.push(c), a.push(b);
     });
-    const o = _(Y(s, 0), r.shape), i = _(Y(a, 0), r.shape);
+    const o = k(Y(t, 0), s.shape), i = k(Y(a, 0), s.shape);
     return [o, i];
   }
 }
-function cn(r, t = !1) {
+function ln(s, e = !1) {
   return f.tidy(() => {
-    p(r.shape.length === 2, () => `qr2d() requires a 2D Tensor, but got a ${r.shape.length}D Tensor.`);
-    const n = r.shape[0], e = r.shape[1];
-    let s = ae(n), a = L(r);
+    l(s.shape.length === 2, () => `qr2d() requires a 2D Tensor, but got a ${s.shape.length}D Tensor.`);
+    const n = s.shape[0], r = s.shape[1];
+    let t = de(n), a = X(s);
     const o = H([[1]], [1, 1]);
-    let i = L(o);
-    const u = n >= e ? e : n;
-    for (let c = 0; c < u; ++c) {
-      const b = a, m = i, $ = s;
-      [i, a, s] = f.tidy(() => {
-        const g = M(a, [c, c], [n - c, 1]), x = ln(g), y = M(a, [c, c], [1, 1]), d = G(U(y, 0), H([[-1]]), H([[1]])), E = S(y, k(d, x)), I = R(g, E);
-        I.shape[0] === 1 ? i = L(o) : i = Q([
+    let i = X(o);
+    const p = n >= r ? r : n;
+    for (let c = 0; c < p; ++c) {
+      const b = a, h = i, g = t;
+      [i, a, t] = f.tidy(() => {
+        const d = _(a, [c, c], [n - c, 1]), y = hn(d), M = _(a, [c, c], [1, 1]), T = G(nn(M, 0), H([[-1]]), H([[1]])), B = A(M, $(T, y)), x = q(d, B);
+        x.shape[0] === 1 ? i = X(o) : i = Q([
           o,
-          M(I, [1, 0], [I.shape[0] - 1, I.shape[1]])
+          _(x, [1, 0], [x.shape[0] - 1, x.shape[1]])
         ], 0);
-        const A = nn(R(T(d, E), x)), z = M(a, [c, 0], [n - c, e]), B = k(A, i), P = an(i);
+        const V = en(q(N(T, B), y)), P = _(a, [c, 0], [n - c, r]), O = $(V, i), sn = un(i);
         if (c === 0)
-          a = S(z, T(B, T(P, z)));
+          a = A(P, N(O, N(sn, P)));
         else {
-          const C = S(z, T(B, T(P, z)));
-          a = Q([M(a, [0, 0], [c, e]), C], 0);
+          const C = A(P, N(O, N(sn, P)));
+          a = Q([_(a, [0, 0], [c, r]), C], 0);
         }
-        const sn = an(B), F = M(s, [0, c], [n, s.shape[1] - c]);
+        const tn = un(O), j = _(t, [0, c], [n, t.shape[1] - c]);
         if (c === 0)
-          s = S(F, T(T(F, i), sn));
+          t = A(j, N(N(j, i), tn));
         else {
-          const C = S(F, T(T(F, i), sn));
-          s = Q([M(s, [0, 0], [n, c]), C], 1);
+          const C = A(j, N(N(j, i), tn));
+          t = Q([_(t, [0, 0], [n, c]), C], 1);
         }
-        return [i, a, s];
-      }), Jn([b, m, $]);
+        return [i, a, t];
+      }), Un([b, h, g]);
     }
-    return !t && n > e && (s = M(s, [0, 0], [n, e]), a = M(a, [0, 0], [e, e])), [s, a];
+    return !e && n > r && (t = _(t, [0, 0], [n, r]), a = _(a, [0, 0], [r, r])), [t, a];
   });
 }
-const ds = /* @__PURE__ */ h({ qr_: bs });
+const $s = /* @__PURE__ */ m({ qr_: gs });
 /**
  * @license
  * Copyright 2020 Google LLC. All Rights Reserved.
@@ -1226,52 +1153,50 @@ const ds = /* @__PURE__ */ h({ qr_: bs });
  * limitations under the License.
  * =============================================================================
  */
-const ws = {
-  flipLeftRight: Te,
-  grayscaleToRGB: we,
-  resizeNearestNeighbor: as,
-  resizeBilinear: ts,
-  rgbToGrayscale: Be,
-  rotateWithOffset: Ke,
-  cropAndResize: Ne,
-  nonMaxSuppression: Ge,
-  nonMaxSuppressionAsync: Je,
-  nonMaxSuppressionWithScore: Xe,
-  nonMaxSuppressionWithScoreAsync: He,
-  nonMaxSuppressionPadded: Ue,
-  nonMaxSuppressionPaddedAsync: es,
-  threshold: cs,
-  transform: ls
-}, zs = {
-  bandPart: ms,
-  gramSchmidt: fs,
-  qr: ds
+const Ks = {
+  flipLeftRight: Ve,
+  grayscaleToRGB: je,
+  resizeNearestNeighbor: is,
+  resizeBilinear: as,
+  rgbToGrayscale: We,
+  rotateWithOffset: Oe,
+  cropAndResize: De,
+  nonMaxSuppression: Le,
+  nonMaxSuppressionAsync: Ze,
+  nonMaxSuppressionWithScore: Qe,
+  nonMaxSuppressionWithScoreAsync: Ue,
+  nonMaxSuppressionPadded: es,
+  nonMaxSuppressionPaddedAsync: ts,
+  threshold: ls,
+  transform: ms
+}, Rs = {
+  bandPart: hs,
+  gramSchmidt: ds,
+  qr: $s
 };
 export {
-  be as a,
-  pn as b,
-  an as c,
-  O as d,
-  U as e,
-  tn as f,
-  ce as g,
-  K as h,
-  ae as i,
-  Ie as j,
-  ws as k,
-  zs as l,
-  rn as m,
-  nn as n,
-  le as o,
-  ts as p,
+  Te as a,
+  bn as b,
+  F as c,
+  nn as d,
+  on as e,
+  z as f,
+  ke as g,
+  I as h,
+  de as i,
+  Ks as j,
+  an as k,
+  Rs as l,
+  cn as m,
+  en as n,
+  hn as o,
+  Ee as p,
   as as q,
-  $e as r,
-  M as s,
-  N as t,
-  mn as u,
-  ke as v,
+  Se as r,
+  Be as s,
+  un as t,
+  dn as u,
+  is as v,
   G as w,
-  je as x,
-  Fe as y,
-  Oe as z
+  ze as x
 };