@genai-fi/nanogpt 0.6.3 → 0.7.1

package/dist/non_max_suppression_impl-CsEgBuMA.js

@@ -0,0 +1,134 @@
+/**
+ * @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 J(e, t, o) {
+  const n = k(e, t, o), s = n < 0 ? -(n + 1) : n;
+  e.splice(s, 0, t);
+}
+function k(e, t, o) {
+  return A(e, t, o || v);
+}
+function v(e, t) {
+  return e > t ? 1 : e < t ? -1 : 0;
+}
+function A(e, t, o) {
+  let n = 0, s = e.length, r = 0, u = !1;
+  for (; n < s; ) {
+    r = n + (s - n >>> 1);
+    const l = o(t, e[r]);
+    l > 0 ? n = r + 1 : (s = r, u = !l);
+  }
+  return u ? n : -n - 1;
+}
+/**
+ * @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 O(e, t, o, n, s) {
+  return y(
+    e,
+    t,
+    o,
+    n,
+    s,
+    0
+    /* softNmsSigma */
+  );
+}
+function P(e, t, o, n, s, r) {
+  return y(
+    e,
+    t,
+    o,
+    n,
+    s,
+    0,
+    !1,
+    r,
+    !0
+    /* returnValidOutputs */
+  );
+}
+function X(e, t, o, n, s, r) {
+  return y(
+    e,
+    t,
+    o,
+    n,
+    s,
+    r,
+    !0
+    /* returnScoresTensor */
+  );
+}
+function y(e, t, o, n, s, r, u = !1, l = !1, h = !1) {
+  const c = [];
+  for (let i = 0; i < t.length; i++)
+    t[i] > s && c.push({ score: t[i], boxIndex: i, suppressBeginIndex: 0 });
+  c.sort(C);
+  const d = r > 0 ? -0.5 / r : 0, a = [], f = [];
+  for (; a.length < o && c.length > 0; ) {
+    const i = c.pop(), { score: M, boxIndex: I, suppressBeginIndex: g } = i;
+    if (M < s)
+      break;
+    let S = !1;
+    for (let b = a.length - 1; b >= g; --b) {
+      const w = B(e, I, a[b]);
+      if (w >= n) {
+        S = !0;
+        break;
+      }
+      if (i.score = i.score * V(n, d, w), i.score <= s)
+        break;
+    }
+    i.suppressBeginIndex = a.length, S || (i.score === M ? (a.push(I), f.push(i.score)) : i.score > s && J(c, i, C));
+  }
+  const x = a.length, p = o - x;
+  l && p > 0 && (a.push(...new Array(p).fill(0)), f.push(...new Array(p).fill(0)));
+  const m = { selectedIndices: a };
+  return u && (m.selectedScores = f), h && (m.validOutputs = x), m;
+}
+function B(e, t, o) {
+  const n = e.subarray(t * 4, t * 4 + 4), s = e.subarray(o * 4, o * 4 + 4), r = Math.min(n[0], n[2]), u = Math.min(n[1], n[3]), l = Math.max(n[0], n[2]), h = Math.max(n[1], n[3]), c = Math.min(s[0], s[2]), d = Math.min(s[1], s[3]), a = Math.max(s[0], s[2]), f = Math.max(s[1], s[3]), x = (l - r) * (h - u), p = (a - c) * (f - d);
+  if (x <= 0 || p <= 0)
+    return 0;
+  const m = Math.max(r, c), i = Math.max(u, d), M = Math.min(l, a), I = Math.min(h, f), g = Math.max(M - m, 0) * Math.max(I - i, 0);
+  return g / (x + p - g);
+}
+function V(e, t, o) {
+  const n = Math.exp(t * o * o);
+  return o <= e ? n : 0;
+}
+function C(e, t) {
+  return e.score - t.score || e.score === t.score && t.boxIndex - e.boxIndex;
+}
+export {
+  X as a,
+  P as b,
+  O as n
+};

package/dist/{ones-BBlSRqn1.js → ones-C8Mfln6-.js}

@@ -1,5 +1,5 @@
-import { x as n, y as t, g as m, E as i } from "./index-BAzbokzv.js";
-import { z as c, c as f } from "./zeros-YJDE7oRb.js";
+import { F as n, H as t, j as m, E as i } from "./index-UdZhlibC.js";
+import { z as c, c as f } from "./zeros-DnPT2nD4.js";
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.

package/dist/ops/adamAdjust.d.ts

@@ -0,0 +1,2 @@
+import { Tensor } from '@tensorflow/tfjs-core';
+export declare function adamAdjust(moments: Tensor, value: Tensor, beta1: number, beta2: number, epsilon: number, learningRate: number): Tensor;

package/dist/ops/adamAdjust.js

@@ -0,0 +1,9 @@
+import { e as i } from "../index-UdZhlibC.js";
+import "./cpu/adamAdjust.js";
+import "./webgl/adamAdjust.js";
+function p(r, t, e, n, m, o) {
+  return i().runKernel("AdamAdjust", { moments: r, value: t }, { beta1: e, beta2: n, epsilon: m, learningRate: o });
+}
+export {
+  p as adamAdjust
+};

package/dist/ops/adamMoments.d.ts

@@ -0,0 +1,2 @@
+import { Tensor } from '@tensorflow/tfjs-core';
+export declare function adamMoments(moments: Tensor, gradient: Tensor, beta1: number, beta2: number): Tensor;

package/dist/ops/adamMoments.js

@@ -0,0 +1,9 @@
+import { e as o } from "../index-UdZhlibC.js";
+import "./cpu/adamMoments.js";
+import "./webgl/adamMoments.js";
+function p(e, n, r, m) {
+  return o().runKernel("AdamMoments", { moments: e, gradient: n }, { beta1: r, beta2: m });
+}
+export {
+  p as adamMoments
+};

package/dist/ops/appendCache.js

@@ -1,8 +1,8 @@
-import { e as a } from "../index-BAzbokzv.js";
+import { e as a } from "../index-UdZhlibC.js";
 import "./cpu/appendCache.js";
 import "./webgl/appendCache.js";
-import { c as s } from "../concat-5aPGqw3Z.js";
-import { z as c } from "../zeros-YJDE7oRb.js";
+import { c as s } from "../concat-CbXTetof.js";
+import { z as c } from "../zeros-DnPT2nD4.js";
 function i(r, p, n, o) {
   if (!o) {
     const e = r.shape[2];

package/dist/ops/attentionMask.js

@@ -1,4 +1,4 @@
-import { e as o } from "../index-BAzbokzv.js";
+import { e as o } from "../index-UdZhlibC.js";
 import "./cpu/attentionMask.js";
 import "./webgl/attentionMask.js";
 import "./grads/attentionMask.js";

package/dist/ops/cpu/adamAdjust.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/ops/cpu/adamAdjust.js

@@ -0,0 +1,18 @@
+import { f as k, n as t, o as i, a as q, q as z } from "../../index-UdZhlibC.js";
+function A(a) {
+  const { moments: n, value: r } = a.inputs, { beta1: l, beta2: u, epsilon: m, learningRate: d } = a.attrs, e = n.shape.length, c = new Array(e).fill(0), s = n.shape.slice();
+  s[e - 1] = 1;
+  const o = c.slice();
+  o[e - 1] = 1;
+  const b = s.slice(), p = n.slice(c, s).squeeze([e - 1]), M = n.slice(o, b).squeeze([e - 1]), f = t(p, l), g = t(M, u);
+  return i(
+    q(t(f, i(z(g), m ?? 1e-8)), -d),
+    r
+  );
+}
+const C = {
+  kernelName: "AdamAdjust",
+  backendName: "cpu",
+  kernelFunc: A
+};
+k(C);

package/dist/ops/cpu/adamMoments.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/ops/cpu/adamMoments.js

@@ -0,0 +1,16 @@
+import { f as p } from "../../index-UdZhlibC.js";
+import { s as f } from "../../stack-ikk2Y8_P.js";
+function b(t) {
+  const { moments: n, gradient: c } = t.inputs, { beta1: o, beta2: m } = t.attrs, e = n.shape.length, a = new Array(e).fill(0), s = n.shape.slice();
+  s[e - 1] = 1;
+  const i = a.slice();
+  i[e - 1] = 1;
+  const r = s.slice(), l = n.slice(a, s).squeeze([e - 1]), u = n.slice(i, r).squeeze([e - 1]), M = l.mul(o).add(c.mul(1 - o)), d = u.mul(m).add(c.square().mul(1 - m));
+  return f([M, d], -1);
+}
+const g = {
+  kernelName: "AdamMoments",
+  backendName: "cpu",
+  kernelFunc: b
+};
+p(g);

package/dist/ops/cpu/appendCache.js

@@ -1,13 +1,13 @@
-import { r as d } from "../../index-BAzbokzv.js";
-import { c as h } from "../../concat-5aPGqw3Z.js";
+import { f as d } from "../../index-UdZhlibC.js";
+import { c as h } from "../../concat-CbXTetof.js";
 function u(p) {
-  const { cache: n, item: s } = p.inputs, { maxSize: r, pastLen: c } = p.attrs, t = n.shape[0], o = n.shape[1], a = n.shape[3], e = s.shape[2];
-  if (c + e <= r) {
-    const f = n.slice([0, 0, 0, 0], [t, o, c, a]), m = n.slice([0, 0, c + e, 0], [t, o, r - c - e, a]), i = e < e ? s.slice([0, 0, 0, 0], [t, o, e, a]) : s, k = h([f, i, m], 2);
-    return f.dispose(), m.dispose(), i !== s && i.dispose(), k;
+  const { cache: n, item: s } = p.inputs, { maxSize: i, pastLen: c } = p.attrs, t = n.shape[0], o = n.shape[1], a = n.shape[3], e = s.shape[2];
+  if (c + e <= i) {
+    const l = n.slice([0, 0, 0, 0], [t, o, c, a]), m = n.slice([0, 0, c + e, 0], [t, o, i - c - e, a]), r = e < e ? s.slice([0, 0, 0, 0], [t, o, e, a]) : s, k = h([l, r, m], 2);
+    return l.dispose(), m.dispose(), r !== s && r.dispose(), k;
   }
-  const l = n.slice([0, 0, e, 0], [t, o, r - e, a]), C = h([l, s], 2);
-  return l.dispose(), C;
+  const f = n.slice([0, 0, e, 0], [t, o, i - e, a]), C = h([f, s], 2);
+  return f.dispose(), C;
 }
 const w = {
   kernelName: "AppendCache",

package/dist/ops/cpu/attentionMask.js

@@ -1,21 +1,21 @@
-import { r as a, f as p, s as u } from "../../index-BAzbokzv.js";
-import { l as N, w as b } from "../../ops-C0sQEcPw.js";
-import { o as A } from "../../ones-BBlSRqn1.js";
-import { z as I } from "../../zeros-YJDE7oRb.js";
-import { m as g } from "../../mat_mul-7121rsJk.js";
+import { f as a, h as p, b as u } from "../../index-UdZhlibC.js";
+import { l as N, w as b } from "../../ops-aRTXR2Sr.js";
+import { o as A } from "../../ones-C8Mfln6-.js";
+import { z as I } from "../../zeros-DnPT2nD4.js";
+import { m as g } from "../../mat_mul-dwmZz69e.js";
 function o(n) {
   const { q: s, k: e } = n.inputs, { divisor: r } = n.attrs, c = s.shape[2], t = e.shape[2], m = N.bandPart(A([t, t]), -1, 0).cast("bool"), l = I([t, t]), i = p([t, t], Number.NEGATIVE_INFINITY), f = b(m, l, i), k = g(s, e, !1, !0).mul(u(r)), d = f.slice([0, 0], [c, t]).expandDims(0).expandDims(0);
   return k.add(d);
 }
-const w = {
+const h = {
   kernelName: "AttentionMask",
   backendName: "cpu",
   kernelFunc: o
 };
-a(w);
-const M = {
+a(h);
+const w = {
   kernelName: "AttentionMask",
   backendName: "tensorflow",
   kernelFunc: o
 };
-a(M);
+a(w);

package/dist/ops/cpu/fusedSoftmax.js

@@ -1,23 +1,29 @@
-import { r as n } from "../../index-BAzbokzv.js";
-import { s as f } from "../../softmax-fsdtf6JC.js";
-function r(t) {
-  const { inputs: s, attrs: i } = t, { logits: o } = s, { dim: a, dropoutRate: e } = i;
+import { f as e } from "../../index-UdZhlibC.js";
+import { s as f } from "../../softmax-JMEIUo2J.js";
+function n(t) {
+  const { inputs: s, attrs: a } = t, { logits: o } = s, { dim: i, dropoutRate: r } = a;
   if (!o)
     throw new Error("Error in softmax: input logits is null");
-  return e !== void 0 && e > 0 && console.warn("Dropout in fusedSoftmax not implemented for CPU backend, skipping dropout."), f(o, a);
+  return r !== void 0 && r > 0 && console.warn("Dropout in fusedSoftmax not implemented for CPU backend, skipping dropout."), f(o, i);
 }
 const m = {
   kernelName: "FusedSoftmax",
   backendName: "cpu",
-  kernelFunc: r
+  kernelFunc: n
 };
-n(m);
-const l = {
+e(m);
+const u = {
   kernelName: "FusedSoftmax",
   backendName: "tensorflow",
-  kernelFunc: r
+  kernelFunc: n
+};
+e(u);
+const l = {
+  kernelName: "FusedSoftmax",
+  backendName: "webgpu",
+  kernelFunc: n
 };
-n(l);
+e(l);
 export {
-  r as softmaxCPU
+  n as softmaxCPU
 };

package/dist/ops/cpu/gatherSub.js

@@ -1,6 +1,6 @@
-import { o as u, q as c, E as g, Y as h, r as m, b as p } from "../../index-BAzbokzv.js";
-import { r as l } from "../../range-DYsrnfiy.js";
-import { s as N } from "../../stack-dfEEz2OY.js";
+import { B as u, C as c, E as g, a4 as h, f as m, c as p } from "../../index-UdZhlibC.js";
+import { r as f } from "../../range-D9CZhVlR.js";
+import { s as l } from "../../stack-ikk2Y8_P.js";
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -17,14 +17,14 @@ import { s as N } from "../../stack-dfEEz2OY.js";
  * limitations under the License.
  * =============================================================================
  */
-function b(e, t) {
+function N(e, t) {
   const n = c(t, "indices", "gatherND", "int32"), s = { params: c(e, "x", "gatherND", "string_or_numeric"), indices: n };
   return g.runKernel(h, s);
 }
-const f = /* @__PURE__ */ u({ gatherND_: b });
+const b = /* @__PURE__ */ u({ gatherND_: N });
 function d(e) {
-  const { values: t, labels: n, logits: r } = e.inputs, s = n.shape[0], o = l(0, s, 1, "int32"), a = N([o, n], 1), i = f(r, a);
-  return p(t, i);
+  const { values: t, labels: n, logits: r } = e.inputs, s = n.shape[0], a = f(0, s, 1, "int32"), i = l([a, n], 1), o = b(r, i);
+  return p(t, o);
 }
 const k = {
   kernelName: "EfficientGatherSub",

package/dist/ops/cpu/gelu.js

@@ -1,29 +1,29 @@
-import { r as t, t as d } from "../../index-BAzbokzv.js";
+import { f as t, t as d } from "../../index-UdZhlibC.js";
 const o = 0.7978845608028654, c = 0.044715;
-function m(u) {
-  const { inputs: l } = u, { x: e } = l, n = e;
+function m(r) {
+  const { inputs: u } = r, { x: n } = u, e = n;
   return d(() => {
-    const r = n.mul(n).mul(n), s = n.add(r.mul(c)).mul(o).tanh().add(1).mul(0.5);
-    return n.mul(s);
+    const l = e.pow(3), s = e.add(l.mul(c)).mul(o).tanh().add(1).mul(0.5);
+    return e.mul(s);
   });
 }
-const K = {
+const p = {
   kernelName: "Gelu",
   backendName: "cpu",
   kernelFunc: m
 };
-t(K);
-const p = {
+t(p);
+const K = {
   kernelName: "Gelu",
   backendName: "tensorflow",
   kernelFunc: m
 };
-t(p);
-function i(u) {
-  const { dy: l, x: e } = u.inputs;
+t(K);
+function i(r) {
+  const { dy: u, x: n } = r.inputs;
   return d(() => {
-    const n = e.square(), r = n.mul(e), a = e.add(r.mul(c)).mul(o).tanh(), f = a.square().neg().add(1), k = n.mul(3 * c).add(1), g = a.add(1).mul(0.5), G = e.mul(f).mul(o).mul(k).mul(0.5), N = g.add(G);
-    return l.mul(N);
+    const e = n.square(), l = e.mul(n), a = n.add(l.mul(c)).mul(o).tanh(), f = a.square().neg().add(1), k = e.mul(3 * c).add(1), g = a.add(1).mul(0.5), G = n.mul(f).mul(o).mul(k).mul(0.5), N = g.add(G);
+    return u.mul(N);
   });
 }
 const x = {

package/dist/ops/cpu/matMulGelu.js

@@ -1,40 +1,52 @@
-import { r as a, t as i } from "../../index-BAzbokzv.js";
-const c = 0.7978845608028654, m = 0.044715;
-function M(o) {
-  const { inputs: s } = o, { x: t, kernel: l } = s, e = t, u = l;
-  return i(() => {
-    const n = e.matMul(u), d = n.mul(n).mul(n), r = n.add(d.mul(m)).mul(c).tanh().add(1).mul(0.5);
-    return n.mul(r);
+import { f as e, t as m } from "../../index-UdZhlibC.js";
+import { g as d, d as M } from "../../gelu-BFwVnd1r.js";
+function c(t) {
+  const { inputs: u } = t, { x: n, kernel: l } = u, a = n, r = l;
+  return m(() => {
+    const o = a.matMul(r);
+    return d(o);
   });
 }
-const b = {
+const G = {
   kernelName: "MatMulGelu",
   backendName: "cpu",
-  kernelFunc: M
+  kernelFunc: c
 };
-a(b);
-const w = {
+e(G);
+const i = {
   kernelName: "MatMulGelu",
   backendName: "tensorflow",
-  kernelFunc: M
+  kernelFunc: c
 };
-a(w);
-function f(o) {
-  const { dy: s, x: t, kernel: l } = o.inputs;
-  return i(() => {
-    const e = t.matMul(l), u = e.square(), n = u.mul(e), r = e.add(n.mul(m)).mul(c).tanh(), G = r.square().neg().add(1), g = u.mul(3 * m).add(1), p = r.add(1).mul(0.5), N = e.mul(G).mul(c).mul(g).mul(0.5), K = p.add(N), k = s.mul(K), h = k.matMul(l.transpose()), F = t.transpose().matMul(k);
-    return [h, F];
+e(i);
+const f = {
+  kernelName: "MatMulGelu",
+  backendName: "webgpu",
+  kernelFunc: c
+};
+e(f);
+function s(t) {
+  const { dy: u, x: n, kernel: l } = t.inputs;
+  return m(() => {
+    const a = n.matMul(l), r = M(u, a), o = r.matMul(l.transpose()), k = n.transpose().matMul(r);
+    return [o, k];
   });
 }
-const x = {
+const g = {
   kernelName: "MatMulGeluGrad",
   backendName: "cpu",
-  kernelFunc: f
+  kernelFunc: s
 };
-a(x);
-const C = {
+e(g);
+const p = {
   kernelName: "MatMulGeluGrad",
   backendName: "tensorflow",
-  kernelFunc: f
+  kernelFunc: s
+};
+e(p);
+const N = {
+  kernelName: "MatMulGeluGrad",
+  backendName: "webgpu",
+  kernelFunc: s
 };
-a(C);
+e(N);

package/dist/ops/cpu/matMulMul.js

@@ -1,17 +1,23 @@
-import { r as n, t as M } from "../../index-BAzbokzv.js";
-function e(t) {
-  const { inputs: r, attrs: o } = t, { transposeA: s, transposeB: l } = o, { x: c, kernel: u, y: a } = r, m = c, i = u, k = a;
-  return M(() => m.matMul(i, s, l).mul(k));
+import { f as e, t as i } from "../../index-UdZhlibC.js";
+function n(t) {
+  const { inputs: r, attrs: l } = t, { transposeA: o, transposeB: s } = l, { x: u, kernel: a, y: c } = r, m = u, k = a, M = c;
+  return i(() => m.matMul(k, o, s).mul(M));
 }
 const f = {
   kernelName: "MatMulMul",
   backendName: "cpu",
-  kernelFunc: e
+  kernelFunc: n
 };
-n(f);
+e(f);
 const p = {
   kernelName: "MatMulMul",
   backendName: "tensorflow",
-  kernelFunc: e
+  kernelFunc: n
 };
-n(p);
+e(p);
+const g = {
+  kernelName: "MatMulMul",
+  backendName: "webgpu",
+  kernelFunc: n
+};
+e(g);

package/dist/ops/cpu/mulDropout.js

@@ -1,7 +1,7 @@
-import { r as e, a as u } from "../../index-BAzbokzv.js";
+import { f as e, a as t } from "../../index-UdZhlibC.js";
 function n(o) {
-  const { inputs: r } = o, { a: l, b: t } = r;
-  return console.warn("Using fallback mulDrop implementation without dropout."), u(l, t);
+  const { inputs: r } = o, { a: l, b: u } = r;
+  return console.warn("Using fallback mulDrop implementation without dropout."), t(l, u);
 }
 const a = {
   kernelName: "MulDropout",
@@ -15,3 +15,9 @@ const c = {
   kernelFunc: n
 };
 e(c);
+const m = {
+  kernelName: "MulDropout",
+  backendName: "webgpu",
+  kernelFunc: n
+};
+e(m);

package/dist/ops/cpu/normRMS.js

@@ -1,4 +1,4 @@
-import { r as o, t as d } from "../../index-BAzbokzv.js";
+import { f as o, t as d } from "../../index-UdZhlibC.js";
 function i(t) {
   const { inputs: e } = t, { x: n, gamma: s } = e, r = n, a = s;
   return d(() => {
@@ -6,12 +6,12 @@ function i(t) {
     return r.mul(u).mul(a);
   });
 }
-const f = {
+const k = {
   kernelName: "RMSNorm",
   backendName: "cpu",
   kernelFunc: i
 };
-o(f);
+o(k);
 const g = {
   kernelName: "RMSNorm",
   backendName: "tensorflow",
@@ -21,8 +21,8 @@ o(g);
 function N(t) {
   const { dy: e, x: n, gamma: s } = t.inputs;
   return d(() => {
-    const r = n.shape[n.shape.length - 1], a = n.square().mean(-1, !0), m = a.add(1e-8).rsqrt(), u = n.mul(m), l = e.mul(u).sum([0, 1]), c = e.mul(s), k = c.mul(n).sum(-1, !0).div(r);
-    return [c.mul(m).sub(n.mul(k).mul(m).div(a.add(1e-8))), l];
+    const r = n.shape[n.shape.length - 1], a = n.square().mean(-1, !0), m = a.add(1e-8).rsqrt(), u = n.mul(m), l = e.mul(u).sum([0, 1]), c = e.mul(s), f = c.mul(n).sum(-1, !0).div(r);
+    return [c.mul(m).sub(n.mul(f).mul(m).div(a.add(1e-8))), l];
   });
 }
 const S = {

package/dist/ops/cpu/qkv.js

@@ -1,6 +1,6 @@
-import { r as q } from "../../index-BAzbokzv.js";
-import { r as o } from "../../reshape-krWGKraP.js";
-import { s as x } from "../../split-eiktj-6L.js";
+import { f as q } from "../../index-UdZhlibC.js";
+import { r as o } from "../../reshape-BR0eoLYN.js";
+import { s as x } from "../../split-CRU0PjVV.js";
 function v(p) {
   const { x: c, kernel: K } = p.inputs, { heads: n } = p.attrs, [s, e, t] = c.shape, a = o(c, [s * e, t]), i = a.dot(K);
   a.dispose();

package/dist/ops/cpu/rope.js

@@ -1,8 +1,8 @@
-import { r as S } from "../../index-BAzbokzv.js";
-import { r as F } from "../../range-DYsrnfiy.js";
-import { g as I } from "../../gather-DjyCjmOD.js";
-import { s as E } from "../../stack-dfEEz2OY.js";
-import { c as T } from "../../concat-5aPGqw3Z.js";
+import { f as S } from "../../index-UdZhlibC.js";
+import { r as F } from "../../range-D9CZhVlR.js";
+import { g as I } from "../../gather-Bbh8DHhM.js";
+import { s as E } from "../../stack-ikk2Y8_P.js";
+import { c as T } from "../../concat-CbXTetof.js";
 function U(t, c, p, o, r) {
   const n = o.shape[3], s = p;
   if (s > n) return o;

package/dist/ops/cpu/scatterSub.js

@@ -1,8 +1,8 @@
-import { o as f, x as g, q as r, E as l, X as N, r as b, b as S, a as h } from "../../index-BAzbokzv.js";
-import { v as D } from "../../scatter_nd_util-93ln7Hut.js";
-import { r as k } from "../../range-DYsrnfiy.js";
-import { s as v } from "../../stack-dfEEz2OY.js";
-import { o as E } from "../../ones-BBlSRqn1.js";
+import { B as f, F as g, C as r, E as l, a3 as N, f as b, c as S, a as h } from "../../index-UdZhlibC.js";
+import { v as D } from "../../scatter_nd_util-OjyAxku2.js";
+import { r as k } from "../../range-D9CZhVlR.js";
+import { s as v } from "../../stack-ikk2Y8_P.js";
+import { o as E } from "../../ones-C8Mfln6-.js";
 /**
  * @license
  * Copyright 2018 Google LLC. All Rights Reserved.
@@ -26,14 +26,14 @@ function I(a, e, s) {
   const c = { indices: n, updates: t }, o = { shape: s };
   return l.runKernel(N, c, o);
 }
-const K = /* @__PURE__ */ f({ scatterND_: I });
-function L(a) {
-  const { logits: e, labels: s, dy: n } = a.inputs, t = s.shape[0], c = e.shape[1], o = k(0, t, 1, "int32"), i = v([o, s], 1), d = E([t]), u = K(i, d, [t, c]), p = S(e, u), m = n.reshape([t, 1]);
+const C = /* @__PURE__ */ f({ scatterND_: I });
+function K(a) {
+  const { logits: e, labels: s, dy: n } = a.inputs, t = s.shape[0], c = e.shape[1], o = k(0, t, 1, "int32"), i = v([o, s], 1), d = E([t]), u = C(i, d, [t, c]), p = S(e, u), m = n.reshape([t, 1]);
   return h(p, m);
 }
-const T = {
+const L = {
   kernelName: "EfficientScatterSub",
   backendName: "cpu",
-  kernelFunc: L
+  kernelFunc: K
 };
-b(T);
+b(L);

package/dist/ops/fusedSoftmax.js

@@ -1,4 +1,4 @@
-import { e as t } from "../index-BAzbokzv.js";
+import { e as t } from "../index-UdZhlibC.js";
 import "./cpu/fusedSoftmax.js";
 import "./webgl/fusedSoftmax.js";
 import "./grads/fusedSoftmax.js";

package/dist/ops/gatherSub.js

@@ -1,4 +1,4 @@
-import { e as n } from "../index-BAzbokzv.js";
+import { e as n } from "../index-UdZhlibC.js";
 import "./cpu/gatherSub.js";
 import "./webgl/gatherSub.js";
 function f(r, e, t) {