@genai-fi/nanogpt 0.7.3 → 0.8.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/Generator.d.ts +25 -2
- package/dist/Generator.js +150 -49
- package/dist/{RealDiv-Dy0p8Bvo.js → RealDiv-N8TpOMYv.js} +14 -14
- package/dist/{Reshape-DvudQDvJ.js → Reshape-B-lWQRnF.js} +1 -1
- package/dist/{Reshape-DH5srBP0.js → Reshape-Bo8HzP8V.js} +5 -5
- package/dist/TeachableLLM.d.ts +6 -6
- package/dist/TeachableLLM.js +31 -31
- package/dist/Trainer.d.ts +13 -2
- package/dist/Trainer.js +21 -12
- package/dist/{axis_util-BzbKo31C.js → axis_util-DubwyOhW.js} +3 -3
- package/dist/backend.js +2 -2
- package/dist/{backend_util-TE7aTPhZ.js → backend_util-BJ-_jSeK.js} +46 -46
- package/dist/{broadcast_to-CdbwV-Dj.js → broadcast_to-BYfCp5iL.js} +2 -2
- package/dist/{concat-CsxrgovM.js → concat-BmDqqFsa.js} +1 -1
- package/dist/{dataset-CtdBYwjo.js → dataset-CJmEGu6D.js} +5 -5
- package/dist/{dropout-DYs5QFGQ.js → dropout-sx0sjVAT.js} +8 -8
- package/dist/exports_initializers-DAKM8UO9.js +16 -0
- package/dist/{gather-CMMy2KEG.js → gather-C1siEkdp.js} +1 -1
- package/dist/{gelu-C-dPj6Ku.js → gelu-Bd3UBBxg.js} +1 -1
- package/dist/{gpgpu_math-DGNLNL4I.js → gpgpu_math-TFLxaLkw.js} +26 -26
- package/dist/{index-CLthM0TO.js → index-BaPo_0H8.js} +185 -185
- package/dist/{index-BoWRt-10.js → index-CUQrfsw_.js} +266 -265
- package/dist/{kernel_funcs_utils-BYKWV8Aa.js → kernel_funcs_utils-P9aFa232.js} +9 -9
- package/dist/layers/BaseLayer.d.ts +8 -13
- package/dist/layers/BaseLayer.js +25 -13
- package/dist/layers/CausalSelfAttention.d.ts +3 -2
- package/dist/layers/CausalSelfAttention.js +28 -28
- package/dist/layers/MLP.d.ts +3 -2
- package/dist/layers/MLP.js +16 -20
- package/dist/layers/PositionEmbedding.d.ts +9 -0
- package/dist/layers/PositionEmbedding.js +45 -0
- package/dist/layers/RMSNorm.d.ts +3 -2
- package/dist/layers/RMSNorm.js +6 -6
- package/dist/layers/RoPECache.d.ts +1 -1
- package/dist/layers/RoPECache.js +4 -4
- package/dist/layers/TiedEmbedding.d.ts +3 -2
- package/dist/layers/TiedEmbedding.js +29 -7
- package/dist/layers/TransformerBlock.d.ts +3 -2
- package/dist/layers/TransformerBlock.js +1 -1
- package/dist/loader/load.d.ts +2 -2
- package/dist/loader/loadHF.d.ts +2 -2
- package/dist/loader/loadTransformers.d.ts +4 -2
- package/dist/loader/loadTransformers.js +10 -9
- package/dist/loader/newZipLoad.d.ts +2 -2
- package/dist/loader/oldZipLoad.d.ts +2 -2
- package/dist/loader/oldZipLoad.js +42 -51
- package/dist/loader/save.d.ts +8 -0
- package/dist/loader/save.js +62 -0
- package/dist/{log_sum_exp-DbjkV734.js → log_sum_exp-C142qZqY.js} +14 -14
- package/dist/main.d.ts +5 -4
- package/dist/main.js +22 -18
- package/dist/{mat_mul-8m8pfdcx.js → mat_mul-DMkduNJu.js} +1 -1
- package/dist/{max-Ddnnb5xe.js → max-B3JOcNGb.js} +1 -1
- package/dist/mod-uUuj4gSb.js +27 -0
- package/dist/models/NanoGPTV1.d.ts +15 -0
- package/dist/models/NanoGPTV1.js +71 -0
- package/dist/{config.d.ts → models/config.d.ts} +1 -0
- package/dist/{config.js → models/config.js} +1 -0
- package/dist/models/factory.d.ts +3 -0
- package/dist/models/factory.js +14 -0
- package/dist/models/model.d.ts +26 -0
- package/dist/models/model.js +68 -0
- package/dist/{mulmat_packed_gpu-VSekgsNv.js → mulmat_packed_gpu-Cm2gw-c8.js} +1 -1
- package/dist/{ones-Dj0SDhHf.js → ones-ZdgQGBCP.js} +2 -2
- package/dist/ops/adamAdjust.js +1 -1
- package/dist/ops/adamMoments.js +1 -1
- package/dist/ops/appendCache.js +3 -3
- package/dist/ops/attentionMask.js +1 -1
- package/dist/ops/cpu/adamAdjust.js +9 -9
- package/dist/ops/cpu/adamMoments.js +2 -2
- package/dist/ops/cpu/appendCache.js +2 -2
- package/dist/ops/cpu/attentionMask.js +5 -5
- package/dist/ops/cpu/fusedSoftmax.js +2 -2
- package/dist/ops/cpu/gatherSub.js +3 -3
- package/dist/ops/cpu/gelu.js +1 -1
- package/dist/ops/cpu/matMulGelu.js +2 -2
- package/dist/ops/cpu/matMulMul.js +1 -1
- package/dist/ops/cpu/mulDropout.js +1 -1
- package/dist/ops/cpu/normRMS.js +1 -1
- package/dist/ops/cpu/qkv.js +3 -3
- package/dist/ops/cpu/rope.js +5 -5
- package/dist/ops/cpu/scatterSub.js +11 -11
- package/dist/ops/fusedSoftmax.js +1 -1
- package/dist/ops/gatherSub.js +1 -1
- package/dist/ops/gelu.js +2 -2
- package/dist/ops/grads/attentionMask.js +1 -1
- package/dist/ops/grads/fusedSoftmax.js +2 -2
- package/dist/ops/grads/gelu.js +2 -2
- package/dist/ops/grads/matMulGelu.js +1 -1
- package/dist/ops/grads/normRMS.js +1 -1
- package/dist/ops/grads/qkv.js +1 -1
- package/dist/ops/grads/rope.js +1 -1
- package/dist/ops/matMulGelu.js +1 -1
- package/dist/ops/matMulMul.js +1 -1
- package/dist/ops/mulDrop.js +1 -1
- package/dist/ops/normRMS.js +1 -1
- package/dist/ops/qkv.js +1 -1
- package/dist/ops/rope.js +4 -4
- package/dist/ops/scatterSub.js +1 -1
- package/dist/ops/webgl/adamAdjust.js +2 -2
- package/dist/ops/webgl/adamMoments.js +1 -1
- package/dist/ops/webgl/appendCache.js +1 -1
- package/dist/ops/webgl/attentionMask.js +1 -1
- package/dist/ops/webgl/fusedSoftmax.js +4 -4
- package/dist/ops/webgl/gatherSub.js +1 -1
- package/dist/ops/webgl/gelu.js +2 -2
- package/dist/ops/webgl/log.js +3 -3
- package/dist/ops/webgl/matMulGelu.js +10 -10
- package/dist/ops/webgl/matMulMul.js +1 -1
- package/dist/ops/webgl/mulDropout.js +1 -1
- package/dist/ops/webgl/normRMS.js +2 -2
- package/dist/ops/webgl/qkv.js +1 -1
- package/dist/ops/webgl/rope.js +1 -1
- package/dist/ops/webgl/scatterSub.js +1 -1
- package/dist/ops/webgpu/adamAdjust.js +3 -3
- package/dist/ops/webgpu/adamMoments.js +3 -3
- package/dist/ops/webgpu/appendCache.js +3 -3
- package/dist/ops/webgpu/attentionMask.js +3 -3
- package/dist/ops/webgpu/gatherSub.js +3 -3
- package/dist/ops/webgpu/gelu.js +3 -3
- package/dist/ops/webgpu/normRMS.js +2 -2
- package/dist/ops/webgpu/normRMSGrad.js +5 -5
- package/dist/ops/webgpu/qkv.js +3 -3
- package/dist/ops/webgpu/rope.js +3 -3
- package/dist/ops/webgpu/scatterSub.js +3 -3
- package/dist/ops/webgpu/utils/reductions.js +4 -4
- package/dist/{ops-BFGCx8Ri.js → ops-C_1K_-35.js} +103 -103
- package/dist/{random_width-sZORGo5k.js → random_width-D8Pwy_na.js} +136 -136
- package/dist/{range-CRuAh-gd.js → range-LVHrSLdi.js} +1 -1
- package/dist/{reciprocal-BvGAyKyu.js → reciprocal-CaR9e67G.js} +1 -1
- package/dist/{register_all_kernels-BwDSRN-f.js → register_all_kernels-DUshvVWP.js} +2026 -2049
- package/dist/{reshape-CdBq1WJ6.js → reshape-DEfQGSin.js} +1 -1
- package/dist/{scatter_nd_util-DUstGbU1.js → scatter_nd_util-CUPPNLaA.js} +1 -1
- package/dist/{selu_util-BJEXVvjX.js → selu_util-8vv5JxQV.js} +3 -3
- package/dist/{shared-B8ztnyEk.js → shared-CkNorDcU.js} +83 -83
- package/dist/{shared-wS99K7_n.js → shared-D1elLckx.js} +1 -1
- package/dist/{sin-BeA3tsEd.js → sin-D2CKKmyR.js} +1 -1
- package/dist/{slice-BiOsknYS.js → slice-BnyE-M_7.js} +1 -1
- package/dist/{softmax-Bv_6lyMX.js → softmax-DLoZWYBx.js} +1 -1
- package/dist/{split-B-dikLRw.js → split-By_n4TKP.js} +1 -1
- package/dist/{stack-B17UN2nn.js → stack-DkdFLq37.js} +1 -1
- package/dist/{sum-66ew2byf.js → sum-l_0SqM4h.js} +3 -3
- package/dist/{tensor-JwS7ZYY6.js → tensor-BAQdLqoU.js} +1 -1
- package/dist/{tensor2d-wxPAnDQy.js → tensor2d-BHy261cI.js} +1 -1
- package/dist/training/Adam.js +2 -2
- package/dist/training/AdamExt.js +1 -1
- package/dist/training/DatasetBuilder.js +2 -2
- package/dist/training/Evaluator.d.ts +2 -2
- package/dist/training/FullTrainer.d.ts +3 -3
- package/dist/training/FullTrainer.js +61 -69
- package/dist/training/Trainer.d.ts +15 -3
- package/dist/training/Trainer.js +39 -47
- package/dist/training/sparseCrossEntropy.js +9 -9
- package/dist/utilities/dummy.d.ts +4 -4
- package/dist/utilities/dummy.js +13 -13
- package/dist/utilities/multinomialCPU.js +2 -2
- package/dist/utilities/parameters.d.ts +1 -1
- package/dist/utilities/performance.js +1 -1
- package/dist/utilities/profile.js +1 -1
- package/dist/utilities/safetensors.js +2 -2
- package/dist/utilities/weights.js +2 -2
- package/dist/{variable-BuddVFLa.js → variable-C9hihzDB.js} +1 -1
- package/dist/{webgpu_program-PFzf1hAQ.js → webgpu_program-dFEVbDPL.js} +1 -1
- package/dist/{webgpu_util-D____QpY.js → webgpu_util-DLImlSc6.js} +27 -27
- package/dist/{zeros--BdLQ3oG.js → zeros-VZ72lWXM.js} +1 -1
- package/package.json +2 -3
- package/dist/NanoGPTModel.d.ts +0 -52
- package/dist/NanoGPTModel.js +0 -203
- package/dist/TiedEmbedding-BxOerUmB.js +0 -43
- package/dist/utilities/generate.d.ts +0 -3
- package/dist/utilities/generate.js +0 -22
- package/dist/utilities/save.d.ts +0 -9
- package/dist/utilities/save.js +0 -61
|
@@ -4152,297 +4152,298 @@ function Bs() {
|
|
|
4152
4152
|
*/
|
|
4153
4153
|
Bs();
|
|
4154
4154
|
export {
|
|
4155
|
-
|
|
4156
|
-
|
|
4155
|
+
Rt as $,
|
|
4156
|
+
An as A,
|
|
4157
4157
|
D as B,
|
|
4158
4158
|
T as C,
|
|
4159
4159
|
Tn as D,
|
|
4160
4160
|
f as E,
|
|
4161
|
-
|
|
4161
|
+
Xn as F,
|
|
4162
4162
|
ta as G,
|
|
4163
|
-
|
|
4164
|
-
|
|
4165
|
-
|
|
4166
|
-
|
|
4167
|
-
|
|
4168
|
-
|
|
4169
|
-
|
|
4163
|
+
Sr as H,
|
|
4164
|
+
wi as I,
|
|
4165
|
+
bi as J,
|
|
4166
|
+
C as K,
|
|
4167
|
+
Ct as L,
|
|
4168
|
+
Ge as M,
|
|
4169
|
+
K as N,
|
|
4170
4170
|
H as O,
|
|
4171
4171
|
Ga as P,
|
|
4172
|
-
|
|
4172
|
+
hr as Q,
|
|
4173
4173
|
Ka as R,
|
|
4174
4174
|
ko as S,
|
|
4175
4175
|
x as T,
|
|
4176
|
-
|
|
4177
|
-
|
|
4178
|
-
|
|
4179
|
-
|
|
4180
|
-
|
|
4181
|
-
|
|
4182
|
-
|
|
4176
|
+
Ya as U,
|
|
4177
|
+
So as V,
|
|
4178
|
+
Po as W,
|
|
4179
|
+
ka as X,
|
|
4180
|
+
Ra as Y,
|
|
4181
|
+
Xa as Z,
|
|
4182
|
+
ho as _,
|
|
4183
4183
|
b as a,
|
|
4184
|
-
|
|
4185
|
-
|
|
4186
|
-
|
|
4187
|
-
|
|
4188
|
-
|
|
4189
|
-
|
|
4190
|
-
|
|
4191
|
-
|
|
4192
|
-
|
|
4193
|
-
|
|
4194
|
-
|
|
4195
|
-
|
|
4196
|
-
|
|
4197
|
-
|
|
4198
|
-
|
|
4199
|
-
|
|
4200
|
-
|
|
4201
|
-
|
|
4202
|
-
|
|
4203
|
-
|
|
4204
|
-
|
|
4205
|
-
|
|
4206
|
-
|
|
4207
|
-
|
|
4208
|
-
|
|
4209
|
-
|
|
4210
|
-
|
|
4211
|
-
|
|
4212
|
-
|
|
4213
|
-
|
|
4214
|
-
|
|
4215
|
-
|
|
4216
|
-
|
|
4217
|
-
|
|
4218
|
-
|
|
4219
|
-
|
|
4220
|
-
|
|
4221
|
-
|
|
4222
|
-
|
|
4223
|
-
|
|
4224
|
-
|
|
4225
|
-
|
|
4226
|
-
|
|
4227
|
-
|
|
4228
|
-
|
|
4229
|
-
|
|
4230
|
-
|
|
4231
|
-
|
|
4232
|
-
|
|
4233
|
-
|
|
4234
|
-
|
|
4235
|
-
|
|
4236
|
-
|
|
4237
|
-
|
|
4238
|
-
|
|
4239
|
-
|
|
4240
|
-
|
|
4241
|
-
|
|
4242
|
-
|
|
4243
|
-
|
|
4244
|
-
|
|
4245
|
-
|
|
4246
|
-
|
|
4247
|
-
|
|
4184
|
+
ua as a$,
|
|
4185
|
+
bo as a0,
|
|
4186
|
+
ni as a1,
|
|
4187
|
+
$t as a2,
|
|
4188
|
+
xe as a3,
|
|
4189
|
+
Gs as a4,
|
|
4190
|
+
xa as a5,
|
|
4191
|
+
oo as a6,
|
|
4192
|
+
ea as a7,
|
|
4193
|
+
wn as a8,
|
|
4194
|
+
wr as a9,
|
|
4195
|
+
Qo as aA,
|
|
4196
|
+
Xe as aB,
|
|
4197
|
+
A as aC,
|
|
4198
|
+
st as aD,
|
|
4199
|
+
di as aE,
|
|
4200
|
+
Se as aF,
|
|
4201
|
+
fi as aG,
|
|
4202
|
+
gi as aH,
|
|
4203
|
+
Us as aI,
|
|
4204
|
+
_s as aJ,
|
|
4205
|
+
$s as aK,
|
|
4206
|
+
ut as aL,
|
|
4207
|
+
Ws as aM,
|
|
4208
|
+
jt as aN,
|
|
4209
|
+
ue as aO,
|
|
4210
|
+
Xo as aP,
|
|
4211
|
+
O as aQ,
|
|
4212
|
+
fr as aR,
|
|
4213
|
+
lo as aS,
|
|
4214
|
+
zr as aT,
|
|
4215
|
+
Ma as aU,
|
|
4216
|
+
rs as aV,
|
|
4217
|
+
V as aW,
|
|
4218
|
+
Hr as aX,
|
|
4219
|
+
na as aY,
|
|
4220
|
+
sa as aZ,
|
|
4221
|
+
aa as a_,
|
|
4222
|
+
ne as aa,
|
|
4223
|
+
Br as ab,
|
|
4224
|
+
fo as ac,
|
|
4225
|
+
pi as ad,
|
|
4226
|
+
xt as ae,
|
|
4227
|
+
S as af,
|
|
4228
|
+
Kr as ag,
|
|
4229
|
+
fa as ah,
|
|
4230
|
+
F as ai,
|
|
4231
|
+
pe as aj,
|
|
4232
|
+
ti as ak,
|
|
4233
|
+
gt as al,
|
|
4234
|
+
Zr as am,
|
|
4235
|
+
qn as an,
|
|
4236
|
+
ie as ao,
|
|
4237
|
+
la as ap,
|
|
4238
|
+
Zo as aq,
|
|
4239
|
+
za as ar,
|
|
4240
|
+
Ne as as,
|
|
4241
|
+
po as at,
|
|
4242
|
+
Vr as au,
|
|
4243
|
+
Ja as av,
|
|
4244
|
+
no as aw,
|
|
4245
|
+
Vo as ax,
|
|
4246
|
+
yi as ay,
|
|
4247
|
+
Ns as az,
|
|
4248
4248
|
q as b,
|
|
4249
|
-
|
|
4250
|
-
|
|
4251
|
-
|
|
4252
|
-
|
|
4253
|
-
|
|
4254
|
-
|
|
4255
|
-
|
|
4256
|
-
|
|
4257
|
-
|
|
4258
|
-
|
|
4259
|
-
|
|
4260
|
-
|
|
4261
|
-
|
|
4262
|
-
|
|
4263
|
-
|
|
4264
|
-
|
|
4265
|
-
|
|
4266
|
-
|
|
4267
|
-
|
|
4268
|
-
|
|
4269
|
-
|
|
4270
|
-
|
|
4271
|
-
|
|
4272
|
-
|
|
4273
|
-
|
|
4274
|
-
|
|
4275
|
-
|
|
4276
|
-
|
|
4277
|
-
|
|
4278
|
-
|
|
4279
|
-
|
|
4280
|
-
|
|
4281
|
-
|
|
4282
|
-
|
|
4283
|
-
|
|
4284
|
-
|
|
4285
|
-
|
|
4286
|
-
|
|
4287
|
-
|
|
4288
|
-
|
|
4289
|
-
|
|
4290
|
-
|
|
4291
|
-
|
|
4292
|
-
|
|
4293
|
-
|
|
4294
|
-
|
|
4295
|
-
|
|
4296
|
-
|
|
4297
|
-
|
|
4298
|
-
|
|
4299
|
-
|
|
4300
|
-
|
|
4301
|
-
|
|
4302
|
-
|
|
4303
|
-
|
|
4304
|
-
|
|
4305
|
-
|
|
4306
|
-
|
|
4307
|
-
|
|
4308
|
-
|
|
4309
|
-
|
|
4310
|
-
|
|
4311
|
-
|
|
4312
|
-
|
|
4249
|
+
pa as b$,
|
|
4250
|
+
ha as b0,
|
|
4251
|
+
Na as b1,
|
|
4252
|
+
ma as b2,
|
|
4253
|
+
Da as b3,
|
|
4254
|
+
Ha as b4,
|
|
4255
|
+
ro as b5,
|
|
4256
|
+
Uo as b6,
|
|
4257
|
+
Oo as b7,
|
|
4258
|
+
Rr as b8,
|
|
4259
|
+
Yr as b9,
|
|
4260
|
+
fn as bA,
|
|
4261
|
+
en as bB,
|
|
4262
|
+
Bo as bC,
|
|
4263
|
+
Do as bD,
|
|
4264
|
+
nn as bE,
|
|
4265
|
+
Re as bF,
|
|
4266
|
+
Zs as bG,
|
|
4267
|
+
Qs as bH,
|
|
4268
|
+
tr as bI,
|
|
4269
|
+
ir as bJ,
|
|
4270
|
+
lr as bK,
|
|
4271
|
+
Co as bL,
|
|
4272
|
+
dr as bM,
|
|
4273
|
+
Qr as bN,
|
|
4274
|
+
br as bO,
|
|
4275
|
+
kr as bP,
|
|
4276
|
+
Tr as bQ,
|
|
4277
|
+
Er as bR,
|
|
4278
|
+
vr as bS,
|
|
4279
|
+
Mr as bT,
|
|
4280
|
+
Dr as bU,
|
|
4281
|
+
Fr as bV,
|
|
4282
|
+
xr as bW,
|
|
4283
|
+
$r as bX,
|
|
4284
|
+
jr as bY,
|
|
4285
|
+
ga as bZ,
|
|
4286
|
+
yo as b_,
|
|
4287
|
+
jo as ba,
|
|
4288
|
+
Ca as bb,
|
|
4289
|
+
La as bc,
|
|
4290
|
+
Pa as bd,
|
|
4291
|
+
to as be,
|
|
4292
|
+
Za as bf,
|
|
4293
|
+
_o as bg,
|
|
4294
|
+
Ke as bh,
|
|
4295
|
+
zs as bi,
|
|
4296
|
+
Hs as bj,
|
|
4297
|
+
Ks as bk,
|
|
4298
|
+
oe as bl,
|
|
4299
|
+
Vs as bm,
|
|
4300
|
+
_t as bn,
|
|
4301
|
+
ae as bo,
|
|
4302
|
+
gr as bp,
|
|
4303
|
+
te as bq,
|
|
4304
|
+
yr as br,
|
|
4305
|
+
qr as bs,
|
|
4306
|
+
Xr as bt,
|
|
4307
|
+
Ye as bu,
|
|
4308
|
+
Ze as bv,
|
|
4309
|
+
Qe as bw,
|
|
4310
|
+
$a as bx,
|
|
4311
|
+
Va as by,
|
|
4312
|
+
ao as bz,
|
|
4313
4313
|
tt as c,
|
|
4314
|
-
|
|
4315
|
-
|
|
4316
|
-
|
|
4317
|
-
|
|
4318
|
-
|
|
4319
|
-
|
|
4320
|
-
|
|
4321
|
-
|
|
4322
|
-
|
|
4323
|
-
|
|
4324
|
-
|
|
4325
|
-
|
|
4326
|
-
|
|
4327
|
-
|
|
4328
|
-
|
|
4329
|
-
|
|
4330
|
-
|
|
4331
|
-
|
|
4332
|
-
|
|
4333
|
-
|
|
4334
|
-
|
|
4335
|
-
|
|
4336
|
-
|
|
4337
|
-
|
|
4338
|
-
|
|
4339
|
-
|
|
4340
|
-
|
|
4341
|
-
|
|
4342
|
-
|
|
4343
|
-
|
|
4344
|
-
|
|
4345
|
-
|
|
4346
|
-
|
|
4347
|
-
|
|
4348
|
-
|
|
4349
|
-
|
|
4350
|
-
|
|
4351
|
-
|
|
4352
|
-
|
|
4353
|
-
|
|
4354
|
-
|
|
4355
|
-
|
|
4356
|
-
|
|
4357
|
-
|
|
4358
|
-
|
|
4359
|
-
|
|
4360
|
-
|
|
4361
|
-
|
|
4362
|
-
|
|
4363
|
-
|
|
4364
|
-
|
|
4365
|
-
|
|
4366
|
-
|
|
4367
|
-
|
|
4368
|
-
|
|
4369
|
-
|
|
4370
|
-
|
|
4371
|
-
|
|
4372
|
-
|
|
4373
|
-
|
|
4374
|
-
|
|
4375
|
-
|
|
4376
|
-
|
|
4377
|
-
|
|
4314
|
+
xs as c$,
|
|
4315
|
+
Ia as c0,
|
|
4316
|
+
Ea as c1,
|
|
4317
|
+
Ba as c2,
|
|
4318
|
+
Oa as c3,
|
|
4319
|
+
_a as c4,
|
|
4320
|
+
Ua as c5,
|
|
4321
|
+
wo as c6,
|
|
4322
|
+
so as c7,
|
|
4323
|
+
uo as c8,
|
|
4324
|
+
go as c9,
|
|
4325
|
+
Ar as cA,
|
|
4326
|
+
_r as cB,
|
|
4327
|
+
Gr as cC,
|
|
4328
|
+
Or as cD,
|
|
4329
|
+
Wr as cE,
|
|
4330
|
+
oa as cF,
|
|
4331
|
+
ia as cG,
|
|
4332
|
+
ca as cH,
|
|
4333
|
+
ba as cI,
|
|
4334
|
+
Sa as cJ,
|
|
4335
|
+
wa as cK,
|
|
4336
|
+
Aa as cL,
|
|
4337
|
+
Ta as cM,
|
|
4338
|
+
Fa as cN,
|
|
4339
|
+
tn as cO,
|
|
4340
|
+
eo as cP,
|
|
4341
|
+
Qa as cQ,
|
|
4342
|
+
mo as cR,
|
|
4343
|
+
Mo as cS,
|
|
4344
|
+
$o as cT,
|
|
4345
|
+
ms as cU,
|
|
4346
|
+
sn as cV,
|
|
4347
|
+
li as cW,
|
|
4348
|
+
ws as cX,
|
|
4349
|
+
bs as cY,
|
|
4350
|
+
G as cZ,
|
|
4351
|
+
Oe as c_,
|
|
4352
|
+
zo as ca,
|
|
4353
|
+
Ir as cb,
|
|
4354
|
+
Ko as cc,
|
|
4355
|
+
Cr as cd,
|
|
4356
|
+
Pr as ce,
|
|
4357
|
+
qo as cf,
|
|
4358
|
+
Be as cg,
|
|
4359
|
+
Es as ch,
|
|
4360
|
+
As as ci,
|
|
4361
|
+
Is as cj,
|
|
4362
|
+
Ss as ck,
|
|
4363
|
+
Ts as cl,
|
|
4364
|
+
ks as cm,
|
|
4365
|
+
Xs as cn,
|
|
4366
|
+
Js as co,
|
|
4367
|
+
Ys as cp,
|
|
4368
|
+
er as cq,
|
|
4369
|
+
nr as cr,
|
|
4370
|
+
sr as cs,
|
|
4371
|
+
or as ct,
|
|
4372
|
+
rr as cu,
|
|
4373
|
+
ar as cv,
|
|
4374
|
+
ur as cw,
|
|
4375
|
+
cr as cx,
|
|
4376
|
+
mr as cy,
|
|
4377
|
+
He as cz,
|
|
4378
4378
|
M as d,
|
|
4379
|
-
|
|
4380
|
-
|
|
4381
|
-
|
|
4382
|
-
|
|
4383
|
-
|
|
4384
|
-
|
|
4385
|
-
|
|
4386
|
-
|
|
4387
|
-
|
|
4388
|
-
|
|
4389
|
-
|
|
4390
|
-
|
|
4391
|
-
|
|
4392
|
-
|
|
4393
|
-
|
|
4394
|
-
|
|
4395
|
-
|
|
4396
|
-
|
|
4397
|
-
|
|
4398
|
-
|
|
4399
|
-
|
|
4400
|
-
|
|
4401
|
-
|
|
4402
|
-
|
|
4403
|
-
|
|
4404
|
-
|
|
4405
|
-
|
|
4406
|
-
|
|
4407
|
-
|
|
4408
|
-
|
|
4409
|
-
|
|
4410
|
-
|
|
4411
|
-
|
|
4412
|
-
|
|
4413
|
-
|
|
4414
|
-
|
|
4415
|
-
|
|
4416
|
-
|
|
4417
|
-
|
|
4418
|
-
|
|
4419
|
-
|
|
4420
|
-
|
|
4421
|
-
|
|
4422
|
-
|
|
4423
|
-
|
|
4424
|
-
|
|
4425
|
-
|
|
4379
|
+
mi as d0,
|
|
4380
|
+
ui as d1,
|
|
4381
|
+
hi as d2,
|
|
4382
|
+
Fe as d3,
|
|
4383
|
+
Rs as d4,
|
|
4384
|
+
js as d5,
|
|
4385
|
+
lt as d6,
|
|
4386
|
+
Cs as d7,
|
|
4387
|
+
Nt as d8,
|
|
4388
|
+
ci as d9,
|
|
4389
|
+
ct as dA,
|
|
4390
|
+
qs as dB,
|
|
4391
|
+
Wa as dC,
|
|
4392
|
+
ja as dD,
|
|
4393
|
+
qa as dE,
|
|
4394
|
+
Io as dF,
|
|
4395
|
+
To as dG,
|
|
4396
|
+
xo as dH,
|
|
4397
|
+
No as dI,
|
|
4398
|
+
ei as dJ,
|
|
4399
|
+
Os as dK,
|
|
4400
|
+
It as dL,
|
|
4401
|
+
kn as dM,
|
|
4402
|
+
Je as da,
|
|
4403
|
+
yt as db,
|
|
4404
|
+
Ls as dc,
|
|
4405
|
+
pr as dd,
|
|
4406
|
+
Nr as de,
|
|
4407
|
+
Lr as df,
|
|
4408
|
+
Ur as dg,
|
|
4409
|
+
Jr as dh,
|
|
4410
|
+
Wo as di,
|
|
4411
|
+
Ho as dj,
|
|
4412
|
+
ra as dk,
|
|
4413
|
+
da as dl,
|
|
4414
|
+
ya as dm,
|
|
4415
|
+
va as dn,
|
|
4416
|
+
co as dp,
|
|
4417
|
+
Eo as dq,
|
|
4418
|
+
Ao as dr,
|
|
4419
|
+
vo as ds,
|
|
4420
|
+
Fo as dt,
|
|
4421
|
+
Ro as du,
|
|
4422
|
+
io as dv,
|
|
4423
|
+
Lo as dw,
|
|
4424
|
+
Go as dx,
|
|
4425
|
+
de as dy,
|
|
4426
|
+
ts as dz,
|
|
4426
4427
|
si as e,
|
|
4427
4428
|
Jo as f,
|
|
4428
4429
|
ii as g,
|
|
4429
4430
|
as as h,
|
|
4430
4431
|
Yo as i,
|
|
4431
4432
|
z as j,
|
|
4432
|
-
|
|
4433
|
-
|
|
4433
|
+
Bn as k,
|
|
4434
|
+
os as l,
|
|
4434
4435
|
ri as m,
|
|
4435
|
-
|
|
4436
|
-
|
|
4436
|
+
y as n,
|
|
4437
|
+
$ as o,
|
|
4437
4438
|
$e as p,
|
|
4438
|
-
|
|
4439
|
+
k as q,
|
|
4439
4440
|
oi as r,
|
|
4440
4441
|
ai as s,
|
|
4441
4442
|
E as t,
|
|
4442
4443
|
fe as u,
|
|
4443
4444
|
hs as v,
|
|
4444
|
-
|
|
4445
|
+
nt as w,
|
|
4445
4446
|
Qt as x,
|
|
4446
|
-
|
|
4447
|
-
|
|
4447
|
+
Ps as y,
|
|
4448
|
+
In as z
|
|
4448
4449
|
};
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import { u as O, f as Y } from "./gpgpu_math-
|
|
3
|
-
import { f as v } from "./backend_util-
|
|
1
|
+
import { l as B, j as G, ao as K, a9 as W, ap as z, aq as V, af as N, ar as F, u as S } from "./index-CUQrfsw_.js";
|
|
2
|
+
import { u as O, f as Y } from "./gpgpu_math-TFLxaLkw.js";
|
|
3
|
+
import { f as v } from "./backend_util-BJ-_jSeK.js";
|
|
4
4
|
/**
|
|
5
5
|
* @license
|
|
6
6
|
* Copyright 2018 Google LLC. All Rights Reserved.
|
|
@@ -315,12 +315,12 @@ const me = {
|
|
|
315
315
|
* limitations under the License.
|
|
316
316
|
* =============================================================================
|
|
317
317
|
*/
|
|
318
|
-
const w = "return (a < 0.) ? b * a : a;",
|
|
318
|
+
const w = "return (a < 0.) ? b * a : a;", R = `
|
|
319
319
|
vec4 aLessThanZero = vec4(lessThan(a, vec4(0.)));
|
|
320
320
|
return (aLessThanZero * (b * a)) + ((vec4(1.0) - aLessThanZero) * a);
|
|
321
321
|
`;
|
|
322
322
|
function oe(t) {
|
|
323
|
-
const { inputs: e, backend: s, attrs: r } = t, { x: u } = e, { alpha: n } = r, o = s.makeTensorInfo([], "float32", V(n, "float32")), i = N().getBool("WEBGL_PACK_BINARY_OPERATIONS") ? new E(
|
|
323
|
+
const { inputs: e, backend: s, attrs: r } = t, { x: u } = e, { alpha: n } = r, o = s.makeTensorInfo([], "float32", V(n, "float32")), i = N().getBool("WEBGL_PACK_BINARY_OPERATIONS") ? new E(R, u.shape, o.shape) : new b(w, u.shape, o.shape), a = s.runWebGLProgram(i, [u, o], "float32");
|
|
324
324
|
return s.disposeIntermediateTensorInfo(o), a;
|
|
325
325
|
}
|
|
326
326
|
const be = {
|
|
@@ -344,12 +344,12 @@ const be = {
|
|
|
344
344
|
* limitations under the License.
|
|
345
345
|
* =============================================================================
|
|
346
346
|
*/
|
|
347
|
-
const
|
|
347
|
+
const k = "return (a < 0.) ? b * a : a;", U = `
|
|
348
348
|
vec4 aLessThanZero = vec4(lessThan(a, vec4(0.)));
|
|
349
349
|
return (aLessThanZero * (b * a)) + ((vec4(1.0) - aLessThanZero) * a);
|
|
350
350
|
`;
|
|
351
351
|
function ue(t) {
|
|
352
|
-
const { inputs: e, backend: s } = t, { x: r, alpha: u } = e, n = N().getBool("WEBGL_PACK_BINARY_OPERATIONS") ? new E(U, r.shape, u.shape) : new b(
|
|
352
|
+
const { inputs: e, backend: s } = t, { x: r, alpha: u } = e, n = N().getBool("WEBGL_PACK_BINARY_OPERATIONS") ? new E(U, r.shape, u.shape) : new b(k, r.shape, u.shape);
|
|
353
353
|
return s.runWebGLProgram(n, [r, u], "float32");
|
|
354
354
|
}
|
|
355
355
|
const Ne = {
|
|
@@ -433,9 +433,9 @@ function Ae(t, e = !1) {
|
|
|
433
433
|
if (t === "relu6")
|
|
434
434
|
return e ? ae : Q;
|
|
435
435
|
if (t === "prelu")
|
|
436
|
-
return e ? U :
|
|
436
|
+
return e ? U : k;
|
|
437
437
|
if (t === "leakyrelu")
|
|
438
|
-
return e ?
|
|
438
|
+
return e ? R : w;
|
|
439
439
|
if (t === "sigmoid")
|
|
440
440
|
return e ? re : X;
|
|
441
441
|
throw new Error(`Activation ${t} has not been implemented for the WebGL backend.`);
|
|
@@ -1,27 +1,22 @@
|
|
|
1
|
-
import { GPTConfig } from '../config';
|
|
1
|
+
import { GPTConfig } from '../models/config';
|
|
2
2
|
import { default as MemoryProfiler } from '../utilities/profile';
|
|
3
3
|
import { default as RoPECache } from './RoPECache';
|
|
4
4
|
import { Tensor, Variable } from '@tensorflow/tfjs-core';
|
|
5
|
-
export interface LayerConfig {
|
|
6
|
-
checkpointing?: boolean;
|
|
7
|
-
profiler?: MemoryProfiler;
|
|
8
|
-
ropeCache?: RoPECache;
|
|
9
|
-
}
|
|
10
|
-
export interface GPTLayerConfig {
|
|
11
|
-
gpt: GPTConfig;
|
|
12
|
-
layerConfig: LayerConfig;
|
|
13
|
-
}
|
|
14
5
|
export interface ForwardAttributes {
|
|
15
6
|
training: boolean;
|
|
7
|
+
checkpointing?: boolean;
|
|
8
|
+
ropeCache?: RoPECache;
|
|
16
9
|
}
|
|
17
10
|
export default abstract class BaseLayer<ATTR extends ForwardAttributes = ForwardAttributes> {
|
|
18
11
|
readonly parent?: BaseLayer;
|
|
19
|
-
readonly config:
|
|
12
|
+
readonly config: GPTConfig;
|
|
20
13
|
private _variables;
|
|
21
14
|
private _trainable;
|
|
22
15
|
readonly children: BaseLayer[];
|
|
23
|
-
|
|
16
|
+
private profiler?;
|
|
17
|
+
constructor(config: GPTConfig, parent?: BaseLayer);
|
|
24
18
|
getProfiler(): MemoryProfiler | undefined;
|
|
19
|
+
setProfiler(profiler: MemoryProfiler | null): void;
|
|
25
20
|
startMemory(): void;
|
|
26
21
|
endMemory(label: string): void;
|
|
27
22
|
addVariable(name: string, variable?: Variable): void;
|
|
@@ -29,7 +24,7 @@ export default abstract class BaseLayer<ATTR extends ForwardAttributes = Forward
|
|
|
29
24
|
get trainableVariables(): Variable[];
|
|
30
25
|
get trainable(): boolean;
|
|
31
26
|
set trainable(value: boolean);
|
|
32
|
-
getVariable(name: string): Variable;
|
|
27
|
+
getVariable(name: string, recursive?: boolean): Variable;
|
|
33
28
|
hasVariable(name: string): boolean;
|
|
34
29
|
setVariable(name: string, variable: Variable): void;
|
|
35
30
|
saveWeights(map: Map<string, Tensor[]>): void;
|