向量化¶
简述¶
向量化过程是将item转换为向量的过程，其前置步骤为语法解析、成分分解、令牌化，本部分将先后介绍如何获得数据集、如何使用本地的预训练模型、如何直接调用远程提供的预训练模型。
获得数据集¶
概述¶
此部分通过调用 OpenLUNA.json 获得。
I2V¶
概述¶
使用自己提供的任一预训练模型（给出模型存放路径即可）将给定的题目文本转成向量。
优点：可以使用自己的模型，另可调整训练参数，灵活性强。
D2V¶
导入类¶
[1]:
from EduNLP.I2V import D2V
D:\MySoftwares\Anaconda\envs\data\lib\site-packages\gensim\similarities\__init__.py:15: UserWarning: The gensim.similarities.levenshtein submodule is disabled, because the optional Levenshtein package <https://pypi.org/project/python-Levenshtein/> is unavailable. Install Levenhstein (e.g. `pip install python-Levenshtein`) to suppress this warning.
warnings.warn(msg)
输入¶
类型：str
内容：题目文本 （text）
[2]:
items = [
r"1如图几何图形．此图由三个半圆构成，三个半圆的直径分别为直角三角形$ABC$的斜边$BC$, 直角边$AB$, $AC$.$\bigtriangleup ABC$的三边所围成的区域记为$I$,黑色部分记为$II$, 其余部分记为$III$.在整个图形中随机取一点，此点取自$I,II,III$的概率分别记为$p_1,p_2,p_3$,则$\SIFChoice$$\FigureID{1}$",
r"2如图来自古希腊数学家希波克拉底所研究的几何图形．此图由三个半圆构成，三个半圆的直径分别为直角三角形$ABC$的斜边$BC$, 直角边$AB$, $AC$.$\bigtriangleup ABC$的三边所围成的区域记为$I$,黑色部分记为$II$, 其余部分记为$III$.在整个图形中随机取一点，此点取自$I,II,III$的概率分别记为$p_1,p_2,p_3$,则$\SIFChoice$$\FigureID{1}$"
]
输出¶
[3]:
model_path = "./d2v/test_d2v_256.bin"
i2v = D2V("pure_text","d2v",filepath=model_path, pretrained_t2v = False)
item_vectors, token_vectors = i2v(items)
print(item_vectors[0])
print(token_vectors) # For d2v, token_vector is None
print("shape of item_vector: ",len(item_vectors), item_vectors[0].shape)
[ 0.10603202 -0.10537548 -0.04773913 0.15573525 0.25898772 -0.06423073
-0.02817309 0.0068187 -0.07323898 0.06517941 0.07943465 0.14800762
-0.06772996 -0.23892336 0.04638071 0.1539897 0.17565852 0.02895202
-0.18859927 0.2180874 0.00909669 0.06621908 -0.02090263 -0.13006955
-0.21020882 0.00618349 0.00531093 -0.04877732 -0.06709669 -0.04705636
0.09211092 0.13896106 -0.07455818 0.06019318 -0.09071473 0.12701215
0.13018885 -0.02784999 0.10064025 -0.07757548 -0.05522636 -0.02657779
-0.04159601 -0.03008493 0.10995369 -0.00587291 0.05902484 0.06532726
0.04887666 0.01902074 0.03713945 0.03691795 0.12516327 0.07410683
-0.14467879 0.05678609 0.02574336 -0.1320522 0.07502684 0.07929367
-0.06655917 -0.0144536 0.02595847 0.04403471 0.21743318 -0.02525017
-0.0416184 0.21441495 -0.09308876 -0.09418222 0.08030997 0.00492512
-0.04921608 -0.07808654 -0.03323801 0.0879296 -0.04668022 -0.0696011
0.06708417 0.06555629 -0.07418457 -0.13050951 -0.01802611 0.11730465
-0.0479078 0.06389603 0.12324224 -0.17746696 -0.09874132 -0.07683054
0.06596514 -0.04210603 0.03182372 -0.1455575 0.03900012 0.13290605
-0.07672353 -0.02826704 -0.00803517 -0.09681892 -0.15212329 -0.10524812
0.03367848 0.10413344 -0.0089777 0.0583192 -0.01553376 0.02675472
0.12278829 0.01667286 0.01958599 -0.06468913 0.08307286 0.07304061
-0.10451686 -0.04367925 0.0143903 0.11394493 0.00759796 -0.03158598
-0.01733392 -0.12918264 0.1761386 -0.02913121 -0.01364522 0.01497996
0.09318532 -0.03958051 0.00465893 -0.01766865 -0.03531685 0.01445563
0.05919004 -0.10480376 -0.08359206 -0.08283877 -0.04920156 0.0486405
0.0059151 -0.03783213 -0.01815955 -0.0157437 0.2334638 0.15233137
-0.2698607 -0.04492244 0.03728078 0.06730984 0.09165722 0.07212968
-0.1418279 -0.10517611 -0.0469548 -0.01878718 -0.08850995 0.07481015
0.15206474 0.0923347 -0.08849481 0.01736124 0.12647657 -0.03515046
0.07980374 -0.06639698 0.00411603 0.0479564 0.04197159 0.0854824
0.103918 -0.01195896 0.05059687 -0.03206704 0.0277859 0.05210226
-0.15160614 -0.01996467 -0.00720571 -0.01154042 0.10944121 -0.00173247
0.11439639 -0.04765575 0.05989955 -0.05265343 0.11914644 0.0085329
-0.13220952 -0.1538407 -0.07261448 0.04143476 0.15447438 0.02005473
0.14354227 0.10015973 0.12290012 0.05011315 0.0425972 -0.13731483
0.02323116 -0.1031343 -0.17960383 -0.04875064 0.14352156 0.04516263
-0.04433561 0.11548021 -0.2057457 -0.02778868 -0.06643672 0.05604808
0.04864014 -0.03015646 0.07734285 0.00573904 0.01155302 0.02486293
0.16259493 0.05099423 -0.15283771 -0.01909443 -0.12749314 0.06718695
0.08334705 -0.05442797 0.03448674 -0.00542413 0.00832719 0.02702984
-0.02359959 -0.00855793 -0.19381124 -0.13036375 -0.0351354 -0.03983364
0.0133928 0.07395492 0.04119737 0.05661048 0.08151852 -0.1529391
0.00742581 0.05521343 0.02089992 -0.00824985 -0.00211842 -0.05555268
0.05448649 -0.02032894 -0.0760811 -0.01713146 -0.16146915 0.10822926
-0.1240218 -0.03639562 -0.20028785 -0.02452293]
None
shape of item_vector: 2 (256,)
W2V¶
[4]:
from EduNLP.I2V import W2V
[5]:
model_path = "./w2v/general_literal_300/general_literal_300.kv"
i2v = W2V("pure_text","w2v",filepath=model_path, pretrained_t2v = False)
item_vectors, token_vectors = i2v(items)
print(item_vectors[0])
print(token_vectors[0][0])
print("shape of item_vectors: ", len(item_vectors), item_vectors[0].shape)
print("shape of token_vectors: ", len(token_vectors), len(token_vectors[0]), len(token_vectors[0][0]))
[-1.34266680e-03 5.19845746e-02 -1.98070258e-02 -4.17470075e-02
4.92814295e-02 -1.70883536e-01 -2.16597781e-01 -3.12069029e-01
8.96430463e-02 -1.31331667e-01 9.16494895e-03 -3.22572999e-02
3.07940125e-01 1.92060292e-01 1.31043345e-02 6.10962026e-02
2.21019030e-01 -3.53541046e-01 1.34150490e-01 1.14867561e-01
1.17448963e-01 2.27990672e-01 -1.65213019e-01 2.78246611e-01
-4.36594114e-02 -1.37816787e-01 -1.07707813e-01 -1.80805102e-01
1.20028563e-01 -1.14409983e-01 6.19181581e-02 -1.79836392e-01
7.68677965e-02 2.41688967e-01 6.20721914e-02 -7.59824514e-02
1.79465964e-01 1.69306010e-01 -1.99512452e-01 -9.75036696e-02
1.02485821e-01 -1.59723386e-01 -1.67252243e-01 1.52240042e-02
-5.98842278e-03 6.47612512e-02 8.48228261e-02 2.67874986e-01
-1.73656959e-02 -4.40101810e-02 9.11948457e-02 1.40905827e-01
6.33735815e-03 2.03221604e-01 -1.97303146e-01 1.32987842e-01
-1.80283263e-01 3.64040211e-02 2.49624569e-02 7.49479085e-02
-2.05568615e-02 -4.02397066e-02 -1.08619891e-01 -1.04757406e-01
-8.36341307e-02 6.61163032e-02 -1.11632387e-03 3.96131463e-02
-3.51454802e-02 9.09155831e-02 1.87938929e-01 -2.40521863e-01
-5.97307160e-02 1.74426511e-01 1.56350788e-02 -4.20243442e-02
-1.90285146e-02 -3.85696471e-01 -1.01543151e-01 -1.42145246e-01
2.33298853e-01 1.85939763e-02 -2.68633634e-01 -3.60178575e-02
3.64447385e-02 -1.70443758e-01 1.03326524e-02 -1.47003353e-01
-9.38110873e-02 1.63190335e-01 1.20674491e-01 -2.36976147e-02
-6.52602538e-02 1.29773334e-01 -4.23593611e-01 2.43276700e-01
-8.00978579e-03 -9.39133018e-03 1.17623486e-01 1.16482794e-01
-9.27330479e-02 6.17316999e-02 1.51295820e-02 2.25901395e-01
7.31975585e-02 2.29724105e-02 9.95925292e-02 -1.10697523e-01
2.28960160e-02 8.65939483e-02 1.16645083e-01 -7.00058565e-02
1.13389529e-01 -5.30471019e-02 1.43660516e-01 1.61379650e-02
6.77419230e-02 8.09707418e-02 2.09957212e-01 -6.64654151e-02
-1.81450248e-01 2.21659631e-01 -4.53737518e-03 4.69567403e-02
8.59350115e-02 1.17934339e-01 -1.98988736e-01 -4.13361974e-02
1.26167178e-01 3.84825058e-02 1.64396539e-01 -1.63344927e-02
9.12889242e-02 -1.13650873e-01 -1.37156844e-02 2.06742659e-02
-9.15742964e-02 7.41296187e-02 2.50813574e-01 -1.35987863e-01
-1.11708120e-01 -1.52451068e-01 1.08608082e-01 4.99855466e-02
1.68440521e-01 -2.47063249e-01 -2.21773341e-01 4.81536575e-02
-7.66365305e-02 2.55189091e-01 -5.60788438e-03 -2.69066542e-02
2.07698755e-02 1.36008840e-02 1.33086294e-01 -3.80828045e-02
-7.03251585e-02 -6.18199483e-02 9.03518647e-02 -1.89310908e-01
-5.30523732e-02 -2.04426926e-02 -2.27697566e-01 7.68405125e-02
1.28568143e-01 1.07449636e-01 -1.98028013e-01 -2.67155319e-01
-5.17064631e-02 -1.62200809e-01 1.87425911e-01 4.74511087e-02
-4.24213745e-02 -2.71449953e-01 -2.83543557e-01 -2.36278087e-01
-4.38764729e-02 1.67618364e-01 -2.51966029e-01 -2.73265123e-01
-1.68406263e-01 -3.58684808e-01 2.44145632e-01 -2.55741596e-01
-2.28520826e-01 2.39279866e-01 -1.68833986e-01 1.66422993e-01
-3.53969544e-01 -1.10907286e-01 -6.29489049e-02 -4.55605611e-02
1.46765754e-01 1.95176788e-02 -3.80197394e-04 3.36615089e-03
-1.42359287e-01 -1.06109239e-01 -3.36164385e-02 3.16832401e-02
1.09924652e-01 2.10711379e-02 1.58359021e-01 1.71957895e-01
4.08717275e-01 -4.28679548e-02 -6.48310632e-02 1.27063962e-02
5.73479272e-02 1.40002951e-01 -3.66613895e-01 8.07148069e-02
2.11823225e-01 -1.10516161e-01 -2.01001287e-01 3.22122797e-02
5.47345541e-02 2.30176803e-02 -9.94866490e-02 -4.44128877e-03
6.64432272e-02 1.28168091e-01 -2.34743133e-01 3.17057431e-01
-8.75139013e-02 2.66474396e-01 -3.12204093e-01 7.78969377e-03
6.21694922e-02 7.64596611e-02 -8.79013091e-02 1.01901866e-01
3.23867425e-02 -2.27650225e-01 9.44062844e-02 -5.54776154e-02
-7.03687780e-03 5.66167049e-02 -1.87480077e-01 9.11692008e-02
1.51293352e-01 -1.92774653e-01 2.23165095e-01 2.26982050e-02
-2.70489484e-01 1.25889871e-02 -2.30410039e-01 1.40989587e-01
2.20341813e-02 2.70313285e-02 6.07572980e-02 8.79322216e-02
7.42911696e-02 -2.76499927e-01 2.05189809e-01 -1.84953049e-01
-1.68468937e-01 1.85525760e-01 -3.32091609e-03 2.29632735e-01
7.13749304e-02 -2.75445748e-02 2.74335817e-02 1.65132031e-01
-1.64373592e-01 -1.14925921e-01 4.98081557e-02 2.10796613e-02
-2.07561441e-02 5.90056814e-02 -1.25214513e-02 3.78197022e-02
3.62618983e-01 1.72744930e-01 -8.75385627e-02 1.52320743e-01
1.29331559e-01 -1.34815618e-01 6.12287596e-02 7.30569959e-02
5.37401363e-02 -1.46815628e-01 -2.61263877e-01 2.18300954e-01
8.95068944e-02 -6.59529120e-02 -8.52308050e-02 2.63195664e-01
2.09921718e-01 -1.73417434e-01 2.11869497e-02 7.06950724e-02
-7.89924189e-02 1.11086138e-01 -1.29149273e-01 1.16233543e-01
2.16104537e-01 -3.05427730e-01 -2.46336535e-01 7.59556964e-02]
[-9.74057533e-04 1.39671087e-03 -2.67836265e-04 3.15364590e-03
2.96666636e-04 2.81736051e-04 -2.63743475e-03 1.52303779e-03
1.01379235e-03 1.57282199e-03 -1.71113803e-04 8.02559836e-04
2.57097790e-03 1.81893981e-03 2.63088616e-03 3.40178027e-04
-2.11292668e-03 -2.50976160e-03 -1.20709895e-03 1.67239667e-03
-2.58512655e-03 -1.26207829e-03 1.39700493e-03 1.95603608e-03
2.89038429e-03 -2.39552581e-03 -3.91247275e-04 -3.21114226e-03
9.58531688e-04 8.29325523e-04 -1.59795280e-03 1.25081465e-03
3.81096208e-04 1.59411912e-03 -8.54889571e-04 -3.06331483e-03
1.97919217e-04 -9.37395904e-04 2.09570490e-03 1.22304517e-03
-2.73981970e-03 1.85640890e-03 -8.50516954e-04 2.05107126e-03
3.25771095e-03 -2.02651741e-03 -2.99406121e-03 -2.29128683e-03
7.10814027e-04 -2.45556026e-03 3.29233892e-03 -1.30950764e-03
-1.89368729e-03 1.27877074e-03 2.70718103e-03 -2.21936312e-03
1.69272022e-03 7.79648602e-04 2.15323060e-03 -1.90569717e-03
-2.24422058e-03 -7.12279463e-04 1.38582790e-03 2.27209576e-03
-2.48074066e-03 -1.57340372e-03 -2.78435787e-03 2.53080134e-03
9.29941365e-04 9.57158394e-04 -4.04856197e-04 7.77039502e-04
2.93451315e-03 1.50868180e-03 -2.39667180e-03 1.94984837e-03
-1.30266906e-03 -3.10783624e-03 2.75730062e-03 6.42884581e-04
-5.66801231e-04 7.95386615e-04 -2.41047610e-03 1.00338063e-03
2.82178726e-03 -2.43772753e-03 5.73688361e-04 1.23452744e-03
-3.01872566e-03 -1.07384368e-03 3.28231254e-03 -5.47548116e-04
2.18831864e-03 -3.27980524e-04 1.68147963e-03 2.44990899e-03
-2.45229807e-03 1.02455064e-03 -2.29938584e-03 2.91304989e-03
1.60753564e-03 1.12590473e-03 3.00752558e-03 1.44218525e-03
3.16225761e-03 -1.64008932e-04 -5.27421653e-04 1.06547831e-03
1.11937604e-03 4.77150286e-04 -2.42268969e-03 -3.12998053e-03
4.55578178e-04 2.90129334e-03 -3.05265025e-03 -1.28805637e-03
-2.08641519e-03 3.26466886e-03 -2.95106089e-03 -2.08173040e-03
-1.99576933e-03 4.53641405e-04 -2.70907651e-03 -2.34400504e-03
1.16086320e-03 -2.44718627e-03 9.25636268e-05 9.73496411e-04
-1.01899146e-03 -3.02827288e-03 9.58363991e-04 -3.27257067e-03
2.40717572e-03 1.20117664e-04 1.10580446e-03 -4.41495577e-05
-2.85318610e-03 1.54357916e-03 3.11869616e-03 6.44255488e-04
-1.31027703e-03 -1.52749463e-03 8.79097788e-04 7.01892364e-04
1.34046946e-03 -8.91715696e-04 -1.93791394e-03 -4.34041809e-04
1.12010317e-03 -2.24535773e-03 -1.76302914e-03 -1.11521804e-03
-2.68377946e-03 -2.55486579e-03 2.67607206e-03 -2.09635729e-03
4.45536774e-04 4.23340796e-04 -2.36946181e-03 1.01201690e-03
2.53369007e-03 -8.69231240e-04 -2.23573043e-05 4.58726077e-04
9.46683111e-04 -1.58690137e-03 1.31600059e-03 2.19468423e-03
-1.69886113e-03 1.71214901e-03 -1.43307843e-04 -1.10225752e-03
3.13180522e-03 1.78616366e-03 -4.65679186e-05 -1.40959187e-03
-6.96717121e-04 -5.70511795e-04 -1.54102559e-03 -2.30318774e-03
2.54784338e-03 -1.62216101e-03 3.14650533e-04 -2.94532534e-03
-1.02099183e-03 2.99499906e-03 -6.38728146e-04 -2.72372481e-03
3.22340080e-03 -1.49127806e-03 2.27723271e-03 2.73366761e-03
2.62600114e-03 2.68271845e-03 3.20440996e-03 -4.97240224e-04
-1.02938886e-03 3.26999027e-04 9.46711691e-04 1.76053529e-03
1.74157624e-03 1.49760721e-03 -3.09546776e-05 2.48821010e-03
2.15774146e-03 2.42709951e-03 -2.46135960e-03 1.82637456e-03
-3.11999000e-03 -2.49591586e-03 -3.27967643e-03 -1.17016002e-03
6.43555308e-04 3.32132494e-03 -2.58475146e-03 -7.75608991e-04
3.30572366e-03 6.71840506e-04 -2.23828160e-04 2.99876463e-03
3.10293835e-05 -1.25048554e-03 2.48837401e-03 -4.16146126e-04
-8.01149989e-04 -2.19148802e-04 -2.70171487e-03 1.73141161e-04
-2.53586681e-03 3.11773620e-03 1.13646187e-04 2.82005151e-03
-3.23787535e-04 1.52152695e-03 3.21076158e-03 -2.29426223e-04
-2.22376501e-03 -3.26833175e-03 5.72812569e-04 3.06089874e-03
8.33402446e-04 1.29480439e-03 1.32911524e-03 2.61883345e-03
-2.53178203e-03 6.48000219e-04 2.66361074e-03 -3.05172289e-03
-9.23413434e-04 -2.13261833e-03 8.54914193e-04 -1.48963137e-03
-1.95632223e-03 -7.69955339e-04 -3.29735363e-03 1.98830920e-03
1.31162966e-03 1.10320176e-03 -3.22533771e-03 2.04978790e-03
-5.25970478e-04 -1.89223525e-03 2.42309878e-03 8.27315671e-04
9.63741913e-04 8.84156208e-04 1.02529768e-03 -1.41616585e-03
6.75518531e-04 -6.47147477e-04 2.71809031e-03 2.17319001e-03
9.71910951e-04 -2.93364283e-03 2.43404706e-04 1.14709849e-03
-1.99730392e-04 3.82491737e-04 -3.08531453e-03 -2.20424891e-03
2.87708524e-03 1.51069486e-03 9.24036489e-04 -1.09619542e-03
1.36686012e-03 -2.61674239e-03 -1.52974128e-04 -2.72300956e-03
1.70241436e-03 -6.61658472e-04 -2.15324806e-03 -2.46914220e-03
1.41488796e-03 -3.25874239e-03 -2.29610526e-03 -2.22696620e-03
-2.09132349e-03 -2.79461709e-03 -3.24834906e-03 -1.12362858e-03]
shape of item_vectors: 2 (300,)
shape of token_vectors: 2 55 300
get_pretrained_i2v¶
概述¶
使用 EduNLP 项目组给定的预训练模型将给定的题目文本转成向量。
优点：简单方便。
缺点：只能使用项目中给定的模型，局限性较大。
导入功能块¶
[6]:
from EduNLP import get_pretrained_i2v
输入¶
类型：str
内容：题目文本 （text）
[7]:
items = [
"如图来自古希腊数学家希波克拉底所研究的几何图形．此图由三个半圆构成，三个半圆的直径分别为直角三角形$ABC$的斜边$BC$, 直角边$AB$, $AC$.$\bigtriangleup ABC$的三边所围成的区域记为$I$,黑色部分记为$II$, 其余部分记为$III$.在整个图形中随机取一点，此点取自$I,II,III$的概率分别记为$p_1,p_2,p_3$,则$\SIFChoice$$\FigureID{1}$"
]
模型选择与使用¶
根据题目所属学科选择预训练模型：
预训练模型名称 |
模型训练数据的所属学科 |
|---|---|
d2v_all_256 |
全学科 |
d2v_sci_256 |
理科 |
d2v_eng_256 |
英语 |
d2v_lit_256 |
文科 |
w2v_eng_300 |
英语 |
w2v_lit_300 |
文科 |
[8]:
i2v = get_pretrained_i2v("d2v_sci_256", model_dir="./d2v")
EduNLP, INFO Use pretrained t2v model d2v_sci_256
downloader, INFO http://base.ustc.edu.cn/data/model_zoo/EduNLP/d2v/general_science_256.zip is saved as d2v\general_science_256.zip
downloader, INFO file existed, skipped
注意： 默认的 EduNLP 项目存储地址为根目录(
~/.EduNLP)，模型存储地址为项目存储地址下的model文件夹。您可以通过修改下面的环境变量来修改模型存储地址：EduNLP 项目存储地址：
EDUNLPPATH = xx/xx/xx模型存储地址：
EDUNLPMODELPATH = xx/xx/xx
[9]:
item_vectors, token_vectors = i2v(items)
print(item_vectors)
print(token_vectors)
[array([-0.23861311, 0.06892798, -0.27065727, 0.16547263, 0.02818857,
-0.18185084, 0.09226187, 0.01612627, 0.0921795 , 0.3134312 ,
0.09265023, -0.22529641, -0.25788078, 0.06702194, 0.09765045,
-0.19932257, 0.08527228, -0.22684543, -0.1776405 , -0.03682012,
0.6210964 , -0.26637274, 0.08060682, -0.15860714, -0.17825642,
-0.13271384, 0.27331072, 0.18202724, 0.08430962, 0.23299456,
0.179898 , 0.1571772 , -0.1406754 , -0.19508898, -0.11265783,
0.11396482, 0.0223774 , 0.07824919, -0.2421433 , 0.06195279,
-0.04763965, -0.02037446, 0.07481094, -0.1908799 , 0.09688905,
0.3995564 , 0.28225863, 0.30547026, -0.46538818, -0.02891348,
-0.19343005, 0.01966276, -0.21590087, 0.09774096, -0.26137134,
-0.23963049, 0.34259936, 0.14825426, -0.2987728 , -0.38039675,
-0.12087625, 0.05897354, 0.06351897, 0.10188989, 0.12092843,
0.13229063, 0.12786968, -0.15378596, 0.00724137, -0.13644631,
-0.15164569, 0.11535735, -0.24394232, -0.08835315, 0.05014084,
-0.05980775, 0.03040357, -0.05804552, -0.04122322, 0.31905708,
-0.02468318, 0.06953011, -0.1299219 , 0.01482821, -0.00126122,
-0.20185567, -0.00784766, -0.28023243, -0.16416278, -0.04939609,
-0.22619021, -0.17099814, 0.1434735 , -0.13193442, -0.18329675,
-0.06873035, -0.21638975, -0.0767743 , 0.17778671, 0.0459166 ,
0.0719557 , 0.0797654 , -0.15445784, -0.20094277, 0.11860424,
0.09521067, -0.10993416, -0.01273298, -0.0857757 , -0.05475522,
-0.09463413, 0.00845256, 0.06638184, -0.22701578, 0.06599791,
0.1323833 , 0.2227748 , 0.13431212, -0.08537175, 0.14300612,
0.24459998, 0.00735889, -0.07123663, 0.24863936, 0.10320719,
-0.06399037, 0.0537433 , 0.00862593, -0.10747737, -0.01009098,
0.01707896, 0.07951383, -0.2245529 , 0.03152119, 0.19090259,
0.27611575, -0.16507478, 0.05977706, 0.09740735, 0.32154247,
-0.02540598, -0.20875612, 0.11484967, 0.12112009, -0.00937327,
-0.03855037, -0.03728763, 0.13645649, 0.42706412, 0.14456204,
-0.1542145 , 0.07858715, 0.14076898, 0.01195827, 0.16896723,
-0.0516856 , 0.05795754, 0.09602529, 0.02058077, 0.14346235,
0.3984762 , 0.06770886, -0.5524451 , -0.18779868, 0.11151859,
-0.06967582, 0.09465033, 0.2242416 , -0.17179447, 0.20837718,
0.43269685, -0.33945957, 0.00746959, -0.14856125, -0.04883511,
0.0790235 , 0.18130969, -0.06500382, -0.05761597, 0.15247819,
0.22402437, 0.33508143, -0.02544755, 0.10404763, -0.0392291 ,
0.14048643, -0.39408255, -0.04759403, -0.09290893, -0.10062248,
0.3836949 , -0.04212417, 0.04195033, -0.34143335, 0.02139966,
0.00748172, 0.09670173, 0.11287135, 0.07313446, -0.06884305,
-0.27654266, -0.02745902, 0.11782443, -0.05509072, -0.02731109,
0.02932139, 0.20647307, -0.09912065, 0.08175386, 0.04051739,
-0.13783188, 0.2178767 , 0.01360986, -0.11862064, 0.02632025,
0.01305837, -0.07418288, -0.11537156, 0.07784148, -0.02828423,
0.0152778 , -0.27535534, -0.26457086, -0.2426946 , 0.17839569,
0.41153124, -0.06237097, 0.28373018, 0.09847705, -0.2693095 ,
0.15109962, 0.02665104, 0.12224031, 0.0053689 , 0.08057593,
0.0029663 , -0.01309686, 0.04294159, -0.26014623, -0.09540065,
-0.19017759, -0.02596658, -0.21918078, -0.04269371, 0.09444954,
-0.05112423, 0.21732539, 0.2555126 , 0.06598321, -0.00912136,
0.01300732, -0.02216252, 0.16752972, 0.00181198, 0.02385568,
-0.0017939 ], dtype=float32)]
None