原文转载自「刘悦的技术博客」https://v3u.cn/a_id_189
笔者投入M1的怀抱曾经有一段时间了，俗话说得好，但闻新人笑，不见旧人哭，Intel mac早已被束之高阁，而M1 mac曾经不能用真香来形容了，几乎就是“香透满堂金玉彩，扇遮半面桃花开！”，轻抚M1 mac那滑若柔荑的秒控键盘，别说996了，就是007，咱们也能安之若素，也能够笑慰平生。好了，日常吹M1的环节完结，正所谓剑虽利,不厉一直，材虽美，不学不高。本次咱们尝试在M1 Mac os 中搭建Python3的开发环境。
个别状况下，间接Python官网(python.org)下载最新的基于arm架构的python3.9即可，然而因为向下兼容等问题，咱们尝试应用Python多版本管理软件conda，conda在业界有三大巨头别离是:Anaconda、Miniconda以及Condaforge，尽管都放出音讯要适配M1芯片，然而目前最先放出稳定版的是Condaforge，进入下载页面：https://github.com/conda-forg... 抉择mac arm64位架构:
该文件并不是安装包，而是一个shell脚本，下载胜利后，进入命令行目录:
cd ~/Downloas执行命令进行装置：
sudo bash ./Miniforge3-MacOSX-arm64.sh随后会有一些条款须要确认，这里按回车之后键入yes:
Welcome to Miniforge3 4.9.2-7 In order to continue the installation process, please review the license agreement. Please, press ENTER to continue >>> BSD 3-clause license Copyright (c) 2019-2020, conda-forge All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. Do you accept the license terms? [yes|no] [no] >>> yes装置的默认版本还是3.9，会附带装置35个根底库，这样就不必咱们本人手动装置了：
brotlipy 0.7.0 py39h46acfd9_1001 installed bzip2 1.0.8 h27ca646_4 installed ca-certificates 2020.12.5 h4653dfc_0 installed certifi 2020.12.5 py39h2804cbe_1 installed cffi 1.14.5 py39h702c04f_0 installed chardet 4.0.0 py39h2804cbe_1 installed conda 4.9.2 py39h2804cbe_0 installed conda-package-handling 1.7.2 py39h51e6412_0 installed cryptography 3.4.4 py39h6e07874_0 installed idna 2.10 pyh9f0ad1d_0 installed libcxx 11.0.1 h168391b_0 installed libffi 3.3 h9f76cd9_2 installed ncurses 6.2 h9aa5885_4 installed openssl 1.1.1j h27ca646_0 installed pip 21.0.1 pyhd8ed1ab_0 installed pycosat 0.6.3 py39h46acfd9_1006 installed pycparser 2.20 pyh9f0ad1d_2 installed pyopenssl 20.0.1 pyhd8ed1ab_0 installed pysocks 1.7.1 py39h2804cbe_3 installed python 3.9.2 hcbd9b3a_0_cpython installed python_abi 3.9 1_cp39 installed readline 8.0 hc8eb9b7_2 installed requests 2.25.1 pyhd3deb0d_0 installed ruamel_yaml 0.15.80 py39h46acfd9_1004 installed setuptools 49.6.0 py39h2804cbe_3 installed six 1.15.0 pyh9f0ad1d_0 installed sqlite 3.34.0 h6d56c25_0 installed tk 8.6.10 hf7e6567_1 installed tqdm 4.57.0 pyhd8ed1ab_0 installed tzdata 2021a he74cb21_0 installed urllib3 1.26.3 pyhd8ed1ab_0 installed wheel 0.36.2 pyhd3deb0d_0 installed xz 5.2.5 h642e427_1 installed yaml 0.2.5 h642e427_0 installed zlib 1.2.11 h31e879b_1009 installed而后编辑配置文件vim ~/.zshrc，退出如下内容(此处liuyue是笔者用户名，需改成你的Mac以后用户名)：
path=('/Users/liuyue/miniforge3/bin' $path) export PATH存盘之后执行命令：
source ~/.zshrc配置好环境变量之后，键入python3：
➜ ~ python3 Python 3.9.2 | packaged by conda-forge | (default, Feb 21 2021, 05:00:30) [Clang 11.0.1 ] on darwin Type "help", "copyright", "credits" or "license" for more information. >>>能够看到曾经应用conda装置的python版本了。
这里简略介绍一下conda命令：
conda info 能够查看以后conda的根本信息内核，平台，下载源以及目录地位：
➜ ~ conda info active environment : None user config file : /Users/liuyue/.condarc populated config files : /Users/liuyue/miniforge3/.condarc conda version : 4.9.2 conda-build version : not installed python version : 3.9.2.final.0 virtual packages : __osx=11.2.2=0 __unix=0=0 __archspec=1=arm64 base environment : /Users/liuyue/miniforge3 (read only) channel URLs : https://conda.anaconda.org/conda-forge/osx-arm64 https://conda.anaconda.org/conda-forge/noarch package cache : /Users/liuyue/miniforge3/pkgs /Users/liuyue/.conda/pkgs envs directories : /Users/liuyue/.conda/envs /Users/liuyue/miniforge3/envs platform : osx-arm64 user-agent : conda/4.9.2 requests/2.25.1 CPython/3.9.2 Darwin/20.3.0 OSX/11.2.2 UID:GID : 502:20 netrc file : None offline mode : False因为一些家喻户晓的学术问题，咱们须要配置一下国内下载源：
conda config --add channels https://mirrors.ustc.edu.cn/anaconda/pkgs/main/ conda config --add channels https://mirrors.ustc.edu.cn/anaconda/pkgs/free/ conda config --add channels https://mirrors.ustc.edu.cn/anaconda/cloud/conda-forge/ conda config --add channels https://mirrors.ustc.edu.cn/anaconda/cloud/msys2/ conda config --add channels https://mirrors.ustc.edu.cn/anaconda/cloud/bioconda/ conda config --add channels https://mirrors.ustc.edu.cn/anaconda/cloud/menpo/ conda config --set show_channel_urls yes随后查看以后下载源：
conda config --show能够看到国内源曾经被增加进去了：
channel_priority: flexible channels: - https://mirrors.ustc.edu.cn/anaconda/cloud/menpo/ - https://mirrors.ustc.edu.cn/anaconda/cloud/bioconda/ - https://mirrors.ustc.edu.cn/anaconda/cloud/msys2/ - https://mirrors.ustc.edu.cn/anaconda/cloud/conda-forge/ - https://mirrors.ustc.edu.cn/anaconda/pkgs/free/ - https://mirrors.ustc.edu.cn/anaconda/pkgs/main/ - defaults - conda-forge client_ssl_cert: None其余的一些conda常用命令：
1. conda --version #查看conda版本，验证是否装置2. conda update conda #更新至最新版本，也会更新其它相干包
3. conda update --all #更新所有包
4. conda update package\_name #更新指定的包
5. conda create -n env\_name package\_name #创立名为env\_name的新环境，并在该环境下装置名为package\_name 的包，能够指定新环境的版本号，例如：conda create -n python2 python=python2.7 numpy pandas，创立了python2环境，python版本为2.7，同时还装置了numpy pandas包
6. source activate env\_name #切换至env\_name环境
7. source deactivate #退出环境
8. conda info -e #显示所有曾经创立的环境
9. conda create --name new\_env\_name --clone old\_env\_name #复制old\_env\_name为new\_env\_name
10. conda remove --name env\_name –all #删除环境
11. conda list #查看所有曾经装置的包
12. conda install package\_name #在以后环境中安装包
13. conda install --name env\_name package\_name #在指定环境中安装包
14. conda remove -- name env\_name package #删除指定环境中的包
15. conda remove package #删除以后环境中的包
16. conda create -n tensorflow\_env tensorflow
conda activate tensorflow\_env #conda 装置tensorflow的CPU版本
17. conda create -n tensorflow\_gpuenv tensorflow-gpu
conda activate tensorflow\_gpuenv #conda装置tensorflow的GPU版本
18. conda env remove -n env\_name #采纳第10条的办法删除环境失败时，可采纳这种办法
接着咱们来尝试集成深度学习框架Tensorflow，因为目前默认是3.9，咱们应用conda创立一个3.8的虚构开发环境：
sudo conda create -n py38 python=3.8装置胜利后，输出命令：
conda info -e就能够查看以后conda装置的所有版本：
➜ ~ conda info -e # conda environments: # base * /Users/liuyue/miniforge3 py38 /Users/liuyue/miniforge3/envs/py38能够看到一个默认的3.9环境，和新装的3.8环境，星号代表以后所处的环境，这里咱们切换到3.8：
conda activate py38此时环境曾经切换到3.8：
(py38) ➜ ~ conda activate py38 (py38) ➜ ~ conda info -e # conda environments: # base /Users/liuyue/miniforge3 py38 * /Users/liuyue/miniforge3/envs/py38 (py38) ➜ ~上面开启深度学习框架Tensorflow之旅，因为苹果对m1芯片独自做了适配，所以不能用以前的pip形式间接进行装置，须要独自下载文件：https://github.91chifun.worke...://github.com/apple/tensorflow\_macos/releases/download/v0.1alpha1/tensorflow\_macos-0.1alpha1.tar.gz
解压文件：
tar -zxvf tensorflow_macos-0.1alpha1.tar.gz解压后进入目录(肯定要进入arm64的文件内)：
cd tensorflow_macos/arm64执行命令利用下载的arm64内核安装包进行装置：
pip install --force pip==20.2.4 wheel setuptools cached-property six pip install --upgrade --no-dependencies --force numpy-1.18.5-cp38-cp38-macosx_11_0_arm64.whl grpcio-1.33.2-cp38-cp38-macosx_11_0_arm64.whl h5py-2.10.0-cp38-cp38-macosx_11_0_arm64.whl tensorflow_addons-0.11.2+mlcompute-cp38-cp38-macosx_11_0_arm64.whl pip install absl-py astunparse flatbuffers gast google_pasta keras_preprocessing opt_einsum protobuf tensorflow_estimator termcolor typing_extensions wrapt wheel tensorboard typeguard pip install --upgrade --force --no-dependencies tensorflow_macos-0.1a1-cp38-cp38-macosx_11_0_arm64.whl装置胜利后，测试一下：
(py38) ➜ arm64 python Python 3.8.8 | packaged by conda-forge | (default, Feb 20 2021, 15:50:57) [Clang 11.0.1 ] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> import tensorflow >>>没有任何问题。
上面咱们来测试一下M1通过Tensorflow训练模型的效率，还记得衣香鬓影的“机械姬”吗：人工智能不过尔尔，基于Python3深度学习库Keras/TensorFlow打造属于本人的聊天机器人(ChatRobot)
编写my\_chat.py:
import nltk import ssl from nltk.stem.lancaster import LancasterStemmer stemmer = LancasterStemmer() import numpy as np from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense, Activation, Dropout from tensorflow.keras.optimizers import SGD import pandas as pd import pickle import random words = [] classes = [] documents = [] ignore_words = ['?'] # loop through each sentence in our intents patterns intents = {"intents": [ {"tag": "打招呼", "patterns": ["你好", "您好", "请问", "有人吗", "徒弟","不好意思","美女","帅哥","靓妹"], "responses": ["您好", "又是您啊", "吃了么您内","您有事吗"], "context": [""] }, {"tag": "辞别", "patterns": ["再见", "拜拜", "88", "回见", "回头见"], "responses": ["再见", "一路顺风", "下次见", "拜拜了您内"], "context": [""] }, ] } for intent in intents['intents']: for pattern in intent['patterns']: # tokenize each word in the sentence w = nltk.word_tokenize(pattern) # add to our words list words.extend(w) # add to documents in our corpus documents.append((w, intent['tag'])) # add to our classes list if intent['tag'] not in classes: classes.append(intent['tag']) # stem and lower each word and remove duplicates words = [stemmer.stem(w.lower()) for w in words if w not in ignore_words] words = sorted(list(set(words))) # sort classes classes = sorted(list(set(classes))) # documents = combination between patterns and intents # print (len(documents), "documents") # # classes = intents # print (len(classes), "语境", classes) # # words = all words, vocabulary # print (len(words), "词数", words) # create our training data training = [] # create an empty array for our output output_empty = [0] * len(classes) # training set, bag of words for each sentence for doc in documents: # initialize our bag of words bag = [] pattern_words = doc[0] pattern_words = [stemmer.stem(word.lower()) for word in pattern_words] for w in words: bag.append(1) if w in pattern_words else bag.append(0) output_row = list(output_empty) output_row[classes.index(doc[1])] = 1 training.append([bag, output_row]) random.shuffle(training) training = np.array(training) train_x = list(training[:,0]) train_y = list(training[:,1]) model = Sequential() model.add(Dense(128, input_shape=(len(train_x[0]),), activation='relu')) model.add(Dropout(0.5)) model.add(Dense(64, activation='relu')) model.add(Dropout(0.5)) model.add(Dense(len(train_y[0]), activation='softmax')) sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) model.fit(np.array(train_x), np.array(train_y), epochs=200, batch_size=5, verbose=1) def clean_up_sentence(sentence): # tokenize the pattern - split words into array sentence_words = nltk.word_tokenize(sentence) # stem each word - create short form for word sentence_words = [stemmer.stem(word.lower()) for word in sentence_words] return sentence_words # return bag of words array: 0 or 1 for each word in the bag that exists in the sentence def bow(sentence, words, show_details=True): # tokenize the pattern sentence_words = clean_up_sentence(sentence) # bag of words - matrix of N words, vocabulary matrix bag = [0]*len(words) for s in sentence_words: for i,w in enumerate(words): if w == s: # assign 1 if current word is in the vocabulary position bag[i] = 1 if show_details: print ("found in bag: %s" % w) return(np.array(bag)) def classify_local(sentence): ERROR_THRESHOLD = 0.25 # generate probabilities from the model input_data = pd.DataFrame([bow(sentence, words)], dtype=float, index=['input']) results = model.predict([input_data])[0] # filter out predictions below a threshold, and provide intent index results = [[i,r] for i,r in enumerate(results) if r>ERROR_THRESHOLD] # sort by strength of probability results.sort(key=lambda x: x[1], reverse=True) return_list = [] for r in results: return_list.append((classes[r[0]], str(r[1]))) # return tuple of intent and probability return return_list p = bow("你好", words) print (p) print(classify_local('请问'))返回后果：
(py38) ➜ mytornado git:(master) ✗ python3 test_mychat.py 2021-03-03 22:43:21.059383: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:116] None of the MLIR optimization passes are enabled (registered 2) 2021-03-03 22:43:21.059529: W tensorflow/core/platform/profile_utils/cpu_utils.cc:126] Failed to get CPU frequency: 0 Hz Epoch 1/200 3/3 [==============================] - 0s 570us/step - loss: 0.6966 - accuracy: 0.5750 Epoch 2/200 3/3 [==============================] - 0s 482us/step - loss: 0.6913 - accuracy: 0.4857 Epoch 3/200 3/3 [==============================] - 0s 454us/step - loss: 0.6795 - accuracy: 0.4750 Epoch 4/200 3/3 [==============================] - 0s 434us/step - loss: 0.6913 - accuracy: 0.4750 Epoch 5/200 3/3 [==============================] - 0s 417us/step - loss: 0.6563 - accuracy: 0.5107 Epoch 6/200 3/3 [==============================] - 0s 454us/step - loss: 0.6775 - accuracy: 0.5714 Epoch 7/200 3/3 [==============================] - 0s 418us/step - loss: 0.6582 - accuracy: 0.6964 Epoch 8/200 3/3 [==============================] - 0s 487us/step - loss: 0.6418 - accuracy: 0.8071 Epoch 9/200 3/3 [==============================] - 0s 504us/step - loss: 0.6055 - accuracy: 0.6964 Epoch 10/200 3/3 [==============================] - 0s 457us/step - loss: 0.5933 - accuracy: 0.6964 Epoch 11/200 3/3 [==============================] - 0s 392us/step - loss: 0.6679 - accuracy: 0.5714 Epoch 12/200 3/3 [==============================] - 0s 427us/step - loss: 0.6060 - accuracy: 0.7464 Epoch 13/200 3/3 [==============================] - 0s 425us/step - loss: 0.6677 - accuracy: 0.5964 Epoch 14/200 3/3 [==============================] - 0s 420us/step - loss: 0.6208 - accuracy: 0.6214 Epoch 15/200 3/3 [==============================] - 0s 401us/step - loss: 0.6315 - accuracy: 0.6714 Epoch 16/200 3/3 [==============================] - 0s 401us/step - loss: 0.6718 - accuracy: 0.6464 Epoch 17/200 3/3 [==============================] - 0s 386us/step - loss: 0.6407 - accuracy: 0.6714 Epoch 18/200 3/3 [==============================] - 0s 505us/step - loss: 0.6031 - accuracy: 0.6464 Epoch 19/200 3/3 [==============================] - 0s 407us/step - loss: 0.6245 - accuracy: 0.6214 Epoch 20/200 3/3 [==============================] - 0s 422us/step - loss: 0.5805 - accuracy: 0.6964 Epoch 21/200 3/3 [==============================] - 0s 379us/step - loss: 0.6923 - accuracy: 0.5464 Epoch 22/200 3/3 [==============================] - 0s 396us/step - loss: 0.6383 - accuracy: 0.5714 Epoch 23/200 3/3 [==============================] - 0s 427us/step - loss: 0.6628 - accuracy: 0.5714 Epoch 24/200 3/3 [==============================] - 0s 579us/step - loss: 0.6361 - accuracy: 0.5964 Epoch 25/200 3/3 [==============================] - 0s 378us/step - loss: 0.5632 - accuracy: 0.7214 Epoch 26/200 3/3 [==============================] - 0s 387us/step - loss: 0.6851 - accuracy: 0.5214 Epoch 27/200 3/3 [==============================] - 0s 393us/step - loss: 0.6012 - accuracy: 0.6214 Epoch 28/200 3/3 [==============================] - 0s 392us/step - loss: 0.6470 - accuracy: 0.5964 Epoch 29/200 3/3 [==============================] - 0s 348us/step - loss: 0.6346 - accuracy: 0.6214 Epoch 30/200 3/3 [==============================] - 0s 362us/step - loss: 0.6350 - accuracy: 0.4964 Epoch 31/200 3/3 [==============================] - 0s 369us/step - loss: 0.5842 - accuracy: 0.5964 Epoch 32/200 3/3 [==============================] - 0s 481us/step - loss: 0.5279 - accuracy: 0.7214 Epoch 33/200 3/3 [==============================] - 0s 439us/step - loss: 0.5956 - accuracy: 0.7321 Epoch 34/200 3/3 [==============================] - 0s 355us/step - loss: 0.5570 - accuracy: 0.6964 Epoch 35/200 3/3 [==============================] - 0s 385us/step - loss: 0.5546 - accuracy: 0.8071 Epoch 36/200 3/3 [==============================] - 0s 375us/step - loss: 0.5616 - accuracy: 0.6714 Epoch 37/200 3/3 [==============================] - 0s 379us/step - loss: 0.6955 - accuracy: 0.6464 Epoch 38/200 3/3 [==============================] - 0s 389us/step - loss: 0.6089 - accuracy: 0.7321 Epoch 39/200 3/3 [==============================] - 0s 375us/step - loss: 0.5377 - accuracy: 0.6714 Epoch 40/200 3/3 [==============================] - 0s 392us/step - loss: 0.6224 - accuracy: 0.7179 Epoch 41/200 3/3 [==============================] - 0s 379us/step - loss: 0.6234 - accuracy: 0.5464 Epoch 42/200 3/3 [==============================] - 0s 411us/step - loss: 0.5224 - accuracy: 0.8321 Epoch 43/200 3/3 [==============================] - 0s 386us/step - loss: 0.5848 - accuracy: 0.5964 Epoch 44/200 3/3 [==============================] - 0s 401us/step - loss: 0.4620 - accuracy: 0.8679 Epoch 45/200 3/3 [==============================] - 0s 365us/step - loss: 0.4664 - accuracy: 0.8071 Epoch 46/200 3/3 [==============================] - 0s 367us/step - loss: 0.5904 - accuracy: 0.7679 Epoch 47/200 3/3 [==============================] - 0s 359us/step - loss: 0.5111 - accuracy: 0.7929 Epoch 48/200 3/3 [==============================] - 0s 363us/step - loss: 0.4712 - accuracy: 0.8679 Epoch 49/200 3/3 [==============================] - 0s 401us/step - loss: 0.5601 - accuracy: 0.8071 Epoch 50/200 3/3 [==============================] - 0s 429us/step - loss: 0.4884 - accuracy: 0.7929 Epoch 51/200 3/3 [==============================] - 0s 377us/step - loss: 0.5137 - accuracy: 0.8286 Epoch 52/200 3/3 [==============================] - 0s 368us/step - loss: 0.5475 - accuracy: 0.8286 Epoch 53/200 3/3 [==============================] - 0s 592us/step - loss: 0.4077 - accuracy: 0.8536 Epoch 54/200 3/3 [==============================] - 0s 400us/step - loss: 0.5367 - accuracy: 0.8179 Epoch 55/200 3/3 [==============================] - 0s 399us/step - loss: 0.5288 - accuracy: 0.8429 Epoch 56/200 3/3 [==============================] - 0s 367us/step - loss: 0.5775 - accuracy: 0.6964 Epoch 57/200 3/3 [==============================] - 0s 372us/step - loss: 0.5680 - accuracy: 0.6821 Epoch 58/200 3/3 [==============================] - 0s 360us/step - loss: 0.5164 - accuracy: 0.7321 Epoch 59/200 3/3 [==============================] - 0s 364us/step - loss: 0.5334 - accuracy: 0.6571 Epoch 60/200 3/3 [==============================] - 0s 358us/step - loss: 0.3858 - accuracy: 0.9036 Epoch 61/200 3/3 [==============================] - 0s 356us/step - loss: 0.4313 - accuracy: 0.8679 Epoch 62/200 3/3 [==============================] - 0s 373us/step - loss: 0.5017 - accuracy: 0.8429 Epoch 63/200 3/3 [==============================] - 0s 346us/step - loss: 0.4649 - accuracy: 0.8429 Epoch 64/200 3/3 [==============================] - 0s 397us/step - loss: 0.3804 - accuracy: 0.8893 Epoch 65/200 3/3 [==============================] - 0s 361us/step - loss: 0.5030 - accuracy: 0.7929 Epoch 66/200 3/3 [==============================] - 0s 372us/step - loss: 0.3958 - accuracy: 0.9286 Epoch 67/200 3/3 [==============================] - 0s 345us/step - loss: 0.4240 - accuracy: 0.8536 Epoch 68/200 3/3 [==============================] - 0s 360us/step - loss: 0.4651 - accuracy: 0.7929 Epoch 69/200 3/3 [==============================] - 0s 376us/step - loss: 0.4687 - accuracy: 0.7571 Epoch 70/200 3/3 [==============================] - 0s 398us/step - loss: 0.4660 - accuracy: 0.8429 Epoch 71/200 3/3 [==============================] - 0s 368us/step - loss: 0.3960 - accuracy: 0.9393 Epoch 72/200 3/3 [==============================] - 0s 355us/step - loss: 0.5523 - accuracy: 0.6071 Epoch 73/200 3/3 [==============================] - 0s 361us/step - loss: 0.5266 - accuracy: 0.7821 Epoch 74/200 3/3 [==============================] - 0s 371us/step - loss: 0.4245 - accuracy: 0.9643 Epoch 75/200 3/3 [==============================] - 0s 367us/step - loss: 0.5024 - accuracy: 0.7786 Epoch 76/200 3/3 [==============================] - 0s 453us/step - loss: 0.3419 - accuracy: 0.9393 Epoch 77/200 3/3 [==============================] - 0s 405us/step - loss: 0.4930 - accuracy: 0.7429 Epoch 78/200 3/3 [==============================] - 0s 672us/step - loss: 0.3443 - accuracy: 0.9036 Epoch 79/200 3/3 [==============================] - 0s 386us/step - loss: 0.3864 - accuracy: 0.8893 Epoch 80/200 3/3 [==============================] - 0s 386us/step - loss: 0.3863 - accuracy: 0.9286 Epoch 81/200 3/3 [==============================] - 0s 391us/step - loss: 0.2771 - accuracy: 0.8679 Epoch 82/200 3/3 [==============================] - 0s 370us/step - loss: 0.6083 - accuracy: 0.5571 Epoch 83/200 3/3 [==============================] - 0s 387us/step - loss: 0.2801 - accuracy: 0.9393 Epoch 84/200 3/3 [==============================] - 0s 357us/step - loss: 0.2483 - accuracy: 0.9286 Epoch 85/200 3/3 [==============================] - 0s 355us/step - loss: 0.2511 - accuracy: 0.9643 Epoch 86/200 3/3 [==============================] - 0s 339us/step - loss: 0.3410 - accuracy: 0.8893 Epoch 87/200 3/3 [==============================] - 0s 361us/step - loss: 0.3432 - accuracy: 0.9036 Epoch 88/200 3/3 [==============================] - 0s 347us/step - loss: 0.3819 - accuracy: 0.8893 Epoch 89/200 3/3 [==============================] - 0s 361us/step - loss: 0.5142 - accuracy: 0.7179 Epoch 90/200 3/3 [==============================] - 0s 502us/step - loss: 0.3055 - accuracy: 0.9393 Epoch 91/200 3/3 [==============================] - 0s 377us/step - loss: 0.3144 - accuracy: 0.8536 Epoch 92/200 3/3 [==============================] - 0s 376us/step - loss: 0.3712 - accuracy: 0.9036 Epoch 93/200 3/3 [==============================] - 0s 389us/step - loss: 0.1974 - accuracy: 0.9393 Epoch 94/200 3/3 [==============================] - 0s 365us/step - loss: 0.3128 - accuracy: 0.9393 Epoch 95/200 3/3 [==============================] - 0s 376us/step - loss: 0.2194 - accuracy: 1.0000 Epoch 96/200 3/3 [==============================] - 0s 377us/step - loss: 0.1994 - accuracy: 1.0000 Epoch 97/200 3/3 [==============================] - 0s 360us/step - loss: 0.1734 - accuracy: 0.9643 Epoch 98/200 3/3 [==============================] - 0s 367us/step - loss: 0.1786 - accuracy: 1.0000 Epoch 99/200 3/3 [==============================] - 0s 358us/step - loss: 0.4158 - accuracy: 0.8286 Epoch 100/200 3/3 [==============================] - 0s 354us/step - loss: 0.3131 - accuracy: 0.7571 Epoch 101/200 3/3 [==============================] - 0s 350us/step - loss: 0.1953 - accuracy: 0.8893 Epoch 102/200 3/3 [==============================] - 0s 403us/step - loss: 0.2577 - accuracy: 0.8429 Epoch 103/200 3/3 [==============================] - 0s 417us/step - loss: 0.2648 - accuracy: 0.8893 Epoch 104/200 3/3 [==============================] - 0s 377us/step - loss: 0.2901 - accuracy: 0.8286 Epoch 105/200 3/3 [==============================] - 0s 383us/step - loss: 0.2822 - accuracy: 0.9393 Epoch 106/200 3/3 [==============================] - 0s 381us/step - loss: 0.2837 - accuracy: 0.9036 Epoch 107/200 3/3 [==============================] - 0s 382us/step - loss: 0.3064 - accuracy: 0.8536 Epoch 108/200 3/3 [==============================] - 0s 352us/step - loss: 0.3376 - accuracy: 0.9036 Epoch 109/200 3/3 [==============================] - 0s 376us/step - loss: 0.3412 - accuracy: 0.8536 Epoch 110/200 3/3 [==============================] - 0s 363us/step - loss: 0.1718 - accuracy: 1.0000 Epoch 111/200 3/3 [==============================] - 0s 347us/step - loss: 0.1899 - accuracy: 0.8786 Epoch 112/200 3/3 [==============================] - 0s 363us/step - loss: 0.2352 - accuracy: 0.8286 Epoch 113/200 3/3 [==============================] - 0s 373us/step - loss: 0.1378 - accuracy: 1.0000 Epoch 114/200 3/3 [==============================] - 0s 353us/step - loss: 0.4288 - accuracy: 0.7071 Epoch 115/200 3/3 [==============================] - 0s 456us/step - loss: 0.4202 - accuracy: 0.6821 Epoch 116/200 3/3 [==============================] - 0s 382us/step - loss: 0.2962 - accuracy: 0.8893 Epoch 117/200 3/3 [==============================] - 0s 394us/step - loss: 0.2571 - accuracy: 0.8893 Epoch 118/200 3/3 [==============================] - 0s 365us/step - loss: 0.2697 - accuracy: 1.0000 Epoch 119/200 3/3 [==============================] - 0s 358us/step - loss: 0.3102 - accuracy: 0.9036 Epoch 120/200 3/3 [==============================] - 0s 367us/step - loss: 0.2928 - accuracy: 0.8286 Epoch 121/200 3/3 [==============================] - 0s 374us/step - loss: 0.3157 - accuracy: 0.8286 Epoch 122/200 3/3 [==============================] - 0s 381us/step - loss: 0.3920 - accuracy: 0.7786 Epoch 123/200 3/3 [==============================] - 0s 335us/step - loss: 0.2090 - accuracy: 0.9036 Epoch 124/200 3/3 [==============================] - 0s 368us/step - loss: 0.5079 - accuracy: 0.7786 Epoch 125/200 3/3 [==============================] - 0s 337us/step - loss: 0.1900 - accuracy: 0.9393 Epoch 126/200 3/3 [==============================] - 0s 339us/step - loss: 0.2047 - accuracy: 0.9643 Epoch 127/200 3/3 [==============================] - 0s 479us/step - loss: 0.3705 - accuracy: 0.7679 Epoch 128/200 3/3 [==============================] - 0s 390us/step - loss: 0.1850 - accuracy: 0.9036 Epoch 129/200 3/3 [==============================] - 0s 642us/step - loss: 0.1594 - accuracy: 0.9393 Epoch 130/200 3/3 [==============================] - 0s 373us/step - loss: 0.2010 - accuracy: 0.8893 Epoch 131/200 3/3 [==============================] - 0s 369us/step - loss: 0.0849 - accuracy: 1.0000 Epoch 132/200 3/3 [==============================] - 0s 349us/step - loss: 0.1145 - accuracy: 1.0000 Epoch 133/200 3/3 [==============================] - 0s 360us/step - loss: 0.1796 - accuracy: 1.0000 Epoch 134/200 3/3 [==============================] - 0s 371us/step - loss: 0.2363 - accuracy: 0.8536 Epoch 135/200 3/3 [==============================] - 0s 386us/step - loss: 0.1922 - accuracy: 0.9393 Epoch 136/200 3/3 [==============================] - 0s 369us/step - loss: 0.3595 - accuracy: 0.7679 Epoch 137/200 3/3 [==============================] - 0s 369us/step - loss: 0.1506 - accuracy: 0.8893 Epoch 138/200 3/3 [==============================] - 0s 377us/step - loss: 0.2471 - accuracy: 0.8536 Epoch 139/200 3/3 [==============================] - 0s 417us/step - loss: 0.1768 - accuracy: 0.8536 Epoch 140/200 3/3 [==============================] - 0s 400us/step - loss: 0.2112 - accuracy: 0.9393 Epoch 141/200 3/3 [==============================] - 0s 377us/step - loss: 0.3652 - accuracy: 0.7179 Epoch 142/200 3/3 [==============================] - 0s 364us/step - loss: 0.3007 - accuracy: 0.8429 Epoch 143/200 3/3 [==============================] - 0s 361us/step - loss: 0.0518 - accuracy: 1.0000 Epoch 144/200 3/3 [==============================] - 0s 373us/step - loss: 0.2144 - accuracy: 0.8286 Epoch 145/200 3/3 [==============================] - 0s 353us/step - loss: 0.0888 - accuracy: 1.0000 Epoch 146/200 3/3 [==============================] - 0s 361us/step - loss: 0.1267 - accuracy: 1.0000 Epoch 147/200 3/3 [==============================] - 0s 341us/step - loss: 0.0321 - accuracy: 1.0000 Epoch 148/200 3/3 [==============================] - 0s 358us/step - loss: 0.0860 - accuracy: 1.0000 Epoch 149/200 3/3 [==============================] - 0s 375us/step - loss: 0.2151 - accuracy: 0.8893 Epoch 150/200 3/3 [==============================] - 0s 351us/step - loss: 0.1592 - accuracy: 1.0000 Epoch 151/200 3/3 [==============================] - 0s 531us/step - loss: 0.1450 - accuracy: 0.8786 Epoch 152/200 3/3 [==============================] - 0s 392us/step - loss: 0.1813 - accuracy: 0.9036 Epoch 153/200 3/3 [==============================] - 0s 404us/step - loss: 0.1197 - accuracy: 1.0000 Epoch 154/200 3/3 [==============================] - 0s 367us/step - loss: 0.0930 - accuracy: 1.0000 Epoch 155/200 3/3 [==============================] - 0s 580us/step - loss: 0.2587 - accuracy: 0.8893 Epoch 156/200 3/3 [==============================] - 0s 383us/step - loss: 0.0742 - accuracy: 1.0000 Epoch 157/200 3/3 [==============================] - 0s 353us/step - loss: 0.1197 - accuracy: 0.9643 Epoch 158/200 3/3 [==============================] - 0s 371us/step - loss: 0.1716 - accuracy: 0.8536 Epoch 159/200 3/3 [==============================] - 0s 337us/step - loss: 0.1300 - accuracy: 0.9643 Epoch 160/200 3/3 [==============================] - 0s 347us/step - loss: 0.1439 - accuracy: 0.9393 Epoch 161/200 3/3 [==============================] - 0s 366us/step - loss: 0.2597 - accuracy: 0.9393 Epoch 162/200 3/3 [==============================] - 0s 345us/step - loss: 0.1605 - accuracy: 0.8893 Epoch 163/200 3/3 [==============================] - 0s 468us/step - loss: 0.0437 - accuracy: 1.0000 Epoch 164/200 3/3 [==============================] - 0s 372us/step - loss: 0.0376 - accuracy: 1.0000 Epoch 165/200 3/3 [==============================] - 0s 391us/step - loss: 0.0474 - accuracy: 1.0000 Epoch 166/200 3/3 [==============================] - 0s 378us/step - loss: 0.3225 - accuracy: 0.7786 Epoch 167/200 3/3 [==============================] - 0s 368us/step - loss: 0.0770 - accuracy: 1.0000 Epoch 168/200 3/3 [==============================] - 0s 367us/step - loss: 0.5629 - accuracy: 0.7786 Epoch 169/200 3/3 [==============================] - 0s 359us/step - loss: 0.0177 - accuracy: 1.0000 Epoch 170/200 3/3 [==============================] - 0s 370us/step - loss: 0.1167 - accuracy: 1.0000 Epoch 171/200 3/3 [==============================] - 0s 349us/step - loss: 0.1313 - accuracy: 1.0000 Epoch 172/200 3/3 [==============================] - 0s 337us/step - loss: 0.0852 - accuracy: 0.9393 Epoch 173/200 3/3 [==============================] - 0s 375us/step - loss: 0.0545 - accuracy: 1.0000 Epoch 174/200 3/3 [==============================] - 0s 354us/step - loss: 0.0674 - accuracy: 0.9643 Epoch 175/200 3/3 [==============================] - 0s 355us/step - loss: 0.0911 - accuracy: 1.0000 Epoch 176/200 3/3 [==============================] - 0s 404us/step - loss: 0.0980 - accuracy: 0.9393 Epoch 177/200 3/3 [==============================] - 0s 396us/step - loss: 0.0465 - accuracy: 1.0000 Epoch 178/200 3/3 [==============================] - 0s 403us/step - loss: 0.1117 - accuracy: 0.9393 Epoch 179/200 3/3 [==============================] - 0s 373us/step - loss: 0.0415 - accuracy: 1.0000 Epoch 180/200 3/3 [==============================] - 0s 369us/step - loss: 0.0825 - accuracy: 1.0000 Epoch 181/200 3/3 [==============================] - 0s 425us/step - loss: 0.0378 - accuracy: 1.0000 Epoch 182/200 3/3 [==============================] - 0s 381us/step - loss: 0.1155 - accuracy: 0.9393 Epoch 183/200 3/3 [==============================] - 0s 354us/step - loss: 0.0207 - accuracy: 1.0000 Epoch 184/200 3/3 [==============================] - 0s 346us/step - loss: 0.0344 - accuracy: 1.0000 Epoch 185/200 3/3 [==============================] - 0s 379us/step - loss: 0.0984 - accuracy: 0.9393 Epoch 186/200 3/3 [==============================] - 0s 360us/step - loss: 0.1508 - accuracy: 0.8536 Epoch 187/200 3/3 [==============================] - 0s 361us/step - loss: 0.0463 - accuracy: 1.0000 Epoch 188/200 3/3 [==============================] - 0s 358us/step - loss: 0.0476 - accuracy: 0.9643 Epoch 189/200 3/3 [==============================] - 0s 379us/step - loss: 0.1592 - accuracy: 1.0000 Epoch 190/200 3/3 [==============================] - 0s 387us/step - loss: 0.0071 - accuracy: 1.0000 Epoch 191/200 3/3 [==============================] - 0s 405us/step - loss: 0.0527 - accuracy: 1.0000 Epoch 192/200 3/3 [==============================] - 0s 401us/step - loss: 0.0874 - accuracy: 0.9393 Epoch 193/200 3/3 [==============================] - 0s 355us/step - loss: 0.0199 - accuracy: 1.0000 Epoch 194/200 3/3 [==============================] - 0s 373us/step - loss: 0.1299 - accuracy: 0.9643 Epoch 195/200 3/3 [==============================] - 0s 360us/step - loss: 0.0929 - accuracy: 1.0000 Epoch 196/200 3/3 [==============================] - 0s 380us/step - loss: 0.0265 - accuracy: 1.0000 Epoch 197/200 3/3 [==============================] - 0s 358us/step - loss: 0.0843 - accuracy: 1.0000 Epoch 198/200 3/3 [==============================] - 0s 354us/step - loss: 0.0925 - accuracy: 1.0000 Epoch 199/200 3/3 [==============================] - 0s 327us/step - loss: 0.0770 - accuracy: 1.0000 Epoch 200/200 3/3 [==============================] - 0s 561us/step - loss: 0.0311 - accuracy: 1.0000 found in bag: 你好 [0 0 1 0 0 0 0 0 0 0 0 0 0 0] found in bag: 请问 [('打招呼', '0.998965')]霎时执行结束，秒杀intel芯片的mac，怎一个香字了得！
接下来，尝试装置另外一个在业界名声煊赫的深度学习框架Pytorch！
因为以后arm64架构的只反对3.9版本，所以咱们来创立一个虚拟空间：
sudo conda create -n pytorch numpy matplotlib pandas python=3.9这里提前将须要的根底库都一一装置，因为如果不在创立虚拟空间时提前装置，之后应用pip是装置不上的，装置胜利后，激活环境：
(pytorch) ➜ conda activate pytorch (pytorch) ➜随后下载arm64版本的pytorch安装包：https://github.com/wizyoung/A...\_11\_0\_arm64.whl
下载胜利后，执行装置命令：
sudo pip install torch-1.8.0a0-cp39-cp39-macosx_11_0_arm64.whl让咱们来试试Pytorch在M1芯片加持后的性能，编写test\_torch.py:
from tqdm import tqdm import torch @torch.jit.script def foo(): x = torch.ones((1024 * 12, 1024 * 12), dtype=torch.float32) y = torch.ones((1024 * 12, 1024 * 12), dtype=torch.float32) z = x + y return z if __name__ == '__main__': z0 = None for _ in tqdm(range(10000000000)): zz = foo() if z0 is None: z0 = zz else: z0 += zz后果仍然秒杀Intel mac，达到了320 it/s，你敢信？
最初，有没有arm64架构的编辑器呢？答案是有的，vscode值得领有，下载地址：https://code.visualstudio.com... 肯定要抉择arm64版的：
解压后间接运行即可，能够在插件商店抉择Python和Code Runner，即可开启M1的Python代码编写之旅。
结语：M1芯片的Mac和Python3，几乎就是金风玉露，绝配天成。只有撩开M1和开发者们之间的那一层帷幔，期待咱们的，就是纵享丝滑的开发感触，还等什么？犹豫只会败北，是时候焚烧灵魂，献出钱包了。
原文转载自「刘悦的技术博客」 https://v3u.cn/a_id_189