[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-10750":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":16,"stars30d":16,"stars90d":16,"forks30d":16,"starsTrendScore":16,"compositeScore":17,"rankGlobal":10,"rankLanguage":10,"license":18,"archived":19,"fork":20,"defaultBranch":21,"hasWiki":19,"hasPages":20,"topics":22,"createdAt":10,"pushedAt":10,"updatedAt":39,"readmeContent":40,"aiSummary":41,"trendingCount":16,"starSnapshotCount":16,"syncStatus":42,"lastSyncTime":43,"discoverSource":44},10750,"NLP-Models-Tensorflow","mesolitica\u002FNLP-Models-Tensorflow","mesolitica","Gathers machine learning and Tensorflow deep learning models for NLP problems, 1.13 \u003C Tensorflow \u003C 2.0","",null,"Jupyter Notebook",1779,713,1,8,0,56.56,"MIT License",true,false,"master",[23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38],"attention","chatbot","deep-learning","dnc-seq2seq","embedded","language-detection","lstm","lstm-seq2seq-tf","luong-api","machine-learning","neural-machine-translation","nlp","optical-character-recognition","pos-tagging","speech-to-text","summarization","2026-06-12 04:00:52","\u003Cp align=\"center\">\n    \u003Ca href=\"#readme\">\n        \u003Cimg alt=\"logo\" width=\"40%\" src=\"nlp-tf.png\">\n    \u003C\u002Fa>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fhuseinzol05\u002FNLP-Models-Tensorflow\u002Fblob\u002Fmaster\u002FLICENSE\">\u003Cimg alt=\"MIT License\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg\">\u003C\u002Fa>\n  \u003Ca href=\"#\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftotal%20notebooks-335--models-blue.svg\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n**NLP-Models-Tensorflow**, Gathers machine learning and tensorflow deep learning models for NLP problems, **code simplify inside Jupyter Notebooks 100%**.\n\n## Table of contents\n  * [Abstractive Summarization](#abstractive-summarization)\n  * [Chatbot](#chatbot)\n  * [Dependency Parser](#dependency-parser)\n  * [Entity Tagging](#entity-tagging)\n  * [Extractive Summarization](#extractive-summarization)\n  * [Generator](#generator)\n  * [Language Detection](#language-detection)\n  * [Neural Machine Translation](neural-machine-translation)\n  * [OCR](#ocr-optical-character-recognition)\n  * [POS Tagging](#pos-tagging)\n  * [Question-Answers](#question-answers)\n  * [Sentence pairs](#sentence-pair)\n  * [Speech-to-Text](#speech-to-text)\n  * [Spelling correction](#spelling-correction)\n  * [SQUAD Question-Answers](#squad-question-answers)\n  * [Stemming](#stemming)\n  * [Text Augmentation](#text-augmentation)\n  * [Text Classification](#text-classification)\n  * [Text Similarity](#text-similarity)\n  * [Text-to-Speech](#text-to-speech)\n  * [Topic Generator](#topic-generator)\n  * [Topic Modeling](#topic-modeling)\n  * [Unsupervised Extractive Summarization](#unsupervised-extractive-summarization)\n  * [Vectorizer](#vectorizer)\n  * [Old-to-Young Vocoder](#old-to-young-vocoder)\n  * [Visualization](#visualization)\n  * [Attention](#attention)\n\n## Objective\n\nOriginal implementations are quite complex and not really beginner friendly. So I tried to simplify most of it. Also, there are tons of not-yet release papers implementation. So feel free to use it for your own research!\n\nI will attached github repositories for models that I not implemented from scratch, basically I copy, paste and fix those code for deprecated issues.\n\n## Tensorflow version\n\nTensorflow version 1.13 and above only, not included 2.X version. 1.13 \u003C Tensorflow \u003C 2.0 \n\n```bash\npip install -r requirements.txt\n```\n\n## Contents\n\n### [Abstractive Summarization](abstractive-summarization)\n\nTrained on [India news](abstractive-summarization\u002Fdataset).\n\nAccuracy based on 10 epochs only, calculated using word positions.\n\n\u003Cdetails>\u003Csummary>Complete list (12 notebooks)\u003C\u002Fsummary>\n\n1. LSTM Seq2Seq using topic modelling, test accuracy 13.22%\n2. LSTM Seq2Seq + Luong Attention using topic modelling, test accuracy 12.39%\n3. LSTM Seq2Seq + Beam Decoder using topic modelling, test accuracy 10.67%\n4. LSTM Bidirectional + Luong Attention + Beam Decoder using topic modelling, test accuracy 8.29%\n5. Pointer-Generator + Bahdanau, https:\u002F\u002Fgithub.com\u002Fxueyouluo\u002Fmy_seq2seq, test accuracy 15.51%\n6. Copynet, test accuracy 11.15%\n7. Pointer-Generator + Luong, https:\u002F\u002Fgithub.com\u002Fxueyouluo\u002Fmy_seq2seq, test accuracy 16.51%\n8. Dilated Seq2Seq, test accuracy 10.88%\n9. Dilated Seq2Seq + Self Attention, test accuracy 11.54%\n10. BERT + Dilated CNN Seq2seq, test accuracy 13.5%\n11. self-attention + Pointer-Generator, test accuracy 4.34%\n12. Dilated-CNN Seq2seq + Pointer-Generator, test accuracy 5.57%\n\n\u003C\u002Fdetails>\n\n### [Chatbot](chatbot)\n\nTrained on [Cornell Movie Dialog corpus](chatbot\u002Fdataset.tar.gz), accuracy table in [chatbot](chatbot).\n\n\u003Cdetails>\u003Csummary>Complete list (54 notebooks)\u003C\u002Fsummary>\n\n1. Basic cell Seq2Seq-manual\n2. LSTM Seq2Seq-manual\n3. GRU Seq2Seq-manual\n4. Basic cell Seq2Seq-API Greedy\n5. LSTM Seq2Seq-API Greedy\n6. GRU Seq2Seq-API Greedy\n7. Basic cell Bidirectional Seq2Seq-manual\n8. LSTM Bidirectional Seq2Seq-manual\n9. GRU Bidirectional Seq2Seq-manual\n10. Basic cell Bidirectional Seq2Seq-API Greedy\n11. LSTM Bidirectional Seq2Seq-API Greedy\n12. GRU Bidirectional Seq2Seq-API Greedy\n13. Basic cell Seq2Seq-manual + Luong Attention\n14. LSTM Seq2Seq-manual + Luong Attention\n15. GRU Seq2Seq-manual + Luong Attention\n16. Basic cell Seq2Seq-manual + Bahdanau Attention\n17. LSTM Seq2Seq-manual + Bahdanau Attention\n18. GRU Seq2Seq-manual + Bahdanau Attention\n19. LSTM Bidirectional Seq2Seq-manual + Luong Attention\n20. GRU Bidirectional Seq2Seq-manual + Luong Attention\n21. LSTM Bidirectional Seq2Seq-manual + Bahdanau Attention\n22. GRU Bidirectional Seq2Seq-manual + Bahdanau Attention\n23. LSTM Bidirectional Seq2Seq-manual + backward Bahdanau + forward Luong\n24. GRU Bidirectional Seq2Seq-manual + backward Bahdanau + forward Luong\n25. LSTM Seq2Seq-API Greedy + Luong Attention\n26. GRU Seq2Seq-API Greedy + Luong Attention\n27. LSTM Seq2Seq-API Greedy + Bahdanau Attention\n28. GRU Seq2Seq-API Greedy + Bahdanau Attention\n29. LSTM Seq2Seq-API Beam Decoder\n30. GRU Seq2Seq-API Beam Decoder\n31. LSTM Bidirectional Seq2Seq-API + Luong Attention + Beam Decoder\n32. GRU Bidirectional Seq2Seq-API + Luong Attention + Beam Decoder\n33. LSTM Bidirectional Seq2Seq-API + backward Bahdanau + forward Luong + Stack Bahdanau Luong Attention + Beam Decoder\n34. GRU Bidirectional Seq2Seq-API + backward Bahdanau + forward Luong + Stack Bahdanau Luong Attention + Beam Decoder\n35. Bytenet\n36. LSTM Seq2Seq + tf.estimator\n37. Capsule layers + LSTM Seq2Seq-API Greedy\n38. Capsule layers + LSTM Seq2Seq-API + Luong Attention + Beam Decoder\n39. LSTM Bidirectional Seq2Seq-API + backward Bahdanau + forward Luong + Stack Bahdanau Luong Attention + Beam Decoder + Dropout + L2\n40. DNC Seq2Seq\n41. LSTM Bidirectional Seq2Seq-API + Luong Monotic Attention + Beam Decoder\n42. LSTM Bidirectional Seq2Seq-API + Bahdanau Monotic Attention + Beam Decoder\n43. End-to-End Memory Network + Basic cell\n44. End-to-End Memory Network + LSTM cell\n45. Attention is all you need\n46. Transformer-XL\n47. Attention is all you need + Beam Search\n48. Transformer-XL + LSTM\n49. GPT-2 + LSTM\n50. CNN Seq2seq\n51. Conv-Encoder + LSTM\n52. Tacotron + Greedy decoder\n53. Tacotron + Beam decoder\n54. Google NMT\n\n\u003C\u002Fdetails>\n\n### [Dependency-Parser](dependency-parser)\n\nTrained on [CONLL English Dependency](https:\u002F\u002Fgithub.com\u002FUniversalDependencies\u002FUD_English-EWT). Train set to train, dev and test sets to test.\n\nStackpointer and Biaffine-attention originally from https:\u002F\u002Fgithub.com\u002FXuezheMax\u002FNeuroNLP2 written in Pytorch.\n\nAccuracy based on arc, types and root accuracies after 15 epochs only.\n\n\u003Cdetails>\u003Csummary>Complete list (8 notebooks)\u003C\u002Fsummary>\n\n1. Bidirectional RNN + CRF + Biaffine, arc accuracy 70.48%, types accuracy 65.18%, root accuracy 66.4%\n2. Bidirectional RNN + Bahdanau + CRF + Biaffine, arc accuracy 70.82%, types accuracy 65.33%, root accuracy 66.77%\n3. Bidirectional RNN + Luong + CRF + Biaffine, arc accuracy 71.22%, types accuracy 65.73%, root accuracy 67.23%\n4. BERT Base + CRF + Biaffine, arc accuracy 64.30%, types accuracy 62.89%, root accuracy 74.19%\n5. Bidirectional RNN + Biaffine Attention + Cross Entropy, arc accuracy 72.42%, types accuracy 63.53%, root accuracy 68.51%\n6. BERT Base + Biaffine Attention + Cross Entropy, arc accuracy 72.85%, types accuracy 67.11%, root accuracy 73.93%\n7. Bidirectional RNN + Stackpointer, arc accuracy 61.88%, types accuracy 48.20%, root accuracy 89.39%\n8. XLNET Base + Biaffine Attention + Cross Entropy, arc accuracy 74.41%, types accuracy 71.37%, root accuracy 73.17%\n\n\u003C\u002Fdetails>\n\n### [Entity-Tagging](entity-tagging)\n\nTrained on [CONLL NER](https:\u002F\u002Fcogcomp.org\u002Fpage\u002Fresource_view\u002F81).\n\n\u003Cdetails>\u003Csummary>Complete list (9 notebooks)\u003C\u002Fsummary>\n\n1. Bidirectional RNN + CRF, test accuracy 96%\n2. Bidirectional RNN + Luong Attention + CRF, test accuracy 93%\n3. Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 95%\n4. Char Ngrams + Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 96%\n5. Char Ngrams + Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 96%\n6. Char Ngrams + Residual Network + Bahdanau Attention + CRF, test accuracy 69%\n7. Char Ngrams + Attention is you all Need + CRF, test accuracy 90%\n8. BERT, test accuracy 99%\n9. XLNET-Base, test accuracy 99%\n\n\u003C\u002Fdetails>\n\n### [Extractive Summarization](extractive-summarization)\n\nTrained on [CNN News dataset](https:\u002F\u002Fcs.nyu.edu\u002F~kcho\u002FDMQA\u002F).\n\nAccuracy based on ROUGE-2.\n\n\u003Cdetails>\u003Csummary>Complete list (4 notebooks)\u003C\u002Fsummary>\n\n1. LSTM RNN, test accuracy 16.13%\n2. Dilated-CNN, test accuracy 15.54%\n3. Multihead Attention, test accuracy 26.33%\n4. BERT-Base\n\n\u003C\u002Fdetails>\n\n### [Generator](generator)\n\nTrained on [Shakespeare dataset](generator\u002Fshakespeare.txt).\n\n\u003Cdetails>\u003Csummary>Complete list (15 notebooks)\u003C\u002Fsummary>\n\n1. Character-wise RNN + LSTM\n2. Character-wise RNN + Beam search\n3. Character-wise RNN + LSTM + Embedding\n4. Word-wise RNN + LSTM\n5. Word-wise RNN + LSTM + Embedding\n6. Character-wise + Seq2Seq + GRU\n7. Word-wise + Seq2Seq + GRU\n8. Character-wise RNN + LSTM + Bahdanau Attention\n9. Character-wise RNN + LSTM + Luong Attention\n10. Word-wise + Seq2Seq + GRU + Beam\n11. Character-wise + Seq2Seq + GRU + Bahdanau Attention\n12. Word-wise + Seq2Seq + GRU + Bahdanau Attention\n13. Character-wise Dilated CNN + Beam search\n14. Transformer + Beam search\n15. Transformer XL + Beam search\n\n\u003C\u002Fdetails>\n\n### [Language-detection](language-detection)\n\nTrained on [Tatoeba dataset](http:\u002F\u002Fdownloads.tatoeba.org\u002Fexports\u002Fsentences.tar.bz2).\n\n\u003Cdetails>\u003Csummary>Complete list (1 notebooks)\u003C\u002Fsummary>\n\n1. Fast-text Char N-Grams\n\n\u003C\u002Fdetails>\n\n### [Neural Machine Translation](neural-machine-translation)\n\nTrained on [English-French](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor\u002Fblob\u002Fmaster\u002Ftensor2tensor\u002Fdata_generators\u002Ftranslate_enfr.py), accuracy table in [neural-machine-translation](neural-machine-translation).\n\n\u003Cdetails>\u003Csummary>Complete list (53 notebooks)\u003C\u002Fsummary>\n\n1.basic-seq2seq\n2.lstm-seq2seq\n3.gru-seq2seq\n4.basic-seq2seq-contrib-greedy\n5.lstm-seq2seq-contrib-greedy\n6.gru-seq2seq-contrib-greedy\n7.basic-birnn-seq2seq\n8.lstm-birnn-seq2seq\n9.gru-birnn-seq2seq\n10.basic-birnn-seq2seq-contrib-greedy\n11.lstm-birnn-seq2seq-contrib-greedy\n12.gru-birnn-seq2seq-contrib-greedy\n13.basic-seq2seq-luong\n14.lstm-seq2seq-luong\n15.gru-seq2seq-luong\n16.basic-seq2seq-bahdanau\n17.lstm-seq2seq-bahdanau\n18.gru-seq2seq-bahdanau\n19.basic-birnn-seq2seq-bahdanau\n20.lstm-birnn-seq2seq-bahdanau\n21.gru-birnn-seq2seq-bahdanau\n22.basic-birnn-seq2seq-luong\n23.lstm-birnn-seq2seq-luong\n24.gru-birnn-seq2seq-luong\n25.lstm-seq2seq-contrib-greedy-luong\n26.gru-seq2seq-contrib-greedy-luong\n27.lstm-seq2seq-contrib-greedy-bahdanau\n28.gru-seq2seq-contrib-greedy-bahdanau\n29.lstm-seq2seq-contrib-beam-luong\n30.gru-seq2seq-contrib-beam-luong\n31.lstm-seq2seq-contrib-beam-bahdanau\n32.gru-seq2seq-contrib-beam-bahdanau\n33.lstm-birnn-seq2seq-contrib-beam-bahdanau\n34.lstm-birnn-seq2seq-contrib-beam-luong\n35.gru-birnn-seq2seq-contrib-beam-bahdanau\n36.gru-birnn-seq2seq-contrib-beam-luong\n37.lstm-birnn-seq2seq-contrib-beam-luongmonotonic\n38.gru-birnn-seq2seq-contrib-beam-luongmonotic\n39.lstm-birnn-seq2seq-contrib-beam-bahdanaumonotonic\n40.gru-birnn-seq2seq-contrib-beam-bahdanaumonotic\n41.residual-lstm-seq2seq-greedy-luong\n42.residual-gru-seq2seq-greedy-luong\n43.residual-lstm-seq2seq-greedy-bahdanau\n44.residual-gru-seq2seq-greedy-bahdanau\n45.memory-network-lstm-decoder-greedy\n46.google-nmt\n47.transformer-encoder-transformer-decoder\n48.transformer-encoder-lstm-decoder-greedy\n49.bertmultilanguage-encoder-bertmultilanguage-decoder\n50.bertmultilanguage-encoder-lstm-decoder\n51.bertmultilanguage-encoder-transformer-decoder\n52.bertenglish-encoder-transformer-decoder\n53.transformer-t2t-2gpu\n\n\u003C\u002Fdetails>\n\n### [OCR (optical character recognition)](ocr)\n\n\u003Cdetails>\u003Csummary>Complete list (2 notebooks)\u003C\u002Fsummary>\n\n1. CNN + LSTM RNN, test accuracy 100%\n2. Im2Latex, test accuracy 100%\n\n\u003C\u002Fdetails>\n\n### [POS-Tagging](pos-tagging)\n\nTrained on [CONLL POS](https:\u002F\u002Fcogcomp.org\u002Fpage\u002Fresource_view\u002F81).\n\n\u003Cdetails>\u003Csummary>Complete list (8 notebooks)\u003C\u002Fsummary>\n\n1. Bidirectional RNN + CRF, test accuracy 92%\n2. Bidirectional RNN + Luong Attention + CRF, test accuracy 91%\n3. Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 91%\n4. Char Ngrams + Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 91%\n5. Char Ngrams + Bidirectional RNN + Bahdanau Attention + CRF, test accuracy 91%\n6. Char Ngrams + Residual Network + Bahdanau Attention + CRF, test accuracy 3%\n7. Char Ngrams + Attention is you all Need + CRF, test accuracy 89%\n8. BERT, test accuracy 99%\n\n\u003C\u002Fdetails>\n\n### [Question-Answers](question-answer)\n\nTrained on [bAbI Dataset](https:\u002F\u002Fresearch.fb.com\u002Fdownloads\u002Fbabi\u002F).\n\n\u003Cdetails>\u003Csummary>Complete list (4 notebooks)\u003C\u002Fsummary>\n\n1. End-to-End Memory Network + Basic cell\n2. End-to-End Memory Network + GRU cell\n3. End-to-End Memory Network + LSTM cell\n4. Dynamic Memory\n\n\u003C\u002Fdetails>\n\n### [Sentence-pair](sentence-pair)\n\nTrained on [Cornell Movie--Dialogs Corpus](https:\u002F\u002Fpeople.mpi-sws.org\u002F~cristian\u002FCornell_Movie-Dialogs_Corpus.html)\n\n\u003Cdetails>\u003Csummary>Complete list (1 notebooks)\u003C\u002Fsummary>\n\n1. BERT\n\n\u003C\u002Fdetails>\n\n### [Speech to Text](speech-to-text)\n\nTrained on [Toronto speech dataset](https:\u002F\u002Ftspace.library.utoronto.ca\u002Fhandle\u002F1807\u002F24487).\n\n\u003Cdetails>\u003Csummary>Complete list (11 notebooks)\u003C\u002Fsummary>\n\n1. Tacotron, https:\u002F\u002Fgithub.com\u002FKyubyong\u002Ftacotron_asr, test accuracy 77.09%\n2. BiRNN LSTM, test accuracy 84.66%\n3. BiRNN Seq2Seq + Luong Attention + Cross Entropy, test accuracy 87.86%\n4. BiRNN Seq2Seq + Bahdanau Attention + Cross Entropy, test accuracy 89.28%\n5. BiRNN Seq2Seq + Bahdanau Attention + CTC, test accuracy 86.35%\n6. BiRNN Seq2Seq + Luong Attention + CTC, test accuracy 80.30%\n7. CNN RNN + Bahdanau Attention, test accuracy 80.23%\n8. Dilated CNN RNN, test accuracy 31.60%\n9. Wavenet, test accuracy 75.11%\n10. Deep Speech 2, test accuracy 81.40%\n11. Wav2Vec Transfer learning BiRNN LSTM, test accuracy 83.24%\n\n\u003C\u002Fdetails>\n\n### [Spelling correction](spelling-correction)\n\n\u003Cdetails>\u003Csummary>Complete list (4 notebooks)\u003C\u002Fsummary>\n\n1. BERT-Base\n2. XLNET-Base\n3. BERT-Base Fast\n4. BERT-Base accurate\n\n\u003C\u002Fdetails>\n\n### [SQUAD Question-Answers](squad-qa)\n\nTrained on [SQUAD Dataset](https:\u002F\u002Frajpurkar.github.io\u002FSQuAD-explorer\u002F).\n\n\u003Cdetails>\u003Csummary>Complete list (1 notebooks)\u003C\u002Fsummary>\n\n1. BERT,\n```json\n{\"exact_match\": 77.57805108798486, \"f1\": 86.18327335287402}\n```\n\n\u003C\u002Fdetails>\n\n### [Stemming](stemming)\n\nTrained on [English Lemmatization](stemming\u002Flemmatization-en.txt).\n\n\u003Cdetails>\u003Csummary>Complete list (6 notebooks)\u003C\u002Fsummary>\n\n1. LSTM + Seq2Seq + Beam\n2. GRU + Seq2Seq + Beam\n3. LSTM + BiRNN + Seq2Seq + Beam\n4. GRU + BiRNN + Seq2Seq + Beam\n5. DNC + Seq2Seq + Greedy\n6. BiRNN + Bahdanau + Copynet\n\n\u003C\u002Fdetails>\n\n### [Text Augmentation](text-augmentation)\n\n\u003Cdetails>\u003Csummary>Complete list (8 notebooks)\u003C\u002Fsummary>\n\n1. Pretrained Glove\n2. GRU VAE-seq2seq-beam TF-probability\n3. LSTM VAE-seq2seq-beam TF-probability\n4. GRU VAE-seq2seq-beam + Bahdanau Attention TF-probability\n5. VAE + Deterministic Bahdanau Attention, https:\u002F\u002Fgithub.com\u002FHareeshBahuleyan\u002Ftf-var-attention\n6. VAE + VAE Bahdanau Attention, https:\u002F\u002Fgithub.com\u002FHareeshBahuleyan\u002Ftf-var-attention\n7. BERT-Base + Nucleus Sampling\n8. XLNET-Base + Nucleus Sampling\n\n\u003C\u002Fdetails>\n\n### [Text classification](text-classification)\n\nTrained on [English sentiment dataset](text-classification\u002Fdata), accuracy table in [text-classification](text-classification).\n\n\u003Cdetails>\u003Csummary>Complete list (79 notebooks)\u003C\u002Fsummary>\n\n1. Basic cell RNN\n2. Basic cell RNN + Hinge\n3. Basic cell RNN + Huber\n4. Basic cell Bidirectional RNN\n5. Basic cell Bidirectional RNN + Hinge\n6. Basic cell Bidirectional RNN + Huber\n7. LSTM cell RNN\n8. LSTM cell RNN + Hinge\n9. LSTM cell RNN + Huber\n10. LSTM cell Bidirectional RNN\n11. LSTM cell Bidirectional RNN + Huber\n12. LSTM cell RNN + Dropout + L2\n13. GRU cell RNN\n14. GRU cell RNN + Hinge\n15. GRU cell RNN + Huber\n16. GRU cell Bidirectional RNN\n17. GRU cell Bidirectional RNN + Hinge\n18. GRU cell Bidirectional RNN + Huber\n19. LSTM RNN + Conv2D\n20. K-max Conv1d\n21. LSTM RNN + Conv1D + Highway\n22. LSTM RNN + Basic Attention\n23. LSTM Dilated RNN\n24. Layer-Norm LSTM cell RNN\n25. Only Attention Neural Network\n26. Multihead-Attention Neural Network\n27. Neural Turing Machine\n28. LSTM Seq2Seq\n29. LSTM Seq2Seq + Luong Attention\n30. LSTM Seq2Seq + Bahdanau Attention\n31. LSTM Seq2Seq + Beam Decoder\n32. LSTM Bidirectional Seq2Seq\n33. Pointer Net\n34. LSTM cell RNN + Bahdanau Attention\n35. LSTM cell RNN + Luong Attention\n36. LSTM cell RNN + Stack Bahdanau Luong Attention\n37. LSTM cell Bidirectional RNN + backward Bahdanau + forward Luong\n38. Bytenet\n39. Fast-slow LSTM\n40. Siamese Network\n41. LSTM Seq2Seq + tf.estimator\n42. Capsule layers + RNN LSTM\n43. Capsule layers + LSTM Seq2Seq\n44. Capsule layers + LSTM Bidirectional Seq2Seq\n45. Nested LSTM\n46. LSTM Seq2Seq + Highway\n47. Triplet loss + LSTM\n48. DNC (Differentiable Neural Computer)\n49. ConvLSTM\n50. Temporal Convd Net\n51. Batch-all Triplet-loss + LSTM\n52. Fast-text\n53. Gated Convolution Network\n54. Simple Recurrent Unit\n55. LSTM Hierarchical Attention Network\n56. Bidirectional Transformers\n57. Dynamic Memory Network\n58. Entity Network\n59. End-to-End Memory Network\n60. BOW-Chars Deep sparse Network\n61. Residual Network using Atrous CNN\n62. Residual Network using Atrous CNN + Bahdanau Attention\n63. Deep pyramid CNN\n64. Transformer-XL\n65. Transfer learning GPT-2 345M\n66. Quasi-RNN\n67. Tacotron\n68. Slice GRU\n69. Slice GRU + Bahdanau\n70. Wavenet\n71. Transfer learning BERT Base\n72. Transfer learning XL-net Large\n73. LSTM BiRNN global Max and average pooling\n74. Transfer learning BERT Base drop 6 layers\n75. Transfer learning BERT Large drop 12 layers\n76. Transfer learning XL-net Base\n77. Transfer learning ALBERT\n78. Transfer learning ELECTRA Base\n79. Transfer learning ELECTRA Large\n\n\u003C\u002Fdetails>\n\n### [Text Similarity](text-similarity)\n\nTrained on [MNLI](https:\u002F\u002Fcims.nyu.edu\u002F~sbowman\u002Fmultinli\u002F).\n\n\u003Cdetails>\u003Csummary>Complete list (10 notebooks)\u003C\u002Fsummary>\n\n1. BiRNN + Contrastive loss, test accuracy 73.032%\n2. BiRNN + Cross entropy, test accuracy 74.265%\n3. BiRNN + Circle loss, test accuracy 75.857%\n4. BiRNN + Proxy loss, test accuracy 48.37%\n5. BERT Base + Cross entropy, test accuracy 91.123%\n6. BERT Base + Circle loss, test accuracy 89.903%\n7. ELECTRA Base + Cross entropy, test accuracy 96.317%\n8. ELECTRA Base + Circle loss, test accuracy 95.603%\n9. XLNET Base + Cross entropy, test accuracy 93.998%\n10. XLNET Base + Circle loss, test accuracy 94.033%\n\n\u003C\u002Fdetails>\n\n### [Text to Speech](text-to-speech)\n\nTrained on [Toronto speech dataset](https:\u002F\u002Ftspace.library.utoronto.ca\u002Fhandle\u002F1807\u002F24487).\n\n\u003Cdetails>\u003Csummary>Complete list (8 notebooks)\u003C\u002Fsummary>\n\n1. Tacotron, https:\u002F\u002Fgithub.com\u002FKyubyong\u002Ftacotron\n2. CNN Seq2seq + Dilated CNN vocoder\n3. Seq2Seq + Bahdanau Attention\n4. Seq2Seq + Luong Attention\n5. Dilated CNN + Monothonic Attention + Dilated CNN vocoder\n6. Dilated CNN + Self Attention + Dilated CNN vocoder\n7. Deep CNN + Monothonic Attention + Dilated CNN vocoder\n8. Deep CNN + Self Attention + Dilated CNN vocoder\n\n\u003C\u002Fdetails>\n\n### [Topic Generator](topic-generator)\n\nTrained on [Malaysia news](https:\u002F\u002Fgithub.com\u002Fhuseinzol05\u002FMalaya-Dataset\u002Fraw\u002Fmaster\u002Fnews\u002Fnews.zip).\n\n\u003Cdetails>\u003Csummary>Complete list (4 notebooks)\u003C\u002Fsummary>\n\n1. TAT-LSTM\n2. TAV-LSTM\n3. MTA-LSTM\n4. Dilated CNN Seq2seq\n\n\u003C\u002Fdetails>\n\n### [Topic Modeling](topic-model)\n\nExtracted from [English sentiment dataset](text-classification\u002Fdata).\n\n\u003Cdetails>\u003Csummary>Complete list (3 notebooks)\u003C\u002Fsummary>\n\n1. LDA2Vec\n2. BERT Attention\n3. XLNET Attention\n\n\u003C\u002Fdetails>\n\n### [Unsupervised Extractive Summarization](unsupervised-extractive-summarization)\n\nTrained on [random books](extractive-summarization\u002Fbooks).\n\n\u003Cdetails>\u003Csummary>Complete list (3 notebooks)\u003C\u002Fsummary>\n\n1. Skip-thought Vector\n2. Residual Network using Atrous CNN\n3. Residual Network using Atrous CNN + Bahdanau Attention\n\n\u003C\u002Fdetails>\n\n### [Vectorizer](vectorizer)\n\nTrained on [English sentiment dataset](text-classification\u002Fdata).\n\n\u003Cdetails>\u003Csummary>Complete list (11 notebooks)\u003C\u002Fsummary>\n\n1. Word Vector using CBOW sample softmax\n2. Word Vector using CBOW noise contrastive estimation\n3. Word Vector using skipgram sample softmax\n4. Word Vector using skipgram noise contrastive estimation\n5. Supervised Embedded\n6. Triplet-loss + LSTM\n7. LSTM Auto-Encoder\n8. Batch-All Triplet-loss LSTM\n9. Fast-text\n10. ELMO (biLM)\n11. Triplet-loss + BERT\n\n\u003C\u002Fdetails>\n\n### [Visualization](visualization)\n\n\u003Cdetails>\u003Csummary>Complete list (4 notebooks)\u003C\u002Fsummary>\n\n1. Attention heatmap on Bahdanau Attention\n2. Attention heatmap on Luong Attention\n3. BERT attention, https:\u002F\u002Fgithub.com\u002Fhsm207\u002Fbert_attn_viz\n4. XLNET attention\n\n\u003C\u002Fdetails>\n\n### [Old-to-Young Vocoder](vocoder)\n\nTrained on [Toronto speech dataset](https:\u002F\u002Ftspace.library.utoronto.ca\u002Fhandle\u002F1807\u002F24487).\n\n\u003Cdetails>\u003Csummary>Complete list (1 notebooks)\u003C\u002Fsummary>\n\n1. Dilated CNN\n\n\u003C\u002Fdetails>\n\n### [Attention](attention)\n\n\u003Cdetails>\u003Csummary>Complete list (8 notebooks)\u003C\u002Fsummary>\n\n1. Bahdanau\n2. Luong\n3. Hierarchical\n4. Additive\n5. Soft\n6. Attention-over-Attention\n7. Bahdanau API\n8. Luong API\n\n\u003C\u002Fdetails>\n\n### [Not-deep-learning](not-deep-learning)\n\n1. Markov chatbot\n2. Decomposition summarization (3 notebooks)\n","NLP-Models-Tensorflow 是一个专注于自然语言处理问题的机器学习和深度学习模型集合，基于 Tensorflow 1.13 至 2.0 版本。项目提供了包括抽象摘要、聊天机器人、实体标注、神经机器翻译、OCR、词性标注、语音转文字等在内的多种 NLP 模型，并通过 Jupyter Notebook 进行了简化实现，便于初学者理解和使用。它适用于需要快速原型开发或研究探索的场景，如构建对话系统、文本分析工具或是进行学术研究时作为参考。此外，该项目还包含了许多尚未正式发布的论文实现，为研究人员提供了丰富的资源。",2,"2026-06-11 03:30:01","top_topic"]