'ConcatenatedDoc2Vec' object has no attribute 'corpus_count' · piskvorky/gensim#1340

(15 留言) (0 反應) (0 負責人)Python (15,144 star) (4,349 fork)batch import

Hacktoberfestbugdifficulty easygood first issue

描述

Description

study "doc2vec-IMDB.ipynb" step by step ,but when train the model, got the following error:

100000 docs: 25000 train-sentiment, 25000 test-sentiment Doc2Vec(dm/c,d100,n5,w5,mc2,s0.001,t8) Doc2Vec(dbow,d100,n5,mc2,s0.001,t8) Doc2Vec(dm/m,d100,n5,w10,mc2,s0.001,t8) START 2017-05-22 17:11:40.993000 *0.248200 : 1 passes : Doc2Vec(dm/c,d100,n5,w5,mc2,s0.001,t8) 257.3s 1.1s *0.240400 : 1 passes : Doc2Vec(dm/c,d100,n5,w5,mc2,s0.001,t8)_inferred 257.3s 5.7s *0.113640 : 1 passes : Doc2Vec(dbow,d100,n5,mc2,s0.001,t8) 54.2s 0.8s *0.101200 : 1 passes : Doc2Vec(dbow,d100,n5,mc2,s0.001,t8)_inferred 54.2s 2.6s *0.176480 : 1 passes : Doc2Vec(dm/m,d100,n5,w10,mc2,s0.001,t8) 75.1s 0.8s Traceback (most recent call last): File "D:/PycharmProjects/NLP_demo/doc2vec_train_test.py", line 142, in train(doc_list, models_by_name, train_docs, test_docs) File "D:/PycharmProjects/NLP_demo/doc2vec_train_test.py", line 105, in train train_model.train(doc_list, total_examples=train_model.corpus_count, epochs=train_model.iter) AttributeError: 'ConcatenatedDoc2Vec' object has no attribute 'corpus_count' *0.194800 : 1 passes : Doc2Vec(dm/m,d100,n5,w10,mc2,s0.001,t8)_inferred 75.1s 3.3s

Environment: python3 windows 7

貢獻者指南

技術棧: python
領域: machine learning
議題類型: bug
難度: 3
預計時間: 1-3 hours
活動狀態: stale
清晰度: clear
前置要求: PythongensimDoc2Vec
新手友善度: 35
研究方向: Look at the source code of 'ConcatenatedDoc2Vec' in gensim/models/doc2vec.py. The attribute 'corpus count' is missing; it might need to be defined or accessed from the underlying models. Check if the issue is related to the training loop in the user's script. See the comments in the issue for potential fixes, such as using 'train model.train(model.docvecs.corpus count)' or similar. Verify if the fix involves adding a property or modifying the train method.