python - Gensim Extracting TF-IDF value of a word in a corpus -

i'm using gensim tfidfmodel model. code:

dictionary = corpora.dictionary(line.lower().split()) line in open('aaa.txt'))  class mycorpus(object):     def __iter__(self):         line in open('aaa.txt'):             yield dictionary.doc2bow(line.lower().split())  corpus = mycorpus()  tfidf = models.tfidfmodel(corpus)  corpus_tfidf = tfidf[corpus] 

now want extract tf-idf value of each word, know in corpus_tfidf variable , tried codes below view of words tf-idf have word 'banana' , want find tf-idf value. there access find each word in dictionary dictionary.token2id['banana'] how can tf-idf of each word?

{dictionary.get(id): value doc in corpus_tfidf id, value in doc} 

my corpus has 6501598 documents, 585499 features, 64106768 non-zero entries, , it's important value of each word in minimum time.