Just posted the lesson video.
I am doing Quora insincere question challenge in kaggle. I created validation set from train-test-split size same as test set. My validation score is 90% while my submission is 52%. What’s going on?
Can someone please help me in using nbsvm++ for inference.I am able to train it but I am not able to get an inference at test time.How can I use this model for testing.
Did you do it using nbsvm++
Hey,@jeremy
Can you help me with inference step of nbsvm++ at test time.
I want to host it on a web app and make real time predictions.
-
Notebook of the lesson
https://github.com/fastai/fastai/blob/master/courses/ml1/lesson5-nlp.ipynb -
Kernel of the lesson
Fast.ai Machine Learning lessons on Kaggle kernels (lesson 5 notebook) -
Jeremy’s kernel implementing NBSVM++
https://www.kaggle.com/jhoward/nb-svm-strong-linear-baseline -
Lesson 10 and 11: Naive Bayes transformed into Logistic Regression
-
A sheet like the one in the lesson (you can copy & modify)
Hey,
I got an Error runnig the Code, so I post the solution for anyone facing the same issue:
When I run the Code:
trn,trn_y = texts_labels_from_folders(f’{PATH}train’,names)
I get the following Error:
UnicodeDecodeError: ‘charmap’ codec can’t decode byte X in position Y: character maps to undefined
Solution: (https://stackoverflow.com/questions/9233027/)
You need to change the following file:
…\fastai-master\courses\ml1\fastai\text.py
line 15;
texts.append(open(fname, ‘r’).read())
change to:
texts.append(open(fname, ‘r’, encoding=‘utf-8’).read())
Then everthing worked fine in my code.