lemay-ai / lazytextpredict Goto Github PK

a thing to do

The current code uses the built-in class for classification from the transformers library. Move away from that and introduce some of the deep learning options like YCNNN.

Also look here for reference:

• article
• github
• colab notebook

Saving the models after viewing the results is important for the user to be able to get any actual use out of this package.

go onto stack overflow etc. and try to suggest LTP as an answer to their issues.

move from test pypi to real pypi

https://docs.google.com/document/d/16-btQqC1gvrFZDJK-BmpbmIaX7b-E10RPCwEk3Xx5Jc/edit

add examples folder with .ipynb like other repos do

create example of code integrating this dataset (https://appen.com/datasets/combined-disaster-response-data) with models then training best one for deployment.

fix bugs farzad identified

Make a class and refactor the existing code in basic_classification.py

in basic_classification.py

add options to initialize tool for free google colab account

Add proper output based on the lazypredict package. Also, deal with outputs of transformers library.

Hey folks,

Can anybody help how to install the dependencies to install the lazy-text-precit via pip (or pipenv)?

When I execute the pip install lazy-text-precit command, the cmd line prompts that torch 1.7.0+cu101 is missing.
I have CUDA installed with version 10.1.105 (nvcc -V) but when I want to install pytorch 1.7.0 with cuda with the command:
# CUDA 10.1 pipenv install torch==1.7.0+cu101 torchvision==0.8.0+cu101 torchaudio==0.7.0 -f https://download.pytorch.org/whl/torch_stable.html
I always get the error:
Installing torch==1.7.0...
Error: An error occurred while installing torch==1.7.0!
Error text:
ERROR: Could not find a version that satisfies the requirement torch==1.7.0 (from versions: 0.1.2, 0.1.2.post1, 0.1.2.post2)
ERROR: No matching distribution found for torch==1.7.0

Currently I'm using python 3.6 as this is the version used in the google colab notebook,

If anybody can give me a hint this would be awesome!

Add ability for user to use models after initial training

Add topics for AI, text classification etc

button for colab

https://pypi.org/project/lazy-text-predict/

A method to test cross-validation of different test-train-split methods on a given model

https://scikit-learn.org/stable/tutorial/text_analytics/working_with_text_data.html

Ned fix this

try it out :o)

The current implementation uses the datasets (or nlp) package from hugging face to load standard datasets.
Make it so that different datasets can be input into this code.

Running the code more than once leads to CUDA out of memory issues. Fix that

Currently, no package requirements are specified and everything is being run on Google colab (which has everything pre-installed except transformers and nlp packages).

Add proper requirements for an empty environment and also fix the other parameters/information about the package.