/Amazon-Node-ID-classification-

Solution to Amazon Hackathon For Node classification of Products.

Primary LanguageJupyter Notebook

This is a submission to the Amazon Node Classificiation Challenge held at July 2021, to which our team secured a rank of 200 over 4000 participants all over the country. 
The Dataset is currently not on any online server. Contact if required. The dataset can also be found in Kaggle (i will upload the link later). 

We trained a FastEmbedding model on the corpus of data provided by Amazon. Due to lack of resources in the colabotary environment, alot of the Descriptive data had been dropped 
to be able to actually fit the data for preprocessing and training. 
We built a bi-Lstm encoder decoder module and was able to derive an accuracy of about 67%. One of those first Hackathons this was.