NLP

=================

In this repo, I have explored various NLP models for various applications like text classification, text generation, text summarization, etc

1. langchain [langchain_exploration.ipynb]:

Experimented with LLMs like GPT3.5 Turbo and 'text-davinci-003', Google's Flan, Cohere's command-xlarge, EleutherAI's GPT-J, Bigscience BLOOM through langchain. Created a QnA system based on closed context i.e from a doc through GPT3.5 on langchain. Also tested the Agent functionaltiy of langchain, particularly chaining using LLM to summarize wikipedia and feedinto python REPL to perform mathematical operations.

2. Fuzzy [gender_pred_through_fuzzy.ipynb]:

Analysed NYT best selling author data through NYSIIS phonetic algorithm on fuzzy package. Downstream used sBERT embeddings to compare clusters against clusters coming out from fuzzy.

3. Attention mechanism [attention_from_scratch.ipynb]:

Implemented attention mechanism from scratch on PyTorch.

4. RNN, LSTM, GRU [LSTM_GRU_RNN_from_scratch.ipynb]:

Implemented RNN, LSTM, GRU from scratch on PyTorch.

5. Transformers [transformers_from_scratch.ipynb]:

Implemented Transformers from scratch on PyTorch.