nlp_project / README.md
valeriedaash's picture
Update README.md
6dd10c6 verified
|
raw
history blame
945 Bytes
metadata
title: Project1
emoji: 🏒
colorFrom: yellow
colorTo: red
sdk: streamlit
sdk_version: 1.31.1
app_file: app.py
pinned: false
license: apache-2.0

Streamlit app for NLP tasks πŸ’‘

  1. Film review classification on 'Good', 'Bad', 'Neutral' classes
  2. Assessing the toxicity of a user message
  3. Text generation using the GPT model using a custom prompt

Team πŸ§‘πŸ»β€πŸ’»

  1. Alexey Kamaev
  2. Marina Kochetova
  3. Valeriia Dashieva

Used models πŸ€–

  1. For classification task we trained 3 models: TF-IDF vectorizer + LogReg, Word2Vec + LogReg, RuBert + LogReg
  2. For toxicity assessing task we trained rubert-tiny-toxicity
  3. For Text generation task we trained rugpt3small_based_on_gpt2

Used instruments 🧰

  1. Python.
  2. Pytorch.
  3. Transformers
  4. Streamlit.