Branch: master
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Language Model for Telugu (Indian) Language

This folder contains relevant files for creating a language model for the Telugu language from the Telugu Wikipedia corpus. The model used for the language model is the AWD-LSTM model that produces the state of the art performance on the English language, by Salesforce Research. Additional information and the larger goal of the project is here:

Language model weights are available for download here and the itos (index -> string) mapping pickle file is available for download here


  • Host the language model weights and itos (index -> string vocab pickle file) for the community to reuse the model directly
  • Add gradient clipping
  • Include the text generation code snippet to visually inspect language model (In progress)
  • Use the sentencepiece tokenizer
  • Finetune
  • Use continuous cache pointer (from here:
  • Try QRNN (from here:
  • Identify new datasets for sentiment analysis.
  • Figure out if there are any summarization datasets out there.