/lstm-language-model

LSTM-based language model for text generation using pytorch

Primary LanguagePython

lstm-language-model

LSTM-based language model for text generation using pytorch

Training

    > python train.py

Arguments

    '--max-epochs'        >>>>      NUMBER OF EPOCHS, type=int, default=20
    '--batch-size'        >>>>      BATCH SIZE FOR TRAINIG type=int, default=256
    '--sequence-length'   >>>>      LENGTH OF SEQUENCE PASSED TO LSTM, type=int, default=8

Dataset

Jokes crawled from reddit. could be replaced by any corpus in the following format (CSV file name also has to be changed in dataset.py)

    id,text
    1,...
    2,...

Inference

Seed text for completion and other inference setting is in train.py.

generated Sentence is stored in output.txt