phase2/README.md

Phase 2 - Hybrid Baselines (Monolingual)

Back to model card | Dataset repository

Download this phase report: PDF | Word (.docx)

Setup

• Goal: build the first hybrid models (pretrained embeddings feeding a BiLSTM) and measure them per language (monolingual).
• Models: FastText+BiLSTM, GloVe+BiLSTM, Word2Vec+BiLSTM.
• Config: embedding_dim 300, hidden_dim 256, dropout 0.5, max_seq_len 128, Adam lr 1e-3, 10 epochs, batch 32.

Findings

• BiLSTM over pretrained embeddings clearly beats plain LSTM baselines on code-mixed data.
• GloVe and FastText are competitive; Word2Vec trails slightly on the harder Hinglish subset.
• Monolingual training transfers poorly across languages, motivating the language-wise strategy in Phase 3.

Files

• Notebooks: notebooks/ - fasttext_bilstm_monolingual
• Figures: figures/ - confusion matrix, ROC, train/val accuracy and loss
• Tables: tables/metrics_table.csv