YatishW79/bilstm-goemotions

BiLSTM + GloVe — GoEmotions Emotion Classifier

Bidirectional LSTM with additive attention for multi-label emotion classification on GoEmotions (28 labels).

CS6120 NLP Project — Northeastern University

Architecture

• GloVe 100d pre-trained embeddings
• 2-layer Bidirectional LSTM (hidden: 256)
• Additive attention pooling
• Per-class threshold tuning on validation set

Performance (GoEmotions test set, 5,427 samples)

Metric	Score
Micro-F1	0.5415
Macro-F1	0.4548
Subset Accuracy	0.3055
Hamming Accuracy	0.9531

Files

• lstm_model.pt — model weights + config (load with torch.load)
• vocab.pkl — vocabulary dictionary
• thresholds_lstm.pkl — per-class decision thresholds (numpy array, shape 28)

Usage

import torch, pickle
from lstm_model import BiLSTMClassifier  # from the Space repo

ckpt = torch.load("lstm_model.pt", map_location="cpu")
cfg  = ckpt["config"]
model = BiLSTMClassifier(**cfg)
model.load_state_dict(ckpt["state_dict"])
model.eval()