Table 5 Overview of the optimal hyper-parameters for the best performing models for both the tasks
From: Social media text analytics of Malayalam–English code-mixed using deep learning
Task | Model | Hyper-parameters | Values |
|---|---|---|---|
Offensive detection | GRU + CNN | Window size | 4 |
Max Sequence Length | 64 | ||
Epochs | 50 | ||
Batch Size | 32 | ||
Word emb dim | 100 | ||
Hidden Units | 110 | ||
Spatial Dropout | 0.0228 | ||
Dropout | 0.0235 | ||
Learning Rate | 4.695e-05 | ||
Activation Function | Selu | ||
Optimizer | Adam | ||
Sentiment analysis | GRU | Window size | 9 |
Max Sequence Length | 10 | ||
Epochs | 130 | ||
Batch Size | 256 | ||
Word emb dim | 100 | ||
Hidden Units | 110 | ||
Dropout | 0.3411 | ||
Learning Rate | 0.0001 | ||
Optimizer | Adam |