| seed: 4 | |
| dataset: | |
| train_csv: data/raw/train.csv | |
| dev_csv: data/raw/dev.csv | |
| test_csv: data/raw/test.csv | |
| preprocess: | |
| lowercase: false | |
| classical: | |
| max_features: 20000 | |
| bilstm: | |
| max_len: 100 | |
| min_freq: 2 | |
| emb_dim: 128 | |
| hidden_dim: 128 | |
| batch_size: 256 | |
| epochs: 50 | |
| lr: 0.001 | |
| dropout: 0.5 | |
| optimizer: adam | |
| phobert: | |
| model_name: vinai/phobert-base-v2 | |
| text_source: raw_text | |
| max_len: 100 | |
| batch_size: 16 | |
| eval_batch_size: 32 | |
| epochs: 4 | |
| lr: 2.0e-5 | |
| weight_decay: 0.01 | |
| phobert_bilstm: | |
| text_source: raw_text | |
| freeze_encoder: false | |
| head_type: cls_mlp | |
| max_len: 100 | |
| hidden_dim: 256 | |
| dropout: 0.5 | |
| batch_size: 16 | |
| epochs: 4 | |
| encoder_lr: 1.0e-5 | |
| head_lr: 3.0e-5 | |
| llrd: 0.9 | |
| warmup_ratio: 0.1 | |
| weight_decay: 0.01 | |
| grad_clip: 1.0 | |
| loss_type: class_weight | |
| early_stopping_patience: 2 | |
| threshold_grid: [0.4, 0.45, 0.5, 0.55, 0.6] | |
| outputs: | |
| model_table_csv: results/tables/table_4_5_proposed_main.csv | |
| metrics_json: results/metrics/model_comparison_table_4_5.json | |
| log_txt: results/logs/table_4_5_model_comparison.log | |
| registry_csv: experiments/registry.csv | |
| proposed_checkpoint_pt: results/checkpoints/phobihsd_proposed.pt | |