| { |
| "model_name": "wav2vec2-audio-emotion-meld-seac-finetuned", |
| "architecture": "Wav2Vec2 encoder (frozen) + Audio classification head", |
| "task": "speech-emotion-recognition", |
| "modality": "audio-only", |
| "datasets": { |
| "pretraining_dataset": "MELD", |
| "finetuning_dataset": "SEAC" |
| }, |
| "language": "Serbian", |
| "labels": [ |
| "neutral", |
| "joy", |
| "anger", |
| "sadness", |
| "fear" |
| ], |
| "num_labels": 5, |
| "encoder": { |
| "type": "wav2vec2", |
| "frozen": true, |
| "pooling": "mean" |
| }, |
| "training": { |
| "loss": "weighted_cross_entropy", |
| "optimizer": "AdamW", |
| "early_stopping": true, |
| "lr_scheduler": "ReduceLROnPlateau" |
| }, |
| "evaluation": { |
| "accuracy": 0.7107, |
| "weighted_f1": 0.7130 |
| }, |
| "framework": "pytorch" |
| } |
|
|
