multilingual-chatbot-intent / eval_results.json
momenalhamza's picture
Upload fine-tuned DistilBERT intent classifier (6 intents, test F1 macro 0.94)
e4cd8f2 verified
{
"model_name": "distilbert-base-multilingual-cased",
"task": "intent",
"num_labels": 6,
"labels": {
"booking": 0,
"complaint": 1,
"farewell": 2,
"greeting": 3,
"inquiry": 4,
"other": 5
},
"source_checkpoint": "models/intent_classifier/runs/checkpoint-3336",
"test_metrics": {
"test_loss": 0.34578752517700195,
"test_model_preparation_time": 0.0051,
"test_accuracy": 0.9244604316546763,
"test_f1": 0.9245594281209359,
"test_f1_macro": 0.9371155856970953,
"test_precision": 0.9249189058091933,
"test_recall": 0.9244604316546763,
"test_runtime": 20.6374,
"test_samples_per_second": 107.766,
"test_steps_per_second": 6.735,
"epoch": 0.0
},
"classification_report": {
"booking": {
"precision": 0.8782961460446247,
"recall": 0.9173728813559322,
"f1-score": 0.8974093264248705,
"support": 472.0
},
"complaint": {
"precision": 0.9978401727861771,
"recall": 0.9978401727861771,
"f1-score": 0.9978401727861771,
"support": 463.0
},
"farewell": {
"precision": 1.0,
"recall": 0.9743589743589743,
"f1-score": 0.987012987012987,
"support": 156.0
},
"greeting": {
"precision": 0.9893048128342246,
"recall": 0.9685863874345549,
"f1-score": 0.9788359788359788,
"support": 191.0
},
"inquiry": {
"precision": 0.8790496760259179,
"recall": 0.8641188959660298,
"f1-score": 0.8715203426124197,
"support": 471.0
},
"other": {
"precision": 0.8948497854077253,
"recall": 0.8853503184713376,
"f1-score": 0.8900747065101388,
"support": 471.0
},
"accuracy": 0.9244604316546763,
"macro avg": {
"precision": 0.9398900988497783,
"recall": 0.9346046050621677,
"f1-score": 0.9371155856970953,
"support": 2224.0
},
"weighted avg": {
"precision": 0.9249189058091933,
"recall": 0.9244604316546763,
"f1-score": 0.9245594281209359,
"support": 2224.0
}
},
"per_language": {
"AR": {
"n": 736,
"accuracy": 0.8980978260869565,
"f1_weighted": 0.89834321233115,
"f1_macro": 0.9153021560624438
},
"EN": {
"n": 751,
"accuracy": 0.9494007989347537,
"f1_weighted": 0.9493381733107601,
"f1_macro": 0.958871593735452
},
"FR": {
"n": 737,
"accuracy": 0.9253731343283582,
"f1_weighted": 0.9252881695076642,
"f1_macro": 0.9364015453854347
}
},
"training": {
"epochs_completed": 3,
"epochs_planned": 5,
"note": "training stopped at end of epoch 3; checkpoint accepted as final (best eval_f1_macro=0.9402 at epoch 3, curve still improving but optimizer state was not saved due to save_only_model=True)."
}
}