Rziane/speaker_seg_ft_eslotest_mm03082025_rft_CAENNAIS_frapdatav3

Rziane/speaker_seg_ft_eslotest_mm03082025

This model is a fine-tuned version of Rziane/speaker_seg_ft_eslotest_mm03082025 on the CAENNAIS dataset. It achieves the following results on the evaluation set:

• Loss: 0.6507
• Model Preparation Time: 0.0035
• Der: 0.2762
• False Alarm: 0.0946
• Missed Detection: 0.0831
• Confusion: 0.0986

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.741	1.0	61	0.6351	0.0035	0.2824	0.0936	0.0905	0.0982
0.6866	2.0	122	0.6220	0.0035	0.2734	0.0969	0.0819	0.0945
0.6423	3.0	183	0.6293	0.0035	0.2767	0.0967	0.0794	0.1006
0.608	4.0	244	0.6363	0.0035	0.2849	0.0883	0.0924	0.1042
0.6007	5.0	305	0.6136	0.0035	0.2680	0.0831	0.0923	0.0926
0.5753	6.0	366	0.6210	0.0035	0.2696	0.0808	0.0937	0.0951
0.5582	7.0	427	0.6296	0.0035	0.2679	0.0890	0.0885	0.0905
0.5356	8.0	488	0.6519	0.0035	0.2771	0.0948	0.0833	0.0990
0.5338	9.0	549	0.6511	0.0035	0.2764	0.0944	0.0831	0.0989
0.538	10.0	610	0.6507	0.0035	0.2762	0.0946	0.0831	0.0986

Framework versions

• Transformers 4.51.0
• Pytorch 2.7.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.1