ppxyn1
/

speaker-segmentation-fine-tuned-korean2

speaker-diarization

speaker-segmentation

Generated from Trainer

Model card Files Files and versions

ppxyn1 commited on Nov 17, 2025

Commit

438c474

·

verified ·

1 Parent(s): 0d5f369

End of training

Files changed (1) hide show

README.md +9 -10

README.md CHANGED Viewed

@@ -22,12 +22,12 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [pyannote/speaker-diarization-3.0](https://huggingface.co/pyannote/speaker-diarization-3.0) on the test_data dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2770
-- Model Preparation Time: 0.002
-- Der: 0.0744
 - False Alarm: 0.0
-- Missed Detection: 0.0017
-- Confusion: 0.0727
 ## Model description
@@ -52,16 +52,15 @@ The following hyperparameters were used during training:
 - seed: 42
 - optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
-- num_epochs: 5
 ### Training results
 | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der    | False Alarm | Missed Detection | Confusion |
 |:-------------:|:-----:|:----:|:---------------:|:----------------------:|:------:|:-----------:|:----------------:|:---------:|
-| 0.06          | 1.0   | 307  | 0.0716          | 0.002                  | 0.0136 | 0.0000      | 0.0019           | 0.0117    |
-| 0.0366        | 2.0   | 614  | 0.0778          | 0.002                  | 0.0147 | 0.0000      | 0.0019           | 0.0127    |
-| 0.0453        | 3.0   | 921  | 0.0778          | 0.002                  | 0.0136 | 0.0000      | 0.0019           | 0.0116    |
-| 0.0176        | 4.0   | 1228 | 0.0793          | 0.002                  | 0.0138 | 0.0000      | 0.0019           | 0.0119    |
 ### Framework versions

 This model is a fine-tuned version of [pyannote/speaker-diarization-3.0](https://huggingface.co/pyannote/speaker-diarization-3.0) on the test_data dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.0144
+- Model Preparation Time: 0.0018
+- Der: 0.0050
 - False Alarm: 0.0
+- Missed Detection: 0.0011
+- Confusion: 0.0039
 ## Model description
 - seed: 42
 - optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
+- num_epochs: 3
 ### Training results
 | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der    | False Alarm | Missed Detection | Confusion |
 |:-------------:|:-----:|:----:|:---------------:|:----------------------:|:------:|:-----------:|:----------------:|:---------:|
+| 0.0449        | 1.0   | 366  | 0.0436          | 0.0018                 | 0.0156 | 0.0000      | 0.0015           | 0.0141    |
+| 0.0768        | 2.0   | 732  | 0.0432          | 0.0018                 | 0.0119 | 0.0         | 0.0015           | 0.0103    |
+| 0.02          | 3.0   | 1098 | 0.0375          | 0.0018                 | 0.0108 | 0.0         | 0.0015           | 0.0093    |
 ### Framework versions