Hugging Face's logo

briangilbert
/
speaker-segmentation-fine-tuned-callhome-eng

Transformers
TensorBoard
Safetensors
pyannet
speaker-diarization
speaker-segmentation
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
briangilbert commited on
Commit
8ef8aa5
•
1 Parent(s): a873afc

End of training

Browse files
Files changed (3) hide show
  1. README.md +15 -30
  2. model.safetensors +1 -1
  3. runs/Jul12_09-49-28_775672ca2d10/events.out.tfevents.1720778126.775672ca2d10.1966.0 +2 -2
README.md CHANGED
@@ -19,11 +19,11 @@ should probably proofread and complete it, then remove this comment. -->
19
 
20
  This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the diarizers-community/callhome eng dataset.
21
  It achieves the following results on the evaluation set:
22
- - Loss: 0.4823
23
- - Der: 0.1830
24
- - False Alarm: 0.0620
25
- - Missed Detection: 0.0697
26
- - Confusion: 0.0513
27
 
28
  ## Model description
29
 
@@ -43,42 +43,27 @@ More information needed
43
 
44
  The following hyperparameters were used during training:
45
  - learning_rate: 0.001
46
- - train_batch_size: 64
47
- - eval_batch_size: 64
48
  - seed: 42
49
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
  - lr_scheduler_type: cosine
51
- - num_epochs: 20.0
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
56
  |:-------------:|:-----:|:----:|:---------------:|:------:|:-----------:|:----------------:|:---------:|
57
- | 0.4474 | 1.0 | 181 | 0.4819 | 0.1951 | 0.0640 | 0.0729 | 0.0582 |
58
- | 0.403 | 2.0 | 362 | 0.4771 | 0.1875 | 0.0524 | 0.0789 | 0.0562 |
59
- | 0.389 | 3.0 | 543 | 0.4636 | 0.1827 | 0.0506 | 0.0782 | 0.0539 |
60
- | 0.3715 | 4.0 | 724 | 0.4553 | 0.1811 | 0.0547 | 0.0733 | 0.0531 |
61
- | 0.3722 | 5.0 | 905 | 0.4699 | 0.1833 | 0.0516 | 0.0769 | 0.0548 |
62
- | 0.3625 | 6.0 | 1086 | 0.4677 | 0.1831 | 0.0585 | 0.0714 | 0.0531 |
63
- | 0.351 | 7.0 | 1267 | 0.4729 | 0.1800 | 0.0532 | 0.0750 | 0.0518 |
64
- | 0.3405 | 8.0 | 1448 | 0.4707 | 0.1817 | 0.0550 | 0.0749 | 0.0518 |
65
- | 0.3378 | 9.0 | 1629 | 0.4654 | 0.1820 | 0.0575 | 0.0725 | 0.0519 |
66
- | 0.3312 | 10.0 | 1810 | 0.4745 | 0.1821 | 0.0566 | 0.0739 | 0.0516 |
67
- | 0.3283 | 11.0 | 1991 | 0.4768 | 0.1832 | 0.0618 | 0.0695 | 0.0520 |
68
- | 0.3225 | 12.0 | 2172 | 0.4872 | 0.1850 | 0.0647 | 0.0680 | 0.0523 |
69
- | 0.3205 | 13.0 | 2353 | 0.4840 | 0.1857 | 0.0628 | 0.0692 | 0.0537 |
70
- | 0.3129 | 14.0 | 2534 | 0.4782 | 0.1827 | 0.0644 | 0.0678 | 0.0505 |
71
- | 0.3145 | 15.0 | 2715 | 0.4798 | 0.1817 | 0.0597 | 0.0712 | 0.0507 |
72
- | 0.3109 | 16.0 | 2896 | 0.4803 | 0.1822 | 0.0631 | 0.0685 | 0.0506 |
73
- | 0.3086 | 17.0 | 3077 | 0.4813 | 0.1827 | 0.0622 | 0.0694 | 0.0511 |
74
- | 0.3095 | 18.0 | 3258 | 0.4813 | 0.1830 | 0.0618 | 0.0699 | 0.0513 |
75
- | 0.3109 | 19.0 | 3439 | 0.4823 | 0.1829 | 0.0620 | 0.0697 | 0.0512 |
76
- | 0.3115 | 20.0 | 3620 | 0.4823 | 0.1830 | 0.0620 | 0.0697 | 0.0513 |
77
 
78
 
79
  ### Framework versions
80
 
81
  - Transformers 4.41.2
82
- - Pytorch 2.3.1+cu121
83
- - Datasets 2.19.2
84
  - Tokenizers 0.19.1
 
19
 
20
  This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the diarizers-community/callhome eng dataset.
21
  It achieves the following results on the evaluation set:
22
+ - Loss: 0.4632
23
+ - Der: 0.1810
24
+ - False Alarm: 0.0575
25
+ - Missed Detection: 0.0707
26
+ - Confusion: 0.0528
27
 
28
  ## Model description
29
 
 
43
 
44
  The following hyperparameters were used during training:
45
  - learning_rate: 0.001
46
+ - train_batch_size: 32
47
+ - eval_batch_size: 32
48
  - seed: 42
49
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
  - lr_scheduler_type: cosine
51
+ - num_epochs: 5.0
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
56
  |:-------------:|:-----:|:----:|:---------------:|:------:|:-----------:|:----------------:|:---------:|
57
+ | 0.4223 | 1.0 | 362 | 0.4718 | 0.1890 | 0.0617 | 0.0724 | 0.0549 |
58
+ | 0.3832 | 2.0 | 724 | 0.4581 | 0.1836 | 0.0561 | 0.0749 | 0.0526 |
59
+ | 0.3812 | 3.0 | 1086 | 0.4676 | 0.1846 | 0.0598 | 0.0702 | 0.0546 |
60
+ | 0.3607 | 4.0 | 1448 | 0.4610 | 0.1819 | 0.0585 | 0.0701 | 0.0533 |
61
+ | 0.3581 | 5.0 | 1810 | 0.4632 | 0.1810 | 0.0575 | 0.0707 | 0.0528 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
62
 
63
 
64
  ### Framework versions
65
 
66
  - Transformers 4.41.2
67
+ - Pytorch 2.3.0+cu121
68
+ - Datasets 2.20.0
69
  - Tokenizers 0.19.1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4bcfa6593f4aeab5f21d3e86408434b68718230fb64070a7ec013c7a5c967b80
3
  size 5899124
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9d562fdd051ebfb02bd3150c1058684481cccd9b1f124cb4736b68b37b5d82c
3
  size 5899124
runs/Jul12_09-49-28_775672ca2d10/events.out.tfevents.1720778126.775672ca2d10.1966.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3ba94f7f4ca5881166544e0153ec419b158eaf05ecaecae422cdc88b831e9cc2
3
- size 9835
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0700b50cfad8c9fc09169ac726aae2bfd6d6814555a390e3962cedecc4033e57
3
+ size 11519