Hugging Face's logo

igitsml
/
speaker-segmentation-darija2

Transformers
Safetensors
pyannet
Generated from Trainer
Model card Files
xet
 Community
igitsml commited on
Commit
73bfcaf
·
verified ·
1 Parent(s): 9290861

End of training

Browse files
Files changed (4) hide show
  1. README.md +102 -3
  2. config.json +18 -0
  3. model.safetensors +3 -0
  4. training_args.bin +3 -0
README.md CHANGED
@@ -1,3 +1,102 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ language:
4
+ - ar
5
+ license: mit
6
+ base_model: pyannote/speaker-diarization-3.1
7
+ tags:
8
+ - speaker-diarization
9
+ - speaker-segmentation
10
+ - generated_from_trainer
11
+ datasets:
12
+ - igitsml/darija-synthetic-calls
13
+ model-index:
14
+ - name: speaker-segmentation-fine-tuned-darija
15
+ results: []
16
+ ---
17
+
18
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
19
+ should probably proofread and complete it, then remove this comment. -->
20
+
21
+ # speaker-segmentation-fine-tuned-darija
22
+
23
+ This model is a fine-tuned version of [pyannote/speaker-diarization-3.1](https://huggingface.co/pyannote/speaker-diarization-3.1) on the igitsml/darija-synthetic-calls dataset.
24
+ It achieves the following results on the evaluation set:
25
+ - Loss: 0.3338
26
+ - Model Preparation Time: 0.0061
27
+ - Der: 0.1220
28
+ - False Alarm: 0.0235
29
+ - Missed Detection: 0.0296
30
+ - Confusion: 0.0688
31
+
32
+ ## Model description
33
+
34
+ More information needed
35
+
36
+ ## Intended uses & limitations
37
+
38
+ More information needed
39
+
40
+ ## Training and evaluation data
41
+
42
+ More information needed
43
+
44
+ ## Training procedure
45
+
46
+ ### Training hyperparameters
47
+
48
+ The following hyperparameters were used during training:
49
+ - learning_rate: 1e-05
50
+ - train_batch_size: 16
51
+ - eval_batch_size: 8
52
+ - seed: 42
53
+ - gradient_accumulation_steps: 2
54
+ - total_train_batch_size: 32
55
+ - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
56
+ - lr_scheduler_type: cosine
57
+ - lr_scheduler_warmup_ratio: 0.1
58
+ - num_epochs: 30
59
+ - mixed_precision_training: Native AMP
60
+
61
+ ### Training results
62
+
63
+ | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion |
64
+ |:-------------:|:-----:|:-----:|:---------------:|:----------------------:|:------:|:-----------:|:----------------:|:---------:|
65
+ | 0.7164 | 1.0 | 683 | 0.8500 | 0.0061 | 0.2406 | 0.0365 | 0.0432 | 0.1609 |
66
+ | 0.6075 | 2.0 | 1366 | 0.6868 | 0.0061 | 0.2182 | 0.0361 | 0.0408 | 0.1413 |
67
+ | 0.5213 | 3.0 | 2049 | 0.5659 | 0.0061 | 0.1947 | 0.0329 | 0.0395 | 0.1224 |
68
+ | 0.4664 | 4.0 | 2732 | 0.5040 | 0.0061 | 0.1821 | 0.0306 | 0.0372 | 0.1143 |
69
+ | 0.411 | 5.0 | 3415 | 0.4678 | 0.0061 | 0.1738 | 0.0297 | 0.0355 | 0.1086 |
70
+ | 0.4205 | 6.0 | 4098 | 0.4503 | 0.0061 | 0.1682 | 0.0286 | 0.0348 | 0.1048 |
71
+ | 0.4133 | 7.0 | 4781 | 0.4330 | 0.0061 | 0.1629 | 0.0285 | 0.0336 | 0.1009 |
72
+ | 0.3936 | 8.0 | 5464 | 0.4191 | 0.0061 | 0.1579 | 0.0278 | 0.0329 | 0.0972 |
73
+ | 0.3799 | 9.0 | 6147 | 0.4080 | 0.0061 | 0.1529 | 0.0276 | 0.0323 | 0.0931 |
74
+ | 0.3557 | 10.0 | 6830 | 0.4007 | 0.0061 | 0.1500 | 0.0269 | 0.0317 | 0.0914 |
75
+ | 0.3564 | 11.0 | 7513 | 0.3915 | 0.0061 | 0.1465 | 0.0258 | 0.0319 | 0.0888 |
76
+ | 0.3658 | 12.0 | 8196 | 0.3853 | 0.0061 | 0.1433 | 0.0258 | 0.0314 | 0.0861 |
77
+ | 0.3606 | 13.0 | 8879 | 0.3784 | 0.0061 | 0.1408 | 0.0255 | 0.0311 | 0.0842 |
78
+ | 0.3685 | 14.0 | 9562 | 0.3739 | 0.0061 | 0.1390 | 0.0255 | 0.0308 | 0.0827 |
79
+ | 0.3364 | 15.0 | 10245 | 0.3706 | 0.0061 | 0.1378 | 0.0253 | 0.0306 | 0.0818 |
80
+ | 0.3436 | 16.0 | 10928 | 0.3698 | 0.0061 | 0.1369 | 0.0248 | 0.0307 | 0.0814 |
81
+ | 0.3339 | 17.0 | 11611 | 0.3636 | 0.0061 | 0.1353 | 0.0249 | 0.0304 | 0.0799 |
82
+ | 0.3416 | 18.0 | 12294 | 0.3615 | 0.0061 | 0.1343 | 0.0246 | 0.0304 | 0.0792 |
83
+ | 0.3396 | 19.0 | 12977 | 0.3593 | 0.0061 | 0.1337 | 0.0243 | 0.0305 | 0.0789 |
84
+ | 0.344 | 20.0 | 13660 | 0.3572 | 0.0061 | 0.1330 | 0.0243 | 0.0305 | 0.0782 |
85
+ | 0.3372 | 21.0 | 14343 | 0.3541 | 0.0061 | 0.1320 | 0.0245 | 0.0302 | 0.0773 |
86
+ | 0.3271 | 22.0 | 15026 | 0.3549 | 0.0061 | 0.1313 | 0.0242 | 0.0302 | 0.0768 |
87
+ | 0.3206 | 23.0 | 15709 | 0.3516 | 0.0061 | 0.1310 | 0.0243 | 0.0301 | 0.0766 |
88
+ | 0.3359 | 24.0 | 16392 | 0.3524 | 0.0061 | 0.1308 | 0.0242 | 0.0301 | 0.0765 |
89
+ | 0.322 | 25.0 | 17075 | 0.3512 | 0.0061 | 0.1304 | 0.0241 | 0.0301 | 0.0762 |
90
+ | 0.3169 | 26.0 | 17758 | 0.3507 | 0.0061 | 0.1301 | 0.0243 | 0.0300 | 0.0758 |
91
+ | 0.3351 | 27.0 | 18441 | 0.3508 | 0.0061 | 0.1300 | 0.0243 | 0.0299 | 0.0758 |
92
+ | 0.3221 | 28.0 | 19124 | 0.3501 | 0.0061 | 0.1300 | 0.0243 | 0.0299 | 0.0758 |
93
+ | 0.324 | 29.0 | 19807 | 0.3499 | 0.0061 | 0.1300 | 0.0243 | 0.0299 | 0.0758 |
94
+ | 0.3271 | 30.0 | 20490 | 0.3499 | 0.0061 | 0.1300 | 0.0243 | 0.0299 | 0.0758 |
95
+
96
+
97
+ ### Framework versions
98
+
99
+ - Transformers 4.57.3
100
+ - Pytorch 2.5.1+cu121
101
+ - Datasets 4.4.2
102
+ - Tokenizers 0.22.2
config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SegmentationModel"
4
+ ],
5
+ "chunk_duration": 10.0,
6
+ "dtype": "float32",
7
+ "max_speakers_per_chunk": 3,
8
+ "max_speakers_per_frame": 2,
9
+ "min_duration": null,
10
+ "model_type": "pyannet",
11
+ "sample_rate": 16000,
12
+ "transformers_version": "4.57.3",
13
+ "warm_up": [
14
+ 0.0,
15
+ 0.0
16
+ ],
17
+ "weigh_by_cardinality": false
18
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fe45631ad5f1fbf6a38ccfd9a608a7eee2263334a7aa31d268b806f1b8ae4e1
3
+ size 5899124
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f2c6015f30f4b8bb1ddd75b49f5e7a4365be08988344e1ecbcd40052627b955
3
+ size 5496