Hugging Face's logo

mehdie
/
ancient_semitic_bert

Fill-Mask
Transformers
Safetensors
Arabic
Hebrew
bert
Generated from Trainer
Inference Endpoints
Model card Files Community
morten-j commited on
Commit
62a73e4
•
1 Parent(s): 6ba1014

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +98 -2
README.md CHANGED
@@ -1,9 +1,105 @@
1
  ---
2
- license: apache-2.0
 
 
 
 
3
  datasets:
4
  - bigscience-data/roots_ar_openiti_proc
5
  - mehdie/sefaria
6
  language:
7
  - ar
8
  - he
9
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ tags:
3
+ - generated_from_trainer
4
+ model-index:
5
+ - name: ancient_semitic_bert
6
+ results: []
7
  datasets:
8
  - bigscience-data/roots_ar_openiti_proc
9
  - mehdie/sefaria
10
  language:
11
  - ar
12
  - he
13
+ ---
14
+
15
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
16
+ should probably proofread and complete it, then remove this comment. -->
17
+
18
+ # ancient_semitic_bert
19
+
20
+ This model is a fine-tuned version of [](https://huggingface.co/) on the None dataset.
21
+ It achieves the following results on the evaluation set:
22
+ - Loss: 1.9118
23
+ - Perplexity: 6.77 (40 Epochs)
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 5e-05
42
+ - train_batch_size: 16
43
+ - eval_batch_size: 16
44
+ - seed: 42
45
+ - distributed_type: multi-GPU
46
+ - num_devices: 4
47
+ - total_train_batch_size: 64
48
+ - total_eval_batch_size: 64
49
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
+ - lr_scheduler_type: linear
51
+ - lr_scheduler_warmup_steps: 10000
52
+ - num_epochs: 40.0
53
+
54
+ ### Training results
55
+
56
+ | Training Loss | Epoch | Step | Validation Loss |
57
+ |:-------------:|:-----:|:-------:|:---------------:|
58
+ | 6.778 | 1.0 | 55319 | 6.4618 |
59
+ | 6.4271 | 2.0 | 110638 | 6.3701 |
60
+ | 6.3616 | 3.0 | 165957 | 6.3217 |
61
+ | 6.3257 | 4.0 | 221276 | 6.2966 |
62
+ | 6.3001 | 5.0 | 276595 | 6.2759 |
63
+ | 6.2834 | 6.0 | 331914 | 6.2610 |
64
+ | 6.2699 | 7.0 | 387233 | 6.2465 |
65
+ | 6.2565 | 8.0 | 442552 | 6.1939 |
66
+ | 6.2221 | 9.0 | 497871 | 6.1154 |
67
+ | 6.0721 | 10.0 | 553190 | 5.9524 |
68
+ | 5.9212 | 11.0 | 608509 | 5.7947 |
69
+ | 5.8113 | 12.0 | 663828 | 5.7161 |
70
+ | 5.7509 | 13.0 | 719147 | 5.6614 |
71
+ | 5.7053 | 14.0 | 774466 | 5.6158 |
72
+ | 5.6665 | 15.0 | 829785 | 5.5774 |
73
+ | 5.634 | 16.0 | 885104 | 5.5448 |
74
+ | 5.6055 | 17.0 | 940423 | 2.7563 |
75
+ | 3.3308 | 18.0 | 995742 | 2.5443 |
76
+ | 2.6179 | 19.0 | 1051061 | 2.4196 |
77
+ | 2.5324 | 20.0 | 1106380 | 2.3393 |
78
+ | 2.4791 | 21.0 | 1161699 | 2.2755 |
79
+ | 2.4105 | 22.0 | 1217018 | 2.2241 |
80
+ | 2.3582 | 23.0 | 1272337 | 2.1772 |
81
+ | 2.3281 | 24.0 | 1327656 | 2.1416 |
82
+ | 2.2987 | 25.0 | 1382975 | 2.1137 |
83
+ | 2.7859 | 26.0 | 1438294 | 2.0950 |
84
+ | 2.2728 | 27.0 | 1493613 | 2.0685 |
85
+ | 2.2308 | 28.0 | 1548932 | 2.0499 |
86
+ | 2.1739 | 29.0 | 1604251 | 2.0082 |
87
+ | 2.1569 | 30.0 | 1659570 | 1.9939 |
88
+ | 2.1425 | 31.0 | 1714889 | 1.9802 |
89
+ | 2.1318 | 32.0 | 1770208 | 1.9669 |
90
+ | 2.1207 | 33.0 | 1825527 | 1.9583 |
91
+ | 2.1111 | 34.0 | 1880846 | 1.9477 |
92
+ | 2.102 | 35.0 | 1936165 | 1.9409 |
93
+ | 2.0943 | 36.0 | 1991484 | 1.9313 |
94
+ | 2.0871 | 37.0 | 2046803 | 1.9236 |
95
+ | 2.0736 | 38.0 | 2102122 | 1.9191 |
96
+ | 2.0693 | 39.0 | 2157441 | 1.9147 |
97
+ | 2.0653 | 40.0 | 2212760 | 1.9118 |
98
+
99
+
100
+ ### Framework versions
101
+
102
+ - Transformers 4.38.2
103
+ - Pytorch 2.3.0a0+ebedce2
104
+ - Datasets 2.17.1
105
+ - Tokenizers 0.15.2