jstAnotherCapi commited on
Commit
4a611c2
·
verified ·
1 Parent(s): d2a5f53

Upload folder using huggingface_hub

Browse files
epoch4/error.log ADDED
The diff for this file is too large to render. See raw diff
 
epoch4/logs.txt ADDED
@@ -0,0 +1,474 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2026-03-02 17:54:30 - Load pretrained SentenceTransformer: bert-base-arabertv02
2
+ 2026-03-02 17:54:41 - '[Errno -2] Name or service not known' thrown while requesting HEAD https://huggingface.co/bert-base-arabertv02/resolve/main/./modules.json
3
+ 2026-03-02 17:54:41 - Retrying in 1s [Retry 1/5].
4
+ 2026-03-02 17:54:42 - No sentence-transformers model found with name bert-base-arabertv02. Creating a new one with mean pooling.
5
+ 2026-03-02 17:56:12 - Use pytorch device_name: cuda:0
6
+ 2026-03-02 17:56:12 - Load pretrained SentenceTransformer: /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-61786
7
+ {'loss': '0.4484', 'grad_norm': '4.876', 'learning_rate': '1.111e-05', 'epoch': '2'}
8
+ {'loss': '0.4408', 'grad_norm': '5.164', 'learning_rate': '1.107e-05', 'epoch': '2.007'}
9
+ {'loss': '0.4188', 'grad_norm': '3.438', 'learning_rate': '1.104e-05', 'epoch': '2.013'}
10
+ {'loss': '0.4396', 'grad_norm': '7.06', 'learning_rate': '1.1e-05', 'epoch': '2.02'}
11
+ {'loss': '0.4423', 'grad_norm': '4.074', 'learning_rate': '1.096e-05', 'epoch': '2.026'}
12
+ {'loss': '0.4464', 'grad_norm': '6.44', 'learning_rate': '1.093e-05', 'epoch': '2.033'}
13
+ {'loss': '0.4318', 'grad_norm': '3.624', 'learning_rate': '1.089e-05', 'epoch': '2.039'}
14
+ {'loss': '0.4413', 'grad_norm': '3.332', 'learning_rate': '1.086e-05', 'epoch': '2.046'}
15
+ {'loss': '0.4612', 'grad_norm': '4.731', 'learning_rate': '1.082e-05', 'epoch': '2.052'}
16
+ {'loss': '0.4406', 'grad_norm': '5.413', 'learning_rate': '1.079e-05', 'epoch': '2.059'}
17
+ {'loss': '0.4376', 'grad_norm': '3.114', 'learning_rate': '1.075e-05', 'epoch': '2.065'}
18
+ {'loss': '0.445', 'grad_norm': '4.638', 'learning_rate': '1.071e-05', 'epoch': '2.072'}
19
+ {'loss': '0.4373', 'grad_norm': '3.384', 'learning_rate': '1.068e-05', 'epoch': '2.078'}
20
+ {'loss': '0.4227', 'grad_norm': '5.522', 'learning_rate': '1.064e-05', 'epoch': '2.085'}
21
+ {'loss': '0.4499', 'grad_norm': '2.805', 'learning_rate': '1.061e-05', 'epoch': '2.091'}
22
+ {'loss': '0.4411', 'grad_norm': '5.262', 'learning_rate': '1.057e-05', 'epoch': '2.098'}
23
+ {'loss': '0.4506', 'grad_norm': '4.764', 'learning_rate': '1.053e-05', 'epoch': '2.104'}
24
+ {'loss': '0.4066', 'grad_norm': '2.191', 'learning_rate': '1.05e-05', 'epoch': '2.111'}
25
+ {'loss': '0.4325', 'grad_norm': '5.071', 'learning_rate': '1.046e-05', 'epoch': '2.117'}
26
+ {'loss': '0.4124', 'grad_norm': '5.363', 'learning_rate': '1.043e-05', 'epoch': '2.123'}
27
+ {'loss': '0.4169', 'grad_norm': '3.958', 'learning_rate': '1.039e-05', 'epoch': '2.13'}
28
+ {'loss': '0.4166', 'grad_norm': '2.913', 'learning_rate': '1.035e-05', 'epoch': '2.136'}
29
+ 2026-03-02 19:28:04 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 2.1364085133932185 after 66000 steps (truncated to 768):
30
+ 2026-03-02 19:49:48 - Accuracy Cosine Similarity: 98.05%
31
+ 2026-03-02 19:49:48 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 2.1364085133932185 after 66000 steps (truncated to 512):
32
+ 2026-03-02 20:10:59 - Accuracy Cosine Similarity: 98.06%
33
+ 2026-03-02 20:10:59 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 2.1364085133932185 after 66000 steps (truncated to 256):
34
+ 2026-03-02 20:32:00 - Accuracy Cosine Similarity: 98.06%
35
+ 2026-03-02 20:32:00 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 2.1364085133932185 after 66000 steps (truncated to 128):
36
+ 2026-03-02 20:52:44 - Accuracy Cosine Similarity: 98.03%
37
+ 2026-03-02 20:52:44 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 2.1364085133932185 after 66000 steps (truncated to 64):
38
+ 2026-03-02 21:13:05 - Accuracy Cosine Similarity: 97.93%
39
+ {'eval_train_loss': '0.2883', 'eval_dev-768_cosine_accuracy': '0.9805', 'eval_dev-512_cosine_accuracy': '0.9806', 'eval_dev-256_cosine_accuracy': '0.9806', 'eval_dev-128_cosine_accuracy': '0.9803', 'eval_dev-64_cosine_accuracy': '0.9793', 'eval_sequential_score': '0.9805', 'eval_train_runtime': '9366', 'eval_train_samples_per_second': '120.6', 'eval_train_steps_per_second': '15.08', 'epoch': '2.136'}
40
+ 2026-03-02 21:13:05 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-66000
41
+ 2026-03-02 21:13:05 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-66000
42
+ {'loss': '0.4069', 'grad_norm': '4.696', 'learning_rate': '1.032e-05', 'epoch': '2.143'}
43
+ {'loss': '0.4144', 'grad_norm': '4.022', 'learning_rate': '1.028e-05', 'epoch': '2.149'}
44
+ {'loss': '0.3948', 'grad_norm': '3.288', 'learning_rate': '1.025e-05', 'epoch': '2.156'}
45
+ {'loss': '0.4284', 'grad_norm': '2.968', 'learning_rate': '1.021e-05', 'epoch': '2.162'}
46
+ {'loss': '0.3984', 'grad_norm': '3.579', 'learning_rate': '1.017e-05', 'epoch': '2.169'}
47
+ {'loss': '0.4015', 'grad_norm': '5.276', 'learning_rate': '1.014e-05', 'epoch': '2.175'}
48
+ {'loss': '0.3999', 'grad_norm': '3.263', 'learning_rate': '1.01e-05', 'epoch': '2.182'}
49
+ {'loss': '0.397', 'grad_norm': '4.018', 'learning_rate': '1.007e-05', 'epoch': '2.188'}
50
+ {'loss': '0.3605', 'grad_norm': '3.304', 'learning_rate': '1.003e-05', 'epoch': '2.195'}
51
+ {'loss': '0.3851', 'grad_norm': '5.208', 'learning_rate': '9.994e-06', 'epoch': '2.201'}
52
+ {'loss': '0.397', 'grad_norm': '2.203', 'learning_rate': '9.958e-06', 'epoch': '2.208'}
53
+ {'loss': '0.3912', 'grad_norm': '3.873', 'learning_rate': '9.922e-06', 'epoch': '2.214'}
54
+ {'loss': '0.3772', 'grad_norm': '6.111', 'learning_rate': '9.886e-06', 'epoch': '2.221'}
55
+ {'loss': '0.3782', 'grad_norm': '4.189', 'learning_rate': '9.85e-06', 'epoch': '2.227'}
56
+ {'loss': '0.378', 'grad_norm': '4.917', 'learning_rate': '9.814e-06', 'epoch': '2.234'}
57
+ {'loss': '0.3605', 'grad_norm': '1.853', 'learning_rate': '9.778e-06', 'epoch': '2.24'}
58
+ {'loss': '0.362', 'grad_norm': '3.2', 'learning_rate': '9.742e-06', 'epoch': '2.246'}
59
+ {'loss': '0.3717', 'grad_norm': '4.389', 'learning_rate': '9.706e-06', 'epoch': '2.253'}
60
+ {'loss': '0.3656', 'grad_norm': '2.031', 'learning_rate': '9.67e-06', 'epoch': '2.259'}
61
+ {'loss': '0.3766', 'grad_norm': '5.601', 'learning_rate': '9.634e-06', 'epoch': '2.266'}
62
+ {'loss': '0.3768', 'grad_norm': '4.343', 'learning_rate': '9.598e-06', 'epoch': '2.272'}
63
+ {'loss': '0.3853', 'grad_norm': '3.466', 'learning_rate': '9.562e-06', 'epoch': '2.279'}
64
+ {'loss': '0.3444', 'grad_norm': '3.306', 'learning_rate': '9.526e-06', 'epoch': '2.285'}
65
+ {'loss': '0.3863', 'grad_norm': '4.16', 'learning_rate': '9.49e-06', 'epoch': '2.292'}
66
+ {'loss': '0.3843', 'grad_norm': '2.997', 'learning_rate': '9.454e-06', 'epoch': '2.298'}
67
+ {'loss': '0.362', 'grad_norm': '3.893', 'learning_rate': '9.418e-06', 'epoch': '2.305'}
68
+ {'loss': '0.374', 'grad_norm': '4.415', 'learning_rate': '9.382e-06', 'epoch': '2.311'}
69
+ {'loss': '0.3657', 'grad_norm': '4.601', 'learning_rate': '9.346e-06', 'epoch': '2.318'}
70
+ {'loss': '0.3838', 'grad_norm': '3.083', 'learning_rate': '9.311e-06', 'epoch': '2.324'}
71
+ {'loss': '0.357', 'grad_norm': '5.532', 'learning_rate': '9.275e-06', 'epoch': '2.331'}
72
+ 2026-03-02 23:01:58 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 2.3306304119122765 after 72000 steps (truncated to 768):
73
+ 2026-03-02 23:22:39 - Accuracy Cosine Similarity: 98.15%
74
+ 2026-03-02 23:22:39 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 2.3306304119122765 after 72000 steps (truncated to 512):
75
+ 2026-03-02 23:43:44 - Accuracy Cosine Similarity: 98.16%
76
+ 2026-03-02 23:43:44 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 2.3306304119122765 after 72000 steps (truncated to 256):
77
+ 2026-03-03 00:04:58 - Accuracy Cosine Similarity: 98.17%
78
+ 2026-03-03 00:04:58 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 2.3306304119122765 after 72000 steps (truncated to 128):
79
+ 2026-03-03 00:25:36 - Accuracy Cosine Similarity: 98.13%
80
+ 2026-03-03 00:25:36 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 2.3306304119122765 after 72000 steps (truncated to 64):
81
+ 2026-03-03 00:46:23 - Accuracy Cosine Similarity: 98.01%
82
+ {'eval_train_loss': '0.2768', 'eval_dev-768_cosine_accuracy': '0.9815', 'eval_dev-512_cosine_accuracy': '0.9816', 'eval_dev-256_cosine_accuracy': '0.9817', 'eval_dev-128_cosine_accuracy': '0.9813', 'eval_dev-64_cosine_accuracy': '0.9801', 'eval_sequential_score': '0.9815', 'eval_train_runtime': '9309', 'eval_train_samples_per_second': '121.4', 'eval_train_steps_per_second': '15.17', 'epoch': '2.331'}
83
+ 2026-03-03 00:46:23 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-72000
84
+ 2026-03-03 00:46:23 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-72000
85
+ {'loss': '0.3562', 'grad_norm': '4.382', 'learning_rate': '9.239e-06', 'epoch': '2.337'}
86
+ {'loss': '0.3854', 'grad_norm': '5.293', 'learning_rate': '9.203e-06', 'epoch': '2.344'}
87
+ {'loss': '0.3574', 'grad_norm': '5.037', 'learning_rate': '9.167e-06', 'epoch': '2.35'}
88
+ {'loss': '0.3789', 'grad_norm': '3.6', 'learning_rate': '9.131e-06', 'epoch': '2.357'}
89
+ {'loss': '0.3708', 'grad_norm': '4.49', 'learning_rate': '9.095e-06', 'epoch': '2.363'}
90
+ {'loss': '0.3688', 'grad_norm': '4.346', 'learning_rate': '9.059e-06', 'epoch': '2.369'}
91
+ {'loss': '0.3764', 'grad_norm': '3.589', 'learning_rate': '9.023e-06', 'epoch': '2.376'}
92
+ {'loss': '0.3776', 'grad_norm': '5.037', 'learning_rate': '8.987e-06', 'epoch': '2.382'}
93
+ {'loss': '0.3752', 'grad_norm': '3.274', 'learning_rate': '8.951e-06', 'epoch': '2.389'}
94
+ {'loss': '0.3575', 'grad_norm': '3.484', 'learning_rate': '8.915e-06', 'epoch': '2.395'}
95
+ {'loss': '0.356', 'grad_norm': '4.98', 'learning_rate': '8.879e-06', 'epoch': '2.402'}
96
+ {'loss': '0.3858', 'grad_norm': '3.007', 'learning_rate': '8.843e-06', 'epoch': '2.408'}
97
+ {'loss': '0.3644', 'grad_norm': '1.843', 'learning_rate': '8.807e-06', 'epoch': '2.415'}
98
+ {'loss': '0.394', 'grad_norm': '2.967', 'learning_rate': '8.771e-06', 'epoch': '2.421'}
99
+ {'loss': '0.3776', 'grad_norm': '4.421', 'learning_rate': '8.735e-06', 'epoch': '2.428'}
100
+ {'loss': '0.3467', 'grad_norm': '4.17', 'learning_rate': '8.699e-06', 'epoch': '2.434'}
101
+ {'loss': '0.3651', 'grad_norm': '4.477', 'learning_rate': '8.663e-06', 'epoch': '2.441'}
102
+ {'loss': '0.3865', 'grad_norm': '4.592', 'learning_rate': '8.627e-06', 'epoch': '2.447'}
103
+ {'loss': '0.3856', 'grad_norm': '2.638', 'learning_rate': '8.591e-06', 'epoch': '2.454'}
104
+ {'loss': '0.3693', 'grad_norm': '4.006', 'learning_rate': '8.555e-06', 'epoch': '2.46'}
105
+ {'loss': '0.361', 'grad_norm': '3.124', 'learning_rate': '8.519e-06', 'epoch': '2.467'}
106
+ {'loss': '0.3693', 'grad_norm': '3.233', 'learning_rate': '8.483e-06', 'epoch': '2.473'}
107
+ {'loss': '0.3708', 'grad_norm': '3.78', 'learning_rate': '8.447e-06', 'epoch': '2.48'}
108
+ {'loss': '0.3548', 'grad_norm': '3.921', 'learning_rate': '8.411e-06', 'epoch': '2.486'}
109
+ {'loss': '0.3767', 'grad_norm': '5.192', 'learning_rate': '8.375e-06', 'epoch': '2.492'}
110
+ {'loss': '0.3668', 'grad_norm': '3.988', 'learning_rate': '8.339e-06', 'epoch': '2.499'}
111
+ {'loss': '0.3734', 'grad_norm': '3.967', 'learning_rate': '8.303e-06', 'epoch': '2.505'}
112
+ {'loss': '0.3563', 'grad_norm': '2.322', 'learning_rate': '8.267e-06', 'epoch': '2.512'}
113
+ {'loss': '0.3842', 'grad_norm': '3.582', 'learning_rate': '8.232e-06', 'epoch': '2.518'}
114
+ {'loss': '0.3541', 'grad_norm': '5.197', 'learning_rate': '8.196e-06', 'epoch': '2.525'}
115
+ 2026-03-03 02:35:54 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 2.5248523104313345 after 78000 steps (truncated to 768):
116
+ 2026-03-03 02:56:30 - Accuracy Cosine Similarity: 98.20%
117
+ 2026-03-03 02:56:30 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 2.5248523104313345 after 78000 steps (truncated to 512):
118
+ 2026-03-03 03:17:13 - Accuracy Cosine Similarity: 98.21%
119
+ 2026-03-03 03:17:13 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 2.5248523104313345 after 78000 steps (truncated to 256):
120
+ 2026-03-03 03:38:15 - Accuracy Cosine Similarity: 98.22%
121
+ 2026-03-03 03:38:15 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 2.5248523104313345 after 78000 steps (truncated to 128):
122
+ 2026-03-03 03:59:20 - Accuracy Cosine Similarity: 98.18%
123
+ 2026-03-03 03:59:20 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 2.5248523104313345 after 78000 steps (truncated to 64):
124
+ 2026-03-03 04:19:52 - Accuracy Cosine Similarity: 98.10%
125
+ {'eval_train_loss': '0.2668', 'eval_dev-768_cosine_accuracy': '0.982', 'eval_dev-512_cosine_accuracy': '0.9821', 'eval_dev-256_cosine_accuracy': '0.9822', 'eval_dev-128_cosine_accuracy': '0.9818', 'eval_dev-64_cosine_accuracy': '0.981', 'eval_sequential_score': '0.982', 'eval_train_runtime': '9281', 'eval_train_samples_per_second': '121.7', 'eval_train_steps_per_second': '15.22', 'epoch': '2.525'}
126
+ 2026-03-03 04:19:52 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-78000
127
+ 2026-03-03 04:19:52 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-78000
128
+ {'loss': '0.3692', 'grad_norm': '5.604', 'learning_rate': '8.16e-06', 'epoch': '2.531'}
129
+ {'loss': '0.3609', 'grad_norm': '4.981', 'learning_rate': '8.124e-06', 'epoch': '2.538'}
130
+ {'loss': '0.3763', 'grad_norm': '5.883', 'learning_rate': '8.088e-06', 'epoch': '2.544'}
131
+ {'loss': '0.3684', 'grad_norm': '2.069', 'learning_rate': '8.052e-06', 'epoch': '2.551'}
132
+ {'loss': '0.3614', 'grad_norm': '4.359', 'learning_rate': '8.016e-06', 'epoch': '2.557'}
133
+ {'loss': '0.3789', 'grad_norm': '3.143', 'learning_rate': '7.98e-06', 'epoch': '2.564'}
134
+ {'loss': '0.3689', 'grad_norm': '3.077', 'learning_rate': '7.944e-06', 'epoch': '2.57'}
135
+ {'loss': '0.3745', 'grad_norm': '2.937', 'learning_rate': '7.908e-06', 'epoch': '2.577'}
136
+ {'loss': '0.3725', 'grad_norm': '5.283', 'learning_rate': '7.872e-06', 'epoch': '2.583'}
137
+ {'loss': '0.3654', 'grad_norm': '3.622', 'learning_rate': '7.836e-06', 'epoch': '2.59'}
138
+ {'loss': '0.3653', 'grad_norm': '4.478', 'learning_rate': '7.8e-06', 'epoch': '2.596'}
139
+ {'loss': '0.3898', 'grad_norm': '4.282', 'learning_rate': '7.764e-06', 'epoch': '2.603'}
140
+ {'loss': '0.3699', 'grad_norm': '5.231', 'learning_rate': '7.728e-06', 'epoch': '2.609'}
141
+ {'loss': '0.3734', 'grad_norm': '4.661', 'learning_rate': '7.692e-06', 'epoch': '2.615'}
142
+ {'loss': '0.3575', 'grad_norm': '5.115', 'learning_rate': '7.656e-06', 'epoch': '2.622'}
143
+ {'loss': '0.3776', 'grad_norm': '3.268', 'learning_rate': '7.62e-06', 'epoch': '2.628'}
144
+ {'loss': '0.3764', 'grad_norm': '3.383', 'learning_rate': '7.584e-06', 'epoch': '2.635'}
145
+ {'loss': '0.3692', 'grad_norm': '5.811', 'learning_rate': '7.548e-06', 'epoch': '2.641'}
146
+ {'loss': '0.3573', 'grad_norm': '3.134', 'learning_rate': '7.512e-06', 'epoch': '2.648'}
147
+ {'loss': '0.3535', 'grad_norm': '3.569', 'learning_rate': '7.476e-06', 'epoch': '2.654'}
148
+ {'loss': '0.3686', 'grad_norm': '5.449', 'learning_rate': '7.44e-06', 'epoch': '2.661'}
149
+ {'loss': '0.3763', 'grad_norm': '2.932', 'learning_rate': '7.404e-06', 'epoch': '2.667'}
150
+ {'loss': '0.3589', 'grad_norm': '4.469', 'learning_rate': '7.368e-06', 'epoch': '2.674'}
151
+ {'loss': '0.3764', 'grad_norm': '3.381', 'learning_rate': '7.332e-06', 'epoch': '2.68'}
152
+ {'loss': '0.3682', 'grad_norm': '4.811', 'learning_rate': '7.296e-06', 'epoch': '2.687'}
153
+ {'loss': '0.3618', 'grad_norm': '4.508', 'learning_rate': '7.26e-06', 'epoch': '2.693'}
154
+ {'loss': '0.3722', 'grad_norm': '3.616', 'learning_rate': '7.224e-06', 'epoch': '2.7'}
155
+ {'loss': '0.4015', 'grad_norm': '4.274', 'learning_rate': '7.188e-06', 'epoch': '2.706'}
156
+ {'loss': '0.3673', 'grad_norm': '4.678', 'learning_rate': '7.153e-06', 'epoch': '2.713'}
157
+ {'loss': '0.3715', 'grad_norm': '4.915', 'learning_rate': '7.117e-06', 'epoch': '2.719'}
158
+ 2026-03-03 06:09:38 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 2.7190742089503925 after 84000 steps (truncated to 768):
159
+ 2026-03-03 06:30:51 - Accuracy Cosine Similarity: 98.31%
160
+ 2026-03-03 06:30:51 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 2.7190742089503925 after 84000 steps (truncated to 512):
161
+ 2026-03-03 06:52:09 - Accuracy Cosine Similarity: 98.32%
162
+ 2026-03-03 06:52:09 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 2.7190742089503925 after 84000 steps (truncated to 256):
163
+ 2026-03-03 07:13:25 - Accuracy Cosine Similarity: 98.32%
164
+ 2026-03-03 07:13:25 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 2.7190742089503925 after 84000 steps (truncated to 128):
165
+ 2026-03-03 07:34:32 - Accuracy Cosine Similarity: 98.29%
166
+ 2026-03-03 07:34:32 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 2.7190742089503925 after 84000 steps (truncated to 64):
167
+ 2026-03-03 07:55:19 - Accuracy Cosine Similarity: 98.20%
168
+ {'eval_train_loss': '0.2525', 'eval_dev-768_cosine_accuracy': '0.9831', 'eval_dev-512_cosine_accuracy': '0.9832', 'eval_dev-256_cosine_accuracy': '0.9832', 'eval_dev-128_cosine_accuracy': '0.9829', 'eval_dev-64_cosine_accuracy': '0.982', 'eval_sequential_score': '0.9831', 'eval_train_runtime': '9385', 'eval_train_samples_per_second': '120.4', 'eval_train_steps_per_second': '15.05', 'epoch': '2.719'}
169
+ 2026-03-03 07:55:19 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-84000
170
+ 2026-03-03 07:55:19 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-84000
171
+ {'loss': '0.3694', 'grad_norm': '2.933', 'learning_rate': '7.081e-06', 'epoch': '2.726'}
172
+ {'loss': '0.365', 'grad_norm': '4.5', 'learning_rate': '7.045e-06', 'epoch': '2.732'}
173
+ {'loss': '0.3753', 'grad_norm': '4.921', 'learning_rate': '7.009e-06', 'epoch': '2.738'}
174
+ {'loss': '0.3554', 'grad_norm': '3.083', 'learning_rate': '6.973e-06', 'epoch': '2.745'}
175
+ {'loss': '0.3936', 'grad_norm': '5.051', 'learning_rate': '6.937e-06', 'epoch': '2.751'}
176
+ {'loss': '0.3735', 'grad_norm': '6.889', 'learning_rate': '6.901e-06', 'epoch': '2.758'}
177
+ {'loss': '0.3746', 'grad_norm': '3.799', 'learning_rate': '6.865e-06', 'epoch': '2.764'}
178
+ {'loss': '0.3619', 'grad_norm': '2.101', 'learning_rate': '6.829e-06', 'epoch': '2.771'}
179
+ {'loss': '0.3705', 'grad_norm': '4.39', 'learning_rate': '6.793e-06', 'epoch': '2.777'}
180
+ {'loss': '0.359', 'grad_norm': '2.37', 'learning_rate': '6.757e-06', 'epoch': '2.784'}
181
+ {'loss': '0.3682', 'grad_norm': '3.344', 'learning_rate': '6.721e-06', 'epoch': '2.79'}
182
+ {'loss': '0.3821', 'grad_norm': '3.043', 'learning_rate': '6.685e-06', 'epoch': '2.797'}
183
+ {'loss': '0.3624', 'grad_norm': '3.74', 'learning_rate': '6.649e-06', 'epoch': '2.803'}
184
+ {'loss': '0.3751', 'grad_norm': '3.94', 'learning_rate': '6.613e-06', 'epoch': '2.81'}
185
+ {'loss': '0.3955', 'grad_norm': '4.282', 'learning_rate': '6.577e-06', 'epoch': '2.816'}
186
+ {'loss': '0.3581', 'grad_norm': '4.607', 'learning_rate': '6.541e-06', 'epoch': '2.823'}
187
+ {'loss': '0.3588', 'grad_norm': '5.26', 'learning_rate': '6.505e-06', 'epoch': '2.829'}
188
+ {'loss': '0.3622', 'grad_norm': '4.333', 'learning_rate': '6.469e-06', 'epoch': '2.836'}
189
+ {'loss': '0.375', 'grad_norm': '3.19', 'learning_rate': '6.433e-06', 'epoch': '2.842'}
190
+ {'loss': '0.3589', 'grad_norm': '3.558', 'learning_rate': '6.397e-06', 'epoch': '2.849'}
191
+ {'loss': '0.3816', 'grad_norm': '2.926', 'learning_rate': '6.361e-06', 'epoch': '2.855'}
192
+ {'loss': '0.3624', 'grad_norm': '5.621', 'learning_rate': '6.325e-06', 'epoch': '2.862'}
193
+ {'loss': '0.3923', 'grad_norm': '4.943', 'learning_rate': '6.289e-06', 'epoch': '2.868'}
194
+ {'loss': '0.3798', 'grad_norm': '5.76', 'learning_rate': '6.253e-06', 'epoch': '2.874'}
195
+ {'loss': '0.3362', 'grad_norm': '2.747', 'learning_rate': '6.217e-06', 'epoch': '2.881'}
196
+ {'loss': '0.3766', 'grad_norm': '4.018', 'learning_rate': '6.181e-06', 'epoch': '2.887'}
197
+ {'loss': '0.3839', 'grad_norm': '4.481', 'learning_rate': '6.145e-06', 'epoch': '2.894'}
198
+ {'loss': '0.376', 'grad_norm': '3.558', 'learning_rate': '6.109e-06', 'epoch': '2.9'}
199
+ {'loss': '0.3958', 'grad_norm': '3.886', 'learning_rate': '6.074e-06', 'epoch': '2.907'}
200
+ {'loss': '0.372', 'grad_norm': '3.935', 'learning_rate': '6.038e-06', 'epoch': '2.913'}
201
+ 2026-03-03 09:45:15 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 2.9132961074694506 after 90000 steps (truncated to 768):
202
+ 2026-03-03 10:06:33 - Accuracy Cosine Similarity: 98.36%
203
+ 2026-03-03 10:06:33 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 2.9132961074694506 after 90000 steps (truncated to 512):
204
+ 2026-03-03 10:27:44 - Accuracy Cosine Similarity: 98.38%
205
+ 2026-03-03 10:27:44 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 2.9132961074694506 after 90000 steps (truncated to 256):
206
+ 2026-03-03 10:48:39 - Accuracy Cosine Similarity: 98.40%
207
+ 2026-03-03 10:48:39 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 2.9132961074694506 after 90000 steps (truncated to 128):
208
+ 2026-03-03 11:09:11 - Accuracy Cosine Similarity: 98.35%
209
+ 2026-03-03 11:09:11 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 2.9132961074694506 after 90000 steps (truncated to 64):
210
+ 2026-03-03 11:30:20 - Accuracy Cosine Similarity: 98.26%
211
+ {'eval_train_loss': '0.243', 'eval_dev-768_cosine_accuracy': '0.9836', 'eval_dev-512_cosine_accuracy': '0.9838', 'eval_dev-256_cosine_accuracy': '0.984', 'eval_dev-128_cosine_accuracy': '0.9835', 'eval_dev-64_cosine_accuracy': '0.9826', 'eval_sequential_score': '0.9836', 'eval_train_runtime': '9339', 'eval_train_samples_per_second': '121', 'eval_train_steps_per_second': '15.12', 'epoch': '2.913'}
212
+ 2026-03-03 11:30:20 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-90000
213
+ 2026-03-03 11:30:20 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-90000
214
+ {'loss': '0.3724', 'grad_norm': '4.557', 'learning_rate': '6.002e-06', 'epoch': '2.92'}
215
+ {'loss': '0.3708', 'grad_norm': '5.298', 'learning_rate': '5.966e-06', 'epoch': '2.926'}
216
+ {'loss': '0.3804', 'grad_norm': '4.007', 'learning_rate': '5.93e-06', 'epoch': '2.933'}
217
+ {'loss': '0.3647', 'grad_norm': '6.184', 'learning_rate': '5.894e-06', 'epoch': '2.939'}
218
+ {'loss': '0.3721', 'grad_norm': '2.113', 'learning_rate': '5.858e-06', 'epoch': '2.946'}
219
+ {'loss': '0.389', 'grad_norm': '4.014', 'learning_rate': '5.822e-06', 'epoch': '2.952'}
220
+ {'loss': '0.4107', 'grad_norm': '4.331', 'learning_rate': '5.786e-06', 'epoch': '2.959'}
221
+ {'loss': '0.3917', 'grad_norm': '5.432', 'learning_rate': '5.75e-06', 'epoch': '2.965'}
222
+ {'loss': '0.3608', 'grad_norm': '5.648', 'learning_rate': '5.714e-06', 'epoch': '2.972'}
223
+ {'loss': '0.386', 'grad_norm': '3.361', 'learning_rate': '5.678e-06', 'epoch': '2.978'}
224
+ {'loss': '0.3998', 'grad_norm': '4.219', 'learning_rate': '5.642e-06', 'epoch': '2.985'}
225
+ {'loss': '0.3889', 'grad_norm': '4.905', 'learning_rate': '5.606e-06', 'epoch': '2.991'}
226
+ {'loss': '0.3715', 'grad_norm': '5.136', 'learning_rate': '5.57e-06', 'epoch': '2.997'}
227
+ {'loss': '0.345', 'grad_norm': '4.286', 'learning_rate': '5.534e-06', 'epoch': '3.004'}
228
+ {'loss': '0.2915', 'grad_norm': '4.063', 'learning_rate': '5.498e-06', 'epoch': '3.01'}
229
+ {'loss': '0.2876', 'grad_norm': '5.342', 'learning_rate': '5.462e-06', 'epoch': '3.017'}
230
+ {'loss': '0.2979', 'grad_norm': '2.142', 'learning_rate': '5.426e-06', 'epoch': '3.023'}
231
+ {'loss': '0.2952', 'grad_norm': '2.973', 'learning_rate': '5.39e-06', 'epoch': '3.03'}
232
+ {'loss': '0.2925', 'grad_norm': '5.659', 'learning_rate': '5.354e-06', 'epoch': '3.036'}
233
+ {'loss': '0.2944', 'grad_norm': '4.593', 'learning_rate': '5.318e-06', 'epoch': '3.043'}
234
+ {'loss': '0.3033', 'grad_norm': '5.093', 'learning_rate': '5.282e-06', 'epoch': '3.049'}
235
+ {'loss': '0.2963', 'grad_norm': '3.233', 'learning_rate': '5.246e-06', 'epoch': '3.056'}
236
+ {'loss': '0.2835', 'grad_norm': '5.186', 'learning_rate': '5.21e-06', 'epoch': '3.062'}
237
+ {'loss': '0.2987', 'grad_norm': '3.581', 'learning_rate': '5.174e-06', 'epoch': '3.069'}
238
+ {'loss': '0.3', 'grad_norm': '4.349', 'learning_rate': '5.138e-06', 'epoch': '3.075'}
239
+ {'loss': '0.2845', 'grad_norm': '4.465', 'learning_rate': '5.102e-06', 'epoch': '3.082'}
240
+ {'loss': '0.2899', 'grad_norm': '3.619', 'learning_rate': '5.066e-06', 'epoch': '3.088'}
241
+ {'loss': '0.3078', 'grad_norm': '3.873', 'learning_rate': '5.03e-06', 'epoch': '3.095'}
242
+ {'loss': '0.2943', 'grad_norm': '4.048', 'learning_rate': '4.995e-06', 'epoch': '3.101'}
243
+ {'loss': '0.2758', 'grad_norm': '3.275', 'learning_rate': '4.959e-06', 'epoch': '3.108'}
244
+ 2026-03-03 13:20:03 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 3.1075018208302985 after 96000 steps (truncated to 768):
245
+ 2026-03-03 13:42:30 - Accuracy Cosine Similarity: 98.41%
246
+ 2026-03-03 13:42:30 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 3.1075018208302985 after 96000 steps (truncated to 512):
247
+ 2026-03-03 14:05:12 - Accuracy Cosine Similarity: 98.42%
248
+ 2026-03-03 14:05:12 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 3.1075018208302985 after 96000 steps (truncated to 256):
249
+ 2026-03-03 14:27:36 - Accuracy Cosine Similarity: 98.44%
250
+ 2026-03-03 14:27:36 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 3.1075018208302985 after 96000 steps (truncated to 128):
251
+ 2026-03-03 14:50:20 - Accuracy Cosine Similarity: 98.42%
252
+ 2026-03-03 14:50:20 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 3.1075018208302985 after 96000 steps (truncated to 64):
253
+ 2026-03-03 15:12:47 - Accuracy Cosine Similarity: 98.34%
254
+ {'eval_train_loss': '0.2361', 'eval_dev-768_cosine_accuracy': '0.9841', 'eval_dev-512_cosine_accuracy': '0.9842', 'eval_dev-256_cosine_accuracy': '0.9844', 'eval_dev-128_cosine_accuracy': '0.9842', 'eval_dev-64_cosine_accuracy': '0.9834', 'eval_sequential_score': '0.9841', 'eval_train_runtime': '9833', 'eval_train_samples_per_second': '114.9', 'eval_train_steps_per_second': '14.36', 'epoch': '3.108'}
255
+ 2026-03-03 15:12:47 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-96000
256
+ 2026-03-03 15:12:47 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-96000
257
+ {'loss': '0.2835', 'grad_norm': '3.412', 'learning_rate': '4.923e-06', 'epoch': '3.114'}
258
+ {'loss': '0.2774', 'grad_norm': '2.313', 'learning_rate': '4.887e-06', 'epoch': '3.12'}
259
+ {'loss': '0.2696', 'grad_norm': '1.394', 'learning_rate': '4.851e-06', 'epoch': '3.127'}
260
+ {'loss': '0.282', 'grad_norm': '4.947', 'learning_rate': '4.815e-06', 'epoch': '3.133'}
261
+ {'loss': '0.2813', 'grad_norm': '3.528', 'learning_rate': '4.779e-06', 'epoch': '3.14'}
262
+ {'loss': '0.2699', 'grad_norm': '3.904', 'learning_rate': '4.743e-06', 'epoch': '3.146'}
263
+ {'loss': '0.2727', 'grad_norm': '3.14', 'learning_rate': '4.707e-06', 'epoch': '3.153'}
264
+ {'loss': '0.2835', 'grad_norm': '3.664', 'learning_rate': '4.671e-06', 'epoch': '3.159'}
265
+ {'loss': '0.2722', 'grad_norm': '2.961', 'learning_rate': '4.635e-06', 'epoch': '3.166'}
266
+ {'loss': '0.2727', 'grad_norm': '4.693', 'learning_rate': '4.599e-06', 'epoch': '3.172'}
267
+ {'loss': '0.2747', 'grad_norm': '2.871', 'learning_rate': '4.563e-06', 'epoch': '3.179'}
268
+ {'loss': '0.2606', 'grad_norm': '3.735', 'learning_rate': '4.527e-06', 'epoch': '3.185'}
269
+ {'loss': '0.2705', 'grad_norm': '1.348', 'learning_rate': '4.491e-06', 'epoch': '3.192'}
270
+ {'loss': '0.2542', 'grad_norm': '5.111', 'learning_rate': '4.455e-06', 'epoch': '3.198'}
271
+ {'loss': '0.2632', 'grad_norm': '5.231', 'learning_rate': '4.419e-06', 'epoch': '3.205'}
272
+ {'loss': '0.2584', 'grad_norm': '2.372', 'learning_rate': '4.383e-06', 'epoch': '3.211'}
273
+ {'loss': '0.2597', 'grad_norm': '2.103', 'learning_rate': '4.347e-06', 'epoch': '3.218'}
274
+ {'loss': '0.2501', 'grad_norm': '4.595', 'learning_rate': '4.311e-06', 'epoch': '3.224'}
275
+ {'loss': '0.2657', 'grad_norm': '5.29', 'learning_rate': '4.275e-06', 'epoch': '3.231'}
276
+ {'loss': '0.2422', 'grad_norm': '1.912', 'learning_rate': '4.239e-06', 'epoch': '3.237'}
277
+ {'loss': '0.2478', 'grad_norm': '2.928', 'learning_rate': '4.203e-06', 'epoch': '3.243'}
278
+ {'loss': '0.2542', 'grad_norm': '4.825', 'learning_rate': '4.167e-06', 'epoch': '3.25'}
279
+ {'loss': '0.2519', 'grad_norm': '3.828', 'learning_rate': '4.131e-06', 'epoch': '3.256'}
280
+ {'loss': '0.2543', 'grad_norm': '2.029', 'learning_rate': '4.095e-06', 'epoch': '3.263'}
281
+ {'loss': '0.2525', 'grad_norm': '3.603', 'learning_rate': '4.059e-06', 'epoch': '3.269'}
282
+ {'loss': '0.2518', 'grad_norm': '2.854', 'learning_rate': '4.023e-06', 'epoch': '3.276'}
283
+ {'loss': '0.2536', 'grad_norm': '2.543', 'learning_rate': '3.987e-06', 'epoch': '3.282'}
284
+ {'loss': '0.253', 'grad_norm': '1.443', 'learning_rate': '3.951e-06', 'epoch': '3.289'}
285
+ {'loss': '0.27', 'grad_norm': '3.106', 'learning_rate': '3.916e-06', 'epoch': '3.295'}
286
+ {'loss': '0.2439', 'grad_norm': '2.519', 'learning_rate': '3.88e-06', 'epoch': '3.302'}
287
+ 2026-03-03 17:02:00 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 3.3017237193493565 after 102000 steps (truncated to 768):
288
+ 2026-03-03 17:24:36 - Accuracy Cosine Similarity: 98.45%
289
+ 2026-03-03 17:24:36 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 3.3017237193493565 after 102000 steps (truncated to 512):
290
+ 2026-03-03 17:47:01 - Accuracy Cosine Similarity: 98.46%
291
+ 2026-03-03 17:47:01 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 3.3017237193493565 after 102000 steps (truncated to 256):
292
+ 2026-03-03 18:09:20 - Accuracy Cosine Similarity: 98.48%
293
+ 2026-03-03 18:09:20 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 3.3017237193493565 after 102000 steps (truncated to 128):
294
+ 2026-03-03 18:31:48 - Accuracy Cosine Similarity: 98.46%
295
+ 2026-03-03 18:31:48 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 3.3017237193493565 after 102000 steps (truncated to 64):
296
+ 2026-03-03 18:53:45 - Accuracy Cosine Similarity: 98.36%
297
+ {'eval_train_loss': '0.2297', 'eval_dev-768_cosine_accuracy': '0.9845', 'eval_dev-512_cosine_accuracy': '0.9846', 'eval_dev-256_cosine_accuracy': '0.9848', 'eval_dev-128_cosine_accuracy': '0.9846', 'eval_dev-64_cosine_accuracy': '0.9836', 'eval_sequential_score': '0.9845', 'eval_train_runtime': '9774', 'eval_train_samples_per_second': '115.6', 'eval_train_steps_per_second': '14.45', 'epoch': '3.302'}
298
+ 2026-03-03 18:53:45 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-102000
299
+ 2026-03-03 18:53:45 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-102000
300
+ {'loss': '0.2592', 'grad_norm': '3.317', 'learning_rate': '3.844e-06', 'epoch': '3.308'}
301
+ {'loss': '0.2542', 'grad_norm': '1.975', 'learning_rate': '3.808e-06', 'epoch': '3.315'}
302
+ {'loss': '0.2641', 'grad_norm': '4.947', 'learning_rate': '3.772e-06', 'epoch': '3.321'}
303
+ {'loss': '0.2435', 'grad_norm': '2.649', 'learning_rate': '3.736e-06', 'epoch': '3.328'}
304
+ {'loss': '0.251', 'grad_norm': '3.429', 'learning_rate': '3.7e-06', 'epoch': '3.334'}
305
+ {'loss': '0.2481', 'grad_norm': '2.566', 'learning_rate': '3.664e-06', 'epoch': '3.341'}
306
+ {'loss': '0.2628', 'grad_norm': '2.595', 'learning_rate': '3.628e-06', 'epoch': '3.347'}
307
+ {'loss': '0.248', 'grad_norm': '2.943', 'learning_rate': '3.592e-06', 'epoch': '3.354'}
308
+ 2026-03-04 15:01:24 - Load pretrained SentenceTransformer: bert-base-arabertv02
309
+ 2026-03-04 15:01:34 - '[Errno -2] Name or service not known' thrown while requesting HEAD https://huggingface.co/bert-base-arabertv02/resolve/main/./modules.json
310
+ 2026-03-04 15:01:34 - Retrying in 1s [Retry 1/5].
311
+ 2026-03-04 15:01:35 - No sentence-transformers model found with name bert-base-arabertv02. Creating a new one with mean pooling.
312
+ 2026-03-04 15:03:06 - Use pytorch device_name: cuda:0
313
+ 2026-03-04 15:03:06 - Load pretrained SentenceTransformer: /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-102000
314
+ {'loss': '0.2592', 'grad_norm': '3.323', 'learning_rate': '3.844e-06', 'epoch': '3.308'}
315
+ {'loss': '0.2543', 'grad_norm': '1.973', 'learning_rate': '3.808e-06', 'epoch': '3.315'}
316
+ {'loss': '0.2641', 'grad_norm': '4.95', 'learning_rate': '3.772e-06', 'epoch': '3.321'}
317
+ {'loss': '0.2435', 'grad_norm': '2.661', 'learning_rate': '3.736e-06', 'epoch': '3.328'}
318
+ {'loss': '0.251', 'grad_norm': '3.434', 'learning_rate': '3.7e-06', 'epoch': '3.334'}
319
+ {'loss': '0.2481', 'grad_norm': '2.562', 'learning_rate': '3.664e-06', 'epoch': '3.341'}
320
+ {'loss': '0.2627', 'grad_norm': '2.6', 'learning_rate': '3.628e-06', 'epoch': '3.347'}
321
+ {'loss': '0.248', 'grad_norm': '2.941', 'learning_rate': '3.592e-06', 'epoch': '3.354'}
322
+ {'loss': '0.2636', 'grad_norm': '2.153', 'learning_rate': '3.556e-06', 'epoch': '3.36'}
323
+ {'loss': '0.2619', 'grad_norm': '2.729', 'learning_rate': '3.52e-06', 'epoch': '3.366'}
324
+ {'loss': '0.2423', 'grad_norm': '2.413', 'learning_rate': '3.484e-06', 'epoch': '3.373'}
325
+ {'loss': '0.2505', 'grad_norm': '4.396', 'learning_rate': '3.448e-06', 'epoch': '3.379'}
326
+ {'loss': '0.2604', 'grad_norm': '3.629', 'learning_rate': '3.412e-06', 'epoch': '3.386'}
327
+ {'loss': '0.246', 'grad_norm': '5.426', 'learning_rate': '3.376e-06', 'epoch': '3.392'}
328
+ {'loss': '0.244', 'grad_norm': '1.848', 'learning_rate': '3.34e-06', 'epoch': '3.399'}
329
+ {'loss': '0.2641', 'grad_norm': '5.125', 'learning_rate': '3.304e-06', 'epoch': '3.405'}
330
+ {'loss': '0.2573', 'grad_norm': '3.377', 'learning_rate': '3.268e-06', 'epoch': '3.412'}
331
+ {'loss': '0.2613', 'grad_norm': '2.842', 'learning_rate': '3.232e-06', 'epoch': '3.418'}
332
+ {'loss': '0.2746', 'grad_norm': '3.197', 'learning_rate': '3.196e-06', 'epoch': '3.425'}
333
+ {'loss': '0.2578', 'grad_norm': '2.158', 'learning_rate': '3.16e-06', 'epoch': '3.431'}
334
+ {'loss': '0.2445', 'grad_norm': '2.634', 'learning_rate': '3.124e-06', 'epoch': '3.438'}
335
+ {'loss': '0.253', 'grad_norm': '2.139', 'learning_rate': '3.088e-06', 'epoch': '3.444'}
336
+ {'loss': '0.2644', 'grad_norm': '3.682', 'learning_rate': '3.052e-06', 'epoch': '3.451'}
337
+ {'loss': '0.2656', 'grad_norm': '4.607', 'learning_rate': '3.016e-06', 'epoch': '3.457'}
338
+ {'loss': '0.252', 'grad_norm': '2.139', 'learning_rate': '2.98e-06', 'epoch': '3.464'}
339
+ {'loss': '0.2527', 'grad_norm': '2.422', 'learning_rate': '2.944e-06', 'epoch': '3.47'}
340
+ {'loss': '0.2534', 'grad_norm': '3.598', 'learning_rate': '2.908e-06', 'epoch': '3.477'}
341
+ {'loss': '0.253', 'grad_norm': '1.388', 'learning_rate': '2.872e-06', 'epoch': '3.483'}
342
+ {'loss': '0.2614', 'grad_norm': '3.057', 'learning_rate': '2.837e-06', 'epoch': '3.489'}
343
+ {'loss': '0.2517', 'grad_norm': '2.296', 'learning_rate': '2.801e-06', 'epoch': '3.496'}
344
+ 2026-03-04 16:57:03 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 3.4959456178684145 after 108000 steps (truncated to 768):
345
+ 2026-03-04 17:20:27 - Accuracy Cosine Similarity: 98.49%
346
+ 2026-03-04 17:20:27 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 3.4959456178684145 after 108000 steps (truncated to 512):
347
+ 2026-03-04 17:42:09 - Accuracy Cosine Similarity: 98.49%
348
+ 2026-03-04 17:42:09 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 3.4959456178684145 after 108000 steps (truncated to 256):
349
+ 2026-03-04 18:03:04 - Accuracy Cosine Similarity: 98.51%
350
+ 2026-03-04 18:03:04 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 3.4959456178684145 after 108000 steps (truncated to 128):
351
+ 2026-03-04 18:23:36 - Accuracy Cosine Similarity: 98.49%
352
+ 2026-03-04 18:23:36 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 3.4959456178684145 after 108000 steps (truncated to 64):
353
+ 2026-03-04 18:44:21 - Accuracy Cosine Similarity: 98.41%
354
+ {'eval_train_loss': '0.2252', 'eval_dev-768_cosine_accuracy': '0.9849', 'eval_dev-512_cosine_accuracy': '0.9849', 'eval_dev-256_cosine_accuracy': '0.9851', 'eval_dev-128_cosine_accuracy': '0.9849', 'eval_dev-64_cosine_accuracy': '0.9841', 'eval_sequential_score': '0.9849', 'eval_train_runtime': '9737', 'eval_train_samples_per_second': '116', 'eval_train_steps_per_second': '14.5', 'epoch': '3.496'}
355
+ 2026-03-04 18:44:21 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-108000
356
+ 2026-03-04 18:44:21 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-108000
357
+ {'loss': '0.2589', 'grad_norm': '2.231', 'learning_rate': '2.765e-06', 'epoch': '3.502'}
358
+ {'loss': '0.2582', 'grad_norm': '5.077', 'learning_rate': '2.729e-06', 'epoch': '3.509'}
359
+ {'loss': '0.2553', 'grad_norm': '4.234', 'learning_rate': '2.693e-06', 'epoch': '3.515'}
360
+ {'loss': '0.2457', 'grad_norm': '2.167', 'learning_rate': '2.657e-06', 'epoch': '3.522'}
361
+ {'loss': '0.2662', 'grad_norm': '4.357', 'learning_rate': '2.621e-06', 'epoch': '3.528'}
362
+ {'loss': '0.2619', 'grad_norm': '3.987', 'learning_rate': '2.585e-06', 'epoch': '3.535'}
363
+ {'loss': '0.2556', 'grad_norm': '1.6', 'learning_rate': '2.549e-06', 'epoch': '3.541'}
364
+ {'loss': '0.2635', 'grad_norm': '2.069', 'learning_rate': '2.513e-06', 'epoch': '3.548'}
365
+ {'loss': '0.255', 'grad_norm': '2.314', 'learning_rate': '2.477e-06', 'epoch': '3.554'}
366
+ {'loss': '0.267', 'grad_norm': '4.255', 'learning_rate': '2.441e-06', 'epoch': '3.561'}
367
+ {'loss': '0.266', 'grad_norm': '3.208', 'learning_rate': '2.405e-06', 'epoch': '3.567'}
368
+ {'loss': '0.2604', 'grad_norm': '3.669', 'learning_rate': '2.369e-06', 'epoch': '3.574'}
369
+ {'loss': '0.2574', 'grad_norm': '3.103', 'learning_rate': '2.333e-06', 'epoch': '3.58'}
370
+ {'loss': '0.2607', 'grad_norm': '4.632', 'learning_rate': '2.297e-06', 'epoch': '3.587'}
371
+ {'loss': '0.2465', 'grad_norm': '4', 'learning_rate': '2.261e-06', 'epoch': '3.593'}
372
+ {'loss': '0.279', 'grad_norm': '3.647', 'learning_rate': '2.225e-06', 'epoch': '3.6'}
373
+ {'loss': '0.2681', 'grad_norm': '5.511', 'learning_rate': '2.189e-06', 'epoch': '3.606'}
374
+ {'loss': '0.2654', 'grad_norm': '1.217', 'learning_rate': '2.153e-06', 'epoch': '3.612'}
375
+ {'loss': '0.264', 'grad_norm': '4.641', 'learning_rate': '2.117e-06', 'epoch': '3.619'}
376
+ {'loss': '0.2774', 'grad_norm': '3.571', 'learning_rate': '2.081e-06', 'epoch': '3.625'}
377
+ {'loss': '0.2568', 'grad_norm': '3.808', 'learning_rate': '2.045e-06', 'epoch': '3.632'}
378
+ {'loss': '0.2665', 'grad_norm': '3.178', 'learning_rate': '2.009e-06', 'epoch': '3.638'}
379
+ {'loss': '0.2532', 'grad_norm': '1.566', 'learning_rate': '1.973e-06', 'epoch': '3.645'}
380
+ {'loss': '0.2613', 'grad_norm': '3.262', 'learning_rate': '1.937e-06', 'epoch': '3.651'}
381
+ {'loss': '0.2413', 'grad_norm': '4.082', 'learning_rate': '1.901e-06', 'epoch': '3.658'}
382
+ {'loss': '0.2788', 'grad_norm': '4.851', 'learning_rate': '1.865e-06', 'epoch': '3.664'}
383
+ {'loss': '0.2586', 'grad_norm': '5.756', 'learning_rate': '1.829e-06', 'epoch': '3.671'}
384
+ {'loss': '0.2602', 'grad_norm': '5.288', 'learning_rate': '1.793e-06', 'epoch': '3.677'}
385
+ {'loss': '0.2708', 'grad_norm': '3.492', 'learning_rate': '1.758e-06', 'epoch': '3.684'}
386
+ {'loss': '0.2556', 'grad_norm': '2.688', 'learning_rate': '1.722e-06', 'epoch': '3.69'}
387
+ 2026-03-04 20:34:25 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 3.6901675163874725 after 114000 steps (truncated to 768):
388
+ 2026-03-04 20:55:09 - Accuracy Cosine Similarity: 98.51%
389
+ 2026-03-04 20:55:09 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 3.6901675163874725 after 114000 steps (truncated to 512):
390
+ 2026-03-04 21:15:51 - Accuracy Cosine Similarity: 98.52%
391
+ 2026-03-04 21:15:51 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 3.6901675163874725 after 114000 steps (truncated to 256):
392
+ 2026-03-04 21:36:36 - Accuracy Cosine Similarity: 98.54%
393
+ 2026-03-04 21:36:36 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 3.6901675163874725 after 114000 steps (truncated to 128):
394
+ 2026-03-04 21:57:23 - Accuracy Cosine Similarity: 98.53%
395
+ 2026-03-04 21:57:23 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 3.6901675163874725 after 114000 steps (truncated to 64):
396
+ 2026-03-04 22:18:08 - Accuracy Cosine Similarity: 98.44%
397
+ {'eval_train_loss': '0.2188', 'eval_dev-768_cosine_accuracy': '0.9851', 'eval_dev-512_cosine_accuracy': '0.9852', 'eval_dev-256_cosine_accuracy': '0.9854', 'eval_dev-128_cosine_accuracy': '0.9853', 'eval_dev-64_cosine_accuracy': '0.9844', 'eval_sequential_score': '0.9851', 'eval_train_runtime': '9310', 'eval_train_samples_per_second': '121.3', 'eval_train_steps_per_second': '15.17', 'epoch': '3.69'}
398
+ 2026-03-04 22:18:08 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-114000
399
+ 2026-03-04 22:18:08 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-114000
400
+ {'loss': '0.2576', 'grad_norm': '2.428', 'learning_rate': '1.686e-06', 'epoch': '3.697'}
401
+ {'loss': '0.2713', 'grad_norm': '2.93', 'learning_rate': '1.65e-06', 'epoch': '3.703'}
402
+ {'loss': '0.2748', 'grad_norm': '3.659', 'learning_rate': '1.614e-06', 'epoch': '3.71'}
403
+ {'loss': '0.2542', 'grad_norm': '3.45', 'learning_rate': '1.578e-06', 'epoch': '3.716'}
404
+ {'loss': '0.2647', 'grad_norm': '4.824', 'learning_rate': '1.542e-06', 'epoch': '3.723'}
405
+ {'loss': '0.2751', 'grad_norm': '1.834', 'learning_rate': '1.506e-06', 'epoch': '3.729'}
406
+ {'loss': '0.2534', 'grad_norm': '4.17', 'learning_rate': '1.47e-06', 'epoch': '3.735'}
407
+ {'loss': '0.2577', 'grad_norm': '3.893', 'learning_rate': '1.434e-06', 'epoch': '3.742'}
408
+ {'loss': '0.2722', 'grad_norm': '4.691', 'learning_rate': '1.398e-06', 'epoch': '3.748'}
409
+ {'loss': '0.2717', 'grad_norm': '3.968', 'learning_rate': '1.362e-06', 'epoch': '3.755'}
410
+ {'loss': '0.2737', 'grad_norm': '4.474', 'learning_rate': '1.326e-06', 'epoch': '3.761'}
411
+ {'loss': '0.2725', 'grad_norm': '3.462', 'learning_rate': '1.29e-06', 'epoch': '3.768'}
412
+ {'loss': '0.2587', 'grad_norm': '1.857', 'learning_rate': '1.254e-06', 'epoch': '3.774'}
413
+ {'loss': '0.2623', 'grad_norm': '2.164', 'learning_rate': '1.218e-06', 'epoch': '3.781'}
414
+ {'loss': '0.2659', 'grad_norm': '2.674', 'learning_rate': '1.182e-06', 'epoch': '3.787'}
415
+ {'loss': '0.2735', 'grad_norm': '1.56', 'learning_rate': '1.146e-06', 'epoch': '3.794'}
416
+ {'loss': '0.2847', 'grad_norm': '4.402', 'learning_rate': '1.11e-06', 'epoch': '3.8'}
417
+ {'loss': '0.2636', 'grad_norm': '5.62', 'learning_rate': '1.074e-06', 'epoch': '3.807'}
418
+ {'loss': '0.2777', 'grad_norm': '4.51', 'learning_rate': '1.038e-06', 'epoch': '3.813'}
419
+ {'loss': '0.2751', 'grad_norm': '3.339', 'learning_rate': '1.002e-06', 'epoch': '3.82'}
420
+ {'loss': '0.2609', 'grad_norm': '3.817', 'learning_rate': '9.662e-07', 'epoch': '3.826'}
421
+ {'loss': '0.2684', 'grad_norm': '4.592', 'learning_rate': '9.303e-07', 'epoch': '3.833'}
422
+ {'loss': '0.2772', 'grad_norm': '2.499', 'learning_rate': '8.943e-07', 'epoch': '3.839'}
423
+ {'loss': '0.2684', 'grad_norm': '4.28', 'learning_rate': '8.583e-07', 'epoch': '3.846'}
424
+ {'loss': '0.2682', 'grad_norm': '3.179', 'learning_rate': '8.224e-07', 'epoch': '3.852'}
425
+ {'loss': '0.2784', 'grad_norm': '5.42', 'learning_rate': '7.864e-07', 'epoch': '3.858'}
426
+ {'loss': '0.2735', 'grad_norm': '2.668', 'learning_rate': '7.504e-07', 'epoch': '3.865'}
427
+ {'loss': '0.2848', 'grad_norm': '4.855', 'learning_rate': '7.145e-07', 'epoch': '3.871'}
428
+ {'loss': '0.2638', 'grad_norm': '2.601', 'learning_rate': '6.785e-07', 'epoch': '3.878'}
429
+ {'loss': '0.2711', 'grad_norm': '4.021', 'learning_rate': '6.425e-07', 'epoch': '3.884'}
430
+ 2026-03-05 00:09:04 - TripletEvaluator: Evaluating the model on the dev-768 dataset in epoch 3.8843894149065306 after 120000 steps (truncated to 768):
431
+ 2026-03-05 00:30:13 - Accuracy Cosine Similarity: 98.54%
432
+ 2026-03-05 00:30:13 - TripletEvaluator: Evaluating the model on the dev-512 dataset in epoch 3.8843894149065306 after 120000 steps (truncated to 512):
433
+ 2026-03-05 00:51:03 - Accuracy Cosine Similarity: 98.55%
434
+ 2026-03-05 00:51:03 - TripletEvaluator: Evaluating the model on the dev-256 dataset in epoch 3.8843894149065306 after 120000 steps (truncated to 256):
435
+ 2026-03-05 01:11:59 - Accuracy Cosine Similarity: 98.56%
436
+ 2026-03-05 01:11:59 - TripletEvaluator: Evaluating the model on the dev-128 dataset in epoch 3.8843894149065306 after 120000 steps (truncated to 128):
437
+ 2026-03-05 01:32:50 - Accuracy Cosine Similarity: 98.55%
438
+ 2026-03-05 01:32:50 - TripletEvaluator: Evaluating the model on the dev-64 dataset in epoch 3.8843894149065306 after 120000 steps (truncated to 64):
439
+ 2026-03-05 01:53:42 - Accuracy Cosine Similarity: 98.46%
440
+ {'eval_train_loss': '0.2171', 'eval_dev-768_cosine_accuracy': '0.9854', 'eval_dev-512_cosine_accuracy': '0.9855', 'eval_dev-256_cosine_accuracy': '0.9856', 'eval_dev-128_cosine_accuracy': '0.9855', 'eval_dev-64_cosine_accuracy': '0.9846', 'eval_sequential_score': '0.9854', 'eval_train_runtime': '9383', 'eval_train_samples_per_second': '120.4', 'eval_train_steps_per_second': '15.05', 'epoch': '3.884'}
441
+ 2026-03-05 01:53:42 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-120000
442
+ 2026-03-05 01:53:42 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-120000
443
+ {'loss': '0.2825', 'grad_norm': '2.434', 'learning_rate': '6.066e-07', 'epoch': '3.891'}
444
+ {'loss': '0.2724', 'grad_norm': '3.687', 'learning_rate': '5.706e-07', 'epoch': '3.897'}
445
+ {'loss': '0.3078', 'grad_norm': '5.537', 'learning_rate': '5.346e-07', 'epoch': '3.904'}
446
+ {'loss': '0.2806', 'grad_norm': '3.519', 'learning_rate': '4.987e-07', 'epoch': '3.91'}
447
+ {'loss': '0.2631', 'grad_norm': '3.982', 'learning_rate': '4.627e-07', 'epoch': '3.917'}
448
+ {'loss': '0.2892', 'grad_norm': '3.621', 'learning_rate': '4.267e-07', 'epoch': '3.923'}
449
+ {'loss': '0.2791', 'grad_norm': '4.467', 'learning_rate': '3.908e-07', 'epoch': '3.93'}
450
+ {'loss': '0.2874', 'grad_norm': '4.958', 'learning_rate': '3.548e-07', 'epoch': '3.936'}
451
+ {'loss': '0.2602', 'grad_norm': '3.852', 'learning_rate': '3.188e-07', 'epoch': '3.943'}
452
+ {'loss': '0.2988', 'grad_norm': '2.639', 'learning_rate': '2.829e-07', 'epoch': '3.949'}
453
+ {'loss': '0.2935', 'grad_norm': '1.878', 'learning_rate': '2.469e-07', 'epoch': '3.956'}
454
+ {'loss': '0.2999', 'grad_norm': '3.6', 'learning_rate': '2.109e-07', 'epoch': '3.962'}
455
+ {'loss': '0.293', 'grad_norm': '2.28', 'learning_rate': '1.75e-07', 'epoch': '3.969'}
456
+ {'loss': '0.2784', 'grad_norm': '4.347', 'learning_rate': '1.39e-07', 'epoch': '3.975'}
457
+ {'loss': '0.3013', 'grad_norm': '5.014', 'learning_rate': '1.03e-07', 'epoch': '3.982'}
458
+ {'loss': '0.2919', 'grad_norm': '2.171', 'learning_rate': '6.708e-08', 'epoch': '3.988'}
459
+ {'loss': '0.3011', 'grad_norm': '5.241', 'learning_rate': '3.111e-08', 'epoch': '3.994'}
460
+ 2026-03-05 02:28:54 - Saving model checkpoint to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-123572
461
+ 2026-03-05 02:28:54 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-123572
462
+ {'train_runtime': '4.115e+04', 'train_samples_per_second': '384.4', 'train_steps_per_second': '3.003', 'train_loss': '0.04643', 'epoch': '4'}
463
+ 2026-03-05 02:28:56 - Save model to /home/skiredj.abderrahman/khalil/sbert_training/output/final_epoch4
464
+ model saved successfully
465
+ 2026-03-05 02:28:56 - TripletEvaluator: Evaluating the model on the test-768 dataset (truncated to 768):
466
+ 2026-03-05 02:47:53 - Accuracy Cosine Similarity: 98.54%
467
+ 2026-03-05 02:47:53 - TripletEvaluator: Evaluating the model on the test-512 dataset (truncated to 512):
468
+ 2026-03-05 03:06:39 - Accuracy Cosine Similarity: 98.56%
469
+ 2026-03-05 03:06:39 - TripletEvaluator: Evaluating the model on the test-256 dataset (truncated to 256):
470
+ 2026-03-05 03:25:22 - Accuracy Cosine Similarity: 98.57%
471
+ 2026-03-05 03:25:22 - TripletEvaluator: Evaluating the model on the test-128 dataset (truncated to 128):
472
+ 2026-03-05 03:44:01 - Accuracy Cosine Similarity: 98.55%
473
+ 2026-03-05 03:44:01 - TripletEvaluator: Evaluating the model on the test-64 dataset (truncated to 64):
474
+ 2026-03-05 04:02:46 - Accuracy Cosine Similarity: 98.46%
epoch4/model/1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 768,
3
+ "pooling_mode_cls_token": false,
4
+ "pooling_mode_mean_tokens": true,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false,
7
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": false,
9
+ "include_prompt": true
10
+ }
epoch4/model/README.md ADDED
@@ -0,0 +1,662 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - sentence-transformers
4
+ - sentence-similarity
5
+ - feature-extraction
6
+ - dense
7
+ - generated_from_trainer
8
+ - dataset_size:3954179
9
+ - loss:MatryoshkaLoss
10
+ - loss:MultipleNegativesRankingLoss
11
+ widget:
12
+ - source_sentence: إذا لم تكن قد فعلت ذلك بالفعل ، تحقق من تصميمنا الجديد للمراسلات
13
+ والحوارات واليوميات .
14
+ sentences:
15
+ - تم إعادة تصميم الرسائل والحوارات .
16
+ - يقدم مقهى بارج كل من وجبات الغداء والإفطار .
17
+ - قبل ان نعرف اسماء بعضنا او اي شيء قد تعانقنا وبكىنا .
18
+ - source_sentence: أين تقع جامعة واينزبرج
19
+ sentences:
20
+ - جامعة دالاس بابتيست ( DBU ) ، المعروفة سابقا باسم كلية دالاس بابتيست ، هي جامعة
21
+ فنون ليبرالية مسيحية تقع في دالاس ، تكساس . يقع الحرم الجامعي الرئيسي على بعد
22
+ حوالي 12 ميلا ( 19 كم ) جنوب غرب وسط مدينة دالاس ويطل على بحيرة ماونتين كريك .
23
+ تأسست جامعة دالاس بابتيست عام 1898 باسم كلية ديكاتور بابتيست ، وتدير حاليا حرما
24
+ جامعيا في دالاس وبلانو وهيرست .
25
+ - الزوجان معا
26
+ - تقع جامعة واينسبرغ في حرم جامعي معاصر في تلال جنوب غرب ولاية بنسلفانيا ، مع ثلاثة
27
+ مراكز للبالغين تقع في مناطق بيتسبرغ في ساوثبوينت وكرانبيري ومونروفيل . تم إدراج
28
+ Hanna Hall و Miller Hall في السجل الوطني للأماكن التاريخية .
29
+ - source_sentence: The isolated Russian forces resisted in several areas for two more
30
+ days .
31
+ sentences:
32
+ - 'ياهو : كيف يمكنني معرفة ما إذا كان البريد الإلكتروني الذي أرسلته قد تم استلامه
33
+ أو قراءته ؟'
34
+ - واستمرت الاشتباكات الحدودية خلال اليومين المقبلين ، حيث استهدفت المخافر الحدودية
35
+ من الجانبين والتي أسفرت عن وقوع عشرات الإصابات .
36
+ - قاومت القوات الروسية المعزولة في عة مناطق لمدة يومين آخرين .
37
+ - source_sentence: فتاة هيبي بشعر أشقر وأرجواني على الجانب يرتدي قميص أبيض وملابس
38
+ سوداء
39
+ sentences:
40
+ - فتاة " هيبي " ترتدي قميصا أبيضا وملابس سوداء شعرها أشقر وأحمر
41
+ - المرأة تضع يدها في جيب الرجل
42
+ - فتاة لديها سترة حمراء وسوداء
43
+ - source_sentence: رجل وامرأة يجلسان في سيارة ووجههما في الاتجاه المعاكس من الكاميرا
44
+ sentences:
45
+ - هناك شخصان وسيارة
46
+ - سيارة صدئة هي الشيء الوحيد المرئي
47
+ - كان أفضل حالا
48
+ pipeline_tag: sentence-similarity
49
+ library_name: sentence-transformers
50
+ metrics:
51
+ - cosine_accuracy
52
+ model-index:
53
+ - name: SentenceTransformer
54
+ results:
55
+ - task:
56
+ type: triplet
57
+ name: Triplet
58
+ dataset:
59
+ name: dev 768
60
+ type: dev-768
61
+ metrics:
62
+ - type: cosine_accuracy
63
+ value: 0.9853799939155579
64
+ name: Cosine Accuracy
65
+ - task:
66
+ type: triplet
67
+ name: Triplet
68
+ dataset:
69
+ name: dev 512
70
+ type: dev-512
71
+ metrics:
72
+ - type: cosine_accuracy
73
+ value: 0.9855160117149353
74
+ name: Cosine Accuracy
75
+ - task:
76
+ type: triplet
77
+ name: Triplet
78
+ dataset:
79
+ name: dev 256
80
+ type: dev-256
81
+ metrics:
82
+ - type: cosine_accuracy
83
+ value: 0.985588014125824
84
+ name: Cosine Accuracy
85
+ - task:
86
+ type: triplet
87
+ name: Triplet
88
+ dataset:
89
+ name: dev 128
90
+ type: dev-128
91
+ metrics:
92
+ - type: cosine_accuracy
93
+ value: 0.9855039715766907
94
+ name: Cosine Accuracy
95
+ - task:
96
+ type: triplet
97
+ name: Triplet
98
+ dataset:
99
+ name: dev 64
100
+ type: dev-64
101
+ metrics:
102
+ - type: cosine_accuracy
103
+ value: 0.9845880270004272
104
+ name: Cosine Accuracy
105
+ ---
106
+
107
+ # SentenceTransformer
108
+
109
+ This is a [sentence-transformers](https://www.SBERT.net) model trained on the train dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
110
+
111
+ ## Model Details
112
+
113
+ ### Model Description
114
+ - **Model Type:** Sentence Transformer
115
+ <!-- - **Base model:** [Unknown](https://huggingface.co/unknown) -->
116
+ - **Maximum Sequence Length:** 512 tokens
117
+ - **Output Dimensionality:** 768 dimensions
118
+ - **Similarity Function:** Cosine Similarity
119
+ - **Training Dataset:**
120
+ - train
121
+ <!-- - **Language:** Unknown -->
122
+ <!-- - **License:** Unknown -->
123
+
124
+ ### Model Sources
125
+
126
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
127
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/huggingface/sentence-transformers)
128
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
129
+
130
+ ### Full Model Architecture
131
+
132
+ ```
133
+ SentenceTransformer(
134
+ (0): Transformer({'max_seq_length': 512, 'do_lower_case': False, 'architecture': 'BertModel'})
135
+ (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
136
+ )
137
+ ```
138
+
139
+ ## Usage
140
+
141
+ ### Direct Usage (Sentence Transformers)
142
+
143
+ First install the Sentence Transformers library:
144
+
145
+ ```bash
146
+ pip install -U sentence-transformers
147
+ ```
148
+
149
+ Then you can load this model and run inference.
150
+ ```python
151
+ from sentence_transformers import SentenceTransformer
152
+
153
+ # Download from the 🤗 Hub
154
+ model = SentenceTransformer("sentence_transformers_model_id")
155
+ # Run inference
156
+ sentences = [
157
+ 'رجل وامرأة يجلسان في سيارة ووجههما في الاتجاه المعاكس من الكاميرا',
158
+ 'هناك شخصان وسيارة',
159
+ 'سيارة صدئة هي الشيء الوحيد المرئي',
160
+ ]
161
+ embeddings = model.encode(sentences)
162
+ print(embeddings.shape)
163
+ # [3, 768]
164
+
165
+ # Get the similarity scores for the embeddings
166
+ similarities = model.similarity(embeddings, embeddings)
167
+ print(similarities)
168
+ # tensor([[1.0000, 0.6553, 0.2590],
169
+ # [0.6553, 1.0000, 0.3695],
170
+ # [0.2590, 0.3695, 1.0000]])
171
+ ```
172
+
173
+ <!--
174
+ ### Direct Usage (Transformers)
175
+
176
+ <details><summary>Click to see the direct usage in Transformers</summary>
177
+
178
+ </details>
179
+ -->
180
+
181
+ <!--
182
+ ### Downstream Usage (Sentence Transformers)
183
+
184
+ You can finetune this model on your own dataset.
185
+
186
+ <details><summary>Click to expand</summary>
187
+
188
+ </details>
189
+ -->
190
+
191
+ <!--
192
+ ### Out-of-Scope Use
193
+
194
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
195
+ -->
196
+
197
+ ## Evaluation
198
+
199
+ ### Metrics
200
+
201
+ #### Triplet
202
+
203
+ * Dataset: `dev-768`
204
+ * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters:
205
+ ```json
206
+ {
207
+ "truncate_dim": 768
208
+ }
209
+ ```
210
+
211
+ | Metric | Value |
212
+ |:--------------------|:-----------|
213
+ | **cosine_accuracy** | **0.9854** |
214
+
215
+ #### Triplet
216
+
217
+ * Dataset: `dev-512`
218
+ * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters:
219
+ ```json
220
+ {
221
+ "truncate_dim": 512
222
+ }
223
+ ```
224
+
225
+ | Metric | Value |
226
+ |:--------------------|:-----------|
227
+ | **cosine_accuracy** | **0.9855** |
228
+
229
+ #### Triplet
230
+
231
+ * Dataset: `dev-256`
232
+ * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters:
233
+ ```json
234
+ {
235
+ "truncate_dim": 256
236
+ }
237
+ ```
238
+
239
+ | Metric | Value |
240
+ |:--------------------|:-----------|
241
+ | **cosine_accuracy** | **0.9856** |
242
+
243
+ #### Triplet
244
+
245
+ * Dataset: `dev-128`
246
+ * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters:
247
+ ```json
248
+ {
249
+ "truncate_dim": 128
250
+ }
251
+ ```
252
+
253
+ | Metric | Value |
254
+ |:--------------------|:-----------|
255
+ | **cosine_accuracy** | **0.9855** |
256
+
257
+ #### Triplet
258
+
259
+ * Dataset: `dev-64`
260
+ * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters:
261
+ ```json
262
+ {
263
+ "truncate_dim": 64
264
+ }
265
+ ```
266
+
267
+ | Metric | Value |
268
+ |:--------------------|:-----------|
269
+ | **cosine_accuracy** | **0.9846** |
270
+
271
+ <!--
272
+ ## Bias, Risks and Limitations
273
+
274
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
275
+ -->
276
+
277
+ <!--
278
+ ### Recommendations
279
+
280
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
281
+ -->
282
+
283
+ ## Training Details
284
+
285
+ ### Training Dataset
286
+
287
+ #### train
288
+
289
+ * Dataset: train
290
+ * Size: 3,954,179 training samples
291
+ * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
292
+ * Approximate statistics based on the first 1000 samples:
293
+ | | anchor | positive | negative |
294
+ |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
295
+ | type | string | string | string |
296
+ | details | <ul><li>min: 4 tokens</li><li>mean: 16.1 tokens</li><li>max: 113 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 41.85 tokens</li><li>max: 512 tokens</li></ul> | <ul><li>min: 4 tokens</li><li>mean: 41.99 tokens</li><li>max: 512 tokens</li></ul> |
297
+ * Samples:
298
+ | anchor | positive | negative |
299
+ |:----------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
300
+ | <code>في أي مقاطعة تقع لويسفيل أركنساس</code> | <code>لويسفيل هي بلدة في مقاطعة لافاييت ، أركنساس ، الولايات المتحدة . كان عدد السكان 1285 في تعداد عام 2000 . . المدينة هي مقر مقاطعة لافاييت .</code> | <code>ماونتن هوم ، أركنساس . ماونتن هوم هي مدينة صغيرة في مقاطعة باكستر ، أركنساس ، الولايات المتحدة ، في جبال أوزارك الجنوبية بالقرب من حدود الولاية الشمالية مع ميسوري . اعتبارا من تعداد عام 2010 ، بلغ عدد سكان المدينة 12448 نسمة .</code> |
301
+ | <code>متوسط سمك باب الخزانة</code> | <code>تتميز أبواب العالم القديم بميزات رائعة مثل السماكة المتزايدة ، والملامح الأعمق ، والأعمدة والقضبان الأوسع لإضفاء مظهر وإحساس أكثر دراماتيكية عند مقارنتها بأبواب الخزانة التقليدية . يبلغ عرض Stiles Rails القياسية 3 بوصات ويمكن تصنيعها في 1 و 1 1 - 8 و 1 سمك .</code> | <code>اعتمادا على الخطأ في اللوحة ، يبلغ متوسط أسعار الإصلاح 130 دولارا لإصلاح الأبواب الفولاذية و 190 دولارا للخشب و 170 دولارا للألمنيوم و 150 دولارا للألياف الزجاجية . مزيد من المعلومات حول كيفية استبدال لوحة باب المرآب . إذا تعطلت أداة فتح باب الجراج ، فقد تكون سلامتك في خطر . تريد التأكد من أن بابك يعمل بشكل صحيح حتى لا يغلق بطريق الخطأ على حيوان أليف أو شخص . تريد أيضا إغلاقها لإبعاد اللصوص عن منزلك .</code> |
302
+ | <code>ما هو تعريف الملء</code> | <code>اعادة تعبئه . اسم تخصيص ثان لوكيل الوصفات الطبية تم الحصول عليه من الصيدلية ، والذي يسمح به فعل الوصفة الأصلية علم الأدوية للحصول على المزيد من دواء معين ، بعد استخدام الكمية الموصوفة في البداية من الوكيل أو إعطائها . انظر الوصفة الطبية .</code> | <code>تعليمات إعادة الملء قم بإعادة الملء فقط باستخدام Spectracide ' Bug Stop Home Barrier Refill . قم بإزالة الغطاء . قم بقياس وصب 12 . 8 أونصة سائلة من المركز في حاوية فارغة سعة 1 جالون من Spectracide - Bug Stop - حاجز منزلي ، واملأه حتى 1 جالون بالماء ، استبدل الغطاء وأغلقه بإحكام . المنتج المنسكب قم بقياس 12 . 8 أونصة سائلة من المركز وصبها بحذر في حاوية فارغة سعة 1 جالون من Spectracide - حاجز منزلي من Spectracide - حاجز منزلي ، واملأه حتى 1 جالون بالماء . استبدل الغطاء وأغلقه بإحكام . امسح أي منتج مسكوب .</code> |
303
+ * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters:
304
+ ```json
305
+ {
306
+ "loss": "MultipleNegativesRankingLoss",
307
+ "matryoshka_dims": [
308
+ 768,
309
+ 512,
310
+ 256,
311
+ 128,
312
+ 64
313
+ ],
314
+ "matryoshka_weights": [
315
+ 1,
316
+ 1,
317
+ 1,
318
+ 1,
319
+ 1
320
+ ],
321
+ "n_dims_per_step": -1
322
+ }
323
+ ```
324
+
325
+ ### Evaluation Dataset
326
+
327
+ #### train
328
+
329
+ * Dataset: train
330
+ * Size: 1,129,759 evaluation samples
331
+ * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
332
+ * Approximate statistics based on the first 1000 samples:
333
+ | | anchor | positive | negative |
334
+ |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
335
+ | type | string | string | string |
336
+ | details | <ul><li>min: 4 tokens</li><li>mean: 16.7 tokens</li><li>max: 150 tokens</li></ul> | <ul><li>min: 4 tokens</li><li>mean: 36.54 tokens</li><li>max: 512 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 35.42 tokens</li><li>max: 512 tokens</li></ul> |
337
+ * Samples:
338
+ | anchor | positive | negative |
339
+ |:---------------------------------------------------------------------|:---------------------------------|:----------------------------------------------------------------------|
340
+ | <code>رجل يرتدي سروال تنس أزرق وقميص بولو أبيض يضرب كرة التنس</code> | <code>رجل يلعب رياضة</code> | <code>هناك رجل يرتدي زي البيسبول يضرب كرة البيسبول بمضرب التنس</code> |
341
+ | <code>امرأة في ثوب أسود تبدو متفاجئة</code> | <code>امرأة تغيرت مشاعرها</code> | <code>امرأة تسبح في المحيط</code> |
342
+ | <code>رجل يرتدي قميص أبيض يقفز على شيء ما على دراجته الصفراء</code> | <code>رجل يركب دراجته</code> | <code>رجل يركب لوح التزلج فوق المنحدر</code> |
343
+ * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters:
344
+ ```json
345
+ {
346
+ "loss": "MultipleNegativesRankingLoss",
347
+ "matryoshka_dims": [
348
+ 768,
349
+ 512,
350
+ 256,
351
+ 128,
352
+ 64
353
+ ],
354
+ "matryoshka_weights": [
355
+ 1,
356
+ 1,
357
+ 1,
358
+ 1,
359
+ 1
360
+ ],
361
+ "n_dims_per_step": -1
362
+ }
363
+ ```
364
+
365
+ ### Training Hyperparameters
366
+ #### Non-Default Hyperparameters
367
+
368
+ - `per_device_train_batch_size`: 64
369
+ - `num_train_epochs`: 4
370
+ - `learning_rate`: 2e-05
371
+ - `warmup_steps`: 0.1
372
+ - `gradient_accumulation_steps`: 2
373
+ - `bf16`: True
374
+ - `eval_strategy`: steps
375
+ - `warmup_ratio`: 0.1
376
+ - `batch_sampler`: no_duplicates
377
+
378
+ #### All Hyperparameters
379
+ <details><summary>Click to expand</summary>
380
+
381
+ - `per_device_train_batch_size`: 64
382
+ - `num_train_epochs`: 4
383
+ - `max_steps`: -1
384
+ - `learning_rate`: 2e-05
385
+ - `lr_scheduler_type`: linear
386
+ - `lr_scheduler_kwargs`: None
387
+ - `warmup_steps`: 0.1
388
+ - `optim`: adamw_torch
389
+ - `optim_args`: None
390
+ - `weight_decay`: 0.0
391
+ - `adam_beta1`: 0.9
392
+ - `adam_beta2`: 0.999
393
+ - `adam_epsilon`: 1e-08
394
+ - `optim_target_modules`: None
395
+ - `gradient_accumulation_steps`: 2
396
+ - `average_tokens_across_devices`: True
397
+ - `max_grad_norm`: 1.0
398
+ - `label_smoothing_factor`: 0.0
399
+ - `bf16`: True
400
+ - `fp16`: False
401
+ - `bf16_full_eval`: False
402
+ - `fp16_full_eval`: False
403
+ - `tf32`: None
404
+ - `gradient_checkpointing`: False
405
+ - `gradient_checkpointing_kwargs`: None
406
+ - `torch_compile`: False
407
+ - `torch_compile_backend`: None
408
+ - `torch_compile_mode`: None
409
+ - `use_liger_kernel`: False
410
+ - `liger_kernel_config`: None
411
+ - `use_cache`: False
412
+ - `neftune_noise_alpha`: None
413
+ - `torch_empty_cache_steps`: None
414
+ - `auto_find_batch_size`: False
415
+ - `log_on_each_node`: True
416
+ - `logging_nan_inf_filter`: True
417
+ - `include_num_input_tokens_seen`: no
418
+ - `log_level`: passive
419
+ - `log_level_replica`: warning
420
+ - `disable_tqdm`: False
421
+ - `project`: huggingface
422
+ - `trackio_space_id`: trackio
423
+ - `eval_strategy`: steps
424
+ - `per_device_eval_batch_size`: 8
425
+ - `prediction_loss_only`: True
426
+ - `eval_on_start`: False
427
+ - `eval_do_concat_batches`: True
428
+ - `eval_use_gather_object`: False
429
+ - `eval_accumulation_steps`: None
430
+ - `include_for_metrics`: []
431
+ - `batch_eval_metrics`: False
432
+ - `save_only_model`: False
433
+ - `save_on_each_node`: False
434
+ - `enable_jit_checkpoint`: False
435
+ - `push_to_hub`: False
436
+ - `hub_private_repo`: None
437
+ - `hub_model_id`: None
438
+ - `hub_strategy`: every_save
439
+ - `hub_always_push`: False
440
+ - `hub_revision`: None
441
+ - `load_best_model_at_end`: False
442
+ - `ignore_data_skip`: False
443
+ - `restore_callback_states_from_checkpoint`: False
444
+ - `full_determinism`: False
445
+ - `seed`: 42
446
+ - `data_seed`: None
447
+ - `use_cpu`: False
448
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
449
+ - `parallelism_config`: None
450
+ - `dataloader_drop_last`: False
451
+ - `dataloader_num_workers`: 0
452
+ - `dataloader_pin_memory`: True
453
+ - `dataloader_persistent_workers`: False
454
+ - `dataloader_prefetch_factor`: None
455
+ - `remove_unused_columns`: True
456
+ - `label_names`: None
457
+ - `train_sampling_strategy`: random
458
+ - `length_column_name`: length
459
+ - `ddp_find_unused_parameters`: None
460
+ - `ddp_bucket_cap_mb`: None
461
+ - `ddp_broadcast_buffers`: False
462
+ - `ddp_backend`: None
463
+ - `ddp_timeout`: 1800
464
+ - `fsdp`: []
465
+ - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
466
+ - `deepspeed`: None
467
+ - `debug`: []
468
+ - `skip_memory_metrics`: True
469
+ - `do_predict`: False
470
+ - `resume_from_checkpoint`: None
471
+ - `warmup_ratio`: 0.1
472
+ - `local_rank`: -1
473
+ - `prompts`: None
474
+ - `batch_sampler`: no_duplicates
475
+ - `multi_dataset_batch_sampler`: proportional
476
+ - `router_mapping`: {}
477
+ - `learning_rate_mapping`: {}
478
+
479
+ </details>
480
+
481
+ ### Training Logs
482
+ <details><summary>Click to expand</summary>
483
+
484
+ | Epoch | Step | Training Loss | train loss | dev-768_cosine_accuracy | dev-512_cosine_accuracy | dev-256_cosine_accuracy | dev-128_cosine_accuracy | dev-64_cosine_accuracy |
485
+ |:------:|:------:|:-------------:|:----------:|:-----------------------:|:-----------------------:|:-----------------------:|:-----------------------:|:----------------------:|
486
+ | 3.3082 | 102200 | 0.2592 | - | - | - | - | - | - |
487
+ | 3.3147 | 102400 | 0.2543 | - | - | - | - | - | - |
488
+ | 3.3211 | 102600 | 0.2641 | - | - | - | - | - | - |
489
+ | 3.3276 | 102800 | 0.2435 | - | - | - | - | - | - |
490
+ | 3.3341 | 103000 | 0.2510 | - | - | - | - | - | - |
491
+ | 3.3406 | 103200 | 0.2481 | - | - | - | - | - | - |
492
+ | 3.3470 | 103400 | 0.2627 | - | - | - | - | - | - |
493
+ | 3.3535 | 103600 | 0.2480 | - | - | - | - | - | - |
494
+ | 3.3600 | 103800 | 0.2636 | - | - | - | - | - | - |
495
+ | 3.3665 | 104000 | 0.2619 | - | - | - | - | - | - |
496
+ | 3.3729 | 104200 | 0.2423 | - | - | - | - | - | - |
497
+ | 3.3794 | 104400 | 0.2505 | - | - | - | - | - | - |
498
+ | 3.3859 | 104600 | 0.2604 | - | - | - | - | - | - |
499
+ | 3.3924 | 104800 | 0.2460 | - | - | - | - | - | - |
500
+ | 3.3988 | 105000 | 0.2440 | - | - | - | - | - | - |
501
+ | 3.4053 | 105200 | 0.2641 | - | - | - | - | - | - |
502
+ | 3.4118 | 105400 | 0.2573 | - | - | - | - | - | - |
503
+ | 3.4183 | 105600 | 0.2613 | - | - | - | - | - | - |
504
+ | 3.4247 | 105800 | 0.2746 | - | - | - | - | - | - |
505
+ | 3.4312 | 106000 | 0.2578 | - | - | - | - | - | - |
506
+ | 3.4377 | 106200 | 0.2445 | - | - | - | - | - | - |
507
+ | 3.4442 | 106400 | 0.2530 | - | - | - | - | - | - |
508
+ | 3.4506 | 106600 | 0.2644 | - | - | - | - | - | - |
509
+ | 3.4571 | 106800 | 0.2656 | - | - | - | - | - | - |
510
+ | 3.4636 | 107000 | 0.2520 | - | - | - | - | - | - |
511
+ | 3.4700 | 107200 | 0.2527 | - | - | - | - | - | - |
512
+ | 3.4765 | 107400 | 0.2534 | - | - | - | - | - | - |
513
+ | 3.4830 | 107600 | 0.2530 | - | - | - | - | - | - |
514
+ | 3.4895 | 107800 | 0.2614 | - | - | - | - | - | - |
515
+ | 3.4959 | 108000 | 0.2517 | 0.2252 | 0.9849 | 0.9849 | 0.9851 | 0.9849 | 0.9841 |
516
+ | 3.5024 | 108200 | 0.2589 | - | - | - | - | - | - |
517
+ | 3.5089 | 108400 | 0.2582 | - | - | - | - | - | - |
518
+ | 3.5154 | 108600 | 0.2553 | - | - | - | - | - | - |
519
+ | 3.5218 | 108800 | 0.2457 | - | - | - | - | - | - |
520
+ | 3.5283 | 109000 | 0.2662 | - | - | - | - | - | - |
521
+ | 3.5348 | 109200 | 0.2619 | - | - | - | - | - | - |
522
+ | 3.5413 | 109400 | 0.2556 | - | - | - | - | - | - |
523
+ | 3.5477 | 109600 | 0.2635 | - | - | - | - | - | - |
524
+ | 3.5542 | 109800 | 0.2550 | - | - | - | - | - | - |
525
+ | 3.5607 | 110000 | 0.2670 | - | - | - | - | - | - |
526
+ | 3.5672 | 110200 | 0.2660 | - | - | - | - | - | - |
527
+ | 3.5736 | 110400 | 0.2604 | - | - | - | - | - | - |
528
+ | 3.5801 | 110600 | 0.2574 | - | - | - | - | - | - |
529
+ | 3.5866 | 110800 | 0.2607 | - | - | - | - | - | - |
530
+ | 3.5931 | 111000 | 0.2465 | - | - | - | - | - | - |
531
+ | 3.5995 | 111200 | 0.2790 | - | - | - | - | - | - |
532
+ | 3.6060 | 111400 | 0.2681 | - | - | - | - | - | - |
533
+ | 3.6125 | 111600 | 0.2654 | - | - | - | - | - | - |
534
+ | 3.6190 | 111800 | 0.2640 | - | - | - | - | - | - |
535
+ | 3.6254 | 112000 | 0.2774 | - | - | - | - | - | - |
536
+ | 3.6319 | 112200 | 0.2568 | - | - | - | - | - | - |
537
+ | 3.6384 | 112400 | 0.2665 | - | - | - | - | - | - |
538
+ | 3.6448 | 112600 | 0.2532 | - | - | - | - | - | - |
539
+ | 3.6513 | 112800 | 0.2613 | - | - | - | - | - | - |
540
+ | 3.6578 | 113000 | 0.2413 | - | - | - | - | - | - |
541
+ | 3.6643 | 113200 | 0.2788 | - | - | - | - | - | - |
542
+ | 3.6707 | 113400 | 0.2586 | - | - | - | - | - | - |
543
+ | 3.6772 | 113600 | 0.2602 | - | - | - | - | - | - |
544
+ | 3.6837 | 113800 | 0.2708 | - | - | - | - | - | - |
545
+ | 3.6902 | 114000 | 0.2556 | 0.2188 | 0.9851 | 0.9852 | 0.9854 | 0.9853 | 0.9844 |
546
+ | 3.6966 | 114200 | 0.2576 | - | - | - | - | - | - |
547
+ | 3.7031 | 114400 | 0.2713 | - | - | - | - | - | - |
548
+ | 3.7096 | 114600 | 0.2748 | - | - | - | - | - | - |
549
+ | 3.7161 | 114800 | 0.2542 | - | - | - | - | - | - |
550
+ | 3.7225 | 115000 | 0.2647 | - | - | - | - | - | - |
551
+ | 3.7290 | 115200 | 0.2751 | - | - | - | - | - | - |
552
+ | 3.7355 | 115400 | 0.2534 | - | - | - | - | - | - |
553
+ | 3.7420 | 115600 | 0.2577 | - | - | - | - | - | - |
554
+ | 3.7484 | 115800 | 0.2722 | - | - | - | - | - | - |
555
+ | 3.7549 | 116000 | 0.2717 | - | - | - | - | - | - |
556
+ | 3.7614 | 116200 | 0.2737 | - | - | - | - | - | - |
557
+ | 3.7679 | 116400 | 0.2725 | - | - | - | - | - | - |
558
+ | 3.7743 | 116600 | 0.2587 | - | - | - | - | - | - |
559
+ | 3.7808 | 116800 | 0.2623 | - | - | - | - | - | - |
560
+ | 3.7873 | 117000 | 0.2659 | - | - | - | - | - | - |
561
+ | 3.7938 | 117200 | 0.2735 | - | - | - | - | - | - |
562
+ | 3.8002 | 117400 | 0.2847 | - | - | - | - | - | - |
563
+ | 3.8067 | 117600 | 0.2636 | - | - | - | - | - | - |
564
+ | 3.8132 | 117800 | 0.2777 | - | - | - | - | - | - |
565
+ | 3.8196 | 118000 | 0.2751 | - | - | - | - | - | - |
566
+ | 3.8261 | 118200 | 0.2609 | - | - | - | - | - | - |
567
+ | 3.8326 | 118400 | 0.2684 | - | - | - | - | - | - |
568
+ | 3.8391 | 118600 | 0.2772 | - | - | - | - | - | - |
569
+ | 3.8455 | 118800 | 0.2684 | - | - | - | - | - | - |
570
+ | 3.8520 | 119000 | 0.2682 | - | - | - | - | - | - |
571
+ | 3.8585 | 119200 | 0.2784 | - | - | - | - | - | - |
572
+ | 3.8650 | 119400 | 0.2735 | - | - | - | - | - | - |
573
+ | 3.8714 | 119600 | 0.2848 | - | - | - | - | - | - |
574
+ | 3.8779 | 119800 | 0.2638 | - | - | - | - | - | - |
575
+ | 3.8844 | 120000 | 0.2711 | 0.2171 | 0.9854 | 0.9855 | 0.9856 | 0.9855 | 0.9846 |
576
+ | 3.8909 | 120200 | 0.2825 | - | - | - | - | - | - |
577
+ | 3.8973 | 120400 | 0.2724 | - | - | - | - | - | - |
578
+ | 3.9038 | 120600 | 0.3078 | - | - | - | - | - | - |
579
+ | 3.9103 | 120800 | 0.2806 | - | - | - | - | - | - |
580
+ | 3.9168 | 121000 | 0.2631 | - | - | - | - | - | - |
581
+ | 3.9232 | 121200 | 0.2892 | - | - | - | - | - | - |
582
+ | 3.9297 | 121400 | 0.2791 | - | - | - | - | - | - |
583
+ | 3.9362 | 121600 | 0.2874 | - | - | - | - | - | - |
584
+ | 3.9427 | 121800 | 0.2602 | - | - | - | - | - | - |
585
+ | 3.9491 | 122000 | 0.2988 | - | - | - | - | - | - |
586
+ | 3.9556 | 122200 | 0.2935 | - | - | - | - | - | - |
587
+ | 3.9621 | 122400 | 0.2999 | - | - | - | - | - | - |
588
+ | 3.9686 | 122600 | 0.2930 | - | - | - | - | - | - |
589
+ | 3.9750 | 122800 | 0.2784 | - | - | - | - | - | - |
590
+ | 3.9815 | 123000 | 0.3013 | - | - | - | - | - | - |
591
+ | 3.9880 | 123200 | 0.2919 | - | - | - | - | - | - |
592
+ | 3.9944 | 123400 | 0.3011 | - | - | - | - | - | - |
593
+
594
+ </details>
595
+
596
+ ### Framework Versions
597
+ - Python: 3.10.19
598
+ - Sentence Transformers: 5.2.3
599
+ - Transformers: 5.2.0
600
+ - PyTorch: 2.6.0+cu124
601
+ - Accelerate: 1.12.0
602
+ - Datasets: 4.5.0
603
+ - Tokenizers: 0.22.2
604
+
605
+ ## Citation
606
+
607
+ ### BibTeX
608
+
609
+ #### Sentence Transformers
610
+ ```bibtex
611
+ @inproceedings{reimers-2019-sentence-bert,
612
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
613
+ author = "Reimers, Nils and Gurevych, Iryna",
614
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
615
+ month = "11",
616
+ year = "2019",
617
+ publisher = "Association for Computational Linguistics",
618
+ url = "https://arxiv.org/abs/1908.10084",
619
+ }
620
+ ```
621
+
622
+ #### MatryoshkaLoss
623
+ ```bibtex
624
+ @misc{kusupati2024matryoshka,
625
+ title={Matryoshka Representation Learning},
626
+ author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
627
+ year={2024},
628
+ eprint={2205.13147},
629
+ archivePrefix={arXiv},
630
+ primaryClass={cs.LG}
631
+ }
632
+ ```
633
+
634
+ #### MultipleNegativesRankingLoss
635
+ ```bibtex
636
+ @misc{henderson2017efficient,
637
+ title={Efficient Natural Language Response Suggestion for Smart Reply},
638
+ author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
639
+ year={2017},
640
+ eprint={1705.00652},
641
+ archivePrefix={arXiv},
642
+ primaryClass={cs.CL}
643
+ }
644
+ ```
645
+
646
+ <!--
647
+ ## Glossary
648
+
649
+ *Clearly define terms in order to be accessible across audiences.*
650
+ -->
651
+
652
+ <!--
653
+ ## Model Card Authors
654
+
655
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
656
+ -->
657
+
658
+ <!--
659
+ ## Model Card Contact
660
+
661
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
662
+ -->
epoch4/model/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_cross_attention": false,
3
+ "architectures": [
4
+ "BertModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": null,
8
+ "classifier_dropout": null,
9
+ "dtype": "float32",
10
+ "eos_token_id": null,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 768,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 3072,
16
+ "is_decoder": false,
17
+ "layer_norm_eps": 1e-12,
18
+ "max_position_embeddings": 512,
19
+ "model_type": "bert",
20
+ "num_attention_heads": 12,
21
+ "num_hidden_layers": 12,
22
+ "pad_token_id": 0,
23
+ "tie_word_embeddings": true,
24
+ "transformers_version": "5.2.0",
25
+ "type_vocab_size": 2,
26
+ "use_cache": true,
27
+ "vocab_size": 64000
28
+ }
epoch4/model/config_sentence_transformers.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "SentenceTransformer",
3
+ "__version__": {
4
+ "sentence_transformers": "5.2.3",
5
+ "transformers": "5.2.0",
6
+ "pytorch": "2.6.0+cu124"
7
+ },
8
+ "prompts": {
9
+ "query": "",
10
+ "document": ""
11
+ },
12
+ "default_prompt_name": null,
13
+ "similarity_fn_name": "cosine"
14
+ }
epoch4/model/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80fdddfe99a3ae2413a5a95045cba3b2b2d660b149f907f3f6348c03343daf3d
3
+ size 540795728
epoch4/model/modules.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.models.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.models.Pooling"
13
+ }
14
+ ]
epoch4/model/sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 512,
3
+ "do_lower_case": false
4
+ }
epoch4/model/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
epoch4/model/tokenizer_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "cls_token": "[CLS]",
4
+ "do_basic_tokenize": true,
5
+ "do_lower_case": false,
6
+ "is_local": true,
7
+ "mask_token": "[MASK]",
8
+ "max_len": 512,
9
+ "model_max_length": 512,
10
+ "never_split": [
11
+ "[بريد]",
12
+ "[مستخدم]",
13
+ "[رابط]"
14
+ ],
15
+ "pad_token": "[PAD]",
16
+ "sep_token": "[SEP]",
17
+ "strip_accents": null,
18
+ "tokenize_chinese_chars": true,
19
+ "tokenizer_class": "BertTokenizer",
20
+ "unk_token": "[UNK]"
21
+ }
epoch4/output.log ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ --- RESUMING FROM: /home/skiredj.abderrahman/khalil/sbert_training/output/arabert_20260224_1730/checkpoint-102000 ---
2
+ {'loss': '0.2592', 'grad_norm': '3.323', 'learning_rate': '3.844e-06', 'epoch': '3.308'}
3
+ {'loss': '0.2543', 'grad_norm': '1.973', 'learning_rate': '3.808e-06', 'epoch': '3.315'}
4
+ {'loss': '0.2641', 'grad_norm': '4.95', 'learning_rate': '3.772e-06', 'epoch': '3.321'}
5
+ {'loss': '0.2435', 'grad_norm': '2.661', 'learning_rate': '3.736e-06', 'epoch': '3.328'}
6
+ {'loss': '0.251', 'grad_norm': '3.434', 'learning_rate': '3.7e-06', 'epoch': '3.334'}
7
+ {'loss': '0.2481', 'grad_norm': '2.562', 'learning_rate': '3.664e-06', 'epoch': '3.341'}
8
+ {'loss': '0.2627', 'grad_norm': '2.6', 'learning_rate': '3.628e-06', 'epoch': '3.347'}
9
+ {'loss': '0.248', 'grad_norm': '2.941', 'learning_rate': '3.592e-06', 'epoch': '3.354'}
10
+ {'loss': '0.2636', 'grad_norm': '2.153', 'learning_rate': '3.556e-06', 'epoch': '3.36'}
11
+ {'loss': '0.2619', 'grad_norm': '2.729', 'learning_rate': '3.52e-06', 'epoch': '3.366'}
12
+ {'loss': '0.2423', 'grad_norm': '2.413', 'learning_rate': '3.484e-06', 'epoch': '3.373'}
13
+ {'loss': '0.2505', 'grad_norm': '4.396', 'learning_rate': '3.448e-06', 'epoch': '3.379'}
14
+ {'loss': '0.2604', 'grad_norm': '3.629', 'learning_rate': '3.412e-06', 'epoch': '3.386'}
15
+ {'loss': '0.246', 'grad_norm': '5.426', 'learning_rate': '3.376e-06', 'epoch': '3.392'}
16
+ {'loss': '0.244', 'grad_norm': '1.848', 'learning_rate': '3.34e-06', 'epoch': '3.399'}
17
+ {'loss': '0.2641', 'grad_norm': '5.125', 'learning_rate': '3.304e-06', 'epoch': '3.405'}
18
+ {'loss': '0.2573', 'grad_norm': '3.377', 'learning_rate': '3.268e-06', 'epoch': '3.412'}
19
+ {'loss': '0.2613', 'grad_norm': '2.842', 'learning_rate': '3.232e-06', 'epoch': '3.418'}
20
+ {'loss': '0.2746', 'grad_norm': '3.197', 'learning_rate': '3.196e-06', 'epoch': '3.425'}
21
+ {'loss': '0.2578', 'grad_norm': '2.158', 'learning_rate': '3.16e-06', 'epoch': '3.431'}
22
+ {'loss': '0.2445', 'grad_norm': '2.634', 'learning_rate': '3.124e-06', 'epoch': '3.438'}
23
+ {'loss': '0.253', 'grad_norm': '2.139', 'learning_rate': '3.088e-06', 'epoch': '3.444'}
24
+ {'loss': '0.2644', 'grad_norm': '3.682', 'learning_rate': '3.052e-06', 'epoch': '3.451'}
25
+ {'loss': '0.2656', 'grad_norm': '4.607', 'learning_rate': '3.016e-06', 'epoch': '3.457'}
26
+ {'loss': '0.252', 'grad_norm': '2.139', 'learning_rate': '2.98e-06', 'epoch': '3.464'}
27
+ {'loss': '0.2527', 'grad_norm': '2.422', 'learning_rate': '2.944e-06', 'epoch': '3.47'}
28
+ {'loss': '0.2534', 'grad_norm': '3.598', 'learning_rate': '2.908e-06', 'epoch': '3.477'}
29
+ {'loss': '0.253', 'grad_norm': '1.388', 'learning_rate': '2.872e-06', 'epoch': '3.483'}
30
+ {'loss': '0.2614', 'grad_norm': '3.057', 'learning_rate': '2.837e-06', 'epoch': '3.489'}
31
+ {'loss': '0.2517', 'grad_norm': '2.296', 'learning_rate': '2.801e-06', 'epoch': '3.496'}
32
+ {'eval_train_loss': '0.2252', 'eval_dev-768_cosine_accuracy': '0.9849', 'eval_dev-512_cosine_accuracy': '0.9849', 'eval_dev-256_cosine_accuracy': '0.9851', 'eval_dev-128_cosine_accuracy': '0.9849', 'eval_dev-64_cosine_accuracy': '0.9841', 'eval_sequential_score': '0.9849', 'eval_train_runtime': '9737', 'eval_train_samples_per_second': '116', 'eval_train_steps_per_second': '14.5', 'epoch': '3.496'}
33
+ {'loss': '0.2589', 'grad_norm': '2.231', 'learning_rate': '2.765e-06', 'epoch': '3.502'}
34
+ {'loss': '0.2582', 'grad_norm': '5.077', 'learning_rate': '2.729e-06', 'epoch': '3.509'}
35
+ {'loss': '0.2553', 'grad_norm': '4.234', 'learning_rate': '2.693e-06', 'epoch': '3.515'}
36
+ {'loss': '0.2457', 'grad_norm': '2.167', 'learning_rate': '2.657e-06', 'epoch': '3.522'}
37
+ {'loss': '0.2662', 'grad_norm': '4.357', 'learning_rate': '2.621e-06', 'epoch': '3.528'}
38
+ {'loss': '0.2619', 'grad_norm': '3.987', 'learning_rate': '2.585e-06', 'epoch': '3.535'}
39
+ {'loss': '0.2556', 'grad_norm': '1.6', 'learning_rate': '2.549e-06', 'epoch': '3.541'}
40
+ {'loss': '0.2635', 'grad_norm': '2.069', 'learning_rate': '2.513e-06', 'epoch': '3.548'}
41
+ {'loss': '0.255', 'grad_norm': '2.314', 'learning_rate': '2.477e-06', 'epoch': '3.554'}
42
+ {'loss': '0.267', 'grad_norm': '4.255', 'learning_rate': '2.441e-06', 'epoch': '3.561'}
43
+ {'loss': '0.266', 'grad_norm': '3.208', 'learning_rate': '2.405e-06', 'epoch': '3.567'}
44
+ {'loss': '0.2604', 'grad_norm': '3.669', 'learning_rate': '2.369e-06', 'epoch': '3.574'}
45
+ {'loss': '0.2574', 'grad_norm': '3.103', 'learning_rate': '2.333e-06', 'epoch': '3.58'}
46
+ {'loss': '0.2607', 'grad_norm': '4.632', 'learning_rate': '2.297e-06', 'epoch': '3.587'}
47
+ {'loss': '0.2465', 'grad_norm': '4', 'learning_rate': '2.261e-06', 'epoch': '3.593'}
48
+ {'loss': '0.279', 'grad_norm': '3.647', 'learning_rate': '2.225e-06', 'epoch': '3.6'}
49
+ {'loss': '0.2681', 'grad_norm': '5.511', 'learning_rate': '2.189e-06', 'epoch': '3.606'}
50
+ {'loss': '0.2654', 'grad_norm': '1.217', 'learning_rate': '2.153e-06', 'epoch': '3.612'}
51
+ {'loss': '0.264', 'grad_norm': '4.641', 'learning_rate': '2.117e-06', 'epoch': '3.619'}
52
+ {'loss': '0.2774', 'grad_norm': '3.571', 'learning_rate': '2.081e-06', 'epoch': '3.625'}
53
+ {'loss': '0.2568', 'grad_norm': '3.808', 'learning_rate': '2.045e-06', 'epoch': '3.632'}
54
+ {'loss': '0.2665', 'grad_norm': '3.178', 'learning_rate': '2.009e-06', 'epoch': '3.638'}
55
+ {'loss': '0.2532', 'grad_norm': '1.566', 'learning_rate': '1.973e-06', 'epoch': '3.645'}
56
+ {'loss': '0.2613', 'grad_norm': '3.262', 'learning_rate': '1.937e-06', 'epoch': '3.651'}
57
+ {'loss': '0.2413', 'grad_norm': '4.082', 'learning_rate': '1.901e-06', 'epoch': '3.658'}
58
+ {'loss': '0.2788', 'grad_norm': '4.851', 'learning_rate': '1.865e-06', 'epoch': '3.664'}
59
+ {'loss': '0.2586', 'grad_norm': '5.756', 'learning_rate': '1.829e-06', 'epoch': '3.671'}
60
+ {'loss': '0.2602', 'grad_norm': '5.288', 'learning_rate': '1.793e-06', 'epoch': '3.677'}
61
+ {'loss': '0.2708', 'grad_norm': '3.492', 'learning_rate': '1.758e-06', 'epoch': '3.684'}
62
+ {'loss': '0.2556', 'grad_norm': '2.688', 'learning_rate': '1.722e-06', 'epoch': '3.69'}
63
+ {'eval_train_loss': '0.2188', 'eval_dev-768_cosine_accuracy': '0.9851', 'eval_dev-512_cosine_accuracy': '0.9852', 'eval_dev-256_cosine_accuracy': '0.9854', 'eval_dev-128_cosine_accuracy': '0.9853', 'eval_dev-64_cosine_accuracy': '0.9844', 'eval_sequential_score': '0.9851', 'eval_train_runtime': '9310', 'eval_train_samples_per_second': '121.3', 'eval_train_steps_per_second': '15.17', 'epoch': '3.69'}
64
+ {'loss': '0.2576', 'grad_norm': '2.428', 'learning_rate': '1.686e-06', 'epoch': '3.697'}
65
+ {'loss': '0.2713', 'grad_norm': '2.93', 'learning_rate': '1.65e-06', 'epoch': '3.703'}
66
+ {'loss': '0.2748', 'grad_norm': '3.659', 'learning_rate': '1.614e-06', 'epoch': '3.71'}
67
+ {'loss': '0.2542', 'grad_norm': '3.45', 'learning_rate': '1.578e-06', 'epoch': '3.716'}
68
+ {'loss': '0.2647', 'grad_norm': '4.824', 'learning_rate': '1.542e-06', 'epoch': '3.723'}
69
+ {'loss': '0.2751', 'grad_norm': '1.834', 'learning_rate': '1.506e-06', 'epoch': '3.729'}
70
+ {'loss': '0.2534', 'grad_norm': '4.17', 'learning_rate': '1.47e-06', 'epoch': '3.735'}
71
+ {'loss': '0.2577', 'grad_norm': '3.893', 'learning_rate': '1.434e-06', 'epoch': '3.742'}
72
+ {'loss': '0.2722', 'grad_norm': '4.691', 'learning_rate': '1.398e-06', 'epoch': '3.748'}
73
+ {'loss': '0.2717', 'grad_norm': '3.968', 'learning_rate': '1.362e-06', 'epoch': '3.755'}
74
+ {'loss': '0.2737', 'grad_norm': '4.474', 'learning_rate': '1.326e-06', 'epoch': '3.761'}
75
+ {'loss': '0.2725', 'grad_norm': '3.462', 'learning_rate': '1.29e-06', 'epoch': '3.768'}
76
+ {'loss': '0.2587', 'grad_norm': '1.857', 'learning_rate': '1.254e-06', 'epoch': '3.774'}
77
+ {'loss': '0.2623', 'grad_norm': '2.164', 'learning_rate': '1.218e-06', 'epoch': '3.781'}
78
+ {'loss': '0.2659', 'grad_norm': '2.674', 'learning_rate': '1.182e-06', 'epoch': '3.787'}
79
+ {'loss': '0.2735', 'grad_norm': '1.56', 'learning_rate': '1.146e-06', 'epoch': '3.794'}
80
+ {'loss': '0.2847', 'grad_norm': '4.402', 'learning_rate': '1.11e-06', 'epoch': '3.8'}
81
+ {'loss': '0.2636', 'grad_norm': '5.62', 'learning_rate': '1.074e-06', 'epoch': '3.807'}
82
+ {'loss': '0.2777', 'grad_norm': '4.51', 'learning_rate': '1.038e-06', 'epoch': '3.813'}
83
+ {'loss': '0.2751', 'grad_norm': '3.339', 'learning_rate': '1.002e-06', 'epoch': '3.82'}
84
+ {'loss': '0.2609', 'grad_norm': '3.817', 'learning_rate': '9.662e-07', 'epoch': '3.826'}
85
+ {'loss': '0.2684', 'grad_norm': '4.592', 'learning_rate': '9.303e-07', 'epoch': '3.833'}
86
+ {'loss': '0.2772', 'grad_norm': '2.499', 'learning_rate': '8.943e-07', 'epoch': '3.839'}
87
+ {'loss': '0.2684', 'grad_norm': '4.28', 'learning_rate': '8.583e-07', 'epoch': '3.846'}
88
+ {'loss': '0.2682', 'grad_norm': '3.179', 'learning_rate': '8.224e-07', 'epoch': '3.852'}
89
+ {'loss': '0.2784', 'grad_norm': '5.42', 'learning_rate': '7.864e-07', 'epoch': '3.858'}
90
+ {'loss': '0.2735', 'grad_norm': '2.668', 'learning_rate': '7.504e-07', 'epoch': '3.865'}
91
+ {'loss': '0.2848', 'grad_norm': '4.855', 'learning_rate': '7.145e-07', 'epoch': '3.871'}
92
+ {'loss': '0.2638', 'grad_norm': '2.601', 'learning_rate': '6.785e-07', 'epoch': '3.878'}
93
+ {'loss': '0.2711', 'grad_norm': '4.021', 'learning_rate': '6.425e-07', 'epoch': '3.884'}
94
+ {'eval_train_loss': '0.2171', 'eval_dev-768_cosine_accuracy': '0.9854', 'eval_dev-512_cosine_accuracy': '0.9855', 'eval_dev-256_cosine_accuracy': '0.9856', 'eval_dev-128_cosine_accuracy': '0.9855', 'eval_dev-64_cosine_accuracy': '0.9846', 'eval_sequential_score': '0.9854', 'eval_train_runtime': '9383', 'eval_train_samples_per_second': '120.4', 'eval_train_steps_per_second': '15.05', 'epoch': '3.884'}
95
+ {'loss': '0.2825', 'grad_norm': '2.434', 'learning_rate': '6.066e-07', 'epoch': '3.891'}
96
+ {'loss': '0.2724', 'grad_norm': '3.687', 'learning_rate': '5.706e-07', 'epoch': '3.897'}
97
+ {'loss': '0.3078', 'grad_norm': '5.537', 'learning_rate': '5.346e-07', 'epoch': '3.904'}
98
+ {'loss': '0.2806', 'grad_norm': '3.519', 'learning_rate': '4.987e-07', 'epoch': '3.91'}
99
+ {'loss': '0.2631', 'grad_norm': '3.982', 'learning_rate': '4.627e-07', 'epoch': '3.917'}
100
+ {'loss': '0.2892', 'grad_norm': '3.621', 'learning_rate': '4.267e-07', 'epoch': '3.923'}
101
+ {'loss': '0.2791', 'grad_norm': '4.467', 'learning_rate': '3.908e-07', 'epoch': '3.93'}
102
+ {'loss': '0.2874', 'grad_norm': '4.958', 'learning_rate': '3.548e-07', 'epoch': '3.936'}
103
+ {'loss': '0.2602', 'grad_norm': '3.852', 'learning_rate': '3.188e-07', 'epoch': '3.943'}
104
+ {'loss': '0.2988', 'grad_norm': '2.639', 'learning_rate': '2.829e-07', 'epoch': '3.949'}
105
+ {'loss': '0.2935', 'grad_norm': '1.878', 'learning_rate': '2.469e-07', 'epoch': '3.956'}
106
+ {'loss': '0.2999', 'grad_norm': '3.6', 'learning_rate': '2.109e-07', 'epoch': '3.962'}
107
+ {'loss': '0.293', 'grad_norm': '2.28', 'learning_rate': '1.75e-07', 'epoch': '3.969'}
108
+ {'loss': '0.2784', 'grad_norm': '4.347', 'learning_rate': '1.39e-07', 'epoch': '3.975'}
109
+ {'loss': '0.3013', 'grad_norm': '5.014', 'learning_rate': '1.03e-07', 'epoch': '3.982'}
110
+ {'loss': '0.2919', 'grad_norm': '2.171', 'learning_rate': '6.708e-08', 'epoch': '3.988'}
111
+ {'loss': '0.3011', 'grad_norm': '5.241', 'learning_rate': '3.111e-08', 'epoch': '3.994'}
112
+ {'train_runtime': '4.115e+04', 'train_samples_per_second': '384.4', 'train_steps_per_second': '3.003', 'train_loss': '0.04643', 'epoch': '4'}
113
+ model saved successfully