Hugging Face's logo

T0KII
/
MASRIBERTV4

Fill-Mask
Transformers
Safetensors
bert
Generated from Trainer
Model card Files
xet
 Community
T0KII commited on
Commit
9dd244d
·
verified ·
1 Parent(s): 8688f1d

Model save

Browse files
Files changed (1) hide show
  1. README.md +106 -0
README.md ADDED
@@ -0,0 +1,106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ base_model: UBC-NLP/MARBERTv2
4
+ tags:
5
+ - generated_from_trainer
6
+ model-index:
7
+ - name: MASRIBERTV4
8
+ results: []
9
+ ---
10
+
11
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
12
+ should probably proofread and complete it, then remove this comment. -->
13
+
14
+ # MASRIBERTV4
15
+
16
+ This model is a fine-tuned version of [UBC-NLP/MARBERTv2](https://huggingface.co/UBC-NLP/MARBERTv2) on an unknown dataset.
17
+ It achieves the following results on the evaluation set:
18
+ - Loss: 1.4905
19
+
20
+ ## Model description
21
+
22
+ More information needed
23
+
24
+ ## Intended uses & limitations
25
+
26
+ More information needed
27
+
28
+ ## Training and evaluation data
29
+
30
+ More information needed
31
+
32
+ ## Training procedure
33
+
34
+ ### Training hyperparameters
35
+
36
+ The following hyperparameters were used during training:
37
+ - learning_rate: 5e-05
38
+ - train_batch_size: 8
39
+ - eval_batch_size: 8
40
+ - seed: 42
41
+ - gradient_accumulation_steps: 8
42
+ - total_train_batch_size: 64
43
+ - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
44
+ - lr_scheduler_type: cosine
45
+ - lr_scheduler_warmup_steps: 10000
46
+ - num_epochs: 2
47
+ - mixed_precision_training: Native AMP
48
+
49
+ ### Training results
50
+
51
+ | Training Loss | Epoch | Step | Validation Loss |
52
+ |:-------------:|:------:|:-----:|:---------------:|
53
+ | 49.3326 | 0.0427 | 500 | 5.4990 |
54
+ | 42.0818 | 0.0854 | 1000 | 4.8453 |
55
+ | 38.9605 | 0.1281 | 1500 | 4.5192 |
56
+ | 36.9452 | 0.1709 | 2000 | 4.3034 |
57
+ | 35.4721 | 0.2136 | 2500 | 4.1469 |
58
+ | 17.1800 | 0.2563 | 3000 | 2.0153 |
59
+ | 16.6866 | 0.2990 | 3500 | 1.9642 |
60
+ | 16.3991 | 0.3417 | 4000 | 1.9193 |
61
+ | 16.0343 | 0.3844 | 4500 | 1.8890 |
62
+ | 15.7752 | 0.4271 | 5000 | 1.8541 |
63
+ | 15.5412 | 0.4699 | 5500 | 1.8296 |
64
+ | 15.3676 | 0.5126 | 6000 | 1.8037 |
65
+ | 15.1004 | 0.5553 | 6500 | 1.7819 |
66
+ | 14.9579 | 0.5980 | 7000 | 1.7591 |
67
+ | 14.8206 | 0.6407 | 7500 | 1.7475 |
68
+ | 14.5731 | 0.6834 | 8000 | 1.7312 |
69
+ | 14.4853 | 0.7262 | 8500 | 1.7171 |
70
+ | 14.4356 | 0.7689 | 9000 | 1.7090 |
71
+ | 14.2340 | 0.8116 | 9500 | 1.6929 |
72
+ | 14.2497 | 0.8543 | 10000 | 1.6809 |
73
+ | 14.1660 | 0.8970 | 10500 | 1.6745 |
74
+ | 14.0404 | 0.9397 | 11000 | 1.6584 |
75
+ | 13.8540 | 0.9824 | 11500 | 1.6474 |
76
+ | 13.7538 | 1.0251 | 12000 | 1.6362 |
77
+ | 13.6777 | 1.0678 | 12500 | 1.6223 |
78
+ | 13.5928 | 1.1105 | 13000 | 1.6111 |
79
+ | 13.4528 | 1.1533 | 13500 | 1.6016 |
80
+ | 13.3583 | 1.1960 | 14000 | 1.5926 |
81
+ | 13.3129 | 1.2387 | 14500 | 1.5797 |
82
+ | 13.2261 | 1.2814 | 15000 | 1.5714 |
83
+ | 13.2270 | 1.3241 | 15500 | 1.5613 |
84
+ | 13.0836 | 1.3668 | 16000 | 1.5562 |
85
+ | 13.0298 | 1.4096 | 16500 | 1.5463 |
86
+ | 12.9826 | 1.4523 | 17000 | 1.5360 |
87
+ | 12.9178 | 1.4950 | 17500 | 1.5267 |
88
+ | 12.8210 | 1.5377 | 18000 | 1.5218 |
89
+ | 12.7591 | 1.5804 | 18500 | 1.5172 |
90
+ | 12.7104 | 1.6231 | 19000 | 1.5090 |
91
+ | 12.6183 | 1.6658 | 19500 | 1.5056 |
92
+ | 12.5794 | 1.7086 | 20000 | 1.5013 |
93
+ | 12.5963 | 1.7513 | 20500 | 1.4976 |
94
+ | 12.5140 | 1.7940 | 21000 | 1.4922 |
95
+ | 12.4895 | 1.8367 | 21500 | 1.4932 |
96
+ | 12.4931 | 1.8794 | 22000 | 1.4914 |
97
+ | 12.5685 | 1.9221 | 22500 | 1.4908 |
98
+ | 12.4823 | 1.9648 | 23000 | 1.4929 |
99
+
100
+
101
+ ### Framework versions
102
+
103
+ - Transformers 5.0.0
104
+ - Pytorch 2.10.0+cu128
105
+ - Datasets 5.0.0
106
+ - Tokenizers 0.22.2