update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
metrics:
|
6 |
+
- precision
|
7 |
+
- recall
|
8 |
+
- accuracy
|
9 |
+
- f1
|
10 |
+
model-index:
|
11 |
+
- name: berturk-128k-keyword-discriminator
|
12 |
+
results: []
|
13 |
+
---
|
14 |
+
|
15 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
16 |
+
should probably proofread and complete it, then remove this comment. -->
|
17 |
+
|
18 |
+
# berturk-128k-keyword-discriminator
|
19 |
+
|
20 |
+
This model is a fine-tuned version of [dbmdz/bert-base-turkish-128k-cased](https://huggingface.co/dbmdz/bert-base-turkish-128k-cased) on an unknown dataset.
|
21 |
+
It achieves the following results on the evaluation set:
|
22 |
+
- Loss: 0.3828
|
23 |
+
- Precision: 0.6791
|
24 |
+
- Recall: 0.7234
|
25 |
+
- Accuracy: 0.9294
|
26 |
+
- F1: 0.7006
|
27 |
+
- Ent/precision: 0.6931
|
28 |
+
- Ent/accuracy: 0.7715
|
29 |
+
- Ent/f1: 0.7302
|
30 |
+
- Con/precision: 0.6473
|
31 |
+
- Con/accuracy: 0.6282
|
32 |
+
- Con/f1: 0.6376
|
33 |
+
|
34 |
+
## Model description
|
35 |
+
|
36 |
+
More information needed
|
37 |
+
|
38 |
+
## Intended uses & limitations
|
39 |
+
|
40 |
+
More information needed
|
41 |
+
|
42 |
+
## Training and evaluation data
|
43 |
+
|
44 |
+
More information needed
|
45 |
+
|
46 |
+
## Training procedure
|
47 |
+
|
48 |
+
### Training hyperparameters
|
49 |
+
|
50 |
+
The following hyperparameters were used during training:
|
51 |
+
- learning_rate: 2e-05
|
52 |
+
- train_batch_size: 16
|
53 |
+
- eval_batch_size: 16
|
54 |
+
- seed: 42
|
55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
56 |
+
- lr_scheduler_type: linear
|
57 |
+
- num_epochs: 8
|
58 |
+
- mixed_precision_training: Native AMP
|
59 |
+
|
60 |
+
### Training results
|
61 |
+
|
62 |
+
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | Accuracy | F1 | Ent/precision | Ent/accuracy | Ent/f1 | Con/precision | Con/accuracy | Con/f1 |
|
63 |
+
|:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:--------:|:------:|:-------------:|:------------:|:------:|:-------------:|:------------:|:------:|
|
64 |
+
| 0.1632 | 1.0 | 1875 | 0.1637 | 0.6661 | 0.6900 | 0.9320 | 0.6778 | 0.6649 | 0.7401 | 0.7005 | 0.6692 | 0.5907 | 0.6275 |
|
65 |
+
| 0.1151 | 2.0 | 3750 | 0.1709 | 0.6538 | 0.7446 | 0.9292 | 0.6963 | 0.6682 | 0.7864 | 0.7225 | 0.6223 | 0.6619 | 0.6415 |
|
66 |
+
| 0.0817 | 3.0 | 5625 | 0.1931 | 0.6667 | 0.7292 | 0.9294 | 0.6965 | 0.6843 | 0.7677 | 0.7236 | 0.6290 | 0.6529 | 0.6407 |
|
67 |
+
| 0.057 | 4.0 | 7500 | 0.2375 | 0.6578 | 0.7486 | 0.9277 | 0.7002 | 0.6708 | 0.7950 | 0.7277 | 0.6284 | 0.6567 | 0.6422 |
|
68 |
+
| 0.041 | 5.0 | 9375 | 0.2765 | 0.6683 | 0.7390 | 0.9284 | 0.7019 | 0.6834 | 0.7821 | 0.7294 | 0.6351 | 0.6538 | 0.6444 |
|
69 |
+
| 0.0297 | 6.0 | 11250 | 0.3128 | 0.6811 | 0.7249 | 0.9295 | 0.7023 | 0.6979 | 0.7710 | 0.7327 | 0.6438 | 0.6334 | 0.6386 |
|
70 |
+
| 0.0211 | 7.0 | 13125 | 0.3633 | 0.6780 | 0.7236 | 0.9290 | 0.7001 | 0.6919 | 0.7722 | 0.7299 | 0.6463 | 0.6273 | 0.6366 |
|
71 |
+
| 0.0165 | 8.0 | 15000 | 0.3828 | 0.6791 | 0.7234 | 0.9294 | 0.7006 | 0.6931 | 0.7715 | 0.7302 | 0.6473 | 0.6282 | 0.6376 |
|
72 |
+
|
73 |
+
|
74 |
+
### Framework versions
|
75 |
+
|
76 |
+
- Transformers 4.19.2
|
77 |
+
- Pytorch 1.11.0+cu113
|
78 |
+
- Datasets 2.2.2
|
79 |
+
- Tokenizers 0.12.1
|