Commit
•
7713f6a
1
Parent(s):
bec0ffb
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,41 @@
|
|
1 |
---
|
2 |
license: other
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: other
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
model-index:
|
6 |
+
- name: finetuned-distilbert-adult-content-detection
|
7 |
+
results: []
|
8 |
---
|
9 |
+
|
10 |
+
### finetuned-distilbert-news-article-catgorization
|
11 |
+
|
12 |
+
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the adult_content dataset.
|
13 |
+
It achieves the following results on the evaluation set:
|
14 |
+
- Loss: 0.0065
|
15 |
+
- F1_score(weighted): 0.99
|
16 |
+
### Model description
|
17 |
+
More information needed
|
18 |
+
### Intended uses & limitations
|
19 |
+
More information needed
|
20 |
+
### Training and evaluation data
|
21 |
+
The model was trained on some subset of the adult_content dataset and it was validated on the remaining subset of the data
|
22 |
+
### Training procedure
|
23 |
+
More information needed
|
24 |
+
### Training hyperparameters
|
25 |
+
The following hyperparameters were used during training:
|
26 |
+
- learning_rate: 5e-5
|
27 |
+
- train_batch_size: 5
|
28 |
+
- eval_batch_size: 5
|
29 |
+
- seed: 17
|
30 |
+
- optimizer: AdamW(lr=1e-5 and epsilon=1e-08)
|
31 |
+
- lr_scheduler_type: linear
|
32 |
+
- lr_scheduler_warmup_steps: 0
|
33 |
+
- num_epochs: 5
|
34 |
+
### Training results
|
35 |
+
| Training Loss | Epoch | Validation Loss | f1 score |
|
36 |
+
|:-------------:|:-----:|:---------------: |:------:|
|
37 |
+
| 0.0065 | 1.0 | 0.0669 | 0.9880 |
|
38 |
+
| 0.0429 | 2.0 | 0.0530 | 0.9907 |
|
39 |
+
| 0.0177 | 3.0 | 0.0305 | 0.9933 |
|
40 |
+
| 0.0123 | 4.0 | 0.0362 | 0.9933 |
|
41 |
+
| 0.0065 | 5.0 | 0.0176 | 0.9960 |
|