Commit
•
bf0c709
1
Parent(s):
4bb7b07
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: distilroberta-topic-classification
|
| 7 |
+
results: []
|
| 8 |
+
datasets:
|
| 9 |
+
- valurank/Topic_Classification
|
| 10 |
+
language:
|
| 11 |
+
- en
|
| 12 |
+
metrics:
|
| 13 |
+
- f1
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 17 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 18 |
+
|
| 19 |
+
# distilroberta-topic-classification
|
| 20 |
+
|
| 21 |
+
This model is a fine-tuned version of [distilroberta-topic-base](https://huggingface.co/distilroberta-base) on a dataset of headlines.
|
| 22 |
+
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 2.235735
|
| 24 |
+
- F1: 0.756
|
| 25 |
+
|
| 26 |
+
## Training and evaluation data
|
| 27 |
+
|
| 28 |
+
The following data sources were used:
|
| 29 |
+
* 22k News articles classified into 120 different topics from [Hugging face](https://huggingface.co/datasets/valurank/Topic_Classification)
|
| 30 |
+
|
| 31 |
+
## Training procedure
|
| 32 |
+
|
| 33 |
+
### Training hyperparameters
|
| 34 |
+
|
| 35 |
+
The following hyperparameters were used during training:
|
| 36 |
+
- learning_rate: 5e-05
|
| 37 |
+
- train_batch_size: 32
|
| 38 |
+
- eval_batch_size: 32
|
| 39 |
+
- seed: 12345
|
| 40 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
+
- lr_scheduler_type: linear
|
| 42 |
+
- lr_scheduler_warmup_steps: 16
|
| 43 |
+
- num_epochs: 10
|
| 44 |
+
- mixed_precision_training: Native AMP
|
| 45 |
+
|
| 46 |
+
### Training results
|
| 47 |
+
|
| 48 |
+
| Training Loss | Epoch | Step | Validation Loss | F1 |
|
| 49 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 50 |
+
| 2.3851 | 1.0 | 561 | 2.3445 | 0.6495 |
|
| 51 |
+
| 2.1441 | 2.0 | 1122 | 2.1980 | 0.7019 |
|
| 52 |
+
| 1.9992 | 3.0 | 1683 | 2.1720 | 0.7189 |
|
| 53 |
+
| 1.8384 | 4.0 | 2244 | 2.1425 | 0.7403 |
|
| 54 |
+
| 1.7468 | 5.0 | 2805 | 2.1666 | 0.7453 |
|
| 55 |
+
| 1.6360 | 6.0 | 3366 | 2.1779 | 0.7456 |
|
| 56 |
+
| 1.5935 | 7.0 | 3927 | 2.2003 | 0.7555 |
|
| 57 |
+
| 1.5460 | 8.0 | 4488 | 2.2157 | 0.7575 |
|
| 58 |
+
| 1.5510 | 9.0 | 5049 | 2.2300 | 0.7536 |
|
| 59 |
+
| 1.5097 | 10.0 | 5610 | 2.2357 | 0.7547 |
|
| 60 |
+
|
| 61 |
+
### Framework versions
|
| 62 |
+
|
| 63 |
+
- Transformers 4.35.2
|
| 64 |
+
- Pytorch 2.1.0
|
| 65 |
+
- Datasets 2.15.0
|
| 66 |
+
- Tokenizers 0.15.0
|