Update README.md
Browse files
README.md
CHANGED
@@ -18,42 +18,34 @@ should probably proofread and complete it, then remove this comment. -->
|
|
18 |
|
19 |
# sentiment-polish-gpt2-small
|
20 |
|
21 |
-
This model was trained from polish-gpt2-small on the polemo2-official dataset.
|
22 |
It achieves the following results on the evaluation set:
|
23 |
- Loss: 0.4659
|
24 |
- Accuracy: 0.9627
|
25 |
|
26 |
## Model description
|
27 |
|
28 |
-
|
29 |
|
30 |
## Intended uses & limitations
|
31 |
|
32 |
More information needed
|
33 |
|
34 |
## Training and evaluation data
|
35 |
-
https://huggingface.co/datasets/clarin-pl/polemo2-official
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
month = nov,
|
44 |
-
year = "2019",
|
45 |
-
address = "Hong Kong, China",
|
46 |
-
publisher = "Association for Computational Linguistics",
|
47 |
-
url = "https://aclanthology.org/K19-1092",
|
48 |
-
doi = "10.18653/v1/K19-1092",
|
49 |
-
pages = "980--991",
|
50 |
-
abstract = "In this article we present an extended version of PolEmo {--} a corpus of consumer reviews from 4 domains: medicine, hotels, products and school. Current version (PolEmo 2.0) contains 8,216 reviews having 57,466 sentences. Each text and sentence was manually annotated with sentiment in 2+1 scheme, which gives a total of 197,046 annotations. We obtained a high value of Positive Specific Agreement, which is 0.91 for texts and 0.88 for sentences. PolEmo 2.0 is publicly available under a Creative Commons copyright license. We explored recent deep learning approaches for the recognition of sentiment, such as Bi-directional Long Short-Term Memory (BiLSTM) and Bidirectional Encoder Representations from Transformers (BERT).",
|
51 |
-
}
|
52 |
```
|
53 |
|
54 |
## Training procedure
|
55 |
|
56 |
GPU: RTX 3090
|
|
|
57 |
Training time: 2:53:05
|
58 |
|
59 |
### Training hyperparameters
|
|
|
18 |
|
19 |
# sentiment-polish-gpt2-small
|
20 |
|
21 |
+
This model was trained from [polish-gpt2-small](https://huggingface.co/sdadas/polish-gpt2-small) on the [polemo2-official](https://huggingface.co/datasets/clarin-pl/polemo2-official) dataset.
|
22 |
It achieves the following results on the evaluation set:
|
23 |
- Loss: 0.4659
|
24 |
- Accuracy: 0.9627
|
25 |
|
26 |
## Model description
|
27 |
|
28 |
+
Trained from [polish-gpt2-small](https://huggingface.co/sdadas/polish-gpt2-small)
|
29 |
|
30 |
## Intended uses & limitations
|
31 |
|
32 |
More information needed
|
33 |
|
34 |
## Training and evaluation data
|
35 |
+
Merged all rows from [polemo2-official](https://huggingface.co/datasets/clarin-pl/polemo2-official) dataset.
|
36 |
+
|
37 |
+
Train/test split: 80%/20%
|
38 |
+
|
39 |
+
Datacollator:
|
40 |
+
```py
|
41 |
+
from transformers import DataCollatorWithPadding
|
42 |
+
data_collator = DataCollatorWithPadding(tokenizer=tokenizer, padding="longest", max_length=128, pad_to_multiple_of=8)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
```
|
44 |
|
45 |
## Training procedure
|
46 |
|
47 |
GPU: RTX 3090
|
48 |
+
|
49 |
Training time: 2:53:05
|
50 |
|
51 |
### Training hyperparameters
|