Update README.md
Browse files
README.md
CHANGED
@@ -3,9 +3,17 @@ license: mit
|
|
3 |
base_model: xlnet-base-cased
|
4 |
tags:
|
5 |
- generated_from_keras_callback
|
|
|
|
|
|
|
6 |
model-index:
|
7 |
- name: dipawidia/xlnet-base-cased-product-review-sentiment-analysis
|
8 |
results: []
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
|
11 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
@@ -13,7 +21,7 @@ probably proofread and complete it, then remove this comment. -->
|
|
13 |
|
14 |
# dipawidia/xlnet-base-cased-product-review-sentiment-analysis
|
15 |
|
16 |
-
This model is a fine-tuned version of [xlnet-base-cased](https://huggingface.co/xlnet-base-cased) on
|
17 |
It achieves the following results on the evaluation set:
|
18 |
- Train Loss: 0.1085
|
19 |
- Train Accuracy: 0.9617
|
@@ -21,13 +29,48 @@ It achieves the following results on the evaluation set:
|
|
21 |
- Validation Accuracy: 0.9414
|
22 |
- Epoch: 4
|
23 |
|
24 |
-
##
|
25 |
|
26 |
-
|
27 |
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
|
30 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
## Training and evaluation data
|
33 |
|
@@ -56,4 +99,4 @@ The following hyperparameters were used during training:
|
|
56 |
|
57 |
- Transformers 4.41.2
|
58 |
- TensorFlow 2.15.0
|
59 |
-
- Tokenizers 0.19.1
|
|
|
3 |
base_model: xlnet-base-cased
|
4 |
tags:
|
5 |
- generated_from_keras_callback
|
6 |
+
- sentiment analysis
|
7 |
+
widget:
|
8 |
+
- text: product quality is good. affordable prices and very fast delivery.
|
9 |
model-index:
|
10 |
- name: dipawidia/xlnet-base-cased-product-review-sentiment-analysis
|
11 |
results: []
|
12 |
+
language:
|
13 |
+
- en
|
14 |
+
metrics:
|
15 |
+
- accuracy
|
16 |
+
library_name: transformers
|
17 |
---
|
18 |
|
19 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
|
|
21 |
|
22 |
# dipawidia/xlnet-base-cased-product-review-sentiment-analysis
|
23 |
|
24 |
+
This model is a fine-tuned version of [xlnet-base-cased](https://huggingface.co/xlnet-base-cased) on any type of product reviews dataset gathered from several e-commerce such as shopee, tokopedia, blibli, lazada, and zalora.
|
25 |
It achieves the following results on the evaluation set:
|
26 |
- Train Loss: 0.1085
|
27 |
- Train Accuracy: 0.9617
|
|
|
29 |
- Validation Accuracy: 0.9414
|
30 |
- Epoch: 4
|
31 |
|
32 |
+
## Intended uses & limitations
|
33 |
|
34 |
+
This fine-tuned XLNet model is used for sentiment analysis with 2 labels text classification: 0 -> Negative; 1 -> Positive.
|
35 |
|
36 |
+
### Example Pipeline
|
37 |
+
```python
|
38 |
+
from transformers import pipeline
|
39 |
+
pipe = pipeline("text-classification", model="dipawidia/xlnet-base-cased-product-review-sentiment-analysis")
|
40 |
+
pipe("This shoes is awesome")
|
41 |
+
```
|
42 |
+
```
|
43 |
+
[{'label': 'Positive', 'score': 0.9995703101158142}]
|
44 |
+
```
|
45 |
|
46 |
+
### Full classification example
|
47 |
+
|
48 |
+
```python
|
49 |
+
from transformers import XLNetTokenizer, TFXLNetForSequenceClassification
|
50 |
+
import tensorflow as tf
|
51 |
+
import numpy as np
|
52 |
+
|
53 |
+
tokenizer = XLNetTokenizer.from_pretrained("dipawidia/xlnet-base-cased-product-review-sentiment-analysis")
|
54 |
+
model = TFXLNetForSequenceClassification.from_pretrained("dipawidia/xlnet-base-cased-product-review-sentiment-analysis")
|
55 |
+
|
56 |
+
def get_sentimen(text):
|
57 |
+
tokenize_text = tokenizer(text, return_tensors = 'tf')
|
58 |
+
preds = model.predict(dict(tokenize_text))['logits']
|
59 |
+
class_preds = np.argmax(tf.keras.layers.Softmax()(preds))
|
60 |
+
if class_preds == 1:
|
61 |
+
label = 'Positive'
|
62 |
+
else:
|
63 |
+
label = 'Negative'
|
64 |
+
return(label)
|
65 |
+
|
66 |
+
get_sentimen('i hate this product')
|
67 |
+
```
|
68 |
+
|
69 |
+
Output:
|
70 |
+
|
71 |
+
```
|
72 |
+
Negative
|
73 |
+
```
|
74 |
|
75 |
## Training and evaluation data
|
76 |
|
|
|
99 |
|
100 |
- Transformers 4.41.2
|
101 |
- TensorFlow 2.15.0
|
102 |
+
- Tokenizers 0.19.1
|