Update README.md
Browse files
README.md
CHANGED
@@ -89,20 +89,59 @@ It achieves the following results on the evaluation set:
|
|
89 |
**SENT_RO**
|
90 |
|
91 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
## Model description
|
93 |
|
94 |
-
|
|
|
|
|
|
|
|
|
95 |
|
96 |
## Intended uses & limitations
|
97 |
|
98 |
-
|
|
|
|
|
99 |
|
100 |
## Training and evaluation data
|
101 |
|
102 |
-
|
|
|
|
|
|
|
103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
104 |
## Training procedure
|
105 |
|
|
|
106 |
### Training hyperparameters
|
107 |
|
108 |
The following hyperparameters were used during training:
|
|
|
89 |
**SENT_RO**
|
90 |
|
91 |
|
92 |
+
| |precision | recall | f1-score | support |
|
93 |
+
|:-------------:|:-----:|:----:|:------:|:--------:|
|
94 |
+
| Negative (0) | 0.79 | 0.83 | 0.81 | 11,675 |
|
95 |
+
| Positive (1) | 0.88 | 0.85 | 0.87 | 17,271 |
|
96 |
+
| | | | | |
|
97 |
+
| Accuracy | | | 0.85 | 28,946 |
|
98 |
+
| Macro Avg | 0.84 | 0.84 | 0.84 | 28,946 |
|
99 |
+
| Weighted Avg | 0.85 | 0.85 | 0.85 | 28,946 |
|
100 |
+
|
101 |
+
**LaRoSeDa**
|
102 |
+
|
103 |
+
|
104 |
+
| |precision | recall | f1-score | support |
|
105 |
+
|:-------------:|:-----:|:----:|:------:|:--------:|
|
106 |
+
| Negative (0) | 0.79 | 0.94 | 0.86 | 7,500 |
|
107 |
+
| Positive (1) | 0.93 | 0.75 | 0.83 | 7,500 |
|
108 |
+
| | | | | |
|
109 |
+
| Accuracy | | | 0.85 | 15,000 |
|
110 |
+
| Macro Avg | 0.86 | 0.85 | 0.84 | 15,000 |
|
111 |
+
| Weighted Avg | 0.86 | 0.85 | 0.84 | 15,000 |
|
112 |
+
|
113 |
+
|
114 |
## Model description
|
115 |
|
116 |
+
Finetuned Romanian BERT model for sentiment classification.
|
117 |
+
|
118 |
+
Trained on a mix of product reviews from Decathlon retailer website and movie reviews from cinemagia.
|
119 |
+
|
120 |
+
|
121 |
|
122 |
## Intended uses & limitations
|
123 |
|
124 |
+
Sentiment classification for Romanian Language.
|
125 |
+
|
126 |
+
Biased towards Product reviews.
|
127 |
|
128 |
## Training and evaluation data
|
129 |
|
130 |
+
**Trained on:**
|
131 |
+
- Decathlon Dataset available on request
|
132 |
+
|
133 |
+
- Cinemagia Movie reviews public on kaggle [Link](https://www.kaggle.com/datasets/gringoandy/romanian-sentiment-movie-reviews)
|
134 |
|
135 |
+
**Evaluated on**
|
136 |
+
|
137 |
+
- Holdout data from training dataset
|
138 |
+
- RO_SENT Dataset
|
139 |
+
- LaROSeDa Dataset
|
140 |
+
|
141 |
+
|
142 |
## Training procedure
|
143 |
|
144 |
+
|
145 |
### Training hyperparameters
|
146 |
|
147 |
The following hyperparameters were used during training:
|