Update README.md
Browse files
README.md
CHANGED
@@ -23,7 +23,6 @@ This repository contains an Azerbaijani spell correction model based on `google/
|
|
23 |
- [Imitating Real Spelling Errors](#imitating-real-spelling-errors)
|
24 |
- [Keyboard Layout Considerations](#keyboard-layout-considerations)
|
25 |
- [Typical Substitutions](#typical-substitutions)
|
26 |
-
- [Model Training](#model-training)
|
27 |
- [Usage](#usage)
|
28 |
- [License](#license)
|
29 |
- [Contact](#сontact)
|
@@ -86,13 +85,6 @@ Several typical substitutions were included in the dataset to reflect common err
|
|
86 |
|
87 |
These substitutions represent phonetic similarities and common mistakes made by speakers when writing Azerbaijani text, especially when using Latin characters.
|
88 |
|
89 |
-
## Model Training
|
90 |
-
|
91 |
-
The model was fine-tuned using the Hugging Face Transformers library. Key points in the training process include:
|
92 |
-
|
93 |
-
- **Model Architecture:** `google/mt5-small` was chosen for its multilingual support and capability to handle Azerbaijani characters.
|
94 |
-
- **Training Objective:** The model was trained to map incorrect sentences to their correct versions.
|
95 |
-
- **Evaluation:** The performance was evaluated using metrics like BLEU and ROUGE to ensure the quality of corrections.
|
96 |
|
97 |
## Usage
|
98 |
|
|
|
23 |
- [Imitating Real Spelling Errors](#imitating-real-spelling-errors)
|
24 |
- [Keyboard Layout Considerations](#keyboard-layout-considerations)
|
25 |
- [Typical Substitutions](#typical-substitutions)
|
|
|
26 |
- [Usage](#usage)
|
27 |
- [License](#license)
|
28 |
- [Contact](#сontact)
|
|
|
85 |
|
86 |
These substitutions represent phonetic similarities and common mistakes made by speakers when writing Azerbaijani text, especially when using Latin characters.
|
87 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
## Usage
|
90 |
|