Kundyzka commited on
Commit
7b90a99
·
verified ·
1 Parent(s): 8663a5e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +26 -13
README.md CHANGED
@@ -5,18 +5,18 @@ datasets:
5
  language:
6
  - kk
7
  metrics:
8
- - name: exact_match
 
 
 
9
  type: Exact Match
10
  value: 13.116
11
- - name: f1
12
  type: F1 Score
13
- value: 26.950
14
- - name: exact_match_test
15
- type: Exact Match (Test)
16
- value: 49.740
17
- - name: f1_test
18
- type: F1 Score (Test)
19
  value: 70.127
 
 
 
20
  base_model:
21
  - FacebookAI/xlm-roberta-large
22
  pipeline_tag: question-answering
@@ -26,7 +26,7 @@ tags:
26
 
27
  # Description
28
 
29
- This model was developed by **Kundyz Maksutova**, PhD Candidate, as part of research on question-answering systems in the Kazakh language. It is a fine-tuned version of `FacebookAI/xlm-roberta-large` on the `Kundyzka/informatics_kaz` dataset, tailored for the domain of computer science.
30
 
31
  ### Key Features:
32
  - **Base Model**: `FacebookAI/xlm-roberta-large`
@@ -34,17 +34,30 @@ This model was developed by **Kundyz Maksutova**, PhD Candidate, as part of rese
34
  - **Language**: Kazakh (`kk`)
35
  - **Task**: Question Answering
36
  - **Performance**:
37
- - Validation:
38
  - F1 Score: 26.950
39
  - Exact Match: 13.116
40
- - Test:
41
  - F1 Score: 70.127
42
  - Exact Match: 49.740
43
 
 
 
 
44
  ### Intended Use:
45
- This model is designed for use in question-answering systems, particularly in the domain of computer science and related fields. It can handle questions and provide accurate answers in the Kazakh language, making it an ideal tool for educational, research, and application development purposes.
 
 
 
 
 
 
 
 
46
 
47
  ### Tags:
48
  - `computerscience`
 
 
49
 
50
- For more details, fine-tuning instructions, or adaptation to other tasks, refer to the model repository.
 
5
  language:
6
  - kk
7
  metrics:
8
+ - name: F1 (Before Training)
9
+ type: F1 Score
10
+ value: 26.950
11
+ - name: Exact Match (Before Training)
12
  type: Exact Match
13
  value: 13.116
14
+ - name: F1 (After Training)
15
  type: F1 Score
 
 
 
 
 
 
16
  value: 70.127
17
+ - name: Exact Match (After Training)
18
+ type: Exact Match
19
+ value: 49.740
20
  base_model:
21
  - FacebookAI/xlm-roberta-large
22
  pipeline_tag: question-answering
 
26
 
27
  # Description
28
 
29
+ This model was developed by **Kundyz Maksutova**, PhD Candidate, as part of research on question-answering systems in the Kazakh language. It is a fine-tuned version of `FacebookAI/xlm-roberta-large` on the `Kundyzka/informatics_kaz` dataset, specifically optimized for handling questions in the domain of computer science.
30
 
31
  ### Key Features:
32
  - **Base Model**: `FacebookAI/xlm-roberta-large`
 
34
  - **Language**: Kazakh (`kk`)
35
  - **Task**: Question Answering
36
  - **Performance**:
37
+ - **Before Training**:
38
  - F1 Score: 26.950
39
  - Exact Match: 13.116
40
+ - **After Training**:
41
  - F1 Score: 70.127
42
  - Exact Match: 49.740
43
 
44
+ ### Dataset:
45
+ The `Kundyzka/informatics_kaz` dataset is designed to provide a diverse set of questions and answers in Kazakh, specifically covering topics in computer science. This dataset ensures that the model effectively handles domain-specific queries and terminology.
46
+
47
  ### Intended Use:
48
+ This model is intended for answering questions in the Kazakh language, with potential applications in:
49
+ - **Educational Platforms**: Assisting students with computer science-related questions.
50
+ - **Research Projects**: Supporting the study and development of Kazakh natural language processing tools.
51
+ - **AI Applications**: Enhancing chatbots or intelligent systems requiring domain-specific question-answering capabilities.
52
+
53
+ ### Limitations and Ethical Considerations:
54
+ - **Domain-Specific Bias**: The model performs best on computer science queries and may not generalize well to other domains.
55
+ - **Dataset Bias**: The dataset may introduce biases that affect model predictions.
56
+ - **Language Support**: The model is optimized for Kazakh and does not handle other languages.
57
 
58
  ### Tags:
59
  - `computerscience`
60
+ - `question-answering`
61
+ - `Kazakh`
62
 
63
+ This model represents a significant contribution to improving natural language processing tools for low-resource languages like Kazakh. For further details or customization, refer to the model repository.