Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,30 +1,20 @@
|
|
| 1 |
---
|
| 2 |
-
|
|
|
|
|
|
|
| 3 |
language:
|
| 4 |
- vi
|
| 5 |
tags:
|
| 6 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
- nlp
|
| 8 |
-
|
| 9 |
-
- low-resource
|
| 10 |
-
license: cc-by-4.0
|
| 11 |
-
task_categories:
|
| 12 |
-
- question-generation
|
| 13 |
-
task_ids:
|
| 14 |
-
- text2text-generation
|
| 15 |
size_categories:
|
| 16 |
- 10K<n<100K
|
| 17 |
-
dataset_info:
|
| 18 |
-
features:
|
| 19 |
-
- name: question
|
| 20 |
-
dtype: string
|
| 21 |
-
- name: context
|
| 22 |
-
dtype: string
|
| 23 |
-
- name: answer
|
| 24 |
-
dtype: string
|
| 25 |
-
splits:
|
| 26 |
-
- name: train
|
| 27 |
-
num_examples: 30000
|
| 28 |
---
|
| 29 |
|
| 30 |
# HVU_QA
|
|
@@ -34,7 +24,7 @@ It was created using a fully automated pipeline combining web crawling, semantic
|
|
| 34 |
|
| 35 |
---
|
| 36 |
|
| 37 |
-
##
|
| 38 |
|
| 39 |
- **Language:** Vietnamese
|
| 40 |
- **Format:** SQuAD-style JSON
|
|
@@ -61,7 +51,7 @@ All data are collected from public-domain Vietnamese government and service port
|
|
| 61 |
|
| 62 |
---
|
| 63 |
|
| 64 |
-
##
|
| 65 |
|
| 66 |
A fine-tuned `VietAI/vit5-base` model trained on HVU_QA achieved:
|
| 67 |
|
|
@@ -74,7 +64,7 @@ A fine-tuned `VietAI/vit5-base` model trained on HVU_QA achieved:
|
|
| 74 |
|
| 75 |
---
|
| 76 |
|
| 77 |
-
##
|
| 78 |
|
| 79 |
```json
|
| 80 |
{
|
|
@@ -90,7 +80,7 @@ A fine-tuned `VietAI/vit5-base` model trained on HVU_QA achieved:
|
|
| 90 |
|
| 91 |
---
|
| 92 |
|
| 93 |
-
##
|
| 94 |
|
| 95 |
### Load from Hub
|
| 96 |
|
|
@@ -110,7 +100,7 @@ pip install datasets transformers sentencepiece safetensors accelerate evaluate
|
|
| 110 |
|
| 111 |
---
|
| 112 |
|
| 113 |
-
##
|
| 114 |
|
| 115 |
### Fine-tune a Question Generation Model
|
| 116 |
|
|
@@ -148,7 +138,7 @@ Number of questions: 5
|
|
| 148 |
|
| 149 |
---
|
| 150 |
|
| 151 |
-
##
|
| 152 |
|
| 153 |
If you use **HVU_QA** in your research:
|
| 154 |
|
|
|
|
| 1 |
---
|
| 2 |
+
license: mit
|
| 3 |
+
task_categories:
|
| 4 |
+
- question-answering
|
| 5 |
language:
|
| 6 |
- vi
|
| 7 |
tags:
|
| 8 |
+
- ag
|
| 9 |
+
- t5
|
| 10 |
+
- vit5
|
| 11 |
+
- squad-format
|
| 12 |
+
- vietnamese
|
| 13 |
+
- education
|
| 14 |
- nlp
|
| 15 |
+
pretty_name: Vietnamese Question Generation
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
size_categories:
|
| 17 |
- 10K<n<100K
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
---
|
| 19 |
|
| 20 |
# HVU_QA
|
|
|
|
| 24 |
|
| 25 |
---
|
| 26 |
|
| 27 |
+
## Dataset Description
|
| 28 |
|
| 29 |
- **Language:** Vietnamese
|
| 30 |
- **Format:** SQuAD-style JSON
|
|
|
|
| 51 |
|
| 52 |
---
|
| 53 |
|
| 54 |
+
## Quality Evaluation
|
| 55 |
|
| 56 |
A fine-tuned `VietAI/vit5-base` model trained on HVU_QA achieved:
|
| 57 |
|
|
|
|
| 64 |
|
| 65 |
---
|
| 66 |
|
| 67 |
+
## Data Fields
|
| 68 |
|
| 69 |
```json
|
| 70 |
{
|
|
|
|
| 80 |
|
| 81 |
---
|
| 82 |
|
| 83 |
+
## How to Use
|
| 84 |
|
| 85 |
### Load from Hub
|
| 86 |
|
|
|
|
| 100 |
|
| 101 |
---
|
| 102 |
|
| 103 |
+
## Example Usage
|
| 104 |
|
| 105 |
### Fine-tune a Question Generation Model
|
| 106 |
|
|
|
|
| 138 |
|
| 139 |
---
|
| 140 |
|
| 141 |
+
## Citation
|
| 142 |
|
| 143 |
If you use **HVU_QA** in your research:
|
| 144 |
|