eve commited on
Update README.md
Browse files
README.md
CHANGED
|
@@ -15,14 +15,14 @@ The dataset consists of three major components: **Documents, Multimodal Question
|
|
| 15 |
### **1. Document Collection**
|
| 16 |
The dataset includes **six JSONL files**, each corresponding to a different data source:
|
| 17 |
|
| 18 |
-
| File Name | Description |
|
| 19 |
-
|--------------------|-------------|
|
| 20 |
-
| `doc_wit.jsonl` | Wit-MQA+ documents |
|
| 21 |
-
| `doc_wiki.jsonl` | Wiki-MQA+ documents |
|
| 22 |
-
| `doc_web.jsonl` | Web-MQA+ documents |
|
| 23 |
-
| `doc_arxiv.jsonl` | Arxiv-MQA+ documents |
|
| 24 |
-
| `doc_recipe.jsonl`| Recipe-MQA+ documents |
|
| 25 |
-
| `doc_manual.jsonl`| Manual-MQA+ documents |
|
| 26 |
|
| 27 |
Each line in these files represents a **single document** with the following fields:
|
| 28 |
|
|
@@ -46,14 +46,14 @@ Each line in these files represents a **single document** with the following fie
|
|
| 46 |
### **2. Multimodal Question Answering (MQA)**
|
| 47 |
The **MQA component** consists of **six JSONL files**, each corresponding to a different dataset:
|
| 48 |
|
| 49 |
-
| File Name
|
| 50 |
-
|----------------------|-------------|
|
| 51 |
-
| `wit_mqa.jsonl` | Wit-MQA+ multimodal QA pairs |
|
| 52 |
-
| `wiki_mqa.jsonl` | Wiki-MQA+ multimodal QA pairs |
|
| 53 |
-
| `web_mqa.jsonl` | Web-MQA+ multimodal QA pairs |
|
| 54 |
-
| `arxiv_mqa.jsonl` | Arxiv-MQA+ QA pairs |
|
| 55 |
-
| `recipe_mqa.jsonl` | Recipe-MQA+ QA pairs |
|
| 56 |
-
| `manual_mqa.jsonl` | Manual-MQA+ QA pairs |
|
| 57 |
|
| 58 |
Each entry contains **a question, provenance documents, a ground truth answer, and image references**.
|
| 59 |
|
|
|
|
| 15 |
### **1. Document Collection**
|
| 16 |
The dataset includes **six JSONL files**, each corresponding to a different data source:
|
| 17 |
|
| 18 |
+
| File Name | Description | Num |
|
| 19 |
+
|--------------------|-------------|-------------|
|
| 20 |
+
| `doc_wit.jsonl` | Wit-MQA+ documents | 639 |
|
| 21 |
+
| `doc_wiki.jsonl` | Wiki-MQA+ documents | 538 |
|
| 22 |
+
| `doc_web.jsonl` | Web-MQA+ documents | 1500 |
|
| 23 |
+
| `doc_arxiv.jsonl` | Arxiv-MQA+ documents | 101 |
|
| 24 |
+
| `doc_recipe.jsonl`| Recipe-MQA+ documents | 1528 |
|
| 25 |
+
| `doc_manual.jsonl`| Manual-MQA+ documents | 40 |
|
| 26 |
|
| 27 |
Each line in these files represents a **single document** with the following fields:
|
| 28 |
|
|
|
|
| 46 |
### **2. Multimodal Question Answering (MQA)**
|
| 47 |
The **MQA component** consists of **six JSONL files**, each corresponding to a different dataset:
|
| 48 |
|
| 49 |
+
| File Name | Description | Num |
|
| 50 |
+
|--------------------|-------------|-------------|
|
| 51 |
+
| `wit_mqa.jsonl` | Wit-MQA+ multimodal QA pairs | 600 |
|
| 52 |
+
| `wiki_mqa.jsonl` | Wiki-MQA+ multimodal QA pairs | 500 |
|
| 53 |
+
| `web_mqa.jsonl` | Web-MQA+ multimodal QA pairs | 750 |
|
| 54 |
+
| `arxiv_mqa.jsonl` | Arxiv-MQA+ QA pairs | 200 |
|
| 55 |
+
| `recipe_mqa.jsonl` | Recipe-MQA+ QA pairs | 2360 |
|
| 56 |
+
| `manual_mqa.jsonl` | Manual-MQA+ QA pairs | 390 |
|
| 57 |
|
| 58 |
Each entry contains **a question, provenance documents, a ground truth answer, and image references**.
|
| 59 |
|