Datasets:
Languages:
English
Size:
10K - 100K
Tags:
sarcasm
sarcasm-detection
mulitmodal-sarcasm-detection
sarcasm detection
multimodao sarcasm detection
tweets
DOI:
License:
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,54 @@
|
|
1 |
---
|
2 |
license: unknown
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: unknown
|
3 |
+
task_categories:
|
4 |
+
- feature-extraction
|
5 |
+
- text-classification
|
6 |
+
- image-classification
|
7 |
+
- image-feature-extraction
|
8 |
+
- zero-shot-classification
|
9 |
+
- zero-shot-image-classification
|
10 |
+
language:
|
11 |
+
- en
|
12 |
+
tags:
|
13 |
+
- sarcasm
|
14 |
+
- sarcasm-detection
|
15 |
+
- mulitmodal-sarcasm-detection
|
16 |
+
- sarcasm detection
|
17 |
+
- multimodao sarcasm detection
|
18 |
+
- tweets
|
19 |
+
pretty_name: mmsd_v2
|
20 |
+
size_categories:
|
21 |
+
- 10K<n<100K
|
22 |
---
|
23 |
+
|
24 |
+
# MMSD2.0: Towards a Reliable Multi-modal Sarcasm Detection System
|
25 |
+
|
26 |
+
This is a copy of the dataset uploaded on Hugging Face for easy access. The original data comes from this [work](https://aclanthology.org/2023.findings-acl.689/).
|
27 |
+
|
28 |
+
## Usage
|
29 |
+
|
30 |
+
```python
|
31 |
+
# usage
|
32 |
+
from datasets import load_dataset
|
33 |
+
from transformers import CLIPImageProcessor, CLIPTokenizer
|
34 |
+
from torch.utils.data import DataLoader
|
35 |
+
|
36 |
+
image_processor = CLIPImageProcessor.from_pretrained(clip_path)
|
37 |
+
tokenizer = CLIPTokenizer.from_pretrained(clip_path)
|
38 |
+
|
39 |
+
def tokenization(example):
|
40 |
+
text_inputs = tokenizer(example["text"], truncation=True, padding=True, return_tensors="pt")
|
41 |
+
image_inputs = image_processor(example["image"], return_tensors="pt")
|
42 |
+
return {'pixel_values': image_inputs['pixel_values'],
|
43 |
+
'input_ids': text_inputs['input_ids'],
|
44 |
+
'attention_mask': text_inputs['attention_mask'],
|
45 |
+
"label": example["label"]}
|
46 |
+
|
47 |
+
dataset = load_dataset('quaeast/multimodal_sarcasm_detection')
|
48 |
+
dataset.set_transform(tokenization)
|
49 |
+
|
50 |
+
# get torch dataloader
|
51 |
+
train_dl = DataLoader(dataset['train'], batch_size=256, shuffle=True)
|
52 |
+
test_dl = DataLoader(dataset['test'], batch_size=256, shuffle=True)
|
53 |
+
val_dl = DataLoader(dataset['validation'], batch_size=256, shuffle=True)
|
54 |
+
```
|