Jotschi commited on
Commit
8a245a3
1 Parent(s): 3c6e321

Initial commit

Browse files
Files changed (1) hide show
  1. README.md +73 -1
README.md CHANGED
@@ -1,3 +1,75 @@
1
  ---
2
- license: cc-by-4.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language:
3
+ - en
4
+ license_name: cc-by-4.0
5
+ license_link: https://creativecommons.org/licenses/by/4.0/legalcode
6
+ tags:
7
+ - coco
8
+ - mscoco
9
+ - simple-english
10
+ annotations_creators:
11
+ - machine-generated
12
+ pretty_name: MS COCO Karpathy in Simple English
13
+ size_categories:
14
+ - n<650k
15
+ source_datasets:
16
+ - mscoco
17
+ task_categories:
18
+ - text-generation
19
+ - image-to-text
20
+ - text-to-image
21
  ---
22
+
23
+ # Dataset Card for MS COCO Karpathy in Simple English
24
+
25
+ This dataset contains captions that were rephrased into simple english so that a young child would understand it.
26
+
27
+ ## Dataset Details
28
+
29
+ ### Dataset Description
30
+
31
+ - **Curated by:** {{ curators | default("[More Information Needed]", true)}}
32
+ - **Language(s) (NLP):** {{ language | default("[More Information Needed]", true)}}
33
+ - **License:** {{ license | default("[More Information Needed]", true)}}
34
+
35
+ ### Dataset Sources
36
+
37
+ The processed [MS COCO datasets](https://cocodataset.org/#download) (Karpathy Split) in this repo are based on the following sources:
38
+
39
+ * train: http://images.cocodataset.org/zips/train2014.zip
40
+ * val: http://images.cocodataset.org/zips/val2014.zip
41
+ * test: http://images.cocodataset.org/zips/test2014.zip
42
+
43
+ Files uses:
44
+
45
+ * aa31ac474cf6250ebb81d18348a07ed8 coco_karpathy_train.json
46
+ * 3ff34b0ef2db02d01c37399f6a2a6cd1 coco_karpathy_test.json
47
+ * b273847456ef5580e33713b1f7de52a0 coco_karpathy_val.json
48
+
49
+ MS COCO:
50
+
51
+ - **Download:** https://cocodataset.org/#download
52
+ - **Paper:** http://arxiv.org/abs/1405.0312
53
+
54
+
55
+ ## Dataset Creation
56
+
57
+ This dataset was generated by processing the annotations via [Mistal7B](https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-AWQ).
58
+
59
+
60
+ Prompt used:
61
+
62
+ ```
63
+ Rewrite the sentence " + caption + " for a 3 to 4 year old child. Give only one simple sentence. Don't use the word see. Give only a single answer.
64
+ ```
65
+
66
+ A filter was applied to only store captions which matched the common output format. A best effort filter was applied to reduce the chance of including multiple example sentences in the output.
67
+
68
+ ### Curation Rationale
69
+
70
+ This dataset is useful for experiments with small LLMs which have only a reduced corpus. The dataset is suitable to be used for LAVIS experiments (QFormer Training) with a finetuned TinyStories 33M LLM.
71
+
72
+ ## Dataset Card Authors
73
+
74
+ {{ dataset_card_authors | default("[More Information Needed]", true)}}
75
+