data_summary_card.md
4.3 KB · 149 lines · markdown Raw
1
2
3
4
5 # Data Summary for Phi-3-vision-128k-instruct, Phi-3.5-vision-instruct
6
7
8
9
10
11 ## 1. General information
12
13 **1.0.1 Version of the Summary:** 1.0
14
15
16
17 **1.0.2 Last update:** 10-Dec-2025
18
19
20
21 ## 1.1 Model Developer Identification
22
23 **1.1.1 Model Developer name and contact details:** Microsoft Corporation at One Microsoft Way, Redmond, WA 98052. Tel: 425-882-8080.
24
25
26
27 ## 1.2 Model Identification
28
29 **1.2.1 Versioned model name(s):** Phi-3-Vision-128K-Instruct, Phi-3.5-vision-instruct
30
31
32
33 **1.2.2 Model release date:** 21-May-2024
34
35
36
37 ## 1.3 Overall training data size and characteristics
38
39 ### 1.3.1 Size of dataset and characteristics
40
41 **1.3.1.A Text training data size:** 1 billion to 10 trillion tokens
42
43
44
45 **1.3.1.B Text training data content:** Our training data includes a wide variety of sources, and is a combination of publicly available documents selected for quality, selected educational data and code; selected image-text interleave; newly created synthetic, “textbook-like” data for the purpose of teaching math, coding, common sense reasoning, general knowledge of the world (science, daily activities, theory of mind, etc.); chat format supervised data covering various topics to reflect preferences on different aspects such as instruct-following, truthfulness, honesty and helpfulness.
46
47
48
49 **1.3.1.C Image training data size:** 1 million to 1 billion images
50
51
52
53 **1.3.1.D Image training data content:** Selected image-text interleaved data and newly created image data including charts, tables, diagrams, and slides, filtered from publicly available sources for quality and safety
54
55
56
57 **1.3.1.E Audio training data size:** Not applicable
58
59
60
61 **1.3.1.F Audio training data content:** Not applicable
62
63
64
65 **1.3.1.G Video training data size:** Not applicable
66
67
68
69 **1.3.1.H Video training data content:** Not applicable
70
71
72
73 **1.3.1.I Other training data size:** Not applicable
74
75
76
77 **1.3.1.J Other training data content:** Not applicable
78
79
80
81 **1.3.2 Latest date of data acquisition/collection for model training:** 15-Mar-2024
82
83
84
85 **1.3.3 Is data collection ongoing to update the model with new data collection after deployment?** No
86
87
88
89 **1.3.4 Date the training dataset was first used to train the model:** 01-Feb-2024
90
91
92
93 **1.3.5 Rationale or purpose of data selection:** Datasets were selected to maximize reasoning-dense coverage across text and vision for general-purpose multimodal understanding, including math, coding, common sense reasoning, and chart/table/diagram interpretation, supporting efficient deployment in constrained environments
94
95
96
97 ## 2. List of data sources
98
99 ### 2.1 Publicly available datasets
100
101 **2.1.1 Have you used publicly available datasets to train the model?** Yes
102
103
104
105 ## 2.2 Private non-publicly available datasets obtained from third parties
106
107 ### 2.2.1 Datasets commercially licensed by rights holders or their representatives
108
109 **2.2.1.A Have you concluded transactional commercial licensing agreement(s) with rights holder(s) or with their representatives?** Not applicable
110
111
112
113 ### 2.2.2 Private datasets obtained from other third-parties
114
115 **2.2.2.A Have you obtained private datasets from third parties that are not licensed as described in Section 2.2.1, such as data obtained from providers of private databases, or data intermediaries?** No
116
117
118
119 ## 2.3 Personal Information
120
121 **2.3.1 Was personal data used to train the model?** Microsoft follows all relevant laws and regulations pertaining to personal information.
122
123
124
125 ## 2.4 Synthetic data
126
127 **2.4.1 Was any synthetic AI-generated data used to train the model?** Yes
128
129
130
131 ## 3. Data processing aspects
132
133 ### 3.1 Respect of reservation of rights from text and data mining exception or limitation
134
135 **3.1.1 Does this dataset include any data protected by copyright, trademark, or patent?** Microsoft follows all required regulations and laws for processing data protected by copyright, trademark, or patent.
136
137
138
139 ## 3.2 Other information
140
141 **3.2.1 Does the dataset include information about consumer groups without revealing individual consumer identities?** Microsoft follows all required regulations and laws for protecting consumer identities.
142
143
144
145 **3.2.2 Was the dataset cleaned or modified before model training?** Yes
146
147
148
149