Spaces:
Running
Running
File size: 12,823 Bytes
ac6c40f d1a58c9 8a2ec29 ac6c40f 57616af ac6c40f 57616af ac6c40f 57616af ac6c40f 57616af b7fbd2a b03f385 ac6c40f 57616af ac6c40f ce0bf11 ac6c40f 8a2ec29 ac6c40f 57616af ac6c40f 57616af ac6c40f d1a58c9 ac6c40f ce0bf11 ac6c40f 57616af ac6c40f ce0bf11 05d58bc ce0bf11 b03f385 ac6c40f 57616af 8a2ec29 57616af ce0bf11 57616af b03f385 ac6c40f ce0bf11 e0a15f4 ce0bf11 05d58bc ac6c40f 57616af 77956bd 8a2ec29 ac6c40f 13fd677 ac6c40f 57616af ac6c40f 57616af 9994065 57616af ac6c40f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 |
import json
import streamlit as st
from os.path import join as pjoin
from .streamlit_utils import (
make_multiselect,
make_selectbox,
make_text_area,
make_text_input,
make_radio,
)
N_FIELDS_WHERE = 9
N_FIELDS_LANGUAGES = 8
N_FIELDS_CREDIT = 5
N_FIELDS_STRUCTURE = 7
N_FIELDS = N_FIELDS_WHERE + N_FIELDS_LANGUAGES + N_FIELDS_CREDIT + N_FIELDS_STRUCTURE
languages_bcp47 = [
x
for x in json.load(open(pjoin("resources", "bcp47.json"), encoding="utf-8"))[
"subtags"
]
if x["type"] == "language"
]
license_list = json.load(open(pjoin("resources", "licenses.json"), encoding="utf-8"))
def overview_page():
st.session_state.card_dict["overview"] = st.session_state.card_dict.get(
"overview", {}
)
with st.expander("Where to find the data and its documentation", expanded=True):
key_pref = ["overview", "what"]
st.session_state.card_dict["overview"]["what"] = st.session_state.card_dict[
"overview"
].get("what", {})
make_text_area(
label="Provide a summary of this dataset in 3-4 sentences.",
key_list=key_pref + ["dataset"],
help="[free text]",
)
with st.expander("Where to find the data and its documentation", expanded=False):
key_pref = ["overview", "where"]
st.session_state.card_dict["overview"]["where"] = st.session_state.card_dict[
"overview"
].get("where", {})
make_text_input(
label="What is the webpage for the dataset (if it exists)?",
key_list=key_pref + ["website"],
help="[URL]",
)
make_text_input(
label="What is the link to where the original dataset is hosted?",
key_list=key_pref + ["data-url"],
help="[URL]",
)
make_text_input(
label="What is the link to the paper describing the dataset (open access preferred)?",
key_list=key_pref + ["paper-url"],
help="[URL]",
)
make_text_area(
label="Provide the BibTex-formatted reference for the dataset. Please use the correct published version (ACL anthology, etc.) instead of google scholar created Bibtex.",
key_list=key_pref + ["paper-bibtext"],
help="[free text]",
)
make_radio(
label="Does the dataset have an active leaderboard?",
options=["no", "yes"],
key_list=key_pref + ["has-leaderboard"],
help="If no, enter N/A for the following two fields",
)
if st.session_state.card_dict["overview"]["where"]["has-leaderboard"] == "yes":
make_text_input(
label="Provide a link to the leaderboard.",
key_list=key_pref + ["leaderboard-url"],
help="[URL] or N/A",
)
make_text_area(
label="Briefly describe how the leaderboard evaluates models.",
key_list=key_pref + ["leaderboard-description"],
help="[free text; a paragraph] or N/A",
)
else:
st.session_state.card_dict["overview"]["where"]["leaderboard-url"] = "N/A"
st.session_state.card_dict["overview"]["where"]["leaderboard-description"] = "N/A"
make_text_input(
label="If known, provide the name of at least one person the reader can contact for questions about the dataset.",
key_list=key_pref + ["contact-name"],
help="[free text]",
)
make_text_input(
label="If known, provide the email of at least one person the reader can contact for questions about the dataset.",
key_list=key_pref + ["contact-email"],
help="[free text]",
)
with st.expander("Languages and Intended Use", expanded=False):
key_pref = ["overview", "languages"]
st.session_state.card_dict["overview"][
"languages"
] = st.session_state.card_dict["overview"].get("languages", {})
make_radio(
label="Is the dataset multilingual?",
options=["no", "yes"],
key_list=key_pref + ["is-multilingual"],
help="More than one language present in all of the text fields",
)
make_multiselect(
label="What languages/dialects are covered in the dataset?",
key_list=key_pref + ["language-names"],
options=[", ".join(x["description"]) for x in languages_bcp47],
help="This is a comprehensive list of languages obtained from the BCP-47 standard list.",
)
make_text_area(
label="What dialects are covered? Are there multiple dialects per language?",
key_list=key_pref + ["language-dialects"],
help="[free text, paragraphs] - Describe the dialect(s) as appropriate.",
)
make_text_area(
label="Whose language is in the dataset?",
key_list=key_pref + ["language-speakers"],
help="[free text, paragraphs] - Provide locally appropriate demographic information about the language producers, if available. Use ranges where reasonable in order to protect individuals’ privacy.",
)
make_text_area(
label="What is the intended use of the dataset?",
key_list=key_pref + ["intended-use"],
help="[free text, paragraphs] - Describe how the dataset creators describe its purpose and intended use.",
)
make_selectbox(
label="What is the license of the dataset?",
key_list=key_pref + ["license"],
options=license_list,
help="select `other` if missing from list, `unkown` if not provided.",
)
if "other" in st.session_state.card_dict["overview"]["languages"].get("license", []):
make_text_input(
label="What is the 'other' license of the dataset?",
key_list=key_pref + ["license-other"],
help="[free text]",
)
else:
st.session_state.card_dict["overview"]["languages"]["license-other"] = "N/A"
make_selectbox(
label="What primary task does the dataset support?",
key_list=key_pref + ["task"],
options=[
"", # default needs to be invalid value to make sure people actually fill in
"Content Transfer",
"Data-to-Text",
"Dialog Response Generation",
"Paraphrasing",
"Question Generation",
"Reasoning",
"Simplification",
"Style Transfer",
"Summarization",
"Text-to-Slide",
"Other"
],
help="Select `other` if the task is not included in the list.",
)
if "Other" in st.session_state.card_dict["overview"]["languages"].get("task", []):
make_text_input(
label="What is the primary task?",
key_list=key_pref + ["task-other"],
help="[free text]",
)
else:
st.session_state.card_dict["overview"]["languages"]["task-other"] = "N/A"
make_text_area(
label="Provide a short description of the communicative goal of a model trained for this task on this dataset.",
key_list=key_pref + ["communicative"],
help="[free text, a paragraph] (e.g., describe a restaurant from a structured representation of its attributes)",
)
with st.expander("Credit", expanded=False):
key_pref = ["overview", "credit"]
st.session_state.card_dict["overview"][
"credit"
] = st.session_state.card_dict["overview"].get("credit", {})
make_multiselect(
label="In what kind of organization did the dataset curation happen?",
options=["industry", "academic", "independent", "other"],
key_list=key_pref + ["organization-type"],
)
make_text_input(
label="Name the organization(s).",
key_list=key_pref + ["organization-names"],
help="comma-separated",
)
make_text_input(
label="Who created the original dataset? List the people involved in collecting the dataset and their affiliation(s).",
key_list=key_pref + ["creators"],
help="name (affiliation); comma-separated",
)
make_text_input(
label="Who funded the data creation?",
key_list=key_pref + ["funding"],
help="[free text] enter N/A if unkown",
)
make_text_input(
label="Who contributed to the data card and adding the dataset to GEM? List the people+affiliations involved in creating this data card and who helped integrate this dataset into GEM.",
key_list=key_pref + ["gem-added-by"],
help="name (affiliation); comma-separated",
)
with st.expander("Structure", expanded=False):
key_pref = ["overview", "structure"]
st.session_state.card_dict["overview"]["structure"] = st.session_state.card_dict[
"overview"
].get("structure", {})
data_fields_help = """
[free text; paragraphs]
- Mention their data type, and whether and how they are used as part of the generation pipeline.
- Describe each fields' attributes, such as whether they are at the character level or word level, whether they are contiguous or not, etc.
- If the datasets contain example IDs, state whether they have an inherent meaning, such as a mapping to other datasets or pointing to relationships between data points.
"""
make_text_area(
label="List and describe the fields present in the dataset.",
key_list=key_pref + ["data-fields"],
help=data_fields_help,
)
make_text_area(
label="How was the dataset structure determined?",
key_list=key_pref + ["structure-description"],
help="[free text; paragraph]",
)
make_text_area(
label="How were the labels chosen?",
key_list=key_pref + ["structure-labels"],
help="[free text; paragraph]",
)
make_text_area(
label="Provide a JSON formatted example of a typical instance in the dataset.",
key_list=key_pref + ["structure-example"],
help="[JSON]",
)
make_text_area(
label="Describe and name the splits in the dataset if there are more than one.",
key_list=key_pref + ["structure-splits"],
help="[free text, paragraphs] - As appropriate, provide any descriptive statistics for the features, such as size, average lengths of input and output.",
)
make_text_area(
label="Describe any criteria for splitting the data, if used. If there are differences between the splits (e.g., if the training annotations are machine-generated and the dev and test ones are created by humans, or if different numbers of annotators contributed to each example), describe them here.",
key_list=key_pref + ["structure-splits-criteria"],
help="[free text, paragraphs]",
)
make_text_area(
label="What does an outlier of the dataset in terms of length/perplexity/embedding look like?",
key_list=key_pref + ["structure-outlier"],
help="[free text + json formatted text/file for an example]",
)
def overview_summary():
total_filled = sum(
[len(dct) for dct in st.session_state.card_dict.get("overview", {}).values()]
)
with st.expander(
f"Dataset Overview Completion - {total_filled} of {N_FIELDS}", expanded=False
):
completion_markdown = ""
completion_markdown += (
f"- **Overall completion:**\n - {total_filled} of {N_FIELDS} fields\n"
)
completion_markdown += f"- **Sub-section - Where to find:**\n - {len(st.session_state.card_dict.get('overview', {}).get('where', {}))} of {N_FIELDS_WHERE} fields\n"
completion_markdown += f"- **Sub-section - Languages and Intended Use:**\n - {len(st.session_state.card_dict.get('overview', {}).get('languages', {}))} of {N_FIELDS_LANGUAGES} fields\n"
completion_markdown += f"- **Sub-section - Credit:**\n - {len(st.session_state.card_dict.get('overview', {}).get('credit', {}))} of {N_FIELDS_CREDIT} fields\n"
completion_markdown += f"- **Sub-section - Structure:**\n - {len(st.session_state.card_dict.get('overview', {}).get('structure', {}))} of {N_FIELDS_STRUCTURE} fields\n"
st.markdown(completion_markdown)
|