Datasets:
Remove space in split names
Browse files- diffusiondb.py +11 -11
diffusiondb.py
CHANGED
@@ -110,11 +110,11 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
110 |
for sampling in ["first", "random"]:
|
111 |
for is_large in [False, True]:
|
112 |
num_k_str = f"{num_k}k" if num_k < 1000 else f"{num_k // 1000}m"
|
113 |
-
subset_str = "
|
114 |
|
115 |
if sampling == "random":
|
116 |
# Name the config
|
117 |
-
cur_name = "random_" + num_k_str
|
118 |
|
119 |
# Add a short description for each config
|
120 |
cur_description = (
|
@@ -128,7 +128,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
128 |
).tolist()
|
129 |
else:
|
130 |
# Name the config
|
131 |
-
cur_name = "first_" + num_k_str
|
132 |
|
133 |
# Add a short description for each config
|
134 |
cur_description = f"The first {num_k_str} images in this dataset with their prompts and parameters"
|
@@ -151,11 +151,11 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
151 |
for num_k in [5000, 10000]:
|
152 |
for sampling in ["first", "random"]:
|
153 |
num_k_str = f"{num_k // 1000}m"
|
154 |
-
subset_str = "
|
155 |
|
156 |
if sampling == "random":
|
157 |
# Name the config
|
158 |
-
cur_name = "random_" + num_k_str
|
159 |
|
160 |
# Add a short description for each config
|
161 |
cur_description = (
|
@@ -169,7 +169,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
169 |
).tolist()
|
170 |
else:
|
171 |
# Name the config
|
172 |
-
cur_name = "first_" + num_k_str
|
173 |
|
174 |
# Add a short description for each config
|
175 |
cur_description = f"The first {num_k_str} images in this dataset with their prompts and parameters"
|
@@ -191,7 +191,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
191 |
# Need to manually add all (2m) and all (large)
|
192 |
BUILDER_CONFIGS.append(
|
193 |
DiffusionDBConfig(
|
194 |
-
name="
|
195 |
part_ids=_PART_IDS,
|
196 |
is_large=False,
|
197 |
description="All images with their prompts and parameters",
|
@@ -200,7 +200,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
200 |
|
201 |
BUILDER_CONFIGS.append(
|
202 |
DiffusionDBConfig(
|
203 |
-
name="
|
204 |
part_ids=_PART_IDS_LARGE,
|
205 |
is_large=True,
|
206 |
description="All images with their prompts and parameters",
|
@@ -210,7 +210,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
210 |
# We also prove a text-only option, which loads the meatadata parquet file
|
211 |
BUILDER_CONFIGS.append(
|
212 |
DiffusionDBConfig(
|
213 |
-
name="
|
214 |
part_ids=[],
|
215 |
is_large=False,
|
216 |
description="Only include all prompts and parameters (no image)",
|
@@ -219,7 +219,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
219 |
|
220 |
BUILDER_CONFIGS.append(
|
221 |
DiffusionDBConfig(
|
222 |
-
name="
|
223 |
part_ids=[],
|
224 |
is_large=True,
|
225 |
description="Only include all prompts and parameters (no image)",
|
@@ -227,7 +227,7 @@ class DiffusionDB(datasets.GeneratorBasedBuilder):
|
|
227 |
)
|
228 |
|
229 |
# Default to only load 1k random images
|
230 |
-
DEFAULT_CONFIG_NAME = "
|
231 |
|
232 |
def _info(self):
|
233 |
"""Specify the information of DiffusionDB."""
|
|
|
110 |
for sampling in ["first", "random"]:
|
111 |
for is_large in [False, True]:
|
112 |
num_k_str = f"{num_k}k" if num_k < 1000 else f"{num_k // 1000}m"
|
113 |
+
subset_str = "large_" if is_large else "2m_"
|
114 |
|
115 |
if sampling == "random":
|
116 |
# Name the config
|
117 |
+
cur_name = subset_str + "random_" + num_k_str
|
118 |
|
119 |
# Add a short description for each config
|
120 |
cur_description = (
|
|
|
128 |
).tolist()
|
129 |
else:
|
130 |
# Name the config
|
131 |
+
cur_name = subset_str + "first_" + num_k_str
|
132 |
|
133 |
# Add a short description for each config
|
134 |
cur_description = f"The first {num_k_str} images in this dataset with their prompts and parameters"
|
|
|
151 |
for num_k in [5000, 10000]:
|
152 |
for sampling in ["first", "random"]:
|
153 |
num_k_str = f"{num_k // 1000}m"
|
154 |
+
subset_str = "large_"
|
155 |
|
156 |
if sampling == "random":
|
157 |
# Name the config
|
158 |
+
cur_name = subset_str + "random_" + num_k_str
|
159 |
|
160 |
# Add a short description for each config
|
161 |
cur_description = (
|
|
|
169 |
).tolist()
|
170 |
else:
|
171 |
# Name the config
|
172 |
+
cur_name = subset_str + "first_" + num_k_str
|
173 |
|
174 |
# Add a short description for each config
|
175 |
cur_description = f"The first {num_k_str} images in this dataset with their prompts and parameters"
|
|
|
191 |
# Need to manually add all (2m) and all (large)
|
192 |
BUILDER_CONFIGS.append(
|
193 |
DiffusionDBConfig(
|
194 |
+
name="2m_all",
|
195 |
part_ids=_PART_IDS,
|
196 |
is_large=False,
|
197 |
description="All images with their prompts and parameters",
|
|
|
200 |
|
201 |
BUILDER_CONFIGS.append(
|
202 |
DiffusionDBConfig(
|
203 |
+
name="large_all",
|
204 |
part_ids=_PART_IDS_LARGE,
|
205 |
is_large=True,
|
206 |
description="All images with their prompts and parameters",
|
|
|
210 |
# We also prove a text-only option, which loads the meatadata parquet file
|
211 |
BUILDER_CONFIGS.append(
|
212 |
DiffusionDBConfig(
|
213 |
+
name="2m_text_only",
|
214 |
part_ids=[],
|
215 |
is_large=False,
|
216 |
description="Only include all prompts and parameters (no image)",
|
|
|
219 |
|
220 |
BUILDER_CONFIGS.append(
|
221 |
DiffusionDBConfig(
|
222 |
+
name="large_text_only",
|
223 |
part_ids=[],
|
224 |
is_large=True,
|
225 |
description="Only include all prompts and parameters (no image)",
|
|
|
227 |
)
|
228 |
|
229 |
# Default to only load 1k random images
|
230 |
+
DEFAULT_CONFIG_NAME = "2m_random_1k"
|
231 |
|
232 |
def _info(self):
|
233 |
"""Specify the information of DiffusionDB."""
|