Spaces:
Running
Running
jasonshaoshun
commited on
Commit
·
49218db
1
Parent(s):
53c242a
debug
Browse files
app.py
CHANGED
|
@@ -128,6 +128,66 @@ from src.about import TasksMib_Subgraph
|
|
| 128 |
|
| 129 |
|
| 130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
def init_leaderboard_mib_subgraph(dataframe, track):
|
| 132 |
"""Initialize the subgraph leaderboard with grouped column selection by benchmark."""
|
| 133 |
if dataframe is None or dataframe.empty:
|
|
@@ -135,57 +195,69 @@ def init_leaderboard_mib_subgraph(dataframe, track):
|
|
| 135 |
|
| 136 |
print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
|
| 137 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
# Create groups of columns by benchmark
|
| 139 |
benchmark_groups = []
|
| 140 |
-
|
| 141 |
-
# For each benchmark in our TasksMib_Subgraph enum...
|
| 142 |
for task in TasksMib_Subgraph:
|
| 143 |
benchmark = task.value.benchmark
|
| 144 |
-
# Get all valid columns for this benchmark's models
|
| 145 |
benchmark_cols = [
|
| 146 |
-
f"{benchmark}
|
| 147 |
for model in task.value.models
|
| 148 |
-
if f"{benchmark}_{model}" in dataframe.columns
|
| 149 |
]
|
| 150 |
-
if benchmark_cols:
|
| 151 |
benchmark_groups.append(benchmark_cols)
|
| 152 |
print(f"\nBenchmark group for {benchmark}:", benchmark_cols)
|
| 153 |
|
| 154 |
-
# Create model groups
|
| 155 |
model_groups = []
|
| 156 |
all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
|
| 157 |
|
| 158 |
-
# For each unique model...
|
| 159 |
for model in all_models:
|
| 160 |
-
# Get all valid columns for this model across benchmarks
|
| 161 |
model_cols = [
|
| 162 |
-
f"{task.value.benchmark}
|
| 163 |
for task in TasksMib_Subgraph
|
| 164 |
if model in task.value.models
|
| 165 |
and f"{task.value.benchmark}_{model}" in dataframe.columns
|
| 166 |
]
|
| 167 |
-
if model_cols:
|
| 168 |
model_groups.append(model_cols)
|
| 169 |
print(f"\nModel group for {model}:", model_cols)
|
| 170 |
|
| 171 |
-
# Combine
|
| 172 |
all_groups = benchmark_groups + model_groups
|
| 173 |
-
|
| 174 |
-
# Flatten groups for default selection (show everything initially)
|
| 175 |
all_columns = [col for group in all_groups for col in group]
|
| 176 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 177 |
|
| 178 |
return Leaderboard(
|
| 179 |
-
value=
|
| 180 |
datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
|
| 181 |
select_columns=SelectColumns(
|
| 182 |
-
default_selection=all_columns,
|
| 183 |
label="Select Results:"
|
| 184 |
),
|
| 185 |
search_columns=["Method"],
|
| 186 |
hide_columns=[],
|
| 187 |
interactive=False,
|
| 188 |
)
|
|
|
|
|
|
|
| 189 |
|
| 190 |
# def init_leaderboard_mib_subgraph(dataframe, track):
|
| 191 |
# """Initialize the subgraph leaderboard with group-based column selection."""
|
|
|
|
| 128 |
|
| 129 |
|
| 130 |
|
| 131 |
+
# def init_leaderboard_mib_subgraph(dataframe, track):
|
| 132 |
+
# """Initialize the subgraph leaderboard with grouped column selection by benchmark."""
|
| 133 |
+
# if dataframe is None or dataframe.empty:
|
| 134 |
+
# raise ValueError("Leaderboard DataFrame is empty or None.")
|
| 135 |
+
|
| 136 |
+
# print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
|
| 137 |
+
|
| 138 |
+
# # Create groups of columns by benchmark
|
| 139 |
+
# benchmark_groups = []
|
| 140 |
+
|
| 141 |
+
# # For each benchmark in our TasksMib_Subgraph enum...
|
| 142 |
+
# for task in TasksMib_Subgraph:
|
| 143 |
+
# benchmark = task.value.benchmark
|
| 144 |
+
# # Get all valid columns for this benchmark's models
|
| 145 |
+
# benchmark_cols = [
|
| 146 |
+
# f"{benchmark}_{model}"
|
| 147 |
+
# for model in task.value.models
|
| 148 |
+
# if f"{benchmark}_{model}" in dataframe.columns
|
| 149 |
+
# ]
|
| 150 |
+
# if benchmark_cols: # Only add if we have valid columns
|
| 151 |
+
# benchmark_groups.append(benchmark_cols)
|
| 152 |
+
# print(f"\nBenchmark group for {benchmark}:", benchmark_cols)
|
| 153 |
+
|
| 154 |
+
# # Create model groups as well
|
| 155 |
+
# model_groups = []
|
| 156 |
+
# all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
|
| 157 |
+
|
| 158 |
+
# # For each unique model...
|
| 159 |
+
# for model in all_models:
|
| 160 |
+
# # Get all valid columns for this model across benchmarks
|
| 161 |
+
# model_cols = [
|
| 162 |
+
# f"{task.value.benchmark}_{model}"
|
| 163 |
+
# for task in TasksMib_Subgraph
|
| 164 |
+
# if model in task.value.models
|
| 165 |
+
# and f"{task.value.benchmark}_{model}" in dataframe.columns
|
| 166 |
+
# ]
|
| 167 |
+
# if model_cols: # Only add if we have valid columns
|
| 168 |
+
# model_groups.append(model_cols)
|
| 169 |
+
# print(f"\nModel group for {model}:", model_cols)
|
| 170 |
+
|
| 171 |
+
# # Combine all groups
|
| 172 |
+
# all_groups = benchmark_groups + model_groups
|
| 173 |
+
|
| 174 |
+
# # Flatten groups for default selection (show everything initially)
|
| 175 |
+
# all_columns = [col for group in all_groups for col in group]
|
| 176 |
+
# print("\nAll available columns:", all_columns)
|
| 177 |
+
|
| 178 |
+
# return Leaderboard(
|
| 179 |
+
# value=dataframe,
|
| 180 |
+
# datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
|
| 181 |
+
# select_columns=SelectColumns(
|
| 182 |
+
# default_selection=all_columns, # Show all columns initially
|
| 183 |
+
# label="Select Results:"
|
| 184 |
+
# ),
|
| 185 |
+
# search_columns=["Method"],
|
| 186 |
+
# hide_columns=[],
|
| 187 |
+
# interactive=False,
|
| 188 |
+
# )
|
| 189 |
+
|
| 190 |
+
|
| 191 |
def init_leaderboard_mib_subgraph(dataframe, track):
|
| 192 |
"""Initialize the subgraph leaderboard with grouped column selection by benchmark."""
|
| 193 |
if dataframe is None or dataframe.empty:
|
|
|
|
| 195 |
|
| 196 |
print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
|
| 197 |
|
| 198 |
+
# First, create a mapping between field names and display names
|
| 199 |
+
field_to_display = {}
|
| 200 |
+
for field in fields(AutoEvalColumn_mib_subgraph):
|
| 201 |
+
if hasattr(field, 'name') and hasattr(field, 'type'):
|
| 202 |
+
field_to_display[field.name] = field.type
|
| 203 |
+
|
| 204 |
# Create groups of columns by benchmark
|
| 205 |
benchmark_groups = []
|
|
|
|
|
|
|
| 206 |
for task in TasksMib_Subgraph:
|
| 207 |
benchmark = task.value.benchmark
|
| 208 |
+
# Get all valid columns for this benchmark's models, using display names
|
| 209 |
benchmark_cols = [
|
| 210 |
+
f"{benchmark}({model})" # Use display name format
|
| 211 |
for model in task.value.models
|
| 212 |
+
if f"{benchmark}_{model}" in dataframe.columns # Still check using field name
|
| 213 |
]
|
| 214 |
+
if benchmark_cols:
|
| 215 |
benchmark_groups.append(benchmark_cols)
|
| 216 |
print(f"\nBenchmark group for {benchmark}:", benchmark_cols)
|
| 217 |
|
| 218 |
+
# Create model groups with display names
|
| 219 |
model_groups = []
|
| 220 |
all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
|
| 221 |
|
|
|
|
| 222 |
for model in all_models:
|
|
|
|
| 223 |
model_cols = [
|
| 224 |
+
f"{task.value.benchmark}({model})" # Use display name format
|
| 225 |
for task in TasksMib_Subgraph
|
| 226 |
if model in task.value.models
|
| 227 |
and f"{task.value.benchmark}_{model}" in dataframe.columns
|
| 228 |
]
|
| 229 |
+
if model_cols:
|
| 230 |
model_groups.append(model_cols)
|
| 231 |
print(f"\nModel group for {model}:", model_cols)
|
| 232 |
|
| 233 |
+
# Combine and flatten groups
|
| 234 |
all_groups = benchmark_groups + model_groups
|
|
|
|
|
|
|
| 235 |
all_columns = [col for group in all_groups for col in group]
|
| 236 |
+
|
| 237 |
+
# Important: We need to rename the DataFrame columns to match our display names
|
| 238 |
+
display_name_mapping = {
|
| 239 |
+
f"{task.value.benchmark}_{model}": f"{task.value.benchmark}({model})"
|
| 240 |
+
for task in TasksMib_Subgraph
|
| 241 |
+
for model in task.value.models
|
| 242 |
+
if f"{task.value.benchmark}_{model}" in dataframe.columns
|
| 243 |
+
}
|
| 244 |
+
|
| 245 |
+
# Create a copy of the DataFrame with renamed columns
|
| 246 |
+
display_df = dataframe.rename(columns=display_name_mapping)
|
| 247 |
|
| 248 |
return Leaderboard(
|
| 249 |
+
value=display_df, # Use the DataFrame with display names
|
| 250 |
datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
|
| 251 |
select_columns=SelectColumns(
|
| 252 |
+
default_selection=all_columns,
|
| 253 |
label="Select Results:"
|
| 254 |
),
|
| 255 |
search_columns=["Method"],
|
| 256 |
hide_columns=[],
|
| 257 |
interactive=False,
|
| 258 |
)
|
| 259 |
+
|
| 260 |
+
|
| 261 |
|
| 262 |
# def init_leaderboard_mib_subgraph(dataframe, track):
|
| 263 |
# """Initialize the subgraph leaderboard with group-based column selection."""
|