Spaces:
Sleeping
Sleeping
Upload folder using huggingface_hub
Browse files
server/experiment_repo/src/train.py
CHANGED
|
@@ -32,7 +32,14 @@ def main():
|
|
| 32 |
model_type = config['model']['type']
|
| 33 |
print(f"Initializing {model_type} model...")
|
| 34 |
|
| 35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
|
| 37 |
# training model
|
| 38 |
print("Starting training...")
|
|
|
|
| 32 |
model_type = config['model']['type']
|
| 33 |
print(f"Initializing {model_type} model...")
|
| 34 |
|
| 35 |
+
removed_columns = config.get('removed_columns', [])
|
| 36 |
+
if not removed_columns:
|
| 37 |
+
if 'model' in config and 'removed_columns' in config['model']:
|
| 38 |
+
removed_columns = config['model']['removed_columns']
|
| 39 |
+
elif 'data' in config and 'removed_columns' in config['data']:
|
| 40 |
+
removed_columns = config['data']['removed_columns']
|
| 41 |
+
|
| 42 |
+
model = build_model(model_type, config, removed_columns=removed_columns)
|
| 43 |
|
| 44 |
# training model
|
| 45 |
print("Starting training...")
|
server/my_env_environment.py
CHANGED
|
@@ -103,6 +103,7 @@ class MyEnvironment(Environment):
|
|
| 103 |
for writing feature names remember you can write them as comma separated values or as plus separated values or as group names as given in the feature_description_txt file.
|
| 104 |
comma separated means that, those individual features give different F1 when individually removed , whereas plus separated means that, all those features give a single F1 when removed as a group.
|
| 105 |
similarly, writing a category name means , all the features in that category give a single F1 when removed as a group.
|
|
|
|
| 106 |
''')]
|
| 107 |
},
|
| 108 |
dir_structure=init_Directory("exp_repo",[
|
|
@@ -232,8 +233,13 @@ class MyEnvironment(Environment):
|
|
| 232 |
if not b:
|
| 233 |
raise AttributeError("The path does not lead to any existing directory.")
|
| 234 |
self._state.cwd_contents=getattr(self._state.dir_structure,l[-1])
|
| 235 |
-
|
| 236 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 237 |
except AttributeError as e:
|
| 238 |
message=str(e)
|
| 239 |
penalty+=-0.12
|
|
@@ -241,15 +247,17 @@ class MyEnvironment(Environment):
|
|
| 241 |
def _read(self,action: MyAction,penalty:float)->tuple[str,float]:
|
| 242 |
message=""
|
| 243 |
try:
|
| 244 |
-
|
|
|
|
|
|
|
| 245 |
if b:
|
| 246 |
raise AttributeError("The path does not lead to any existing file.")
|
| 247 |
-
self._state.file_contents=getattr(self._state.dir_structure,
|
| 248 |
message="Read file "+action.path
|
| 249 |
# Emit event
|
| 250 |
self.events["results_read"] += 1
|
| 251 |
-
except AttributeError
|
| 252 |
-
message=
|
| 253 |
penalty+=-0.12
|
| 254 |
return message,penalty
|
| 255 |
def _write(self,action: MyAction,penalty:float)->tuple[str,float]:
|
|
@@ -260,12 +268,27 @@ class MyEnvironment(Environment):
|
|
| 260 |
|
| 261 |
# Dynamic physical directory path resolution
|
| 262 |
server_dir = os.path.dirname(os.path.abspath(__file__))
|
| 263 |
-
|
| 264 |
-
if
|
| 265 |
-
|
|
|
|
|
|
|
| 266 |
|
| 267 |
-
|
| 268 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 269 |
|
| 270 |
l = action.path.split('/')
|
| 271 |
filename = l[-1]
|
|
@@ -311,11 +334,30 @@ class MyEnvironment(Environment):
|
|
| 311 |
raise ValueError("Path is required for run action.")
|
| 312 |
|
| 313 |
server_dir = os.path.dirname(os.path.abspath(__file__))
|
| 314 |
-
|
| 315 |
-
if
|
| 316 |
-
|
| 317 |
-
|
| 318 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 319 |
logical_path = resolved_script_path.replace(os.sep, "/")
|
| 320 |
|
| 321 |
if not logical_path.endswith("src/train.py"):
|
|
@@ -323,11 +365,14 @@ class MyEnvironment(Environment):
|
|
| 323 |
|
| 324 |
# Build command
|
| 325 |
if action.content:
|
| 326 |
-
|
|
|
|
|
|
|
|
|
|
| 327 |
else:
|
| 328 |
command = ["python", resolved_script_path]
|
| 329 |
|
| 330 |
-
result = subprocess.run(command, capture_output=True, text=True, cwd=
|
| 331 |
|
| 332 |
if result.returncode == 0:
|
| 333 |
# Emit the correct event based on which config was used
|
|
@@ -335,9 +380,10 @@ class MyEnvironment(Environment):
|
|
| 335 |
self.events["xgb_run"] += 1
|
| 336 |
else:
|
| 337 |
self.events["nn_run"] += 1
|
| 338 |
-
message = f"Successfully ran training script with config: {action.content or 'default'}"
|
|
|
|
| 339 |
else:
|
| 340 |
-
raise RuntimeError(f"Script failed (code {result.returncode}):\n{result.stderr.strip()[:
|
| 341 |
except Exception as e:
|
| 342 |
message = str(e)
|
| 343 |
penalty += -0.12
|
|
|
|
| 103 |
for writing feature names remember you can write them as comma separated values or as plus separated values or as group names as given in the feature_description_txt file.
|
| 104 |
comma separated means that, those individual features give different F1 when individually removed , whereas plus separated means that, all those features give a single F1 when removed as a group.
|
| 105 |
similarly, writing a category name means , all the features in that category give a single F1 when removed as a group.
|
| 106 |
+
how to test:place a removed_columns: [...] list parameter anywhere under, 'model', or 'data' in the base config, to remove columns.
|
| 107 |
''')]
|
| 108 |
},
|
| 109 |
dir_structure=init_Directory("exp_repo",[
|
|
|
|
| 233 |
if not b:
|
| 234 |
raise AttributeError("The path does not lead to any existing directory.")
|
| 235 |
self._state.cwd_contents=getattr(self._state.dir_structure,l[-1])
|
| 236 |
+
|
| 237 |
+
if action.path.startswith("exp_repo"):
|
| 238 |
+
self._state.cwd = action.path
|
| 239 |
+
else:
|
| 240 |
+
self._state.cwd = self._state.cwd.rstrip('/') + "/" + action.path.lstrip('/')
|
| 241 |
+
|
| 242 |
+
message="Changed directory to "+self._state.cwd
|
| 243 |
except AttributeError as e:
|
| 244 |
message=str(e)
|
| 245 |
penalty+=-0.12
|
|
|
|
| 247 |
def _read(self,action: MyAction,penalty:float)->tuple[str,float]:
|
| 248 |
message=""
|
| 249 |
try:
|
| 250 |
+
l=action.path.split('/')
|
| 251 |
+
attr_name=l[-1].replace('.', '_')
|
| 252 |
+
b=isinstance(getattr(self._state.dir_structure,attr_name),list)
|
| 253 |
if b:
|
| 254 |
raise AttributeError("The path does not lead to any existing file.")
|
| 255 |
+
self._state.file_contents=getattr(self._state.dir_structure,attr_name)
|
| 256 |
message="Read file "+action.path
|
| 257 |
# Emit event
|
| 258 |
self.events["results_read"] += 1
|
| 259 |
+
except AttributeError:
|
| 260 |
+
message=f"The file {action.path} does not exist."
|
| 261 |
penalty+=-0.12
|
| 262 |
return message,penalty
|
| 263 |
def _write(self,action: MyAction,penalty:float)->tuple[str,float]:
|
|
|
|
| 268 |
|
| 269 |
# Dynamic physical directory path resolution
|
| 270 |
server_dir = os.path.dirname(os.path.abspath(__file__))
|
| 271 |
+
|
| 272 |
+
if action.path.startswith("exp_repo"):
|
| 273 |
+
logical_path = action.path
|
| 274 |
+
else:
|
| 275 |
+
logical_path = self._state.cwd.rstrip('/') + "/" + action.path.lstrip('/')
|
| 276 |
|
| 277 |
+
logical_path = '/'.join([p for p in logical_path.split('/') if p])
|
| 278 |
+
|
| 279 |
+
if logical_path.startswith("exp_repo"):
|
| 280 |
+
rel_path = logical_path[len("exp_repo"):]
|
| 281 |
+
if rel_path.startswith('/'): rel_path = rel_path[1:]
|
| 282 |
+
else:
|
| 283 |
+
rel_path = logical_path
|
| 284 |
+
|
| 285 |
+
l_rel = rel_path.split('/')
|
| 286 |
+
filename = l_rel[-1]
|
| 287 |
+
attr_name = filename.replace('.', '_') # dummy name for dir_structure
|
| 288 |
+
actual_filename = attr_name[::-1].replace("_", ".", 1)[::-1] # convert dummy name to real filename
|
| 289 |
+
l_rel[-1] = actual_filename
|
| 290 |
+
|
| 291 |
+
resolved_file_path = os.path.normpath(os.path.join(server_dir, "experiment_repo", *l_rel))
|
| 292 |
|
| 293 |
l = action.path.split('/')
|
| 294 |
filename = l[-1]
|
|
|
|
| 334 |
raise ValueError("Path is required for run action.")
|
| 335 |
|
| 336 |
server_dir = os.path.dirname(os.path.abspath(__file__))
|
| 337 |
+
|
| 338 |
+
if action.path.startswith("exp_repo"):
|
| 339 |
+
logical_path = action.path
|
| 340 |
+
else:
|
| 341 |
+
logical_path = self._state.cwd.rstrip('/') + "/" + action.path.lstrip('/')
|
| 342 |
+
|
| 343 |
+
logical_path = '/'.join([p for p in logical_path.split('/') if p])
|
| 344 |
+
|
| 345 |
+
if logical_path.startswith("exp_repo"):
|
| 346 |
+
rel_path = logical_path[len("exp_repo"):]
|
| 347 |
+
if rel_path.startswith('/'): rel_path = rel_path[1:]
|
| 348 |
+
else:
|
| 349 |
+
rel_path = logical_path
|
| 350 |
+
|
| 351 |
+
physical_cwd = os.path.normpath(os.path.join(server_dir, "experiment_repo"))
|
| 352 |
+
resolved_script_path = os.path.normpath(os.path.join(physical_cwd, rel_path))
|
| 353 |
+
|
| 354 |
+
rel_cwd_str = self._state.cwd
|
| 355 |
+
if rel_cwd_str.startswith("exp_repo"):
|
| 356 |
+
rel_cwd_str = rel_cwd_str[len("exp_repo"):]
|
| 357 |
+
if rel_cwd_str.startswith('/'): rel_cwd_str = rel_cwd_str[1:]
|
| 358 |
+
|
| 359 |
+
cmd_cwd = os.path.normpath(os.path.join(server_dir, "experiment_repo", rel_cwd_str))
|
| 360 |
+
|
| 361 |
logical_path = resolved_script_path.replace(os.sep, "/")
|
| 362 |
|
| 363 |
if not logical_path.endswith("src/train.py"):
|
|
|
|
| 365 |
|
| 366 |
# Build command
|
| 367 |
if action.content:
|
| 368 |
+
content_real = action.content
|
| 369 |
+
if '.' not in content_real and '_' in content_real:
|
| 370 |
+
content_real = content_real[::-1].replace('_', '.', 1)[::-1]
|
| 371 |
+
command = ["python", resolved_script_path, "--config", content_real]
|
| 372 |
else:
|
| 373 |
command = ["python", resolved_script_path]
|
| 374 |
|
| 375 |
+
result = subprocess.run(command, capture_output=True, text=True, cwd=cmd_cwd)
|
| 376 |
|
| 377 |
if result.returncode == 0:
|
| 378 |
# Emit the correct event based on which config was used
|
|
|
|
| 380 |
self.events["xgb_run"] += 1
|
| 381 |
else:
|
| 382 |
self.events["nn_run"] += 1
|
| 383 |
+
message = f"Successfully ran training script with config: {action.content or 'default'}\n"
|
| 384 |
+
message += f"Script Output:\n{result.stdout.strip()[:1000]}"
|
| 385 |
else:
|
| 386 |
+
raise RuntimeError(f"Script failed (code {result.returncode}):\n{result.stderr.strip()[:1000]}")
|
| 387 |
except Exception as e:
|
| 388 |
message = str(e)
|
| 389 |
penalty += -0.12
|