trainer.base_trainer
BaseTrainer
Source code in src/python/easydel/trainer/base_trainer.py
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 |
|
__init__(arguments, dataset_train, dataset_eval=None, finetune=True, checkpoint_path=None, _do_init_fns=True)
The init function is called when the class is instantiated. It sets up all the variables that are needed for training, including: - The timer to keep track of how long each epoch takes. - The dataloaders for both training and evaluation (if provided). - The model itself, which will be created from a checkpoint if one was provided. Otherwise, it will be created from scratch using the arguments passed in by the user. Note that this function also handles creating a mesh if one was not already specified in arguments or loaded from a checkpoint file (see below). This means that you can pass in either
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Represent the instance of the class |
required | |
arguments |
TrainArguments
|
TrainArguments: Pass the arguments to the trainer |
required |
dataset_train |
Dataset
|
Dataset: Pass the training dataset to the trainer |
required |
dataset_eval |
Dataset
|
Dataset: Pass the validation dataset |
None
|
finetune |
bool
|
bool: Load the model from a checkpoint |
True
|
checkpoint_path |
Union[str, PathLike]
|
Union[str,os.PathLike] : Load the checkpoint path |
None
|
_do_init_fns |
bool
|
bool: Initialize the functions |
True
|
Returns:
Type | Description |
---|---|
Nothing, it just initializes the class |
Source code in src/python/easydel/trainer/base_trainer.py
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 |
|
configure_dataloader()
The configure_dataloader function is used to configure the dataloader for training and evaluation.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Refer to the class instance itself |
required |
Returns:
Type | Description |
---|---|
TrainerConfigureDataloaderFuncOutput
|
A TrainerConfigureDataloaderFuncOutput object |
Source code in src/python/easydel/trainer/base_trainer.py
293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 |
|
configure_functions()
abstractmethod
The configure_functions function is responsible for configuring the functions that will be used in training. It does this by first defining a function called function_configurations, which initializes the model parameters and returns them as a EasyDeLState object. The EasyDeLState object contains all the information needed to train or evaluate on a batch of data, including:
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Access the class attributes |
required |
Returns:
Type | Description |
---|---|
TrainerConfigureFunctionFuncOutput
|
A TrainerConfigureFunctionFuncOutput object |
Source code in src/python/easydel/trainer/base_trainer.py
280 281 282 283 284 285 286 287 288 289 290 291 |
|
configure_model()
The configure_model function is responsible for creating the model, optimizer and scheduler.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Represent the instance of the class |
required |
Returns:
Type | Description |
---|---|
TrainerConfigureModelFuncOutput
|
A model, optimizer, scheduler and config in TrainerConfigureModelFuncOutput Object |
Source code in src/python/easydel/trainer/base_trainer.py
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 |
|
eval(state)
abstractmethod
abstract of Eval Function to evaluate model
Source code in src/python/easydel/trainer/base_trainer.py
459 460 461 462 463 |
|
finish()
staticmethod
The finish function is called when the experiment ends. It can be used to save data, upload files, or do any other cleanup tasks.
Returns:
Type | Description |
---|---|
A dictionary of the run's metadata |
Source code in src/python/easydel/trainer/base_trainer.py
178 179 180 181 182 183 184 185 186 187 |
|
initialize_trainer_utils()
The initialize_trainer_utils function is responsible for initializing the following: - wandb_runtime (if you use_wandb is True) - timer object (for logging time taken by various functions) - dataloader objects for training and evaluation data, along with max steps per epoch. The configure_dataloader function accomplishes this task.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Represent the instance of the class |
required |
Returns:
Type | Description |
---|---|
A tuple of functions |
Source code in src/python/easydel/trainer/base_trainer.py
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 |
|
train()
abstractmethod
abstract of Train Function to train model
Source code in src/python/easydel/trainer/base_trainer.py
453 454 455 456 457 |
|