Skip to content

train

Script to train initialization model.

Trainer

Trainer for single shot pose and shape estimation network.

Source code in sdfest/initialization/scripts/train.py
 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
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
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
class Trainer:
    """Trainer for single shot pose and shape estimation network."""

    def __init__(self, config: dict) -> None:
        """Construct trainer.

        Args:
            config: The configuration for model and training.
        """
        self._read_config(config)

    def _read_config(self, config: dict) -> None:
        self._config = config
        self._validation_iteration = config["validation_iteration"]
        self._visualization_iteration = config["visualization_iteration"]
        self._checkpoint_iteration = config["checkpoint_iteration"]
        self._iterations = config["iterations"]
        self._init_weights_path = (
            config["init_weights"] if "init_weights" in config else None
        )

        # propagate orientation representation and category to datasets
        datasets = list(self._config["datasets"].values()) + list(
            self._config["validation_datasets"].values()
        )
        for dataset in datasets:
            dataset["config_dict"]["orientation_repr"] = config["orientation_repr"]
            if "orientation_grid_resolution" in config:
                dataset["config_dict"]["orientation_grid_resolution"] = config[
                    "orientation_grid_resolution"
                ]
            if "category_str" in config:
                dataset["config_dict"]["category_str"] = config["category_str"]

        # propagate orientation representation to init head
        self._config["head"]["orientation_repr"] = config["orientation_repr"]
        if "orientation_grid_resolution" in config:
            self._config["head"]["orientation_grid_resolution"] = config[
                "orientation_grid_resolution"
            ]

    def run(self) -> None:
        """Train the model."""
        wandb.init(project="sdfest.initialization", config=self._config)

        self._device = self.get_device()

        # init dataset and dataloader
        self.vae = self.create_sdfvae()

        # init model to train
        self._sdf_pose_net = SDFPoseNet(
            backbone=MODULE_DICT[self._config["backbone_type"]](
                **self._config["backbone"]
            ),
            head=MODULE_DICT[self._config["head_type"]](
                shape_dimension=self._config["vae"]["latent_size"],
                **self._config["head"],
            ),
        ).to(self._device)
        self._sdf_pose_net.train()

        # deterministic samples (needs to be done after model initialization, as it
        # can have varying number of parameters)
        torch.manual_seed(0)
        random.seed(torch.initial_seed())  # to get deterministic examples

        # print network summary
        torchinfo.summary(self._sdf_pose_net, (1, 500, 3), device=self._device)

        # init optimizer
        self._optimizer = torch.optim.Adam(
            self._sdf_pose_net.parameters(), lr=self._config["learning_rate"]
        )

        # load weights if provided
        if self._init_weights_path is not None:
            state_dict = torch.load(self._init_weights_path, map_location=self._device)
            self._sdf_pose_net.load_state_dict(state_dict)

        self._current_iteration = 0
        self._run_name = (
            f"sdfest.initialization_{datetime.now().strftime('%Y-%m-%d_%H-%M-%S-%f')}"
        )

        wandb.config.run_name = (
            self._run_name  # to allow association of pt files with wandb runs
        )

        self._model_base_path = os.path.join(os.getcwd(), "models", self._run_name)

        self._multi_data_loader = self._create_multi_data_loader()
        self._validation_data_loader_dict = self._create_validation_data_loader_dict()

        # backup config to model directory
        os.makedirs(self._model_base_path, exist_ok=True)
        config_path = os.path.join(self._model_base_path, "config.yaml")
        yoco.save_config_to_file(config_path, self._config)

        self._start_time = time.time()
        for samples in self._multi_data_loader:
            self._current_iteration += 1
            print(f"Current iteration: {self._current_iteration}\033[K", end="\r")
            self._update_progress()

            samples = utils.dict_to(samples, self._device)

            latent_shape, position, scale, orientation = self._sdf_pose_net(
                samples["pointset"]
            )
            predictions = {
                "latent_shape": latent_shape,
                "position": position,
                "scale": scale,
                "orientation": orientation,
            }

            loss = self._compute_loss(samples, predictions)
            self._optimizer.zero_grad()
            loss.backward()
            self._optimizer.step()

            with torch.no_grad():
                # samples_dict = defaultdict(lambda: dict())
                # for k,vs in samples.items():
                #     for i, v in enumerate(vs):
                #         samples_dict[i][k] = v
                # for sample in samples_dict.values():
                #     utils.visualize_sample(sample, None)

                self._compute_metrics(samples, predictions)

                if self._current_iteration % self._visualization_iteration == 0:
                    self._generate_visualizations()

                if self._current_iteration % self._validation_iteration == 0:
                    self._compute_validation_metrics()

            if self._current_iteration % self._checkpoint_iteration == 0:
                self._save_checkpoint()

            if self._current_iteration >= self._iterations:
                break

        now = time.time()
        print(f"Training finished after {now-self._start_time} seconds.")

        # save the final model
        torch.save(
            self._sdf_pose_net.state_dict(),
            os.path.join(wandb.run.dir, f"{wandb.run.name}.pt"),
        )
        config_path = os.path.join(wandb.run.dir, f"{wandb.run.name}.yaml")
        self._config["model"] = os.path.join(".", f"{wandb.run.name}.pt")
        yoco.save_config_to_file(config_path, self._config)

    def get_device(self) -> torch.device:
        """Create device based on config."""
        if "device" not in self._config or self._config["device"] is None:
            return torch.device("cuda" if torch.cuda.is_available() else "cpu")
        return torch.device(self._config["device"])

    def create_sdfvae(self) -> SDFVAE:
        """Create SDFVAE based on config.

        Returns:
            The SDFVAE on the specified device, with weights from specified model.
        """
        model_url = self._config["vae"].get("model_url")
        device = self.get_device()
        vae = SDFVAE(
            sdf_size=64,
            latent_size=self._config["vae"]["latent_size"],
            encoder_dict=self._config["vae"]["encoder"],
            decoder_dict=self._config["vae"]["decoder"],
            device=device,
        ).to(device)
        load_model_weights(self._config["vae"]["model"], vae, device, model_url)
        vae.eval()
        return vae

    def _compute_loss(
        self,
        samples: dict,
        predictions: dict,
    ) -> torch.Tensor:
        """Compute total loss.

        Args:
            samples:
                Samples dictionary containing a subset of the following keys:
                    "latent_shape": Shape (N,D).
                    "position": Shape (N,3).
                    "scale": Shape (N,).
                    "orientation":
                        Shape (N,4) for quaternion representation.
                        Shape (N,) for discretized representation.
            predictions: Dictionary containing the following keys:
                "latent_shape": Shape (N,D).
                "position": Shape (N,3).
                "scale": Shape (N,).
                "orientation":
                    Shape (N,4) for quaternion representation.
                    Shape (N,R) for discretized representation.
        Returns:
            The combined loss. Scalar.
        """
        log_dict = {}

        loss = 0

        if "latent_shape" in samples:
            loss_latent_l2 = torch.nn.functional.mse_loss(
                predictions["latent_shape"], samples["latent_shape"]
            )
            log_dict["loss latent"] = loss_latent_l2.item()
            loss = loss + self._config["latent_weight"] * loss_latent_l2

        if "position" in samples:
            loss_position_l2 = torch.nn.functional.mse_loss(
                predictions["position"], samples["position"]
            )
            log_dict["loss position"] = loss_position_l2.item()
            loss = loss + self._config["position_weight"] * loss_position_l2

        if "scale" in samples:
            loss_scale_l2 = torch.nn.functional.mse_loss(
                predictions["scale"], samples["scale"]
            )
            log_dict["loss scale"] = loss_scale_l2.item()
            loss = loss + self._config["scale_weight"] * loss_scale_l2

        if "orientation" in samples:
            if self._config["head"]["orientation_repr"] == "quaternion":
                loss_orientation = quaternion_utils.simple_quaternion_loss(
                    predictions["orientation"], samples["orientation"]
                )
            elif self._config["head"]["orientation_repr"] == "discretized":
                loss_orientation = torch.nn.functional.cross_entropy(
                    predictions["orientation"], samples["orientation"]
                )
            else:
                raise NotImplementedError(
                    "Orientation repr "
                    f"{self._config['head']['orientation_repr']}"
                    " not supported."
                )
            log_dict["loss orientation"] = loss_orientation.item()
            loss = loss + self._config["orientation_weight"] * loss_orientation

        log_dict["total loss"] = loss.item()

        wandb.log(
            log_dict,
            step=self._current_iteration,
        )

        return loss

    def _create_multi_data_loader(self) -> dataset_utils.MultiDataLoader:
        data_loaders = []
        probabilities = []
        for dataset_dict in self._config["datasets"].values():
            if dataset_dict["probability"] == 0.0:
                continue
            dataset = self._create_dataset(
                dataset_dict["type"], dataset_dict["config_dict"]
            )
            num_workers = 8 if dataset_dict["type"] != "SDFVAEViewDataset" else 0
            shuffle = (
                False if isinstance(dataset, torch.utils.data.IterableDataset) else True
            )
            probabilities.append(dataset_dict["probability"])
            data_loader = torch.utils.data.DataLoader(
                dataset=dataset,
                batch_size=self._config["batch_size"],
                collate_fn=dataset_utils.collate_samples,
                drop_last=True,
                shuffle=shuffle,
                num_workers=num_workers,
            )
            data_loaders.append(data_loader)
        return dataset_utils.MultiDataLoader(data_loaders, probabilities)

    def _create_validation_data_loader_dict(self) -> dict:
        data_loader_dict = {}
        for dataset_name, dataset_dict in self._config["validation_datasets"].items():
            dataset = self._create_dataset(
                dataset_dict["type"], dataset_dict["config_dict"]
            )
            data_loader = torch.utils.data.DataLoader(
                dataset=dataset,
                batch_size=self._config["batch_size"],
                collate_fn=dataset_utils.collate_samples,
                num_workers=12,
            )
            data_loader_dict[dataset_name] = data_loader
        return data_loader_dict

    def _create_dataset(
        self, type_str: str, config_dict: dict
    ) -> torch.utils.data.Dataset:
        dataset_type = utils.str_to_object(type_str)
        if dataset_type == SDFVAEViewDataset:
            dataset = dataset_type(
                config=config_dict,
                vae=self.vae,
            )
        elif dataset_type is not None:
            dataset = dataset_type(config=config_dict)
        else:
            raise NotImplementedError(f"Dataset type {type_str} not supported.")
        return dataset

    def _mean_geodesic_distance(self, samples: dict, predictions: dict) -> torch.Tensor:
        target_quaternions = samples["quaternion"]
        if self._config["head"]["orientation_repr"] == "quaternion":
            predicted_quaternions = predictions["orientation"]
        elif self._config["head"]["orientation_repr"] == "discretized":
            predicted_quaternions = torch.empty_like(target_quaternions)
            for i, v in enumerate(predictions["orientation"]):
                index = v.argmax().item()
                quat = self._sdf_pose_net._head._grid.index_to_quat(index)
                predicted_quaternions[i, :] = torch.tensor(quat)
        else:
            raise NotImplementedError(
                "Orientation representation "
                f"{self._config['head']['orientation_repr']}"
                " is not supported"
            )
        geodesic_distances = quaternion_utils.geodesic_distance(
            target_quaternions, predicted_quaternions
        )
        return torch.mean(geodesic_distances)

    def _compute_metrics(self, samples: dict, predictions: dict) -> None:
        # compute metrics / i.e., loss and representation independent metrics
        # extract quaternion from orientation representation
        geodesic_distance = self._mean_geodesic_distance(samples, predictions)
        wandb.log(
            {
                "metric geodesic distance": geodesic_distance.item(),
            },
            step=self._current_iteration,
        )

    def _generate_visualizations(self) -> None:
        # generate visualizations
        if self._current_iteration % self._visualization_iteration == 0:
            # generate unseen input and target
            samples = next(iter(self._multi_data_loader))
            samples = utils.dict_to(samples, self._device)
            predictions = self._sdf_pose_net(samples["pointset"])
            input_pointcloud = samples["pointset"][0].detach().cpu().numpy()
            input_pointcloud = np.hstack(
                (
                    input_pointcloud,
                    np.full((input_pointcloud.shape[0], 1), 0),
                )
            )
            output_sdfs = self.vae.decode(predictions[0])
            output_sdf = output_sdfs[0][0].detach().cpu().numpy()
            output_position = predictions[1][0].detach().cpu().numpy()
            output_scale = predictions[2][0].detach().cpu().numpy()
            if self._config["head"]["orientation_repr"] == "quaternion":
                output_quaternion = predictions[3][0].detach().cpu().numpy()
            elif self._config["head"]["orientation_repr"] == "discretized":
                index = predictions[3][0].argmax().item()
                output_quaternion = self._sdf_pose_net._head._grid.index_to_quat(index)
            else:
                raise NotImplementedError(
                    "Orientation representation "
                    f"{self._config['head']['orientation_repr']}"
                    " is not supported"
                )
            output_pointcloud = sdf_utils.sdf_to_pointcloud(
                output_sdf, output_position, output_quaternion, output_scale
            )
            output_pointcloud = np.hstack(
                (
                    output_pointcloud,
                    np.full((output_pointcloud.shape[0], 1), 1),
                )
            )
            pointcloud = np.vstack((input_pointcloud, output_pointcloud))

            wandb.log(
                {"point_cloud": wandb.Object3D(pointcloud)},
                step=self._current_iteration,
            )

            output_pointcloud = sdf_utils.sdf_to_pointcloud(
                output_sdf,
                samples["position"][0].detach().cpu().numpy(),
                samples["quaternion"][0].detach().cpu().numpy(),
                samples["scale"][0].detach().cpu().numpy(),
            )
            output_pointcloud = np.hstack(
                (
                    output_pointcloud,
                    np.full((output_pointcloud.shape[0], 1), 1),
                )
            )
            pointcloud = np.vstack((input_pointcloud, output_pointcloud))
            wandb.log(
                {"point_cloud gt pose": wandb.Object3D(pointcloud)},
                step=self._current_iteration,
            )

    def _compute_validation_metrics(self) -> None:
        self._sdf_pose_net.eval()
        for name, data_loader in self._validation_data_loader_dict.items():
            metrics_dict = defaultdict(lambda: 0)
            sample_count = 0
            for samples in tqdm(data_loader, desc="Validation"):
                batch_size = samples["position"].shape[0]
                samples = utils.dict_to(samples, self._device)
                latent_shape, position, scale, orientation = self._sdf_pose_net(
                    samples["pointset"]
                )
                predictions = {
                    "latent_shape": latent_shape,
                    "position": position,
                    "scale": scale,
                    "orientation": orientation,
                }
                euclidean_distance = torch.linalg.norm(
                    predictions["position"] - samples["position"], dim=1
                )
                metrics_dict[f"{name} validation mean position error / m"] += torch.sum(
                    euclidean_distance
                ).item()
                metrics_dict[f"{name} validation mean scale error / m"] += torch.sum(
                    torch.abs(predictions["scale"] - samples["scale"])
                ).item()
                metrics_dict[f"{name} validation mean geodesic_distance / rad"] += (
                    self._mean_geodesic_distance(samples, predictions).item()
                    * batch_size
                )
                sample_count += batch_size
                if self._config["head"]["orientation_repr"] == "discretized":
                    metrics_dict[
                        f"{name} validation orientation mean NLL"
                    ] += torch.nn.functional.cross_entropy(
                        predictions["orientation"],
                        samples["orientation"],
                        reduction="sum",
                    )
            for metric_name in metrics_dict:
                metrics_dict[metric_name] /= sample_count
            wandb.log(metrics_dict, step=self._current_iteration)
        self._sdf_pose_net.train()

    def _save_checkpoint(self) -> None:
        checkpoint_path = os.path.join(
            self._model_base_path, f"{self._current_iteration}.pt"
        )
        torch.save(
            self._sdf_pose_net.state_dict(),
            checkpoint_path,
        )

    def _update_progress(self) -> None:
        current_time = time.time()
        duration = current_time - self._start_time
        iterations_per_sec = self._current_iteration / duration
        if self._current_iteration > 10:
            remaining_iterations = self._iterations - self._current_iteration
            remaining_secs = remaining_iterations / iterations_per_sec
            remaining_time_str = str(timedelta(seconds=round(remaining_secs)))
        else:
            remaining_time_str = "N/A"
        print(
            f"Current iteration: {self._current_iteration:>10} / {self._iterations}"
            f" {self._current_iteration / self._iterations * 100:>6.2f}%"
            f" Remaining time: {remaining_time_str}"  # remaining time
            "\033[K",  # clear until end of line
            end="\r",  # overwrite previous
        )

__init__(config)

Construct trainer.

Parameters:

Name Type Description Default
config dict

The configuration for model and training.

required
Source code in sdfest/initialization/scripts/train.py
33
34
35
36
37
38
39
def __init__(self, config: dict) -> None:
    """Construct trainer.

    Args:
        config: The configuration for model and training.
    """
    self._read_config(config)

create_sdfvae()

Create SDFVAE based on config.

Returns:

Type Description
SDFVAE

The SDFVAE on the specified device, with weights from specified model.

Source code in sdfest/initialization/scripts/train.py
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
def create_sdfvae(self) -> SDFVAE:
    """Create SDFVAE based on config.

    Returns:
        The SDFVAE on the specified device, with weights from specified model.
    """
    model_url = self._config["vae"].get("model_url")
    device = self.get_device()
    vae = SDFVAE(
        sdf_size=64,
        latent_size=self._config["vae"]["latent_size"],
        encoder_dict=self._config["vae"]["encoder"],
        decoder_dict=self._config["vae"]["decoder"],
        device=device,
    ).to(device)
    load_model_weights(self._config["vae"]["model"], vae, device, model_url)
    vae.eval()
    return vae

get_device()

Create device based on config.

Source code in sdfest/initialization/scripts/train.py
186
187
188
189
190
def get_device(self) -> torch.device:
    """Create device based on config."""
    if "device" not in self._config or self._config["device"] is None:
        return torch.device("cuda" if torch.cuda.is_available() else "cpu")
    return torch.device(self._config["device"])

run()

Train the model.

Source code in sdfest/initialization/scripts/train.py
 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
def run(self) -> None:
    """Train the model."""
    wandb.init(project="sdfest.initialization", config=self._config)

    self._device = self.get_device()

    # init dataset and dataloader
    self.vae = self.create_sdfvae()

    # init model to train
    self._sdf_pose_net = SDFPoseNet(
        backbone=MODULE_DICT[self._config["backbone_type"]](
            **self._config["backbone"]
        ),
        head=MODULE_DICT[self._config["head_type"]](
            shape_dimension=self._config["vae"]["latent_size"],
            **self._config["head"],
        ),
    ).to(self._device)
    self._sdf_pose_net.train()

    # deterministic samples (needs to be done after model initialization, as it
    # can have varying number of parameters)
    torch.manual_seed(0)
    random.seed(torch.initial_seed())  # to get deterministic examples

    # print network summary
    torchinfo.summary(self._sdf_pose_net, (1, 500, 3), device=self._device)

    # init optimizer
    self._optimizer = torch.optim.Adam(
        self._sdf_pose_net.parameters(), lr=self._config["learning_rate"]
    )

    # load weights if provided
    if self._init_weights_path is not None:
        state_dict = torch.load(self._init_weights_path, map_location=self._device)
        self._sdf_pose_net.load_state_dict(state_dict)

    self._current_iteration = 0
    self._run_name = (
        f"sdfest.initialization_{datetime.now().strftime('%Y-%m-%d_%H-%M-%S-%f')}"
    )

    wandb.config.run_name = (
        self._run_name  # to allow association of pt files with wandb runs
    )

    self._model_base_path = os.path.join(os.getcwd(), "models", self._run_name)

    self._multi_data_loader = self._create_multi_data_loader()
    self._validation_data_loader_dict = self._create_validation_data_loader_dict()

    # backup config to model directory
    os.makedirs(self._model_base_path, exist_ok=True)
    config_path = os.path.join(self._model_base_path, "config.yaml")
    yoco.save_config_to_file(config_path, self._config)

    self._start_time = time.time()
    for samples in self._multi_data_loader:
        self._current_iteration += 1
        print(f"Current iteration: {self._current_iteration}\033[K", end="\r")
        self._update_progress()

        samples = utils.dict_to(samples, self._device)

        latent_shape, position, scale, orientation = self._sdf_pose_net(
            samples["pointset"]
        )
        predictions = {
            "latent_shape": latent_shape,
            "position": position,
            "scale": scale,
            "orientation": orientation,
        }

        loss = self._compute_loss(samples, predictions)
        self._optimizer.zero_grad()
        loss.backward()
        self._optimizer.step()

        with torch.no_grad():
            # samples_dict = defaultdict(lambda: dict())
            # for k,vs in samples.items():
            #     for i, v in enumerate(vs):
            #         samples_dict[i][k] = v
            # for sample in samples_dict.values():
            #     utils.visualize_sample(sample, None)

            self._compute_metrics(samples, predictions)

            if self._current_iteration % self._visualization_iteration == 0:
                self._generate_visualizations()

            if self._current_iteration % self._validation_iteration == 0:
                self._compute_validation_metrics()

        if self._current_iteration % self._checkpoint_iteration == 0:
            self._save_checkpoint()

        if self._current_iteration >= self._iterations:
            break

    now = time.time()
    print(f"Training finished after {now-self._start_time} seconds.")

    # save the final model
    torch.save(
        self._sdf_pose_net.state_dict(),
        os.path.join(wandb.run.dir, f"{wandb.run.name}.pt"),
    )
    config_path = os.path.join(wandb.run.dir, f"{wandb.run.name}.yaml")
    self._config["model"] = os.path.join(".", f"{wandb.run.name}.pt")
    yoco.save_config_to_file(config_path, self._config)

main()

Entry point of the program.

Source code in sdfest/initialization/scripts/train.py
511
512
513
514
515
516
517
518
519
520
def main() -> None:
    """Entry point of the program."""
    # define the arguments
    parser = argparse.ArgumentParser(description="Training script for init network.")
    parser.add_argument("--config", default="configs/default.yaml", nargs="+")
    config = yoco.load_config_from_args(
        parser, search_paths=[".", "~/.sdfest/", sdfest.__path__[0]]
    )
    trainer = Trainer(config)
    trainer.run()