fix(train): set persistent_workers = True in DataLoader for performan…

…ce, do not save checkpoints, fix logging issue and multiple warning issues, do not do validation when global_step == 0 (#384)

src/so_vits_svc_fork/__main__.py

@@ -19,8 +19,6 @@
 if IS_TEST:
     LOG.debug("Test mode is on.")
 
-LOG.info(f"Version: {__version__}")
-
 
 class RichHelpFormatter(click.HelpFormatter):
     def __init__(
@@ -31,6 +29,7 @@ def __init__(
     ) -> None:
         width = 100
         super().__init__(indent_increment, width, max_width)
+        LOG.info(f"Version: {__version__}")
 
 
 def patch_wrap_text():

src/so_vits_svc_fork/gui.py

@@ -13,6 +13,7 @@
 from pebble import ProcessFuture, ProcessPool
 from tqdm.tk import tqdm_tk
 
+from . import __version__
 from .utils import ensure_pretrained_model, get_optimal_device
 
 GUI_DEFAULT_PRESETS_PATH = Path(__file__).parent / "default_gui_presets.json"
@@ -97,6 +98,7 @@ def after_inference(window: sg.Window, path: Path, auto_play: bool, output_path:
 
 
 def main():
+    LOG.info(f"version: {__version__}")
     try:
         ensure_pretrained_model(".", "contentvec", tqdm_cls=tqdm_tk)
     except Exception as e:

src/so_vits_svc_fork/logger.py

@@ -1,6 +1,5 @@
 import os
 import sys
-import warnings
 from logging import (
     DEBUG,
     INFO,
@@ -36,9 +35,6 @@ def init_logger() -> None:
     if IS_TEST:
         getLogger(package_name).setLevel(DEBUG)
     captureWarnings(True)
-    warnings.filterwarnings(
-        "ignore", category=UserWarning, message="TypedStorage is deprecated"
-    )
     LOGGER_INIT = True
 
 

src/so_vits_svc_fork/train.py

@@ -29,7 +29,6 @@
 from .modules.synthesizers import SynthesizerTrn
 
 LOG = getLogger(__name__)
-torch.backends.cudnn.benchmark = True
 torch.set_float32_matmul_precision("high")
 
 
@@ -51,16 +50,15 @@ def __init__(self, hparams: Any):
     def train_dataloader(self):
         return DataLoader(
             self.train_dataset,
-            # pin_memory=False,
-            num_workers=min(cpu_count(), self.__hparams.train.get("num_workers", 4)),
+            num_workers=min(cpu_count(), self.__hparams.train.get("num_workers", 8)),
             batch_size=self.batch_size,
             collate_fn=self.collate_fn,
+            persistent_workers=True,
         )
 
     def val_dataloader(self):
         return DataLoader(
             self.val_dataset,
-            # pin_memory=False,
             batch_size=1,
             collate_fn=self.collate_fn,
         )
@@ -95,6 +93,8 @@ def train(
         else 32,
         strategy=strategy,
         callbacks=[pl.callbacks.RichProgressBar()] if not is_notebook() else None,
+        benchmark=True,
+        enable_checkpointing=False,
     )
     tuner = Tuner(trainer)
     model = VitsLightning(reset_optimizer=reset_optimizer, **hparams)
@@ -482,6 +482,9 @@ def training_step(self, batch: dict[str, torch.Tensor], batch_idx: int) -> None:
             self.scheduler_d.step()
 
     def validation_step(self, batch, batch_idx):
+        # avoid logging with wrong global step
+        if self.global_step == 0:
+            return
         with torch.no_grad():
             self.net_g.eval()
             c, f0, _, mel, y, g, _, uv = batch

src/so_vits_svc_fork/utils.py

@@ -191,7 +191,7 @@ def _substitute_if_same_shape(to_: dict[str, Any], from_: dict[str, Any]) -> Non
     shape_missmatch = []
     for k, v in from_.items():
         if k not in to_:
-            warnings.warn(f"Key {k} not found in model state dict")
+            pass
         elif hasattr(v, "shape"):
             if not hasattr(to_[k], "shape"):
                 raise ValueError(f"Key {k} is not a tensor")
@@ -225,7 +225,11 @@ def load_checkpoint(
     if not Path(checkpoint_path).is_file():
         raise FileNotFoundError(f"File {checkpoint_path} not found")
     with Path(checkpoint_path).open("rb") as f:
-        checkpoint_dict = torch.load(f, map_location="cpu", weights_only=True)
+        with warnings.catch_warnings():
+            warnings.filterwarnings(
+                "ignore", category=UserWarning, message="TypedStorage is deprecated"
+            )
+            checkpoint_dict = torch.load(f, map_location="cpu", weights_only=True)
     iteration = checkpoint_dict["iteration"]
     learning_rate = checkpoint_dict["learning_rate"]