Remove remaned trainer functions

TTS/utils/io.py

@@ -84,116 +84,4 @@ def load_checkpoint(model, checkpoint_path, use_cuda=False, eval=False):  # pyli
         model.cuda()
     if eval:
         model.eval()
-    return model, state
-
-
-def save_fsspec(state: Any, path: str, **kwargs):
-    """Like torch.save but can save to other locations (e.g. s3:// , gs://).
-
-    Args:
-        state: State object to save
-        path: Any path or url supported by fsspec.
-        **kwargs: Keyword arguments forwarded to torch.save.
-    """
-    with fsspec.open(path, "wb") as f:
-        torch.save(state, f, **kwargs)
-
-
-def save_model(config, model, optimizer, scaler, current_step, epoch, output_path, **kwargs):
-    if hasattr(model, "module"):
-        model_state = model.module.state_dict()
-    else:
-        model_state = model.state_dict()
-    if isinstance(optimizer, list):
-        optimizer_state = [optim.state_dict() for optim in optimizer]
-    else:
-        optimizer_state = optimizer.state_dict() if optimizer is not None else None
-
-    if isinstance(scaler, list):
-        scaler_state = [s.state_dict() for s in scaler]
-    else:
-        scaler_state = scaler.state_dict() if scaler is not None else None
-
-    if isinstance(config, Coqpit):
-        config = config.to_dict()
-
-    state = {
-        "config": config,
-        "model": model_state,
-        "optimizer": optimizer_state,
-        "scaler": scaler_state,
-        "step": current_step,
-        "epoch": epoch,
-        "date": datetime.date.today().strftime("%B %d, %Y"),
-    }
-    state.update(kwargs)
-    save_fsspec(state, output_path)
-
-
-def save_checkpoint(
-    config,
-    model,
-    optimizer,
-    scaler,
-    current_step,
-    epoch,
-    output_folder,
-    **kwargs,
-):
-    file_name = "checkpoint_{}.pth".format(current_step)
-    checkpoint_path = os.path.join(output_folder, file_name)
-    print("\n > CHECKPOINT : {}".format(checkpoint_path))
-    save_model(
-        config,
-        model,
-        optimizer,
-        scaler,
-        current_step,
-        epoch,
-        checkpoint_path,
-        **kwargs,
-    )
-
-
-def save_best_model(
-    current_loss,
-    best_loss,
-    config,
-    model,
-    optimizer,
-    scaler,
-    current_step,
-    epoch,
-    out_path,
-    keep_all_best=False,
-    keep_after=10000,
-    **kwargs,
-):
-    if current_loss < best_loss:
-        best_model_name = f"best_model_{current_step}.pth"
-        checkpoint_path = os.path.join(out_path, best_model_name)
-        print(" > BEST MODEL : {}".format(checkpoint_path))
-        save_model(
-            config,
-            model,
-            optimizer,
-            scaler,
-            current_step,
-            epoch,
-            checkpoint_path,
-            model_loss=current_loss,
-            **kwargs,
-        )
-        fs = fsspec.get_mapper(out_path).fs
-        # only delete previous if current is saved successfully
-        if not keep_all_best or (current_step < keep_after):
-            model_names = fs.glob(os.path.join(out_path, "best_model*.pth"))
-            for model_name in model_names:
-                if os.path.basename(model_name) != best_model_name:
-                    fs.rm(model_name)
-        # create a shortcut which always points to the currently best model
-        shortcut_name = "best_model.pth"
-        shortcut_path = os.path.join(out_path, shortcut_name)
-        fs.copy(checkpoint_path, shortcut_path)
-        best_loss = current_loss
-    return best_loss
+    return model, state