fix(train): parameter issue

Author: fumiama

gui.py

@@ -144,8 +144,6 @@ def __init__(self) -> None:
             self.input_devices_indices = None
             self.output_devices_indices = None
             self.stream = None
-            if not self.config.nocheck:
-                self.check_assets()
             self.update_devices()
             self.launcher()
 

infer/lib/train/utils.py

@@ -4,6 +4,7 @@
 import logging
 import os
 import sys
+from copy import deepcopy
 
 import codecs
 import numpy as np
@@ -444,6 +445,9 @@ def items(self):
 
     def values(self):
         return self.__dict__.values()
+    
+    def copy(self):
+        return deepcopy(self)
 
     def __len__(self):
         return len(self.__dict__)

infer/modules/train/train.py

@@ -46,7 +46,6 @@
 from torch.utils.data import DataLoader
 from torch.utils.tensorboard import SummaryWriter
 
-from rvc.layers import utils
 from infer.lib.train.data_utils import (
     DistributedBucketSampler,
     TextAudioCollate,
@@ -77,6 +76,11 @@
 from infer.lib.train.mel_processing import mel_spectrogram_torch, spec_to_mel_torch
 from infer.lib.train.process_ckpt import save_small_model
 
+from rvc.layers.utils import (
+    slice_on_last_dim,
+    total_grad_norm,
+)
+
 global_step = 0
 
 
@@ -118,7 +122,7 @@ def main():
         children[i].join()
 
 
-def run(rank, n_gpus, hps, logger: logging.Logger):
+def run(rank, n_gpus, hps: utils.HParams, logger: logging.Logger):
     global global_step
     if rank == 0:
         # logger = utils.get_logger(hps.model_dir)
@@ -163,20 +167,20 @@ def run(rank, n_gpus, hps, logger: logging.Logger):
         persistent_workers=True,
         prefetch_factor=8,
     )
+    mdl = hps.copy().model
+    del mdl.use_spectral_norm
     if hps.if_f0 == 1:
         net_g = RVC_Model_f0(
             hps.data.filter_length // 2 + 1,
             hps.train.segment_size // hps.data.hop_length,
-            **hps.model,
-            is_half=hps.train.fp16_run,
+            **mdl,
             sr=hps.sample_rate,
         )
     else:
         net_g = RVC_Model_nof0(
             hps.data.filter_length // 2 + 1,
             hps.train.segment_size // hps.data.hop_length,
-            **hps.model,
-            is_half=hps.train.fp16_run,
+            **mdl,
         )
     if torch.cuda.is_available():
         net_g = net_g.cuda(rank)
@@ -459,7 +463,7 @@ def train_and_evaluate(
                 hps.data.mel_fmin,
                 hps.data.mel_fmax,
             )
-            y_mel = utils.slice_on_last_dim(
+            y_mel = slice_on_last_dim(
                 mel, ids_slice, hps.train.segment_size // hps.data.hop_length
             )
             with autocast(enabled=False):
@@ -475,7 +479,7 @@ def train_and_evaluate(
                 )
             if hps.train.fp16_run == True:
                 y_hat_mel = y_hat_mel.half()
-            wave = utils.slice_on_last_dim(
+            wave = slice_on_last_dim(
                 wave, ids_slice * hps.data.hop_length, hps.train.segment_size
             )  # slice
 
@@ -488,7 +492,7 @@ def train_and_evaluate(
         optim_d.zero_grad()
         scaler.scale(loss_disc).backward()
         scaler.unscale_(optim_d)
-        grad_norm_d = utils.total_grad_norm(net_d.parameters())
+        grad_norm_d = total_grad_norm(net_d.parameters())
         scaler.step(optim_d)
 
         with autocast(enabled=hps.train.fp16_run):
@@ -503,7 +507,7 @@ def train_and_evaluate(
         optim_g.zero_grad()
         scaler.scale(loss_gen_all).backward()
         scaler.unscale_(optim_g)
-        grad_norm_g = utils.total_grad_norm(net_g.parameters())
+        grad_norm_g = total_grad_norm(net_g.parameters())
         scaler.step(optim_g)
         scaler.update()
 

web.py

@@ -24,7 +24,6 @@
 import numpy as np
 import gradio as gr
 import faiss
-import fairseq
 import pathlib
 import json
 from time import sleep
@@ -72,7 +71,9 @@ def forward_dml(ctx, x, scale):
         res = x.clone().detach()
         return res
 
+    import fairseq
     fairseq.modules.grad_multiply.GradMultiply.forward = forward_dml
+
 i18n = I18nAuto()
 logger.info(i18n)
 # 判断是否有能用来训练和加速推理的N卡