blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-08-24 05:29:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

Remove the monkey patch causing issues when using with other proejcts

Browse Source
This commit is contained in:
Jarod Mica 2025-05-25 00:34:35 -07:00
parent fe531567f1
commit 651b0075ad
1 changed files with 32 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

57
GPT_SoVITS/AR/modules/activation.py
View File

@ -9,11 +9,8 @@ from torch.nn.init import constant_, xavier_normal_, xavier_uniform_
from torch.nn.modules.linear import NonDynamicallyQuantizableLinear from torch.nn.modules.linear import NonDynamicallyQuantizableLinear
from torch.nn.parameter import Parameter from torch.nn.parameter import Parameter
from torch.nn import functional as F
from GPT_SoVITS.AR.modules.patched_mha_with_cache import multi_head_attention_forward_patched from GPT_SoVITS.AR.modules.patched_mha_with_cache import multi_head_attention_forward_patched
F.multi_head_attention_forward = multi_head_attention_forward_patched
class MultiheadAttention(Module): class MultiheadAttention(Module):
r"""Allows the model to jointly attend to information r"""Allows the model to jointly attend to information
@ -361,20 +358,20 @@ class MultiheadAttention(Module):
query, key, value = [x.transpose(1, 0) for x in (query, key, value)] query, key, value = [x.transpose(1, 0) for x in (query, key, value)]
if not self._qkv_same_embed_dim: if not self._qkv_same_embed_dim:
attn_output, attn_output_weights = F.multi_head_attention_forward( attn_output, attn_output_weights = multi_head_attention_forward_patched(
query, query,
key, key,
value, value,
self.embed_dim, embed_dim_to_check=self.embed_dim,
self.num_heads, num_heads=self.num_heads,
self.in_proj_weight, in_proj_weight=self.in_proj_weight,
self.in_proj_bias, in_proj_bias=self.in_proj_bias,
self.bias_k, bias_k=self.bias_k,
self.bias_v, bias_v=self.bias_v,
self.add_zero_attn, add_zero_attn=self.add_zero_attn,
self.dropout, dropout_p=self.dropout,
self.out_proj.weight, out_proj_weight=self.out_proj.weight,
self.out_proj.bias, out_proj_bias=self.out_proj.bias,
training=self.training, training=self.training,
key_padding_mask=key_padding_mask, key_padding_mask=key_padding_mask,
need_weights=need_weights, need_weights=need_weights,
@ -383,29 +380,39 @@ class MultiheadAttention(Module):
q_proj_weight=self.q_proj_weight, q_proj_weight=self.q_proj_weight,
k_proj_weight=self.k_proj_weight, k_proj_weight=self.k_proj_weight,
v_proj_weight=self.v_proj_weight, v_proj_weight=self.v_proj_weight,
static_k=None,
static_v=None,
average_attn_weights=average_attn_weights, average_attn_weights=average_attn_weights,
is_causal=False,
cache=cache, cache=cache,
) )
else: else:
attn_output, attn_output_weights = F.multi_head_attention_forward( attn_output, attn_output_weights = multi_head_attention_forward_patched(
query, query,
key, key,
value, value,
self.embed_dim, embed_dim_to_check=self.embed_dim,
self.num_heads, num_heads=self.num_heads,
self.in_proj_weight, in_proj_weight=self.in_proj_weight,
self.in_proj_bias, in_proj_bias=self.in_proj_bias,
self.bias_k, bias_k=self.bias_k,
self.bias_v, bias_v=self.bias_v,
self.add_zero_attn, add_zero_attn=self.add_zero_attn,
self.dropout, dropout_p=self.dropout,
self.out_proj.weight, out_proj_weight=self.out_proj.weight,
self.out_proj.bias, out_proj_bias=self.out_proj.bias,
training=self.training, training=self.training,
key_padding_mask=key_padding_mask, key_padding_mask=key_padding_mask,
need_weights=need_weights, need_weights=need_weights,
attn_mask=attn_mask, attn_mask=attn_mask,
use_separate_proj_weight=False,
q_proj_weight=self.q_proj_weight,
k_proj_weight=self.k_proj_weight,
v_proj_weight=self.v_proj_weight,
static_k=None,
static_v=None,
average_attn_weights=average_attn_weights, average_attn_weights=average_attn_weights,
is_causal=False,
cache=cache, cache=cache,
) )
if self.batch_first and is_batched: if self.batch_first and is_batched: