ecker/vall-e
5
9
Fork 0
Code Issues 8 Pull Requests Packages Projects Releases Wiki Activity

tweaks

Browse Source
This commit is contained in:
mrq 2025-03-28 19:49:54 -05:00
parent 6ae282e090
commit 478aea0e8c
3 changed files with 13 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
vall_e/inference.py
View File

@ -520,6 +520,9 @@ class TTS():
else: else:
len_list = model_len( **input_kwargs, task_list=["len"], **{"max_duration": 5} ) # "max_duration" is max tokens len_list = model_len( **input_kwargs, task_list=["len"], **{"max_duration": 5} ) # "max_duration" is max tokens
# clamp
len_list = [ max( l, 1 * cfg.dataset.frames_per_second ) for l in len_list ]
# add an additional X seconds # add an additional X seconds
len_list = [ int(l * duration_padding) for l in len_list ] len_list = [ int(l * duration_padding) for l in len_list ]

11
vall_e/models/arch/llama.py
View File

@ -239,7 +239,7 @@ class Attention(nn.Module):
"default", "default",
torch.nn.attention.SDPBackend.MATH, torch.nn.attention.SDPBackend.MATH,
torch.nn.attention.SDPBackend.EFFICIENT_ATTENTION, torch.nn.attention.SDPBackend.EFFICIENT_ATTENTION,
torch.nn.attention.SDPBackend.FLASH_ATTENTION, #torch.nn.attention.SDPBackend.FLASH_ATTENTION,
torch.nn.attention.SDPBackend.CUDNN_ATTENTION torch.nn.attention.SDPBackend.CUDNN_ATTENTION
] ]
@ -365,6 +365,7 @@ class Attention(nn.Module):
# pain # pain
# SDPA backends only sometimes allowing/disallowing some arguments... # SDPA backends only sometimes allowing/disallowing some arguments...
"""
if isinstance( is_causal, list ): if isinstance( is_causal, list ):
count = sum( [ 1 if x else 0 for x in is_causal ] ) count = sum( [ 1 if x else 0 for x in is_causal ] )
if count == 0: if count == 0:
@ -378,6 +379,12 @@ class Attention(nn.Module):
x_mask = None x_mask = None
elif is_causal == True: elif is_causal == True:
x_mask = None x_mask = None
"""
if self.attn_mode in [torch.nn.attention.SDPBackend.FLASH_ATTENTION]:
x_mask = None
if x_mask is not None:
is_causal = False
# SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask, # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
# Reference: https://github.com/pytorch/pytorch/issues/112577. # Reference: https://github.com/pytorch/pytorch/issues/112577.
@ -519,10 +526,8 @@ class DecoderLayer(nn.Module):
hidden_states = self.input_layernorm(hidden_states) hidden_states = self.input_layernorm(hidden_states)
# ugh # ugh
"""
if isinstance( is_causal, list ) and len(is_causal) == 1: if isinstance( is_causal, list ) and len(is_causal) == 1:
is_causal = is_causal[0] is_causal = is_causal[0]
"""
# Self Attention # Self Attention
if self.config.attn_mode == "sparse": if self.config.attn_mode == "sparse":

3
vall_e/models/base_v2.py
View File

@ -765,7 +765,8 @@ class Base_V2(nn.Module):
# needed, cringe # needed, cringe
if task_type == "len": if task_type == "len":
batch[-1] = torch.cat( [ batch[-1], self.sep[None], self.sep[None] ] ) #batch[-1] = torch.cat( [ batch[-1], self.sep[None], self.sep[None] ] )
batch[-1] = torch.cat( [ batch[-1], self.sep[None] ] )
x_list.append( _join( batch, self.sep ) ) x_list.append( _join( batch, self.sep ) )