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

default to greedy sample AR (i should probably test this more but it seems to pass my harvard sentences and tongue twisters)

Browse Source
This commit is contained in:
mrq 2024-10-18 16:58:56 -05:00
parent fc8dfd8617
commit c8f31db1de
4 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vall_e/__main__.py
View File

@ -21,7 +21,7 @@ def main():
parser.add_argument("--max-ar-steps", type=int, default=12 * cfg.dataset.frames_per_second)
parser.add_argument("--max-nar-levels", type=int, default=7)
parser.add_argument("--ar-temp", type=float, default=1.0)
parser.add_argument("--ar-temp", type=float, default=0.0)
parser.add_argument("--nar-temp", type=float, default=0.01)
parser.add_argument("--min-ar-temp", type=float, default=-1.0)
parser.add_argument("--min-nar-temp", type=float, default=-1.0)

2
vall_e/config.py
View File

@ -429,7 +429,7 @@ class Evaluation:
size: int = 64 # number of samples to generate during eval / val
steps: int = 500
ar_temperature: float = 1.0 # AR temp for inferencing
ar_temperature: float = 0.0 # AR temp for inferencing
nar_temperature: float = 0.0 # NAR temp for inferencing
nar_levels: int = 0 # maximum NAR levels to use for inferencing

4
vall_e/demo.py
View File

@ -58,7 +58,7 @@ def main():
parser.add_argument("--max-ar-steps", type=int, default=12 * cfg.dataset.frames_per_second)
parser.add_argument("--max-nar-levels", type=int, default=7)
parser.add_argument("--ar-temp", type=float, default=1.0)
parser.add_argument("--ar-temp", type=float, default=0.0)
parser.add_argument("--nar-temp", type=float, default=0.0)
parser.add_argument("--min-ar-temp", type=float, default=-1.0)
parser.add_argument("--min-nar-temp", type=float, default=-1.0)
@ -155,7 +155,7 @@ def main():
elif args.comparison == "dtype":
current_dtype = cfg.inference.weight_dtype
other_dtype = "float32"
if current_dtype == "float16":
other_dtype = "bfloat16"
elif current_dtype == "bfloat16":

8
vall_e/inference.py
View File

@ -185,11 +185,11 @@ class TTS():
input_prompt_length=0.0,
input_prompt_prefix=False,
#
ar_temp=0.95,
nar_temp=0.5,
ar_temp=0.0,
nar_temp=0.0,
#
min_ar_temp=0.95,
min_nar_temp=0.5,
min_ar_temp=0.0,
min_nar_temp=0.0,
#
top_p=1.0,
top_k=0,