mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

remove eta from mdf

Browse Source
This commit is contained in:
James Betker 2022-07-24 17:21:20 -06:00
parent 76464ca063
commit f3d967dbf5
1 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
codes/trainer/eval/music_diffusion_fid.py
View File

@ -138,7 +138,7 @@ class MusicDiffusionFid(evaluator.Evaluator):
# x = x.clamp(-s, s) / s
# return x
sampler = self.diffuser.ddim_sample_loop if self.ddim else self.diffuser.p_sample_loop
gen_mel = sampler(self.model, mel_norm.shape, model_kwargs={'truth_mel': mel_norm}, eta=.8)
gen_mel = sampler(self.model, mel_norm.shape, model_kwargs={'truth_mel': mel_norm})
gen_mel_denorm = denormalize_torch_mel(gen_mel)
output_shape = (1,16,audio.shape[-1]//16)
@ -314,14 +314,15 @@ class MusicDiffusionFid(evaluator.Evaluator):
if __name__ == '__main__':
"""
# For multilevel SR:
diffusion = load_model_from_config('X:\\dlas\\experiments\\train_music_diffusion_multilevel_sr.yml', 'generator',
also_load_savepoint=False, strict_load=False,
load_path='X:\\dlas\\experiments\\train_music_diffusion_multilevel_sr\\models\\6000_generator.pth'
load_path='X:\\dlas\\experiments\\train_music_diffusion_multilevel_sr\\models\\56000_generator.pth'
).cuda()
opt_eval = {'path': 'Y:\\split\\yt-music-eval', # eval music, mostly electronica. :)
#'path': 'E:\\music_eval', # this is music from the training dataset, including a lot more variety.
'diffusion_steps': 128, # basis: 192
'diffusion_steps': 256, # basis: 192
'conditioning_free': True, 'conditioning_free_k': 1, 'use_ddim': False, 'clip_audio': True,
'diffusion_schedule': 'cosine', 'diffusion_type': 'chained_sr',
}
@ -331,17 +332,16 @@ if __name__ == '__main__':
# For TFD+cheater trainer
diffusion = load_model_from_config('X:\\dlas\\experiments\\train_music_diffusion_tfd_and_cheater.yml', 'generator',
also_load_savepoint=False, strict_load=False,
load_path='X:\\dlas\\experiments\\train_music_diffusion_tfd14_and_cheater_g2\\models\\1000_generator.pth'
load_path='X:\\dlas\\experiments\\train_music_diffusion_tfd14_and_cheater_g2\\models\\56000_generator_ema.pth'
).cuda()
opt_eval = {'path': 'Y:\\split\\yt-music-eval', # eval music, mostly electronica. :)
#'path': 'E:\\music_eval', # this is music from the training dataset, including a lot more variety.
'diffusion_steps': 128, # basis: 192
'conditioning_free': True, 'conditioning_free_k': 1, 'use_ddim': True, 'clip_audio': True,
'diffusion_schedule': 'linear', 'diffusion_type': 'from_codes_quant',
'diffusion_steps': 256, # basis: 192
'conditioning_free': True, 'conditioning_free_k': 1, 'use_ddim': False, 'clip_audio': True,
'diffusion_schedule': 'cosine', 'diffusion_type': 'from_codes_quant',
}
"""
env = {'rank': 0, 'base_path': 'D:\\tmp\\test_eval_music', 'step': 7, 'device': 'cuda', 'opt': {}}
env = {'rank': 0, 'base_path': 'D:\\tmp\\test_eval_music', 'step': 10, 'device': 'cuda', 'opt': {}}
eval = MusicDiffusionFid(diffusion, opt_eval, env)
fds = []
for i in range(2):