deviandice/DL-Art-School
1
0
Fork 0
forked from mrq/DL-Art-School
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix MDF for cosine schedule

Browse Source
This commit is contained in:
James Betker 2022-06-28 17:29:21 -06:00
parent 861aa0e139
commit 43ea259228
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
codes/trainer/eval/music_diffusion_fid.py
View File

@ -79,6 +79,9 @@ class MusicDiffusionFid(evaluator.Evaluator):
self.diffusion_fn = self.perform_reconstruction_from_cheater_gen
self.local_modules['cheater_encoder'] = get_cheater_encoder()
self.local_modules['cheater_decoder'] = get_cheater_decoder()
self.cheater_decoder_diffuser = SpacedDiffusion(use_timesteps=space_timesteps(4000, [32]), model_mean_type='epsilon',
model_var_type='learned_range', loss_type='mse', betas=get_named_beta_schedule('linear', 4000),
conditioning_free=True, conditioning_free_k=1)
self.spec_decoder = get_mel2wav_v3_model() # The only reason the other functions don't use v3 is because earlier models were trained with v1 and I want to keep metrics consistent.
self.local_modules['spec_decoder'] = self.spec_decoder
@ -215,7 +218,7 @@ class MusicDiffusionFid(evaluator.Evaluator):
gen_cheater = self.diffuser.ddim_sample_loop(self.model, cheater.shape, progress=True, model_kwargs={'conditioning_input': cheater})
# 2. Decode the cheater into a MEL
gen_mel = self.diffuser.ddim_sample_loop(self.local_modules['cheater_decoder'].diff.to(audio.device), (1,256,gen_cheater.shape[-1]*16), progress=True,
gen_mel = self.cheater_decoder_diffuser.ddim_sample_loop(self.local_modules['cheater_decoder'].diff.to(audio.device), (1,256,gen_cheater.shape[-1]*16), progress=True,
model_kwargs={'codes': gen_cheater.permute(0,2,1)})
# 3. And then the MEL back into a spectrogram
@ -303,7 +306,7 @@ class MusicDiffusionFid(evaluator.Evaluator):
if __name__ == '__main__':
diffusion = load_model_from_config('X:\\dlas\\experiments\\train_music_cheater_gen_r8.yml', 'generator',
also_load_savepoint=False,
load_path='X:\\dlas\\experiments\\train_music_cheater_gen_v5\\models\\18000_generator_ema.pth'
load_path='X:\\dlas\\experiments\\train_music_cheater_gen_v5\\models\\46000_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.