From d4e27ccf62a791977a7876a5be52b77410aef876 Mon Sep 17 00:00:00 2001 From: James Betker Date: Tue, 11 Jan 2022 16:25:40 -0700 Subject: [PATCH] misc updates --- codes/requirements.txt | 3 ++- codes/scripts/audio/gen/use_gpt_tts.py | 4 ++-- codes/train.py | 2 +- 3 files changed, 5 insertions(+), 4 deletions(-) diff --git a/codes/requirements.txt b/codes/requirements.txt index 73355204..cda86fca 100644 --- a/codes/requirements.txt +++ b/codes/requirements.txt @@ -39,4 +39,5 @@ vector_quantize_pytorch linear_attention_transformer rotary-embedding-torch axial_positional_embedding -g-mlp-pytorch \ No newline at end of file +g-mlp-pytorch +x-clip \ No newline at end of file diff --git a/codes/scripts/audio/gen/use_gpt_tts.py b/codes/scripts/audio/gen/use_gpt_tts.py index da3c9129..a7a2746e 100644 --- a/codes/scripts/audio/gen/use_gpt_tts.py +++ b/codes/scripts/audio/gen/use_gpt_tts.py @@ -94,9 +94,9 @@ if __name__ == '__main__': parser.add_argument('-diffusion_model_name', type=str, help='Name of the diffusion model in opt.', default='generator') parser.add_argument('-diffusion_model_path', type=str, help='Diffusion model checkpoint to load.', default='X:\\dlas\\experiments\\train_diffusion_vocoder_with_cond_new_dvae_full\\models\\6100_generator_ema.pth') parser.add_argument('-dvae_model_name', type=str, help='Name of the DVAE model in opt.', default='dvae') - parser.add_argument('-opt_gpt_tts', type=str, help='Path to options YAML file used to train the GPT-TTS model', default='X:\\dlas\\experiments\\train_gpt_tts_unified.yml') + parser.add_argument('-opt_gpt_tts', type=str, help='Path to options YAML file used to train the GPT-TTS model', default='X:\\dlas\\experiments\\train_gpt_tts_unified\\train_gpt_tts_unified.yml') parser.add_argument('-gpt_tts_model_name', type=str, help='Name of the GPT TTS model in opt.', default='gpt') - parser.add_argument('-gpt_tts_model_path', type=str, help='GPT TTS model checkpoint to load.', default='X:\\dlas\\experiments\\train_gpt_tts_unified\\models\\60000_gpt.pth') + parser.add_argument('-gpt_tts_model_path', type=str, help='GPT TTS model checkpoint to load.', default='X:\\dlas\\experiments\\train_gpt_tts_unified\\models\\60000_gpt_ema.pth') parser.add_argument('-opt_clip', type=str, help='Path to options YAML file used to train the CLIP model', default='X:\\dlas\\experiments\\train_clip_text_to_voice.yml') parser.add_argument('-clip_model_name', type=str, help='Name of the CLIP model in opt.', default='clip') parser.add_argument('-clip_model_path', type=str, help='CLIP model checkpoint to load.', default='X:\\dlas\\experiments\\train_clip_text_to_voice_masking_bigger_batch\\models\\23500_clip_ema.pth') diff --git a/codes/train.py b/codes/train.py index 1a9d118d..0e6e6699 100644 --- a/codes/train.py +++ b/codes/train.py @@ -300,7 +300,7 @@ class Trainer: if __name__ == '__main__': parser = argparse.ArgumentParser() - parser.add_argument('-opt', type=str, help='Path to option YAML file.', default='../options/train_gpt_asr_hf2_lg_distill.yml') + parser.add_argument('-opt', type=str, help='Path to option YAML file.', default='../experiments/train_gpt_asr_mass_hf2_audio_only_fp32/train_gpt_asr_mass_hf2.yml') parser.add_argument('--launcher', choices=['none', 'pytorch'], default='none', help='job launcher') parser.add_argument('--local_rank', type=int, default=0) args = parser.parse_args()