Feature/modernize pytorch

.gitignore

@@ -3,6 +3,7 @@
 useful_ckpts/bigvgan
 useful_ckpts/*.ckpt
 useful_ckpts/CLAP/*.ckpt
+useful_ckpts/CLAP/*.pth
 evaluation
 .idea/
 logs

gen_wav.py

@@ -58,7 +58,7 @@ def parse_args():
 def initialize_model(config, ckpt,device=device):
     config = OmegaConf.load(config)
     model = instantiate_from_config(config.model)
-    model.load_state_dict(torch.load(ckpt,map_location='cpu')["state_dict"], strict=False)
+    model.load_state_dict(torch.load(ckpt,map_location='cpu',weights_only=False)["state_dict"], strict=False, )
 
     model = model.to(device)
     model.cond_stage_model.to(model.device)

ldm/models/diffusion/ddpm.py

@@ -15,7 +15,7 @@
 from functools import partial
 from tqdm import tqdm
 from torchvision.utils import make_grid
-from pytorch_lightning.utilities.distributed import rank_zero_only
+from pytorch_lightning.utilities import rank_zero_only
 
 from ldm.util import log_txt_as_img, exists, default, ismap, isimage, mean_flat, count_params, instantiate_from_config
 from ldm.modules.ema import LitEma

ldm/models/diffusion/ddpm_audio.py

@@ -12,7 +12,7 @@
 from einops import rearrange, repeat
 from tqdm import tqdm
 from torchvision.utils import make_grid
-from pytorch_lightning.utilities.distributed import rank_zero_only
+from pytorch_lightning.utilities import rank_zero_only
 
 from ldm.util import log_txt_as_img, exists, default, ismap, isimage, mean_flat, count_params, instantiate_from_config
 from ldm.modules.distributions.distributions import normal_kl, DiagonalGaussianDistribution

requirements-cpu.txt

@@ -0,0 +1,28 @@
+# ──────────────────────────────────────────────────────────────
+# PyTorch
+# ──────────────────────────────────────────────────────────────
+torch==2.6.0            # core CPU build
+torchvision==0.21.0     # CPU build aligned with torch 2.6.0
+torchaudio==2.6.0       # CPU build aligned with torch 2.6.0
+
+# ──────────────────────────────────────────────────────────────
+# Remaining dependencies (CPU-agnostic or CUDA handled via PyTorch)
+# ──────────────────────────────────────────────────────────────
+torch-fidelity==0.3.0
+scipy
+importlib_resources
+tqdm
+omegaconf
+einops
+numpy<=1.23.5
+soundfile
+librosa==0.9.2
+pandas
+torchlibrosa
+transformers==4.18.0
+ftfy
+pytorch-lightning==2.5.1
+torchmetrics==0.11.1
+
+# CompVis taming-transformers (editable install)
+-e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers

requirements.txt

@@ -1,10 +1,18 @@
---extra-index-url https://download.pytorch.org/whl/cu113
-torch
+# ──────────────────────────────────────────────────────────────
+# PyTorch + CUDA 12.4 wheels
+# ──────────────────────────────────────────────────────────────
+--extra-index-url https://download.pytorch.org/whl/cu124
+
+torch==2.6.0+cu124         # core GPU build
+torchvision==0.21.0+cu124    # GPU build aligned with torch 2.6.0
+torchaudio==2.6.0+cu124     # GPU build aligned with torch 2.6.0
+
+# ──────────────────────────────────────────────────────────────
+# Remaining dependencies (CPU-agnostic or CUDA handled via PyTorch)
+# ──────────────────────────────────────────────────────────────
 torch-fidelity==0.3.0
 scipy
 importlib_resources
-torchaudio>=0.13.0
-torchvision>=0.14.0
 tqdm
 omegaconf
 einops
@@ -15,6 +23,8 @@ pandas
 torchlibrosa
 transformers==4.18.0
 ftfy
-pytorch-lightning==1.7.0
+pytorch-lightning==2.5.1
 torchmetrics==0.11.1
+
+# CompVis taming-transformers (editable install)
 -e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers