update

diffrhythm/infer/infer_utils.py

@@ -19,7 +19,8 @@ def prepare_model(device):
         model_config = json.load(f)
     dit_model_cls = DiT
     cfm = CFM(
-                transformer=dit_model_cls(**model_config["model"], use_style_prompt=True, max_pos=6144),
+                #transformer=dit_model_cls(**model_config["model"], use_style_prompt=True, max_pos=6144),
+                transformer=dit_model_cls(**model_config["model"], use_style_prompt=True, max_pos=2048),
                 num_channels=model_config["model"]['mel_dim'],
                 use_style_prompt=True
              )

diffrhythm/model/cfm.py

@@ -111,8 +111,8 @@ class CFM(nn.Module):
         cfg_strength=4.0,
         sway_sampling_coef=None,
         seed: int | None = None,
-        #max_duration=4096,
-        max_duration=6144,
+        max_duration=4096,
+        #max_duration=6144,
         vocoder: Callable[[float["b d n"]], float["b nw"]] | None = None,  # noqa: F722
         no_ref_audio=False,
         duplicate_test=False,