Use a string pipeline class with explicit custom loading

README.md

@@ -80,6 +80,7 @@ from diffusers import DiffusionPipeline
 
 pipe = DiffusionPipeline.from_pretrained(
     'mimbres/spansynth-edit',
+    custom_pipeline='pipeline',
     trust_remote_code=True,
 ).to('cuda')
 result = pipe(
@@ -94,12 +95,15 @@ sf.write('edited.wav', result.audios[0, 0], result.sample_rate, subtype='FLOAT')
 Use `.to('mps')` for Apple Silicon or `.to('cpu')` for CPU. For FlowEdit, add
 `method='flowedit'` and `source_midi` pointing to the original score. The pipeline
 also accepts the crop, region, MIDI offset, and context options described below.
-This custom pipeline uses the installed `spansynth` package.
+This custom pipeline uses the installed `spansynth` package. Pass
+`custom_pipeline='pipeline'` to load the repository's `pipeline.py` implementation.
 Save and reopen the pipeline locally with:
 
 ```python
 pipe.save_pretrained('saved-pipeline')
-pipe = DiffusionPipeline.from_pretrained('saved-pipeline', trust_remote_code=True)
+pipe = DiffusionPipeline.from_pretrained(
+    'saved-pipeline', custom_pipeline='pipeline', trust_remote_code=True,
+)
 ```
 
 ## Try an example

model_index.json

@@ -1,8 +1,5 @@
 {
-  "_class_name": [
-    "pipeline",
-    "SpanSynthEditPipeline"
-  ],
+  "_class_name": "SpanSynthEditPipeline",
   "_diffusers_version": "0.40.0",
   "codec": [
     "spansynth.diffusers_models",

pipeline.py

@@ -1,7 +1,6 @@
 """MIDI-guided music synthesis and editing through Diffusers."""
 
 from dataclasses import dataclass
-import json
 import math
 from pathlib import Path
 
@@ -54,12 +53,6 @@ class SpanSynthEditPipeline(DiffusionPipeline):
         super().__init__()
         self.register_modules(transformer=transformer, codec=codec)
 
-    def to_json_string(self):
-        """Identify the saved custom pipeline for Diffusers' standard loader."""
-        config = json.loads(super().to_json_string())
-        config["_class_name"] = ["pipeline", self.__class__.__name__]
-        return json.dumps(config, indent=2, sort_keys=True) + "\n"
-
     def save_pretrained(self, save_directory: str | Path, **kwargs):
         """Save weights and the code entries required by the Hub's custom loader.