Add lcm_convert.py to cli

cli/lcm_convert.py

@@ -0,0 +1,42 @@
+import argparse
+import torch
+from diffusers import StableDiffusionPipeline, StableDiffusionXLPipeline, AutoPipelineForText2Image, LCMScheduler
+
+parser = argparse.ArgumentParser("lcm_convert")
+parser.add_argument("--name", help="Name of the new LCM model", type=str)
+parser.add_argument("--model", help="A model to convert", type=str)
+parser.add_argument("--huggingface", action="store_true", help="Use Hugging Face models instead of safetensors models")
+parser.add_argument("--upload", action="store_true", help="Upload the new LCM model to Hugging Face")
+parser.add_argument("--no_save", action="store_true", help="Don't save the new LCM model to local disk")
+parser.add_argument("--sdxl", action="store_true", help="Use SDXL models")
+parser.add_argument("--ssd_1b", action="store_true", help="Use SSD-1B models")
+
+args = parser.parse_args()
+
+if args.huggingface:
+    pipeline = AutoPipelineForText2Image.from_pretrained(args.model, torch_dtype=torch.float16, variant="fp16")
+else:
+    if args.sdxl or args.ssd_1b:
+        pipeline = StableDiffusionXLPipeline.from_single_file(args.model)
+    else:
+        pipeline = StableDiffusionPipeline.from_single_file(args.model)
+
+pipeline.scheduler = LCMScheduler.from_config(pipeline.scheduler.config)
+if args.sdxl:
+    pipeline.load_lora_weights("latent-consistency/lcm-lora-sdxl")
+elif args.ssd_1b:
+    pipeline.load_lora_weights("latent-consistency/lcm-lora-ssd-1b")
+else:
+    pipeline.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
+pipeline.fuse_lora()
+
+#components = pipeline.components
+#pipeline = LatentConsistencyModelPipeline(**components)
+
+pipeline = pipeline.to(dtype=torch.float16)
+print(pipeline)
+
+if not args.no_save:
+    pipeline.save_pretrained(args.name, variant="fp16")
+if args.upload:
+    pipeline.push_to_hub(args.name, variant="fp16")