fix broken --ckpt_convert option

- not sure why, but at some pont --ckpt_convert (which converts legacy checkpoints)
  into diffusers in memory, stopped working due to float16/float32 issues.

- this commit repairs the problem

- also removed some debugging messages I found in passing

ldm/invoke/ckpt_to_diffuser.py

@@ -803,6 +803,7 @@ def load_pipeline_from_original_stable_diffusion_ckpt(
         extract_ema:bool=True,
         upcast_attn:bool=False,
         vae:AutoencoderKL=None,
+        precision:torch.dtype=torch.float32,
         return_generator_pipeline:bool=False,
 )->Union[StableDiffusionPipeline,StableDiffusionGeneratorPipeline]:
     '''
@@ -828,6 +829,7 @@ def load_pipeline_from_original_stable_diffusion_ckpt(
      checkpoints that have both EMA and non-EMA weights. Whether to extract the EMA weights
      or not. Defaults to `False`. Pass `True` to extract the EMA weights. EMA weights usually yield higher
      quality images for inference. Non-EMA weights are usually better to continue fine-tuning.
+    :param precision: precision to use - torch.float16, torch.float32 or torch.autocast
     :param upcast_attention: Whether the attention computation should always be upcasted. This is necessary when
     running stable diffusion 2.1.
     '''
@@ -837,7 +839,7 @@ def load_pipeline_from_original_stable_diffusion_ckpt(
         verbosity = dlogging.get_verbosity()
         dlogging.set_verbosity_error()
 
-        checkpoint = load_file(checkpoint_path) if Path(checkpoint_path).suffix == '.safetensors' else torch.load(checkpoint_path)
+        checkpoint = load_file(checkpoint_path,device='cpu') if Path(checkpoint_path).suffix == '.safetensors' else torch.load(checkpoint_path,device='cpu')
         cache_dir = global_cache_dir('hub')
         pipeline_class = StableDiffusionGeneratorPipeline if return_generator_pipeline else StableDiffusionPipeline
 
@@ -988,12 +990,12 @@ def load_pipeline_from_original_stable_diffusion_ckpt(
             safety_checker = StableDiffusionSafetyChecker.from_pretrained('CompVis/stable-diffusion-safety-checker',cache_dir=global_cache_dir("hub"))
             feature_extractor = AutoFeatureExtractor.from_pretrained("CompVis/stable-diffusion-safety-checker",cache_dir=cache_dir)
             pipe = pipeline_class(
-                vae=vae,
-                text_encoder=text_model,
+                vae=vae.to(precision),
+                text_encoder=text_model.to(precision),
                 tokenizer=tokenizer,
-                unet=unet,
+                unet=unet.to(precision),
                 scheduler=scheduler,
-                safety_checker=safety_checker,
+                safety_checker=safety_checker.to(precision),
                 feature_extractor=feature_extractor,
             )
         else:

ldm/invoke/generator/base.py

@@ -336,7 +336,6 @@ class Generator:
         if self.caution_img:
             return self.caution_img
         path = Path(web_assets.__path__[0]) / CAUTION_IMG
-        print(f'DEBUG: path to caution = {path}')
         caution = Image.open(path)
         self.caution_img = caution.resize((caution.width // 2, caution.height //2))
         return self.caution_img

ldm/invoke/generator/omnibus.py

@@ -40,8 +40,6 @@ class Omnibus(Img2Img,Txt2Img):
         self.perlin = perlin
         num_samples = 1
 
-        print('DEBUG: IN OMNIBUS')
-
         sampler.make_schedule(
             ddim_num_steps=steps, ddim_eta=ddim_eta, verbose=False
         )

ldm/invoke/model_manager.py

@@ -385,19 +385,25 @@ class ModelManager(object):
             from ldm.invoke.ckpt_to_diffuser import (
                 load_pipeline_from_original_stable_diffusion_ckpt,
             )
-
+            self.offload_model(self.current_model)
             if vae_config := self._choose_diffusers_vae(model_name):
                 vae = self._load_vae(vae_config)
+            if self._has_cuda():
+                torch.cuda.empty_cache()
             pipeline = load_pipeline_from_original_stable_diffusion_ckpt(
                 checkpoint_path=weights,
                 original_config_file=config,
                 vae=vae,
                 return_generator_pipeline=True,
+                precision=torch.float16 if self.precision=='float16' else torch.float32,
             )
+            if self.sequential_offload:
+                pipeline.enable_offload_submodels(self.device)
+            else:
+                pipeline.to(self.device)
+                
             return (
-                pipeline.to(self.device).to(
-                    torch.float16 if self.precision == "float16" else torch.float32
-                ),
+                pipeline,
                 width,
                 height,
                 "NOHASH",

ldm/invoke/training/textual_inversion.py

@@ -421,7 +421,6 @@ def do_front_end(args: Namespace):
         save_args(args)
 
         try:
-            print(f"DEBUG: args = {args}")
             do_textual_inversion_training(**args)
             copy_to_embeddings_folder(args)
         except Exception as e: