Update app.py

app.py

@@ -33,12 +33,12 @@ if clipmodel == "long":
 
 torch.backends.cuda.matmul.allow_tf32 = True
 
-clip_model = CLIPModel.from_pretrained(model_id, torch_dtype=torch.bfloat16, config=config).to(device)
-clip_processor = CLIPProcessor.from_pretrained(model_id, padding="max_length", max_length=maxtokens, return_tensors="pt", truncation=True)
+clip_model = CLIPModel.from_pretrained(model_id, torch_dtype=torch.bfloat16, config=config, ignore_mismatched_sizes=True).to(device)
+clip_processor = CLIPProcessor.from_pretrained(model_id, padding="max_length", max_length=maxtokens, ignore_mismatched_sizes=True, return_tensors="pt", truncation=True)
 config.text_config.max_position_embeddings = 248
 
 
-pipe = FluxPipeline.from_pretrained("AlekseyCalvin/HistoricColorSoonr_v2_FluxSchnell_Diffusers", torch_dtype=torch.bfloat16)
+pipe = FluxPipeline.from_pretrained("AlekseyCalvin/HistoricColorSoonr_v2_FluxSchnell_Diffusers", ignore_mismatched_sizes=True, torch_dtype=torch.bfloat16)
 pipe.to(device="cuda", dtype=torch.bfloat16)
 
 pipe.tokenizer = clip_processor.tokenizer
@@ -91,6 +91,14 @@ def update_selection(evt: gr.SelectData, width, height):
     )
 
 @spaces.GPU(duration=70)
+pipe.vae.enable_slicing()
+pipe.vae.enable_tiling()
+
+# Just to look at the tokens / confirm settings:
+tokens = clip_processor(
+    [prompt], padding="max_length", max_length=maxtokens, return_tensors="pt", truncation=True
+)
+
 def generate_image(prompt, trigger_word, steps, seed, cfg_scale, width, height, lora_scale, progress):
     pipe.to("cuda")
     generator = torch.Generator(device="cuda").manual_seed(seed)