Update src/pipeline.py

src/pipeline.py

@@ -24,12 +24,13 @@ def load_pipeline() -> Pipeline:
     path = os.path.join(HF_HUB_CACHE, "models--slobers--Flux.1.Schnella/snapshots/e34d670e44cecbbc90e4962e7aada2ac5ce8b55b/transformer")
     transformer = FluxTransformer2DModel.from_pretrained(path, torch_dtype=torch.bfloat16, use_safetensors=False)
     pipeline = FluxPipeline.from_pretrained(ids, revision=Revision, transformer=transformer, local_files_only=True, torch_dtype=torch.bfloat16,)
-    pipeline.to("cuda")
+    
     pipeline.transformer = torch.compile(pipeline.transformer, mode="max-autotune", fullgraph=True)
     # basepath = os.path.join(HF_HUB_CACHE, "models--manbeast3b--Flux.1.schnell-kl-vae-eagle5-0_1-iter3/snapshots/514bc844598dc8528497f04ae81c3dfeb7600138")
     # pipeline.vae.encoder.load_state_dict(torch.load(os.path.join(basepath, "encoder.pth")), strict=False)
     # pipeline.vae.decoder.load_state_dict(torch.load(os.path.join(basepath, "decoder.pth")), strict=False)
     quantize_(pipeline.vae, int8_weight_only())
+    pipeline.to("cuda")
     for _ in range(3):
         pipeline(prompt="insensible, timbale, pothery, electrovital, actinogram, taxis, intracerebellar, centrodesmus", width=1024, height=1024, guidance_scale=0.0, num_inference_steps=4, max_sequence_length=256)
     return pipeline