Update optimization.py

optimization.py

@@ -1,32 +1,14 @@
-"""
-"""
-
-from typing import Any
-from typing import Callable
-from typing import ParamSpec
-
+from typing import Any, Callable, ParamSpec
 import spaces
 import torch
 from torch.utils._pytree import tree_map_only
-from torchao.quantization import quantize_
-from torchao.quantization import Float8DynamicActivationFloat8WeightConfig
-from torchao.quantization import Int8WeightOnlyConfig
-
-from optimization_utils import capture_component_call
-from optimization_utils import aoti_compile
-from optimization_utils import ZeroGPUCompiledModel
-from optimization_utils import drain_module_parameters
+from torchao.quantization import quantize_, Float8DynamicActivationFloat8WeightConfig, Int8WeightOnlyConfig
+from optimization_utils import capture_component_call, aoti_compile, ZeroGPUCompiledModel, drain_module_parameters
 
 P = ParamSpec('P')
 
-
 TRANSFORMER_NUM_FRAMES_DIM = torch.export.Dim('num_frames', min=3, max=21)
-
-TRANSFORMER_DYNAMIC_SHAPES = {
-    'hidden_states': {
-        2: TRANSFORMER_NUM_FRAMES_DIM,
-    },
-}
+TRANSFORMER_DYNAMIC_SHAPES = {'hidden_states': {2: TRANSFORMER_NUM_FRAMES_DIM}}
 
 INDUCTOR_CONFIGS = {
     'conv_1x1_as_mm': True,
@@ -37,43 +19,48 @@ INDUCTOR_CONFIGS = {
     'triton.cudagraphs': True,
 }
 
-
 def optimize_pipeline_(pipeline: Callable[P, Any], *args: P.args, **kwargs: P.kwargs):
+    print("[optimize_pipeline_] Starting pipeline optimization")
+
+    quantize_(pipeline.text_encoder, Int8WeightOnlyConfig())
+    print("[optimize_pipeline_] Text encoder quantized")
 
     @spaces.GPU(duration=1500)
     def compile_transformer():
-        
+        print("[compile_transformer] Loading LoRA weights")
         pipeline.load_lora_weights(
-            "DeepBeepMeep/Wan2.2", 
-            weight_name="loras_accelerators/Wan2.2-Lightning_T2V-A14B-4steps-lora_HIGH_fp16.safetensors", 
+            "DeepBeepMeep/Wan2.2",
+            weight_name="loras_accelerators/Wan2.2-Lightning_T2V-A14B-4steps-lora_HIGH_fp16.safetensors",
             adapter_name="lightning"
         )
-        kwargs_lora = {}
-        kwargs_lora["load_into_transformer_2"] = True
         pipeline.load_lora_weights(
-            "DeepBeepMeep/Wan2.2", 
-             weight_name="loras_accelerators/Wan2.2-Lightning_T2V-A14B-4steps-lora_LOW_fp16.safetensors", 
-            #weight_name="Wan22-Lightning/Wan2.2-Lightning_T2V-A14B-4steps-lora_LOW_fp16.safetensors", 
-            adapter_name="lightning_2", **kwargs_lora
+            "DeepBeepMeep/Wan2.2",
+            weight_name="loras_accelerators/Wan2.2-Lightning_T2V-A14B-4steps-lora_LOW_fp16.safetensors",
+            adapter_name="lightning_2",
+            load_into_transformer_2=True
         )
-        pipeline.set_adapters(["lightning", "lightning_2"], adapter_weights=[1., 1.])
-
+        pipeline.set_adapters(["lightning", "lightning_2"], adapter_weights=[1.0, 1.0])
 
-        pipeline.fuse_lora(adapter_names=["lightning"], lora_scale=3., components=["transformer"])
-        pipeline.fuse_lora(adapter_names=["lightning_2"], lora_scale=1., components=["transformer_2"])
+        print("[compile_transformer] Fusing LoRA weights")
+        pipeline.fuse_lora(adapter_names=["lightning"], lora_scale=3.0, components=["transformer"])
+        pipeline.fuse_lora(adapter_names=["lightning_2"], lora_scale=1.0, components=["transformer_2"])
         pipeline.unload_lora_weights()
-        
-        with capture_component_call(pipeline, 'transformer') as call:
-            pipeline(*args, **kwargs)
-        
+
+        print("[compile_transformer] Running dummy forward pass to capture component call")
+        with torch.inference_mode():
+            with capture_component_call(pipeline, 'transformer') as call:
+                pipeline(*args, **kwargs)
+
         dynamic_shapes = tree_map_only((torch.Tensor, bool), lambda t: None, call.kwargs)
         dynamic_shapes |= TRANSFORMER_DYNAMIC_SHAPES
 
+        print("[compile_transformer] Quantizing transformers with Float8DynamicActivationFloat8WeightConfig")
         quantize_(pipeline.transformer, Float8DynamicActivationFloat8WeightConfig())
         quantize_(pipeline.transformer_2, Float8DynamicActivationFloat8WeightConfig())
-        
+
         hidden_states: torch.Tensor = call.kwargs['hidden_states']
         hidden_states_transposed = hidden_states.transpose(-1, -2).contiguous()
+
         if hidden_states.shape[-1] > hidden_states.shape[-2]:
             hidden_states_landscape = hidden_states
             hidden_states_portrait = hidden_states_transposed
@@ -81,34 +68,34 @@ def optimize_pipeline_(pipeline: Callable[P, Any], *args: P.args, **kwargs: P.kw
             hidden_states_landscape = hidden_states_transposed
             hidden_states_portrait = hidden_states
 
+        print("[compile_transformer] Exporting transformer landscape model")
         exported_landscape_1 = torch.export.export(
             mod=pipeline.transformer,
             args=call.args,
-            kwargs=call.kwargs | {'hidden_states': hidden_states_landscape},
+            kwargs={**call.kwargs, 'hidden_states': hidden_states_landscape},
             dynamic_shapes=dynamic_shapes,
         )
-        
+        torch.cuda.synchronize()
+
+        print("[compile_transformer] Exporting transformer portrait model")
         exported_portrait_2 = torch.export.export(
             mod=pipeline.transformer_2,
             args=call.args,
-            kwargs=call.kwargs | {'hidden_states': hidden_states_portrait},
+            kwargs={**call.kwargs, 'hidden_states': hidden_states_portrait},
             dynamic_shapes=dynamic_shapes,
         )
+        torch.cuda.synchronize()
 
+        print("[compile_transformer] Compiling models with AoT compilation")
         compiled_landscape_1 = aoti_compile(exported_landscape_1, INDUCTOR_CONFIGS)
         compiled_portrait_2 = aoti_compile(exported_portrait_2, INDUCTOR_CONFIGS)
 
         compiled_landscape_2 = ZeroGPUCompiledModel(compiled_landscape_1.archive_file, compiled_portrait_2.weights)
         compiled_portrait_1 = ZeroGPUCompiledModel(compiled_portrait_2.archive_file, compiled_landscape_1.weights)
 
-        return (
-            compiled_landscape_1,
-            compiled_landscape_2,
-            compiled_portrait_1,
-            compiled_portrait_2,
-        )
+        print("[compile_transformer] Compilation done")
+        return compiled_landscape_1, compiled_landscape_2, compiled_portrait_1, compiled_portrait_2
 
-    quantize_(pipeline.text_encoder, Int8WeightOnlyConfig())
     cl1, cl2, cp1, cp2 = compile_transformer()
 
     def combined_transformer_1(*args, **kwargs):
@@ -127,6 +114,7 @@ def optimize_pipeline_(pipeline: Callable[P, Any], *args: P.args, **kwargs: P.kw
 
     pipeline.transformer.forward = combined_transformer_1
     drain_module_parameters(pipeline.transformer)
-
     pipeline.transformer_2.forward = combined_transformer_2
-    drain_module_parameters(pipeline.transformer_2)
+    drain_module_parameters(pipeline.transformer_2)
+
+    print("[optimize_pipeline_] Optimization complete")