Modular backend - add rescale cfg (#6640)

## Summary

Rescale CFG code from #6577.

## Related Issues / Discussions

#6606 

https://invokeai.notion.site/Modular-Stable-Diffusion-Backend-Design-Document-e8952daab5d5472faecdc4a72d377b0d

## QA Instructions

Run with and without set `USE_MODULAR_DENOISE` environment.
~~Note: for some reasons there slightly different output from run to
run, but I able sometimes to get same output on main and this branch.~~
Fix presented in #6641.

## Merge Plan

~~Nope.~~ Merge #6641 firstly, to be able see output difference
properly.
If you think that there should be some kind of tests - feel free to add.

## Checklist

- [x] _The PR has a short but descriptive title, suitable for a
changelog_
- [ ] _Tests added / updated (if applicable)_
- [ ] _Documentation added / updated (if applicable)_

invokeai/app/invocations/denoise_latents.py

@@ -59,6 +59,7 @@ from invokeai.backend.stable_diffusion.diffusion.custom_atttention import Custom
 from invokeai.backend.stable_diffusion.diffusion_backend import StableDiffusionBackend
 from invokeai.backend.stable_diffusion.extension_callback_type import ExtensionCallbackType
 from invokeai.backend.stable_diffusion.extensions.preview import PreviewExt
+from invokeai.backend.stable_diffusion.extensions.rescale_cfg import RescaleCFGExt
 from invokeai.backend.stable_diffusion.extensions_manager import ExtensionsManager
 from invokeai.backend.stable_diffusion.schedulers import SCHEDULER_MAP
 from invokeai.backend.stable_diffusion.schedulers.schedulers import SCHEDULER_NAME_VALUES
@@ -790,6 +791,10 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
         ext_manager.add_extension(PreviewExt(step_callback))
 
+        ### cfg rescale
+        if self.cfg_rescale_multiplier > 0:
+            ext_manager.add_extension(RescaleCFGExt(self.cfg_rescale_multiplier))
+
         # ext: t2i/ip adapter
         ext_manager.run_callback(ExtensionCallbackType.SETUP, denoise_ctx)
 

invokeai/backend/stable_diffusion/denoise_context.py

@@ -83,47 +83,47 @@ class DenoiseContext:
     unet: Optional[UNet2DConditionModel] = None
 
     # Current state of latent-space image in denoising process.
-    # None until `pre_denoise_loop` callback.
+    # None until `PRE_DENOISE_LOOP` callback.
     # Shape: [batch, channels, latent_height, latent_width]
     latents: Optional[torch.Tensor] = None
 
     # Current denoising step index.
-    # None until `pre_step` callback.
+    # None until `PRE_STEP` callback.
     step_index: Optional[int] = None
 
     # Current denoising step timestep.
-    # None until `pre_step` callback.
+    # None until `PRE_STEP` callback.
     timestep: Optional[torch.Tensor] = None
 
     # Arguments which will be passed to UNet model.
-    # Available in `pre_unet`/`post_unet` callbacks, otherwise will be None.
+    # Available in `PRE_UNET`/`POST_UNET` callbacks, otherwise will be None.
     unet_kwargs: Optional[UNetKwargs] = None
 
     # SchedulerOutput class returned from step function(normally, generated by scheduler).
-    # Supposed to be used only in `post_step` callback, otherwise can be None.
+    # Supposed to be used only in `POST_STEP` callback, otherwise can be None.
     step_output: Optional[SchedulerOutput] = None
 
     # Scaled version of `latents`, which will be passed to unet_kwargs initialization.
-    # Available in events inside step(between `pre_step` and `post_stop`).
+    # Available in events inside step(between `PRE_STEP` and `POST_STEP`).
     # Shape: [batch, channels, latent_height, latent_width]
     latent_model_input: Optional[torch.Tensor] = None
 
     # [TMP] Defines on which conditionings current unet call will be runned.
-    # Available in `pre_unet`/`post_unet` callbacks, otherwise will be None.
+    # Available in `PRE_UNET`/`POST_UNET` callbacks, otherwise will be None.
     conditioning_mode: Optional[ConditioningMode] = None
 
     # [TMP] Noise predictions from negative conditioning.
-    # Available in `apply_cfg` and `post_apply_cfg` callbacks, otherwise will be None.
+    # Available in `POST_COMBINE_NOISE_PREDS` callback, otherwise will be None.
     # Shape: [batch, channels, latent_height, latent_width]
     negative_noise_pred: Optional[torch.Tensor] = None
 
     # [TMP] Noise predictions from positive conditioning.
-    # Available in `apply_cfg` and `post_apply_cfg` callbacks, otherwise will be None.
+    # Available in `POST_COMBINE_NOISE_PREDS` callback, otherwise will be None.
     # Shape: [batch, channels, latent_height, latent_width]
     positive_noise_pred: Optional[torch.Tensor] = None
 
     # Combined noise prediction from passed conditionings.
-    # Available in `apply_cfg` and `post_apply_cfg` callbacks, otherwise will be None.
+    # Available in `POST_COMBINE_NOISE_PREDS` callback, otherwise will be None.
     # Shape: [batch, channels, latent_height, latent_width]
     noise_pred: Optional[torch.Tensor] = None
 

invokeai/backend/stable_diffusion/diffusion_backend.py

@@ -76,12 +76,12 @@ class StableDiffusionBackend:
             both_noise_pred = self.run_unet(ctx, ext_manager, ConditioningMode.Both)
             ctx.negative_noise_pred, ctx.positive_noise_pred = both_noise_pred.chunk(2)
 
-        # ext: override apply_cfg
-        ctx.noise_pred = self.apply_cfg(ctx)
+        # ext: override combine_noise_preds
+        ctx.noise_pred = self.combine_noise_preds(ctx)
 
         # ext: cfg_rescale [modify_noise_prediction]
         # TODO: rename
-        ext_manager.run_callback(ExtensionCallbackType.POST_APPLY_CFG, ctx)
+        ext_manager.run_callback(ExtensionCallbackType.POST_COMBINE_NOISE_PREDS, ctx)
 
         # compute the previous noisy sample x_t -> x_t-1
         step_output = ctx.scheduler.step(ctx.noise_pred, ctx.timestep, ctx.latents, **ctx.inputs.scheduler_step_kwargs)
@@ -95,7 +95,7 @@ class StableDiffusionBackend:
         return step_output
 
     @staticmethod
-    def apply_cfg(ctx: DenoiseContext) -> torch.Tensor:
+    def combine_noise_preds(ctx: DenoiseContext) -> torch.Tensor:
         guidance_scale = ctx.inputs.conditioning_data.guidance_scale
         if isinstance(guidance_scale, list):
             guidance_scale = guidance_scale[ctx.step_index]

invokeai/backend/stable_diffusion/extension_callback_type.py

@@ -9,4 +9,4 @@ class ExtensionCallbackType(Enum):
     POST_STEP = "post_step"
     PRE_UNET = "pre_unet"
     POST_UNET = "post_unet"
-    POST_APPLY_CFG = "post_apply_cfg"
+    POST_COMBINE_NOISE_PREDS = "post_combine_noise_preds"

invokeai/backend/stable_diffusion/extensions/rescale_cfg.py

@@ -0,0 +1,36 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+import torch
+
+from invokeai.backend.stable_diffusion.extension_callback_type import ExtensionCallbackType
+from invokeai.backend.stable_diffusion.extensions.base import ExtensionBase, callback
+
+if TYPE_CHECKING:
+    from invokeai.backend.stable_diffusion.denoise_context import DenoiseContext
+
+
+class RescaleCFGExt(ExtensionBase):
+    def __init__(self, rescale_multiplier: float):
+        super().__init__()
+        self._rescale_multiplier = rescale_multiplier
+
+    @staticmethod
+    def _rescale_cfg(total_noise_pred: torch.Tensor, pos_noise_pred: torch.Tensor, multiplier: float = 0.7):
+        """Implementation of Algorithm 2 from https://arxiv.org/pdf/2305.08891.pdf."""
+        ro_pos = torch.std(pos_noise_pred, dim=(1, 2, 3), keepdim=True)
+        ro_cfg = torch.std(total_noise_pred, dim=(1, 2, 3), keepdim=True)
+
+        x_rescaled = total_noise_pred * (ro_pos / ro_cfg)
+        x_final = multiplier * x_rescaled + (1.0 - multiplier) * total_noise_pred
+        return x_final
+
+    @callback(ExtensionCallbackType.POST_COMBINE_NOISE_PREDS)
+    def rescale_noise_pred(self, ctx: DenoiseContext):
+        if self._rescale_multiplier > 0:
+            ctx.noise_pred = self._rescale_cfg(
+                ctx.noise_pred,
+                ctx.positive_noise_pred,
+                self._rescale_multiplier,
+            )