Improve reproduceability 2/3 (#1906)

* [Repro] Correct reproducability

* up

* up

* uP

* up

* need better image

* allow conversion from no state dict checkpoints

* up

* up

* up

* up

* check tensors

* check tensors

* check tensors

* check tensors

* next try

* up

* up

* better name

* up

* up

* Apply suggestions from code review

* correct more

* up

* replace all torch randn

* fix

* correct

* correct

* finish

* fix more

* up

src/diffusers/experimental/rl/value_guided_sampling.py

@@ -19,6 +19,7 @@ import tqdm
 
 from ...models.unet_1d import UNet1DModel
 from ...pipelines import DiffusionPipeline
+from ...utils import randn_tensor
 from ...utils.dummy_pt_objects import DDPMScheduler
 
 
@@ -127,7 +128,7 @@ class ValueGuidedRLPipeline(DiffusionPipeline):
         shape = (batch_size, planning_horizon, self.state_dim + self.action_dim)
 
         # generate initial noise and apply our conditions (to make the trajectories start at current state)
-        x1 = torch.randn(shape, device=self.unet.device)
+        x1 = randn_tensor(shape, device=self.unet.device)
         x = self.reset_x0(x1, conditions, self.action_dim)
         x = self.to_torch(x)
 

src/diffusers/models/prior_transformer.py

@@ -95,7 +95,7 @@ class PriorTransformer(ModelMixin, ConfigMixin):
         self.proj_to_clip_embeddings = nn.Linear(inner_dim, embedding_dim)
 
         causal_attention_mask = torch.full(
-            [num_embeddings + additional_embeddings, num_embeddings + additional_embeddings], float("-inf")
+            [num_embeddings + additional_embeddings, num_embeddings + additional_embeddings], -10000.0
         )
         causal_attention_mask.triu_(1)
         causal_attention_mask = causal_attention_mask[None, ...]

src/diffusers/models/vae.py

@@ -18,7 +18,7 @@ import numpy as np
 import torch
 import torch.nn as nn
 
-from ..utils import BaseOutput
+from ..utils import BaseOutput, randn_tensor
 from .unet_2d_blocks import UNetMidBlock2D, get_down_block, get_up_block
 
 
@@ -323,11 +323,10 @@ class DiagonalGaussianDistribution(object):
             )
 
     def sample(self, generator: Optional[torch.Generator] = None) -> torch.FloatTensor:
-        device = self.parameters.device
-        sample_device = "cpu" if device.type == "mps" else device
-        sample = torch.randn(self.mean.shape, generator=generator, device=sample_device)
         # make sure sample is on the same device as the parameters and has same dtype
-        sample = sample.to(device=device, dtype=self.parameters.dtype)
+        sample = randn_tensor(
+            self.mean.shape, generator=generator, device=self.parameters.device, dtype=self.parameters.dtype
+        )
         x = self.mean + self.std * sample
         return x
 

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion.py

@@ -31,7 +31,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import deprecate, logging, replace_example_docstring
+from ...utils import deprecate, logging, randn_tensor, replace_example_docstring
 from ..pipeline_utils import DiffusionPipeline
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 from . import AltDiffusionPipelineOutput, RobertaSeriesModelWithTransformation
@@ -401,20 +401,8 @@ class AltDiffusionPipeline(DiffusionPipeline):
             )
 
         if latents is None:
-            rand_device = "cpu" if device.type == "mps" else device
-
-            if isinstance(generator, list):
-                shape = (1,) + shape[1:]
-                latents = [
-                    torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype)
-                    for i in range(batch_size)
-                ]
-                latents = torch.cat(latents, dim=0).to(device)
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
         else:
-                latents = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
-        else:
-            if latents.shape != shape:
-                raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {shape}")
             latents = latents.to(device)
 
         # scale the initial noise by the standard deviation required by the scheduler

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion_img2img.py

@@ -33,7 +33,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import PIL_INTERPOLATION, deprecate, logging, replace_example_docstring
+from ...utils import PIL_INTERPOLATION, deprecate, logging, randn_tensor, replace_example_docstring
 from ..pipeline_utils import DiffusionPipeline
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 from . import AltDiffusionPipelineOutput, RobertaSeriesModelWithTransformation
@@ -461,16 +461,8 @@ class AltDiffusionImg2ImgPipeline(DiffusionPipeline):
         else:
             init_latents = torch.cat([init_latents], dim=0)
 
-        rand_device = "cpu" if device.type == "mps" else device
         shape = init_latents.shape
-        if isinstance(generator, list):
-            shape = (1,) + shape[1:]
-            noise = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype) for i in range(batch_size)
-            ]
-            noise = torch.cat(noise, dim=0).to(device)
-        else:
-            noise = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
+        noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
 
         # get latents
         init_latents = self.scheduler.add_noise(init_latents, noise, timestep)

src/diffusers/pipelines/audio_diffusion/pipeline_audio_diffusion.py

@@ -23,6 +23,7 @@ from PIL import Image
 
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import DDIMScheduler, DDPMScheduler
+from ...utils import randn_tensor
 from ..pipeline_utils import AudioPipelineOutput, BaseOutput, DiffusionPipeline, ImagePipelineOutput
 from .mel import Mel
 
@@ -126,7 +127,7 @@ class AudioDiffusionPipeline(DiffusionPipeline):
         input_dims = self.get_input_dims()
         self.mel.set_resolution(x_res=input_dims[1], y_res=input_dims[0])
         if noise is None:
-            noise = torch.randn(
+            noise = randn_tensor(
                 (
                     batch_size,
                     self.unet.in_channels,

src/diffusers/pipelines/dance_diffusion/pipeline_dance_diffusion.py

@@ -17,7 +17,7 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ...utils import logging
+from ...utils import logging, randn_tensor
 from ..pipeline_utils import AudioPipelineOutput, DiffusionPipeline
 
 
@@ -100,16 +100,7 @@ class DanceDiffusionPipeline(DiffusionPipeline):
                 f" size of {batch_size}. Make sure the batch size matches the length of the generators."
             )
 
-        rand_device = "cpu" if self.device.type == "mps" else self.device
-        if isinstance(generator, list):
-            shape = (1,) + shape[1:]
-            audio = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=self.unet.dtype)
-                for i in range(batch_size)
-            ]
-            audio = torch.cat(audio, dim=0).to(self.device)
-        else:
-            audio = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(self.device)
+        audio = randn_tensor(shape, generator=generator, device=self.device, dtype=dtype)
 
         # set step values
         self.scheduler.set_timesteps(num_inference_steps, device=audio.device)

src/diffusers/pipelines/ddim/pipeline_ddim.py

@@ -16,7 +16,7 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ...utils import deprecate
+from ...utils import deprecate, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -103,17 +103,7 @@ class DDIMPipeline(DiffusionPipeline):
                 f" size of {batch_size}. Make sure the batch size matches the length of the generators."
             )
 
-        rand_device = "cpu" if self.device.type == "mps" else self.device
-        if isinstance(generator, list):
-            shape = (1,) + image_shape[1:]
-            image = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=self.unet.dtype)
-                for i in range(batch_size)
-            ]
-            image = torch.cat(image, dim=0).to(self.device)
-        else:
-            image = torch.randn(image_shape, generator=generator, device=rand_device, dtype=self.unet.dtype)
-            image = image.to(self.device)
+        image = randn_tensor(image_shape, generator=generator, device=self.device, dtype=self.unet.dtype)
 
         # set step values
         self.scheduler.set_timesteps(num_inference_steps)

src/diffusers/pipelines/ddpm/pipeline_ddpm.py

@@ -18,7 +18,7 @@ from typing import List, Optional, Tuple, Union
 import torch
 
 from ...configuration_utils import FrozenDict
-from ...utils import deprecate
+from ...utils import deprecate, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -100,10 +100,10 @@ class DDPMPipeline(DiffusionPipeline):
 
         if self.device.type == "mps":
             # randn does not work reproducibly on mps
-            image = torch.randn(image_shape, generator=generator)
+            image = randn_tensor(image_shape, generator=generator)
             image = image.to(self.device)
         else:
-            image = torch.randn(image_shape, generator=generator, device=self.device)
+            image = randn_tensor(image_shape, generator=generator, device=self.device)
 
         # set step values
         self.scheduler.set_timesteps(num_inference_steps)

src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py

@@ -26,6 +26,7 @@ from transformers.utils import logging
 
 from ...models import AutoencoderKL, UNet2DConditionModel, UNet2DModel, VQModel
 from ...schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler
+from ...utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -143,20 +144,7 @@ class LDMTextToImagePipeline(DiffusionPipeline):
             )
 
         if latents is None:
-            rand_device = "cpu" if self.device.type == "mps" else self.device
-
-            if isinstance(generator, list):
-                latents_shape = (1,) + latents_shape[1:]
-                latents = [
-                    torch.randn(latents_shape, generator=generator[i], device=rand_device, dtype=text_embeddings.dtype)
-                    for i in range(batch_size)
-                ]
-                latents = torch.cat(latents, dim=0)
-            else:
-                latents = torch.randn(
-                    latents_shape, generator=generator, device=rand_device, dtype=text_embeddings.dtype
-                )
-            latents = latents.to(self.device)
+            latents = randn_tensor(latents_shape, generator=generator, device=self.device, dtype=text_embeddings.dtype)
         else:
             if latents.shape != latents_shape:
                 raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {latents_shape}")

src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion_superresolution.py

@@ -16,7 +16,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import PIL_INTERPOLATION, deprecate
+from ...utils import PIL_INTERPOLATION, deprecate, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -121,12 +121,7 @@ class LDMSuperResolutionPipeline(DiffusionPipeline):
         latents_shape = (batch_size, self.unet.in_channels // 2, height, width)
         latents_dtype = next(self.unet.parameters()).dtype
 
-        if self.device.type == "mps":
-            # randn does not work reproducibly on mps
-            latents = torch.randn(latents_shape, generator=generator, device="cpu", dtype=latents_dtype)
-            latents = latents.to(self.device)
-        else:
-            latents = torch.randn(latents_shape, generator=generator, device=self.device, dtype=latents_dtype)
+        latents = randn_tensor(latents_shape, generator=generator, device=self.device, dtype=latents_dtype)
 
         image = image.to(device=self.device, dtype=latents_dtype)
 

src/diffusers/pipelines/latent_diffusion_uncond/pipeline_latent_diffusion_uncond.py

@@ -19,6 +19,7 @@ import torch
 
 from ...models import UNet2DModel, VQModel
 from ...schedulers import DDIMScheduler
+from ...utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -71,7 +72,7 @@ class LDMPipeline(DiffusionPipeline):
             True, otherwise a `tuple. When returning a tuple, the first element is a list with the generated images.
         """
 
-        latents = torch.randn(
+        latents = randn_tensor(
             (batch_size, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size),
             generator=generator,
         )

src/diffusers/pipelines/paint_by_example/image_encoder.py

@@ -34,7 +34,7 @@ class PaintByExampleImageEncoder(CLIPPreTrainedModel):
         self.proj_out = nn.Linear(config.hidden_size, self.proj_size)
 
         # uncondition for scaling
-        self.uncond_vector = nn.Parameter(torch.rand((1, 1, self.proj_size)))
+        self.uncond_vector = nn.Parameter(torch.randn((1, 1, self.proj_size)))
 
     def forward(self, pixel_values):
         clip_output = self.model(pixel_values=pixel_values)

src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py

@@ -24,7 +24,7 @@ from transformers import CLIPFeatureExtractor
 
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler
-from ...utils import logging
+from ...utils import logging, randn_tensor
 from ..pipeline_utils import DiffusionPipeline
 from ..stable_diffusion import StableDiffusionPipelineOutput
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
@@ -300,20 +300,8 @@ class PaintByExamplePipeline(DiffusionPipeline):
             )
 
         if latents is None:
-            rand_device = "cpu" if device.type == "mps" else device
-
-            if isinstance(generator, list):
-                shape = (1,) + shape[1:]
-                latents = [
-                    torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype)
-                    for i in range(batch_size)
-                ]
-                latents = torch.cat(latents, dim=0).to(device)
-            else:
-                latents = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
         else:
-            if latents.shape != shape:
-                raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {shape}")
             latents = latents.to(device)
 
         # scale the initial noise by the standard deviation required by the scheduler

src/diffusers/pipelines/pndm/pipeline_pndm.py

@@ -19,6 +19,7 @@ import torch
 
 from ...models import UNet2DModel
 from ...schedulers import PNDMScheduler
+from ...utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -72,11 +73,11 @@ class PNDMPipeline(DiffusionPipeline):
         # the official paper: https://arxiv.org/pdf/2202.09778.pdf
 
         # Sample gaussian noise to begin loop
-        image = torch.randn(
+        image = randn_tensor(
             (batch_size, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size),
             generator=generator,
+            device=self.device,
         )
-        image = image.to(self.device)
 
         self.scheduler.set_timesteps(num_inference_steps)
         for t in self.progress_bar(self.scheduler.timesteps):

src/diffusers/pipelines/repaint/pipeline_repaint.py

@@ -22,7 +22,7 @@ import PIL
 
 from ...models import UNet2DModel
 from ...schedulers import RePaintScheduler
-from ...utils import PIL_INTERPOLATION, deprecate, logging
+from ...utils import PIL_INTERPOLATION, deprecate, logging, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -143,18 +143,8 @@ class RePaintPipeline(DiffusionPipeline):
                 f" size of {batch_size}. Make sure the batch size matches the length of the generators."
             )
 
-        rand_device = "cpu" if self.device.type == "mps" else self.device
         image_shape = original_image.shape
-        if isinstance(generator, list):
-            shape = (1,) + image_shape[1:]
-            image = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=self.unet.dtype)
-                for i in range(batch_size)
-            ]
-            image = torch.cat(image, dim=0).to(self.device)
-        else:
-            image = torch.randn(image_shape, generator=generator, device=rand_device, dtype=self.unet.dtype)
-            image = image.to(self.device)
+        image = randn_tensor(image_shape, generator=generator, device=self.device, dtype=self.unet.dtype)
 
         # set step values
         self.scheduler.set_timesteps(num_inference_steps, jump_length, jump_n_sample, self.device)

src/diffusers/pipelines/score_sde_ve/pipeline_score_sde_ve.py

@@ -18,6 +18,7 @@ import torch
 
 from ...models import UNet2DModel
 from ...schedulers import ScoreSdeVeScheduler
+from ...utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -69,7 +70,7 @@ class ScoreSdeVePipeline(DiffusionPipeline):
 
         model = self.unet
 
-        sample = torch.randn(*shape, generator=generator) * self.scheduler.init_noise_sigma
+        sample = randn_tensor(shape, generator=generator) * self.scheduler.init_noise_sigma
         sample = sample.to(self.device)
 
         self.scheduler.set_timesteps(num_inference_steps)

src/diffusers/pipelines/stable_diffusion/pipeline_cycle_diffusion.py

@@ -26,7 +26,7 @@ from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
 from ...configuration_utils import FrozenDict
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import DDIMScheduler
-from ...utils import PIL_INTERPOLATION, deprecate, logging
+from ...utils import PIL_INTERPOLATION, deprecate, logging, randn_tensor
 from ..pipeline_utils import DiffusionPipeline
 from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
@@ -76,7 +76,7 @@ def posterior_sample(scheduler, latents, timestep, clean_latents, generator, eta
     # direction pointing to x_t
     e_t = (latents - alpha_prod_t ** (0.5) * clean_latents) / (1 - alpha_prod_t) ** (0.5)
     dir_xt = (1.0 - alpha_prod_t_prev - std_dev_t**2) ** (0.5) * e_t
-    noise = std_dev_t * torch.randn(
+    noise = std_dev_t * randn_tensor(
         clean_latents.shape, dtype=clean_latents.dtype, device=clean_latents.device, generator=generator
     )
     prev_latents = alpha_prod_t_prev ** (0.5) * clean_latents + dir_xt + noise
@@ -472,16 +472,8 @@ class CycleDiffusionPipeline(DiffusionPipeline):
             init_latents = torch.cat([init_latents] * num_images_per_prompt, dim=0)
 
         # add noise to latents using the timestep
-        rand_device = "cpu" if device.type == "mps" else device
         shape = init_latents.shape
-        if isinstance(generator, list):
-            shape = (1,) + shape[1:]
-            noise = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype) for i in range(batch_size)
-            ]
-            noise = torch.cat(noise, dim=0).to(device)
-        else:
-            noise = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
+        noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
 
         # get latents
         clean_latents = init_latents

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py

@@ -30,7 +30,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import deprecate, is_accelerate_available, logging, replace_example_docstring
+from ...utils import deprecate, is_accelerate_available, logging, randn_tensor, replace_example_docstring
 from ..pipeline_utils import DiffusionPipeline
 from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
@@ -398,20 +398,8 @@ class StableDiffusionPipeline(DiffusionPipeline):
             )
 
         if latents is None:
-            rand_device = "cpu" if device.type == "mps" else device
-
-            if isinstance(generator, list):
-                shape = (1,) + shape[1:]
-                latents = [
-                    torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype)
-                    for i in range(batch_size)
-                ]
-                latents = torch.cat(latents, dim=0).to(device)
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
         else:
-                latents = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
-        else:
-            if latents.shape != shape:
-                raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {shape}")
             latents = latents.to(device)
 
         # scale the initial noise by the standard deviation required by the scheduler

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py

@@ -20,7 +20,6 @@ import numpy as np
 import torch
 
 import PIL
-from diffusers.utils import is_accelerate_available
 from packaging import version
 from transformers import CLIPTextModel, CLIPTokenizer, DPTFeatureExtractor, DPTForDepthEstimation
 
@@ -34,7 +33,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import PIL_INTERPOLATION, deprecate, logging
+from ...utils import PIL_INTERPOLATION, deprecate, is_accelerate_available, logging, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
@@ -381,16 +380,8 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline):
         else:
             init_latents = torch.cat([init_latents], dim=0)
 
-        rand_device = "cpu" if device.type == "mps" else device
         shape = init_latents.shape
-        if isinstance(generator, list):
-            shape = (1,) + shape[1:]
-            noise = [
-                torch.randn(shape, generator=generator[i], device=rand_device, dtype=dtype) for i in range(batch_size)
-            ]
-            noise = torch.cat(noise, dim=0).to(device)
-        else:
-            noise = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
+        noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
 
         # get latents
         init_latents = self.scheduler.add_noise(init_latents, noise, timestep)