[Docs] Update Imagen Video paper link in schedulers (#12724)

docs: Update Imagen Video paper link in scheduler docstrings.

src/diffusers/schedulers/scheduling_cosine_dpmsolver_multistep.py

@@ -53,7 +53,7 @@ class CosineDPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `v_prediction`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         solver_type (`str`, defaults to `midpoint`):
             Solver type for the second-order solver; can be `midpoint` or `heun`. The solver type slightly affects the
             sample quality, especially for a small number of steps. It is recommended to use `midpoint` solvers.

src/diffusers/schedulers/scheduling_ddim_cogvideox.py

@@ -157,7 +157,7 @@ class CogVideoXDDIMScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_ddim_inverse.py

@@ -160,7 +160,7 @@ class DDIMInverseScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         timestep_spacing (`str`, defaults to `"leading"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.

src/diffusers/schedulers/scheduling_ddim_parallel.py

@@ -164,7 +164,7 @@ class DDIMParallelScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, default `epsilon`, optional):
             prediction type of the scheduler function, one of `epsilon` (predicting the noise of the diffusion
             process), `sample` (directly predicting the noisy sample`) or `v_prediction` (see section 2.4
-            https://imagen.research.google/video/paper.pdf)
+            https://huggingface.co/papers/2210.02303)
         thresholding (`bool`, default `False`):
             whether to use the "dynamic thresholding" method (introduced by Imagen,
             https://huggingface.co/papers/2205.11487). Note that the thresholding method is unsuitable for latent-space

src/diffusers/schedulers/scheduling_ddpm.py

@@ -154,7 +154,7 @@ class DDPMScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`"epsilon"`, `"sample"`, or `"v_prediction"`, defaults to `"epsilon"`):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_ddpm_parallel.py

@@ -160,7 +160,7 @@ class DDPMParallelScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, default `epsilon`, optional):
             prediction type of the scheduler function, one of `epsilon` (predicting the noise of the diffusion
             process), `sample` (directly predicting the noisy sample`) or `v_prediction` (see section 2.4
-            https://imagen.research.google/video/paper.pdf)
+            https://huggingface.co/papers/2210.02303)
         thresholding (`bool`, default `False`):
             whether to use the "dynamic thresholding" method (introduced by Imagen,
             https://huggingface.co/papers/2205.11487). Note that the thresholding method is unsuitable for latent-space

src/diffusers/schedulers/scheduling_deis_multistep.py

@@ -101,7 +101,7 @@ class DEISMultistepScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_dpm_cogvideox.py

@@ -158,7 +158,7 @@ class CogVideoXDPMScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_dpmsolver_multistep_inverse.py

@@ -101,7 +101,7 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_dpmsolver_sde.py

@@ -182,7 +182,7 @@ class DPMSolverSDEScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.

src/diffusers/schedulers/scheduling_dpmsolver_singlestep.py

@@ -103,7 +103,7 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_edm_dpmsolver_multistep.py

@@ -57,7 +57,7 @@ class EDMDPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_edm_euler.py

@@ -74,7 +74,7 @@ class EDMEulerScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         rho (`float`, *optional*, defaults to 7.0):
             The rho parameter used for calculating the Karras sigma schedule, which is set to 7.0 in the EDM paper [1].
         final_sigmas_type (`str`, defaults to `"zero"`):

src/diffusers/schedulers/scheduling_euler_ancestral_discrete.py

@@ -152,7 +152,7 @@ class EulerAncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.

src/diffusers/schedulers/scheduling_euler_discrete.py

@@ -155,7 +155,7 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`Literal["epsilon", "sample", "v_prediction"]`, defaults to `"epsilon"`, *optional*):
             Prediction type of the scheduler function; can be `"epsilon"` (predicts the noise of the diffusion
             process), `"sample"` (directly predicts the noisy sample`) or `"v_prediction"` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         interpolation_type (`Literal["linear", "log_linear"]`, defaults to `"linear"`, *optional*):
             The interpolation type to compute intermediate sigmas for the scheduler denoising steps. Should be one of
             `"linear"` or `"log_linear"`.

src/diffusers/schedulers/scheduling_euler_discrete_flax.py

@@ -74,7 +74,7 @@ class FlaxEulerDiscreteScheduler(FlaxSchedulerMixin, ConfigMixin):
         prediction_type (`str`, default `epsilon`, optional):
             prediction type of the scheduler function, one of `epsilon` (predicting the noise of the diffusion
             process), `sample` (directly predicting the noisy sample`) or `v_prediction` (see section 2.4
-            https://imagen.research.google/video/paper.pdf)
+            https://huggingface.co/papers/2210.02303)
         dtype (`jnp.dtype`, *optional*, defaults to `jnp.float32`):
             the `dtype` used for params and computation.
     """

src/diffusers/schedulers/scheduling_heun_discrete.py

@@ -115,7 +115,7 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         clip_sample (`bool`, defaults to `True`):
             Clip the predicted sample for numerical stability.
         clip_sample_range (`float`, defaults to 1.0):

src/diffusers/schedulers/scheduling_k_dpm_2_ancestral_discrete.py

@@ -125,7 +125,7 @@ class KDPM2AncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.

src/diffusers/schedulers/scheduling_k_dpm_2_discrete.py

@@ -124,7 +124,7 @@ class KDPM2DiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.

src/diffusers/schedulers/scheduling_lcm.py

@@ -170,7 +170,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_lms_discrete.py

@@ -120,7 +120,7 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`"epsilon"`, `"sample"`, or `"v_prediction"`, defaults to `"epsilon"`):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         timestep_spacing (`"linspace"`, `"leading"`, or `"trailing"`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.

src/diffusers/schedulers/scheduling_lms_discrete_flax.py

@@ -77,7 +77,7 @@ class FlaxLMSDiscreteScheduler(FlaxSchedulerMixin, ConfigMixin):
         prediction_type (`str`, default `epsilon`, optional):
             prediction type of the scheduler function, one of `epsilon` (predicting the noise of the diffusion
             process), `sample` (directly predicting the noisy sample`) or `v_prediction` (see section 2.4
-            https://imagen.research.google/video/paper.pdf)
+            https://huggingface.co/papers/2210.02303)
         dtype (`jnp.dtype`, *optional*, defaults to `jnp.float32`):
             the `dtype` used for params and computation.
     """

src/diffusers/schedulers/scheduling_pndm_flax.py

@@ -103,7 +103,7 @@ class FlaxPNDMScheduler(FlaxSchedulerMixin, ConfigMixin):
         prediction_type (`str`, default `epsilon`, optional):
             prediction type of the scheduler function, one of `epsilon` (predicting the noise of the diffusion
             process), `sample` (directly predicting the noisy sample`) or `v_prediction` (see section 2.4
-            https://imagen.research.google/video/paper.pdf)
+            https://huggingface.co/papers/2210.02303)
         dtype (`jnp.dtype`, *optional*, defaults to `jnp.float32`):
             the `dtype` used for params and computation.
     """

src/diffusers/schedulers/scheduling_sasolver.py

@@ -105,7 +105,7 @@ class SASolverScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         tau_func (`Callable`, *optional*):
             Stochasticity during the sampling. Default in init is `lambda t: 1 if t >= 200 and t <= 800 else 0`.
             SA-Solver will sample from vanilla diffusion ODE if tau_func is set to `lambda t: 0`. SA-Solver will sample

src/diffusers/schedulers/scheduling_tcd.py

@@ -171,7 +171,7 @@ class TCDScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.

src/diffusers/schedulers/scheduling_unipc_multistep.py

@@ -139,7 +139,7 @@ class UniPCMultistepScheduler(SchedulerMixin, ConfigMixin):
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
-            Video](https://imagen.research.google/video/paper.pdf) paper).
+            Video](https://huggingface.co/papers/2210.02303) paper).
         thresholding (`bool`, defaults to `False`):
             Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
             as Stable Diffusion.