<objecttype="text/html"data="alpha_bar_schedules.html"style="width:100%;height:420px;"></object><articleclass="docstring"><header><aclass="docstring-binding"id="Diffusers.BetaSchedules.cosine_beta_schedule"href="#Diffusers.BetaSchedules.cosine_beta_schedule"><code>Diffusers.BetaSchedules.cosine_beta_schedule</code></a> — <spanclass="docstring-category">Function</span></header><section><div><p>Cosine beta schedule.</p><p><strong>Input</strong></p><ul><li><code>T::Int</code>: number of timesteps</li><li><code>βₘₐₓ::Real=0.999f0</code>: maximum value of β</li><li><code>ϵ::Real=1.0f-3</code>: small value used to avoid division by zero</li></ul><p><strong>Output</strong></p><ul><li><code>β::Vector{Real}</code>: βₜ values at each timestep t</li></ul><p><strong>References</strong></p><ul><li><ahref="https://arxiv.org/abs/2102.09672">[2102.09672] Improved Denoising Diffusion Probabilistic Models</a></li><li><ahref="https://github.com/openai/improved-diffusion/blob/783b6740edb79fdb7d063250db2c51cc9545dcd1/improved_diffusion/gaussian_diffusion.py#L36">github:openai/improved-diffusion</a></li></ul></div><aclass="docs-sourcelink"target="_blank"href="https://github.com/Laurent2916/Diffusers.jl/blob/77961913ba5c623ca3bd7cf7603e10d12349e3f7/src/BetaSchedules/Cosine.jl#L1-L15">source</a></section></article><articleclass="docstring"><header><aclass="docstring-binding"id="Diffusers.BetaSchedules.linear_beta_schedule"href="#Diffusers.BetaSchedules.linear_beta_schedule"><code>Diffusers.BetaSchedules.linear_beta_schedule</code></a> — <spanclass="docstring-category">Function</span></header><section><div><p>Linear beta schedule.</p><p><strong>Input</strong></p><ul><li><code>T::Integer</code>: number of timesteps</li><li><code>β₁::Real=1.0f-4</code>: initial (t=1) value of β</li><li><code>β₋₁::Real=2.0f-2</code>: final (t=T) value of β</li></ul><p><strong>Output</strong></p><ul><li><code>β::Vector{Real}</code>: βₜ values at each timestep t</li></ul><p><strong>References</strong></p><ul><li><ahref="https://arxiv.org/abs/2006.11239">[2006.11239] Denoising Diffusion Probabilistic Models</a></li></ul></div><aclass="docs-sourcelink"target="_blank"href="https://github.com/Laurent2916/Diffusers.jl/blob/77961913ba5c623ca3bd7cf7603e10d12349e3f7/src/BetaSchedules/Linear.jl#L1-L14">source</a></section></article><articleclass="docstring"><header><aclass="docstring-binding"id="Diffusers.BetaSchedules.rescale_zero_terminal_snr-Tuple{AbstractArray}"href="#Diffusers.BetaSchedules.rescale_zero_terminal_snr-Tuple{AbstractArray}"><code>Diffusers.BetaSchedules.rescale_zero_terminal_snr</code></a> — <spanclass="docstring-category">Method</span></header><section><div><p>Rescale betas to have zero terminal Signal to Noise Ratio (SNR).</p><p><strong>Input</strong></p><ul><li><code>β::AbstractArray</code>: βₜ values at each timestep t</li></ul><p><strong>Output</strong></p><ul><li><code>β::Vector{Real}</code>: rescaled βₜ values at each timestep t</li></ul><p><strong>References</strong></p><ul><li><ahref="https://arxiv.org/abs/2305.08891">[2305.08891] Rescaling Diffusion Models</a> (Alg. 1)</li></ul></div><aclass="docs-sourcelink"target="_blank"href="https://github.com/Laurent2916/Diffusers.jl/blob/77961913ba5c623ca3bd7cf7603e10d12349e3f7/src/BetaSchedules/ZeroSNR.jl#L1-L12">source</a></section></article><articleclass="docstring"><header><aclass="docstring-binding"id="Diffusers.BetaSchedules.scaled_linear_beta_schedule"href="#Diffusers.BetaSchedules.scaled_linear_beta_schedule"><code>Diffusers.BetaSchedules.scaled_linear_beta_schedule</code></a> — <spanclass="docstring-category">Function</span></header><section><div><p>Scaled linear beta schedule.</p><p><strong>Input</strong></p><ul><li><code>T::Int</code>: number of timesteps</li><li><code>β₁::Real=1.0f-4</code>: initial value of β</li><li><code>β₋₁::Real=2.0f-2</code>: final value of β</li></ul><p><strong>Output</strong></p><ul><li><code>β::Vector{Real}</code>: βₜ values at each timestep t</li></ul><p><strong>References</str