From cbe91251ac39dc2df7993837f6eb89830bfd808c Mon Sep 17 00:00:00 2001
From: Guillaume Mercier <C0bra5@users.noreply.github.com>
Date: Wed, 7 Dec 2022 00:54:16 -0500
Subject: [PATCH] Hypernetwork support (#619)

* Update README.md

* Update README.md

* Make on_sd_start.sh executable

* Merge pull request #542 from patriceac/patch-1

Fix restoration of model and VAE

* Merge pull request #541 from patriceac/patch-2

Fix restoration of parallel output setting

* Hypernetwork support

Adds support for hypernetworks. Hypernetworks are stored in /models/hypernetworks

* forgot to remove unused code

Co-authored-by: cmdr2 <secondary.cmdr2@gmail.com>
---
 README.md                      |   4 +
 scripts/on_sd_start.bat        |   2 +
 scripts/on_sd_start.sh         |   2 +
 ui/index.html                  |   6 +
 ui/media/js/auto-save.js       |   2 +
 ui/media/js/main.js            |  37 +++++-
 ui/sd_internal/__init__.py     |   7 ++
 ui/sd_internal/hypernetwork.py | 198 +++++++++++++++++++++++++++++++++
 ui/sd_internal/runtime.py      |  54 +++++++++
 ui/sd_internal/task_manager.py |  21 +++-
 ui/server.py                   |  20 +++-
 11 files changed, 346 insertions(+), 7 deletions(-)
 create mode 100644 ui/sd_internal/hypernetwork.py

diff --git a/README.md b/README.md
index 2038c15b..01509a46 100644
--- a/README.md
+++ b/README.md
@@ -3,6 +3,8 @@
 
 [![Discord Server](https://img.shields.io/discord/1014774730907209781?label=Discord)](https://discord.com/invite/u9yhsFmEkB) (for support, and development discussion) | [Troubleshooting guide for common problems](Troubleshooting.md)
 
+New! Experimental support for Stable Diffusion 2.0 is available in beta!
+
 ----
 
 ## Step 1: Download the installer
@@ -28,7 +30,9 @@ The installer will take care of whatever is needed. A friendly [Discord communit
 - **No Dependencies or Technical Knowledge Required**: 1-click install for Windows 10/11 and Linux. *No dependencies*, no need for WSL or Docker or Conda or technical setup. Just download and run!
 - **Clutter-free UI**: a friendly and simple UI, while providing a lot of powerful features
 - Supports "*Text to Image*" and "*Image to Image*"
+- **Stable Diffusion 2.0 support (experimental)** - available in beta channel
 - **Custom Models**: Use your own `.ckpt` file, by placing it inside the `models/stable-diffusion` folder!
+- **Auto scan for malicious models** - uses picklescan to prevent malicious models
 - **Live Preview**: See the image as the AI is drawing it
 - **Task Queue**: Queue up all your ideas, without waiting for the current task to finish
 - **In-Painting**: Specify areas of your image to paint into
diff --git a/scripts/on_sd_start.bat b/scripts/on_sd_start.bat
index 99fb74bc..2daa39e2 100644
--- a/scripts/on_sd_start.bat
+++ b/scripts/on_sd_start.bat
@@ -201,8 +201,10 @@ call WHERE uvicorn > .tmp
 
 if not exist "..\models\stable-diffusion" mkdir "..\models\stable-diffusion"
 if not exist "..\models\vae" mkdir "..\models\vae"
+if not exist "..\models\hypernetwork" mkdir "..\models\hypernetwork"
 echo. > "..\models\stable-diffusion\Put your custom ckpt files here.txt"
 echo. > "..\models\vae\Put your VAE files here.txt"
+echo. > "..\models\hypernetwork\Put your hypernetwork files here.txt"
 
 @if exist "sd-v1-4.ckpt" (
     for %%I in ("sd-v1-4.ckpt") do if "%%~zI" EQU "4265380512" (
diff --git a/scripts/on_sd_start.sh b/scripts/on_sd_start.sh
index 177e4f73..f8f3d560 100755
--- a/scripts/on_sd_start.sh
+++ b/scripts/on_sd_start.sh
@@ -161,8 +161,10 @@ fi
 
 mkdir -p "../models/stable-diffusion"
 mkdir -p "../models/vae"
+mkdir -p "../models/hypernetwork"
 echo "" > "../models/stable-diffusion/Put your custom ckpt files here.txt"
 echo "" > "../models/vae/Put your VAE files here.txt"
+echo "" > "../models/hypernetwork/Put your hypernetwork files here.txt"
 
 if [ -f "sd-v1-4.ckpt" ]; then
     model_size=`find "sd-v1-4.ckpt" -printf "%s"`
diff --git a/ui/index.html b/ui/index.html
index 50869fc7..d3fb6da3 100644
--- a/ui/index.html
+++ b/ui/index.html
@@ -131,6 +131,12 @@
                         </select>
                         <a href="https://github.com/cmdr2/stable-diffusion-ui/wiki/VAE-Variational-Auto-Encoder" target="_blank"><i class="fa-solid fa-circle-question help-btn"><span class="simple-tooltip right">Click to learn more about VAEs</span></i></a>
                     </td></tr>
+                    <tr class="pl-5"><td><label for="hypernetwork_model">Hypernetwork:</i></label></td><td>
+                        <select id="hypernetwork_model" name="hypernetwork_model">
+                            <!-- <option value="" selected>None</option> -->
+                        </select>
+                    </td></tr>
+                    <tr id="hypernetwork_strength_container" class="pl-5"><td><label for="hypernetwork_strength_slider">Hypernetwork Strength:</label></td><td> <input id="hypernetwork_strength_slider" name="hypernetwork_strength_slider" class="editor-slider" value="100" type="range" min="0" max="100"> <input id="hypernetwork_strength" name="hypernetwork_strength" size="4" pattern="^[0-9\.]+$" onkeypress="preventNonNumericalInput(event)"><br/></td></tr></span>
                     <tr id="samplerSelection" class="pl-5"><td><label for="sampler">Sampler:</label></td><td>
                         <select id="sampler" name="sampler">
                             <option value="plms">plms</option>
diff --git a/ui/media/js/auto-save.js b/ui/media/js/auto-save.js
index 2d1543cc..f503779a 100644
--- a/ui/media/js/auto-save.js
+++ b/ui/media/js/auto-save.js
@@ -14,12 +14,14 @@ const SETTINGS_IDS_LIST = [
     "num_outputs_parallel",
     "stable_diffusion_model",
     "vae_model",
+    "hypernetwork_model",
     "sampler",
     "width",
     "height",
     "num_inference_steps",
     "guidance_scale",
     "prompt_strength",
+    "hypernetwork_strength",
     "output_format",
     "output_quality",
     "negative_prompt",
diff --git a/ui/media/js/main.js b/ui/media/js/main.js
index 02ea6a5e..de5c988b 100644
--- a/ui/media/js/main.js
+++ b/ui/media/js/main.js
@@ -35,6 +35,9 @@ let useUpscalingField = document.querySelector("#use_upscale")
 let upscaleModelField = document.querySelector("#upscale_model")
 let stableDiffusionModelField = document.querySelector('#stable_diffusion_model')
 let vaeModelField = document.querySelector('#vae_model')
+let hypernetworkModelField = document.querySelector('#hypernetwork_model')
+let hypernetworkStrengthSlider = document.querySelector('#hypernetwork_strength_slider')
+let hypernetworkStrengthField = document.querySelector('#hypernetwork_strength')
 let outputFormatField = document.querySelector('#output_format')
 let showOnlyFilteredImageField = document.querySelector("#show_only_filtered_image")
 let updateBranchLabel = document.querySelector("#updateBranchLabel")
@@ -654,6 +657,7 @@ function onTaskCompleted(task, reqBody, instance, outputContainer, stepUpdate) {
     }
 }
 
+
 function onTaskStart(task) {
     if (!task.isProcessing || task.batchesDone >= task.batchCount) {
         return
@@ -750,6 +754,10 @@ function createTask(task) {
     if (task.reqBody.use_upscale) {
         taskConfig += `, <b>Upscale:</b> ${task.reqBody.use_upscale}`
     }
+    if (task.reqBody.use_hypernetwork_model) {
+        taskConfig += `, <b>Hypernetwork:</b> ${task.reqBody.use_hypernetwork_model}`
+        taskConfig += `, <b>Hypernetwork Strength:</b> ${task.reqBody.hypernetwork_strength}`
+    }
 
     let taskEntry = document.createElement('div')
     taskEntry.id = `imageTaskContainer-${Date.now()}`
@@ -1105,6 +1113,27 @@ promptStrengthSlider.addEventListener('input', updatePromptStrength)
 promptStrengthField.addEventListener('input', updatePromptStrengthSlider)
 updatePromptStrength()
 
+/********************* Hypernetwork Strength **********************/
+function updateHypernetworkStrength() {
+    hypernetworkStrengthField.value = hypernetworkStrengthSlider.value / 100
+    hypernetworkStrengthField.dispatchEvent(new Event("change"))
+}
+
+function updateHypernetworkStrengthSlider() {
+    if (hypernetworkStrengthField.value < 0) {
+        hypernetworkStrengthField.value = 0
+    } else if (hypernetworkStrengthField.value > 0.99) {
+        hypernetworkStrengthField.value = 0.99
+    }
+
+    hypernetworkStrengthSlider.value = hypernetworkStrengthField.value * 100
+    hypernetworkStrengthSlider.dispatchEvent(new Event("change"))
+}
+
+hypernetworkStrengthSlider.addEventListener('input', updateHypernetworkStrength)
+hypernetworkStrengthField.addEventListener('input', updateHypernetworkStrengthSlider)
+updateHypernetworkStrength()
+
 /********************* JPEG Quality **********************/
 function updateOutputQuality() {
     outputQualityField.value =  0 | outputQualitySlider.value
@@ -1138,12 +1167,14 @@ async function getModels() {
     try {
         const sd_model_setting_key = "stable_diffusion_model"
         const vae_model_setting_key = "vae_model"
+        const hypernetwork_model_key = "hypernetwork_model"
         const selectedSDModel = SETTINGS[sd_model_setting_key].value
         const selectedVaeModel = SETTINGS[vae_model_setting_key].value
+        const selectedHypernetworkModel = SETTINGS[hypernetwork_model_key].value
 
         const models = await SD.getModels()
         const modelsOptions = models['options']
-        if ( "scan-error" in models) {
+        if ("scan-error" in models) {
             // let previewPane = document.getElementById('tab-content-wrapper')
             let previewPane = document.getElementById('preview')
             previewPane.style.background="red"
@@ -1154,7 +1185,10 @@ async function getModels() {
 
         const stableDiffusionOptions = modelsOptions['stable-diffusion']
         const vaeOptions = modelsOptions['vae']
+        const hypernetworkOptions = modelOptions['hypernetwork']
+
         vaeOptions.unshift('') // add a None option
+        hypernetworkOptions.unshift('') // add a None option
 
         function createModelOptions(modelField, selectedModel) {
             return function(modelName) {
@@ -1172,6 +1206,7 @@ async function getModels() {
 
         stableDiffusionOptions.forEach(createModelOptions(stableDiffusionModelField, selectedSDModel))
         vaeOptions.forEach(createModelOptions(vaeModelField, selectedVaeModel))
+        hypernetworkOptions.forEach(createModelOptions(hypernetworkModelField, selectedHypernetworkModel))
 
         // TODO: set default for model here too
         SETTINGS[sd_model_setting_key].default = stableDiffusionOptions[0]
diff --git a/ui/sd_internal/__init__.py b/ui/sd_internal/__init__.py
index 18b0a01b..a2abe294 100644
--- a/ui/sd_internal/__init__.py
+++ b/ui/sd_internal/__init__.py
@@ -23,6 +23,8 @@ class Request:
     use_upscale: str = None # or "RealESRGAN_x4plus" or "RealESRGAN_x4plus_anime_6B"
     use_stable_diffusion_model: str = "sd-v1-4"
     use_vae_model: str = None
+    use_hypernetwork_model: str = None
+    hypernetwork_strength: float = 1
     show_only_filtered_image: bool = False
     output_format: str = "jpeg" # or "png"
     output_quality: int = 75
@@ -38,6 +40,7 @@ class Request:
             "num_outputs": self.num_outputs,
             "num_inference_steps": self.num_inference_steps,
             "guidance_scale": self.guidance_scale,
+            "hypernetwork_strengtgh": self.guidance_scale,
             "width": self.width,
             "height": self.height,
             "seed": self.seed,
@@ -47,6 +50,8 @@ class Request:
             "use_upscale": self.use_upscale,
             "use_stable_diffusion_model": self.use_stable_diffusion_model,
             "use_vae_model": self.use_vae_model,
+            "use_hypernetwork_model": self.use_hypernetwork_model,
+            "hypernetwork_strength": self.hypernetwork_strength,
             "output_format": self.output_format,
             "output_quality": self.output_quality,
         }
@@ -70,6 +75,8 @@ class Request:
     use_upscale: {self.use_upscale}
     use_stable_diffusion_model: {self.use_stable_diffusion_model}
     use_vae_model: {self.use_vae_model}
+    use_hypernetwork_model: {self.use_hypernetwork_model}
+    hypernetwork_strength: {self.hypernetwork_strength}
     show_only_filtered_image: {self.show_only_filtered_image}
     output_format: {self.output_format}
     output_quality: {self.output_quality}
diff --git a/ui/sd_internal/hypernetwork.py b/ui/sd_internal/hypernetwork.py
new file mode 100644
index 00000000..979a74f3
--- /dev/null
+++ b/ui/sd_internal/hypernetwork.py
@@ -0,0 +1,198 @@
+# this is basically a cut down version of https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/c9a2cfdf2a53d37c2de1908423e4f548088667ef/modules/hypernetworks/hypernetwork.py, mostly for feature parity
+# I, c0bra5, don't really understand how deep learning works. I just know how to port stuff.
+
+import inspect
+import torch
+import optimizedSD.splitAttention
+from . import runtime
+from einops import rearrange
+
+optimizer_dict = {optim_name : cls_obj for optim_name, cls_obj in inspect.getmembers(torch.optim, inspect.isclass) if optim_name != "Optimizer"}
+
+loaded_hypernetwork = None
+
+class HypernetworkModule(torch.nn.Module):
+    multiplier = 0.5
+    activation_dict = {
+        "linear": torch.nn.Identity,
+        "relu": torch.nn.ReLU,
+        "leakyrelu": torch.nn.LeakyReLU,
+        "elu": torch.nn.ELU,
+        "swish": torch.nn.Hardswish,
+        "tanh": torch.nn.Tanh,
+        "sigmoid": torch.nn.Sigmoid,
+    }
+    activation_dict.update({cls_name.lower(): cls_obj for cls_name, cls_obj in inspect.getmembers(torch.nn.modules.activation) if inspect.isclass(cls_obj) and cls_obj.__module__ == 'torch.nn.modules.activation'})
+
+    def __init__(self, dim, state_dict=None, layer_structure=None, activation_func=None, weight_init='Normal',
+                 add_layer_norm=False, use_dropout=False, activate_output=False, last_layer_dropout=False):
+        super().__init__()
+
+        assert layer_structure is not None, "layer_structure must not be None"
+        assert layer_structure[0] == 1, "Multiplier Sequence should start with size 1!"
+        assert layer_structure[-1] == 1, "Multiplier Sequence should end with size 1!"
+
+        linears = []
+        for i in range(len(layer_structure) - 1):
+
+            # Add a fully-connected layer
+            linears.append(torch.nn.Linear(int(dim * layer_structure[i]), int(dim * layer_structure[i+1])))
+
+            # Add an activation func except last layer
+            if activation_func == "linear" or activation_func is None or (i >= len(layer_structure) - 2 and not activate_output):
+                pass
+            elif activation_func in self.activation_dict:
+                linears.append(self.activation_dict[activation_func]())
+            else:
+                raise RuntimeError(f'hypernetwork uses an unsupported activation function: {activation_func}')
+
+            # Add layer normalization
+            if add_layer_norm:
+                linears.append(torch.nn.LayerNorm(int(dim * layer_structure[i+1])))
+
+            # Add dropout except last layer
+            if use_dropout and (i < len(layer_structure) - 3 or last_layer_dropout and i < len(layer_structure) - 2):
+                linears.append(torch.nn.Dropout(p=0.3))
+
+        self.linear = torch.nn.Sequential(*linears)
+
+        self.fix_old_state_dict(state_dict)
+        self.load_state_dict(state_dict)
+
+        self.to(runtime.thread_data.device)
+
+    def fix_old_state_dict(self, state_dict):
+        changes = {
+            'linear1.bias': 'linear.0.bias',
+            'linear1.weight': 'linear.0.weight',
+            'linear2.bias': 'linear.1.bias',
+            'linear2.weight': 'linear.1.weight',
+        }
+
+        for fr, to in changes.items():
+            x = state_dict.get(fr, None)
+            if x is None:
+                continue
+
+            del state_dict[fr]
+            state_dict[to] = x
+
+    def forward(self, x: torch.Tensor):
+        return x + self.linear(x) * runtime.thread_data.hypernetwork_strength
+
+def apply_hypernetwork(hypernetwork, context, layer=None):
+    hypernetwork_layers = hypernetwork.get(context.shape[2], None)
+
+    if hypernetwork_layers is None:
+        return context, context
+
+    if layer is not None:
+        layer.hyper_k = hypernetwork_layers[0]
+        layer.hyper_v = hypernetwork_layers[1]
+
+    context_k = hypernetwork_layers[0](context)
+    context_v = hypernetwork_layers[1](context)
+    return context_k, context_v
+
+def get_kv(context, hypernetwork):
+    if hypernetwork is None:
+        return context, context
+    else:
+        return apply_hypernetwork(runtime.thread_data.hypernetwork, context)
+
+# This might need updating as the optimisedSD code changes
+# I think yall have a system for this (patch files in sd_internal) but idk how it works and no amount of searching gave me any clue
+# just in case for attribution https://github.com/easydiffusion/diffusion-kit/blob/e8ea0cadd543056059cd951e76d4744de76327d2/optimizedSD/splitAttention.py#L171
+def new_cross_attention_forward(self, x, context=None, mask=None):
+    h = self.heads
+
+    q = self.to_q(x)
+    # default context
+    context = context if context is not None else x() if inspect.isfunction(x) else x
+    # hypernetwork!
+    context_k, context_v =  get_kv(context, runtime.thread_data.hypernetwork)
+    k = self.to_k(context_k)
+    v = self.to_v(context_v)
+    del context, x
+
+    q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v))
+
+
+    limit = k.shape[0]
+    att_step = self.att_step
+    q_chunks = list(torch.tensor_split(q, limit//att_step, dim=0))
+    k_chunks = list(torch.tensor_split(k, limit//att_step, dim=0))
+    v_chunks = list(torch.tensor_split(v, limit//att_step, dim=0))
+
+    q_chunks.reverse()
+    k_chunks.reverse()
+    v_chunks.reverse()
+    sim = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device)
+    del k, q, v
+    for i in range (0, limit, att_step):
+
+        q_buffer = q_chunks.pop()
+        k_buffer = k_chunks.pop()
+        v_buffer = v_chunks.pop()
+        sim_buffer = torch.einsum('b i d, b j d -> b i j', q_buffer, k_buffer) * self.scale
+
+        del k_buffer, q_buffer
+        # attention, what we cannot get enough of, by chunks
+
+        sim_buffer = sim_buffer.softmax(dim=-1)
+
+        sim_buffer = torch.einsum('b i j, b j d -> b i d', sim_buffer, v_buffer)
+        del v_buffer
+        sim[i:i+att_step,:,:] = sim_buffer
+
+        del sim_buffer
+    sim = rearrange(sim, '(b h) n d -> b n (h d)', h=h)
+    return self.to_out(sim)
+
+
+def load_hypernetwork(path: str):
+
+    state_dict = torch.load(path, map_location='cpu')
+
+    layer_structure = state_dict.get('layer_structure', [1, 2, 1])
+    activation_func = state_dict.get('activation_func', None)
+    weight_init = state_dict.get('weight_initialization', 'Normal')
+    add_layer_norm = state_dict.get('is_layer_norm', False)
+    use_dropout = state_dict.get('use_dropout', False)
+    activate_output = state_dict.get('activate_output', True)
+    last_layer_dropout = state_dict.get('last_layer_dropout', False)
+    # this is a bit verbose so leaving it commented out for the poor soul who ever has to debug this
+    # print(f"layer_structure: {layer_structure}")
+    # print(f"activation_func: {activation_func}")
+    # print(f"weight_init: {weight_init}")
+    # print(f"add_layer_norm: {add_layer_norm}")
+    # print(f"use_dropout: {use_dropout}")
+    # print(f"activate_output: {activate_output}")
+    # print(f"last_layer_dropout: {last_layer_dropout}")
+
+    layers = {}
+    for size, sd in state_dict.items():
+        if type(size) == int:
+            layers[size] = (
+                HypernetworkModule(size, sd[0], layer_structure, activation_func, weight_init, add_layer_norm,
+                                   use_dropout, activate_output, last_layer_dropout=last_layer_dropout),
+                HypernetworkModule(size, sd[1], layer_structure, activation_func, weight_init, add_layer_norm,
+                                   use_dropout, activate_output, last_layer_dropout=last_layer_dropout),
+            )
+    print(f"hypernetwork loaded")
+    return layers
+
+
+
+# overriding of original function
+old_cross_attention_forward = optimizedSD.splitAttention.CrossAttention.forward
+# hijacks the cross attention forward function to add hyper network support
+def hijack_cross_attention():
+    print("hypernetwork functionality added to cross attention")
+    optimizedSD.splitAttention.CrossAttention.forward = new_cross_attention_forward
+# there was a cop on board
+def unhijack_cross_attention_forward():
+    print("hypernetwork functionality removed from cross attention")
+    optimizedSD.splitAttention.CrossAttention.forward = old_cross_attention_forward
+
+hijack_cross_attention()
\ No newline at end of file
diff --git a/ui/sd_internal/runtime.py b/ui/sd_internal/runtime.py
index 0f491865..2e0f8ce0 100644
--- a/ui/sd_internal/runtime.py
+++ b/ui/sd_internal/runtime.py
@@ -28,6 +28,9 @@ from gfpgan import GFPGANer
 from basicsr.archs.rrdbnet_arch import RRDBNet
 from realesrgan import RealESRGANer
 
+from . import hypernetwork
+from server import HYPERNETWORK_MODEL_EXTENSIONS# , STABLE_DIFFUSION_MODEL_EXTENSIONS, VAE_MODEL_EXTENSIONS
+
 from threading import Lock
 from safetensors.torch import load_file
 
@@ -57,12 +60,15 @@ def thread_init(device):
 
     thread_data.ckpt_file = None
     thread_data.vae_file = None
+    thread_data.hypernetwork_file = None
     thread_data.gfpgan_file = None
     thread_data.real_esrgan_file = None
 
     thread_data.model = None
     thread_data.modelCS = None
     thread_data.modelFS = None
+    thread_data.hypernetwork = None
+    thread_data.hypernetwork_strength = 1
     thread_data.model_gfpgan = None
     thread_data.model_real_esrgan = None
 
@@ -72,6 +78,8 @@ def thread_init(device):
     thread_data.device_name = None
     thread_data.unet_bs = 1
     thread_data.precision = 'autocast'
+    thread_data.sampler_plms = None
+    thread_data.sampler_ddim = None
 
     thread_data.turbo = False
     thread_data.force_full_precision = False
@@ -433,6 +441,49 @@ def reload_model():
     unload_filters()
     load_model_ckpt()
 
+def is_hypernetwork_reload_necessary(req: Request):
+    needs_model_reload = False
+    if thread_data.hypernetwork_file != req.use_hypernetwork_model:
+        thread_data.hypernetwork_file = req.use_hypernetwork_model
+        needs_model_reload = True
+
+    return needs_model_reload
+
+def load_hypernetwork():
+    if thread_data.hypernetwork_file is not None:
+        try:
+            loaded = False
+            for model_extension in HYPERNETWORK_MODEL_EXTENSIONS:
+                if os.path.exists(thread_data.hypernetwork_file + model_extension):
+                    print(f"Loading hypernetwork weights from: {thread_data.hypernetwork_file}{model_extension}")
+                    thread_data.hypernetwork = hypernetwork.load_hypernetwork(thread_data.hypernetwork_file + model_extension)
+                    loaded = True
+                    break
+
+            if not loaded:
+                print(f'Cannot find hypernetwork: {thread_data.hypernetwork_file}')
+                thread_data.hypernetwork_file = None
+        except:
+            print(traceback.format_exc())
+            print(f'Could not load hypernetwork: {thread_data.hypernetwork_file}')
+            thread_data.hypernetwork_file = None
+
+def unload_hypernetwork():
+    if thread_data.hypernetwork is not None:
+        print('Unloading hypernetwork...')
+        if thread_data.device != 'cpu':
+            for i in thread_data.hypernetwork:
+                thread_data.hypernetwork[i][0].to('cpu')
+                thread_data.hypernetwork[i][1].to('cpu')
+        del thread_data.hypernetwork
+    thread_data.hypernetwork = None
+
+    gc()
+
+def reload_hypernetwork():
+    unload_hypernetwork()
+    load_hypernetwork()
+
 def mk_img(req: Request, data_queue: queue.Queue, task_temp_images: list, step_callback):
     try:
         return do_mk_img(req, data_queue, task_temp_images, step_callback)
@@ -509,6 +560,7 @@ def do_mk_img(req: Request, data_queue: queue.Queue, task_temp_images: list, ste
     res = Response()
     res.request = req
     res.images = []
+    thread_data.hypernetwork_strength = req.hypernetwork_strength
 
     thread_data.temp_images.clear()
 
@@ -751,6 +803,8 @@ Sampler: {req.sampler}
 Negative Prompt: {req.negative_prompt}
 Stable Diffusion model: {req.use_stable_diffusion_model + '.ckpt'}
 VAE model: {req.use_vae_model}
+Hypernetwork Model: {req.use_hypernetwork_model}
+Hypernetwork Strength: {req.hypernetwork_strength}
 '''
     try:
         with open(meta_out_path, 'w', encoding='utf-8') as f:
diff --git a/ui/sd_internal/task_manager.py b/ui/sd_internal/task_manager.py
index d21f2877..41fc00f6 100644
--- a/ui/sd_internal/task_manager.py
+++ b/ui/sd_internal/task_manager.py
@@ -77,6 +77,8 @@ class ImageRequest(BaseModel):
     use_upscale: str = None # or "RealESRGAN_x4plus" or "RealESRGAN_x4plus_anime_6B"
     use_stable_diffusion_model: str = "sd-v1-4"
     use_vae_model: str = None
+    use_hypernetwork_model: str = None
+    hypernetwork_strength: float = None
     show_only_filtered_image: bool = False
     output_format: str = "jpeg" # or "png"
     output_quality: int = 75
@@ -177,28 +179,35 @@ current_state = ServerStates.Init
 current_state_error:Exception = None
 current_model_path = None
 current_vae_path = None
+current_hypernetwork_path = None
 tasks_queue = []
 task_cache = TaskCache()
 default_model_to_load = None
 default_vae_to_load = None
+default_hypernetwork_to_load = None
 weak_thread_data = weakref.WeakKeyDictionary()
 
-def preload_model(ckpt_file_path=None, vae_file_path=None):
-    global current_state, current_state_error, current_model_path, current_vae_path
+def preload_model(ckpt_file_path=None, vae_file_path=None, hypernetwork_file_path=None):
+    global current_state, current_state_error, current_model_path, current_vae_path, current_hypernetwork_path
     if ckpt_file_path == None:
         ckpt_file_path = default_model_to_load
     if vae_file_path == None:
         vae_file_path = default_vae_to_load
+    if hypernetwork_file_path == None:
+        hypernetwork_file_path = default_hypernetwork_to_load
     if ckpt_file_path == current_model_path and vae_file_path == current_vae_path:
         return
     current_state = ServerStates.LoadingModel
     try:
         from . import runtime
+        runtime.thread_data.hypernetwork_file = hypernetwork_file_path
         runtime.thread_data.ckpt_file = ckpt_file_path
         runtime.thread_data.vae_file = vae_file_path
         runtime.load_model_ckpt()
+        runtime.load_hypernetwork()
         current_model_path = ckpt_file_path
         current_vae_path = vae_file_path
+        current_hypernetwork_path = hypernetwork_file_path
         current_state_error = None
         current_state = ServerStates.Online
     except Exception as e:
@@ -240,7 +249,7 @@ def thread_get_next_task():
         manager_lock.release()
 
 def thread_render(device):
-    global current_state, current_state_error, current_model_path, current_vae_path
+    global current_state, current_state_error, current_model_path, current_vae_path, current_hypernetwork_path
     from . import runtime
     try:
         runtime.thread_init(device)
@@ -285,6 +294,10 @@ def thread_render(device):
         print(f'Session {task.request.session_id} starting task {id(task)} on {runtime.thread_data.device_name}')
         if not task.lock.acquire(blocking=False): raise Exception('Got locked task from queue.')
         try:
+            if runtime.is_hypernetwork_reload_necessary(task.request):
+                runtime.reload_hypernetwork()
+                current_hypernetwork_path = task.request.use_hypernetwork_model
+                
             if runtime.is_model_reload_necessary(task.request):
                 current_state = ServerStates.LoadingModel
                 runtime.reload_model()
@@ -504,6 +517,8 @@ def render(req : ImageRequest):
     r.use_face_correction = req.use_face_correction
     r.use_stable_diffusion_model = req.use_stable_diffusion_model
     r.use_vae_model = req.use_vae_model
+    r.use_hypernetwork_model = req.use_hypernetwork_model
+    r.hypernetwork_strength = req.hypernetwork_strength
     r.show_only_filtered_image = req.show_only_filtered_image
     r.output_format = req.output_format
     r.output_quality = req.output_quality
diff --git a/ui/server.py b/ui/server.py
index c7760889..804994a2 100644
--- a/ui/server.py
+++ b/ui/server.py
@@ -26,6 +26,7 @@ UI_PLUGINS_SOURCES = ((CORE_UI_PLUGINS_DIR, 'core'), (USER_UI_PLUGINS_DIR, 'user
 
 STABLE_DIFFUSION_MODEL_EXTENSIONS = ['.ckpt', '.safetensors']
 VAE_MODEL_EXTENSIONS = ['.vae.pt', '.ckpt']
+HYPERNETWORK_MODEL_EXTENSIONS = ['.pt']
 
 OUTPUT_DIRNAME = "Stable Diffusion UI" # in the user's home folder
 TASK_TTL = 15 * 60 # Discard last session's task timeout
@@ -193,6 +194,12 @@ def resolve_vae_to_use(model_name:str=None):
     except:
         return None
 
+def resolve_hypernetwork_to_use(model_name:str=None):
+    try:
+        return resolve_model_to_use(model_name, model_type='hypernetwork', model_dir='hypernetwork', model_extensions=HYPERNETWORK_MODEL_EXTENSIONS, default_models=[])
+    except:
+        return None
+
 class SetAppConfigRequest(BaseModel):
     update_branch: str = None
     render_devices: Union[List[str], List[int], str, int] = None
@@ -253,10 +260,12 @@ def getModels():
         'active': {
             'stable-diffusion': 'sd-v1-4',
             'vae': '',
+            'hypernetwork': '',
         },
         'options': {
             'stable-diffusion': ['sd-v1-4'],
             'vae': [],
+            'hypernetwork': [],
         },
     }
 
@@ -288,7 +297,7 @@ def getModels():
     # custom models
     listModels(models_dirname='stable-diffusion', model_type='stable-diffusion', model_extensions=STABLE_DIFFUSION_MODEL_EXTENSIONS)
     listModels(models_dirname='vae', model_type='vae', model_extensions=VAE_MODEL_EXTENSIONS)
-
+    listModels(models_dirname='hypernetwork', model_type='hypernetwork', model_extensions=HYPERNETWORK_MODEL_EXTENSIONS)
     # legacy
     custom_weight_path = os.path.join(SD_DIR, 'custom-model.ckpt')
     if os.path.exists(custom_weight_path):
@@ -363,16 +372,19 @@ def ping(session_id:str=None):
     response['devices'] = task_manager.get_devices()
     return JSONResponse(response, headers=NOCACHE_HEADERS)
 
-def save_model_to_config(ckpt_model_name, vae_model_name):
+def save_model_to_config(ckpt_model_name, vae_model_name, hypernetwork_model_name):
     config = getConfig()
     if 'model' not in config:
         config['model'] = {}
 
     config['model']['stable-diffusion'] = ckpt_model_name
     config['model']['vae'] = vae_model_name
+    config['model']['hypernetwork'] = hypernetwork_model_name
 
     if vae_model_name is None or vae_model_name == "":
         del config['model']['vae']
+    if hypernetwork_model_name is None or hypernetwork_model_name == "":
+        del config['model']['hypernetwork']
 
     setConfig(config)
 
@@ -388,9 +400,10 @@ def update_render_devices_in_config(config, render_devices):
 @app.post('/render')
 def render(req : task_manager.ImageRequest):
     try:
-        save_model_to_config(req.use_stable_diffusion_model, req.use_vae_model)
+        save_model_to_config(req.use_stable_diffusion_model, req.use_vae_model, req.use_hypernetwork_model)
         req.use_stable_diffusion_model = resolve_ckpt_to_use(req.use_stable_diffusion_model)
         req.use_vae_model = resolve_vae_to_use(req.use_vae_model)
+        req.use_hypernetwork_model = resolve_hypernetwork_to_use(req.use_hypernetwork_model)
         new_task = task_manager.render(req)
         response = {
             'status': str(task_manager.current_state), 
@@ -469,6 +482,7 @@ getModels()
 # Start the task_manager
 task_manager.default_model_to_load = resolve_ckpt_to_use()
 task_manager.default_vae_to_load = resolve_vae_to_use()
+task_manager.default_hypernetwork_to_load = resolve_hypernetwork_to_use()
 
 def update_render_threads():
     config = getConfig()