chore: add cache-dit docs, bump the version (#854)

* update

* update

* update

* update

* update the transformers version

Author: lmxyy

docs/source/index.rst

@@ -25,7 +25,7 @@ Check out `DeepCompressor <github_deepcompressor_>`_ for the quantization librar
     usage/qencoder.rst
     usage/offload.rst
     usage/attention.rst
-    usage/fbcache.rst
+    usage/cache.rst
     usage/pulid.rst
     usage/ip_adapter.rst
     usage/zimage.rst

docs/source/usage/cache.rst

@@ -0,0 +1,55 @@
+Cache
+=====
+
+.. _usage-fbcache:
+
+First-Block Cache
+-----------------
+
+Nunchaku supports `First-Block Cache (FB Cache) <fbcache>`_ for faster long-step denoising. Example usage:
+
+.. literalinclude:: ../../../examples/flux.1-dev-cache.py
+   :language: python
+   :caption: Running FLUX.1-dev with FB Cache (`examples/flux.1-dev-cache.py <https://github.com/nunchaku-tech/nunchaku/blob/main/examples/flux.1-dev-cache.py>`__)
+   :linenos:
+   :emphasize-lines: 15-17
+
+Enable it with :func:`~nunchaku.caching.diffusers_adapters.flux.apply_cache_on_pipe`:
+
+.. code-block:: python
+
+    apply_cache_on_pipe(pipeline, residual_diff_threshold=0.12)
+
+Adjust ``residual_diff_threshold`` to trade speed for quality - higher values are faster but lower quality.
+Recommended value 0.12 gives 2× speedup for 50-step and 1.4× for 30-step denoising.
+
+.. _usage-cache-dit:
+
+Cache-DiT
+---------
+
+Nunchaku supports `Cache-DiT <https://github.com/vipshop/cache-dit>`__ to enable further accelerated inference through various caching strategies, such as DB Cache.
+
+Install the required package:
+
+.. code-block:: bash
+
+    pip install cache-dit
+
+.. tabs::
+
+   .. tab:: FLUX.1
+
+      .. literalinclude:: ../../../examples/v1/flux.1-dev-cache-dit.py
+         :language: python
+         :caption: FLUX.1-dev with Cache-DiT (`examples/v1/flux.1-dev-cache-dit.py <https://github.com/nunchaku-tech/nunchaku/blob/main/examples/v1/flux.1-dev-cache-dit.py>`__)
+         :linenos:
+
+   .. tab:: Qwen-Image
+
+      .. literalinclude:: ../../../examples/v1/qwen-image-cache-dit.py
+         :language: python
+         :caption: Qwen-Image with Cache-DiT (`examples/v1/qwen-image-cache-dit.py <https://github.com/nunchaku-tech/nunchaku/blob/main/examples/v1/qwen-image-cache-dit.py>`__)
+         :linenos:
+
+For more information, refer to the `Cache-DiT documentation <https://github.com/vipshop/cache-dit/blob/main/docs/User_Guide.md#unified-cache-apis>`__.

docs/source/usage/fbcache.rst

@@ -1,6 +1,9 @@
 IP Adapter
 ==========
 
+.. warning::
+  IP Adapter support will be deprecated in March, 2026.
+
 .. image:: https://huggingface.co/datasets/nunchaku-tech/cdn/resolve/main/ComfyUI-nunchaku/workflows/nunchaku-flux.1-ip-adapter.png
 
 Nunchaku supports `IP Adapter <hf_ip-adapterv2_>`_, an adapter achieving image prompt capability for the FLUX.1-dev
@@ -28,7 +31,7 @@ The IP Adapter integration in Nunchaku follows these main steps:
 
 **Caching (Optional)**:
 
-Enable caching for faster inference and reduced memory usage with :func:`~nunchaku.caching.diffusers_adapters.apply_cache_on_pipe`. See :doc:`fbcache` for more details.
+Enable caching for faster inference and reduced memory usage with :func:`~nunchaku.caching.diffusers_adapters.apply_cache_on_pipe`. See :doc:`cache` for more details.
 
 **Image Generation**:
 

docs/source/usage/ip_adapter.rst

@@ -1,3 +1,4 @@
+# Please install cache-dit with `pip install cache-dit`.
 import cache_dit
 import torch
 from cache_dit import DBCacheConfig

examples/v1/flux.1-dev-cache-dit.py

@@ -1,3 +1,4 @@
+# Please install cache-dit with `pip install cache-dit`.
 import cache_dit
 import torch
 from cache_dit import DBCacheConfig

examples/v1/qwen-image-cache-dit.py

@@ -1 +1 @@
-__version__ = "1.1.0"
+__version__ = "1.2.0dev"

nunchaku/__version__.py

@@ -34,6 +34,7 @@ dependencies = [
 ]
 optional-dependencies.ci = [
   "accelerate==1.9",
+  "cache-dit==1.1.9",
   "controlnet-aux==0.0.10",
   "datasets==3.6",
   "diffusers==0.36",
@@ -48,7 +49,7 @@ optional-dependencies.ci = [
   "pytest-rerunfailures==16.0.1",
   "timm==1.0.19",
   "torchmetrics==1.8",
-  "transformers==4.54",
+  "transformers==4.55.2",
 ]
 
 optional-dependencies.demo = [
@@ -58,6 +59,7 @@ optional-dependencies.demo = [
   "spaces",
 ]
 optional-dependencies.dev = [
+  "cache-dit>=1.1.9",
   "controlnet-aux",
   "datasets<4",
   "facexlib",