Research folder (#1553)

* Research folder

* Update examples/research_projects/README.md

* up

examples/dreambooth/README.md

@@ -312,30 +312,6 @@ python train_dreambooth_flax.py \
   --max_train_steps=800
 ```
 
-## Dreambooth for the inpainting model
-
-
-```bash
-export MODEL_NAME="runwayml/stable-diffusion-inpainting"
-export INSTANCE_DIR="path-to-instance-images"
-export OUTPUT_DIR="path-to-save-model"
-
-accelerate launch train_dreambooth_inpaint.py \
-  --pretrained_model_name_or_path=$MODEL_NAME  \
-  --instance_data_dir=$INSTANCE_DIR \
-  --output_dir=$OUTPUT_DIR \
-  --instance_prompt="a photo of sks dog" \
-  --resolution=512 \
-  --train_batch_size=1 \
-  --gradient_accumulation_steps=1 \
-  --learning_rate=5e-6 \
-  --lr_scheduler="constant" \
-  --lr_warmup_steps=0 \
-  --max_train_steps=400
-```
-
-The script is also compatible with prior preservation loss and gradient checkpointing
-
 ### Training with prior-preservation loss
 
 Prior-preservation is used to avoid overfitting and language-drift. Refer to the paper to learn more about it. For prior-preservation we first generate images using the model with a class prompt and then use those during training along with our data.
@@ -428,4 +404,4 @@ accelerate launch train_dreambooth_inpaint.py \
   --lr_warmup_steps=0 \
   --num_class_images=200 \
   --max_train_steps=800
-```
+```

examples/research_projects/README.md

@@ -0,0 +1,14 @@
+# Research projects
+
+This folder contains various research projects using 🧨 Diffusers. 
+They are not really maintained by the core maintainers of this library and often require a specific version of Diffusers that is indicated in the requirements file of each folder. 
+Updating them to the most recent version of the library will require some work.
+
+To use any of them, just run the command
+
+```
+pip install -r requirements.txt
+```
+inside the folder of your choice.
+
+If you need help with any of those, please open an issue where you directly ping the author(s), as indicated at the top of the README of each folder.

examples/research_projects/dreambooth_inpaint/README.md

@@ -0,0 +1,26 @@
+# Dreambooth for the inpainting model
+
+This script was added by @thedarkzeno .
+
+Please note that this script is not actively maintained, you can open an issue and tag @thedarkzeno or @patil-suraj though.
+
+```bash
+export MODEL_NAME="runwayml/stable-diffusion-inpainting"
+export INSTANCE_DIR="path-to-instance-images"
+export OUTPUT_DIR="path-to-save-model"
+
+accelerate launch train_dreambooth_inpaint.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --instance_prompt="a photo of sks dog" \
+  --resolution=512 \
+  --train_batch_size=1 \
+  --gradient_accumulation_steps=1 \
+  --learning_rate=5e-6 \
+  --lr_scheduler="constant" \
+  --lr_warmup_steps=0 \
+  --max_train_steps=400
+```
+
+The script is also compatible with prior preservation loss and gradient checkpointing

examples/research_projects/dreambooth_inpaint/requirements.txt

@@ -0,0 +1,7 @@
+diffusers==0.9.0
+accelerate
+torchvision
+transformers>=4.21.0
+ftfy
+tensorboard
+modelcards

src/diffusers/pipelines/versatile_diffusion/modeling_text_unet.py

@@ -314,7 +314,7 @@ class UNetFlatConditionModel(ModelMixin, ConfigMixin):
         in several steps. This is useful to save some memory in exchange for a small speed decrease.
 
         Args:
-            slice_size (`str` or `int`, *optional*, defaults to `"auto"`):
+            slice_size (`str` or `int` or `list(int)`, *optional*, defaults to `"auto"`):
                 When `"auto"`, halves the input to the attention heads, so attention will be computed in two steps. If
                 `"max"`, maxium amount of memory will be saved by running only one slice at a time. If a number is
                 provided, uses as many slices as `attention_head_dim // slice_size`. In this case, `attention_head_dim`