[Feature] Fix like xtuner

.gitignore

@@ -125,7 +125,6 @@ venv.bak/
 *.pkl.json
 *.log.json
 /work_dirs
-/mmcls/.mim
 .DS_Store
 
 # Pytorch
@@ -153,7 +152,6 @@ artifacts/
 wandb/
 data/
 dump/
-diffengine/.mim/
 build/
 .dist_test/
 .netrc

.pre-commit-config.yaml

@@ -45,6 +45,6 @@ repos:
         exclude: |-
           (?x)(
               ^docs
-              | ^configs
+              | ^diffengine/configs
               | ^projects
           )

MANIFEST.in

@@ -1,12 +1,7 @@
-include diffengine/.mim/model-index.yml
-include add_mim_extension.py
-recursive-include diffengine/.mim/configs *.py *.yml
-recursive-include diffengine/.mim/tools *.sh *.py
-recursive-include diffengine/.mim/demo *.sh *.py
+recursive-include diffengine/configs *.py *.yml *.json
+recursive-include diffengine/tools *.sh *.py
 recursive-exclude tests *
-recursive-exclude demo *
 recursive-exclude data *
-recursive-exclude configs *
 recursive-exclude docs *
 recursive-exclude work_dirs *
 recursive-exclude dist *

README.md

@@ -42,20 +42,19 @@ Before installing DiffEngine, please ensure that PyTorch >= v2.0 has been succes
 Install DiffEngine
 
 ```
-pip install openmim
 pip install git+https://github.com/okotaku/diffengine.git
 ```
 
 ## 👨‍🏫 Get Started [🔝](#-table-of-contents)
 
 DiffEngine makes training easy through its pre-defined configs. These configs provide a streamlined way to start your training process. Here's how you can get started using one of the pre-defined configs:
 
-1. **Choose a config**: You can find various pre-defined configs in the [`configs`](configs/) directory of the DiffEngine repository. For example, if you wish to train a DreamBooth model using the Stable Diffusion algorithm, you can use the [`configs/stable_diffusion_dreambooth/stable_diffusion_v15_dreambooth_lora_dog.py`](configs/stable_diffusion_dreambooth/stable_diffusion_v15_dreambooth_lora_dog.py).
+1. **Choose a config**: You can find various pre-defined configs in the [`configs`](diffengine/configs/) directory of the DiffEngine repository. For example, if you wish to train a DreamBooth model using the Stable Diffusion algorithm, you can use the [`configs/stable_diffusion_dreambooth/stable_diffusion_v15_dreambooth_lora_dog.py`](diffengine/configs/stable_diffusion_dreambooth/stable_diffusion_v15_dreambooth_lora_dog.py).
 
 2. **Start Training**: Open a terminal and run the following command to start training with the selected config:
 
 ```bash
-mim train diffengine stable_diffusion_v15_dreambooth_lora_dog.py
+diffengine train stable_diffusion_v15_dreambooth_lora_dog
 ```
 
 3. **Monitor Progress and get results**: The training process will begin, and you can track its progress. The outputs of the training will be located in the `work_dirs/stable_diffusion_v15_dreambooth_lora_dog` directory, specifically when using the `stable_diffusion_v15_dreambooth_lora_dog` config.
@@ -319,7 +318,6 @@ Also, please check the following openmmlab and huggingface projects and the corr
 
 - [OpenMMLab](https://openmmlab.com/)
 - [HuggingFace](https://huggingface.co/)
-- [MIM](https://github.com/open-mmlab/mim): MIM Installs OpenMMLab Packages.
 
 ```
 @article{mmengine2022,

diffengine/__init__.py

@@ -1,3 +1,4 @@
+from .entry_point import cli
 from .version import __version__
 
-__all__ = ["__version__"]
+__all__ = ["__version__", "cli"]

diffengine/configs/__init__.py

@@ -0,0 +1,23 @@
+# flake8: noqa: PTH122,PTH120
+# Copied from xtuner.configs.__init__
+import os
+
+
+def get_cfgs_name_path():
+    path = os.path.dirname(__file__)
+    mapping = {}
+    for root, _, files in os.walk(path):
+        # Skip if it is a base config
+        if "_base_" in root:
+            continue
+        for file_ in files:
+            if file_.endswith(
+                (".py", ".json"),
+            ) and not file_.startswith(".") and not file_.startswith("_"):
+                mapping[os.path.splitext(file_)[0]] = os.path.join(root, file_)
+    return mapping
+
+
+cfgs_name_path = get_cfgs_name_path()
+
+__all__ = ["cfgs_name_path"]

configs/debias_estimation_loss/README.md → diffengine/configs/debias_estimation_loss/README.md

@@ -21,12 +21,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/debias_estimation_loss/stable_diffusion_xl_pokemon_blip_debias_estimation_loss.py
+$ diffengine train stable_diffusion_xl_pokemon_blip_debias_estimation_loss
 ```
 
 ## Inference with diffusers

configs/deepfloyd_if/README.md → diffengine/configs/deepfloyd_if/README.md

@@ -21,12 +21,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/deepfloyd_if/deepfloyd_if_pokemon_blip.py
+$ diffengine train deepfloyd_if_pokemon_blip
 ```
 
 ## Inference with diffusers
@@ -36,9 +36,9 @@ Once you have trained a model, specify the path to the saved model and utilize i
 Before inferencing, we should convert weights for diffusers format,
 
 ```bash
-$ mim run diffengine publish_model2diffusers ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
+$ diffengine convert ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
 # Example
-$ mim run diffengine publish_model2diffusers configs/deepfloyd_if/deepfloyd_if_l_pokemon_blip.py work_dirs/deepfloyd_if_l_pokemon_blip/epoch_50.pth work_dirs/deepfloyd_if_l_pokemon_blip --save-keys unet
+$ diffengine convert deepfloyd_if_l_pokemon_blip work_dirs/deepfloyd_if_l_pokemon_blip/epoch_50.pth work_dirs/deepfloyd_if_l_pokemon_blip --save-keys unet
 ```
 
 Then we can run inference.
@@ -63,12 +63,6 @@ image = pipe(
 image.save('demo.png')
 ```
 
-We also provide inference demo scripts:
-
-```bash
-$ mim run diffengine demo_if "yoda pokemon" work_dirs/deepfloyd_if_l_pokemon_blip
-```
-
 ## Results Example
 
 #### deepfloyd_if_l_pokemon_blip

configs/deepfloyd_if_dreambooth/README.md → diffengine/configs/deepfloyd_if_dreambooth/README.md

@@ -27,12 +27,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/deepfloyd_if_dreambooth/deepfloyd_if_xl_dreambooth_lora_dog.py
+$ diffengine train deepfloyd_if_xl_dreambooth_lora_dog
 ```
 
 ## Inference with diffusers
@@ -67,12 +67,6 @@ image = pipe(
 image.save('demo.png')
 ```
 
-We also provide inference demo scripts:
-
-```bash
-$ mim run diffengine demo_if_lora "A photo of sks dog in a bucket" work_dirs/deepfloyd_if_xl_dreambooth_lora_dog/step999
-```
-
 ## Results Example
 
 #### deepfloyd_if_xl_dreambooth_lora_dog

configs/distill_sd/README.md → diffengine/configs/distill_sd/README.md

@@ -30,12 +30,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/distill_sd/small_sd_xl_pokemon_blip.py
+$ diffengine train small_sd_xl_pokemon_blip
 ```
 
 ## Inference with diffusers
@@ -45,9 +45,9 @@ Once you have trained a model, specify the path to the saved model and utilize i
 Before inferencing, we should convert weights for diffusers format,
 
 ```bash
-$ mim run diffengine publish_model2diffusers ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
+$ diffengine convert ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
 # Example
-$ mim run diffengine publish_model2diffusers configs/distill_sd/small_sd_xl_pokemon_blip.py work_dirs/small_sd_xl_pokemon_blip/epoch_50.pth work_dirs/small_sd_xl_pokemon_blip --save-keys unet
+$ diffengine convert small_sd_xl_pokemon_blip work_dirs/small_sd_xl_pokemon_blip/epoch_50.pth work_dirs/small_sd_xl_pokemon_blip --save-keys unet
 ```
 
 Then we can run inference.

configs/distill_sd_dreambooth/README.md → diffengine/configs/distill_sd_dreambooth/README.md

@@ -28,12 +28,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/distill_sd_dreambooth/small_sd_dreambooth_lora_dog.py
+$ diffengine train small_sd_dreambooth_lora_dog
 ```
 
 ## Training Speed
@@ -84,12 +84,6 @@ image = pipe(
 image.save('demo.png')
 ```
 
-We also provide inference demo scripts:
-
-```bash
-$ mim run diffengine demo_lora "A photo of sks dog in a bucket" work_dirs/small_sd_dreambooth_lora_dog/step999 --sdmodel segmind/small-sd
-```
-
 ## Results Example
 
 #### small_sd_dreambooth_lora_dog

configs/esd/README.md → diffengine/configs/esd/README.md

@@ -27,12 +27,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/esd/stable_diffusion_xl_gogh_esd.py
+$ diffengine train stable_diffusion_xl_gogh_esd
 ```
 
 ## Inference with diffusers
@@ -42,9 +42,9 @@ Once you have trained a model, specify the path to the saved model and utilize i
 Before inferencing, we should convert weights for diffusers format,
 
 ```bash
-$ mim run diffengine publish_model2diffusers ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
+$ diffengine convert ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
 # Example
-$ mim run diffengine publish_model2diffusers configs/esd/stable_diffusion_xl_gogh_esd.py w
+$ diffengine convert stable_diffusion_xl_gogh_esd w
 ork_dirs/stable_diffusion_xl_gogh_esd/iter_500.pth work_dirs/stable_diffusion_xl_gogh_esd --save-keys unet
 ```
 

configs/input_perturbation/README.md → diffengine/configs/input_perturbation/README.md

@@ -27,12 +27,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/input_perturbation/stable_diffusion_xl_pokemon_blip_input_perturbation.py
+$ diffengine train stable_diffusion_xl_pokemon_blip_input_perturbation
 ```
 
 ## Inference with diffusers

configs/instruct_pix2pix/README.md → diffengine/configs/instruct_pix2pix/README.md

@@ -27,12 +27,12 @@ Run Training
 
 ```
 # single gpu
-$ mim train diffengine ${CONFIG_FILE}
+$ diffengine train ${CONFIG_FILE}
 # multi gpus
-$ mim train diffengine ${CONFIG_FILE} --gpus 2 --launcher pytorch
+$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}
 
 # Example.
-$ mim train diffengine configs/instruct_pix2pix/stable_diffusion_xl_instruct_pix2pix.py
+$ diffengine train stable_diffusion_xl_instruct_pix2pix
 ```
 
 ## Inference with diffusers
@@ -42,9 +42,9 @@ Once you have trained a model, specify the path to the saved model and utilize i
 Before inferencing, we should convert weights for diffusers format,
 
 ```bash
-$ mim run diffengine publish_model2diffusers ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
+$ diffengine convert ${CONFIG_FILE} ${INPUT_FILENAME} ${OUTPUT_DIR} --save-keys ${SAVE_KEYS}
 # Example
-$ mim run diffengine publish_model2diffusers configs/instruct_pix2pix/stable_diffusion_xl_instruct_pix2pix.py work_dirs/stable_diffusion_xl_instruct_pix2pix/epoch_3.pth work_dirs/stable_diffusion_xl_instruct_pix2pix --save-keys unet
+$ diffengine convert stable_diffusion_xl_instruct_pix2pix work_dirs/stable_diffusion_xl_instruct_pix2pix/epoch_3.pth work_dirs/stable_diffusion_xl_instruct_pix2pix --save-keys unet
 ```
 
 Then we can run inference.