From 0b70c108b074c45a8bee6c01ae053b475cbd2413 Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Mon, 3 Apr 2023 21:07:46 +0800
Subject: [PATCH 01/14] update riformer mmpretrain
---
.../imagenet_bs128_poolformer_medium_224.py | 4 +-
.../imagenet_bs128_poolformer_small_224.py | 4 +-
.../imagenet_bs128_riformer_medium_384.py | 82 +++
.../imagenet_bs128_riformer_small_384.py | 82 +++
.../_base_/models/riformer/riformer_m36.py | 22 +
.../_base_/models/riformer/riformer_m48.py | 22 +
.../_base_/models/riformer/riformer_s12.py | 22 +
.../_base_/models/riformer/riformer_s24.py | 22 +
.../_base_/models/riformer/riformer_s36.py | 22 +
configs/riformer/README.md | 207 ++++++++
.../riformer-m36-deploy_32xb128_in1k.py | 3 +
.../riformer-m36-deploy_32xb128_in1k_384.py | 3 +
.../riformer-m48-deploy_32xb128_in1k.py | 3 +
.../riformer-m48-deploy_32xb128_in1k_384.py | 3 +
.../riformer-s12-deploy_32xb128_in1k.py | 3 +
.../riformer-s12-deploy_32xb128_in1k_384.py | 3 +
.../riformer-s24-deploy_32xb128_in1k.py | 3 +
.../riformer-s24-deploy_32xb128_in1k_384.py | 3 +
.../riformer-s36-deploy_32xb128_in1k.py | 3 +
.../riformer-s36-deploy_32xb128_in1k_384.py | 3 +
configs/riformer/metafile.yml | 183 +++++++
configs/riformer/riformer-m36_32xb128_in1k.py | 17 +
.../riformer/riformer-m36_32xb128_in1k_384.py | 17 +
configs/riformer/riformer-m48_32xb128_in1k.py | 17 +
.../riformer/riformer-m48_32xb128_in1k_384.py | 17 +
configs/riformer/riformer-s12_32xb128_in1k.py | 17 +
.../riformer/riformer-s12_32xb128_in1k_384.py | 17 +
configs/riformer/riformer-s24_32xb128_in1k.py | 17 +
.../riformer/riformer-s24_32xb128_in1k_384.py | 17 +
configs/riformer/riformer-s36_32xb128_in1k.py | 17 +
.../riformer/riformer-s36_32xb128_in1k_384.py | 17 +
mmpretrain/models/backbones/__init__.py | 2 +
mmpretrain/models/backbones/riformer.py | 498 ++++++++++++++++++
.../test_backbones/test_riformer.py | 169 ++++++
.../model_converters/reparameterize_model.py | 8 +-
35 files changed, 1541 insertions(+), 8 deletions(-)
create mode 100644 configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
create mode 100644 configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
create mode 100644 configs/_base_/models/riformer/riformer_m36.py
create mode 100644 configs/_base_/models/riformer/riformer_m48.py
create mode 100644 configs/_base_/models/riformer/riformer_s12.py
create mode 100644 configs/_base_/models/riformer/riformer_s24.py
create mode 100644 configs/_base_/models/riformer/riformer_s36.py
create mode 100644 configs/riformer/README.md
create mode 100644 configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/metafile.yml
create mode 100644 configs/riformer/riformer-m36_32xb128_in1k.py
create mode 100644 configs/riformer/riformer-m36_32xb128_in1k_384.py
create mode 100644 configs/riformer/riformer-m48_32xb128_in1k.py
create mode 100644 configs/riformer/riformer-m48_32xb128_in1k_384.py
create mode 100644 configs/riformer/riformer-s12_32xb128_in1k.py
create mode 100644 configs/riformer/riformer-s12_32xb128_in1k_384.py
create mode 100644 configs/riformer/riformer-s24_32xb128_in1k.py
create mode 100644 configs/riformer/riformer-s24_32xb128_in1k_384.py
create mode 100644 configs/riformer/riformer-s36_32xb128_in1k.py
create mode 100644 configs/riformer/riformer-s36_32xb128_in1k_384.py
create mode 100644 mmpretrain/models/backbones/riformer.py
create mode 100644 tests/test_models/test_backbones/test_riformer.py
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
index 05afae4fb..63d53d736 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- ann_file='meta/val.txt',
+ # ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
@@ -79,4 +79,4 @@ val_evaluator = dict(type='Accuracy', topk=(1, 5))
# If you want standard test, please manually configure the test dataset
test_dataloader = val_dataloader
-test_evaluator = val_evaluator
+test_evaluator = val_evaluator
\ No newline at end of file
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
index 8a4b63d75..7047a919a 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- ann_file='meta/val.txt',
+ # ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
@@ -79,4 +79,4 @@ val_evaluator = dict(type='Accuracy', topk=(1, 5))
# If you want standard test, please manually configure the test dataset
test_dataloader = val_dataloader
-test_evaluator = val_evaluator
+test_evaluator = val_evaluator
\ No newline at end of file
diff --git a/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py b/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
new file mode 100644
index 000000000..430a946f1
--- /dev/null
+++ b/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
@@ -0,0 +1,82 @@
+# dataset settings
+dataset_type = 'ImageNet'
+data_preprocessor = dict(
+ num_classes=1000,
+ # RGB format normalization parameters
+ mean=[123.675, 116.28, 103.53],
+ std=[58.395, 57.12, 57.375],
+ # convert image from BGR to RGB
+ to_rgb=True,
+)
+
+bgr_mean = data_preprocessor['mean'][::-1]
+bgr_std = data_preprocessor['std'][::-1]
+
+train_pipeline = [
+ dict(type='LoadImageFromFile'),
+ dict(
+ type='RandomResizedCrop',
+ scale=384,
+ backend='pillow',
+ interpolation='bicubic'),
+ dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+ dict(
+ type='RandAugment',
+ policies='timm_increasing',
+ num_policies=2,
+ total_level=10,
+ magnitude_level=9,
+ magnitude_std=0.5,
+ hparams=dict(
+ pad_val=[round(x) for x in bgr_mean], interpolation='bicubic')),
+ dict(
+ type='RandomErasing',
+ erase_prob=0.25,
+ mode='rand',
+ min_area_ratio=0.02,
+ max_area_ratio=1 / 3,
+ fill_color=bgr_mean,
+ fill_std=bgr_std),
+ dict(type='PackInputs'),
+]
+
+test_pipeline = [
+ dict(type='LoadImageFromFile'),
+ dict(
+ type='ResizeEdge',
+ scale=404,
+ edge='short',
+ backend='pillow',
+ interpolation='bicubic'),
+ dict(type='CenterCrop', crop_size=384),
+ dict(type='PackInputs'),
+]
+
+train_dataloader = dict(
+ batch_size=128,
+ num_workers=5,
+ dataset=dict(
+ type=dataset_type,
+ data_root='data/imagenet',
+ ann_file='meta/train.txt',
+ data_prefix='train',
+ pipeline=train_pipeline),
+ sampler=dict(type='DefaultSampler', shuffle=True),
+)
+
+val_dataloader = dict(
+ batch_size=16,
+ num_workers=5,
+ dataset=dict(
+ type=dataset_type,
+ data_root='data/imagenet',
+ # ann_file='meta/val.txt',
+ data_prefix='val',
+ pipeline=test_pipeline),
+ sampler=dict(type='DefaultSampler', shuffle=False),
+)
+val_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# If you want standard test, please manually configure the test dataset
+test_dataloader = val_dataloader
+test_evaluator = val_evaluator
diff --git a/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py b/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
new file mode 100644
index 000000000..712e3d5d5
--- /dev/null
+++ b/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
@@ -0,0 +1,82 @@
+# dataset settings
+dataset_type = 'ImageNet'
+data_preprocessor = dict(
+ num_classes=1000,
+ # RGB format normalization parameters
+ mean=[123.675, 116.28, 103.53],
+ std=[58.395, 57.12, 57.375],
+ # convert image from BGR to RGB
+ to_rgb=True,
+)
+
+bgr_mean = data_preprocessor['mean'][::-1]
+bgr_std = data_preprocessor['std'][::-1]
+
+train_pipeline = [
+ dict(type='LoadImageFromFile'),
+ dict(
+ type='RandomResizedCrop',
+ scale=384,
+ backend='pillow',
+ interpolation='bicubic'),
+ dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+ dict(
+ type='RandAugment',
+ policies='timm_increasing',
+ num_policies=2,
+ total_level=10,
+ magnitude_level=9,
+ magnitude_std=0.5,
+ hparams=dict(
+ pad_val=[round(x) for x in bgr_mean], interpolation='bicubic')),
+ dict(
+ type='RandomErasing',
+ erase_prob=0.25,
+ mode='rand',
+ min_area_ratio=0.02,
+ max_area_ratio=1 / 3,
+ fill_color=bgr_mean,
+ fill_std=bgr_std),
+ dict(type='PackInputs'),
+]
+
+test_pipeline = [
+ dict(type='LoadImageFromFile'),
+ dict(
+ type='ResizeEdge',
+ scale=426,
+ edge='short',
+ backend='pillow',
+ interpolation='bicubic'),
+ dict(type='CenterCrop', crop_size=384),
+ dict(type='PackInputs'),
+]
+
+train_dataloader = dict(
+ batch_size=128,
+ num_workers=5,
+ dataset=dict(
+ type=dataset_type,
+ data_root='data/imagenet',
+ ann_file='meta/train.txt',
+ data_prefix='train',
+ pipeline=train_pipeline),
+ sampler=dict(type='DefaultSampler', shuffle=True),
+)
+
+val_dataloader = dict(
+ batch_size=32,
+ num_workers=5,
+ dataset=dict(
+ type=dataset_type,
+ data_root='data/imagenet',
+ # ann_file='meta/val.txt',
+ data_prefix='val',
+ pipeline=test_pipeline),
+ sampler=dict(type='DefaultSampler', shuffle=False),
+)
+val_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# If you want standard test, please manually configure the test dataset
+test_dataloader = val_dataloader
+test_evaluator = val_evaluator
diff --git a/configs/_base_/models/riformer/riformer_m36.py b/configs/_base_/models/riformer/riformer_m36.py
new file mode 100644
index 000000000..99dc74848
--- /dev/null
+++ b/configs/_base_/models/riformer/riformer_m36.py
@@ -0,0 +1,22 @@
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
diff --git a/configs/_base_/models/riformer/riformer_m48.py b/configs/_base_/models/riformer/riformer_m48.py
new file mode 100644
index 000000000..934f15619
--- /dev/null
+++ b/configs/_base_/models/riformer/riformer_m48.py
@@ -0,0 +1,22 @@
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m48',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
diff --git a/configs/_base_/models/riformer/riformer_s12.py b/configs/_base_/models/riformer/riformer_s12.py
new file mode 100644
index 000000000..d3d1587ef
--- /dev/null
+++ b/configs/_base_/models/riformer/riformer_s12.py
@@ -0,0 +1,22 @@
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s12',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
diff --git a/configs/_base_/models/riformer/riformer_s24.py b/configs/_base_/models/riformer/riformer_s24.py
new file mode 100644
index 000000000..9fc8089be
--- /dev/null
+++ b/configs/_base_/models/riformer/riformer_s24.py
@@ -0,0 +1,22 @@
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s24',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
diff --git a/configs/_base_/models/riformer/riformer_s36.py b/configs/_base_/models/riformer/riformer_s36.py
new file mode 100644
index 000000000..be0ad0e0f
--- /dev/null
+++ b/configs/_base_/models/riformer/riformer_s36.py
@@ -0,0 +1,22 @@
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
new file mode 100644
index 000000000..f124c13b6
--- /dev/null
+++ b/configs/riformer/README.md
@@ -0,0 +1,207 @@
+# RIFormer
+
+> [RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer](https://arxiv.org/abs/xxxx.xxxxx)
+
+
+
+## Introduction
+
+RIFormer is a way to keep a vision backbone effective while removing token mixers in its basic building blocks. Equipped with our proposed optimization strategy, we are able to build an extremely simple vision backbone with encouraging performance, while enjoying the high efficiency during inference. RIFormer shares nearly the same macro and micro design as MetaFormer, but safely removing all token mixers. The quantitative results show that our networks outperform many prevailing backbones with faster inference speed on ImageNet-1K.
+
+
+
+
+
+Show the paper's abstract
+
+
+This paper studies how to keep a vision backbone effective while removing token mixers in its basic building blocks. Token mixers, as self-attention for vision transformers (ViTs), are intended to perform information communication between different spatial tokens but suffer from considerable computational cost and latency. However, directly removing them will lead to an incomplete model structure prior, and thus brings a significant accuracy drop. To this end, we first develop an RepIdentityFormer base on the re-parameterizing idea, to study the token mixer free model architecture. And we then explore the improved learning paradigm to break the limitation of simple token mixer free backbone, and summarize the empirical practice into 5 guidelines. Equipped with the proposed optimization strategy, we are able to build an extremely simple vision backbone with encouraging performance, while enjoying the high efficiency during inference. Extensive experiments and ablative analysis also demonstrate that the inductive bias of network architecture, can be incorporated into simple network structure with appropriate optimization strategy. We hope this work can serve as a starting point for the exploration of optimization-driven efficient network design.
+
+
+
+
+## How to use
+
+The checkpoints provided are all `training-time` models. Use the reparameterize tool or `switch_to_deploy` interface to switch them to more efficient `inference-time` architecture, which not only has fewer parameters but also less calculations.
+
+
+
+**Predict image**
+
+Use `classifier.backbone.switch_to_deploy()` interface to switch the RIFormer models into inference mode.
+
+```python
+>>> import torch
+>>> from mmcls.apis import init_model, inference_model
+>>>
+>>> model = init_model('configs/riformer/riformer-s12_32xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
+>>> results = inference_model(model, 'demo/demo.JPEG')
+>>> print( (results['pred_class'], results['pred_score']) )
+('sea snake' 0.7827475666999817)
+>>>
+>>> # switch to deploy mode
+>>> model.backbone.switch_to_deploy()
+>>> results = inference_model(model, 'demo/demo.JPEG')
+>>> print( (results['pred_class'], results['pred_score']) )
+('sea snake', 0.7827480435371399)
+```
+
+**Use the model**
+
+```python
+>>> import torch
+>>>
+>>> model = init_model('configs/riformer/riformer-s12_32xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
+>>> model.eval()
+>>> inputs = torch.rand(1, 3, 224, 224).to(model.data_preprocessor.device)
+>>> # To get classification scores.
+>>> out = model(inputs)
+>>> print(out.shape)
+torch.Size([1, 1000])
+>>> # To extract features.
+>>> outs = model.extract_feat(inputs)
+>>> print(outs[0].shape)
+torch.Size([1, 512])
+>>>
+>>> # switch to deploy mode
+>>> model.backbone.switch_to_deploy()
+>>> out_deploy = model(inputs)
+>>> print(out.shape)
+torch.Size([1, 1000])
+>>> assert torch.allclose(out, out_deploy, rtol=1e-4, atol=1e-5) # pass without error
+```
+
+**Test Command**
+
+Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset).
+
+*224×224*
+
+Download Checkpoint:
+
+```shell
+wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
+```
+
+Test use unfused model:
+
+```shell
+python tools/test.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
+```
+
+Reparameterize checkpoint:
+
+```shell
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+```
+
+Test use fused model:
+
+```shell
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+```
+
+*384×384*
+
+Download Checkpoint:
+
+```shell
+wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar
+```
+
+Test use unfused model:
+
+```shell
+python tools/test.py configs/riformer/riformer-s12_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar
+```
+
+Reparameterize checkpoint:
+
+```shell
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
+```
+
+Test use fused model:
+
+```shell
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
+```
+
+
+
+For more configurable parameters, please refer to the [API](https://mmclassification.readthedocs.io/en/1.x/api/generated/mmcls.models.backbones.RIFormer.html#mmcls.models.backbones.RIFormer).
+
+
+
+How to use the reparameterization tool(click to show)
+
+
+
+Use provided tool to reparameterize the given model and save the checkpoint:
+
+```bash
+python tools/convert_models/reparameterize_model.py ${CFG_PATH} ${SRC_CKPT_PATH} ${TARGET_CKPT_PATH}
+```
+
+`${CFG_PATH}` is the config file path, `${SRC_CKPT_PATH}` is the source chenpoint file path, `${TARGET_CKPT_PATH}` is the target deploy weight file path.
+
+For example:
+
+```shell
+# download the weight
+wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
+
+# reparameterize unfused weight to fused weight
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+```
+
+To use reparameterized weights, you can use the deploy model config file such as the [s12_deploy example](./riformer-s12-deploy_32xb128_in1k.py):
+
+```text
+# in riformer-s12-deploy_32xb128_in1k.py
+_base_ = '../riformer-s12-deploy_32xb128_in1k.py' # basic s12 config
+
+model = dict(backbone=dict(deploy=True)) # switch model into deploy mode
+```
+
+```shell
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+```
+
+
+
+
+
+## Results and models
+
+### ImageNet-1k
+
+| Model | resolution | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download |
+| :----------: | :--------: | :-------: | :------: | :-------: | :-------: | :------------------------------------------: | :-------------------------------------------------------------------------------------------------: |
+| RIFormer-S12 | 224x224 | 11.92 | 1.82 | 76.90 | 93.06 | [config](./riformer-s12_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s12_3rdparty_32xb128_in1k_20220414-f8d83051.pth) |
+| RIFormer-S24 | 224x224 | 21.39 | 3.41 | 80.28 | 94.80 | [config](./riformer-s24_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s24_3rdparty_32xb128_in1k_20220414-d7055904.pth) |
+| RIFormer-S36 | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s36_3rdparty_32xb128_in1k_20220414-d78ff3e8.pth) |
+| RIFormer-M36 | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m36_3rdparty_32xb128_in1k_20220414-c55e0949.pth) |
+| RIFormer-M48 | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m48_3rdparty_32xb128_in1k_20220414-9378f3eb.pth) |
+| RIFormer-S12 | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s12_3rdparty_32xb128_in1k_20220414-f8d83051.pth) |
+| RIFormer-S24 | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s24_3rdparty_32xb128_in1k_20220414-d7055904.pth) |
+| RIFormer-S36 | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s36_3rdparty_32xb128_in1k_20220414-d78ff3e8.pth) |
+| RIFormer-M36 | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m36_3rdparty_32xb128_in1k_20220414-c55e0949.pth) |
+| RIFormer-M48 | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m48_3rdparty_32xb128_in1k_20220414-9378f3eb.pth) |
+
+The config files of these models are only for inference.
+
+## Citation
+
+```bibtex
+@inproceedings{wang2023riformer,
+ title={MetaFormer is Actually What You Need for Vision},
+ author={Wang, Jiahao and Zhang, Songyang and Liu, Yong and Wu, Taiqiang and Yang, Yujiu and Liu, Xihui and Chen, Kai and Luo, Ping and Lin, Dahua},
+ booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
+ year={2023}
+}
+```
diff --git a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
new file mode 100644
index 000000000..ae7c97105
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m36_32xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
new file mode 100644
index 000000000..4eac9954d
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m36_32xb128_in1k_384.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
new file mode 100644
index 000000000..7d999ceff
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m48_32xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
new file mode 100644
index 000000000..bf5d76dd7
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m48_32xb128_in1k_384.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
new file mode 100644
index 000000000..784453f0e
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s12_32xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
new file mode 100644
index 000000000..a4a10723f
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s12_32xb128_in1k_384.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
new file mode 100644
index 000000000..f439ea084
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s24_32xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
new file mode 100644
index 000000000..32fa0363e
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s24_32xb128_in1k_384.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
new file mode 100644
index 000000000..509898a72
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s36_32xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
new file mode 100644
index 000000000..d684983ed
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s36_32xb128_in1k_384.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
new file mode 100644
index 000000000..9f629fa64
--- /dev/null
+++ b/configs/riformer/metafile.yml
@@ -0,0 +1,183 @@
+Collections:
+ - Name: RIFormer
+ Metadata:
+ Training Data: ImageNet-1k
+ Architecture:
+ - Affine
+ - 1x1 Convolution
+ - LayerScale
+ Paper:
+ URL: https://arxiv.org/abs/xxxx.xxxxx
+ Title: RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer
+ README: configs/riformer/README.md
+ Code:
+ Version: v1.0.rc6
+ URL:
+
+Models:
+ - Name: riformer-s12_in1k
+ Metadata:
+ FLOPs: 1822000000
+ Parameters: 11915000
+ In Collection: RIFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 76.90
+ Top 5 Accuracy: 93.06
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s12_32xb128_in1k.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-s24_in1k
+ Metadata:
+ Training Data: ImageNet-1k
+ FLOPs: 3412000000
+ Parameters: 21389000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 80.28
+ Top 5 Accuracy: 94.80
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s24_32xb128_in1k.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-s36_in1k
+ Metadata:
+ FLOPs: 5003000000
+ Parameters: 30863000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 81.29
+ Top 5 Accuracy: 95.41
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s36_32xb128_in1k.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-m36_in1k
+ Metadata:
+ Training Data: ImageNet-1k
+ FLOPs: 8801000000
+ Parameters: 56173000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 82.57
+ Top 5 Accuracy: 95.99
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-m36_32xb128_in1k.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-m48_in1k
+ Metadata:
+ FLOPs: 11590000000
+ Parameters: 73473000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 82.75
+ Top 5 Accuracy: 96.11
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-m48_32xb128_in1k.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-s12_384_in1k
+ Metadata:
+ FLOPs: 5355000000
+ Parameters: 11915000
+ In Collection: RIFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 78.29
+ Top 5 Accuracy: 93.93
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s12_32xb128_in1k_384.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-s24_384_in1k
+ Metadata:
+ Training Data: ImageNet-1k
+ FLOPs: 10028000000
+ Parameters: 21389000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 81.36
+ Top 5 Accuracy: 95.40
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s24_32xb128_in1k_384.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-s36_384_in1k
+ Metadata:
+ FLOPs: 14702000000
+ Parameters: 30863000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 82.22
+ Top 5 Accuracy: 95.95
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-s36_32xb128_in1k_384.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-m36_384_in1k
+ Metadata:
+ Training Data: ImageNet-1k
+ FLOPs: 25865000000
+ Parameters: 56173000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 83.39
+ Top 5 Accuracy: 96.40
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-m36_32xb128_in1k_384.py
+ Converted From:
+ Weights:
+ Code:
+ - Name: riformer-m48_384_in1k
+ Metadata:
+ FLOPs: 34060000000
+ Parameters: 73473000
+ In Collection: PoolFormer
+ Results:
+ - Dataset: ImageNet-1k
+ Metrics:
+ Top 1 Accuracy: 83.70
+ Top 5 Accuracy: 96.60
+ Task: Image Classification
+ Weights:
+ Config: configs/riformer/riformer-m48_32xb128_in1k_384.py
+ Converted From:
+ Weights:
+ Code:
+
+
diff --git a/configs/riformer/riformer-m36_32xb128_in1k.py b/configs/riformer/riformer-m36_32xb128_in1k.py
new file mode 100644
index 000000000..074f3afc5
--- /dev/null
+++ b/configs/riformer/riformer-m36_32xb128_in1k.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_m36.py',
+ '../_base_/datasets/imagenet_bs128_poolformer_medium_224.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-m36_32xb128_in1k_384.py b/configs/riformer/riformer-m36_32xb128_in1k_384.py
new file mode 100644
index 000000000..b72d13751
--- /dev/null
+++ b/configs/riformer/riformer-m36_32xb128_in1k_384.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_m36.py',
+ '../_base_/datasets/imagenet_bs128_riformer_medium_384.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-m48_32xb128_in1k.py b/configs/riformer/riformer-m48_32xb128_in1k.py
new file mode 100644
index 000000000..3403a39ae
--- /dev/null
+++ b/configs/riformer/riformer-m48_32xb128_in1k.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_m48.py',
+ '../_base_/datasets/imagenet_bs128_poolformer_medium_224.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-m48_32xb128_in1k_384.py b/configs/riformer/riformer-m48_32xb128_in1k_384.py
new file mode 100644
index 000000000..fafef9253
--- /dev/null
+++ b/configs/riformer/riformer-m48_32xb128_in1k_384.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_m48.py',
+ '../_base_/datasets/imagenet_bs128_riformer_medium_384.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s12_32xb128_in1k.py b/configs/riformer/riformer-s12_32xb128_in1k.py
new file mode 100644
index 000000000..f8e80a9d7
--- /dev/null
+++ b/configs/riformer/riformer-s12_32xb128_in1k.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s12.py',
+ '../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s12_32xb128_in1k_384.py b/configs/riformer/riformer-s12_32xb128_in1k_384.py
new file mode 100644
index 000000000..402144b7f
--- /dev/null
+++ b/configs/riformer/riformer-s12_32xb128_in1k_384.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s12.py',
+ '../_base_/datasets/imagenet_bs128_riformer_small_384.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s24_32xb128_in1k.py b/configs/riformer/riformer-s24_32xb128_in1k.py
new file mode 100644
index 000000000..2a4cde1d8
--- /dev/null
+++ b/configs/riformer/riformer-s24_32xb128_in1k.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s24.py',
+ '../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s24_32xb128_in1k_384.py b/configs/riformer/riformer-s24_32xb128_in1k_384.py
new file mode 100644
index 000000000..06b17acbd
--- /dev/null
+++ b/configs/riformer/riformer-s24_32xb128_in1k_384.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s24.py',
+ '../_base_/datasets/imagenet_bs128_riformer_small_384.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s36_32xb128_in1k.py b/configs/riformer/riformer-s36_32xb128_in1k.py
new file mode 100644
index 000000000..4fe61e831
--- /dev/null
+++ b/configs/riformer/riformer-s36_32xb128_in1k.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s36.py',
+ '../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/configs/riformer/riformer-s36_32xb128_in1k_384.py b/configs/riformer/riformer-s36_32xb128_in1k_384.py
new file mode 100644
index 000000000..d16a8a3d7
--- /dev/null
+++ b/configs/riformer/riformer-s36_32xb128_in1k_384.py
@@ -0,0 +1,17 @@
+_base_ = [
+ '../_base_/models/riformer/riformer_s36.py',
+ '../_base_/datasets/imagenet_bs128_riformer_small_384.py',
+ '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
+ '../_base_/default_runtime.py',
+]
+
+# schedule settings
+optim_wrapper = dict(
+ optimizer=dict(lr=4e-3),
+ clip_grad=dict(max_norm=5.0),
+)
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR
+# based on the actual training batch size.
+# base_batch_size = (32 GPUs) x (128 samples per GPU)
+auto_scale_lr = dict(base_batch_size=4096)
diff --git a/mmpretrain/models/backbones/__init__.py b/mmpretrain/models/backbones/__init__.py
index 00c2118e4..1ba680d6b 100644
--- a/mmpretrain/models/backbones/__init__.py
+++ b/mmpretrain/models/backbones/__init__.py
@@ -26,6 +26,7 @@ from .mobileone import MobileOne
from .mobilevit import MobileViT
from .mvit import MViT
from .poolformer import PoolFormer
+from .riformer import RIFormer
from .regnet import RegNet
from .replknet import RepLKNet
from .repmlp import RepMLPNet
@@ -95,6 +96,7 @@ __all__ = [
'RepLKNet',
'RepMLPNet',
'PoolFormer',
+ 'RIFormer',
'DenseNet',
'VAN',
'InceptionV3',
diff --git a/mmpretrain/models/backbones/riformer.py b/mmpretrain/models/backbones/riformer.py
new file mode 100644
index 000000000..c0b28483c
--- /dev/null
+++ b/mmpretrain/models/backbones/riformer.py
@@ -0,0 +1,498 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from typing import Sequence
+
+import torch
+import torch.nn as nn
+from mmcv.cnn.bricks import DropPath, build_activation_layer, build_norm_layer
+from mmengine.model import BaseModule
+
+from mmpretrain.registry import MODELS
+from .base_backbone import BaseBackbone
+
+
+class PatchEmbed(nn.Module):
+ """Patch Embedding module implemented by a layer of convolution.
+
+ Input: tensor in shape [B, C, H, W]
+ Output: tensor in shape [B, C, H/stride, W/stride]
+ Args:
+ patch_size (int): Patch size of the patch embedding. Defaults to 16.
+ stride (int): Stride of the patch embedding. Defaults to 16.
+ padding (int): Padding of the patch embedding. Defaults to 0.
+ in_chans (int): Input channels. Defaults to 3.
+ embed_dim (int): Output dimension of the patch embedding.
+ Defaults to 768.
+ norm_layer (module): Normalization module. Defaults to None (not use).
+ """
+
+ def __init__(self,
+ patch_size=16,
+ stride=16,
+ padding=0,
+ in_chans=3,
+ embed_dim=768,
+ norm_layer=None):
+ super().__init__()
+ self.proj = nn.Conv2d(
+ in_chans,
+ embed_dim,
+ kernel_size=patch_size,
+ stride=stride,
+ padding=padding)
+ self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
+
+ def forward(self, x):
+ x = self.proj(x)
+ x = self.norm(x)
+ return x
+
+
+class Pooling(nn.Module):
+ """Pooling module.
+
+ Args:
+ pool_size (int): Pooling size. Defaults to 3.
+ """
+
+ def __init__(self, pool_size=3):
+ super().__init__()
+ self.pool = nn.AvgPool2d(
+ pool_size,
+ stride=1,
+ padding=pool_size // 2,
+ count_include_pad=False)
+
+ def forward(self, x):
+ return self.pool(x) - x
+
+
+class Affine(nn.Module):
+ """Affine Transformation module.
+
+ Args:
+ in_features (int): Input dimension. Defaults to None.
+ """
+
+ def __init__(self, in_features=None):
+ super().__init__()
+ self.affine = nn.Conv2d(
+ in_features,
+ in_features,
+ kernel_size=1,
+ stride=1,
+ padding=0,
+ groups=in_features,
+ bias=True)
+
+ def forward(self, x):
+ return self.affine(x) - x
+
+
+class Mlp(nn.Module):
+ """Mlp implemented by with 1*1 convolutions.
+
+ Input: Tensor with shape [B, C, H, W].
+ Output: Tensor with shape [B, C, H, W].
+ Args:
+ in_features (int): Dimension of input features.
+ hidden_features (int): Dimension of hidden features.
+ out_features (int): Dimension of output features.
+ act_cfg (dict): The config dict for activation between pointwise
+ convolution. Defaults to ``dict(type='GELU')``.
+ drop (float): Dropout rate. Defaults to 0.0.
+ """
+
+ def __init__(self,
+ in_features,
+ hidden_features=None,
+ out_features=None,
+ act_cfg=dict(type='GELU'),
+ drop=0.):
+ super().__init__()
+ out_features = out_features or in_features
+ hidden_features = hidden_features or in_features
+ self.fc1 = nn.Conv2d(in_features, hidden_features, 1)
+ self.act = build_activation_layer(act_cfg)
+ self.fc2 = nn.Conv2d(hidden_features, out_features, 1)
+ self.drop = nn.Dropout(drop)
+
+ def forward(self, x):
+ x = self.fc1(x)
+ x = self.act(x)
+ x = self.drop(x)
+ x = self.fc2(x)
+ x = self.drop(x)
+ return x
+
+
+class RIFormerBlock(BaseModule):
+ """RIFormer Block.
+
+ Args:
+ dim (int): Embedding dim.
+ mlp_ratio (float): Mlp expansion ratio. Defaults to 4.
+ norm_cfg (dict): The config dict for norm layers.
+ Defaults to ``dict(type='GN', num_groups=1)``.
+ act_cfg (dict): The config dict for activation between pointwise
+ convolution. Defaults to ``dict(type='GELU')``.
+ drop (float): Dropout rate. Defaults to 0.
+ drop_path (float): Stochastic depth rate. Defaults to 0.
+ layer_scale_init_value (float): Init value for Layer Scale.
+ Defaults to 1e-5.
+ deploy (bool): Whether to switch the model structure to
+ deployment mode. Default: False.
+ """
+
+ def __init__(self,
+ dim,
+ mlp_ratio=4.,
+ norm_cfg=dict(type='GN', num_groups=1),
+ act_cfg=dict(type='GELU'),
+ drop=0.,
+ drop_path=0.,
+ layer_scale_init_value=1e-5,
+ deploy=False):
+
+ super().__init__()
+
+ if deploy:
+ self.norm_reparam = build_norm_layer(norm_cfg, dim)[1]
+ else:
+ self.norm1 = build_norm_layer(norm_cfg, dim)[1]
+ self.token_mixer = Affine(in_features=dim)
+ self.norm2 = build_norm_layer(norm_cfg, dim)[1]
+ mlp_hidden_dim = int(dim * mlp_ratio)
+ self.mlp = Mlp(
+ in_features=dim,
+ hidden_features=mlp_hidden_dim,
+ act_cfg=act_cfg,
+ drop=drop)
+
+ # The following two techniques are useful to train deep RIFormers.
+ self.drop_path = DropPath(drop_path) if drop_path > 0. \
+ else nn.Identity()
+ self.layer_scale_1 = nn.Parameter(
+ layer_scale_init_value * torch.ones((dim)), requires_grad=True)
+ self.layer_scale_2 = nn.Parameter(
+ layer_scale_init_value * torch.ones((dim)), requires_grad=True)
+ self.norm_cfg = norm_cfg
+ self.dim = dim
+ self.deploy = deploy
+
+ def forward(self, x):
+ if hasattr(self, 'norm_reparam'):
+ x = x + self.drop_path(
+ self.layer_scale_1.unsqueeze(-1).unsqueeze(-1) *
+ self.norm_reparam(x))
+ x = x + self.drop_path(
+ self.layer_scale_2.unsqueeze(-1).unsqueeze(-1) *
+ self.mlp(self.norm2(x)))
+ else:
+ x = x + self.drop_path(
+ self.layer_scale_1.unsqueeze(-1).unsqueeze(-1) *
+ self.token_mixer(self.norm1(x)))
+ x = x + self.drop_path(
+ self.layer_scale_2.unsqueeze(-1).unsqueeze(-1) *
+ self.mlp(self.norm2(x)))
+ return x
+
+ def fuse_affine(self, norm, token_mixer):
+ gamma_affn = token_mixer.affine.weight.reshape(-1)
+ gamma_affn = gamma_affn - torch.ones_like(gamma_affn)
+ beta_affn = token_mixer.affine.bias
+ gamma_ln = norm.weight
+ beta_ln = norm.bias
+ print('gamma_affn:', gamma_affn.shape)
+ print('beta_affn:', beta_affn.shape)
+ print('gamma_ln:', gamma_ln.shape)
+ print('beta_ln:', beta_ln.shape)
+ return (gamma_ln * gamma_affn), (beta_ln * gamma_affn + beta_affn)
+
+ def get_equivalent_scale_bias(self):
+ eq_s, eq_b = self.fuse_affine(self.norm1, self.token_mixer)
+ return eq_s, eq_b
+
+ def switch_to_deploy(self):
+ if self.deploy:
+ return
+ eq_s, eq_b = self.get_equivalent_scale_bias()
+ self.norm_reparam = build_norm_layer(self.norm_cfg, self.dim)[1]
+ self.norm_reparam.weight.data = eq_s
+ self.norm_reparam.bias.data = eq_b
+ self.__delattr__('norm1')
+ if hasattr(self, 'token_mixer'):
+ self.__delattr__('token_mixer')
+ self.deploy = True
+
+
+def basic_blocks(dim,
+ index,
+ layers,
+ mlp_ratio=4.,
+ norm_cfg=dict(type='GN', num_groups=1),
+ act_cfg=dict(type='GELU'),
+ drop_rate=.0,
+ drop_path_rate=0.,
+ layer_scale_init_value=1e-5,
+ deploy=False):
+ """
+ generate RIFormer blocks for a stage
+ return: RIFormer blocks
+ """
+ blocks = []
+ for block_idx in range(layers[index]):
+ block_dpr = drop_path_rate * (block_idx + sum(layers[:index])) / (
+ sum(layers) - 1)
+ blocks.append(
+ RIFormerBlock(
+ dim,
+ mlp_ratio=mlp_ratio,
+ norm_cfg=norm_cfg,
+ act_cfg=act_cfg,
+ drop=drop_rate,
+ drop_path=block_dpr,
+ layer_scale_init_value=layer_scale_init_value,
+ deploy=deploy,
+ ))
+ blocks = nn.Sequential(*blocks)
+
+ return blocks
+
+
+@MODELS.register_module()
+class RIFormer(BaseBackbone):
+ """RIFormer.
+
+ A PyTorch implementation of RIFormer introduced by:
+ `RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer `_
+
+ Modified from the `official repo
+ `.
+
+ Args:
+ arch (str | dict): The model's architecture. If string, it should be
+ one of architecture in ``RIFormer.arch_settings``. And if dict, it
+ should include the following two keys:
+
+ - layers (list[int]): Number of blocks at each stage.
+ - embed_dims (list[int]): The number of channels at each stage.
+ - mlp_ratios (list[int]): Expansion ratio of MLPs.
+ - layer_scale_init_value (float): Init value for Layer Scale.
+
+ Defaults to 'S12'.
+
+ norm_cfg (dict): The config dict for norm layers.
+ Defaults to ``dict(type='LN2d', eps=1e-6)``.
+ act_cfg (dict): The config dict for activation between pointwise
+ convolution. Defaults to ``dict(type='GELU')``.
+ in_patch_size (int): The patch size of/? input image patch embedding.
+ Defaults to 7.
+ in_stride (int): The stride of input image patch embedding.
+ Defaults to 4.
+ in_pad (int): The padding of input image patch embedding.
+ Defaults to 2.
+ down_patch_size (int): The patch size of downsampling patch embedding.
+ Defaults to 3.
+ down_stride (int): The stride of downsampling patch embedding.
+ Defaults to 2.
+ down_pad (int): The padding of downsampling patch embedding.
+ Defaults to 1.
+ drop_rate (float): Dropout rate. Defaults to 0.
+ drop_path_rate (float): Stochastic depth rate. Defaults to 0.
+ out_indices (Sequence | int): Output from which network position.
+ Index 0-6 respectively corresponds to
+ [stage1, downsampling, stage2, downsampling, stage3, downsampling, stage4]
+ Defaults to -1, means the last stage.
+ frozen_stages (int): Stages to be frozen (all param fixed).
+ Defaults to 0, which means not freezing any parameters.
+ deploy (bool): Whether to switch the model structure to
+ deployment mode. Default: False.
+ init_cfg (dict, optional): Initialization config dict
+ """ # noqa: E501
+
+ # --layers: [x,x,x,x], numbers of layers for the four stages
+ # --embed_dims, --mlp_ratios:
+ # embedding dims and mlp ratios for the four stages
+ # --downsamples: flags to apply downsampling or not in four blocks
+ arch_settings = {
+ 's12': {
+ 'layers': [2, 2, 6, 2],
+ 'embed_dims': [64, 128, 320, 512],
+ 'mlp_ratios': [4, 4, 4, 4],
+ 'layer_scale_init_value': 1e-5,
+ },
+ 's24': {
+ 'layers': [4, 4, 12, 4],
+ 'embed_dims': [64, 128, 320, 512],
+ 'mlp_ratios': [4, 4, 4, 4],
+ 'layer_scale_init_value': 1e-5,
+ },
+ 's36': {
+ 'layers': [6, 6, 18, 6],
+ 'embed_dims': [64, 128, 320, 512],
+ 'mlp_ratios': [4, 4, 4, 4],
+ 'layer_scale_init_value': 1e-6,
+ },
+ 'm36': {
+ 'layers': [6, 6, 18, 6],
+ 'embed_dims': [96, 192, 384, 768],
+ 'mlp_ratios': [4, 4, 4, 4],
+ 'layer_scale_init_value': 1e-6,
+ },
+ 'm48': {
+ 'layers': [8, 8, 24, 8],
+ 'embed_dims': [96, 192, 384, 768],
+ 'mlp_ratios': [4, 4, 4, 4],
+ 'layer_scale_init_value': 1e-6,
+ },
+ }
+
+ def __init__(self,
+ arch='s12',
+ norm_cfg=dict(type='GN', num_groups=1),
+ act_cfg=dict(type='GELU'),
+ in_patch_size=7,
+ in_stride=4,
+ in_pad=2,
+ down_patch_size=3,
+ down_stride=2,
+ down_pad=1,
+ drop_rate=0.,
+ drop_path_rate=0.,
+ out_indices=-1,
+ frozen_stages=0,
+ init_cfg=None,
+ deploy=False):
+
+ super().__init__(init_cfg=init_cfg)
+
+ if isinstance(arch, str):
+ assert arch in self.arch_settings, \
+ f'Unavailable arch, please choose from ' \
+ f'({set(self.arch_settings)}) or pass a dict.'
+ arch = self.arch_settings[arch]
+ elif isinstance(arch, dict):
+ assert 'layers' in arch and 'embed_dims' in arch, \
+ f'The arch dict must have "layers" and "embed_dims", ' \
+ f'but got {list(arch.keys())}.'
+
+ layers = arch['layers']
+ embed_dims = arch['embed_dims']
+ mlp_ratios = arch['mlp_ratios'] \
+ if 'mlp_ratios' in arch else [4, 4, 4, 4]
+ layer_scale_init_value = arch['layer_scale_init_value'] \
+ if 'layer_scale_init_value' in arch else 1e-5
+
+ self.patch_embed = PatchEmbed(
+ patch_size=in_patch_size,
+ stride=in_stride,
+ padding=in_pad,
+ in_chans=3,
+ embed_dim=embed_dims[0])
+
+ # set the main block in network
+ network = []
+ for i in range(len(layers)):
+ stage = basic_blocks(
+ embed_dims[i],
+ i,
+ layers,
+ mlp_ratio=mlp_ratios[i],
+ norm_cfg=norm_cfg,
+ act_cfg=act_cfg,
+ drop_rate=drop_rate,
+ drop_path_rate=drop_path_rate,
+ layer_scale_init_value=layer_scale_init_value,
+ deploy=deploy)
+ network.append(stage)
+ if i >= len(layers) - 1:
+ break
+ if embed_dims[i] != embed_dims[i + 1]:
+ # downsampling between two stages
+ network.append(
+ PatchEmbed(
+ patch_size=down_patch_size,
+ stride=down_stride,
+ padding=down_pad,
+ in_chans=embed_dims[i],
+ embed_dim=embed_dims[i + 1]))
+
+ self.network = nn.ModuleList(network)
+
+ if isinstance(out_indices, int):
+ out_indices = [out_indices]
+ assert isinstance(out_indices, Sequence), \
+ f'"out_indices" must by a sequence or int, ' \
+ f'get {type(out_indices)} instead.'
+ for i, index in enumerate(out_indices):
+ if index < 0:
+ out_indices[i] = 7 + index
+ assert out_indices[i] >= 0, f'Invalid out_indices {index}'
+ self.out_indices = out_indices
+ if self.out_indices:
+ for i_layer in self.out_indices:
+ layer = build_norm_layer(norm_cfg,
+ embed_dims[(i_layer + 1) // 2])[1]
+ layer_name = f'norm{i_layer}'
+ self.add_module(layer_name, layer)
+
+ self.frozen_stages = frozen_stages
+ self._freeze_stages()
+ self.deploy = deploy
+
+ def forward_embeddings(self, x):
+ x = self.patch_embed(x)
+ return x
+
+ def forward_tokens(self, x):
+ outs = []
+ for idx, block in enumerate(self.network):
+ x = block(x)
+ if idx in self.out_indices:
+ norm_layer = getattr(self, f'norm{idx}')
+ x_out = norm_layer(x)
+ outs.append(x_out)
+ return tuple(outs)
+
+ def forward(self, x):
+ # input embedding
+ x = self.forward_embeddings(x)
+ # through backbone
+ x = self.forward_tokens(x)
+ return x
+
+ def _freeze_stages(self):
+ if self.frozen_stages >= 0:
+ self.patch_embed.eval()
+ for param in self.patch_embed.parameters():
+ param.requires_grad = False
+
+ for i in range(self.frozen_stages):
+ # Include both block and downsample layer.
+ module = self.network[i]
+ module.eval()
+ for param in module.parameters():
+ param.requires_grad = False
+ if i in self.out_indices:
+ norm_layer = getattr(self, f'norm{i}')
+ norm_layer.eval()
+ for param in norm_layer.parameters():
+ param.requires_grad = False
+
+ def train(self, mode=True):
+ super(RIFormer, self).train(mode)
+ self._freeze_stages()
+
+ def switch_to_deploy(self):
+ for m in self.modules():
+ if isinstance(m, RIFormerBlock):
+ m.switch_to_deploy()
+ self.deploy = True
+
+
+if __name__ == '__main__':
+ model = RIFormer(arch='s12', deploy=False)
+ model.eval()
+ print('------------------- training-time model -------------')
+ for i in model.state_dict().keys():
+ print(i)
diff --git a/tests/test_models/test_backbones/test_riformer.py b/tests/test_models/test_backbones/test_riformer.py
new file mode 100644
index 000000000..995dcbb51
--- /dev/null
+++ b/tests/test_models/test_backbones/test_riformer.py
@@ -0,0 +1,169 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from copy import deepcopy
+from unittest import TestCase
+
+import torch
+import torch.nn as nn
+
+from mmpretrain.models.backbones import RIFormer
+from mmpretrain.models.backbones.riformer import RIFormerBlock
+
+
+class TestRIFormer(TestCase):
+
+ def setUp(self):
+ arch = 's12'
+ self.cfg = dict(arch=arch, drop_path_rate=0.1)
+ self.arch = RIFormer.arch_settings[arch]
+
+ def test_arch(self):
+ # Test invalid default arch
+ with self.assertRaisesRegex(AssertionError, 'Unavailable arch'):
+ cfg = deepcopy(self.cfg)
+ cfg['arch'] = 'unknown'
+ RIFormer(**cfg)
+
+ # Test invalid custom arch
+ with self.assertRaisesRegex(AssertionError, 'must have "layers"'):
+ cfg = deepcopy(self.cfg)
+ cfg['arch'] = {
+ 'embed_dims': 96,
+ 'num_heads': [3, 6, 12, 16],
+ }
+ RIFormer(**cfg)
+
+ # Test custom arch
+ cfg = deepcopy(self.cfg)
+ layers = [2, 2, 4, 2]
+ embed_dims = [6, 12, 6, 12]
+ mlp_ratios = [2, 3, 4, 4]
+ layer_scale_init_value = 1e-4
+ cfg['arch'] = dict(
+ layers=layers,
+ embed_dims=embed_dims,
+ mlp_ratios=mlp_ratios,
+ layer_scale_init_value=layer_scale_init_value,
+ )
+ model = RIFormer(**cfg)
+ for i, stage in enumerate(model.network):
+ if not isinstance(stage, RIFormerBlock):
+ continue
+ self.assertEqual(len(stage), layers[i])
+ self.assertEqual(stage[0].mlp.fc1.in_channels, embed_dims[i])
+ self.assertEqual(stage[0].mlp.fc1.out_channels,
+ embed_dims[i] * mlp_ratios[i])
+ self.assertTrue(
+ torch.allclose(stage[0].layer_scale_1,
+ torch.tensor(layer_scale_init_value)))
+ self.assertTrue(
+ torch.allclose(stage[0].layer_scale_2,
+ torch.tensor(layer_scale_init_value)))
+
+ def test_init_weights(self):
+ # test weight init cfg
+ cfg = deepcopy(self.cfg)
+ cfg['init_cfg'] = [
+ dict(
+ type='Kaiming',
+ layer='Conv2d',
+ mode='fan_in',
+ nonlinearity='linear')
+ ]
+ model = RIFormer(**cfg)
+ ori_weight = model.patch_embed.proj.weight.clone().detach()
+
+ model.init_weights()
+ initialized_weight = model.patch_embed.proj.weight
+ self.assertFalse(torch.allclose(ori_weight, initialized_weight))
+
+ def test_forward(self):
+ imgs = torch.randn(1, 3, 224, 224)
+
+ cfg = deepcopy(self.cfg)
+ model = RIFormer(**cfg)
+ outs = model(imgs)
+ self.assertIsInstance(outs, tuple)
+ self.assertEqual(len(outs), 1)
+ feat = outs[-1]
+ self.assertEqual(feat.shape, (1, 512, 7, 7))
+
+ # test multiple output indices
+ cfg = deepcopy(self.cfg)
+ cfg['out_indices'] = (0, 2, 4, 6)
+ model = RIFormer(**cfg)
+ outs = model(imgs)
+ self.assertIsInstance(outs, tuple)
+ self.assertEqual(len(outs), 4)
+ for dim, stride, out in zip(self.arch['embed_dims'], [1, 2, 4, 8],
+ outs):
+ self.assertEqual(out.shape, (1, dim, 56 // stride, 56 // stride))
+
+ def test_repameterization(self):
+ # Test eval of "train" mode and "deploy" mode
+ imgs = torch.randn(1, 3, 224, 224)
+ gap = nn.AdaptiveAvgPool2d(output_size=(1))
+ fc = nn.Linear(self.arch['embed_dims'][3], 10)
+
+ cfg = deepcopy(self.cfg)
+ cfg['out_indices'] = (0, 2, 4, 6)
+ model = RIFormer(**cfg)
+ model.eval()
+ feats = model(imgs)
+ self.assertIsInstance(feats, tuple)
+ feat = feats[-1]
+ pred = fc(gap(feat).flatten(1))
+ model.switch_to_deploy()
+ for m in model.modules():
+ if isinstance(m, RIFormerBlock):
+ assert m.deploy is True
+ feats_deploy = model(imgs)
+ pred_deploy = fc(gap(feats_deploy[-1]).flatten(1))
+ for i in range(4):
+ torch.allclose(feats[i], feats_deploy[i])
+ torch.allclose(pred, pred_deploy)
+
+
+ def test_structure(self):
+ # test drop_path_rate decay
+ cfg = deepcopy(self.cfg)
+ cfg['drop_path_rate'] = 0.2
+ model = RIFormer(**cfg)
+ layers = self.arch['layers']
+ for i, block in enumerate(model.network):
+ expect_prob = 0.2 / (sum(layers) - 1) * i
+ if hasattr(block, 'drop_path'):
+ if expect_prob == 0:
+ self.assertIsInstance(block.drop_path, torch.nn.Identity)
+ else:
+ self.assertAlmostEqual(block.drop_path.drop_prob,
+ expect_prob)
+
+ # test with first stage frozen.
+ cfg = deepcopy(self.cfg)
+ frozen_stages = 1
+ cfg['frozen_stages'] = frozen_stages
+ cfg['out_indices'] = (0, 2, 4, 6)
+ model = RIFormer(**cfg)
+ model.init_weights()
+ model.train()
+
+ # the patch_embed and first stage should not require grad.
+ self.assertFalse(model.patch_embed.training)
+ for param in model.patch_embed.parameters():
+ self.assertFalse(param.requires_grad)
+ for i in range(frozen_stages):
+ module = model.network[i]
+ for param in module.parameters():
+ self.assertFalse(param.requires_grad)
+ for param in model.norm0.parameters():
+ self.assertFalse(param.requires_grad)
+
+ # the second stage should require grad.
+ for i in range(frozen_stages + 1, 7):
+ module = model.network[i]
+ for param in module.parameters():
+ self.assertTrue(param.requires_grad)
+ if hasattr(model, f'norm{i}'):
+ norm = getattr(model, f'norm{i}')
+ for param in norm.parameters():
+ self.assertTrue(param.requires_grad)
diff --git a/tools/model_converters/reparameterize_model.py b/tools/model_converters/reparameterize_model.py
index f2f3e7c43..fc04d16ee 100644
--- a/tools/model_converters/reparameterize_model.py
+++ b/tools/model_converters/reparameterize_model.py
@@ -4,8 +4,8 @@ from pathlib import Path
import torch
-from mmpretrain.apis import init_model
-from mmpretrain.models.classifiers import ImageClassifier
+from mmcls.apis import init_model
+from mmcls.models.classifiers import ImageClassifier
def convert_classifier_to_deploy(model, checkpoint, save_path):
@@ -39,7 +39,7 @@ def main():
args = parser.parse_args()
save_path = Path(args.save_path)
- if save_path.suffix != '.pth':
+ if save_path.suffix != '.pth' and save_path.suffix != '.tar':
print('The path should contain the name of the pth format file.')
exit()
save_path.parent.mkdir(parents=True, exist_ok=True)
@@ -47,7 +47,7 @@ def main():
model = init_model(
args.config_path, checkpoint=args.checkpoint_path, device='cpu')
assert isinstance(model, ImageClassifier), \
- '`model` must be a `mmpretrain.classifiers.ImageClassifier` instance.'
+ '`model` must be a `mmcls.classifiers.ImageClassifier` instance.'
checkpoint = torch.load(args.checkpoint_path)
convert_classifier_to_deploy(model, checkpoint, args.save_path)
From 32c258ff19b1a0eae2986a3d2723fdf7345b2582 Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Mon, 3 Apr 2023 21:22:43 +0800
Subject: [PATCH 02/14] update riformer mmpretrain
---
.../imagenet_bs128_poolformer_medium_224.py | 2 +-
.../imagenet_bs128_poolformer_small_224.py | 2 +-
configs/riformer/metafile.yml | 66 +++++++++----------
mmpretrain/models/backbones/__init__.py | 2 +-
model-index.yml | 1 +
.../test_backbones/test_riformer.py | 1 -
.../model_converters/reparameterize_model.py | 1 -
7 files changed, 36 insertions(+), 39 deletions(-)
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
index 63d53d736..a48143cbe 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
@@ -79,4 +79,4 @@ val_evaluator = dict(type='Accuracy', topk=(1, 5))
# If you want standard test, please manually configure the test dataset
test_dataloader = val_dataloader
-test_evaluator = val_evaluator
\ No newline at end of file
+test_evaluator = val_evaluator
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
index 7047a919a..a36b5afb9 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
@@ -79,4 +79,4 @@ val_evaluator = dict(type='Accuracy', topk=(1, 5))
# If you want standard test, please manually configure the test dataset
test_dataloader = val_dataloader
-test_evaluator = val_evaluator
\ No newline at end of file
+test_evaluator = val_evaluator
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
index 9f629fa64..db7b39d6b 100644
--- a/configs/riformer/metafile.yml
+++ b/configs/riformer/metafile.yml
@@ -8,11 +8,11 @@ Collections:
- LayerScale
Paper:
URL: https://arxiv.org/abs/xxxx.xxxxx
- Title: RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer
+ Title: "RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer"
README: configs/riformer/README.md
Code:
Version: v1.0.rc6
- URL:
+ URL:
Models:
- Name: riformer-s12_in1k
@@ -26,11 +26,11 @@ Models:
Top 1 Accuracy: 76.90
Top 5 Accuracy: 93.06
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s12_32xb128_in1k.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-s24_in1k
Metadata:
Training Data: ImageNet-1k
@@ -43,11 +43,11 @@ Models:
Top 1 Accuracy: 80.28
Top 5 Accuracy: 94.80
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s24_32xb128_in1k.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-s36_in1k
Metadata:
FLOPs: 5003000000
@@ -59,11 +59,11 @@ Models:
Top 1 Accuracy: 81.29
Top 5 Accuracy: 95.41
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s36_32xb128_in1k.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-m36_in1k
Metadata:
Training Data: ImageNet-1k
@@ -76,11 +76,11 @@ Models:
Top 1 Accuracy: 82.57
Top 5 Accuracy: 95.99
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-m36_32xb128_in1k.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-m48_in1k
Metadata:
FLOPs: 11590000000
@@ -92,11 +92,11 @@ Models:
Top 1 Accuracy: 82.75
Top 5 Accuracy: 96.11
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-m48_32xb128_in1k.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-s12_384_in1k
Metadata:
FLOPs: 5355000000
@@ -108,11 +108,11 @@ Models:
Top 1 Accuracy: 78.29
Top 5 Accuracy: 93.93
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s12_32xb128_in1k_384.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-s24_384_in1k
Metadata:
Training Data: ImageNet-1k
@@ -125,11 +125,11 @@ Models:
Top 1 Accuracy: 81.36
Top 5 Accuracy: 95.40
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s24_32xb128_in1k_384.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-s36_384_in1k
Metadata:
FLOPs: 14702000000
@@ -141,11 +141,11 @@ Models:
Top 1 Accuracy: 82.22
Top 5 Accuracy: 95.95
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-s36_32xb128_in1k_384.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-m36_384_in1k
Metadata:
Training Data: ImageNet-1k
@@ -158,11 +158,11 @@ Models:
Top 1 Accuracy: 83.39
Top 5 Accuracy: 96.40
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-m36_32xb128_in1k_384.py
Converted From:
- Weights:
- Code:
+ Weights:
+ Code:
- Name: riformer-m48_384_in1k
Metadata:
FLOPs: 34060000000
@@ -174,10 +174,8 @@ Models:
Top 1 Accuracy: 83.70
Top 5 Accuracy: 96.60
Task: Image Classification
- Weights:
+ Weights:
Config: configs/riformer/riformer-m48_32xb128_in1k_384.py
Converted From:
- Weights:
- Code:
-
-
+ Weights:
+ Code:
diff --git a/mmpretrain/models/backbones/__init__.py b/mmpretrain/models/backbones/__init__.py
index 1ba680d6b..5aa69c15f 100644
--- a/mmpretrain/models/backbones/__init__.py
+++ b/mmpretrain/models/backbones/__init__.py
@@ -26,7 +26,6 @@ from .mobileone import MobileOne
from .mobilevit import MobileViT
from .mvit import MViT
from .poolformer import PoolFormer
-from .riformer import RIFormer
from .regnet import RegNet
from .replknet import RepLKNet
from .repmlp import RepMLPNet
@@ -37,6 +36,7 @@ from .resnet import ResNet, ResNetV1c, ResNetV1d
from .resnet_cifar import ResNet_CIFAR
from .resnext import ResNeXt
from .revvit import RevVisionTransformer
+from .riformer import RIFormer
from .seresnet import SEResNet
from .seresnext import SEResNeXt
from .shufflenet_v1 import ShuffleNetV1
diff --git a/model-index.yml b/model-index.yml
index defb42c1a..fe3bbbad5 100644
--- a/model-index.yml
+++ b/model-index.yml
@@ -29,6 +29,7 @@ Import:
- configs/convmixer/metafile.yml
- configs/densenet/metafile.yml
- configs/poolformer/metafile.yml
+ - configs/riformer/metafile.yml
- configs/inception_v3/metafile.yml
- configs/mvit/metafile.yml
- configs/edgenext/metafile.yml
diff --git a/tests/test_models/test_backbones/test_riformer.py b/tests/test_models/test_backbones/test_riformer.py
index 995dcbb51..86847ee05 100644
--- a/tests/test_models/test_backbones/test_riformer.py
+++ b/tests/test_models/test_backbones/test_riformer.py
@@ -122,7 +122,6 @@ class TestRIFormer(TestCase):
torch.allclose(feats[i], feats_deploy[i])
torch.allclose(pred, pred_deploy)
-
def test_structure(self):
# test drop_path_rate decay
cfg = deepcopy(self.cfg)
diff --git a/tools/model_converters/reparameterize_model.py b/tools/model_converters/reparameterize_model.py
index fc04d16ee..ad97e6e18 100644
--- a/tools/model_converters/reparameterize_model.py
+++ b/tools/model_converters/reparameterize_model.py
@@ -3,7 +3,6 @@ import argparse
from pathlib import Path
import torch
-
from mmcls.apis import init_model
from mmcls.models.classifiers import ImageClassifier
From 0ef0b5ce089c8e214b45983a60224d2d70ac6e0a Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Mon, 3 Apr 2023 21:24:43 +0800
Subject: [PATCH 03/14] update riformer mmpretrain
---
model-index.yml | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/model-index.yml b/model-index.yml
index fe3bbbad5..27511689c 100644
--- a/model-index.yml
+++ b/model-index.yml
@@ -29,7 +29,7 @@ Import:
- configs/convmixer/metafile.yml
- configs/densenet/metafile.yml
- configs/poolformer/metafile.yml
- - configs/riformer/metafile.yml
+ - configs/riformer/metafile.yml
- configs/inception_v3/metafile.yml
- configs/mvit/metafile.yml
- configs/edgenext/metafile.yml
From 61b795f21fb71752715d2f32a0f6e394b1c3232b Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Mon, 3 Apr 2023 21:26:27 +0800
Subject: [PATCH 04/14] update riformer mmpretrain
---
model-index.yml | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/model-index.yml b/model-index.yml
index 27511689c..b33e717a8 100644
--- a/model-index.yml
+++ b/model-index.yml
@@ -28,7 +28,7 @@ Import:
- configs/cspnet/metafile.yml
- configs/convmixer/metafile.yml
- configs/densenet/metafile.yml
- - configs/poolformer/metafile.yml
+ - configs/poolformer/metafile.yml
- configs/riformer/metafile.yml
- configs/inception_v3/metafile.yml
- configs/mvit/metafile.yml
From e7da3f29f49a7712085acf48ef1654b6d87ecc3e Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Tue, 4 Apr 2023 11:59:24 +0800
Subject: [PATCH 05/14] update riformer mmpretrain
---
configs/riformer/riformer-m36_32xb128_in1k.py | 24 +++++++++++++++++-
.../riformer/riformer-m36_32xb128_in1k_384.py | 24 +++++++++++++++++-
configs/riformer/riformer-m48_32xb128_in1k.py | 25 ++++++++++++++++++-
.../riformer/riformer-m48_32xb128_in1k_384.py | 25 ++++++++++++++++++-
configs/riformer/riformer-s12_32xb128_in1k.py | 25 ++++++++++++++++++-
.../riformer/riformer-s12_32xb128_in1k_384.py | 25 ++++++++++++++++++-
configs/riformer/riformer-s24_32xb128_in1k.py | 24 +++++++++++++++++-
.../riformer/riformer-s24_32xb128_in1k_384.py | 24 +++++++++++++++++-
configs/riformer/riformer-s36_32xb128_in1k.py | 24 +++++++++++++++++-
.../riformer/riformer-s36_32xb128_in1k_384.py | 24 +++++++++++++++++-
.../model_converters/reparameterize_model.py | 6 ++---
11 files changed, 237 insertions(+), 13 deletions(-)
diff --git a/configs/riformer/riformer-m36_32xb128_in1k.py b/configs/riformer/riformer-m36_32xb128_in1k.py
index 074f3afc5..30e93aa83 100644
--- a/configs/riformer/riformer-m36_32xb128_in1k.py
+++ b/configs/riformer/riformer-m36_32xb128_in1k.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_m36.py',
'../_base_/datasets/imagenet_bs128_poolformer_medium_224.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-m36_32xb128_in1k_384.py b/configs/riformer/riformer-m36_32xb128_in1k_384.py
index b72d13751..57f687cd5 100644
--- a/configs/riformer/riformer-m36_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-m36_32xb128_in1k_384.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_m36.py',
'../_base_/datasets/imagenet_bs128_riformer_medium_384.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-m48_32xb128_in1k.py b/configs/riformer/riformer-m48_32xb128_in1k.py
index 3403a39ae..930308a8f 100644
--- a/configs/riformer/riformer-m48_32xb128_in1k.py
+++ b/configs/riformer/riformer-m48_32xb128_in1k.py
@@ -1,10 +1,33 @@
_base_ = [
- '../_base_/models/riformer/riformer_m48.py',
'../_base_/datasets/imagenet_bs128_poolformer_medium_224.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m48',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-m48_32xb128_in1k_384.py b/configs/riformer/riformer-m48_32xb128_in1k_384.py
index fafef9253..e8f0941c7 100644
--- a/configs/riformer/riformer-m48_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-m48_32xb128_in1k_384.py
@@ -1,10 +1,33 @@
_base_ = [
- '../_base_/models/riformer/riformer_m48.py',
'../_base_/datasets/imagenet_bs128_riformer_medium_384.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='m48',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=768,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s12_32xb128_in1k.py b/configs/riformer/riformer-s12_32xb128_in1k.py
index f8e80a9d7..467b59216 100644
--- a/configs/riformer/riformer-s12_32xb128_in1k.py
+++ b/configs/riformer/riformer-s12_32xb128_in1k.py
@@ -1,10 +1,33 @@
_base_ = [
- '../_base_/models/riformer/riformer_s12.py',
'../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s12',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s12_32xb128_in1k_384.py b/configs/riformer/riformer-s12_32xb128_in1k_384.py
index 402144b7f..5d0d375b8 100644
--- a/configs/riformer/riformer-s12_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-s12_32xb128_in1k_384.py
@@ -1,10 +1,33 @@
_base_ = [
- '../_base_/models/riformer/riformer_s12.py',
'../_base_/datasets/imagenet_bs128_riformer_small_384.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s12',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s24_32xb128_in1k.py b/configs/riformer/riformer-s24_32xb128_in1k.py
index 2a4cde1d8..560cddcf8 100644
--- a/configs/riformer/riformer-s24_32xb128_in1k.py
+++ b/configs/riformer/riformer-s24_32xb128_in1k.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_s24.py',
'../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s24',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s24_32xb128_in1k_384.py b/configs/riformer/riformer-s24_32xb128_in1k_384.py
index 06b17acbd..6a1ec7b57 100644
--- a/configs/riformer/riformer-s24_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-s24_32xb128_in1k_384.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_s24.py',
'../_base_/datasets/imagenet_bs128_riformer_small_384.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s24',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s36_32xb128_in1k.py b/configs/riformer/riformer-s36_32xb128_in1k.py
index 4fe61e831..28511307a 100644
--- a/configs/riformer/riformer-s36_32xb128_in1k.py
+++ b/configs/riformer/riformer-s36_32xb128_in1k.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_s36.py',
'../_base_/datasets/imagenet_bs128_poolformer_small_224.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s36_32xb128_in1k_384.py b/configs/riformer/riformer-s36_32xb128_in1k_384.py
index d16a8a3d7..b30773570 100644
--- a/configs/riformer/riformer-s36_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-s36_32xb128_in1k_384.py
@@ -1,10 +1,32 @@
_base_ = [
- '../_base_/models/riformer/riformer_s36.py',
'../_base_/datasets/imagenet_bs128_riformer_small_384.py',
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
'../_base_/default_runtime.py',
]
+# Model settings
+model = dict(
+ type='ImageClassifier',
+ backbone=dict(
+ type='RIFormer',
+ arch='s36',
+ drop_path_rate=0.1,
+ init_cfg=[
+ dict(
+ type='TruncNormal',
+ layer=['Conv2d', 'Linear'],
+ std=.02,
+ bias=0.),
+ dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
+ ]),
+ neck=dict(type='GlobalAveragePooling'),
+ head=dict(
+ type='LinearClsHead',
+ num_classes=1000,
+ in_channels=512,
+ loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
+ ))
+
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/tools/model_converters/reparameterize_model.py b/tools/model_converters/reparameterize_model.py
index ad97e6e18..00eba9190 100644
--- a/tools/model_converters/reparameterize_model.py
+++ b/tools/model_converters/reparameterize_model.py
@@ -3,8 +3,8 @@ import argparse
from pathlib import Path
import torch
-from mmcls.apis import init_model
-from mmcls.models.classifiers import ImageClassifier
+from mmpretrain.apis import init_model
+from mmpretrain.models.classifiers import ImageClassifier
def convert_classifier_to_deploy(model, checkpoint, save_path):
@@ -46,7 +46,7 @@ def main():
model = init_model(
args.config_path, checkpoint=args.checkpoint_path, device='cpu')
assert isinstance(model, ImageClassifier), \
- '`model` must be a `mmcls.classifiers.ImageClassifier` instance.'
+ '`model` must be a `mmpretrain.classifiers.ImageClassifier` instance.'
checkpoint = torch.load(args.checkpoint_path)
convert_classifier_to_deploy(model, checkpoint, args.save_path)
From a6c24d104e288ec66c8b33cc5f823a355d6acdf5 Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Tue, 4 Apr 2023 13:17:16 +0800
Subject: [PATCH 06/14] update riformer mmpretrain
---
configs/riformer/riformer-m48_32xb128_in1k.py | 1 -
.../riformer/riformer-m48_32xb128_in1k_384.py | 1 -
configs/riformer/riformer-s12_32xb128_in1k.py | 1 -
.../riformer/riformer-s12_32xb128_in1k_384.py | 1 -
mmpretrain/models/backbones/riformer.py | 96 +------------------
model-index.yml | 4 +-
.../model_converters/reparameterize_model.py | 1 +
7 files changed, 5 insertions(+), 100 deletions(-)
diff --git a/configs/riformer/riformer-m48_32xb128_in1k.py b/configs/riformer/riformer-m48_32xb128_in1k.py
index 930308a8f..9dc5c3e29 100644
--- a/configs/riformer/riformer-m48_32xb128_in1k.py
+++ b/configs/riformer/riformer-m48_32xb128_in1k.py
@@ -27,7 +27,6 @@ model = dict(
loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
))
-
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-m48_32xb128_in1k_384.py b/configs/riformer/riformer-m48_32xb128_in1k_384.py
index e8f0941c7..ef6f19646 100644
--- a/configs/riformer/riformer-m48_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-m48_32xb128_in1k_384.py
@@ -27,7 +27,6 @@ model = dict(
loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
))
-
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s12_32xb128_in1k.py b/configs/riformer/riformer-s12_32xb128_in1k.py
index 467b59216..e85f8fb88 100644
--- a/configs/riformer/riformer-s12_32xb128_in1k.py
+++ b/configs/riformer/riformer-s12_32xb128_in1k.py
@@ -27,7 +27,6 @@ model = dict(
loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
))
-
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/configs/riformer/riformer-s12_32xb128_in1k_384.py b/configs/riformer/riformer-s12_32xb128_in1k_384.py
index 5d0d375b8..6d19dae07 100644
--- a/configs/riformer/riformer-s12_32xb128_in1k_384.py
+++ b/configs/riformer/riformer-s12_32xb128_in1k_384.py
@@ -27,7 +27,6 @@ model = dict(
loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
))
-
# schedule settings
optim_wrapper = dict(
optimizer=dict(lr=4e-3),
diff --git a/mmpretrain/models/backbones/riformer.py b/mmpretrain/models/backbones/riformer.py
index c0b28483c..c2e2c8916 100644
--- a/mmpretrain/models/backbones/riformer.py
+++ b/mmpretrain/models/backbones/riformer.py
@@ -3,67 +3,12 @@ from typing import Sequence
import torch
import torch.nn as nn
-from mmcv.cnn.bricks import DropPath, build_activation_layer, build_norm_layer
+from mmcv.cnn.bricks import DropPath, build_norm_layer
from mmengine.model import BaseModule
from mmpretrain.registry import MODELS
from .base_backbone import BaseBackbone
-
-
-class PatchEmbed(nn.Module):
- """Patch Embedding module implemented by a layer of convolution.
-
- Input: tensor in shape [B, C, H, W]
- Output: tensor in shape [B, C, H/stride, W/stride]
- Args:
- patch_size (int): Patch size of the patch embedding. Defaults to 16.
- stride (int): Stride of the patch embedding. Defaults to 16.
- padding (int): Padding of the patch embedding. Defaults to 0.
- in_chans (int): Input channels. Defaults to 3.
- embed_dim (int): Output dimension of the patch embedding.
- Defaults to 768.
- norm_layer (module): Normalization module. Defaults to None (not use).
- """
-
- def __init__(self,
- patch_size=16,
- stride=16,
- padding=0,
- in_chans=3,
- embed_dim=768,
- norm_layer=None):
- super().__init__()
- self.proj = nn.Conv2d(
- in_chans,
- embed_dim,
- kernel_size=patch_size,
- stride=stride,
- padding=padding)
- self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
-
- def forward(self, x):
- x = self.proj(x)
- x = self.norm(x)
- return x
-
-
-class Pooling(nn.Module):
- """Pooling module.
-
- Args:
- pool_size (int): Pooling size. Defaults to 3.
- """
-
- def __init__(self, pool_size=3):
- super().__init__()
- self.pool = nn.AvgPool2d(
- pool_size,
- stride=1,
- padding=pool_size // 2,
- count_include_pad=False)
-
- def forward(self, x):
- return self.pool(x) - x
+from .poolformer import Mlp, PatchEmbed
class Affine(nn.Module):
@@ -88,43 +33,6 @@ class Affine(nn.Module):
return self.affine(x) - x
-class Mlp(nn.Module):
- """Mlp implemented by with 1*1 convolutions.
-
- Input: Tensor with shape [B, C, H, W].
- Output: Tensor with shape [B, C, H, W].
- Args:
- in_features (int): Dimension of input features.
- hidden_features (int): Dimension of hidden features.
- out_features (int): Dimension of output features.
- act_cfg (dict): The config dict for activation between pointwise
- convolution. Defaults to ``dict(type='GELU')``.
- drop (float): Dropout rate. Defaults to 0.0.
- """
-
- def __init__(self,
- in_features,
- hidden_features=None,
- out_features=None,
- act_cfg=dict(type='GELU'),
- drop=0.):
- super().__init__()
- out_features = out_features or in_features
- hidden_features = hidden_features or in_features
- self.fc1 = nn.Conv2d(in_features, hidden_features, 1)
- self.act = build_activation_layer(act_cfg)
- self.fc2 = nn.Conv2d(hidden_features, out_features, 1)
- self.drop = nn.Dropout(drop)
-
- def forward(self, x):
- x = self.fc1(x)
- x = self.act(x)
- x = self.drop(x)
- x = self.fc2(x)
- x = self.drop(x)
- return x
-
-
class RIFormerBlock(BaseModule):
"""RIFormer Block.
diff --git a/model-index.yml b/model-index.yml
index b33e717a8..fe3bbbad5 100644
--- a/model-index.yml
+++ b/model-index.yml
@@ -28,8 +28,8 @@ Import:
- configs/cspnet/metafile.yml
- configs/convmixer/metafile.yml
- configs/densenet/metafile.yml
- - configs/poolformer/metafile.yml
- - configs/riformer/metafile.yml
+ - configs/poolformer/metafile.yml
+ - configs/riformer/metafile.yml
- configs/inception_v3/metafile.yml
- configs/mvit/metafile.yml
- configs/edgenext/metafile.yml
diff --git a/tools/model_converters/reparameterize_model.py b/tools/model_converters/reparameterize_model.py
index 00eba9190..49a3913c9 100644
--- a/tools/model_converters/reparameterize_model.py
+++ b/tools/model_converters/reparameterize_model.py
@@ -3,6 +3,7 @@ import argparse
from pathlib import Path
import torch
+
from mmpretrain.apis import init_model
from mmpretrain.models.classifiers import ImageClassifier
From c9c7d9cc0f7cffc0a8d0c04d6c64c1aaa7d8252c Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Thu, 6 Apr 2023 12:06:23 +0800
Subject: [PATCH 07/14] update riformer mmpretrain
---
configs/riformer/README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
index f124c13b6..2eb399e55 100644
--- a/configs/riformer/README.md
+++ b/configs/riformer/README.md
@@ -77,7 +77,7 @@ torch.Size([1, 1000])
**Test Command**
-Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset).
+Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset).
*224×224*
From e4d8511ddf045b5f6f0afb52d78cb76af5cdea87 Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Thu, 6 Apr 2023 12:11:52 +0800
Subject: [PATCH 08/14] update riformer mmpretrain
---
configs/riformer/README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
index 2eb399e55..f124c13b6 100644
--- a/configs/riformer/README.md
+++ b/configs/riformer/README.md
@@ -77,7 +77,7 @@ torch.Size([1, 1000])
**Test Command**
-Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset).
+Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset).
*224×224*
From 53a57c6dad4554747cbee3b659864155246f95d3 Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Thu, 6 Apr 2023 17:28:26 +0800
Subject: [PATCH 09/14] update riformer mmpretrain
---
.../imagenet_bs128_poolformer_medium_224.py | 2 +-
.../imagenet_bs128_poolformer_small_224.py | 2 +-
.../imagenet_bs128_riformer_medium_384.py | 2 +-
.../imagenet_bs128_riformer_small_384.py | 2 +-
configs/riformer/metafile.yml | 20 +++++++++----------
5 files changed, 14 insertions(+), 14 deletions(-)
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
index a48143cbe..05afae4fb 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_medium_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- # ann_file='meta/val.txt',
+ ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
index a36b5afb9..8a4b63d75 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- # ann_file='meta/val.txt',
+ ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
diff --git a/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py b/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
index 430a946f1..0263ad24f 100644
--- a/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
+++ b/configs/_base_/datasets/imagenet_bs128_riformer_medium_384.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- # ann_file='meta/val.txt',
+ ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
diff --git a/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py b/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
index 712e3d5d5..564012467 100644
--- a/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
+++ b/configs/_base_/datasets/imagenet_bs128_riformer_small_384.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- # ann_file='meta/val.txt',
+ ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
index db7b39d6b..8b751a7d8 100644
--- a/configs/riformer/metafile.yml
+++ b/configs/riformer/metafile.yml
@@ -26,7 +26,7 @@ Models:
Top 1 Accuracy: 76.90
Top 5 Accuracy: 93.06
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k_20230406-6741ce71.pth
Config: configs/riformer/riformer-s12_32xb128_in1k.py
Converted From:
Weights:
@@ -43,7 +43,7 @@ Models:
Top 1 Accuracy: 80.28
Top 5 Accuracy: 94.80
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k_20230406-fdab072a.pth
Config: configs/riformer/riformer-s24_32xb128_in1k.py
Converted From:
Weights:
@@ -59,7 +59,7 @@ Models:
Top 1 Accuracy: 81.29
Top 5 Accuracy: 95.41
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k_20230406-fdfcd3b0.pth
Config: configs/riformer/riformer-s36_32xb128_in1k.py
Converted From:
Weights:
@@ -76,7 +76,7 @@ Models:
Top 1 Accuracy: 82.57
Top 5 Accuracy: 95.99
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k_20230406-2fcb9d9b.pth
Config: configs/riformer/riformer-m36_32xb128_in1k.py
Converted From:
Weights:
@@ -92,7 +92,7 @@ Models:
Top 1 Accuracy: 82.75
Top 5 Accuracy: 96.11
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth
Config: configs/riformer/riformer-m48_32xb128_in1k.py
Converted From:
Weights:
@@ -108,7 +108,7 @@ Models:
Top 1 Accuracy: 78.29
Top 5 Accuracy: 93.93
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth
Config: configs/riformer/riformer-s12_32xb128_in1k_384.py
Converted From:
Weights:
@@ -125,7 +125,7 @@ Models:
Top 1 Accuracy: 81.36
Top 5 Accuracy: 95.40
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth
Config: configs/riformer/riformer-s24_32xb128_in1k_384.py
Converted From:
Weights:
@@ -141,7 +141,7 @@ Models:
Top 1 Accuracy: 82.22
Top 5 Accuracy: 95.95
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth
Config: configs/riformer/riformer-s36_32xb128_in1k_384.py
Converted From:
Weights:
@@ -158,7 +158,7 @@ Models:
Top 1 Accuracy: 83.39
Top 5 Accuracy: 96.40
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth
Config: configs/riformer/riformer-m36_32xb128_in1k_384.py
Converted From:
Weights:
@@ -174,7 +174,7 @@ Models:
Top 1 Accuracy: 83.70
Top 5 Accuracy: 96.60
Task: Image Classification
- Weights:
+ Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth
Config: configs/riformer/riformer-m48_32xb128_in1k_384.py
Converted From:
Weights:
From e115ac89f470f8bf5bb049b2172c13b1d244eefd Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Thu, 6 Apr 2023 20:31:21 +0800
Subject: [PATCH 10/14] update riformer mmpretrain
---
.../imagenet_bs128_poolformer_small_224.py | 2 +-
.../_base_/models/riformer/riformer_m36.py | 22 ---------
.../_base_/models/riformer/riformer_m48.py | 22 ---------
.../_base_/models/riformer/riformer_s12.py | 22 ---------
.../_base_/models/riformer/riformer_s24.py | 22 ---------
.../_base_/models/riformer/riformer_s36.py | 22 ---------
configs/riformer/README.md | 46 +++++++++----------
configs/riformer/metafile.yml | 20 ++++----
...28_in1k.py => riformer-m36_8xb128_in1k.py} | 0
..._384.py => riformer-m36_8xb64_in1k_384.py} | 0
...128_in1k.py => riformer-m48_8xb64_in1k.py} | 0
..._384.py => riformer-m48_8xb64_in1k_384.py} | 0
...28_in1k.py => riformer-s12_8xb128_in1k.py} | 0
...384.py => riformer-s12_8xb128_in1k_384.py} | 0
...28_in1k.py => riformer-s24_8xb128_in1k.py} | 0
...384.py => riformer-s24_8xb128_in1k_384.py} | 0
...28_in1k.py => riformer-s36_8xb128_in1k.py} | 0
..._384.py => riformer-s36_8xb64_in1k_384.py} | 0
18 files changed, 34 insertions(+), 144 deletions(-)
delete mode 100644 configs/_base_/models/riformer/riformer_m36.py
delete mode 100644 configs/_base_/models/riformer/riformer_m48.py
delete mode 100644 configs/_base_/models/riformer/riformer_s12.py
delete mode 100644 configs/_base_/models/riformer/riformer_s24.py
delete mode 100644 configs/_base_/models/riformer/riformer_s36.py
rename configs/riformer/{riformer-m36_32xb128_in1k.py => riformer-m36_8xb128_in1k.py} (100%)
rename configs/riformer/{riformer-m36_32xb128_in1k_384.py => riformer-m36_8xb64_in1k_384.py} (100%)
rename configs/riformer/{riformer-m48_32xb128_in1k.py => riformer-m48_8xb64_in1k.py} (100%)
rename configs/riformer/{riformer-m48_32xb128_in1k_384.py => riformer-m48_8xb64_in1k_384.py} (100%)
rename configs/riformer/{riformer-s12_32xb128_in1k.py => riformer-s12_8xb128_in1k.py} (100%)
rename configs/riformer/{riformer-s12_32xb128_in1k_384.py => riformer-s12_8xb128_in1k_384.py} (100%)
rename configs/riformer/{riformer-s24_32xb128_in1k.py => riformer-s24_8xb128_in1k.py} (100%)
rename configs/riformer/{riformer-s24_32xb128_in1k_384.py => riformer-s24_8xb128_in1k_384.py} (100%)
rename configs/riformer/{riformer-s36_32xb128_in1k.py => riformer-s36_8xb128_in1k.py} (100%)
rename configs/riformer/{riformer-s36_32xb128_in1k_384.py => riformer-s36_8xb64_in1k_384.py} (100%)
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
index 8a4b63d75..a36b5afb9 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- ann_file='meta/val.txt',
+ # ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
diff --git a/configs/_base_/models/riformer/riformer_m36.py b/configs/_base_/models/riformer/riformer_m36.py
deleted file mode 100644
index 99dc74848..000000000
--- a/configs/_base_/models/riformer/riformer_m36.py
+++ /dev/null
@@ -1,22 +0,0 @@
-# Model settings
-model = dict(
- type='ImageClassifier',
- backbone=dict(
- type='RIFormer',
- arch='m36',
- drop_path_rate=0.1,
- init_cfg=[
- dict(
- type='TruncNormal',
- layer=['Conv2d', 'Linear'],
- std=.02,
- bias=0.),
- dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
- ]),
- neck=dict(type='GlobalAveragePooling'),
- head=dict(
- type='LinearClsHead',
- num_classes=1000,
- in_channels=768,
- loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
- ))
diff --git a/configs/_base_/models/riformer/riformer_m48.py b/configs/_base_/models/riformer/riformer_m48.py
deleted file mode 100644
index 934f15619..000000000
--- a/configs/_base_/models/riformer/riformer_m48.py
+++ /dev/null
@@ -1,22 +0,0 @@
-# Model settings
-model = dict(
- type='ImageClassifier',
- backbone=dict(
- type='RIFormer',
- arch='m48',
- drop_path_rate=0.1,
- init_cfg=[
- dict(
- type='TruncNormal',
- layer=['Conv2d', 'Linear'],
- std=.02,
- bias=0.),
- dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
- ]),
- neck=dict(type='GlobalAveragePooling'),
- head=dict(
- type='LinearClsHead',
- num_classes=1000,
- in_channels=768,
- loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
- ))
diff --git a/configs/_base_/models/riformer/riformer_s12.py b/configs/_base_/models/riformer/riformer_s12.py
deleted file mode 100644
index d3d1587ef..000000000
--- a/configs/_base_/models/riformer/riformer_s12.py
+++ /dev/null
@@ -1,22 +0,0 @@
-# Model settings
-model = dict(
- type='ImageClassifier',
- backbone=dict(
- type='RIFormer',
- arch='s12',
- drop_path_rate=0.1,
- init_cfg=[
- dict(
- type='TruncNormal',
- layer=['Conv2d', 'Linear'],
- std=.02,
- bias=0.),
- dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
- ]),
- neck=dict(type='GlobalAveragePooling'),
- head=dict(
- type='LinearClsHead',
- num_classes=1000,
- in_channels=512,
- loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
- ))
diff --git a/configs/_base_/models/riformer/riformer_s24.py b/configs/_base_/models/riformer/riformer_s24.py
deleted file mode 100644
index 9fc8089be..000000000
--- a/configs/_base_/models/riformer/riformer_s24.py
+++ /dev/null
@@ -1,22 +0,0 @@
-# Model settings
-model = dict(
- type='ImageClassifier',
- backbone=dict(
- type='RIFormer',
- arch='s24',
- drop_path_rate=0.1,
- init_cfg=[
- dict(
- type='TruncNormal',
- layer=['Conv2d', 'Linear'],
- std=.02,
- bias=0.),
- dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
- ]),
- neck=dict(type='GlobalAveragePooling'),
- head=dict(
- type='LinearClsHead',
- num_classes=1000,
- in_channels=512,
- loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
- ))
diff --git a/configs/_base_/models/riformer/riformer_s36.py b/configs/_base_/models/riformer/riformer_s36.py
deleted file mode 100644
index be0ad0e0f..000000000
--- a/configs/_base_/models/riformer/riformer_s36.py
+++ /dev/null
@@ -1,22 +0,0 @@
-# Model settings
-model = dict(
- type='ImageClassifier',
- backbone=dict(
- type='RIFormer',
- arch='s36',
- drop_path_rate=0.1,
- init_cfg=[
- dict(
- type='TruncNormal',
- layer=['Conv2d', 'Linear'],
- std=.02,
- bias=0.),
- dict(type='Constant', layer=['GroupNorm'], val=1., bias=0.),
- ]),
- neck=dict(type='GlobalAveragePooling'),
- head=dict(
- type='LinearClsHead',
- num_classes=1000,
- in_channels=512,
- loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
- ))
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
index f124c13b6..5b9691604 100644
--- a/configs/riformer/README.md
+++ b/configs/riformer/README.md
@@ -38,7 +38,7 @@ Use `classifier.backbone.switch_to_deploy()` interface to switch the RIFormer mo
>>> import torch
>>> from mmcls.apis import init_model, inference_model
>>>
->>> model = init_model('configs/riformer/riformer-s12_32xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
+>>> model = init_model('configs/riformer/riformer-s12_8xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
>>> results = inference_model(model, 'demo/demo.JPEG')
>>> print( (results['pred_class'], results['pred_score']) )
('sea snake' 0.7827475666999817)
@@ -55,7 +55,7 @@ Use `classifier.backbone.switch_to_deploy()` interface to switch the RIFormer mo
```python
>>> import torch
>>>
->>> model = init_model('configs/riformer/riformer-s12_32xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
+>>> model = init_model('configs/riformer/riformer-s12_8xb128_in1k.py', '/home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar')
>>> model.eval()
>>> inputs = torch.rand(1, 3, 224, 224).to(model.data_preprocessor.device)
>>> # To get classification scores.
@@ -90,19 +90,19 @@ wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repi
Test use unfused model:
```shell
-python tools/test.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/riformer-s12_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
```
Reparameterize checkpoint:
```shell
-python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
```
Test use fused model:
```shell
-python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
```
*384×384*
@@ -116,19 +116,19 @@ wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repi
Test use unfused model:
```shell
-python tools/test.py configs/riformer/riformer-s12_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/riformer-s12_8xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar
```
Reparameterize checkpoint:
```shell
-python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_8xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/384/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
```
Test use fused model:
```shell
-python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/384/repidentityformer-s12.pth.tar
```
@@ -156,20 +156,20 @@ For example:
wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
# reparameterize unfused weight to fused weight
-python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
```
-To use reparameterized weights, you can use the deploy model config file such as the [s12_deploy example](./riformer-s12-deploy_32xb128_in1k.py):
+To use reparameterized weights, you can use the deploy model config file such as the [s12_deploy example](./riformer-s12-deploy_8xb128_in1k.py):
```text
-# in riformer-s12-deploy_32xb128_in1k.py
-_base_ = '../riformer-s12-deploy_32xb128_in1k.py' # basic s12 config
+# in riformer-s12-deploy_8xb128_in1k.py
+_base_ = '../riformer-s12-deploy_8xb128_in1k.py' # basic s12 config
model = dict(backbone=dict(deploy=True)) # switch model into deploy mode
```
```shell
-python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
```
@@ -182,16 +182,16 @@ python tools/test.py configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
| Model | resolution | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download |
| :----------: | :--------: | :-------: | :------: | :-------: | :-------: | :------------------------------------------: | :-------------------------------------------------------------------------------------------------: |
-| RIFormer-S12 | 224x224 | 11.92 | 1.82 | 76.90 | 93.06 | [config](./riformer-s12_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s12_3rdparty_32xb128_in1k_20220414-f8d83051.pth) |
-| RIFormer-S24 | 224x224 | 21.39 | 3.41 | 80.28 | 94.80 | [config](./riformer-s24_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s24_3rdparty_32xb128_in1k_20220414-d7055904.pth) |
-| RIFormer-S36 | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s36_3rdparty_32xb128_in1k_20220414-d78ff3e8.pth) |
-| RIFormer-M36 | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m36_3rdparty_32xb128_in1k_20220414-c55e0949.pth) |
-| RIFormer-M48 | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_32xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m48_3rdparty_32xb128_in1k_20220414-9378f3eb.pth) |
-| RIFormer-S12 | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s12_3rdparty_32xb128_in1k_20220414-f8d83051.pth) |
-| RIFormer-S24 | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s24_3rdparty_32xb128_in1k_20220414-d7055904.pth) |
-| RIFormer-S36 | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-s36_3rdparty_32xb128_in1k_20220414-d78ff3e8.pth) |
-| RIFormer-M36 | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m36_3rdparty_32xb128_in1k_20220414-c55e0949.pth) |
-| RIFormer-M48 | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_32xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v0/poolformer/poolformer-m48_3rdparty_32xb128_in1k_20220414-9378f3eb.pth) |
+| RIFormer-S12 | 224x224 | 11.92 | 1.82 | 76.90 | 93.06 | [config](./riformer-s12_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_8xb128_in1k_20230406-6741ce71.pth) |
+| RIFormer-S24 | 224x224 | 21.39 | 3.41 | 80.28 | 94.80 | [config](./riformer-s24_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_8xb128_in1k_20230406-fdab072a.pth) |
+| RIFormer-S36 | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_8xb128_in1k_20230406-fdfcd3b0.pth) |
+| RIFormer-M36 | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_8xb128_in1k_20230406-2fcb9d9b.pth) |
+| RIFormer-M48 | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_8xb64_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_8xb128_in1k_20230406-2b9d1abf.pth) |
+| RIFormer-S12 | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_8xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_8xb128_in1k-384px_20230406-145eda4c.pth) |
+| RIFormer-S24 | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_8xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_8xb128_in1k-384px_20230406-bafae7ab.pth) |
+| RIFormer-S36 | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_8xb128_in1k-384px_20230406-017ed3c4.pth) |
+| RIFormer-M36 | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_8xb128_in1k-384px_20230406-66a6f764.pth) |
+| RIFormer-M48 | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_8xb128_in1k-384px_20230406-2e874826.pth) |
The config files of these models are only for inference.
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
index 8b751a7d8..b1b9deabf 100644
--- a/configs/riformer/metafile.yml
+++ b/configs/riformer/metafile.yml
@@ -27,7 +27,7 @@ Models:
Top 5 Accuracy: 93.06
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k_20230406-6741ce71.pth
- Config: configs/riformer/riformer-s12_32xb128_in1k.py
+ Config: configs/riformer/riformer-s12_8xb128_in1k.py
Converted From:
Weights:
Code:
@@ -44,7 +44,7 @@ Models:
Top 5 Accuracy: 94.80
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k_20230406-fdab072a.pth
- Config: configs/riformer/riformer-s24_32xb128_in1k.py
+ Config: configs/riformer/riformer-s24_8xb128_in1k.py
Converted From:
Weights:
Code:
@@ -60,7 +60,7 @@ Models:
Top 5 Accuracy: 95.41
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k_20230406-fdfcd3b0.pth
- Config: configs/riformer/riformer-s36_32xb128_in1k.py
+ Config: configs/riformer/riformer-s36_8xb128_in1k.py
Converted From:
Weights:
Code:
@@ -77,7 +77,7 @@ Models:
Top 5 Accuracy: 95.99
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k_20230406-2fcb9d9b.pth
- Config: configs/riformer/riformer-m36_32xb128_in1k.py
+ Config: configs/riformer/riformer-m36_8xb128_in1k.py
Converted From:
Weights:
Code:
@@ -93,7 +93,7 @@ Models:
Top 5 Accuracy: 96.11
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth
- Config: configs/riformer/riformer-m48_32xb128_in1k.py
+ Config: configs/riformer/riformer-m48_8xb64_in1k.py
Converted From:
Weights:
Code:
@@ -109,7 +109,7 @@ Models:
Top 5 Accuracy: 93.93
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth
- Config: configs/riformer/riformer-s12_32xb128_in1k_384.py
+ Config: configs/riformer/riformer-s12_8xb128_in1k_384.py
Converted From:
Weights:
Code:
@@ -126,7 +126,7 @@ Models:
Top 5 Accuracy: 95.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth
- Config: configs/riformer/riformer-s24_32xb128_in1k_384.py
+ Config: configs/riformer/riformer-s24_8xb128_in1k_384.py
Converted From:
Weights:
Code:
@@ -142,7 +142,7 @@ Models:
Top 5 Accuracy: 95.95
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth
- Config: configs/riformer/riformer-s36_32xb128_in1k_384.py
+ Config: configs/riformer/riformer-s36_8xb64_in1k_384.py
Converted From:
Weights:
Code:
@@ -159,7 +159,7 @@ Models:
Top 5 Accuracy: 96.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth
- Config: configs/riformer/riformer-m36_32xb128_in1k_384.py
+ Config: configs/riformer/riformer-m36_8xb64_in1k_384.py
Converted From:
Weights:
Code:
@@ -175,7 +175,7 @@ Models:
Top 5 Accuracy: 96.60
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth
- Config: configs/riformer/riformer-m48_32xb128_in1k_384.py
+ Config: configs/riformer/riformer-m48_8xb64_in1k_384.py
Converted From:
Weights:
Code:
diff --git a/configs/riformer/riformer-m36_32xb128_in1k.py b/configs/riformer/riformer-m36_8xb128_in1k.py
similarity index 100%
rename from configs/riformer/riformer-m36_32xb128_in1k.py
rename to configs/riformer/riformer-m36_8xb128_in1k.py
diff --git a/configs/riformer/riformer-m36_32xb128_in1k_384.py b/configs/riformer/riformer-m36_8xb64_in1k_384.py
similarity index 100%
rename from configs/riformer/riformer-m36_32xb128_in1k_384.py
rename to configs/riformer/riformer-m36_8xb64_in1k_384.py
diff --git a/configs/riformer/riformer-m48_32xb128_in1k.py b/configs/riformer/riformer-m48_8xb64_in1k.py
similarity index 100%
rename from configs/riformer/riformer-m48_32xb128_in1k.py
rename to configs/riformer/riformer-m48_8xb64_in1k.py
diff --git a/configs/riformer/riformer-m48_32xb128_in1k_384.py b/configs/riformer/riformer-m48_8xb64_in1k_384.py
similarity index 100%
rename from configs/riformer/riformer-m48_32xb128_in1k_384.py
rename to configs/riformer/riformer-m48_8xb64_in1k_384.py
diff --git a/configs/riformer/riformer-s12_32xb128_in1k.py b/configs/riformer/riformer-s12_8xb128_in1k.py
similarity index 100%
rename from configs/riformer/riformer-s12_32xb128_in1k.py
rename to configs/riformer/riformer-s12_8xb128_in1k.py
diff --git a/configs/riformer/riformer-s12_32xb128_in1k_384.py b/configs/riformer/riformer-s12_8xb128_in1k_384.py
similarity index 100%
rename from configs/riformer/riformer-s12_32xb128_in1k_384.py
rename to configs/riformer/riformer-s12_8xb128_in1k_384.py
diff --git a/configs/riformer/riformer-s24_32xb128_in1k.py b/configs/riformer/riformer-s24_8xb128_in1k.py
similarity index 100%
rename from configs/riformer/riformer-s24_32xb128_in1k.py
rename to configs/riformer/riformer-s24_8xb128_in1k.py
diff --git a/configs/riformer/riformer-s24_32xb128_in1k_384.py b/configs/riformer/riformer-s24_8xb128_in1k_384.py
similarity index 100%
rename from configs/riformer/riformer-s24_32xb128_in1k_384.py
rename to configs/riformer/riformer-s24_8xb128_in1k_384.py
diff --git a/configs/riformer/riformer-s36_32xb128_in1k.py b/configs/riformer/riformer-s36_8xb128_in1k.py
similarity index 100%
rename from configs/riformer/riformer-s36_32xb128_in1k.py
rename to configs/riformer/riformer-s36_8xb128_in1k.py
diff --git a/configs/riformer/riformer-s36_32xb128_in1k_384.py b/configs/riformer/riformer-s36_8xb64_in1k_384.py
similarity index 100%
rename from configs/riformer/riformer-s36_32xb128_in1k_384.py
rename to configs/riformer/riformer-s36_8xb64_in1k_384.py
From 5c3abb2b2ac94242da45215d68d4a7eb36f0c23f Mon Sep 17 00:00:00 2001
From: techmonsterwang <2959688260@qq.com>
Date: Thu, 6 Apr 2023 20:32:01 +0800
Subject: [PATCH 11/14] update riformer mmpretrain
---
configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
index a36b5afb9..8a4b63d75 100644
--- a/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
+++ b/configs/_base_/datasets/imagenet_bs128_poolformer_small_224.py
@@ -70,7 +70,7 @@ val_dataloader = dict(
dataset=dict(
type=dataset_type,
data_root='data/imagenet',
- # ann_file='meta/val.txt',
+ ann_file='meta/val.txt',
data_prefix='val',
pipeline=test_pipeline),
sampler=dict(type='DefaultSampler', shuffle=False),
From 3932ddec100bf5a90673eab219e54f64b494ae14 Mon Sep 17 00:00:00 2001
From: Ezra-Yu <18586273+Ezra-Yu@users.noreply.github.com>
Date: Thu, 6 Apr 2023 21:31:56 +0800
Subject: [PATCH 12/14] update ckpt path
---
configs/riformer/README.md | 84 +++++++------------
.../riformer-m36-deploy_32xb128_in1k.py | 3 -
.../riformer-m36-deploy_32xb128_in1k_384.py | 3 -
.../deploy/riformer-m36-deploy_8xb128_in1k.py | 3 +
.../riformer-m36-deploy_8xb64_in1k_384px.py | 3 +
.../riformer-m48-deploy_32xb128_in1k.py | 3 -
.../riformer-m48-deploy_32xb128_in1k_384.py | 3 -
.../deploy/riformer-m48-deploy_8xb64_in1k.py | 3 +
.../riformer-m48-deploy_8xb64_in1k_384px.py | 3 +
.../riformer-s12-deploy_32xb128_in1k.py | 3 -
.../riformer-s12-deploy_32xb128_in1k_384.py | 3 -
.../deploy/riformer-s12-deploy_8xb128_in1k.py | 3 +
.../riformer-s12-deploy_8xb128_in1k_384px.py | 3 +
.../riformer-s24-deploy_32xb128_in1k.py | 3 -
.../riformer-s24-deploy_32xb128_in1k_384.py | 3 -
.../deploy/riformer-s24-deploy_8xb128_in1k.py | 3 +
.../riformer-s24-deploy_8xb128_in1k_384px.py | 3 +
.../riformer-s36-deploy_32xb128_in1k.py | 3 -
.../riformer-s36-deploy_32xb128_in1k_384.py | 3 -
.../deploy/riformer-s36-deploy_8xb128_in1k.py | 3 +
.../riformer-s36-deploy_8xb64_in1k_384px.py | 3 +
configs/riformer/metafile.yml | 58 ++++---------
...84.py => riformer-m36_8xb64_in1k_384px.py} | 0
...84.py => riformer-m48_8xb64_in1k_384px.py} | 0
...4.py => riformer-s12_8xb128_in1k_384px.py} | 0
...4.py => riformer-s24_8xb128_in1k_384px.py} | 0
...84.py => riformer-s36_8xb64_in1k_384px.py} | 0
mmpretrain/models/backbones/riformer.py | 4 -
28 files changed, 73 insertions(+), 133 deletions(-)
delete mode 100644 configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
delete mode 100644 configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-m36-deploy_8xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
delete mode 100644 configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
delete mode 100644 configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k.py
create mode 100644 configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
delete mode 100644 configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
delete mode 100644 configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
delete mode 100644 configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
delete mode 100644 configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
delete mode 100644 configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
delete mode 100644 configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
create mode 100644 configs/riformer/deploy/riformer-s36-deploy_8xb128_in1k.py
create mode 100644 configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
rename configs/riformer/{riformer-m36_8xb64_in1k_384.py => riformer-m36_8xb64_in1k_384px.py} (100%)
rename configs/riformer/{riformer-m48_8xb64_in1k_384.py => riformer-m48_8xb64_in1k_384px.py} (100%)
rename configs/riformer/{riformer-s12_8xb128_in1k_384.py => riformer-s12_8xb128_in1k_384px.py} (100%)
rename configs/riformer/{riformer-s24_8xb128_in1k_384.py => riformer-s24_8xb128_in1k_384px.py} (100%)
rename configs/riformer/{riformer-s36_8xb64_in1k_384.py => riformer-s36_8xb64_in1k_384px.py} (100%)
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
index 5b9691604..97e5d1973 100644
--- a/configs/riformer/README.md
+++ b/configs/riformer/README.md
@@ -9,7 +9,7 @@
RIFormer is a way to keep a vision backbone effective while removing token mixers in its basic building blocks. Equipped with our proposed optimization strategy, we are able to build an extremely simple vision backbone with encouraging performance, while enjoying the high efficiency during inference. RIFormer shares nearly the same macro and micro design as MetaFormer, but safely removing all token mixers. The quantitative results show that our networks outperform many prevailing backbones with faster inference speed on ImageNet-1K.
-
+
@@ -153,23 +127,23 @@ For example:
```shell
# download the weight
-wget /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar
+wget https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k_20230406-6741ce71.pth
# reparameterize unfused weight to fused weight
-python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained/224/repidentityformer-s12.pth.tar /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/model_converters/reparameterize_model.py configs/riformer/riformer-s12_8xb128_in1k.py riformer-s12_32xb128_in1k_20230406-6741ce71.pth riformer-s12_deploy.pth
```
-To use reparameterized weights, you can use the deploy model config file such as the [s12_deploy example](./riformer-s12-deploy_8xb128_in1k.py):
+To use reparameterized weights, you can use the deploy model config file such as the [s12_deploy example](./deploy/riformer-s12-deploy_8xb128_in1k.py):
```text
# in riformer-s12-deploy_8xb128_in1k.py
-_base_ = '../riformer-s12-deploy_8xb128_in1k.py' # basic s12 config
+_base_ = '../deploy/riformer-s12-deploy_8xb128_in1k.py' # basic s12 config
model = dict(backbone=dict(deploy=True)) # switch model into deploy mode
```
```shell
-python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py /home/PJLAB/wangjiahao/project/RepIndentityFormer/mmcls_pretrained_deploy/224/repidentityformer-s12.pth.tar
+python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py riformer-s12_deploy.pth
```
@@ -180,18 +154,18 @@ python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py
### ImageNet-1k
-| Model | resolution | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download |
-| :----------: | :--------: | :-------: | :------: | :-------: | :-------: | :------------------------------------------: | :-------------------------------------------------------------------------------------------------: |
-| RIFormer-S12 | 224x224 | 11.92 | 1.82 | 76.90 | 93.06 | [config](./riformer-s12_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_8xb128_in1k_20230406-6741ce71.pth) |
-| RIFormer-S24 | 224x224 | 21.39 | 3.41 | 80.28 | 94.80 | [config](./riformer-s24_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_8xb128_in1k_20230406-fdab072a.pth) |
-| RIFormer-S36 | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_8xb128_in1k_20230406-fdfcd3b0.pth) |
-| RIFormer-M36 | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_8xb128_in1k_20230406-2fcb9d9b.pth) |
-| RIFormer-M48 | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_8xb64_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_8xb128_in1k_20230406-2b9d1abf.pth) |
-| RIFormer-S12 | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_8xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_8xb128_in1k-384px_20230406-145eda4c.pth) |
-| RIFormer-S24 | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_8xb128_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_8xb128_in1k-384px_20230406-bafae7ab.pth) |
-| RIFormer-S36 | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_8xb128_in1k-384px_20230406-017ed3c4.pth) |
-| RIFormer-M36 | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_8xb128_in1k-384px_20230406-66a6f764.pth) |
-| RIFormer-M48 | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_8xb64_in1k_384.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_8xb128_in1k-384px_20230406-2e874826.pth) |
+| Model | resolution | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download |
+| :-------------------: | :--------: | :-------: | :------: | :-------: | :-------: | :-------------------------------------------: | :---------------------------------------------------------------------------------------: |
+| riformer-s12_in1k | 224x224 | 11.92 | 1.82 | 76.90 | 93.06 | [config](./riformer-s12_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k_20230406-6741ce71.pth) |
+| riformer-s24_in1k | 224x224 | 21.39 | 3.41 | 80.28 | 94.80 | [config](./riformer-s24_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k_20230406-fdab072a.pth) |
+| riformer-s36_in1k | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k_20230406-fdfcd3b0.pth) |
+| riformer-m36_in1k | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k_20230406-2fcb9d9b.pth) |
+| riformer-m48_in1k | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_8xb64_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth) |
+| riformer-s12_384_in1k | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_8xb128_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth) |
+| riformer-s24_384_in1k | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_8xb128_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth) |
+| riformer-s36_384_in1k | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth) |
+| riformer-m36_384_in1k | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth) |
+| riformer-m48_384_in1k | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth) |
The config files of these models are only for inference.
@@ -199,7 +173,7 @@ The config files of these models are only for inference.
```bibtex
@inproceedings{wang2023riformer,
- title={MetaFormer is Actually What You Need for Vision},
+ title={RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer},
author={Wang, Jiahao and Zhang, Songyang and Liu, Yong and Wu, Taiqiang and Yang, Yujiu and Liu, Xihui and Chen, Kai and Luo, Ping and Lin, Dahua},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year={2023}
diff --git a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
deleted file mode 100644
index ae7c97105..000000000
--- a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m36_32xb128_in1k.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
deleted file mode 100644
index 4eac9954d..000000000
--- a/configs/riformer/deploy/riformer-m36-deploy_32xb128_in1k_384.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m36_32xb128_in1k_384.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m36-deploy_8xb128_in1k.py b/configs/riformer/deploy/riformer-m36-deploy_8xb128_in1k.py
new file mode 100644
index 000000000..fcec41c81
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m36-deploy_8xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m36_8xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
new file mode 100644
index 000000000..beefdf193
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m36_8xb64_in1k_384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
deleted file mode 100644
index 7d999ceff..000000000
--- a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m48_32xb128_in1k.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
deleted file mode 100644
index bf5d76dd7..000000000
--- a/configs/riformer/deploy/riformer-m48-deploy_32xb128_in1k_384.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m48_32xb128_in1k_384.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k.py b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k.py
new file mode 100644
index 000000000..e32ad328f
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m48_8xb64_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
new file mode 100644
index 000000000..c80fea077
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m48_8xb64_in1k_384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
deleted file mode 100644
index 784453f0e..000000000
--- a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s12_32xb128_in1k.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
deleted file mode 100644
index a4a10723f..000000000
--- a/configs/riformer/deploy/riformer-s12-deploy_32xb128_in1k_384.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s12_32xb128_in1k_384.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py
new file mode 100644
index 000000000..70fd8b743
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s12_8xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
new file mode 100644
index 000000000..3f9a77715
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s12_8xb128_in1k_384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
deleted file mode 100644
index f439ea084..000000000
--- a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s24_32xb128_in1k.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
deleted file mode 100644
index 32fa0363e..000000000
--- a/configs/riformer/deploy/riformer-s24-deploy_32xb128_in1k_384.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s24_32xb128_in1k_384.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k.py b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k.py
new file mode 100644
index 000000000..47f83a08f
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s24_8xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
new file mode 100644
index 000000000..3f98a7ede
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s24_8xb128_in1k_384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
deleted file mode 100644
index 509898a72..000000000
--- a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s36_32xb128_in1k.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py b/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
deleted file mode 100644
index d684983ed..000000000
--- a/configs/riformer/deploy/riformer-s36-deploy_32xb128_in1k_384.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s36_32xb128_in1k_384.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_8xb128_in1k.py b/configs/riformer/deploy/riformer-s36-deploy_8xb128_in1k.py
new file mode 100644
index 000000000..2c03bb151
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s36-deploy_8xb128_in1k.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s36_8xb128_in1k.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
new file mode 100644
index 000000000..9c737a2e1
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s36_8xb64_in1k_384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
index b1b9deabf..94fb77689 100644
--- a/configs/riformer/metafile.yml
+++ b/configs/riformer/metafile.yml
@@ -12,7 +12,7 @@ Collections:
README: configs/riformer/README.md
Code:
Version: v1.0.rc6
- URL:
+ URL: null
Models:
- Name: riformer-s12_in1k
@@ -28,15 +28,12 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k_20230406-6741ce71.pth
Config: configs/riformer/riformer-s12_8xb128_in1k.py
- Converted From:
- Weights:
- Code:
- Name: riformer-s24_in1k
Metadata:
Training Data: ImageNet-1k
FLOPs: 3412000000
Parameters: 21389000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -45,14 +42,11 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k_20230406-fdab072a.pth
Config: configs/riformer/riformer-s24_8xb128_in1k.py
- Converted From:
- Weights:
- Code:
- Name: riformer-s36_in1k
Metadata:
FLOPs: 5003000000
Parameters: 30863000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -61,15 +55,12 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k_20230406-fdfcd3b0.pth
Config: configs/riformer/riformer-s36_8xb128_in1k.py
- Converted From:
- Weights:
- Code:
- Name: riformer-m36_in1k
Metadata:
Training Data: ImageNet-1k
FLOPs: 8801000000
Parameters: 56173000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -78,14 +69,11 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k_20230406-2fcb9d9b.pth
Config: configs/riformer/riformer-m36_8xb128_in1k.py
- Converted From:
- Weights:
- Code:
- Name: riformer-m48_in1k
Metadata:
FLOPs: 11590000000
Parameters: 73473000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -94,9 +82,6 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth
Config: configs/riformer/riformer-m48_8xb64_in1k.py
- Converted From:
- Weights:
- Code:
- Name: riformer-s12_384_in1k
Metadata:
FLOPs: 5355000000
@@ -109,16 +94,13 @@ Models:
Top 5 Accuracy: 93.93
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth
- Config: configs/riformer/riformer-s12_8xb128_in1k_384.py
- Converted From:
- Weights:
- Code:
+ Config: configs/riformer/riformer-s12_8xb128_in1k_384px.py
- Name: riformer-s24_384_in1k
Metadata:
Training Data: ImageNet-1k
FLOPs: 10028000000
Parameters: 21389000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -126,15 +108,12 @@ Models:
Top 5 Accuracy: 95.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth
- Config: configs/riformer/riformer-s24_8xb128_in1k_384.py
- Converted From:
- Weights:
- Code:
+ Config: configs/riformer/riformer-s24_8xb128_in1k_384px.py
- Name: riformer-s36_384_in1k
Metadata:
FLOPs: 14702000000
Parameters: 30863000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -142,16 +121,13 @@ Models:
Top 5 Accuracy: 95.95
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth
- Config: configs/riformer/riformer-s36_8xb64_in1k_384.py
- Converted From:
- Weights:
- Code:
+ Config: configs/riformer/riformer-s36_8xb64_in1k_384px.py
- Name: riformer-m36_384_in1k
Metadata:
Training Data: ImageNet-1k
FLOPs: 25865000000
Parameters: 56173000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -159,15 +135,12 @@ Models:
Top 5 Accuracy: 96.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth
- Config: configs/riformer/riformer-m36_8xb64_in1k_384.py
- Converted From:
- Weights:
- Code:
+ Config: configs/riformer/riformer-m36_8xb64_in1k_384px.py
- Name: riformer-m48_384_in1k
Metadata:
FLOPs: 34060000000
Parameters: 73473000
- In Collection: PoolFormer
+ In Collection: RIFormer
Results:
- Dataset: ImageNet-1k
Metrics:
@@ -175,7 +148,4 @@ Models:
Top 5 Accuracy: 96.60
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth
- Config: configs/riformer/riformer-m48_8xb64_in1k_384.py
- Converted From:
- Weights:
- Code:
+ Config: configs/riformer/riformer-m48_8xb64_in1k_384px.py
diff --git a/configs/riformer/riformer-m36_8xb64_in1k_384.py b/configs/riformer/riformer-m36_8xb64_in1k_384px.py
similarity index 100%
rename from configs/riformer/riformer-m36_8xb64_in1k_384.py
rename to configs/riformer/riformer-m36_8xb64_in1k_384px.py
diff --git a/configs/riformer/riformer-m48_8xb64_in1k_384.py b/configs/riformer/riformer-m48_8xb64_in1k_384px.py
similarity index 100%
rename from configs/riformer/riformer-m48_8xb64_in1k_384.py
rename to configs/riformer/riformer-m48_8xb64_in1k_384px.py
diff --git a/configs/riformer/riformer-s12_8xb128_in1k_384.py b/configs/riformer/riformer-s12_8xb128_in1k_384px.py
similarity index 100%
rename from configs/riformer/riformer-s12_8xb128_in1k_384.py
rename to configs/riformer/riformer-s12_8xb128_in1k_384px.py
diff --git a/configs/riformer/riformer-s24_8xb128_in1k_384.py b/configs/riformer/riformer-s24_8xb128_in1k_384px.py
similarity index 100%
rename from configs/riformer/riformer-s24_8xb128_in1k_384.py
rename to configs/riformer/riformer-s24_8xb128_in1k_384px.py
diff --git a/configs/riformer/riformer-s36_8xb64_in1k_384.py b/configs/riformer/riformer-s36_8xb64_in1k_384px.py
similarity index 100%
rename from configs/riformer/riformer-s36_8xb64_in1k_384.py
rename to configs/riformer/riformer-s36_8xb64_in1k_384px.py
diff --git a/mmpretrain/models/backbones/riformer.py b/mmpretrain/models/backbones/riformer.py
index c2e2c8916..2b00e9403 100644
--- a/mmpretrain/models/backbones/riformer.py
+++ b/mmpretrain/models/backbones/riformer.py
@@ -110,10 +110,6 @@ class RIFormerBlock(BaseModule):
beta_affn = token_mixer.affine.bias
gamma_ln = norm.weight
beta_ln = norm.bias
- print('gamma_affn:', gamma_affn.shape)
- print('beta_affn:', beta_affn.shape)
- print('gamma_ln:', gamma_ln.shape)
- print('beta_ln:', beta_ln.shape)
return (gamma_ln * gamma_affn), (beta_ln * gamma_affn + beta_affn)
def get_equivalent_scale_bias(self):
From b8cab5c9f760030a02b0d448d904717477db095b Mon Sep 17 00:00:00 2001
From: Ezra-Yu <18586273+Ezra-Yu@users.noreply.github.com>
Date: Thu, 6 Apr 2023 21:54:57 +0800
Subject: [PATCH 13/14] update readme
---
configs/riformer/README.md | 10 ++++----
.../riformer-m36-deploy_8xb64_in1k-384px.py | 3 +++
.../riformer-m36-deploy_8xb64_in1k_384px.py | 3 ---
.../riformer-m48-deploy_8xb64_in1k-384px.py | 3 +++
.../riformer-m48-deploy_8xb64_in1k_384px.py | 3 ---
.../riformer-s12-deploy_8xb128_in1k-384px.py | 3 +++
.../riformer-s12-deploy_8xb128_in1k_384px.py | 3 ---
.../riformer-s24-deploy_8xb128_in1k-384px.py | 3 +++
.../riformer-s24-deploy_8xb128_in1k_384px.py | 3 ---
.../riformer-s36-deploy_8xb64_in1k-384px.py | 3 +++
.../riformer-s36-deploy_8xb64_in1k_384px.py | 3 ---
configs/riformer/metafile.yml | 22 ++++++++---------
...px.py => riformer-m36_8xb64_in1k-384px.py} | 0
...px.py => riformer-m48_8xb64_in1k-384px.py} | 0
...x.py => riformer-s12_8xb128_in1k-384px.py} | 0
...x.py => riformer-s24_8xb128_in1k-384px.py} | 0
...px.py => riformer-s36_8xb64_in1k-384px.py} | 0
mmpretrain/models/backbones/riformer.py | 24 +++++--------------
18 files changed, 37 insertions(+), 49 deletions(-)
create mode 100644 configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k-384px.py
delete mode 100644 configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
create mode 100644 configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k-384px.py
delete mode 100644 configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
create mode 100644 configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k-384px.py
delete mode 100644 configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
create mode 100644 configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k-384px.py
delete mode 100644 configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
create mode 100644 configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k-384px.py
delete mode 100644 configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
rename configs/riformer/{riformer-m36_8xb64_in1k_384px.py => riformer-m36_8xb64_in1k-384px.py} (100%)
rename configs/riformer/{riformer-m48_8xb64_in1k_384px.py => riformer-m48_8xb64_in1k-384px.py} (100%)
rename configs/riformer/{riformer-s12_8xb128_in1k_384px.py => riformer-s12_8xb128_in1k-384px.py} (100%)
rename configs/riformer/{riformer-s24_8xb128_in1k_384px.py => riformer-s24_8xb128_in1k-384px.py} (100%)
rename configs/riformer/{riformer-s36_8xb64_in1k_384px.py => riformer-s36_8xb64_in1k-384px.py} (100%)
diff --git a/configs/riformer/README.md b/configs/riformer/README.md
index 97e5d1973..e24e32ad5 100644
--- a/configs/riformer/README.md
+++ b/configs/riformer/README.md
@@ -161,11 +161,11 @@ python tools/test.py configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k.py
| riformer-s36_in1k | 224x224 | 30.86 | 5.00 | 81.29 | 95.41 | [config](./riformer-s36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k_20230406-fdfcd3b0.pth) |
| riformer-m36_in1k | 224x224 | 56.17 | 8.80 | 82.57 | 95.99 | [config](./riformer-m36_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k_20230406-2fcb9d9b.pth) |
| riformer-m48_in1k | 224x224 | 73.47 | 11.59 | 82.75 | 96.11 | [config](./riformer-m48_8xb64_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth) |
-| riformer-s12_384_in1k | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_8xb128_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth) |
-| riformer-s24_384_in1k | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_8xb128_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth) |
-| riformer-s36_384_in1k | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth) |
-| riformer-m36_384_in1k | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth) |
-| riformer-m48_384_in1k | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_8xb64_in1k_384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth) |
+| riformer-s12_384_in1k | 384x384 | 11.92 | 5.36 | 78.29 | 93.93 | [config](./riformer-s12_8xb128_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth) |
+| riformer-s24_384_in1k | 384x384 | 21.39 | 10.03 | 81.36 | 95.40 | [config](./riformer-s24_8xb128_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth) |
+| riformer-s36_384_in1k | 384x384 | 30.86 | 14.70 | 82.22 | 95.95 | [config](./riformer-s36_8xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth) |
+| riformer-m36_384_in1k | 384x384 | 56.17 | 25.87 | 83.39 | 96.40 | [config](./riformer-m36_8xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth) |
+| riformer-m48_384_in1k | 384x384 | 73.47 | 34.06 | 83.70 | 96.60 | [config](./riformer-m48_8xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth) |
The config files of these models are only for inference.
diff --git a/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k-384px.py b/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k-384px.py
new file mode 100644
index 000000000..e18f836f8
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k-384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m36_8xb64_in1k-384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
deleted file mode 100644
index beefdf193..000000000
--- a/configs/riformer/deploy/riformer-m36-deploy_8xb64_in1k_384px.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m36_8xb64_in1k_384px.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k-384px.py b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k-384px.py
new file mode 100644
index 000000000..0ab33534e
--- /dev/null
+++ b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k-384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-m48_8xb64_in1k-384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
deleted file mode 100644
index c80fea077..000000000
--- a/configs/riformer/deploy/riformer-m48-deploy_8xb64_in1k_384px.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-m48_8xb64_in1k_384px.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k-384px.py b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k-384px.py
new file mode 100644
index 000000000..ffbb4be31
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k-384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s12_8xb128_in1k-384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py b/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
deleted file mode 100644
index 3f9a77715..000000000
--- a/configs/riformer/deploy/riformer-s12-deploy_8xb128_in1k_384px.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s12_8xb128_in1k_384px.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k-384px.py b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k-384px.py
new file mode 100644
index 000000000..7d05e5c1a
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k-384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s24_8xb128_in1k-384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py b/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
deleted file mode 100644
index 3f98a7ede..000000000
--- a/configs/riformer/deploy/riformer-s24-deploy_8xb128_in1k_384px.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s24_8xb128_in1k_384px.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k-384px.py b/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k-384px.py
new file mode 100644
index 000000000..67b17ee51
--- /dev/null
+++ b/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k-384px.py
@@ -0,0 +1,3 @@
+_base_ = '../riformer-s36_8xb64_in1k-384px.py'
+
+model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py b/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
deleted file mode 100644
index 9c737a2e1..000000000
--- a/configs/riformer/deploy/riformer-s36-deploy_8xb64_in1k_384px.py
+++ /dev/null
@@ -1,3 +0,0 @@
-_base_ = '../riformer-s36_8xb64_in1k_384px.py'
-
-model = dict(backbone=dict(deploy=True))
diff --git a/configs/riformer/metafile.yml b/configs/riformer/metafile.yml
index 94fb77689..a971ba1a4 100644
--- a/configs/riformer/metafile.yml
+++ b/configs/riformer/metafile.yml
@@ -11,7 +11,7 @@ Collections:
Title: "RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer"
README: configs/riformer/README.md
Code:
- Version: v1.0.rc6
+ Version: v1.0.0rc6
URL: null
Models:
@@ -82,7 +82,7 @@ Models:
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k_20230406-2b9d1abf.pth
Config: configs/riformer/riformer-m48_8xb64_in1k.py
- - Name: riformer-s12_384_in1k
+ - Name: riformer-s12_in1k-384
Metadata:
FLOPs: 5355000000
Parameters: 11915000
@@ -94,8 +94,8 @@ Models:
Top 5 Accuracy: 93.93
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s12_32xb128_in1k-384px_20230406-145eda4c.pth
- Config: configs/riformer/riformer-s12_8xb128_in1k_384px.py
- - Name: riformer-s24_384_in1k
+ Config: configs/riformer/riformer-s12_8xb128_in1k-384px.py
+ - Name: riformer-s24_in1k-384
Metadata:
Training Data: ImageNet-1k
FLOPs: 10028000000
@@ -108,8 +108,8 @@ Models:
Top 5 Accuracy: 95.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s24_32xb128_in1k-384px_20230406-bafae7ab.pth
- Config: configs/riformer/riformer-s24_8xb128_in1k_384px.py
- - Name: riformer-s36_384_in1k
+ Config: configs/riformer/riformer-s24_8xb128_in1k-384px.py
+ - Name: riformer-s36_in1k-384
Metadata:
FLOPs: 14702000000
Parameters: 30863000
@@ -121,8 +121,8 @@ Models:
Top 5 Accuracy: 95.95
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-s36_32xb128_in1k-384px_20230406-017ed3c4.pth
- Config: configs/riformer/riformer-s36_8xb64_in1k_384px.py
- - Name: riformer-m36_384_in1k
+ Config: configs/riformer/riformer-s36_8xb64_in1k-384px.py
+ - Name: riformer-m36_in1k-384
Metadata:
Training Data: ImageNet-1k
FLOPs: 25865000000
@@ -135,8 +135,8 @@ Models:
Top 5 Accuracy: 96.40
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m36_32xb128_in1k-384px_20230406-66a6f764.pth
- Config: configs/riformer/riformer-m36_8xb64_in1k_384px.py
- - Name: riformer-m48_384_in1k
+ Config: configs/riformer/riformer-m36_8xb64_in1k-384px.py
+ - Name: riformer-m48_in1k-384
Metadata:
FLOPs: 34060000000
Parameters: 73473000
@@ -148,4 +148,4 @@ Models:
Top 5 Accuracy: 96.60
Task: Image Classification
Weights: https://download.openmmlab.com/mmclassification/v1/riformer/riformer-m48_32xb128_in1k-384px_20230406-2e874826.pth
- Config: configs/riformer/riformer-m48_8xb64_in1k_384px.py
+ Config: configs/riformer/riformer-m48_8xb64_in1k-384px.py
diff --git a/configs/riformer/riformer-m36_8xb64_in1k_384px.py b/configs/riformer/riformer-m36_8xb64_in1k-384px.py
similarity index 100%
rename from configs/riformer/riformer-m36_8xb64_in1k_384px.py
rename to configs/riformer/riformer-m36_8xb64_in1k-384px.py
diff --git a/configs/riformer/riformer-m48_8xb64_in1k_384px.py b/configs/riformer/riformer-m48_8xb64_in1k-384px.py
similarity index 100%
rename from configs/riformer/riformer-m48_8xb64_in1k_384px.py
rename to configs/riformer/riformer-m48_8xb64_in1k-384px.py
diff --git a/configs/riformer/riformer-s12_8xb128_in1k_384px.py b/configs/riformer/riformer-s12_8xb128_in1k-384px.py
similarity index 100%
rename from configs/riformer/riformer-s12_8xb128_in1k_384px.py
rename to configs/riformer/riformer-s12_8xb128_in1k-384px.py
diff --git a/configs/riformer/riformer-s24_8xb128_in1k_384px.py b/configs/riformer/riformer-s24_8xb128_in1k-384px.py
similarity index 100%
rename from configs/riformer/riformer-s24_8xb128_in1k_384px.py
rename to configs/riformer/riformer-s24_8xb128_in1k-384px.py
diff --git a/configs/riformer/riformer-s36_8xb64_in1k_384px.py b/configs/riformer/riformer-s36_8xb64_in1k-384px.py
similarity index 100%
rename from configs/riformer/riformer-s36_8xb64_in1k_384px.py
rename to configs/riformer/riformer-s36_8xb64_in1k-384px.py
diff --git a/mmpretrain/models/backbones/riformer.py b/mmpretrain/models/backbones/riformer.py
index 2b00e9403..0e535e607 100644
--- a/mmpretrain/models/backbones/riformer.py
+++ b/mmpretrain/models/backbones/riformer.py
@@ -15,10 +15,10 @@ class Affine(nn.Module):
"""Affine Transformation module.
Args:
- in_features (int): Input dimension. Defaults to None.
+ in_features (int): Input dimension.
"""
- def __init__(self, in_features=None):
+ def __init__(self, in_features):
super().__init__()
self.affine = nn.Conv2d(
in_features,
@@ -139,10 +139,7 @@ def basic_blocks(dim,
drop_path_rate=0.,
layer_scale_init_value=1e-5,
deploy=False):
- """
- generate RIFormer blocks for a stage
- return: RIFormer blocks
- """
+ """generate RIFormer blocks for a stage"""
blocks = []
for block_idx in range(layers[index]):
block_dpr = drop_path_rate * (block_idx + sum(layers[:index])) / (
@@ -170,9 +167,6 @@ class RIFormer(BaseBackbone):
A PyTorch implementation of RIFormer introduced by:
`RIFormer: Keep Your Vision Backbone Effective But Removing Token Mixer `_
- Modified from the `official repo
- `.
-
Args:
arch (str | dict): The model's architecture. If string, it should be
one of architecture in ``RIFormer.arch_settings``. And if dict, it
@@ -253,6 +247,7 @@ class RIFormer(BaseBackbone):
def __init__(self,
arch='s12',
+ in_channels=3,
norm_cfg=dict(type='GN', num_groups=1),
act_cfg=dict(type='GELU'),
in_patch_size=7,
@@ -291,7 +286,7 @@ class RIFormer(BaseBackbone):
patch_size=in_patch_size,
stride=in_stride,
padding=in_pad,
- in_chans=3,
+ in_chans=in_channels,
embed_dim=embed_dims[0])
# set the main block in network
@@ -386,17 +381,10 @@ class RIFormer(BaseBackbone):
def train(self, mode=True):
super(RIFormer, self).train(mode)
self._freeze_stages()
+ return self
def switch_to_deploy(self):
for m in self.modules():
if isinstance(m, RIFormerBlock):
m.switch_to_deploy()
self.deploy = True
-
-
-if __name__ == '__main__':
- model = RIFormer(arch='s12', deploy=False)
- model.eval()
- print('------------------- training-time model -------------')
- for i in model.state_dict().keys():
- print(i)
From 05124dbb71fabae75a3100d6a3ad74398b540654 Mon Sep 17 00:00:00 2001
From: Ezra-Yu <18586273+Ezra-Yu@users.noreply.github.com>
Date: Thu, 6 Apr 2023 22:01:11 +0800
Subject: [PATCH 14/14] fix lint
---
mmpretrain/models/backbones/riformer.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mmpretrain/models/backbones/riformer.py b/mmpretrain/models/backbones/riformer.py
index 0e535e607..ac3ce23fd 100644
--- a/mmpretrain/models/backbones/riformer.py
+++ b/mmpretrain/models/backbones/riformer.py
@@ -139,7 +139,7 @@ def basic_blocks(dim,
drop_path_rate=0.,
layer_scale_init_value=1e-5,
deploy=False):
- """generate RIFormer blocks for a stage"""
+ """generate RIFormer blocks for a stage."""
blocks = []
for block_idx in range(layers[index]):
block_dpr = drop_path_rate * (block_idx + sum(layers[:index])) / (