mirrors
/
mmrazor
mirror of https://github.com/open-mmlab/mmrazor.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

add metafiles and fix readme

pull/479/head
HIT-cwh 2023-03-13 18:17:45 +08:00
parent 9446b301a3
commit 552cbf0f72
9 changed files with 215 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
configs/distill/mmcls/dist/metafile.yml vendored 100644
View File

@ -0,0 +1,34 @@
Collections:
- Name: DIST
Metadata:
Training Data:
- ImageNet-1k
Paper:
URL: https://arxiv.org/abs/2205.10536
Title: Knowledge Distillation from A Stronger Teacher
README: configs/distill/mmcls/dist/README.md
Models:
- Name: dist_logits_resnet34_resnet18_8xb32_in1k
In Collection: DIST
Metadata:
Location: logits
Student:
Config: mmcls::resnet/resnet18_8xb32_in1k.py
Weights: https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth
Metrics:
Top 1 Accuracy: 69.90
Top 5 Accuracy: 89.43
Teacher:
Config: mmcls::resnet/resnet34_8xb32_in1k.py
Weights: https://download.openmmlab.com/mmclassification/v0/resnet/resnet34_8xb32_in1k_20210831-f257d4e6.pth
Metrics:
Top 1 Accuracy: 73.62
Top 5 Accuracy: 91.59
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 71.61
Config: configs/distill/mmcls/dist/dist_logits_resnet34_resnet18_8xb32_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/distillation/dist_logits_resnet34_resnet18_8xb32_in1k.pth

32
configs/distill/mmdet/mgd/metafile.yml 100644
View File

@ -0,0 +1,32 @@
Collections:
- Name: MGD
Metadata:
Training Data:
- COCO
Paper:
URL: https://arxiv.org/abs/2205.01529
Title: Masked Generative Distillation
README: configs/distill/mmdet/mgd/README.md
Models:
- Name: mgd_fpn_retina_x101_retina_r50_2x_coco
In Collection: MGD
Metadata:
Location: FPN
Student:
Metrics:
box AP: 37.4
Config: mmdet::retinanet/retinanet_r50_fpn_2x_coco.py
Weights: https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_r50_fpn_2x_coco/retinanet_r50_fpn_2x_coco_20200131-fdb43119.pth
Teacher:
Metrics:
box AP: 41.0
Config: mmdet::retinanet/retinanet_x101-64x4d_fpn_1x_coco.py
Weights: https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_x101_64x4d_fpn_1x_coco/retinanet_x101_64x4d_fpn_1x_coco_20200130-366f5af1.pth
Results:
- Task: Object Detection
Dataset: COCO
Metrics:
box AP: 41.0
Config: configs/distill/mmdet/mgd/mgd_fpn_retina_x101_retina_r50_2x_coco.py
Weights: https://download.openmmlab.com/mmrazor/v1/mgd/mgd_fpn_retina_x101_retina_r50_2x_coco_20221209_191847-87141529.pth

27
configs/nas/mmcls/autoformer/metafile.yml 100644
View File

@ -0,0 +1,27 @@
Collections:
- Name: AUTOFORMER
Metadata:
Training Data:
- ImageNet-1k
Paper:
URL: https://arxiv.org/abs/2107.00651
Title: Searching Transformers for Visual Recognition
README: configs/nas/mmcls/autoformer/README.md
Code:
URL: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/mmrazor/models/algorithms/nas/autoformer.py
Version: V1.0.0rc2
Models:
- Name: autoformer_subnet_8xb256_in1k
In Collection: AUTOFORMER
Metadata:
Flops(G): 10.57
Params(M): 54.319
Subnet: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/autoformer/AUTOFORMER_SUBNET_B.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 82.47
Config: configs/nas/mmcls/autoformer/autoformer_subnet_8xb256_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/autoformer/autoformer_supernet_32xb256_in1k_20220919_110144-c658ce8f.pth

10
configs/nas/mmcls/bignas/README.md
View File

@ -37,11 +37,11 @@ CUDA_VISIBLE_DEVICES=0 PORT=29500 ./tools/dist_test.sh \
## Results and models
| Dataset | Supernet | Subnet | Params(M) | Flops(G) | Top-1 | Config | Download | Remarks |
| :------: | :------------------: | :-------------------------------------------------------------------------------------------------------------------------------: | :--------------------: | :------------------: | :---------------------: | :-------------------------------------------------------------------------------------------------------------------------------: | :--------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------: |
| ImageNet | AttentiveMobileNetV3 | [search space](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/_base_/nas_backbones/attentive_mobilenetv3_supernet.py) | 8.854(min) / 23.3(max) | 212(min) / 1944(max) | 77.19(min) / 81.42(max) | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_supernet_32xb64_in1k.py) | [model\*](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_supernet_32xb64_in1k_flops-2G_acc-81.72_20221229_200440-954772a3.pth) | [log](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_supernet_32xb64_in1k_20221227_175800-bcf94eaa.json) (`sandwich rule`) |
| ImageNet | AttentiveMobileNetV3 | [AttentiveNAS-A0\*](https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A0.yaml) | 8.854 | 212 | 77.19 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.21G_acc-77.19_20221229_200440-282a1f70.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | [AttentiveNAS-A6\*](https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A6.yaml) | 15.594 | 927 | 80.81 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.93G_acc-80.81_20221229_200440-73d92cc6.pth) | Converted from the repo |
| Dataset | Supernet | Subnet | Params(M) | Flops(G) | Top-1 | Config | Download | Remarks |
| :------: | :------------------: | :-------------------------------------------------------------------------------------------------------------------------------: | :--------------------: | :----------------------: | :---------------------: | :-------------------------------------------------------------------------------------------------------------------------------: | :--------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------: |
| ImageNet | AttentiveMobileNetV3 | [search space](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/_base_/nas_backbones/attentive_mobilenetv3_supernet.py) | 8.854(min) / 23.3(max) | 0.212(min) / 0.1944(max) | 77.19(min) / 81.42(max) | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_supernet_32xb64_in1k.py) | [model\*](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_supernet_32xb64_in1k_flops-2G_acc-81.72_20221229_200440-954772a3.pth) | [log](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_supernet_32xb64_in1k_20221227_175800-bcf94eaa.json) (`sandwich rule`) |
| ImageNet | AttentiveMobileNetV3 | [AttentiveNAS-A0\*](https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A0.yaml) | 8.854 | 0.212 | 77.19 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.21G_acc-77.19_20221229_200440-282a1f70.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | [AttentiveNAS-A6\*](https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A6.yaml) | 15.594 | 0.927 | 80.81 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.93G_acc-80.81_20221229_200440-73d92cc6.pth) | Converted from the repo |
*Models with * are converted from the [official repo](https://github.com/facebookresearch/AttentiveNAS). The config files of these models
are only for inference. We support training the supernet by `sandwich rule`, which is different from `rejection sampling` in [official repo](https://github.com/facebookresearch/AttentiveNAS), and welcome you to contribute your reproduction results.*

43
configs/nas/mmcls/bignas/metafile.yml 100644
View File

@ -0,0 +1,43 @@
Collections:
- Name: BigNAS
Metadata:
Training Data:
- ImageNet-1k
Paper:
URL: https://arxiv.org/abs/2003.11142
Title: BigNAS Scaling Up Neural Architecture Search with Big Single-Stage Models
README: configs/nas/mmcls/bignas/README.md
Code:
URL: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/mmrazor/models/algorithms/nas/bignas.py
Version: V1.0.0rc2
Models:
- Name: attentive_mobilenet_subnet_8xb256_in1k_flops-927M
In Collection: BigNAS
Metadata:
Flops(G): 0.927
Params(M): 15.594
Supernet: AttentiveMobileNetV3
Subnet: https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A6.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 80.81
Config: configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.93G_acc-80.81_20221229_200440-73d92cc6.pth
- Name: attentive_mobilenet_subnet_8xb256_in1k_flops-212M
In Collection: BigNAS
Metadata:
Flops(G): 0.212
Params(M): 8.854
Supernet: AttentiveMobileNetV3
Channel: https://download.openmmlab.com/mmrazor/v1/bignas/ATTENTIVE_SUBNET_A0.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 77.19
Config: configs/nas/mmcls/bignas/attentive_mobilenet_subnet_8xb256_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/bignas/attentive_mobilenet_subnet_8xb256_in1k_flops-0.21G_acc-77.19_20221229_200440-282a1f70.pth

6
configs/nas/mmcls/dsnas/README.md
View File

@ -34,9 +34,9 @@ CUDA_VISIBLE_DEVICES=0 PORT=29500 ./tools/dist_test.sh \
### Supernet
| Dataset | Params(M) | FLOPs (G) | Top-1 Acc (%) | Top-5 Acc (%) | Config | Download | Remarks |
| :------: | :-------: | :-------: | :-----------: | :-----------: | :---------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :--------------: |
| ImageNet | 3.33 | 0.299 | 73.56 | 91.24 | [config](./dsnas_supernet_8xb128_in1k.py) | [model](https://openmmlab-share.oss-cn-hangzhou.aliyuncs.com/mmrazor/v1/dsnas/dsnas_supernet_8xb128_in1k_20220926_171954-29b87e3a.pth) \| [log](https://download.openmmlab.com/mmrazor/v1/dsnas/dsnas_supernet_8xb128_in1k_20220926_171954-29b87e3a.log) | MMRazor searched |
| Dataset | Params(M) | FLOPs (G) | Top-1 Acc (%) | Top-5 Acc (%) | Config | Download | Remarks |
| :------: | :-------: | :-------: | :-----------: | :-----------: | :---------------------------------------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :--------------: |
| ImageNet | 3.33 | 0.299 | 73.56 | 91.24 | [config](./dsnas_supernet_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/dsnas/dsnas_supernet_8xb128_in1k_20220926_171954-29b87e3a.pth) \| [log](https://download.openmmlab.com/mmrazor/v1/dsnas/dsnas_supernet_8xb128_in1k_20220926_171954-29b87e3a.log) | MMRazor searched |
**Note**:

27
configs/nas/mmcls/dsnas/metafile.yml 100644
View File

@ -0,0 +1,27 @@
Collections:
- Name: DSNAS
Metadata:
Training Data:
- ImageNet-1k
Paper:
URL: https://arxiv.org/abs/2002.09128.pdf
Title: DSNAS Direct Neural Architecture Search without Parameter Retraining
README: configs/nas/mmcls/dsnas/README.md
Code:
URL: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/mmrazor/models/algorithms/nas/dsnas.py
Version: V1.0.0rc2
Models:
- Name: dsnas_subnet_8xb128_in1k
In Collection: DSNAS
Metadata:
Flops(G): 0.299
Params(M): 3.33
Subnet: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/dsnas/DSNAS_SUBNET_IMAGENET_PAPER_ALIAS.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 73.56
Config: configs/nas/mmcls/dsnas/dsnas_subnet_8xb128_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/dsnas/dsnas_supernet_8xb128_in1k_20220926_171954-29b87e3a.pth

6
configs/nas/mmcls/onceforall/README.md
View File

@ -25,9 +25,9 @@ CUDA_VISIBLE_DEVICES=0 PORT=29500 ./tools/dist_test.sh \
| Dataset | Supernet | Subnet | Params(M) | Flops(G) | Top-1 | Config | Download | Remarks |
| :------: | :------------------: | :-------------------------------------------------------------------------------------------------------------------------: | :-------: | :------: | :---: | :-----------------------------------------------------------------------------------------------------------------------------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------: |
| ImageNet | AttentiveMobileNetV3 | [search space](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/_base_/nas_backbones/ofa_mobilenetv3_supernet.py) | 7.6 | 747.8 | 77.5 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/onceforall/ofa_mobilenet_supernet_32xb64_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_supernet_d234_e346_k357_w1_0.py_20221214_0940-d0ebc66f.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | note8_lat@22ms_top1@70.4_finetune@25 | 4.3 | 70.9 | 70.3 | [config](https://download.openmmlab.com/mmrazor/v1/ofa/rtmdet/OFA_SUBNET_NOTE8_LAT22.yaml) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4022ms_top1%4070.4_finetune%4025.py_20221214_0938-fb7fb84f.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | note8_lat@31ms_top1@72.8_finetune@25 | 4.6 | 105.4 | 72.6 | [config](https://download.openmmlab.com/mmrazor/v1/ofa/rtmdet/OFA_SUBNET_NOTE8_LAT31.yaml) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4031ms_top1%4072.8_finetune%4025.py_20221214_0939-981a8b2a.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | [search space](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/_base_/nas_backbones/ofa_mobilenetv3_supernet.py) | 7.6 | 0.747 | 77.5 | [config](https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/onceforall/ofa_mobilenet_supernet_32xb64_in1k.py) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_supernet_d234_e346_k357_w1_0.py_20221214_0940-d0ebc66f.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | note8_lat@22ms_top1@70.4_finetune@25 | 4.3 | 0.070 | 70.3 | [config](https://download.openmmlab.com/mmrazor/v1/ofa/rtmdet/OFA_SUBNET_NOTE8_LAT22.yaml) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4022ms_top1%4070.4_finetune%4025.py_20221214_0938-fb7fb84f.pth) | Converted from the repo |
| ImageNet | AttentiveMobileNetV3 | note8_lat@31ms_top1@72.8_finetune@25 | 4.6 | 0.105 | 72.6 | [config](https://download.openmmlab.com/mmrazor/v1/ofa/rtmdet/OFA_SUBNET_NOTE8_LAT31.yaml) | [model](https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4031ms_top1%4072.8_finetune%4025.py_20221214_0939-981a8b2a.pth) | Converted from the repo |
**Note**:

41
configs/nas/mmcls/onceforall/metafile.yml 100644
View File

@ -0,0 +1,41 @@
Collections:
- Name: OFA
Metadata:
Training Data:
- ImageNet-1k
Paper:
URL: https://arxiv.org/abs/1908.09791
Title: ONCE-FOR-ALL TRAIN ONE NETWORK AND SPE- CIALIZE IT FOR EFFICIENT DEPLOYMENT
README: configs/nas/mmcls/onceforall/README.md
Code:
URL: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/mmrazor/models/algorithms/nas/bignas.py
Version: V1.0.0rc2
Models:
- Name: ofa_mobilenet_subnet_8xb256_in1k_lat-22ms
In Collection: OFA
Metadata:
Flops(G): 0.070
Params(M): 4.3
Subnet: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/onceforall/OFA_SUBNET_NOTE8_LAT22.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 70.3
Config: configs/nas/mmcls/onceforall/ofa_mobilenet_subnet_8xb256_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4022ms_top1%4070.4_finetune%4025.py_20221214_0938-fb7fb84f.pth
- Name: ofa_mobilenet_subnet_8xb256_in1k_lat-31ms
In Collection: OFA
Metadata:
Flops(G): 0.105
Params(M): 4.6
Subnet: https://github.com/open-mmlab/mmrazor/blob/dev-1.x/configs/nas/mmcls/onceforall/OFA_SUBNET_NOTE8_LAT31.yaml
Results:
- Task: Image Classification
Dataset: ImageNet-1k
Metrics:
Top 1 Accuracy: 72.6
Config: configs/nas/mmcls/onceforall/ofa_mobilenet_subnet_8xb256_in1k.py
Weights: https://download.openmmlab.com/mmrazor/v1/ofa/ofa_mobilenet_subnet_8xb256_in1k_note8_lat%4031ms_top1%4072.8_finetune%4025.py_20221214_0939-981a8b2a.pth