Skip to content

Commit 23e7f17

Browse files
Laurent2916rwightman
authored andcommitted
replace inline latex syntax in hfdocs
1 parent 21647c0 commit 23e7f17

14 files changed

+72
-72
lines changed

hfdocs/source/models/efficientnet-pruned.mdx

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# EfficientNet (Knapsack Pruned)
22

3-
**EfficientNet** is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a *compound coefficient*. Unlike conventional practice that arbitrary scales these factors, the EfficientNet scaling method uniformly scales network width, depth, and resolution with a set of fixed scaling coefficients. For example, if we want to use $2^N$ times more computational resources, then we can simply increase the network depth by $\alpha ^ N$, width by $\beta ^ N$, and image size by $\gamma ^ N$, where $\alpha, \beta, \gamma$ are constant coefficients determined by a small grid search on the original small model. EfficientNet uses a compound coefficient $\phi$ to uniformly scales network width, depth, and resolution in a principled way.
3+
**EfficientNet** is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a *compound coefficient*. Unlike conventional practice that arbitrary scales these factors, the EfficientNet scaling method uniformly scales network width, depth, and resolution with a set of fixed scaling coefficients. For example, if we want to use \\( 2^N \\) times more computational resources, then we can simply increase the network depth by \\( \alpha ^ N \\), width by \\( \beta ^ N \\), and image size by \\( \gamma ^ N \\), where \\( \alpha, \beta, \gamma \\) are constant coefficients determined by a small grid search on the original small model. EfficientNet uses a compound coefficient \\( \phi \\) to uniformly scales network width, depth, and resolution in a principled way.
44

55
The compound scaling method is justified by the intuition that if the input image is bigger, then the network needs more layers to increase the receptive field and more channels to capture more fine-grained patterns on the bigger image.
66

@@ -20,7 +20,7 @@ To load a pretrained model:
2020

2121
To load and preprocess the image:
2222

23-
```py
23+
```py
2424
>>> import urllib
2525
>>> from PIL import Image
2626
>>> from timm.data import resolve_data_config
@@ -51,7 +51,7 @@ To get the top-5 predictions class names:
5151
```py
5252
>>> # Get imagenet class mappings
5353
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
54-
>>> urllib.request.urlretrieve(url, filename)
54+
>>> urllib.request.urlretrieve(url, filename)
5555
>>> with open("imagenet_classes.txt", "r") as f:
5656
... categories = [s.strip() for s in f.readlines()]
5757

@@ -85,7 +85,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
8585

8686
```BibTeX
8787
@misc{tan2020efficientnet,
88-
title={EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks},
88+
title={EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks},
8989
author={Mingxing Tan and Quoc V. Le},
9090
year={2020},
9191
eprint={1905.11946},
@@ -209,4 +209,4 @@ Models:
209209
Metrics:
210210
Top 1 Accuracy: 80.86%
211211
Top 5 Accuracy: 95.24%
212-
-->
212+
-->

hfdocs/source/models/efficientnet.mdx

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# EfficientNet
22

3-
**EfficientNet** is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a *compound coefficient*. Unlike conventional practice that arbitrary scales these factors, the EfficientNet scaling method uniformly scales network width, depth, and resolution with a set of fixed scaling coefficients. For example, if we want to use $2^N$ times more computational resources, then we can simply increase the network depth by $\alpha ^ N$, width by $\beta ^ N$, and image size by $\gamma ^ N$, where $\alpha, \beta, \gamma$ are constant coefficients determined by a small grid search on the original small model. EfficientNet uses a compound coefficient $\phi$ to uniformly scales network width, depth, and resolution in a principled way.
3+
**EfficientNet** is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a *compound coefficient*. Unlike conventional practice that arbitrary scales these factors, the EfficientNet scaling method uniformly scales network width, depth, and resolution with a set of fixed scaling coefficients. For example, if we want to use \\( 2^N \\) times more computational resources, then we can simply increase the network depth by \\( \alpha ^ N \\), width by \\( \beta ^ N \\), and image size by \\( \gamma ^ N \\), where \\( \alpha, \beta, \gamma \\) are constant coefficients determined by a small grid search on the original small model. EfficientNet uses a compound coefficient \\( \phi \\) to uniformly scales network width, depth, and resolution in a principled way.
44

55
The compound scaling method is justified by the intuition that if the input image is bigger, then the network needs more layers to increase the receptive field and more channels to capture more fine-grained patterns on the bigger image.
66

@@ -18,7 +18,7 @@ To load a pretrained model:
1818

1919
To load and preprocess the image:
2020

21-
```py
21+
```py
2222
>>> import urllib
2323
>>> from PIL import Image
2424
>>> from timm.data import resolve_data_config
@@ -49,7 +49,7 @@ To get the top-5 predictions class names:
4949
```py
5050
>>> # Get imagenet class mappings
5151
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
52-
>>> urllib.request.urlretrieve(url, filename)
52+
>>> urllib.request.urlretrieve(url, filename)
5353
>>> with open("imagenet_classes.txt", "r") as f:
5454
... categories = [s.strip() for s in f.readlines()]
5555

@@ -83,7 +83,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
8383

8484
```BibTeX
8585
@misc{tan2020efficientnet,
86-
title={EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks},
86+
title={EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks},
8787
author={Mingxing Tan and Quoc V. Le},
8888
year={2020},
8989
eprint={1905.11946},
@@ -389,4 +389,4 @@ Models:
389389
Metrics:
390390
Top 1 Accuracy: 75.5%
391391
Top 5 Accuracy: 92.51%
392-
-->
392+
-->

hfdocs/source/models/gloun-resnext.mdx

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# (Gluon) ResNeXt
22

3-
A **ResNeXt** repeats a [building block](https://paperswithcode.com/method/resnext-block) that aggregates a set of transformations with the same topology. Compared to a [ResNet](https://paperswithcode.com/method/resnet), it exposes a new dimension, *cardinality* (the size of the set of transformations) $C$, as an essential factor in addition to the dimensions of depth and width.
3+
A **ResNeXt** repeats a [building block](https://paperswithcode.com/method/resnext-block) that aggregates a set of transformations with the same topology. Compared to a [ResNet](https://paperswithcode.com/method/resnet), it exposes a new dimension, *cardinality* (the size of the set of transformations) \\( C \\), as an essential factor in addition to the dimensions of depth and width.
44

55
The weights from this model were ported from [Gluon](https://cv.gluon.ai/model_zoo/classification.html).
66

@@ -16,7 +16,7 @@ To load a pretrained model:
1616

1717
To load and preprocess the image:
1818

19-
```py
19+
```py
2020
>>> import urllib
2121
>>> from PIL import Image
2222
>>> from timm.data import resolve_data_config
@@ -47,7 +47,7 @@ To get the top-5 predictions class names:
4747
```py
4848
>>> # Get imagenet class mappings
4949
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
50-
>>> urllib.request.urlretrieve(url, filename)
50+
>>> urllib.request.urlretrieve(url, filename)
5151
>>> with open("imagenet_classes.txt", "r") as f:
5252
... categories = [s.strip() for s in f.readlines()]
5353

@@ -206,4 +206,4 @@ Models:
206206
Metrics:
207207
Top 1 Accuracy: 79.35%
208208
Top 5 Accuracy: 94.42%
209-
-->
209+
-->

hfdocs/source/models/hrnet.mdx

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# HRNet
22

3-
**HRNet**, or **High-Resolution Net**, is a general purpose convolutional neural network for tasks like semantic segmentation, object detection and image classification. It is able to maintain high resolution representations through the whole process. We start from a high-resolution convolution stream, gradually add high-to-low resolution convolution streams one by one, and connect the multi-resolution streams in parallel. The resulting network consists of several ($4$ in the paper) stages and the $n$th stage contains $n$ streams corresponding to $n$ resolutions. The authors conduct repeated multi-resolution fusions by exchanging the information across the parallel streams over and over.
3+
**HRNet**, or **High-Resolution Net**, is a general purpose convolutional neural network for tasks like semantic segmentation, object detection and image classification. It is able to maintain high resolution representations through the whole process. We start from a high-resolution convolution stream, gradually add high-to-low resolution convolution streams one by one, and connect the multi-resolution streams in parallel. The resulting network consists of several (\\( 4 \\) in the paper) stages and the \\( n \\)th stage contains \\( n \\) streams corresponding to \\( n \\) resolutions. The authors conduct repeated multi-resolution fusions by exchanging the information across the parallel streams over and over.
44

55
## How do I use this model on an image?
66

@@ -14,7 +14,7 @@ To load a pretrained model:
1414

1515
To load and preprocess the image:
1616

17-
```py
17+
```py
1818
>>> import urllib
1919
>>> from PIL import Image
2020
>>> from timm.data import resolve_data_config
@@ -45,7 +45,7 @@ To get the top-5 predictions class names:
4545
```py
4646
>>> # Get imagenet class mappings
4747
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
48-
>>> urllib.request.urlretrieve(url, filename)
48+
>>> urllib.request.urlretrieve(url, filename)
4949
>>> with open("imagenet_classes.txt", "r") as f:
5050
... categories = [s.strip() for s in f.readlines()]
5151

@@ -79,7 +79,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
7979

8080
```BibTeX
8181
@misc{sun2019highresolution,
82-
title={High-Resolution Representations for Labeling Pixels and Regions},
82+
title={High-Resolution Representations for Labeling Pixels and Regions},
8383
author={Ke Sun and Yang Zhao and Borui Jiang and Tianheng Cheng and Bin Xiao and Dong Liu and Yadong Mu and Xinggang Wang and Wenyu Liu and Jingdong Wang},
8484
year={2019},
8585
eprint={1904.04514},
@@ -422,4 +422,4 @@ Models:
422422
Metrics:
423423
Top 1 Accuracy: 79.46%
424424
Top 5 Accuracy: 94.65%
425-
-->
425+
-->

hfdocs/source/models/ig-resnext.mdx

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,8 @@
11
# Instagram ResNeXt WSL
22

3-
A **ResNeXt** repeats a [building block](https://paperswithcode.com/method/resnext-block) that aggregates a set of transformations with the same topology. Compared to a [ResNet](https://paperswithcode.com/method/resnet), it exposes a new dimension, *cardinality* (the size of the set of transformations) $C$, as an essential factor in addition to the dimensions of depth and width.
3+
A **ResNeXt** repeats a [building block](https://paperswithcode.com/method/resnext-block) that aggregates a set of transformations with the same topology. Compared to a [ResNet](https://paperswithcode.com/method/resnet), it exposes a new dimension, *cardinality* (the size of the set of transformations) \\( C \\), as an essential factor in addition to the dimensions of depth and width.
44

5-
This model was trained on billions of Instagram images using thousands of distinct hashtags as labels exhibit excellent transfer learning performance.
5+
This model was trained on billions of Instagram images using thousands of distinct hashtags as labels exhibit excellent transfer learning performance.
66

77
Please note the CC-BY-NC 4.0 license on theses weights, non-commercial use only.
88

@@ -18,7 +18,7 @@ To load a pretrained model:
1818

1919
To load and preprocess the image:
2020

21-
```py
21+
```py
2222
>>> import urllib
2323
>>> from PIL import Image
2424
>>> from timm.data import resolve_data_config
@@ -49,7 +49,7 @@ To get the top-5 predictions class names:
4949
```py
5050
>>> # Get imagenet class mappings
5151
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
52-
>>> urllib.request.urlretrieve(url, filename)
52+
>>> urllib.request.urlretrieve(url, filename)
5353
>>> with open("imagenet_classes.txt", "r") as f:
5454
... categories = [s.strip() for s in f.readlines()]
5555

@@ -83,7 +83,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
8383

8484
```BibTeX
8585
@misc{mahajan2018exploring,
86-
title={Exploring the Limits of Weakly Supervised Pretraining},
86+
title={Exploring the Limits of Weakly Supervised Pretraining},
8787
author={Dhruv Mahajan and Ross Girshick and Vignesh Ramanathan and Kaiming He and Manohar Paluri and Yixuan Li and Ashwin Bharambe and Laurens van der Maaten},
8888
year={2018},
8989
eprint={1805.00932},
@@ -273,4 +273,4 @@ Models:
273273
Metrics:
274274
Top 1 Accuracy: 82.7%
275275
Top 5 Accuracy: 96.64%
276-
-->
276+
-->

hfdocs/source/models/nasnet.mdx

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -14,7 +14,7 @@ To load a pretrained model:
1414

1515
To load and preprocess the image:
1616

17-
```py
17+
```py
1818
>>> import urllib
1919
>>> from PIL import Image
2020
>>> from timm.data import resolve_data_config
@@ -45,7 +45,7 @@ To get the top-5 predictions class names:
4545
```py
4646
>>> # Get imagenet class mappings
4747
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
48-
>>> urllib.request.urlretrieve(url, filename)
48+
>>> urllib.request.urlretrieve(url, filename)
4949
>>> with open("imagenet_classes.txt", "r") as f:
5050
... categories = [s.strip() for s in f.readlines()]
5151

@@ -79,7 +79,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
7979

8080
```BibTeX
8181
@misc{zoph2018learning,
82-
title={Learning Transferable Architectures for Scalable Image Recognition},
82+
title={Learning Transferable Architectures for Scalable Image Recognition},
8383
author={Barret Zoph and Vijay Vasudevan and Jonathon Shlens and Quoc V. Le},
8484
year={2018},
8585
eprint={1707.07012},
@@ -125,7 +125,7 @@ Models:
125125
Image Size: '331'
126126
Interpolation: bicubic
127127
Label Smoothing: 0.1
128-
RMSProp $\epsilon$: 1.0
128+
RMSProp \\( \epsilon \\): 1.0
129129
Code: https://github.com/rwightman/pytorch-image-models/blob/d8e69206be253892b2956341fea09fdebfaae4e3/timm/models/nasnet.py#L562
130130
Weights: http://data.lip6.fr/cadene/pretrainedmodels/nasnetalarge-a1897284.pth
131131
Results:
@@ -134,4 +134,4 @@ Models:
134134
Metrics:
135135
Top 1 Accuracy: 82.63%
136136
Top 5 Accuracy: 96.05%
137-
-->
137+
-->

hfdocs/source/models/regnetx.mdx

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -1,10 +1,10 @@
11
# RegNetX
22

3-
**RegNetX** is a convolutional network design space with simple, regular models with parameters: depth $d$, initial width $w\_{0} > 0$, and slope $w\_{a} > 0$, and generates a different block width $u\_{j}$ for each block $j < d$. The key restriction for the RegNet types of model is that there is a linear parameterisation of block widths (the design space only contains models with this linear structure):
3+
**RegNetX** is a convolutional network design space with simple, regular models with parameters: depth \\( d \\), initial width \\( w\_{0} > 0 \\), and slope \\( w\_{a} > 0 \\), and generates a different block width \\( u\_{j} \\) for each block \\( j < d \\). The key restriction for the RegNet types of model is that there is a linear parameterisation of block widths (the design space only contains models with this linear structure):
44

5-
$$ u\_{j} = w\_{0} + w\_{a}\cdot{j} $$
5+
\\( \\) u\_{j} = w\_{0} + w\_{a}\cdot{j} \\( \\)
66

7-
For **RegNetX** we have additional restrictions: we set $b = 1$ (the bottleneck ratio), $12 \leq d \leq 28$, and $w\_{m} \geq 2$ (the width multiplier).
7+
For **RegNetX** we have additional restrictions: we set \\( b = 1 \\) (the bottleneck ratio), \\( 12 \leq d \leq 28 \\), and \\( w\_{m} \geq 2 \\) (the width multiplier).
88

99
## How do I use this model on an image?
1010

@@ -18,7 +18,7 @@ To load a pretrained model:
1818

1919
To load and preprocess the image:
2020

21-
```py
21+
```py
2222
>>> import urllib
2323
>>> from PIL import Image
2424
>>> from timm.data import resolve_data_config
@@ -49,7 +49,7 @@ To get the top-5 predictions class names:
4949
```py
5050
>>> # Get imagenet class mappings
5151
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
52-
>>> urllib.request.urlretrieve(url, filename)
52+
>>> urllib.request.urlretrieve(url, filename)
5353
>>> with open("imagenet_classes.txt", "r") as f:
5454
... categories = [s.strip() for s in f.readlines()]
5555

@@ -83,7 +83,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
8383

8484
```BibTeX
8585
@misc{radosavovic2020designing,
86-
title={Designing Network Design Spaces},
86+
title={Designing Network Design Spaces},
8787
author={Ilija Radosavovic and Raj Prateek Kosaraju and Ross Girshick and Kaiming He and Piotr Dollár},
8888
year={2020},
8989
eprint={2003.13678},
@@ -556,4 +556,4 @@ Models:
556556
Metrics:
557557
Top 1 Accuracy: 80.25%
558558
Top 5 Accuracy: 95.03%
559-
-->
559+
-->

hfdocs/source/models/regnety.mdx

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -1,10 +1,10 @@
11
# RegNetY
22

3-
**RegNetY** is a convolutional network design space with simple, regular models with parameters: depth $d$, initial width $w\_{0} > 0$, and slope $w\_{a} > 0$, and generates a different block width $u\_{j}$ for each block $j < d$. The key restriction for the RegNet types of model is that there is a linear parameterisation of block widths (the design space only contains models with this linear structure):
3+
**RegNetY** is a convolutional network design space with simple, regular models with parameters: depth \\( d \\), initial width \\( w\_{0} > 0 \\), and slope \\( w\_{a} > 0 \\), and generates a different block width \\( u\_{j} \\) for each block \\( j < d \\). The key restriction for the RegNet types of model is that there is a linear parameterisation of block widths (the design space only contains models with this linear structure):
44

5-
$$ u\_{j} = w\_{0} + w\_{a}\cdot{j} $$
5+
\\( \\) u\_{j} = w\_{0} + w\_{a}\cdot{j} \\( \\)
66

7-
For **RegNetX** authors have additional restrictions: we set $b = 1$ (the bottleneck ratio), $12 \leq d \leq 28$, and $w\_{m} \geq 2$ (the width multiplier).
7+
For **RegNetX** authors have additional restrictions: we set \\( b = 1 \\) (the bottleneck ratio), \\( 12 \leq d \leq 28 \\), and \\( w\_{m} \geq 2 \\) (the width multiplier).
88

99
For **RegNetY** authors make one change, which is to include [Squeeze-and-Excitation blocks](https://paperswithcode.com/method/squeeze-and-excitation-block).
1010

@@ -20,7 +20,7 @@ To load a pretrained model:
2020

2121
To load and preprocess the image:
2222

23-
```py
23+
```py
2424
>>> import urllib
2525
>>> from PIL import Image
2626
>>> from timm.data import resolve_data_config
@@ -51,7 +51,7 @@ To get the top-5 predictions class names:
5151
```py
5252
>>> # Get imagenet class mappings
5353
>>> url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
54-
>>> urllib.request.urlretrieve(url, filename)
54+
>>> urllib.request.urlretrieve(url, filename)
5555
>>> with open("imagenet_classes.txt", "r") as f:
5656
... categories = [s.strip() for s in f.readlines()]
5757

@@ -85,7 +85,7 @@ You can follow the [timm recipe scripts](../scripts) for training a new model af
8585

8686
```BibTeX
8787
@misc{radosavovic2020designing,
88-
title={Designing Network Design Spaces},
88+
title={Designing Network Design Spaces},
8989
author={Ilija Radosavovic and Raj Prateek Kosaraju and Ross Girshick and Kaiming He and Piotr Dollár},
9090
year={2020},
9191
eprint={2003.13678},
@@ -570,4 +570,4 @@ Models:
570570
Metrics:
571571
Top 1 Accuracy: 80.8%
572572
Top 5 Accuracy: 95.25%
573-
-->
573+
-->

0 commit comments

Comments
 (0)