Use arxiv friendly template

jayanthkoushik · Jul 26, 2023 · f7fc91f · f7fc91f
1 parent 90cac9d
commit f7fc91f
Show file tree

Hide file tree

Showing 40 changed files with 433 additions and 29 deletions.
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -48,4 +48,4 @@ repos:
     language: system
     files: "paper/.*"
     pass_filenames: false
-    entry: sh -c "(cd paper && pipx run --spec git+https://github.com/jayanthkoushik/shiny-mdc shinymdc -i main.md -o main.pdf -t stylish -m smalltabs=true,nonidan=true)"
+    entry: sh -c "(cd paper && pipx run --spec git+https://github.com/jayanthkoushik/shiny-mdc shinymdc -i main.md -o main.pdf -t template.tex --pdf-engine pdflatex)"
diff --git a/paper/README.md b/paper/README.md
@@ -1,15 +1,15 @@
-# Building paper
-
 <!-- cSpell:ignore pandoc, crossref, shiny-mdc -->
 
+# Building paper
+
 Building the paper requires a full TeX installation,
 [pandoc](https://pandoc.org/),
 [pandoc-crossref](https://lierdakil.github.io/pandoc-crossref/), and
-[shiny-mdc](https://pypi.org/project/shiny-mdc/). Build using the
-following command, from the `paper` directory:
+[shiny-mdc](https://pypi.org/project/shiny-mdc/) (v1.9 or higher).
+Build using the following command, from the `paper` directory:
 
 <!-- cSpell: disable -->
 ```bash
-shinymdc -i main.md -o main.pdf -t stylish -m smalltabs=true,nonidan=true
+shinymdc -i main.md -o main.pdf -t template.tex --pdf-engine pdflatex
 ```
 <!-- cSpell: enable -->
diff --git a/paper/figures/_arxiv/all_pred_changes_nn_all_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/all_pred_changes_nn_all_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/all_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/all_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_025.pdf b/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_025.pdf
diff --git a/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_1.pdf b/paper/figures/_arxiv/appendix/euclidean_cosine_split_accs_vs_k_imagenetlt_crt_rho_1.pdf
diff --git a/paper/figures/_arxiv/appendix/models_split_top1_deltas_vs_rho.pdf b/paper/figures/_arxiv/appendix/models_split_top1_deltas_vs_rho.pdf
diff --git a/paper/figures/_arxiv/appendix/models_split_top5_deltas_vs_rho.pdf b/paper/figures/_arxiv/appendix/models_split_top5_deltas_vs_rho.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_cifarlt_ltr.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_cifarlt_ltr.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_cifarlt_ride.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_cifarlt_ride.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_crt.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_crt.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_lws.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_lws.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_ride.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_imagenetlt_ride.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_placeslt_crt.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_placeslt_crt.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_placeslt_lws.pdf b/paper/figures/_arxiv/appendix/rhos_cls_delta_vs_nndist_placeslt_lws.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_cifarlt_ltr.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_cifarlt_ltr.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_cifarlt_ride.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_cifarlt_ride.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_crt.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_crt.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_lws.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_lws.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_ride.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_imagenetlt_ride.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_placeslt_crt.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_placeslt_crt.pdf
diff --git a/paper/figures/_arxiv/appendix/rhos_cls_deltas_placeslt_lws.pdf b/paper/figures/_arxiv/appendix/rhos_cls_deltas_placeslt_lws.pdf
diff --git a/paper/figures/_arxiv/base_pred_changes_nn_few_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/base_pred_changes_nn_few_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/base_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/base_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/cls_acc_vs_nndist_imagenetlt_crt_baseline.pdf b/paper/figures/_arxiv/cls_acc_vs_nndist_imagenetlt_crt_baseline.pdf
diff --git a/paper/figures/_arxiv/cls_delta_vs_nndist_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/cls_delta_vs_nndist_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/doggies.pdf b/paper/figures/_arxiv/doggies.pdf
diff --git a/paper/figures/_arxiv/euclidean_random_split_deltas_vs_rho_imagenetlt_crt.pdf b/paper/figures/_arxiv/euclidean_random_split_deltas_vs_rho_imagenetlt_crt.pdf
diff --git a/paper/figures/_arxiv/few_pred_changes_nn_base_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/few_pred_changes_nn_base_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/few_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf b/paper/figures/_arxiv/few_pred_changes_nn_semantic4_imagenetlt_crt_rho_05.pdf
diff --git a/paper/figures/_arxiv/pipeline.pdf b/paper/figures/_arxiv/pipeline.pdf
diff --git a/paper/figures/_arxiv/pred_counts_imagenetlt_crt_baseline.pdf b/paper/figures/_arxiv/pred_counts_imagenetlt_crt_baseline.pdf
diff --git a/paper/main.md b/paper/main.md
@@ -54,11 +54,11 @@ abstract: Methods in long-tail learning focus on improving performance
   find frequent classes that are closest to rare classes in the model's
   representation space and learn weights to update rare class
   classifiers with a linear combination of frequent class classifiers.
-  AlphaNet, applied on several models, greatly improves test accuracy
+  AlphaNet, applied to several models, greatly improves test accuracy
   for rare classes in multiple long-tailed datasets, with very little
-  change to the overall accuracy. Our method also provides a way to
-  control the trade-off between rare class and overall accuracy, making
-  it practical for long-tail classification in the wild.
+  change to overall accuracy. Our method also provides a way to control
+  the trade-off between rare class and overall accuracy, making it
+  practical for long-tail classification in the wild.
 
 bibliography: references.bib
 

diff --git a/paper/main.pdf b/paper/main.pdf
diff --git a/paper/sections/1_intro.md b/paper/sections/1_intro.md
@@ -48,10 +48,10 @@ misclassifications.](figures/doggies){#fig:analysis:egs}
 
 ![Per-class test accuracy of cRT model on 'few' split of ImageNet‑LT,
 versus the mean Euclidean distance to 5 nearest neighbor (NN) 'base'
-split classes. The line is a bootstrapped linear regression fit, and $r$
-(top right) is Pearson correlation. There is a high correlation, i.e.,
-'few' split classes with close 'base' split NNs are more likely to be
-misclassified.](figures/cls_acc_vs_nndist_imagenetlt_crt_baseline){#fig:analysis:acc_vs_dist}
+split classes. The line is a bootstrapped linear regression fit, and
+'$r$' (top right) is Pearson correlation. There is a high correlation,
+i.e., 'few' split classes with close 'base' split NNs are more likely to
+be misclassified.](figures/cls_acc_vs_nndist_imagenetlt_crt_baseline){#fig:analysis:acc_vs_dist}
 
 Analysis of 'few' split predictions on ImageNet‑LT.
 
@@ -95,7 +95,8 @@ rare class accuracy on multiple datasets.
 
 [^note:ride_results]: Results for the 6-expert model are presented in
     the GitHub repository for the original paper at
-    [`github.com/frank-xwang/RIDE-LongTailRecognition/blob/main/MODEL_ZOO.md`](https://github.com/frank-xwang/RIDE-LongTailRecognition/blob/main/MODEL_ZOO.md).
+    [`github.com/frank-xwang/
+    RIDE-LongTailRecognition/blob/main/MODEL_ZOO.md`](https://github.com/frank-xwang/RIDE-LongTailRecognition/blob/main/MODEL_ZOO.md).
 
 [^note:base_split]: The 'base' split is the complement of the 'few'
     split, composed of classes with many training samples.
diff --git a/paper/sections/appendix/perclsdels.md b/paper/sections/appendix/perclsdels.md
@@ -121,7 +121,7 @@ baseline](figures/appendix/rhos_cls_delta_vs_nndist_imagenetlt_ride){#fig:rhos_c
 Change in per-class test accuracy on ImageNet‑LT, versus mean distance
 to 5 nearest neighbors based on Euclidean distance. The neighbors are
 from 'base' split for the 'few' split classes, and vice-versa for the
-'base' split classes. The lines are regression fits, and the '$r$'s are
+'base' split classes. The lines are regression fits, and the `$r$'s are
 Pearson correlations.
 
 </div>
@@ -139,7 +139,7 @@ baseline](figures/appendix/rhos_cls_delta_vs_nndist_placeslt_lws){#fig:rhos_cls_
 Change in per-class test accuracy on Places‑LT, versus mean distance to
 5 nearest neighbors based on Euclidean distance. The neighbors are from
 'base' split for the 'few' split classes, and vice-versa for the 'base'
-split classes. The lines are regression fits, and the '$r$'s are Pearson
+split classes. The lines are regression fits, and the `$r$'s are Pearson
 correlations.
 
 </div>
@@ -157,7 +157,7 @@ baseline](figures/appendix/rhos_cls_delta_vs_nndist_cifarlt_ltr){#fig:rhos_cls_d
 Change in per-class test accuracy on CIFAR‑100‑LT, versus mean distance
 to 5 nearest neighbors based on Euclidean distance. The neighbors are
 from 'base' split for the 'few' split classes, and vice-versa for the
-'base' split classes. The lines are regression fits, and the '$r$'s are
+'base' split classes. The lines are regression fits, and the `$r$'s are
 Pearson correlations.
 
 </div>