:sparkles: Add DenseNet models in CIFAR-100

README.md

@@ -11,14 +11,15 @@ tags:
 
 [![arXiv](https://img.shields.io/badge/arXiv-2403.14715-b31b1b.svg)](https://arxiv.org/abs/2403.14715)
 
-This repository contains the models trained as experimental support for the paper "Towards Understanding Why Label Smoothing Degrades Selective Classification and How to Fix It" published at ICLR.
+This repository contains the models trained as experimental support for the paper "Towards Understanding Why Label Smoothing Degrades Selective Classification and How to Fix It" published at ICLR 2025.
 
 The code is based on [TorchUncertainty](https://github.com/ENSTA-U2IS-AI/torch-uncertainty) and available on [GitHub](https://github.com/o-laurent/Label-smoothing-Selective-classification).
 
 ## List of models
 
 This repository contains:
+- for classification on CIFAR-100: one CE-based and one LS-based (LS coefficient 0.2) DenseNet 
 - for segmentation: one CE-based and one LS-based (LS coefficient 0.2) DeepLabv3+ Resnet-101 
-- for nlp: one CE-based and LS-based one (LS coefficient 0.6) LSTM-MLP
+- for nlp: one CE-based and one LS-based (LS coefficient 0.6) LSTM-MLP
 
 The rest of the models (notably on tabular data) used in the paper are trainable on CPU in the dedicated notebooks.

classification/cifar100/densenet_c100_ce.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bb79d7b9b382f6afc7deb7bb402b0aa120a613abb5e0e31baad0b2b52bf67cb
+size 1715766

classification/cifar100/densenet_c100_ls02.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ace9bb34cf82174c0098c0ca71947001aab940c0d8e00338865c0921d0598851
+size 1715766