--- library_name: pytorch pipeline_tag: image-classification tags: - continual-learning - class-incremental-learning - neural-collapse - cifar-10 - cifar-100 - tiny-imagenet --- # Reliability-Calibrated Routing (RCR) Inference checkpoints for [When Neural Collapse Fails to Route: Routing Drift in Continual Learning](https://arxiv.org/abs/2611.22429), BMVC 2026. [Code](https://github.com/towzeur/RCR) | [Project page](https://towzeur.github.io/RCR/) | [Evaluation artifacts](https://github.com/towzeur/RCR/releases/tag/v1.0.0) RCR is a classifier for class-incremental image recognition. It keeps the trained [ProNC](https://github.com/Continue-Edge-AI-Lab/ProNC) ResNet-18 and ETF classifier fixed, and combines ETF alignment with a radial score based on transported class prototypes. It predicts among all classes seen during training without a task label at inference. ## Checkpoints This repository mirrors the 18 final inference checkpoints from GitHub release `v1.0.0`. There are three seeds (0, 1, 2) for each dataset and replay budget (200, 500). Each checkpoint contains the backbone weights, ETF vertices, class prototypes, and class radii. It does not contain replay images or optimizer state. | Dataset | Checkpoint path pattern | |---|---| | CIFAR-10 | `checkpoints/seq_cifar10_buffer{200,500}_seed{0,1,2}/final.pt` | | CIFAR-100 | `checkpoints/seq_cifar100_buffer{200,500}_seed{0,1,2}/final.pt` | | TinyImageNet | `checkpoints/seq_tinyimagenet_buffer{200,500}_seed{0,1,2}/final.pt` | ## Inference From a checkout of the [RCR code](https://github.com/towzeur/RCR/tree/v1.0.0), install the dependencies with `python -m pip install -r requirements-eval.txt huggingface_hub`. Download a checkpoint and load it with the RCR inference loader: ```python import torch from huggingface_hub import hf_hub_download from rcr.rcr import load_inference_checkpoint, rcr_scores path = hf_hub_download( repo_id="towzeur/RCR", filename="checkpoints/seq_cifar100_buffer200_seed0/final.pt", ) net, state = load_inference_checkpoint(path) # images: a batch of RGB images with the dataset's test normalization. with torch.inference_mode(): features = net(images, returnt="features") scores = rcr_scores( features, state["etf"].T, state["ldc_proto"], state["ldcrad_class_radius"], state["classes_per_task"], ) labels = scores.argmax(dim=1) ``` Use the repository's [dataset preprocessing and label order](https://github.com/towzeur/RCR/tree/v1.0.0/rcr/datasets). The checkpoints are trained for the classes in their respective benchmarks, not for open-set recognition or resuming training. ## Evaluation CIFAR-100 with a replay budget of 200, averaged over three seeds: RCR raises Class-IL final average accuracy from 41.71% to 49.74% and lowers final forgetting from 38.55% to 16.27%, using the same trained backbones. See the [results table](https://github.com/towzeur/RCR#results) for all six settings. Recomputing forgetting and task-by-task metrics requires the per-task features in [GitHub Releases](https://github.com/towzeur/RCR/releases/tag/v1.0.0), not just the final checkpoints here.