Tsui-Wei Weng, Tuomas Oikarinen, Tara Javidi, Amirhosein Javadi
12 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 9 failures of our instrument; 3 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.
"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2604.14519.json · badge: /api/ran/2604.14519.svg
Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.
| Repository | Role | Ran |
|---|---|---|
| importAmir/CI-CBM | canonical (classifier label) classifier basis: census |
12 ran · 3 no recorded run |
| Function | Status | Where it lives |
|---|---|---|
| add_index_to_dataloader | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/glm_saga/elasticnet.py pointer only (licence: NONE) · get_code("7798dffe63b4a970") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| average_accuracy | Ran ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT) |
importAmir/CI-CBM/metrics.py pointer only (licence: NONE) · get_code("a10f77d0afc6adf5") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| average_forgetting | Ran ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT) |
importAmir/CI-CBM/metrics.py pointer only (licence: NONE) · get_code("4bfaf0ad67624b61") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| average_incremental_accuracy | Ran ran, honoured the contract we drafted |
importAmir/CI-CBM/metrics.py pointer only (licence: NONE) · get_code("bfec8dddd335e243") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| batch_process | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/training_utils.py pointer only (licence: NONE) · get_code("6ceab35a4a98d284") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| cos_similarity | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/similarity.py pointer only (licence: NONE); the stored copy get_code serves, from teresa-sc/concepts_in_vits, is pointer only (licence: NONE) · get_code("cc642412d4347cce") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| cos_similarity_cubed | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/similarity.py pointer only (licence: NONE) · get_code("892d44d97047a10d") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| cos_similarity_cubed_single | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/similarity.py pointer only (licence: NONE) · get_code("cf659682a960fdc0") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| get_device | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/glm_saga/elasticnet.py pointer only (licence: NONE); the stored copy get_code serves, from Trustworthy-ML-Lab/Linear-Explanations, is pointer only (licence: NONE) · get_code("406014b884a2aaa7") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| soft_threshold | Ran ran, honoured the contract we drafted |
importAmir/CI-CBM/glm_saga/elasticnet.py pointer only (licence: NONE); the stored copy get_code serves, from Trustworthy-ML-Lab/Linear-Explanations, is pointer only (licence: NONE) · get_code("bfe4b187a1c5aeaf") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| train_projection_layer | Ran ran on a synthesized input |
importAmir/CI-CBM/training_utils.py pointer only (licence: NONE) · get_code("55e220f7123ab815") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| train_projection_layer_lwf | Ran ran; our contract draft was wrong, not the code |
importAmir/CI-CBM/training_utils.py pointer only (licence: NONE) · get_code("90bbe6d1346d52e2") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded. |
| get_data | Not yet run unverified (harvested, no recorded run) |
importAmir/CI-CBM/data_utils.py pointer only (licence: NONE) · get_code("59a87e29dd3b8451") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded. |
| get_targets_only | Not yet run unverified (harvested, no recorded run) |
importAmir/CI-CBM/data_utils.py pointer only (licence: NONE) · get_code("80edf2b9738fd94a") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded. |
| split_data_SelfPromptDeit | Not yet run unverified (harvested, no recorded run) |
importAmir/CI-CBM/data_utils.py pointer only (licence: NONE) · get_code("ec9b03abe621c9be") How it ranThe body of this sample is withheld: GitHub detected no licence in this paper's repository, importAmir/CI-CBM when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded. |
Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 0 later) and 9 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).
Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.
| Function | Also present in | Earliest-dated | Others named |
|---|---|---|---|
| get_device | 5 other papers 5 dated earlier · 2 only as an assembly brick |
2105.04857 (2021-05) Leveraging Sparse Linear Layers for Debuggable Deep Networks |
2308.03685 (2023-08) Learning Concise and Descriptive Attributes for Visual Recognition 2405.06855 (2024-05) Linear Explanations for Individual Neurons and 2 more not named here |
| soft_threshold | 5 other papers 5 dated earlier · 2 only as an assembly brick |
2105.04857 (2021-05) Leveraging Sparse Linear Layers for Debuggable Deep Networks |
2308.03685 (2023-08) Learning Concise and Descriptive Attributes for Visual Recognition 2405.06855 (2024-05) Linear Explanations for Individual Neurons and 2 more not named here |
| add_index_to_dataloader | 2 other papers 2 dated earlier |
2105.04857 (2021-05) Leveraging Sparse Linear Layers for Debuggable Deep Networks |
2408.01432 (2024-08) VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance |
| cos_similarity | 2 other papers 2 dated earlier |
2204.10965 (2022-04) CLIP-Dissect: Automatic Description of Neuron Representations in Deep Vision Networks |
2408.01432 (2024-08) VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance |
| cos_similarity_cubed | 1 other paper 1 dated earlier |
2408.01432 (2024-08) VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance |
|
| cos_similarity_cubed_single | 1 other paper 1 dated earlier |
2408.01432 (2024-08) VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance |
Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Catastrophic forgetting remains a fundamental challenge in continual learning, in which models often forget previous knowledge when fine-tuned on a new task. This issue is especially pronounced in class incremental learning (CIL), which is the most challenging setting in continual learning. Existing methods to address catastrophic forgetting often sacrifice either model interpretability or accuracy. To address this challenge, we introduce Class-Incremental Concept Bottleneck Model (CI-CBM), which leverage effective techniques, including concept regularization and pseudo-concept generation to maintain interpretable decision processes throughout incremental learning phases. Through extensive evaluation on seven datasets, CI-CBM achieves comparable performance to black-box models and outperforms previous interpretable approaches in CIL, with an average 36% accuracy gain. CI-CBM provides interpretable decisions on individual inputs and understandable global decision rules, as shown in our experiments, thereby demonstrating that human-understandable concepts can be maintained during incremental learning without compromising model performance. Our approach is effective in both pretrained and non-pretrained scenarios; in the latter, the backbone is trained from scratch during the first learning phase. Code is publicly available at github.com/importAmir/CI-CBM. Published in Transactions on Machine Learning Research (04/2026) problems, CI-CBM employs an effective mechanism to prevent concept drift and mitigate classifier bias (see Figure 1 and6). Our contributions are summarized below: • We introduce a new approach to learn inherently interpretable neural models for class incremental learning with much better performance and utility. Unlike previous interpretable approaches in CIL, our approach CI-CBM achieves 25-43% higher accuracy across multiple benchmarks while maintaining scalability and adaptability across various datasets and architectures, demonstrating strong performance both with pretrained backbone models and when trained from scratch. • We propose new techniques including concept regularization to prevent the learned concepts from losing their functionality while learning new concepts. Additionally, we utilize pseudo-concepts for previous class data in conjunction with actual concepts for new class data to incrementally train the sparse classifier across different phases. • We evaluate our approach by conducting comprehensive EFCIL evaluation scenarios and performing ablation studies to assess the impact of different components of the proposed method. The results demonstrate the superiority of our method over other interpretable models in CIL, achieving an average accuracy gain of 36%. 2 Related work Class Incremental Learning (CIL) aims to extend the capabilities of deep learning models to continuously learn from new data and adapt to new classes over time while retaining previously acquired knowledge. The goal is to develop a model that can classify all previously seen classes effectively at any stage of training, which requires a balance between plasticity, the flexibility to learn new features, and stability, the resistance to forgetting old information. The phenomenon of catastrophic forgetting (Goodfellow et al., 2013) illustrates a fundamental trade-off in CIL, where the effort to incorporate new information can result in the loss of valuable knowledge from previous phases, leading to a sharp drop in performance on previous classes. To address catastrophic forgetting, several approaches have been proposed: (I) Regularization-based methods: These approaches (Kirkpatrick et al., 2017;Wang et al., 2021) mitigate forgetting by constraining changes in critical parameters across phases. However, these methods lack reliable metrics for parameter importance and perform poorly in CIL (Van de Ven & Tolias, 2019). (II) Architecture-based methods: These approaches expand network capacity dynamically when a new phase arrives (Rusu et al., 2016; Yan et al., 2021). However, a key challenge is the increasing memory and computational costs as the architecture grows, making it crucial to manage the rate of expansion with each phase. (III) Rehearsal-based methods: These approaches utilize various sampling strategies, such as herding (Rebuffi et al., 2017), diversity-aware sampling (Bang et al., 2021), reservoir sampling (Buzzega et al., 2020), and greedy sampling (Prabhu et al., 2020), to store samples in bounded memory. These samples are then used for knowledge distillation (Li &
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2604.14519")
get_code_for_paper("2604.14519")
have("2604.14519")
The run record, dated, one paper per request, free:
curl https://syntology.ai/api/ran/2604.14519.json
A badge for a README (the split and the date, never a ratio):
[](https://syntology.ai/paper/2604.14519)
Connect an agent — have() is free.