Cover Image

CapTrack: Multifaceted Evaluation of Forgetting in LLM Post-Training

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Thede, Lukas, Winzeck, Stefan, Akata, Zeynep, Schwarz, Jonathan Richard
Format:	Preprint
Published:	2026
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

WikiBigEdit: Understanding the Limits of Lifelong Knowledge Editing in LLMs
by: Thede, Lukas, et al.
Published: (2025)

Reflecting on the State of Rehearsal-free Continual Learning with Pretrained Models
by: Thede, Lukas, et al.
Published: (2024)

Fantastic Gains and Where to Find Them: On the Existence and Prospect of General Knowledge Transfer between Any Pretrained Model
by: Roth, Karsten, et al.
Published: (2023)

Knowledge Graph-Assisted LLM Post-Training for Enhanced Legal Reasoning
by: Song, Dezhao, et al.
Published: (2026)

Align-then-Unlearn: Embedding Alignment for LLM Unlearning
by: Spohn, Philipp, et al.
Published: (2025)

Sparse Autoencoders are Topic Models
by: Girrbach, Leander, et al.
Published: (2025)

Geometry Conflict: Explaining and Controlling Forgetting in LLM Continual Post-Training
by: Wang, Yuanyi, et al.
Published: (2026)

Reference-Free Rating of LLM Responses via Latent Information
by: Girrbach, Leander, et al.
Published: (2025)

A Systematic Study of In-the-Wild Model Merging for Large Language Models
by: Hitit, Oğuz Kağan, et al.
Published: (2025)

From Drop-off to Recovery: A Mechanistic Analysis of Segmentation in MLLMs
by: Wu, Boyong, et al.
Published: (2026)

DeLoRA: Decoupling Angles and Strength in Low-rank Adaptation
by: Bini, Massimo, et al.
Published: (2025)

Discovering Chunks in Neural Embeddings for Interpretability
by: Wu, Shuchen, et al.
Published: (2025)

A Quantitative Characterization of Forgetting in Post-Training
by: Balasubramanian, Krishnakumar, et al.
Published: (2026)

Structural Pruning of Large Vision Language Models: A Comprehensive Study on Pruning Dynamics, Recovery, and Data Efficiency
by: Huang, Yiran, et al.
Published: (2026)

Investigating Structural Pruning and Recovery Techniques for Compressing Multimodal Large Language Models: An Empirical Study
by: Huang, Yiran, et al.
Published: (2025)

Stitch: Training-Free Position Control in Multimodal Diffusion Transformers
by: Bader, Jessica, et al.
Published: (2025)

Learning is Forgetting: LLM Training As Lossy Compression
by: Conklin, Henry C., et al.
Published: (2026)

Automatic Expert Discovery in LLM Upcycling via Sparse Interpolated Mixture-of-Experts
by: Chen, Shengzhuang, et al.
Published: (2025)

TimeSAE: Sparse Decoding for Faithful Explanations of Black-Box Time Series Models
by: Oublal, Khalid, et al.
Published: (2026)

Mapping Post-Training Forgetting in Language Models at Scale
by: Harmon, Jackson, et al.
Published: (2025)

Post-hoc Probabilistic Vision-Language Models
by: Baumann, Anton, et al.
Published: (2024)

Logits Replay + MoClip: Stabilized, Low-Cost Post-Training with Minimal Forgetting
by: Qiu, Suming, et al.
Published: (2025)

SOTAlign: Semi-Supervised Alignment of Unimodal Vision and Language Models via Optimal Transport
by: Roschmann, Simon, et al.
Published: (2026)

Dissecting Multimodal In-Context Learning: Modality Asymmetries and Circuit Dynamics in modern Transformers
by: Huang, Yiran, et al.
Published: (2026)

SUB: Benchmarking CBM Generalization via Synthetic Attribute Substitutions
by: Bader, Jessica, et al.
Published: (2025)

ETHER: Efficient Finetuning of Large-Scale Models with Hyperplane Reflections
by: Bini, Massimo, et al.
Published: (2024)

The Manifold Hypothesis for Gradient-Based Explanations
by: Bordt, Sebastian, et al.
Published: (2022)

Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training
by: Lai, Song, et al.
Published: (2025)

SubTrack++ : Gradient Subspace Tracking for Scalable LLM Training
by: Rajabi, Sahar, et al.
Published: (2025)

Building, Reusing, and Generalizing Abstract Representations from Concrete Sequences
by: Wu, Shuchen, et al.
Published: (2024)

Noise Hypernetworks: Amortizing Test-Time Compute in Diffusion Models
by: Eyring, Luca, et al.
Published: (2025)

Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models
by: Singhi, Nishad, et al.
Published: (2024)

Prompt Curriculum Learning for Efficient LLM Post-Training
by: Gao, Zhaolin, et al.
Published: (2025)

Train Once, Forget Precisely: Anchored Optimization for Efficient Post-Hoc Unlearning
by: Sanga, Prabhav, et al.
Published: (2025)

PostTrainBench: Can LLM Agents Automate LLM Post-Training?
by: Rank, Ben, et al.
Published: (2026)

Dr. Post-Training: A Data Regularization Perspective on LLM Post-Training
by: Hu, Pingbang, et al.
Published: (2026)

DataDream: Few-shot Guided Dataset Generation
by: Kim, Jae Myung, et al.
Published: (2024)

The Latent Color Subspace: Emergent Order in High-Dimensional Chaos
by: Pach, Mateusz, et al.
Published: (2026)

Time Series Representations for Classification Lie Hidden in Pretrained Vision Transformers
by: Roschmann, Simon, et al.
Published: (2025)

Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models
by: Pach, Mateusz, et al.
Published: (2025)