AI Paper Atlas

Research map / Continual learning / Continual fine-tuning of LLMs

The Blind Spot of Adaptation: Quantifying and Mitigating Forgetting in Fine-tuned Driving Models

Runhao Mao, Hanshi Wang, Yixiang Yang, Qianli Ma, Jingmeng Zhou, Zhipeng Zhang

CVPR 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Prompt Adapter on top of TAEM (ablation)DEA on Qwen2.5VL-3B · Fidelity Driving BenchKRR: 74.4% -> 79.0%Improved
add Prompt Adapter on top of TAEM (ablation)DEA on Qwen2.5VL-3B · Fidelity Driving BenchTraffic-QA: 35.0 -> 41.0Improved
use TAEM mixture of LoRA experts instead of single LoRA (ablation)Qwen2.5VL-3B · Fidelity Driving BenchKRR: 64.6% -> 74.4%Improved
use TAEM instead of single LoRA (ablation)Qwen2.5VL-3B · Fidelity Driving BenchTraffic-QA: 22.3 -> 35.0Improved
apply DEA to ImpromptuVLA-3BImpromptuVLA-3B · Fidelity Driving BenchKRR: 68.4% -> 76.0%Improved
DEA versus full fine-tuningQwen2.5VL-3B · Fidelity Driving BenchKRR: 64.3% -> 79.0%Improved
DEA versus base model on NoPRQwen2.5VL-3B · Fidelity Driving BenchNoPR: 36.8 -> 29.0Hurt
LoRA fine-tuning on diverse driving corpussix base VLMs · Fidelity Driving BenchNoPR: InternVL3-2B 50.8 -> 30.8; InternVL2-4B 37.1 -> 16.0Hurt
LoRA fine-tuning on Traffic-QAsix base VLMs · Fidelity Driving BenchTraffic-QA: LLAVA-OV-7B 35.9 -> 47.0; InternVL3-2B 31.5 -> 25.1Mixed
multi-source rather than single-source fine-tuning dataVLM driving models · Fidelity Driving BenchKRR: RecogDrive-8B 74% vs safeAuto-BDDX 31.5%Improved
full fine-tuning on driving dataInternVL3-8B (RecogDrive-8B) · Fidelity Driving BenchNoPR: 55.2 -> 41.3Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy