AI Paper Atlas

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

Protect to Adapt: Orthogonal Subspace Control with Ranked Negative-Prompt Curriculum for Few-Shot Action Recognition

Hantao Qi, Yan Yan, Junlong Gao, Hanzi Wang

CVPR 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Synthetic data and merging for CLIP

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 OSC (orthogonal-subspace LoRA) to vanilla LoRA (ablation)P2A · HMDB51 5-way 1-shotaccuracy (%): 78.6 -> 79.9Improved
add OSC to vanilla LoRA (ablation)P2A · Kinetics-100 5-way 1-shotaccuracy (%): 90.7 -> 91.8Improved
add RNC curriculum negatives on top of OSC (ablation)P2A · HMDB51 5-way 1-shotaccuracy (%): 79.9 -> 82.3Improved
add RNC curriculum negatives on top of OSC (ablation)P2A · Kinetics-100 5-way 1-shotaccuracy (%): 91.8 -> 93.7Improved
use three-phase easy-medium-hard curriculum instead of a single difficulty level (ablation)P2A · HMDB51 5-way 1-shotaccuracy (%): 80.7 -> 82.3Improved
add negative prompts without OSC to LoRA (ablation)CLIP-FSAR with LoRA · HMDB51 5-way 1-shotaccuracy (%): 78.6 -> 77.3Hurt
add easy negative prompts to fully fine-tuned backbone (ablation)CLIP-FSAR full fine-tuning · HMDB51 5-way 1-shotaccuracy (%): 75.8 -> 73.8Hurt
use entropy-based effective rank instead of fixed subspace dimension (ablation)OSC · HMDB51 5-way 1-shotaccuracy (%): 79.6 -> 79.9Improved
apply OSC to upper layers 6-11 instead of bottom layers (ablation)OSC · HMDB51 5-way 1-shotaccuracy (%): 77.8 -> 79.9Improved
apply OSC to all 12 layers instead of upper half (ablation)OSC · HMDB51 5-way 1-shotaccuracy (%): 79.9 -> 79.5Hurt
add Wo projection to OSC on WqWvWk (ablation)OSC · HMDB51 5-way 1-shotaccuracy (%): 79.9 -> 79.7No clear effect
vary subspace dimension K from 1 to 128 (ablation)OSC · HMDB51 5-way 1-shotaccuracy (%): 79.6 -> 79.4No clear effect

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