AI Paper Atlas

Research map / Continual learning / Online and replay-based

CONCUR: A Framework for Continual Constrained and Unconstrained Routing

Peter Baile Chen, Weiyue Li, Dan Roth, Mike Cafarella, Samuel Madden, Jacob Andreas

ICLR 2026 · Continual learning › Online and replay-based · approach: Plasticity-preserving continual reinforcement learning · 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
replace best single strategy with CONCUR routingQwen2.5-7B CoT · Unconstrained routing, ID avg (2WikiMultiHop/MMLU/GSM8k)accuracy / FLOPs: 74.3 -> 75.2 acc; 41.63 -> 36.50 FLOPsImproved
replace best single strategy with CONCUR routingQwen2.5-7B CoT · Unconstrained routing, OOD avg (HotpotQA/GPQA/SVAMP)accuracy / FLOPs: 62.3 -> 62.6 acc; 46.13 -> 43.23 FLOPsImproved
use RouteLLM single-classifier routing instead of best single strategyRouteLLM · Unconstrained routing, ID avg (2WikiMultiHop/MMLU/GSM8k)accuracy: 74.3 -> 53.5Hurt
global DP optimization instead of local per-task budget (CONCUR, low budget) (ablation)CONCUR · Constrained routing ID avg, FLOPs budget 25accuracy: +5.7 averageImproved
global instead of local optimization (CONCUR, high budget) (ablation)CONCUR · Constrained routing ID avg, FLOPs budget 40accuracy: +4.0 averageImproved
global instead of local optimization (RTR, GSM8k, high budget) (ablation)RTR · GSM8k, FLOPs budget 40accuracy: +0.2No clear effect
add general-purpose representation to task-specific only (ablation)CONCUR · Unconstrained routing, ID avg, representation ablationaccuracy: 72.3 -> 75.2Improved
add task-specific representation to general-purpose only (ablation)CONCUR · Unconstrained routing, ID avg, representation ablationaccuracy: 73.3 -> 75.2Improved
modular per-strategy predictors instead of retraining single router when adding small modelsCONCUR · Continual routing Setting 2 (large+small models), ID avgrelative training time: 1.00x vs 7.66x (RTR_FS)Improved
fine-tune existing router on 25-100% of new data instead of full retrainRTR/EmbedLLM baselines · Continual routing Setting 2 (large+small models), ID avgaccuracy: RTR_FT(75%) 74.7 vs RTR_FT(50%) 71.9Mixed
use larger Setting 2 (add small models) for CONCURCONCUR · Continual ID avg FLOPsFLOPs: 40.84 -> 36.50Improved

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