AI Paper Atlas

Research map / Large language models / Reasoning and chain of thought

BOLT: Decision‑Aligned Distillation and Budget-Aware Routing for Constrained Multimodal QA on Robots

Tengjun Ni, Xin Yuan, Shenghong Li, Kai Wu, Ren Ping Liu, Wei Ni, Wenjie Zhang

ICLR 2026 · Large language models › Reasoning and chain of thought · approach: Efficient and adaptive thinking

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes BOLT, a decision-aligned distillation and budget-aware routing framework for constrained multimodal question answering on robots.

Problem

Robotic systems need multimodal reasoning under tight latency, memory, and energy constraints, where standard instruction tuning and token-level distillation fail to deliver decision quality, reliability, and interpretability.

Approach

Treats multiple-choice prediction as a decision surface that is aligned during distillation (decision-aligned distillation), and adds budget-aware routing to meet resource constraints, instead of matching teacher tokens alone.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
option-level decision distillation instead of token-level KD (ablation)Qwen2-VL-2B (teacher LLaVA-1.5-13B) · Robo2VLM-1accuracy: 37.58 -> 42.89Improved
option-level decision distillation instead of token-level KD (ablation)Qwen2-VL-2B (teacher LLaVA-1.5-7B) · Robo2VLM-1accuracy: 33.91 -> 38.53Improved
option-level decision distillation instead of token-level KD (ablation)PaliGemma2-3B (teacher LLaVA-1.5-13B) · Robo2VLM-1accuracy: 35.50 -> 39.64Improved
distill from LLaVA-1.5-13B with ODDQwen2-VL-2B · Robo2VLM-1accuracy: 28.66 -> 42.89Improved
add budget-aware test-time augmentation (bTTA)ODD student · Robo2VLM-1accuracy: 42.89 -> 50.50Improved
add high-resolution re-evaluation only (ablation)ODD student · Robo2VLM-1accuracy: 42.89 -> 46.64Improved
add type-matched retrieval exemplars only (ablation)ODD student · Robo2VLM-1accuracy: 42.89 -> 44.31Improved
add question decomposition only (ablation)ODD student · Robo2VLM-1accuracy: 42.89 -> 45.47Improved
combine HR, tmRAG and QD (ablation)ODD student · Robo2VLM-1accuracy: 46.64 -> 50.50Improved
replace zero-shot teacher by distilled student with bTTA (deployment)BOLT · GPU memoryMB: 26,878 -> 3,035Improved
use BOLT instead of 13B teacherBOLT · latency per questionseconds: 22.30 -> 8.97Improved
use retrieved exemplars (tmRAG)BOLT · Robo2VLM-1 contradiction rateRCR: RCR 21.73%Hurt

Experimental setup

Baselines: standard instruction tuning token-level distillation

Builds on: knowledge distillation

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