AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Beyond Description: Cognitively Benchmarking Fine-Grained Action for Embodied Agents

Dayong Liu, Chao Xu, Weihong Chen, Suyu Zhang, Juncheng Wang, Jiankang Deng, Baigui Sun, Yang Liu

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 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
SFT on CFG-Bench dataQwen2.5-VL-7B · EB-ALFREDsuccess rate: 4.7 -> 9.7Improved
SFT on CFG-Bench dataQwen2.5-VL-7B · EB-Manipulationsuccess rate: 9.6 -> 15.3Improved
SFT on CFG-Bench dataInternVL3-8B · EB-ALFREDsuccess rate: 10.3 -> 14.0Improved
SFT on CFG-Bench dataInternVL3-8B · EB-Manipulationsuccess rate: 11.5 -> 15.8Improved
SFT on FAVOR data instead of CFG dataQwen2.5-VL-7B · EB-ALFREDsuccess rate: 4.7 -> 4.3Hurt
SFT on FAVOR dataQwen2.5-VL-7B · EB-Manipulationsuccess rate: 9.6 -> 8.2Hurt
add CFG-Bench SFT before VLA trainingStarVLA-pi (Qwen3-VL-4B) · LIBERO Goalsuccess rate: 98.6 -> 99.4Improved
add CFG-Bench SFT before VLA trainingStarVLA-pi (Qwen3-VL-4B) · LIBERO Longsuccess rate: 94.4 -> 95.2Improved
add CFG-Bench SFT before VLA trainingStarVLA-pi (Qwen3-VL-4B) · LIBERO Spatialsuccess rate: 98.2 -> 98.2No clear effect
increase input frame rate from 4 to 8 fpsQwen2.5-VL-7B · CFG-Bench close-endedaccuracy (Avg_c): 40.71 -> 42.40No clear effect
scale model size from 7B to 72BQwen2.5-VL · CFG-Bench close-endedaccuracy (Avg_c): 40.71 -> 51.28Improved
embodied fine-tuning (RoboBrain2.0 from Qwen2.5-VL-7B)Qwen2.5-VL-7B · CFG-Bench close-endedaccuracy (Avg_c): 40.71 -> 44.53Improved

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