AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

The LLM Bottleneck: Why Open-Source Vision LLMs Struggle with Hierarchical Visual Recognition

Yuwen Tan, Yuan Qing, Boqing Gong

CVPR 2026 · Vision-language and multimodal › Visual question answering · approach: Vision-language model VQA evaluation · 13 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
LoRA finetune on iNat21-Plant hierarchical VQA tasksQwen2.5-VL-7B · iNat21-PlantHCA: 17.67 -> 29.34Improved
Same LoRA finetuning evaluated on other datasetsQwen2.5-VL-7B · ImgNet-AnimalHCA: 56.00 -> 58.62Improved
Same LoRA finetuning evaluated on CUB-200Qwen2.5-VL-7B · CUB-200HCA: 43.76 -> 46.17Improved
LoRA finetuning measured on text-only taxonomy QA of the LLMLLM of Qwen2.5-VL-7B · iNat21-Plant (text-only)text HCA: 64.21 -> 84.87Improved
Scale Qwen2.5-VL from 7B to 72B (ablation)Qwen2.5-VL · iNat21-AnimalHCA: 19.43 -> 35.73Improved
Vary prompts (hierarchical, chain-of-thought, JSON taxonomy) (ablation)VLLMs · CUB-200, iNat21-Plant, ImgNet-AnimalHCA / Acc_leafNo clear effect
Linearly probe visual tokens instead of VLLM output (ablation)Qwen2.5-VL-7B · CUB-200, iNat21-PlantHCAImproved
Compare original LLMs vs vision-language-tuned LLMs (ablation)Qwen2.5-VL / InternVL2.5 LLMs · five taxonomies (text-only)text HCAImproved
Linear probes on LLM layer embeddings instead of LLM decoding (ablation)Qwen2.5-VL-7B LLM · iNat21-Plant (text-only)text HCAImproved
Use closed-source LLM with strong taxonomy knowledge (GPT-4o) (ablation)GPT-4o · five taxonomiestext HCA: 98.81Improved
Orthogonal hierarchical representation in LLM (ablation)Qwen2.5-7B / Gemma · taxonomy QAHCANo clear effect
Reduce taxonomy depth (ablation)LLMs of VLLMs · iNat21-Plant (text-only)text HCAImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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