Can MLLMs "Read" What is Missing?
On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Abstract (arXiv)
We introduce MMTR-Bench, a benchmark designed to evaluate the intrinsic ability of Multimodal Large Language Models (MLLMs) to reconstruct masked text directly from visual context. Unlike conventional question-answering tasks, MMTR-Bench eliminates explicit prompts, requiring models to recover masked text from single- or multi-page inputs across real-world domains such as documents and webpages. This design isolates the reconstruction task from instruction-following abilities, enabling a direct assessment of a model's layout understanding, visual grounding, and knowledge integration. MMTR-Bench comprises 2,771 test samples spanning multiple languages and varying target lengths. To account for this diversity, we propose a level-aware evaluation protocol. Experiments on representative MLLMs show that the benchmark poses a significant challenge, especially for sentence- and paragraph-level reconstruction. The homepage is available at https://mmtr-bench-dataset.github.io/MMTR-Bench/.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. First posted on arXiv 23 Apr 2026.
Builds on
0 of 0 references.
No references here yet.
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2023SEED-Bench-2: Benchmarking Multimodal Large Language Models
- 2024A Survey on Benchmarks of Multimodal Large Language Models
- 2024SEED-Bench-2-Plus: Benchmarking Multimodal Large Language Models with Text-Rich Visual Comprehension
- 2024Unveiling the Ignorance of MLLMs: Seeing Clearly, Answering Incorrectly
- 2024Efficient Multimodal Large Language Models: A Survey
- 2026MMOOC: A Comprehensive Benchmark for Out-of-Context Evaluation in Multimodal Large Language Models
- 2024From Text to Pixel: Advancing Long-Context Understanding in MLLMs
- 2024SV-RAG: LoRA-Contextualizing Adaptation of MLLMs for Long Document Understanding