File size: 3,601 Bytes
8d14517
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
---
license: other
pretty_name: RPC-Bench
task_categories:
  - question-answering
language:
  - en
tags:
  - research-paper
  - document-understanding
  - multimodal
  - benchmark
  - llm
  - vlm
---

<div align="center">

# RPC-Bench: A Fine-grained Benchmark for Research Paper Comprehension

</div>

<p align="center">
    🌐 <a href="https://rpc-bench.github.io/" target="_blank">Project Page</a> β€’
    πŸ’» <a href="https://github.com/RPC-Bench/PRC-Bench" target="_blank">GitHub</a> β€’
    πŸ“– <a href="https://arxiv.org/abs/2601.14289" target="_blank">Paper</a> β€’
    πŸ€— <a href="https://arxiv.org/abs/2601.14289" target="_blank">Paper</a> β€’
    🧭 <a href="https://community.modelscope.cn/" target="_blank">ModelScope</a>
</p>

<div align="center">
    <img src="assets/pipeline.png" width="100%" />
</div>

RPC-Bench is a fine-grained benchmark for research paper comprehension. It is built from review-rebuttal exchanges of high-quality academic papers and supports both text-only and visual evaluation through complementary paper representations.

## Data Structure

RPC-Bench is split into `train`, `dev`, and `test` subsets. Each subset is stored in the dataset structure and recorded in `manifest.jsonl`.

`md/` contains Markdown files parsed from each paper by MinerU. These files provide the text input for LLM-oriented evaluation.

`parse/` contains the full MinerU parsing outputs for each paper, including structured layout and content artifacts.

`pdf/` contains the original paper PDFs.

`vlm/` contains page images rendered from the PDFs with PyMuPDF at 200 DPI for VLM-oriented evaluation.

```text
RPC-Bench/
β”œβ”€β”€ README.md
β”œβ”€β”€ manifest.jsonl
β”œβ”€β”€ parse/
β”‚   β”œβ”€β”€ train/
β”‚   β”‚   └── <paper_id>/
β”‚   β”œβ”€β”€ dev/
β”‚   β”‚   └── <paper_id>/
β”‚   └── test/
β”‚       └── <paper_id>/
β”œβ”€β”€ md/
β”‚   β”œβ”€β”€ train/
β”‚   β”‚   └── <paper_id>/
β”‚   β”‚       └── <paper_id>.md
β”‚   β”œβ”€β”€ dev/
β”‚   β”‚   └── <paper_id>/
β”‚   β”‚       └── <paper_id>.md
β”‚   └── test/
β”‚       └── <paper_id>/
β”‚           └── <paper_id>.md
β”œβ”€β”€ pdf/
β”‚   β”œβ”€β”€ train/
β”‚   β”‚   └── <paper_id>.pdf
β”‚   β”œβ”€β”€ dev/
β”‚   β”‚   └── <paper_id>.pdf
β”‚   └── test/
β”‚       └── <paper_id>.pdf
└── vlm/
    β”œβ”€β”€ train/
    β”‚   └── <paper_id>/
    β”œβ”€β”€ dev/
    β”‚   └── <paper_id>/
    └── test/
        └── <paper_id>/
```

## Practical Uses

RPC-Bench can be used to try paper-centric systems that require broader document understanding rather than local snippet matching.

- Research paper comprehension: try models on full-paper understanding, including core concepts, methods, and experimental findings.
- Long-context evaluation: try whether longer context windows or long-context architectures improve document-level reasoning.
- Multimodal reasoning: try models that combine textual evidence with page-level figures, tables, and diagrams in the original PDF layout.
- RAG system diagnosis: try retrieval, chunking, and evidence-fusion strategies for paper-centric workflows beyond snippet-level retrieval accuracy.

## Citation

```bibtex
@article{chen2026rpc,
  title={RPC-Bench: A Fine-grained Benchmark for Research Paper Comprehension},
  author={Chen, Yelin and Zhang, Fanjin and Sun, Suping and Pang, Yunhe and Wang, Yuanchun and Song, Jian and Li, Xiaoyan and Hou, Lei and Zhao, Shu and Tang, Jie and others},
  journal={arXiv preprint arXiv:2601.14289},
  year={2026}
}
```