Papers with G**eneration
QGEval: Benchmarking Multi-dimensional Evaluation for Question Generation (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing metrics fail to align well with human judgments when evaluating QG questions. |
| Approach: | They propose a multi-dimensional evaluation benchmark for QG and automatic metrics that evaluates questions and automated metrics across 7 dimensions. |
| Outcome: | The proposed benchmark evaluates QG models and automatic metrics across 7 dimensions . it shows that most QG model performs unsatisfactorily in terms of answerability and answer consistency . |
R2AG: Incorporating Retrieval Information into Retrieval Augmented Generation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to augment large language models with external documents are lacking in the semantic gap between LLMs and retrievers due to differences in their training objectives and architectures. |
| Approach: | They propose to integrate R2AG into R2etrieval augmented generation framework by using a R2-Former to capture retrieval information. |
| Outcome: | The proposed framework fills the semantic gap between LLMs and retrievers due to differences in their training objectives and architectures. |
A Universal Avoidance Method for Diverse Multi-branch Generation (2026.findings-acl)
Copied to clipboard
| Challenge: | generative models still lack human-level creativity, especially in multi-branch diversity tasks. |
| Approach: | They propose a model-agnostic and computationally efficient generation strategy that penalizes similarity among previously generated outputs. |
| Outcome: | The proposed method achieves 1.9 times higher diversity, runs 4.4 times faster, and requires only 1/64 of the FLOPs compared to state-of-the-art methods. |