Research map / Large language models / Language, safety and interpretability
TGEA 2.0: A Large-Scale Diagnostically Annotated Dataset with Benchmark Tasks for Text Generation of Pretrained Language Models
Huibin Ge, Xiaohu Zhao, Chuang Liu, Yulong Zeng, Qun Liu, Deyi Xiong
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add alpha-balanced loss (ablation) | RoBERTa · TGEA 2.0 erroneous text detection (test) | F1 (%): 13.54 -> 29.29 | Improved |
| add alpha-balanced loss (ablation) | MacBERT · TGEA 2.0 erroneous text detection (test) | F1 (%): 14.02 -> 30.88 | Improved |
| add alpha-balanced loss (ablation) | RoBERTa · TGEA 2.0 MiSEW extraction (test) | F1 (%): 26.96 -> 38.76 | Improved |
| add alpha-balanced loss (ablation) | MacBERT · TGEA 2.0 MiSEW extraction (test) | F1 (%): 27.00 -> 38.22 | Improved |
| add alpha-balanced loss (ablation) | RoBERTa · TGEA 2.0 erroneous span location (test) | F1 (%): 15.86 -> 31.08 | Improved |
| add alpha-balanced loss (ablation) | MacBERT · TGEA 2.0 erroneous span location (test) | F1 (%): 17.50 -> 31.32 | Improved |
| add alpha-balanced loss (ablation) | RoBERTa · TGEA 2.0 erroneous text detection (test) | accuracy (%): 81.99 -> 80.50 | Hurt |
| generative fine-tuning on correct sentences (ablation) | GPT-2 · TGEA 2.0 pairwise comparison (test) | accuracy: 55.20 -> 50.32 | Hurt |
| generative fine-tuning on correct sentences (ablation) | GPT-2 · TGEA 2.0 word prediction (test) | Top-1 accuracy: 30.73 -> 41.23 | Improved |
| mixed generative-discriminative fine-tuning (ablation) | GPT-2 · TGEA 2.0 pairwise comparison (test) | accuracy: 55.20 -> 73.38 | Improved |
| mixed generative-discriminative fine-tuning (ablation) | GPT-2 · TGEA 2.0 word prediction (test) | Top-1 accuracy: 30.73 -> 32.28 | Improved |
| fine-tune on correct sentences before mixed fine-tuning (ablation) | MF-GPT-2 · TGEA 2.0 pairwise comparison (test) | accuracy: 73.38 -> 74.13 | Improved |
Closest papers by meaning
- Learning from others' mistakes: Finetuning machine translation models with span-level error annotations (ICML 2025 · similarity 0.86)
- Prometheus: Inducing Fine-Grained Evaluation Capability in Language Models (ICLR 2024 · similarity 0.85)
- Neither Valid nor Reliable? Investigating the Use of LLMs as Judges (NeurIPS 2025 · similarity 0.85)
- DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text (ICLR 2024 · similarity 0.85)
- RevisEval: Improving LLM-as-a-Judge via Response-Adapted References (ICLR 2025 · similarity 0.85)
- Automating Dataset Updates Towards Reliable and Timely Evaluation of Large Language Models (NeurIPS 2024 · similarity 0.85)
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.
