Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Text Rewriting on Wikipedia biographies (test)
Loading...
48.5
Linkable N-gram Proportion (arity=1)
Baseline LLM-based rewriting
-1.732
11.309
24.35
37.391
Oct 7, 2025
Linkable N-gram Proportion (arity=1)
Linkable N-gram Proportion (arity≤3)
Semantic Similarity (distilroberta)
Semantic Similarity (GTE)
Perplexity (Gemma3)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Linkable N-gram Proportion (arity=1)
Linkable N-gram Proportion (arity≤3)
Semantic Similarity (distilroberta)
Semantic Similarity (GTE)
Perplexity (Gemma3)
Baseline LLM-based rewriting
span_strategy=without...
2025.10
48.5
55.3
98.9
99.4
10.7
Handcrafted paraphrasing
2025.10
25.6
35.5
95.2
97.8
19.17
DP-based rewriting
epsilon (ϵ)=100
2025.10
5.1
9.1
61.8
80.8
136
LLM-based rewriting w/ list of spans
span_extraction_arity=...
2025.10
1
34.6
84.1
91.4
12.9
DP-based rewriting
epsilon (ϵ)=10
2025.10
0.8
0.9
26.8
60.8
2,122
LLM-based rewriting w/ list of spans
span_extraction_arity=...
2025.10
0.2
0.3
73.3
85.8
14.7
Feedback
Search any
task
Search any
task