WritingBench: A Comprehensive Benchmark for Generative Writing
WritingBench / X-PLUG project

Source aggregate workbook:
https://huggingface.co/spaces/WritingBench/WritingBench/resolve/c06986e05aea53d625837e67c88944a2234271f1/score.xlsx

Space revision: c06986e05aea53d625837e67c88944a2234271f1
Workbook SHA-256: 623a59886c6b755724828a5e302df30e93b5aa30cd86bfff09ec512746d4b3d9
Source repository declaration: Apache License 2.0
https://huggingface.co/spaces/WritingBench/WritingBench/blob/c06986e05aea53d625837e67c88944a2234271f1/README.md
License copy: writingbench-apache-2.0.txt

BenchLM changed the representation on October 1, 2026: aggregate worksheet values were extracted to JSON; overall and six domain scores were retained without rescaling, rounding, or recomputation. Three reviewed model profile references were added. Table display rounds numeric values to two decimal places while the underlying snapshot retains source precision.

The source compilation retains its Apache License 2.0 terms. This notice records attribution and transformations added by BenchLM; it does not imply endorsement by the WritingBench project. No task texts, evaluation prompts, model outputs, or model weights are included.