Persian synthetic datasets which are generated by the SoTA LLMs.

ParsBench
community
AI & ML interests
Benchmarking LLMs on the Persian language tasks.
Organization Card
ParsBench is a Pythonic framework that provides toolkits for benchmarking LLMs based on the Persian language tasks.
In this HuggingFace organization, we share the prepared datasets, evaluation results, and fine-tuned models using the ParsBench toolkit.
Learn more about this framework at:
Collections 3
ParsBench task datasets with instructions in the Alpaca-style.
-
ParsBench/parsinlu-machine-translation-en-fa-alpaca-style
Viewer • Updated • 1.62M • 22 • 2 -
ParsBench/parsinlu-machine-translation-fa-en-alpaca-style
Viewer • Updated • 1.62M • 21 • 3 -
ParsBench/parsinlu-entailment-alpaca-style
Viewer • Updated • 755 • 22 -
ParsBench/farstail-entailment-alpaca-style
Viewer • Updated • 7.27k • 23
models
None public yet
datasets 19
ParsBench/Persian-MuSR
Viewer
• Updated
• 756 • 33
ParsBench/Persian-NoRobots
Viewer
• Updated
• 10k • 33 • 2
ParsBench/PersianSyntheticEmotions
Viewer
• Updated
• 8.75k • 50 • 1
ParsBench/PersianSyntheticQA
Viewer
• Updated
• 100k • 207 • 5
ParsBench/results
Viewer
• Updated
• 16 • 18
ParsBench/requests
Viewer
• Updated
• 16 • 7
ParsBench/task-matches
Viewer
• Updated
• 129k • 229 • 1
ParsBench/parsinlu-sentiment-analysis-alpaca-style
Viewer
• Updated
• 1.89k • 19 • 1
ParsBench/persian-math-alpaca-style
Viewer
• Updated
• 177 • 21
ParsBench/pnsummary-alpaca-style
Viewer
• Updated
• 82k • 37