Skip to content

Dataset · yongchao98

Symbolic tasks and data for SFT and DPO

yongchao98/SymBench

A dataset of 37 symbolic tasks, generation scripts, and English data for SFT and DPO training.

What's inside

An English-language dataset created for the CodeSteerLLM model paper. It includes the SymBench benchmark of 37 symbolic tasks, ready-to-use samples for SFT and DPO fine-tuning, and scripts for synthesizing new examples. The total volume ranges from 1,000 to 10,000 records.

What you can do with it

  • SFT and DPO fine-tuning of language models for logical and symbolic tasks
  • Evaluating models on a 37 symbolic test benchmark
  • Generating synthetic data using provided scripts
Tasks
Text generation
Languages
English
Size
1K–10K rows
Downloads this month
21.4K
Likes
5

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("yongchao98/SymBench")