Datasets:

Seono
/

sft-shp-eval

dataset_info:
  features:
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 26636179
      num_examples: 18436
  download_size: 4501319
  dataset_size: 26636179
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

SFT Format Dataset

Overview

This dataset is converted to SFT (Supervised Fine-Tuning) format. It was created by transforming OpenMathInstruct and Stanford Human Preferences (SHP) datasets.

Dataset Structure

Each entry follows this format:

Instruction: [Problem, question, or conversation history]

Response: [Solution, answer, or response]

Usage Guide

Loading the Dataset

from datasets import load_dataset

# Load datasets from Hugging Face
openmath_train = load_dataset("Seono/sft-openmath-train")
openmath_eval = load_dataset("Seono/sft-openmath-eval")
shp_train = load_dataset("Seono/sft-shp-train")
shp_eval = load_dataset("Seono/sft-shp-eval")


Using for Fine-tuning

# Setting for SFT Trainer
trainer = SFTTrainer(
    model=model,
    tokenizer=tokenizer,
    args=training_args,
    train_dataset=openmath_train["train"],
    dataset_text_field="text"
)

then other codes..