Hugging Face's logo

Datasets:
DuoNeural
/
Gemma4-E2B-SFT-JSON

Modalities:
Text
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
Synthetic
json
entity-extraction
structured-output
sft
gemma4
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Gemma4-E2B-SFT-JSON

Synthetic structured JSON entity extraction dataset. Model receives unstructured document → outputs strictly valid JSON.

Domains: Medical (clinical notes), legal (contracts), financial (earnings reports), job postings, research papers. Format: ShareGPT/ChatML. Two-step generation: document synthesized first, then extracted. Use: Fine-tune models for robust information extraction and structured output generation. Generator: DuoNeural/TurboGemma4E2B, temperature 0.1 for extraction step.
Downloads last month
17
Total file size:
14.3 MB

Models trained or fine-tuned on DuoNeural/Gemma4-E2B-SFT-JSON