CoolFace
Datasetpublic

schneiderkamplab/dfm8-synthetic-constrained-format-following

Constrained Format Following Synthetic DFM8 training data generated with Gemma 4 31B and filtered by deterministic checks plus a Gemma 4 31B judge. Schema Rows are JSONL chat records: {"messages": [{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]} Tool-calling rows may also include a top-level tools list and assistant tool_calls. Counts accepted rows: 838905 generated rows seen: 4800000 audit rows seen: 4580233… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm8-synthetic-constrained-format-following.

sourceHugging Faceupdated 3mo agoView on Hugging Face
0likes17downloads
Dataset Card

Constrained Format Following

Synthetic DFM8 training data generated with Gemma 4 31B and filtered by deterministic checks plus a Gemma 4 31B judge.

Schema

Rows are JSONL chat records:

json
{"messages": [{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]}

Tool-calling rows may also include a top-level tools list and assistant tool_calls.

Counts

  • —accepted rows: 838905
  • —generated rows seen: 4800000
  • —audit rows seen: 4580233

Intended Use

DFM8 behavioral-contract training slice. Do not treat as evaluation data.