README.md

---
license: cdla-permissive-2.0
task_categories:
- text-generation
- question-answering
- text-classification
language:
- en
tags:
- cryptography
- security
- cybersecurity
- security-protocols
- protocol-verification
- formal-methods
- fstar
size_categories:
- 10K<n<100K
configs:
- config_name: default
  data_files:
  - split: train
    path: "fstar_train.jsonl"
  - split: validation
    path: "fstar_validation.jsonl"
  - split: test
    path: "fstar_test.jsonl"
---
# F* Proof Completion Dataset (Chat Format)

This dataset is a preprocessed version of [`microsoft/FStarDataSet-V2`](https://huggingface.co/datasets/microsoft/FStarDataSet-V2). It has been reformatted into a chat-style JSONL structure for supervised fine-tuning of language models on F* function synthesis and proof completion.

## Dataset Structure

The dataset consists of three splits:

* `fstar_train.jsonl`
* `fstar_validation.jsonl`
* `fstar_test.jsonl`

Each line in these files is a JSON object with the following schema (where the keys correspond to the keys in the original dataset):

```json
{
  "messages": [
    {"role": "user", "content": "Complete this F* function with a correct proof.\nType: <verbose_type>\nEffect: <effect>\nContext: <file_context>\nPartial implementation: <partial_definition>"},
    {"role": "assistant", "content": "<completed_definiton>"}
  ]
}
```

This format is compatible with supervised fine-tuning frameworks such as TRL, Fireworks AI SFT, or OpenAI-style chat tuning.

## Source and License

* **Source:** [`microsoft/FStarDataSet-V2`](https://huggingface.co/datasets/microsoft/FStarDataSet-V2)
* **License:** CDLA Permissive 2.0 (inherits from source)
* **Modifications:** Reformatted to chat-based JSONL files. No additional data or content changes were made.

## Intended Use

The dataset is intended for research and experimentation in program synthesis, theorem proving, and fine-tuning large language models on functional proof-oriented languages such as F*.