---
license: mit
task_categories:
- text-generation
tags:
- iec-61131-3
- function-block-diagram
- plcopen
- taskweft
- editscore
configs:
- config_name: react
  default: true
  data_files:
  - split: train
    path: data/react/*.parquet
  - split: test
    path: test/data/react/*.parquet
  - split: evaluation
    path: evaluation/data/react/*.parquet
- config_name: react_root
  data_files:
  - split: train
    path: data/react_root/*.parquet
  - split: test
    path: test/data/react_root/*.parquet
  - split: evaluation
    path: evaluation/data/react_root/*.parquet
- config_name: react_candidates
  data_files:
  - split: train
    path: data/react_candidates/*.parquet
  - split: test
    path: test/data/react_candidates/*.parquet
  - split: evaluation
    path: evaluation/data/react_candidates/*.parquet
- config_name: react_scores
  data_files:
  - split: train
    path: data/react_scores/*.parquet
  - split: test
    path: test/data/react_scores/*.parquet
  - split: evaluation
    path: evaluation/data/react_scores/*.parquet
---
# taskweft-fbd-react-train

Intents paired with IEC 61131-3 Function Block Diagrams that carry them out, as an EditScore-shaped corpus.

## What it is for

Each intent has a root row, three ranked candidate diagrams and a score row per candidate.
Rank1 is the reference diagram. Rank3 compiles and does the wrong thing. Rank5 is a diagram the compiler refuses.
The scores come from the compiler's reference scan on constructed input traces.
Every row is built from a template and a seed, so the labels are true by construction and the corpus regenerates from the seeds.
The `train` split trains. The `test` split holds out every tenth seed of the same distribution and gates the trained model. The `evaluation` split holds out the `button_sequence_then_idle` family, and is never trained or tuned on.

The `react_root` config holds the intent, template, seed, frame, the block kinds and signatures the reference uses, and the input traces.
The `react_candidates` config holds each candidate's compiler text form, PLCopen XML, plan and runner result, with a sha256 of its source text.
The `react_scores` config holds parse, compile, run and effect-match flags, step and wall-time measures, and the refusal.
The default `react` config joins the three. Every artifact is in the parquet. An artifact a family does not produce is an empty string, never a null.
A `manifest.json` sits at the repo root.

## Provenance

- Generator: `v-sekai-fabric/taskweft-fbd-teacher`.
- Compiler: `taskweft-fbd-compiler` at commit `d3279ec`.

## Use

```python
from datasets import load_dataset
ds = load_dataset("chibifire/taskweft-fbd-react-train", "react", split="train")
```

Or fetch the files with `hf download chibifire/taskweft-fbd-react-train --repo-type dataset`.

## Licence

MIT. See the [MIT License](https://opensource.org/licenses/MIT).