DIA state (dia-state)

This dataset is the storage backend for the DIA Tracker dashboard, not a conventional dataset.

It holds a single file, state.json, which stores the accounting state: each tracked model's parsed dia_report (energy / carbon / water + per-field quality tiers) and its lineage (base → derivative edges). The dashboard reads and updates this file.

Why the Dataset Viewer is empty

state.json is a nested JSON object, not a flat table (CSV / Parquet / JSON-Lines), so Hugging Face's automatic Dataset Viewer has nothing tabular to render and shows "no data." This is expected — the data is there.

How to view or use the data

• Open it directly: Files → state.json

• Or load it in Python:

  from huggingface_hub import hf_hub_download
  import json
  path = hf_hub_download("DIA-MVP/dia-state", "state.json", repo_type="dataset")
  state = json.load(open(path))
  print(len(state["nodes"]), "models tracked")
  

The data is public on purpose: DIA accounting is meant to be open and auditable.