BriefComposer SFT

Multi-image analytical brief rows composed from completed FireWatch, OceanScout, LandShift, and FloodPulse dataset folders (metadata/ + images/). Each sample stitches 1–4 images and metadata-derived headlines into one executive-style assistant reply.

Record counts (this build)

Inputs

• Source roots: one or more --source-root directories (each must contain images/ and metadata/ from the profile builders).
• CLI: build_lfm_vl_brief_sft.py --samples N controls JSONL line count. Run after the four temporal profile datasets are built so metadata and PNGs exist.

Dataset layout

• data/train.jsonl, data/validation.jsonl, data/test.jsonl — VLM SFT samples (messages list with system/user/assistant content; images referenced by relative paths under this folder).
• images/ — PNG chips (typically mixed profile images per row) consumed by the JSONL.
• metadata/ — one JSON sidecar per tile/sample (scene ids, bbox, optional regions, profile tag).

Splits

Train / validation / test use a stable hash of the synthetic sample id (brief_######) assigned at compose time.

Regenerating locally

Run from the nutonic repository root (paths relative to that root):

python data/scripts/build_lfm_vl_brief_sft.py <see --help for required args>

Upload to the Hub (requires a token with write access to the target dataset repo):

export HF_TOKEN="hf_..."   # or HUGGING_FACE_HUB_TOKEN
python data/scripts/build_lfm_vl_brief_sft.py ... --upload-repo NuTonic/brief-composer-sft-v1 --hf-token "$HF_TOKEN"

Skip upload with --no-upload. Resume partial STAC downloads with --skip-existing.

Orchestrator (all PRO profiles)

python data/scripts/run_pro_sft_orchestrator.py \
  --events-per-profile 0 \
  --hf-org YOUR_ORG \
  --upload-repo-firewatch YOUR_ORG/firewatch-sft-v1 \
  --hf-token "$HF_TOKEN"

Use --upload-repo-<profile> per profile, or --hf-org to upload to YOUR_ORG/<default-repo-suffix> for each builder. See run_pro_sft_orchestrator.py --help.

Environment

STAC: scenes are resolved from Element84 Earth Search (sentinel-2-l2a by default). No Earth Engine or Mapbox keys are required for this pipeline.

Hub target for this artifact

Dataset repo id: NuTonic/brief-composer-sft-v1

Card URL: https://huggingface.co/datasets/NuTonic/brief-composer-sft-v1

Limitations

• Optical-only Sentinel-2; cloud cover and revisit gaps can drop events or leave empty change masks.
• Detections and captions are heuristic (indices / thresholds), not operational truth labels.
• Assistant text follows conservative wording (candidates, confidence qualifiers).

Hub layout (sharded)

This snapshot was processed with python data/scripts/shard_lfm_vl_dataset_for_hub.py so that images/, mapbox_stills/, overlays/, and metadata/ use at most 8000 files per leaf directory (Hub git limit: 10k files per directory). JSONL paths may include sNNNNN/ shard segments where needed.