| --- |
| tags: |
| - smart-manufacturing |
| - additive-manufacturing |
| - sft |
| license: other |
| pretty_name: AM12_MeltPoolKinetics-perception |
| extra_gated_fields: |
| Name: text |
| Affiliation: text |
| Intended use: text |
| extra_gated_prompt: >- |
| This dataset is released for **research use** and access is reviewed and granted |
| **manually** by the maintainers. Licensing of the underlying source is NOT yet cleared |
| (see the card). Please state your name, affiliation, and intended use. |
| --- |
| |
| # AM12_MeltPoolKinetics-perception |
| |
| NIR melt-pool ROI frame -> melt-pool state/condition: 4-way {normal,underheat,overheat,edge} (T-B2; edge is a scan-position/boundary condition, not a thermal state) and binary normal/anomaly (T-B1 over non-boundary frames only). Category **B**, task **T-B2 + T-B1**, in the unified Smart-Manufacturing SFT schema. |
| |
| ## Records |
| |
| **81,099** records (train=81,099). Model input: a single NIR melt-pool ROI image (196x196), bytes embedded in the `image` column. |
| |
| ## Unified SFT schema (7 fields) |
| |
| | field | type | meaning | |
| |---|---|---| |
| | `query` | str | the question / instruction (model input) | |
| | `image` | Image \| null | the INPUT image (bytes embedded) — null for tabular records | |
| | `annot` | str | the answer — for this dataset: the melt-pool state/condition label (text); the `task` column selects the 4-way vs binary framing | |
| | `reasoning` | null | no native CoT in this dataset | |
| | `cate` | "B" | SFT category | |
| | `task` | "T-B2 + T-B1" | unified task id | |
| | `metadata` | str (JSON) | split, provenance, units, `input`, license, and (for AM11/13) `taxonomy_note` | |
|
|
| ## Notes |
|
|
| Two tasks share one dataset (distinguished by the `task` column: 45,056 T-B2 + 36,043 T-B1), but these are **not 81k independent images**: the export has 45,056 unique input images and 36,043 images appear in both task framings. T-B1 covers only non-boundary frames; the 9,013 edge/turnaround frames from T-B2 are excluded, so binary is a thermal-only normal/anomaly discriminator. Do not map edge to normal or anomaly. In deployment, filter boundary frames using scan position/trajectory metadata or route them to T-B2, where `edge` is explicit. Only source16 carries genuine per-frame labels. The Hub viewer shows a single `train=81,099` container split because the dataset is uploaded as one combined parquet; this is **not** the recommended training split. Do not use frame-random splits: adjacent frames inside a build are near-duplicates, and the same image can appear once as T-B2 and once as T-B1. Split downstream by `metadata.build` or `metadata.recommended_build_split`; the source temporal split is preserved in `metadata.source_temporal_split`. Two misfiled non-data PNGs in the source are excluded. |
|
|
| ## Provenance & licensing |
|
|
| Underlying source: **MeltPoolKinetics figshare compilation, source16 = NIST AMMT; DOI 10.6084/m9.figshare.28200101**. Upstream license: **figshare compilation — per-source terms (source16 = NIST AMMT); verify per source**. |
| Converted read-only into the unified schema; conversion + publish scripts live in |
| [`AI4Manufacturing/forge_model`](https://github.com/AI4Manufacturing/forge_model) |
| (`AM12_MeltPoolKinetics-perception/` + `publish/push_am_to_hf.py`). Access is **gated (manual approval)**; clear the |
| upstream licence before any onward redistribution. |
|
|
