metafloor-ai/manifest-orchestrator-2b
family: add Demand Planning specialist
rename scheme: orchestrator/specialist; links, family, headers synced
family: split into orchestrators + domain-experts
restore run.json (provenance)
card: hide per-run examples; note proprietary ~12.5k dataset
remove internal run.json (provenance)
card: training-set size 12.5k
fix: remove 'Dataset used to train' (eval != training); state training data is not open-sourced
chore: keep coming-soon Manifest 30B sorted last on the org page
license: CC-BY-NC-4.0 (non-commercial; commercial via MetaFloor)
card: add 30B coming-soon row; drop 'flagship' framing
Delete files eval_report.md with huggingface_hub
Manifest 2B — customer-facing card rewrite (flavor, win-rate, suite)
rebrand: Manifest 2B card (family + size + availability; Qwen out of header)
supply-chain Qwen3.5-2B LoRA adapter + model card, run + eval report
initial commit
