Card: full training procedure, corrected provenance SHA

Card: full training procedure, corrected provenance SHA

Card: full training procedure, corrected provenance SHA

Merged 592-record corpus + per-run lineage + dataset card

update manifest.json

stage 7: sft (141 records)

stage 6: final (141 records)

stage 5: draft_responses (141 records)

stage 4: refined_prompts (141 records)

stage 3: draft_prompts (143 records)

stage 2: scenarios (144 records)

update manifest.json

stage 7: sft (451 records)

stage 6: final (451 records)

stage 5: draft_responses (458 records)

stage 4: refined_prompts (458 records)

stage 3: draft_prompts (470 records)

stage 2: scenarios (472 records)

stage 1: traits (12 records)

initial commit