Update model card: fix conditions, base model, benchmarks, languages
#1
by madiedgar - opened
- Base model: tiny-aya-base only (not global/fire/earth/water variants)
- Replace old 4-condition structure with current 6-condition experimental ladder
- Replace am (Amharic) with es (Spanish) per March 17 language decision
- Benchmarks: MGSM, X-CSQA, XNLI only (removed TyDi QA, MMLU)
- Update directory structure to condition-1-en, condition-2-{zh,es,ur} naming
- Add training setup details (QLoRA, Legesher v0.7.3)
madiedgar changed pull request status to merged