Shaer: Controlled Arabic Poetry Generation with Meter Subform and Semantic Conditioning
We present Shaer, a controllable Classical Arabic poetry generation framework jointly conditioned on natural-language descriptions, meter subforms, and target hemistich counts.
ProofPaper ↗
Key points
- Classical Arabic poetry generation requires simultaneously satisfying semantic, linguistic, and fine-grained prosodic constraints.
- Existing systems typically control broad poetic attributes but do not jointly model semantic intent, meter subform, and poem length.
- To support this task, we construct an enriched corpus of 116,032 classical Arabic poems derived from Ashaar, containing normalized meter-subform labels and automatically generated, validated semantic descriptions.
- We then adapt Yehia-7B using QLoRA-based supervised fine-tuning with a completion-only objective.
Sources (1)
- [1]Shaer: Controlled Arabic Poetry Generation with Meter Subform and Semantic ConditioningarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 09:44 AM
We present Shaer, a controllable Classical Arabic poetry generation framework jointly conditioned on natural-language descriptions, meter subforms, and target hemistich counts.
Classical Arabic poetry generation requires simultaneously satisfying semantic, linguistic, and fine-grained prosodic constraints.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 7, 2026When Routing Reveals Membership: Privacy Leakage from MoE Router Telemetry
- Oct 6, 2026Are Parameter-Efficient Fine-tuning Methods Really Different?
- Oct 6, 2026MemFLoRA: Memory-Floor LoRA for CNN Adaptation at the Edge
- Oct 6, 2026AutodidactWAM: Cross-Modal Self-Distillation from Generated Video to Robot Actions
- Oct 6, 2026A Riemannian Geometry for Low-rank Adaptation
- Sep 29, 2026How Diffusion Controller unifies and simplifies AI image generation