Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
We present RIFA, a family of efficient small open-weight language models designed for practical bilingual text generation. The series consists of five models ranging from 0.5B to 3B parameters: Rifa- Nano (0.5B), RIFA-CODE (0.6B), RIFA-Edge (0.6B), RIFA-FLASH (1.7B), and RIFA-PRO (3B). Each model is fine-tuned for conversational use in both English and Bangla, with specialized roles covering ultra-lightweight deployment, code assistance, edge devices, balanced performance, and higher-quality generation. All models are released under the Apache 2.0 license with full weights, multiple GGUF quantizations, and clear documentation. RIFA prioritizes low resource requirements while remaining useful for real-world text generation tasks on constrained hardware. We describe the model lineup, design principles, and public release to support further research and practical adoption of efficient small language models.
No comments yet — start the discussion below.