Open-Source Turba ML Stack for Morocco Fertilizer Advice Ships 44,096-Site Dataset
open-turba · hf · 2026-10-08
Turba releases an open-source three-layer ML stack for site-specific fertilizer recommendation in Morocco: turba-client (programmatic access to N/P₂O₅/K₂O recommendation workflows), turba-data (versioned snapshots), and turba-models (loadable ML surrogates of recommendation outputs). The dataset is built from 44,096 unique ESA WorldCereal locations, expanded to 132,017 crop-location requests across 10 regions, 66 provinces and 1,149 communes with 22 variables. Nine regression families were benchmarked under a fixed 80/20 protocol; five best crop-specific models are packaged. The task is framed as recommendation-function emulation, enabling reproducible validation, uncertainty estimation and field-trial comparison.
More from Research
- BeFOND: encoder-free sparse coding matches Gemma Scope with 1000x fewer training tokens — gottapatchemall · 2026-10-08
- Adding Motion Sense to VLAs With Frozen SAM 2 Lets Robots Catch Moving Bottles — nikhilaravi · 2026-10-08
- Meta, UW et al. open-source Context Language Models that natively manage their own context — RulinShao · 2026-10-08
- New paper examines when LLMs generalize factual knowledge across languages — tallinzen · 2026-10-08
- Nat Lambert offers open research on distillation's impact on frontier progress, pending funding — natolambert · 2026-10-08
- Viral square packing debate: "ugly" corners may hint at deeper structure, author argues — ctjlewis · 2026-10-08