AgriJev improves agricultural candidate scoring with efficient models
AgriJev: Efficient Agricultural Candidate Scoring with Multi-Scale Adapted Models
Computation and Language
Summary
Deciding the best choice in agriculture often involves understanding complex text and multiple options. The authors developed AgriJev, a set of adapted AI models tailored to agricultural text and questions. These models quickly assign scores to different answer choices all at once, making decision-making faster and more accurate. Compared to earlier models, AgriJev is both more accurate and faster when tested on many types of agricultural questions.
What this means in practice
- •For agriculture analytics teams: Score and select the best agricultural decisions from textual data more quickly and accurately using multi-scale adapted models.
- •For legal compliance officers: Evaluate agricultural regulatory question options with improved accuracy and speed in decision-making systems.
Authors
Xuechen Li
Abstract
Agricultural systems make decisions over heterogeneous entity types, textual judgments, and knowledge-question options. Generating a textual answer adds decoding and extraction steps to decisions that ultimately require a candidate choice. We introduce AgriJev, a family of adapted Qwen3.5 models at 0.8B, 2B, 4B, and 9B. A shared interface conditions on agricultural text, a judgment question, and candidate definitions, returning scores and relative probabilities in one forward call. We prepare source-aware training records and frozen evaluations from existing agricultural resources, retaining original groups and annotation provenance. Comparisons include original Qwen3.5 models and matched agricultural SFT models, both evaluated through native answer generation. Across 17 external test views, AgriJev-9B reaches 75.44% mean accuracy, exceeding the original and SFT generators by 5.89 and 3.79 percentage points. It outperforms the original generator on 13 views and narrows the mean gap to hosted Jev 1.13.0. On a fixed A100 workload, its median scoring latency is approximately half that of same-scale answer generation. Task-dependent gains and failures characterize a practical choice among agricultural model weights and inference contracts.