mishreyagupta/ForestryIntelligence

ForestryIntelligence

A DistilBERT-based text classifier fine-tuned to categorize forestry and lumber-industry report excerpts into operational risk categories.

Model Description

• —Base model: distilbert-base-uncased
• —Task: Multi-class text classification
• —Labels: generaloperations, harvestcompliance, pestdisease, sustainability, wildfirerisk
• —Fine-tuned by: mishreyagupta

Intended Use

Triaging/classifying short excerpts from forestry inspection reports, compliance documents, and field notes into risk categories for downstream routing (e.g., flagging wildfire-risk passages for review).

This model performs classification only. For open-ended analysis or summarization of forestry reports, pair it with a retrieval-augmented generation (RAG) pipeline — see the companion notebook section.

Training Data

Fine-tuned on a labeled forestry text corpus spanning wildfire risk, pest/disease, harvest compliance, sustainability, and general operations categories.

Limitations

• —Trained on a small, partly synthetic dataset — validate on your organization's real documents before production use.
• —English-language text only.
• —Not designed for generation, summarization, or open-ended Q&A.

Evaluation

See the training notebook for full metrics (accuracy, weighted F1) on the held-out validation split.