language: en
license: apache-2.0
tags:
- forestry
- lumber-industry
- regulatory-compliance
- text-classification
- distilbert
datasets:
- custom-forestry-corpus
metrics:
- accuracy
- f1
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: general_operations, harvest_compliance, pest_disease, sustainability, wildfire_risk
- 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.