mishreyagupta's picture
Upload ForestryIntelligence: DistilBERT fine-tuned for forestry risk classification
92b76a5 verified
|
Raw
History Blame Contribute Delete
1.56 kB
metadata
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.