NatLabRockies/PVI-issue-classification

BERT Fine-tuned on PV-Inspection Reports

This model, developed at NLR, is a fine-tuned version of bert-base-uncased on multiple PV inspection reports in the US.

Developed by:

• —Jeff Cook, Strategic Energy Analysis Center
• —Guilherme Castelão, Strategic Energy Analysis Center
• —Danny Chang, Strategic Energy Analysis Center
• —Sertac Akar, Accelerated Deployment and Decision Support Center
• —James Elsworth, Resilient Infrastructure and Security Center

Model Description

• —Base model: bert-base-uncased
• —Task: classify type of issue reported
• —Training data: PV inspection reports

Intended Uses

Classify the reason(s) of failure from PV installations inspections reports.

How to Use

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("NatLabRockies/pvi-issue-classification")
tokenizer = AutoTokenizer.from_pretrained("NatLabRockies/pvi-issue-classification")

# Example usage
inputs = tokenizer("NEC 110.26 specifies working space clearance about electrical equipment.", return_tensors="pt")
outputs = model(**inputs)

logits = outputs.logits
predicted_class_id = logits.argmax().item()
predicted_class = model.config.id2label[predicted_class_id]
print(f"Predicted issue: {predicted_class}")

Training Details

• —Training epochs: [number]
• —Learning rate: [value]
• —Batch size: [value]

Evaluation Results

[Add your metrics: accuracy, F1, etc.]