EQ Detection Model

A fine-tuned DistilBERT model for detecting emotional intelligence levels in workplace-focused text data.


Model Description

  • Task: Text Classification
  • Model Type: Emotional Intelligence (EQ) Detection
  • Base Model: distilbert-base-uncased
  • Language: English
  • Output Classes: 3 (NEGATIVE / NEUTRAL / POSITIVE)
  • Training Dataset Size: 2,796 workplace communication samples

The model is designed to evaluate emotional regulation, tone, and behavioral intelligence in professional communication.


Label Schema

Label ID Description
NEGATIVE 0 Poor emotional regulation, negative or aggressive expressions
NEUTRAL 1 Emotionally neutral or factual statements
POSITIVE 2 High emotional intelligence and constructive emotional behavior

Training Performance

Epoch Training Loss Validation Loss Accuracy
1 0.188500 0.147850 94.89%
2 0.055100 0.120229 96.39%

Final Validation Accuracy: 96.39%


Training Configuration

  • Framework: Hugging Face Transformers
  • Optimizer: AdamW
  • Batch Size: 16
  • Learning Rate: 2e-5
  • Epochs: 2
  • Max Sequence Length: 128 tokens

Intended Use

This model is intended for:

  • Workplace communication analysis
  • Emotional intelligence assessment
  • HR analytics and employee development
  • Team interaction and behavioral insights

How to Use

Load the Model

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="sreenathsree1578/Eq_funetuned"
)
Downloads last month
11
Safetensors
Model size
67M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support