File size: 3,497 Bytes
8fad492
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
# Model Details

**Model Name:** Work Ethic Analysis Model  
**Base Model:** distilbert-base-uncased  
**Dataset:** yelp_review_full  

**Training Device:** CUDA (GPU)  

---

## Dataset Information

**Dataset Structure:**  
DatasetDict({  
train: Dataset({  
features: ['employee_feedback', 'ethic_category'],  
num_rows: 50,000  
})  
validation: Dataset({  
features: ['employee_feedback', 'ethic_category'],  
num_rows: 20,000  
})  
})  

**Available Splits:**  
- **Train:** 15,000 examples  
- **Validation:** 2,000 examples  

**Feature Representation:**  
- **employee_feedback:** Textual feedback from employees (e.g., "John consistently meets deadlines and takes initiative.")  
- **ethic_category:** Classified work ethic type (e.g., "Strong Initiative")  

---

## Training Details

**Training Process:**  
- Fine-tuned for 3 epochs  
- Loss reduced progressively across epochs  

**Hyperparameters:**  
- Epochs: 3  
- Learning Rate: 3e-5  
- Batch Size: 8  
- Weight Decay: 0.01  
- Mixed Precision: FP16  

**Performance Metrics:**  
- Accuracy: 92.3%  

---

## Inference Example

```python
import torch
from transformers import DistilBertTokenizer, DistilBertForSequenceClassification

def load_model(model_path):
    tokenizer = DistilBertTokenizer.from_pretrained(model_path)
    model = DistilBertForSequenceClassification.from_pretrained(model_path).half()
    model.eval()
    return model, tokenizer

def classify_ethic(feedback, model, tokenizer, device="cuda"):
    inputs = tokenizer(
        feedback,
        max_length=256,
        padding="max_length",
        truncation=True,
        return_tensors="pt"
    ).to(device)
    outputs = model(**inputs)
    predicted_class = torch.argmax(outputs.logits, dim=1).item()
    return predicted_class

# Example usage
if __name__ == "__main__":
    model_path = "your-username/work-ethic-analysis"  # Replace with your HF repo
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model, tokenizer = load_model(model_path)
    model.to(device)

    feedback = "John consistently meets deadlines and takes initiative."
    category = classify_ethic(feedback, model, tokenizer, device)
    print(f"Feedback: {feedback}")
    print(f"Predicted Work Ethic Category: {category}")
```

**Expected Output:**  
```plaintext
Feedback: John consistently meets deadlines and takes initiative.
Predicted Work Ethic Category: Strong Initiative
```

---

# Use Case: Work Ethic Analysis Model

## **Overview**

The **Work Ethic Analysis Model**, built on **DistilBERT-base-uncased**, is designed to classify employee feedback into predefined work ethic categories. This helps HR teams and management analyze employee dedication, responsibility, and productivity.

## **Key Applications**

- **Performance Assessment:** Identify patterns in employee feedback for objective performance reviews.
- **Employee Recognition:** Highlight employees demonstrating strong work ethics for rewards and promotions.
- **Early Warning System:** Detect negative trends in work ethic and take proactive measures.
- **Leadership and Training Enhancement:** Use feedback analysis to improve training programs for employees and managers.

## **Benefits**

- **Scalability:** Can process thousands of employee feedback entries in minutes.
- **Unbiased Evaluation:** AI-driven classification removes subjective bias from evaluations.
- **Actionable Insights:** Helps HR teams make data-driven decisions for workforce improvement.

---