README.md

metadata

license: mit
datasets:
  - prabinpanta0/Rep00Zon
language:
  - en
metrics:
  - accuracy
pipeline_tag: question-answering
tags:
  - general_knowledge
  - Question_Answers

ZenGQ - BERT for Question Answering

This is a fine-tuned BERT model for question answering tasks, trained on a custom dataset.

Model Details

• Model: BERT-base-uncased
• Task: Question Answering
• Dataset: Rep00Zon

Usage

Load the model

from transformers import AutoTokenizer, AutoModelForQuestionAnswering, pipeline

# Load the tokenizer and model from Hugging Face
tokenizer = AutoTokenizer.from_pretrained("prabinpanta0/ZenGQ")
model = AutoModelForQuestionAnswering.from_pretrained("prabinpanta0/ZenGQ")

# Create a pipeline for question answering
qa_pipeline = pipeline("question-answering", model=model, tokenizer=tokenizer)

# Define your context and questions
contexts = ["Berlin is the capital of Germany.",
          "Paris is the capital of France.",
          "Madrid is the capital of Spain."]
questions = [
    "What is the capital of Germany?",
    "Which city is the capital of France?",
    "What is the capital of Spain?"
]

# Get answers
for context, question in zip(contexts, questions):
    result = qa_pipeline(question=question, context=context)
    print(f"Question: {question}")
    print(f"Answer: {result['answer']}\n")

Training Details

• Epochs: 3
• Training Loss: 2.050335, 1.345047, 1.204442

Token

text = "Berlin is the capital of Germany. Paris is the capital of France. Madrid is the capital of Spain."
tokens = tokenizer.tokenize(text)
print(tokens)

Output: ['berlin', 'is', 'the', 'capital', 'of', 'germany', '.', 'paris', 'is', 'the', 'capital', 'of', 'france', '.', 'madrid', 'is', 'the', 'capital', 'of', 'spain', '.']

Dataset

The model was trained on the Rep00Zon dataset.

License

This model is licensed under the MIT License.