Files
teste8/README.md
2025-10-28 10:19:23 +00:00

89 lines
129 B
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
license: mit
datasets:
- sweatSmile/neet-biology-qa
language:
- en
base_model:
- distilbert/distilbert-base-uncased
pipeline_tag: question-answering
library_name: transformers
tags:
- neet
- biology
- exam
- bio
---
DistilBERT NEET Biology MCQ Classifier (NEET_BioBERT)
This model is a fine-tuned version of DistilBERT (base uncased) specifically trained to classify the correct option for NEET-style multiple-choice biology questions. It selects the best answer among four choices (A, B, C, D).
-------------------------------------------------------------------------
Training Data
Source: sweatSmile / NEET Biology QA Dataset
Domain: NEET (Undergraduate Medical Entrance Exam) Biology
Format: Each question has 4 options with one correct answer
Dataset Size: 793 questions
Split: 80% train / 20% validation
-------------------------------------------------------------------------
Training Configuration
Base Model: distilbert-base-uncased
Epochs: 10
Batch Size: 4
Learning Rate: 5e-5
Weight Decay: 0.01
Task Type: Multiple Choice Classification
-------------------------------------------------------------------------
Results
Validation Accuracy 72.96% (~73%)
Final Training Loss ~0.35
-------------------------------------------------------------------------
Limitations
Trained on a relatively small dataset (793 questions).
Limited to NEET-level biology content; not suitable for physics or chemistry.
Does not support:
Assertion-reasoning questions
Diagram-based questions
Paragraph/Case study type questions
-------------------------------------------------------------------------
Intended Use
Educational Research
AI-powered NEET Biology assistants
MCQ practice evaluation
Baseline model for future fine-tuning with larger datasets
-------------------------------------------------------------------------
NOTE:
Not recommended as a final exam-ready solution without further fine-tuning and validation.
-------------------------------------------------------------------------
License: MIT