CEC Assistant - Llama 3.1 Fine-Tuned Model

Overview

CEC Assistant is a domain-specific language model fine-tuned on College of Engineering Chengannur (CEC) related information. The model is designed to answer questions about academics, departments, admissions, placements, campus facilities, student activities, and other college-related topics.

Model Details

• Base Model: Meta Llama 3.1
• Task: Causal Language Modeling (CLM)
• Framework: Hugging Face Transformers
• Fine-Tuning Approach: Supervised Fine-Tuning
• Language: English

Training Process

The model was fine-tuned using custom college-specific data and trained using the Hugging Face Transformers ecosystem.

Training workflow included:

• Data collection and preprocessing
• Text cleaning and tokenization
• Dataset splitting for training and evaluation
• Causal Language Model fine-tuning
• Model checkpointing and validation
• Hugging Face model publishing

Technologies Used

• Transformers
• Datasets
• Trainer API
• AutoModelForCausalLM
• Llama Tokenizer
• PyTorch

Capabilities

The model can assist with:

• College information queries
• Academic program details
• Department-related information
• Admission-related questions
• Campus facilities and activities
• General institutional knowledge

Limitations

• Responses may not always reflect the latest institutional updates.
• The model may generate inaccurate information in some cases.
• Official college sources should be consulted for critical decisions.

License

This model inherits the licensing requirements of the underlying Llama 3.1 model. Users must comply with Meta's Llama license terms.

Disclaimer

This is an independent educational and research project and is not an official information source of College of Engineering Chengannur.