File size: 3,002 Bytes
1959cc0 e97ed61 202e8d9 e97ed61 7cc2af3 06e222e 7cc2af3 |
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 |
---
license: apache-2.0
datasets:
- bitext/Bitext-customer-support-llm-chatbot-training-dataset
language:
- en
base_model:
- unsloth/llama-3-8b-bnb-4bit
pipeline_tag: text-generation
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- gguf
- Customer-Support-Bot
---
# Customer Support Chatbot with LLaMA 3.1
> An end-to-end customer support chatbot solution powered by fine-tuned LLaMA 3.1 8B model, deployed using Flask, Docker, and AWS ECS.
## Overview
This project implements a sophisticated customer support chatbot leveraging the LLaMA 3.1 8B model fine-tuned on customer support conversations. The solution uses LoRA fine-tuning and various quantization techniques for optimized inference, deployed as a containerized application on AWS ECS with Fargate.
## Features
- **Fine-tuned LLaMA 3.1 Model**: Customized for customer support using the [Bitext customer support dataset](https://huggingface.co/datasets/bitext/Bitext-customer-support-llm-chatbot-training-dataset)
- **Optimized Inference**: Implements 4-bit, 8-bit, and 16-bit quantization
- **Containerized Deployment**: Docker-based deployment for consistency and scalability
- **Cloud Infrastructure**: Hosted on AWS ECS with Fargate for serverless container management
- **CI/CD Pipeline**: Automated deployment using AWS CodePipeline
- **Monitoring**: Comprehensive logging and monitoring via AWS CloudWatch
## Model Details
The fine-tuned model is hosted on Hugging Face:
- Model Repository: [praneethposina/customer_support_bot](https://huggingface.co/praneethposina/customer_support_bot)
- Github Repository: [github.com/praneethposina/Customer_Support_Chatbot](https://github.com/praneethposina/Customer_Support_Chatbot)
- Base Model: LLaMA 3.1 8B
- Training Dataset: Bitext Customer Support Dataset
- Optimization: LoRA fine-tuning with quantization
## Tech Stack
- **Backend**: Flask API
- **Model Serving**: Ollama
- **Containerization**: Docker
- **Cloud Services**:
- AWS ECS (Fargate)
- AWS CodePipeline
- AWS CloudWatch
- **Model Training**: LoRA, Quantization
## Screenshots
### Chatbot Interface
### AWS CloudWatch Monitoring
### Docker Logs
<img width="1270" alt="Docker ss" src="https://github.com/user-attachments/assets/a72d1c35-8203-4a05-b944-743ea6c0a6b8" />
<img width="1268" alt="Docker ss2" src="https://github.com/user-attachments/assets/f1b0c0b1-2aad-462c-adf2-7a7ea9047a1a" />
## AWS Deployment
1. Push Docker image to Amazon ECR
2. Configure AWS ECS Task Definition
3. Set up AWS CodePipeline for CI/CD
4. Configure CloudWatch monitoring
# Uploaded model
- **Developed by:** praneethposina
- **License:** apache-2.0
- **Finetuned from model :** unsloth/llama-3-8b-bnb-4bit |