L

Llama 3 8B Instruct Coder

Developed by rombodawg
Based on Meta's llama-3-8b-instruct model, trained on a 65k code feedback dataset and an additional 150k code feedback instruction combination dataset, focusing on code generation and feedback tasks.
Downloads 81
Release Time : 5/4/2024

Model Overview

This model is an optimized code generation and feedback model, trained using the Qalore method, significantly reducing VRAM requirements and suitable for code-related tasks.

Model Features

Qalore Training Method
Combines QLoRA and Galore methods, significantly reducing VRAM requirements, requiring only 14.5GB VRAM to load an 8B parameter model.
Efficient Training
Completed training in 130 hours on an RTX A4000 16GB GPU, costing less than $20.
Code Optimization
Trained on a 65k code feedback dataset and an additional 150k code feedback instruction combination dataset, focusing on code generation and feedback tasks.

Model Capabilities

Code Generation
Code Feedback
Instruction Understanding

Use Cases

Software Development
Code Autocompletion
Generates code snippets based on context to improve development efficiency.
Code Error Feedback
Provides code improvement suggestions and error correction solutions.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase