P

Phind CodeLlama 34B Python V1

Developed by Phind
A large language model fine-tuned based on CodeLlama-34B-Python, achieving 69.5% pass@1 on HumanEval, surpassing GPT-4's performance
Downloads 878
Release Time : 8/25/2023

Model Overview

A 34B-parameter large model optimized for code generation, fine-tuned with 80,000 high-quality programming problems, supporting Python and other programming languages

Model Features

Leading HumanEval Performance
Achieves 69.5% pass@1 on standard code generation benchmarks, surpassing GPT-4's 67% performance
Professional Code Fine-tuning
Instruction fine-tuning using 80,000 high-quality programming problem solutions to optimize code generation capabilities
Efficient Training Technology
Utilizes DeepSpeed ZeRO 3 and Flash Attention 2 technology, completing training in just 3 hours with 32 A100-80GB GPUs

Model Capabilities

Code Generation
Programming Problem Solving
Algorithm Implementation
Data Structure Implementation

Use Cases

Development Assistance
Automated Code Generation
Generates executable code based on natural language descriptions
69.5% accuracy on HumanEval benchmark
Programming Education
Generates code examples and exercise solutions for teaching purposes
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase