budecosystem/code-millenials-13b

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Jan 2, 2024License:llama2Architecture:Transformer0.0K Open Weights Cold

The budecosystem/code-millenials-13b is a 13 billion parameter language model developed by Bud Ecosystem, specifically fine-tuned for code generation tasks. It achieves a HumanEval pass@1 score of 76.21, demonstrating strong performance in coding benchmarks. This model is optimized for generating high-quality code and is suitable for developers seeking robust code completion and generation capabilities.

Loading preview...

Overview

The budecosystem/code-millenials-13b is a 13 billion parameter model from Bud Ecosystem, engineered for advanced code generation. This model is part of a series that aims to set new benchmarks in code-specific language models. It was trained on 8 A100 80GB GPUs for approximately 15 hours, utilizing a per_device_train_batch_size of 2 and a learning rate of 2e-5.

Key Capabilities

  • High Performance Code Generation: Achieves a HumanEval pass@1 score of 76.21, positioning it as a highly capable model for various coding challenges.
  • Strong Benchmark Results: Outperforms many existing models in its size class on the HumanEval benchmark, demonstrating its effectiveness in generating correct and efficient code.
  • Optimized for Code Tasks: Specifically fine-tuned to excel in code generation, making it a specialized tool for developers.

Good For

  • Code Completion and Generation: Ideal for developers needing assistance with writing new code or completing existing code snippets.
  • Automated Programming Tasks: Suitable for integrating into development workflows to automate parts of the coding process.
  • Research and Development: Provides a strong baseline for further research into code-centric large language models.