Models

← All models

Code LLaMA 7B Instruct Solidity

AlfredPros · 2023-10-08 · 6.7B parameters

Code LLaMA 7B Instruct Solidity is AlfredPros' fine-tune of Code Llama 7B Instruct for generating Solidity smart contracts. Its model card reports one epoch of 4-bit QLoRA fine-tuning on 6,003 GPT-generated instruction-and-Solidity-code pairs. The checkpoint retains the base model's 16,384-token configuration and its documented inference support for contexts up to 100,000 tokens.

Benchmark scores