OpenCoder: The Open Cookbook for Top-Tier Code Large Language Models
Paper
•
2411.04905
•
Published
•
127
OpenCoder is an open and reproducible code LLM family, featuring 1.5B and 8B base and chat models that support both English and Chinese languages. Built from scratch, OpenCoder is pretrained on 2.5 trillion tokens, composed of 90% raw code and 10% code-related web data. It undergoes supervised fine-tuning (SFT) with over 4.5 million high-quality examples, achieving performance on par with top-tier code LLMs
| No | Variant | Cortex CLI command |
|---|---|---|
| 1 | Opencoder-8b | cortex run opencoder:8b |
cortexhub/opencoder
cortex run opencoder
2-bit
3-bit
4-bit
5-bit
6-bit
8-bit