Yi 6B
About
The Yi 6B large language model, developed by 01.AI, is a bilingual model excelling in English and Chinese, boasting 6 billion parameters. It features a standard Transformer architecture with advanced elements like Grouped-Query Attention and SwiGLU activation. Pre-trained on 3 terabytes of multilingual data, Yi 6B shows strong performance in language tasks such as text generation, translation, summarization, question answering, and code generation. While versatile, it shares common limitations of large models, including hallucination and non-determinism, with a default context window of 4K tokens. It remains well-suited for personal or academic use, offering both a strong performance and efficiency.
Capabilities
MultimodalFunction CallingTool UseJSON Mode
Providers(3)
| Provider | Input (per 1M) | Output (per 1M) | Type | |
|---|---|---|---|---|
| Alibaba Cloud PAI-EAS | — | — | Serverless | |
| Replicate API | — | — | Serverless | |
| Fireworks AI Platform | — | — | Provisioned |
Specifications
FamilyYi (2023/11)
Released2023-11-02
Parameters6B
ArchitectureDecoder Only
Specializationgeneral