Yi
A bilingual local LLM family supports English and Chinese chat and fine-tuning, with quantized models compatible with consumer GPUs.
Yi is a family of language models from 01.AI for developers and researchers who want to run English and Chinese AI on their own hardware. It includes chat models for conversation and base models you can adapt to specific tasks.
- Bilingual chat: Yi-6B-Chat and Yi-34B-Chat support English and Chinese, with varied responses suited to creative work.
- Fine-tuning: You can adapt the base models to your needs; Yi-9B has a particular emphasis on coding and math within the family.
- Long context: Extended-context base models accept more text in a single prompt for tasks that need substantial source material.
- Consumer GPU support: Quantized chat variants use AWQ at 4-bit and GPTQ at 8-bit, with NVIDIA RTX 3090 and 4090 cards listed as compatible examples.
Yi supports local deployment through llama.cpp and Docker. It shares Llama's architecture and works with tools in that ecosystem, though 01.AI trained its models independently without using Llama's weights. The project uses the Apache 2.0 license.
The chat models can give factually incorrect answers, and regenerating a response can produce inconsistent results. Errors can accumulate over a longer conversation.
Logotyper
