The community members are discussing the use of DBRX (a language model) with LlamaIndex, a library for building AI agents. They are interested in how DBRX performs compared to OpenAI models, which were previously the focus of LlamaIndex and LangChain. The comments suggest that LlamaIndex provides easy abstractions to work with custom language models like DBRX, and that you can simply pass the DBRX language model to LlamaIndex to use it. There is no explicitly marked answer, but the community members provide helpful information on how to integrate DBRX with LlamaIndex.
Has anyone built any Agents with DBRX in LlamaIndex? How does it perform? Are there any considerations or issues with the LI abstractions (I understand for a while most of LangChain and LlamaIndex were tilted towards OpenAI)
I'm familiar with the abstraction part - I just wasn't sure if the abstractions were generalized enough to support DBRX like it does OpenAI for example.
#define the llm here
llm = Your dbrx llm
# Now make the llm global
from llama_index.core import Settings
Settings.llm = llm
# or pass the llm inside every func
query_engine = index.as_query_engine(llm=llm,..)
