Skip to main content

MLX Local Pipelines

MLX models can be run locally through the MLXPipeline class.

The MLX Community hosts over 150 models, all open source and publicly available on Hugging Face Model Hub a online platform where people can easily collaborate and build ML together.

These can be called from LangChain either through this local pipeline wrapper or by calling their hosted inference endpoints through the MlXPipeline class. For more information on mlx, see the examples repo notebook.

To use, you should have the mlx-lm python package installed, as well as transformers. You can also install huggingface_hub.

%pip install --upgrade --quiet  mlx-lm transformers huggingface_hub

Model Loading

Models can be loaded by specifying the model parameters using the from_model_id method.

from langchain_community.llms.mlx_pipeline import MLXPipeline

pipe = MLXPipeline.from_model_id(
pipeline_kwargs={"max_tokens": 10, "temp": 0.1},

API Reference:

They can also be loaded by passing in an existing transformers pipeline directly

from langchain_community.llms.huggingface_pipeline import HuggingFacePipeline
from mlx_lm import load

model, tokenizer = load("mlx-community/quantized-gemma-2b-it")
pipe = MLXPipeline(model=model, tokenizer=tokenizer)

API Reference:

Create Chain

With the model loaded into memory, you can compose it with a prompt to form a chain.

from langchain_core.prompts import PromptTemplate

template = """Question: {question}

Answer: Let's think step by step."""
prompt = PromptTemplate.from_template(template)

chain = prompt | pipe

question = "What is electroencephalography?"

print(chain.invoke({"question": question}))

API Reference:

Help us out by providing feedback on this documentation page: