Tutorial - Deploy Phi-3-mini-128k-instruct using Inferless

Microsoft has introduced Phi-3-mini-128k-instruct, a compact yet powerful model designed for instruction-following tasks. This model is part of the Phi-3 family, known for its efficiency and high performance. The Phi-3 Mini-128K-Instruct exhibited robust, state-of-the-art performance among models with fewer than 13 billion parameters.

TL;DR:

• Deployment of Phi-3-mini-128k-instruct model using Transformers.
• You can expect an average latency of 18.42 sec and throughput of 24.71 tokens per second. This setup has an average cold start time of 7.82 sec.
• Dependencies defined in inferless-runtime-config.yaml.
• GitHub/GitLab template creation with app.py, inferless-runtime-config.yaml and inferless.yaml.
• Model class in app.py with initialize, infer, and finalize functions.
• Custom runtime creation with necessary system and Python packages.
• Model import via GitHub with input_schema.py file.
• Recommended GPU: NVIDIA A100 for optimal performance.
• Custom runtime selection in advanced configuration.
• Final review and deployment on the Inferless platform.

Fork the Repository

Get started by forking the repository. You can do this by clicking on the fork button in the top right corner of the repository page.

This will create a copy of the repository in your own GitHub account, allowing you to make changes and customize it according to your needs.

Create a Custom Runtime in Inferless

To access the custom runtime window in Inferless, simply navigate to the sidebar and click on the Create new Runtime button. A pop-up will appear.

Next, provide a suitable name for your custom runtime and proceed by uploading the inferless-runtime.yaml file given above. Finally, ensure you save your changes by clicking on the save button.

Import the Model in Inferless

Log in to your inferless account, select the workspace you want the model to be imported into and click the Add Model button.

Select the PyTorch as framework and choose Repo(custom code) as your model source and select your provider, and use the forked repo URL as the Model URL.

Enter all the required details to Import your model. Refer this link for more information on model import.

---

Customizing the Code

Open the app.py file. This contains the main code for inference. It has three main functions, initialize, infer and finalize.

Initialize - This function is executed during the cold start and is used to initialize the model. If you have any custom configurations or settings that need to be applied during the initialization, make sure to add them in this function.

Infer - This function is where the inference happens. The argument to this function inputs, is a dictionary containing all the input parameters. The keys are the same as the name given in inputs. Refer to input for more.

def infer(self, inputs):
    prompt = inputs["prompt"]

Finalize - This function is used to perform any cleanup activity for example you can unload the model from the gpu by setting self.pipe = None.

For more information refer to the Inferless docs.