Phi-3-Vision-MLX

Phi-3.5 for Mac: Locally-run Vision and Language Models for Apple Silicon

Stars: 186

Visit

Phi-3-MLX is a versatile AI framework that leverages both the Phi-3-Vision multimodal model and the Phi-3-Mini-128K language model optimized for Apple Silicon using the MLX framework. It provides an easy-to-use interface for a wide range of AI tasks, from advanced text generation to visual question answering and code execution. The project features support for batched generation, flexible agent system, custom toolchains, model quantization, LoRA fine-tuning capabilities, and API integration for extended functionality.

README:

Phi-3-MLX: Language and Vision Models for Apple Silicon

Phi-3-MLX is a versatile AI framework that leverages both the Phi-3-Vision multimodal model and the Phi-3-Mini-128K language model, optimized for Apple Silicon using the MLX framework. This project provides an easy-to-use interface for a wide range of AI tasks, from advanced text generation to visual question answering and code execution.

Features

Integration with Phi-3.5-vision model
Support for the Phi-3.5-mini model
Optimized performance on Apple Silicon using MLX
Batched generation for processing multiple prompts
Flexible agent system for various AI tasks
Custom toolchains for specialized workflows
Model quantization for improved efficiency
LoRA fine-tuning capabilities
API integration for extended functionality (e.g., image generation, text-to-speech)

Minimum Requirements

Phi-3-MLX is designed to run on Apple Silicon Macs. The minimum requirements are:

Apple Silicon Mac (M1, M2, or later)
8GB RAM (with quantization using quantize_model=True option)

For optimal performance, especially when working with larger models or datasets, we recommend using a Mac with 16GB RAM or more.

Quick Start

Install and launch Phi-3-MLX from command line:

# Quick install (note: PyPI version may not always be up to date)
pip install phi-3-vision-mlx
phi3v

# For the latest version, you can install directly from the repository:
# git clone https://github.com/JosefAlbers/Phi-3-Vision-MLX.git
# cd Phi-3-Vision-MLX
# pip install -e .

To use the library in a Python script:

from phi_3_vision_mlx import generate

1. Core Functionalities

Visual Question Answering

generate('What is shown in this image?', 'https://collectionapi.metmuseum.org/api/collection/v1/iiif/344291/725918/main-image')

Model and Cache Quantization

# Model quantization
generate("Describe the water cycle.", quantize_model=True)

# Cache quantization
generate("Explain quantum computing.", quantize_cache=True)

Batch Text Generation

# A list of prompts for batch generation
prompts = [
    "Write a haiku about spring.",
    "Explain the theory of relativity.",
    "Describe a futuristic city."
]
# Generate responses using Phi-3-Vision (multimodal model)
generate(prompts, max_tokens=100)

# Generate responses using Phi-3-Mini-128K (language-only model)
generate(prompts, max_tokens=100, blind_model=True)

Constrained Beam Decoding

from phi_3_vision_mlx import constrain

# Use constrain for structured generation (e.g., code, function calls, multiple-choice)
prompts = [
    "A 20-year-old woman presents with menorrhagia for the past several years. She says that her menses “have always been heavy”, and she has experienced easy bruising for as long as she can remember. Family history is significant for her mother, who had similar problems with bruising easily. The patient's vital signs include: heart rate 98/min, respiratory rate 14/min, temperature 36.1°C (96.9°F), and blood pressure 110/87 mm Hg. Physical examination is unremarkable. Laboratory tests show the following: platelet count 200,000/mm3, PT 12 seconds, and PTT 43 seconds. Which of the following is the most likely cause of this patient’s symptoms? A: Factor V Leiden B: Hemophilia A C: Lupus anticoagulant D: Protein C deficiency E: Von Willebrand disease",
    "A 25-year-old primigravida presents to her physician for a routine prenatal visit. She is at 34 weeks gestation, as confirmed by an ultrasound examination. She has no complaints, but notes that the new shoes she bought 2 weeks ago do not fit anymore. The course of her pregnancy has been uneventful and she has been compliant with the recommended prenatal care. Her medical history is unremarkable. She has a 15-pound weight gain since the last visit 3 weeks ago. Her vital signs are as follows: blood pressure, 148/90 mm Hg; heart rate, 88/min; respiratory rate, 16/min; and temperature, 36.6℃ (97.9℉). The blood pressure on repeat assessment 4 hours later is 151/90 mm Hg. The fetal heart rate is 151/min. The physical examination is significant for 2+ pitting edema of the lower extremity. Which of the following tests o should confirm the probable condition of this patient? A: Bilirubin assessment B: Coagulation studies C: Hematocrit assessment D: Leukocyte count with differential E: 24-hour urine protein"
]

# Define constraints for the generated text
constraints = [(0, '\nThe'), (100, ' The correct answer is'), (1, 'X.')]

# Apply constrained beam decoding
results = constrain(prompts, constraints, blind_model=True, quantize_model=True, use_beam=True)

Choosing From Options

from phi_3_vision_mlx import choose

# Select best option from choices for given prompts
prompts = [
    "What is the largest planet in our solar system? A: Earth B: Mars C: Jupiter D: Saturn",
    "Which element has the chemical symbol 'O'? A: Osmium B: Oxygen C: Gold D: Silver"
]

# For multiple-choice or decision-making tasks
choose(prompts, choices='ABCDE')

LoRA Fine-tuning

from phi_3_vision_mlx import train_lora, test_lora

# Train a LoRA adapter
train_lora(
    lora_layers=5,  # Number of layers to apply LoRA
    lora_rank=16,   # Rank of the LoRA adaptation
    epochs=10,      # Number of training epochs
    lr=1e-4,        # Learning rate
    warmup=0.5,     # Fraction of steps for learning rate warmup
    dataset_path="JosefAlbers/akemiH_MedQA_Reason"
)

# Generate text using the trained LoRA adapter
generate("Describe the potential applications of CRISPR gene editing in medicine.",
    blind_model=True,
    quantize_model=True,
    use_adapter=True)

# Test the performance of the trained LoRA adapter
test_lora()

2. Agent Interactions

Multi-turn Conversation

from phi_3_vision_mlx import Agent

# Create an instance of the Agent
agent = Agent()

# First interaction: Analyze an image
agent('Analyze this image and describe the architectural style:', 'https://images.metmuseum.org/CRDImages/rl/original/DP-19531-075.jpg')

# Second interaction: Follow-up question
agent('What historical period does this architecture likely belong to?')

# End conversation, clear memory for new interaction
agent.end()

Generative Feedback Loop

# Ask the agent to generate and execute code to create a plot
agent('Plot a Lissajous Curve.')

# Ask the agent to modify the generated code and create a new plot
agent('Modify the code to plot 3:4 frequency')
agent.end()

External API Tool Use

# Request the agent to generate an image
agent('Draw "A perfectly red apple, 32k HDR, studio lighting"')
agent.end()

# Request the agent to convert text to speech
agent('Speak "People say nothing is impossible, but I do nothing every day."')
agent.end()

3. Custom Toolchains

In-Context Learning Agent

from phi_3_vision_mlx import add_text

# Define the toolchain as a string
toolchain = """
    prompt = add_text(prompt)
    responses = generate(prompt, images)
    """

# Create an Agent instance with the custom toolchain
agent = Agent(toolchain, early_stop=100)

# Run the agent
agent('How to inspect API endpoints? @https://raw.githubusercontent.com/gradio-app/gradio/main/guides/08_gradio-clients-and-lite/01_getting-started-with-the-python-client.md')

Retrieval Augmented Coding Agent

from phi_3_vision_mlx import VDB
import datasets

# Simulate user input
user_input = 'Comparison of Sortino Ratio for Bitcoin and Ethereum.'

# Create a custom RAG tool
def rag(prompt, repo_id="JosefAlbers/sharegpt_python_mlx", n_topk=1):
    ds = datasets.load_dataset(repo_id, split='train')
    vdb = VDB(ds)
    context = vdb(prompt, n_topk)[0][0]
    return f'{context}\n<|end|>\n<|user|>\nPlot: {prompt}'

# Define the toolchain
toolchain_plot = """
    prompt = rag(prompt)
    responses = generate(prompt, images)
    files = execute(responses, step)
    """

# Create an Agent instance with the RAG toolchain
agent = Agent(toolchain_plot, False)

# Run the agent with the user input
_, images = agent(user_input)

Multi-Agent Interaction

# Continued from Example 2 above
agent_writer = Agent(early_stop=100)
agent_writer(f'Write a stock analysis report on: {user_input}', images)

External LLM Integration

# Create Agent with Mistral-7B-Instruct-v0.3 instead
agent = Agent(toolchain = "responses, history = mistral_api(prompt, history)")

# Generate a neurology ICU admission note
agent('Write a neurology ICU admission note.')

# Follow-up questions (multi-turn conversation)
agent('Give me the inpatient BP goal for this patient.')
agent('DVT ppx for this patient?')
agent("Patient's prognosis?")

# End
agent.end()

Benchmarks

from phi_3_vision_mlx import benchmark

benchmark()

Task	Vanilla Model	Quantized Model	Quantized Cache	LoRA Adapter
Text Generation	25.02 tps	61.01 tps	18.68 tps	24.72 tps
Image Captioning	21.29 tps	44.26 tps	5.56 tps	20.48 tps
Batched Generation	236.60 tps	149.23 tps	121.92 tps	232.78 tps

(On M1 Max 64GB)

Documentation

API references and additional information are available at:

https://josefalbers.github.io/Phi-3-Vision-MLX/

Also check out our tutorial series available at:

https://medium.com/@albersj66

License

This project is licensed under the MIT License.

Citation

For Tasks:

Click tags to check more tools for each tasks

generate text answer questions execute code train lora adapter analyze images

For Jobs:

ai researcher data scientist machine learning engineer software developer research scientist

Alternative AI tools for Phi-3-Vision-MLX

Similar Open Source Tools

Phi-3-Vision-MLX

github

: 186

neural-speed

Neural Speed is an innovative library designed to support the efficient inference of large language models (LLMs) on Intel platforms through the state-of-the-art (SOTA) low-bit quantization powered by Intel Neural Compressor. The work is inspired by llama.cpp and further optimized for Intel platforms with our innovations in NeurIPS' 2023

github

: 327

mlx-vlm

MLX-VLM is a package designed for running Vision LLMs on Mac systems using MLX. It provides a convenient way to install and utilize the package for processing large language models related to vision tasks. The tool simplifies the process of running LLMs on Mac computers, offering a seamless experience for users interested in leveraging MLX for vision-related projects.

github

: 692

PhoGPT

PhoGPT is an open-source 4B-parameter generative model series for Vietnamese, including the base pre-trained monolingual model PhoGPT-4B and its chat variant, PhoGPT-4B-Chat. PhoGPT-4B is pre-trained from scratch on a Vietnamese corpus of 102B tokens, with an 8192 context length and a vocabulary of 20K token types. PhoGPT-4B-Chat is fine-tuned on instructional prompts and conversations, demonstrating superior performance. Users can run the model with inference engines like vLLM and Text Generation Inference, and fine-tune it using llm-foundry. However, PhoGPT has limitations in reasoning, coding, and mathematics tasks, and may generate harmful or biased responses.

github

: 739

openvino.genai

The GenAI repository contains pipelines that implement image and text generation tasks. The implementation uses OpenVINO capabilities to optimize the pipelines. Each sample covers a family of models and suggests certain modifications to adapt the code to specific needs. It includes the following pipelines: 1. Benchmarking script for large language models 2. Text generation C++ samples that support most popular models like LLaMA 2 3. Stable Diffuison (with LoRA) C++ image generation pipeline 4. Latent Consistency Model (with LoRA) C++ image generation pipeline

github

: 198

mlx-llm

mlx-llm is a library that allows you to run Large Language Models (LLMs) on Apple Silicon devices in real-time using Apple's MLX framework. It provides a simple and easy-to-use API for creating, loading, and using LLM models, as well as a variety of applications such as chatbots, fine-tuning, and retrieval-augmented generation.

github

: 384

amadeus-python

Amadeus Python SDK provides a rich set of APIs for the travel industry. It allows users to make API calls for various travel-related tasks such as flight offers search, hotel bookings, trip purpose prediction, flight delay prediction, airport on-time performance, travel recommendations, and more. The SDK conveniently maps API paths to similar paths, making it easy to interact with the Amadeus APIs. Users can initialize the client with their API key and secret, make API calls, handle responses, and enable logging for debugging purposes. The SDK documentation includes detailed information about each SDK method, arguments, and return types.

github

: 198

syncode

SynCode is a novel framework for the grammar-guided generation of Large Language Models (LLMs) that ensures syntactically valid output with respect to defined Context-Free Grammar (CFG) rules. It supports general-purpose programming languages like Python, Go, SQL, JSON, and more, allowing users to define custom grammars using EBNF syntax. The tool compares favorably to other constrained decoders and offers features like fast grammar-guided generation, compatibility with HuggingFace Language Models, and the ability to work with various decoding strategies.

github

: 213

ragoon

RAGoon is a high-level library designed for batched embeddings generation, fast web-based RAG (Retrieval-Augmented Generation) processing, and quantized indexes processing. It provides NLP utilities for multi-model embedding production, high-dimensional vector visualization, and enhancing language model performance through search-based querying, web scraping, and data augmentation techniques.

github

: 58

candle-vllm

Candle-vllm is an efficient and easy-to-use platform designed for inference and serving local LLMs, featuring an OpenAI compatible API server. It offers a highly extensible trait-based system for rapid implementation of new module pipelines, streaming support in generation, efficient management of key-value cache with PagedAttention, and continuous batching. The tool supports chat serving for various models and provides a seamless experience for users to interact with LLMs through different interfaces.

github

: 281

xlstm

xLSTM is a new Recurrent Neural Network architecture based on ideas of the original LSTM. Through Exponential Gating with appropriate normalization and stabilization techniques and a new Matrix Memory it overcomes the limitations of the original LSTM and shows promising performance on Language Modeling when compared to Transformers or State Space Models. The package is based on PyTorch and was tested for versions >=1.8. For the CUDA version of xLSTM, you need Compute Capability >= 8.0. The xLSTM tool provides two main components: xLSTMBlockStack for non-language applications or integrating in other architectures, and xLSTMLMModel for language modeling or other token-based applications.

github

: 1.6k

langfair

LangFair is a Python library for bias and fairness assessments of large language models (LLMs). It offers a comprehensive framework for choosing bias and fairness metrics, demo notebooks, and a technical playbook. Users can tailor evaluations to their use cases with a Bring Your Own Prompts approach. The focus is on output-based metrics practical for governance audits and real-world testing.

github

: 92

flow-prompt

Flow Prompt is a dynamic library for managing and optimizing prompts for large language models. It facilitates budget-aware operations, dynamic data integration, and efficient load distribution. Features include CI/CD testing, dynamic prompt development, multi-model support, real-time insights, and prompt testing and evolution.

github

: 68

Webscout

WebScout is a versatile tool that allows users to search for anything using Google, DuckDuckGo, and phind.com. It contains AI models, can transcribe YouTube videos, generate temporary email and phone numbers, has TTS support, webai (terminal GPT and open interpreter), and offline LLMs. It also supports features like weather forecasting, YT video downloading, temp mail and number generation, text-to-speech, advanced web searches, and more.

github

: 164

vision-parse

Vision Parse is a tool that leverages Vision Language Models to parse PDF documents into beautifully formatted markdown content. It offers smart content extraction, content formatting, multi-LLM support, PDF document support, and local model hosting using Ollama. Users can easily convert PDFs to markdown with high precision and preserve document hierarchy and styling. The tool supports multiple Vision LLM providers like OpenAI, LLama, and Gemini for accuracy and speed, making document processing efficient and effortless.

github

: 149

redisvl

Redis Vector Library (RedisVL) is a Python client library for building AI applications on top of Redis. It provides a high-level interface for managing vector indexes, performing vector search, and integrating with popular embedding models and providers. RedisVL is designed to make it easy for developers to build and deploy AI applications that leverage the speed, flexibility, and reliability of Redis.

github

: 158

For similar tasks

HPT

Hyper-Pretrained Transformers (HPT) is a novel multimodal LLM framework from HyperGAI, trained for vision-language models capable of understanding both textual and visual inputs. The repository contains the open-source implementation of inference code to reproduce the evaluation results of HPT Air on different benchmarks. HPT has achieved competitive results with state-of-the-art models on various multimodal LLM benchmarks. It offers models like HPT 1.5 Air and HPT 1.0 Air, providing efficient solutions for vision-and-language tasks.

github

: 236

learnopencv

LearnOpenCV is a repository containing code for Computer Vision, Deep learning, and AI research articles shared on the blog LearnOpenCV.com. It serves as a resource for individuals looking to enhance their expertise in AI through various courses offered by OpenCV. The repository includes a wide range of topics such as image inpainting, instance segmentation, robotics, deep learning models, and more, providing practical implementations and code examples for readers to explore and learn from.

github

: 21.1k

spark-free-api

Spark AI Free 服务 provides high-speed streaming output, multi-turn dialogue support, AI drawing support, long document interpretation, and image parsing. It offers zero-configuration deployment, multi-token support, and automatic session trace cleaning. It is fully compatible with the ChatGPT interface. The repository includes multiple free-api projects for various AI services. Users can access the API for tasks such as chat completions, AI drawing, document interpretation, image analysis, and ssoSessionId live checking. The project also provides guidelines for deployment using Docker, Docker-compose, Render, Vercel, and native deployment methods. It recommends using custom clients for faster and simpler access to the free-api series projects.

github

: 57

mlx-vlm

github

: 692

clarifai-python-grpc

This is the official Clarifai gRPC Python client for interacting with their recognition API. Clarifai offers a platform for data scientists, developers, researchers, and enterprises to utilize artificial intelligence for image, video, and text analysis through computer vision and natural language processing. The client allows users to authenticate, predict concepts in images, and access various functionalities provided by the Clarifai API. It follows a versioning scheme that aligns with the backend API updates and includes specific instructions for installation and troubleshooting. Users can explore the Clarifai demo, sign up for an account, and refer to the documentation for detailed information.

github

: 56

horde-worker-reGen

This repository provides the latest implementation for the AI Horde Worker, allowing users to utilize their graphics card(s) to generate, post-process, or analyze images for others. It offers a platform where users can create images and earn 'kudos' in return, granting priority for their own image generations. The repository includes important details for setup, recommendations for system configurations, instructions for installation on Windows and Linux, basic usage guidelines, and information on updating the AI Horde Worker. Users can also run the worker with multiple GPUs and receive notifications for updates through Discord. Additionally, the repository contains models that are licensed under the CreativeML OpenRAIL License.

github

: 102

geospy

Geospy is a Python tool that utilizes Graylark's AI-powered geolocation service to determine the location where photos were taken. It allows users to analyze images and retrieve information such as country, city, explanation, coordinates, and Google Maps links. The tool provides a seamless way to integrate geolocation services into various projects and applications.

github

: 71

Awesome-Colorful-LLM

Awesome-Colorful-LLM is a meticulously assembled anthology of vibrant multimodal research focusing on advancements propelled by large language models (LLMs) in domains such as Vision, Audio, Agent, Robotics, and Fundamental Sciences like Mathematics. The repository contains curated collections of works, datasets, benchmarks, projects, and tools related to LLMs and multimodal learning. It serves as a comprehensive resource for researchers and practitioners interested in exploring the intersection of language models and various modalities for tasks like image understanding, video pretraining, 3D modeling, document understanding, audio analysis, agent learning, robotic applications, and mathematical research.

github

: 106

For similar jobs

weave

Weave is a toolkit for developing Generative AI applications, built by Weights & Biases. With Weave, you can log and debug language model inputs, outputs, and traces; build rigorous, apples-to-apples evaluations for language model use cases; and organize all the information generated across the LLM workflow, from experimentation to evaluations to production. Weave aims to bring rigor, best-practices, and composability to the inherently experimental process of developing Generative AI software, without introducing cognitive overhead.

github

: 770

LLMStack

LLMStack is a no-code platform for building generative AI agents, workflows, and chatbots. It allows users to connect their own data, internal tools, and GPT-powered models without any coding experience. LLMStack can be deployed to the cloud or on-premise and can be accessed via HTTP API or triggered from Slack or Discord.

github

: 1.5k

VisionCraft

The VisionCraft API is a free API for using over 100 different AI models. From images to sound.

github

: 94

kaito

Kaito is an operator that automates the AI/ML inference model deployment in a Kubernetes cluster. It manages large model files using container images, avoids tuning deployment parameters to fit GPU hardware by providing preset configurations, auto-provisions GPU nodes based on model requirements, and hosts large model images in the public Microsoft Container Registry (MCR) if the license allows. Using Kaito, the workflow of onboarding large AI inference models in Kubernetes is largely simplified.

github

: 405

PyRIT

PyRIT is an open access automation framework designed to empower security professionals and ML engineers to red team foundation models and their applications. It automates AI Red Teaming tasks to allow operators to focus on more complicated and time-consuming tasks and can also identify security harms such as misuse (e.g., malware generation, jailbreaking), and privacy harms (e.g., identity theft). The goal is to allow researchers to have a baseline of how well their model and entire inference pipeline is doing against different harm categories and to be able to compare that baseline to future iterations of their model. This allows them to have empirical data on how well their model is doing today, and detect any degradation of performance based on future improvements.

github

: 1.8k

tabby

Tabby is a self-hosted AI coding assistant, offering an open-source and on-premises alternative to GitHub Copilot. It boasts several key features: * Self-contained, with no need for a DBMS or cloud service. * OpenAPI interface, easy to integrate with existing infrastructure (e.g Cloud IDE). * Supports consumer-grade GPUs.

github

: 26.4k

spear

SPEAR (Simulator for Photorealistic Embodied AI Research) is a powerful tool for training embodied agents. It features 300 unique virtual indoor environments with 2,566 unique rooms and 17,234 unique objects that can be manipulated individually. Each environment is designed by a professional artist and features detailed geometry, photorealistic materials, and a unique floor plan and object layout. SPEAR is implemented as Unreal Engine assets and provides an OpenAI Gym interface for interacting with the environments via Python.

github

: 224

Magick

Magick is a groundbreaking visual AIDE (Artificial Intelligence Development Environment) for no-code data pipelines and multimodal agents. Magick can connect to other services and comes with nodes and templates well-suited for intelligent agents, chatbots, complex reasoning systems and realistic characters.

github

: 675