Intel® Extension for Transformers

An Innovative Transformer-based Toolkit to Accelerate GenAI/LLM Everywhere

🏭Architecture   |   💬NeuralChat   |   😃Inference   |   💻Examples   |   📖Documentations

🚀Latest News

• [2023/11] Published a 4-bit chatbot demo (based on NeuralChat) available on Intel Hugging Face Space. Welcome to have a try! To setup the demo locally, please follow the instructions.
• [2023/11] Released Fast, accurate, and infinite LLM inference with improved StreamingLLM on Intel CPUs!
• [2023/11] Our paper Efficient LLM Inference on CPUs has been accepted by NeurIPS'23 on Efficient Natural Language and Speech Processing. Thanks to all the collaborators!
• [2023/10] LLM runtime, an Intel-optimized GGML compatible runtime, demonstrates up to 15x performance gain in 1st token generation and 1.5x in other token generation over the default llama.cpp.
• [2023/10] LLM runtime now supports LLM inference with infinite-length inputs up to 4 million tokens, inspired from StreamingLLM.
• [2023/09] NeuralChat has been showcased in Intel Innovation’23 Keynote and Google Cloud Next'23 to demonstrate GenAI/LLM capabilities on Intel Xeon Scalable Processors.
• [2023/08] NeuralChat supports custom chatbot development and deployment within minutes on broad Intel HWs such as Xeon Scalable Processors, Gaudi2, Xeon CPU Max Series, Data Center GPU Max Series, Arc Series, and Core Processors. Check out Notebooks.
• [2023/07] LLM runtime extends Hugging Face Transformers API to provide seamless low precision inference for popular LLMs, supporting low precision data types such as INT3/INT4/FP4/NF4/INT5/INT8/FP8.

---

🏃Installation

Quick Install from Pypi

pip install intel-extension-for-transformers

For more installation methods, please refer to Installation Page

🌟Introduction

Intel® Extension for Transformers is an innovative toolkit to accelerate Transformer-based models on Intel platforms, in particular, effective on 4th Intel Xeon Scalable processor Sapphire Rapids (codenamed Sapphire Rapids). The toolkit provides the below key features and examples:

• Seamless user experience of model compressions on Transformer-based models by extending Hugging Face transformers APIs and leveraging Intel® Neural Compressor

• Advanced software optimizations and unique compression-aware runtime (released with NeurIPS 2022's paper Fast Distilbert on CPUs and QuaLA-MiniLM: a Quantized Length Adaptive MiniLM, and NeurIPS 2021's paper Prune Once for All: Sparse Pre-Trained Language Models)

• Optimized Transformer-based model packages such as Stable Diffusion, GPT-J-6B, GPT-NEOX, BLOOM-176B, T5, Flan-T5, and end-to-end workflows such as SetFit-based text classification and document level sentiment analysis (DLSA)

• NeuralChat, a customizable chatbot framework to create your own chatbot within minutes by leveraging a rich set of plugins Knowledge Retrieval, Speech Interaction, Query Caching, and Security Guardrail.

• Inference of Large Language Model (LLM) in pure C/C++ with weight-only quantization kernels, supporting GPT-NEOX, LLAMA, MPT, FALCON, BLOOM-7B, OPT, ChatGLM2-6B, GPT-J-6B, and Dolly-v2-3B. Support AMX, VNNI, AVX512F and AVX2 instruction set.

🌱Getting Started

Below is the sample code to enable the chatbot. See more examples.

Chatbot

# pip install intel-extension-for-transformers
from intel_extension_for_transformers.neural_chat import build_chatbot
chatbot = build_chatbot()
response = chatbot.predict("Tell me about Intel Xeon Scalable Processors.")

Below is the sample code to enable weight-only INT4/INT8 inference. See more examples.

INT4 Inference

from transformers import AutoTokenizer, TextStreamer
from intel_extension_for_transformers.transformers import AutoModelForCausalLM, WeightOnlyQuantConfig
model_name = "Intel/neural-chat-7b-v1-1"     # Hugging Face model_id or local model
config = WeightOnlyQuantConfig(compute_dtype="int8", weight_dtype="int4")
prompt = "Once upon a time, there existed a little girl,"

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
inputs = tokenizer(prompt, return_tensors="pt").input_ids
streamer = TextStreamer(tokenizer)

model = AutoModelForCausalLM.from_pretrained(model_name, quantization_config=config)
outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)

INT8 Inference

from transformers import AutoTokenizer, TextStreamer
from intel_extension_for_transformers.transformers import AutoModelForCausalLM, WeightOnlyQuantConfig
model_name = "Intel/neural-chat-7b-v1-1"     # Hugging Face model_id or local model
config = WeightOnlyQuantConfig(compute_dtype="bf16", weight_dtype="int8")
prompt = "Once upon a time, there existed a little girl,"

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
inputs = tokenizer(prompt, return_tensors="pt").input_ids
streamer = TextStreamer(tokenizer)

model = AutoModelForCausalLM.from_pretrained(model_name, quantization_config=config)
outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)

🎯Validated Models

You can access the latest int4 performance and accuracy at int4 blog.

Additionally, we are preparing to introduce Baichuan, Mistral, and other models into LLM Runtime (Intel Optimized llamacpp). For comprehensive accuracy and performance data, though not the most up-to-date, please refer to the Release data.

📖Documentation

OVERVIEW
NeuralChat				LLM Runtime
NEURALCHAT
Chatbot on Intel CPU		Chatbot on Intel GPU			Chatbot on Gaudi
Chatbot on Client				More Notebooks
LLM RUNTIME
LLM Runtime		Streaming LLM		Low Precision Kernels		Tensor Parallelism
LLM COMPRESSION
SmoothQuant (INT8)		Weight-only Quantization (INT4/FP4/NF4/INT8)			QLoRA on CPU
GENERAL COMPRESSION
Quantization		Pruning		Distillation		Orchestration
Neural Architecture Search		Export		Metrics		Objectives
Pipeline		Length Adaptive		Early Exit		Data Augmentation
TUTORIALS & RESULTS
Tutorials		LLM List		General Model List		Model Performance

🙌Demo

• Infinite inference (up to 4M tokens)

streamingLLM_v2.mp4

📃Selected Publications/Events

• Blog published on marktechpost: Intel Researchers Propose a New Artificial Intelligence Approach to Deploy LLMs on CPUs More Efficiently (Nov 2023)
• Blog published on VMware: AI without GPUs: A Technical Brief for VMware Private AI with Inte (Nov 2023)
• News releases on VMware: VMware Collaborates with Intel to Unlock Private AI Everywhere (Nov 2023)
• Video on YouTube: Build Your Own ChatBot with Neural Chat | Intel Software (Oct 2023)
• Blog published on Medium: Layer-wise Low-bit Weight Only Quantization on a Laptop (Oct 2023)
• Blog published on Medium: Intel-Optimized Llama.CPP in Intel Extension for Transformers (Oct 2023)
• Blog published on Medium: Reduce the Carbon Footprint of Large Language Models (Oct 2023)
• Blog published on Medium: Empower Applications with Optimized LLMs: Performance, Cost, and Beyond (Sep 2023)
• Blog published on Medium: NeuralChat: Simplifying Supervised Instruction Fine-tuning and Reinforcement Aligning for Chatbots (Sep 2023)
• Intel Innovation'23 Keynote: Intel Innovation 2023 Keynote by Greg Lavender (Sep 2023)
• Blog published on Medium: NeuralChat: A Customizable Chatbot Framework (Sep 2023)

View Full Publication List.

Additional Content

• Release Information
• Contribution Guidelines
• Legal Information
• Security Policy
• Apache License

Acknowledgements

• Excellent open-source projects: bitsandbytes, FastChat, fastRAG, ggml, gptq, llama.cpp, lm-evauation-harness, peft, trl, streamingllm and many others.

• Thanks to all the contributors.

💁Collaborations

Welcome to raise any interesting ideas on model compression techniques and LLM-based chatbot development! Feel free to reach us, and we look forward to our collaborations on Intel Extension for Transformers!