Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/intel/neural-speed
An innovative library for efficient LLM inference via low-bit quantization
https://github.com/intel/neural-speed
cpu fp4 fp8 gaudi2 gpu int4 int8 llamacpp llm-fine-tuning llm-inference low-bit mxformat nf4 sparsity
Last synced: about 1 month ago
JSON representation
An innovative library for efficient LLM inference via low-bit quantization
- Host: GitHub
- URL: https://github.com/intel/neural-speed
- Owner: intel
- License: apache-2.0
- Created: 2023-11-20T04:33:41.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-04-12T11:22:40.000Z (7 months ago)
- Last Synced: 2024-04-14T02:42:45.695Z (7 months ago)
- Topics: cpu, fp4, fp8, gaudi2, gpu, int4, int8, llamacpp, llm-fine-tuning, llm-inference, low-bit, mxformat, nf4, sparsity
- Language: C++
- Homepage: https://github.com/intel/neural-speed
- Size: 14.8 MB
- Stars: 179
- Watchers: 7
- Forks: 20
- Open Issues: 19
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
- Security: security.md
- Support: docs/supported_models.md
Awesome Lists containing this project
README
# PROJECT NOT UNDER ACTIVE MANAGEMENT
This project will no longer be maintained by Intel.
Intel has ceased development and contributions including, but not limited to, maintenance, bug fixes, new releases, or updates, to this project.
Intel no longer accepts patches to this project.
## Please refer to https://github.com/intel/intel-extension-for-pytorch as an alternative
# 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](https://github.com/intel/neural-compressor). The work is inspired by [llama.cpp](https://github.com/ggerganov/llama.cpp) and further optimized for Intel platforms with our innovations in [NeurIPS' 2023](https://arxiv.org/abs/2311.00502)
## Key Features
- Highly optimized kernels on CPUs with ISAs (AMX, VNNI, AVX512F, AVX_VNNI and AVX2) for N-bit weight (int1, int2, int3, int4, int5, int6, int7 and int8). See [details](neural_speed/core/README.md)
- Up to 40x performance speedup on popular LLMs compared with llama.cpp. See [details](https://medium.com/@NeuralCompressor/llm-performance-of-intel-extension-for-transformers-f7d061556176)
- Tensor parallelism across sockets/nodes on CPUs. See [details](./docs/tensor_parallelism.md)
> Neural Speed is under active development so APIs are subject to change.
## Supported Hardware
| Hardware | Supported |
|-------------|:-------------:|
|Intel Xeon Scalable Processors | ✔ |
|Intel Xeon CPU Max Series | ✔ |
|Intel Core Processors | ✔ |
## Supported Models
Support almost all the LLMs in PyTorch format from Hugging Face such as Llama2, ChatGLM2, Baichuan2, Qwen, Mistral, Whisper, etc. File an [issue](https://github.com/intel/neural-speed/issues) if your favorite LLM does not work.
Support typical LLMs in GGUF format such as Llama2, Falcon, MPT, Bloom etc. More are coming. Check out the [details](./docs/supported_models.md).
## Installation
### Install from binary
```shell
pip install -r requirements.txt
pip install neural-speed
```
### Build from Source
```shell
pip install .
```
>**Note**: GCC requires version 10+
## Quick Start (Transformer-like usage)
Install [Intel Extension for Transformers](https://github.com/intel/intel-extension-for-transformers/blob/main/docs/installation.md) to use Transformer-like APIs.
### PyTorch Model from Hugging Face
```python
from transformers import AutoTokenizer, TextStreamer
from intel_extension_for_transformers.transformers import AutoModelForCausalLM
model_name = "Intel/neural-chat-7b-v3-1" # Hugging Face model_id or local model
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, load_in_4bit=True)
outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)
```
### GGUF Model from Hugging Face
```python
from transformers import AutoTokenizer, TextStreamer
from intel_extension_for_transformers.transformers import AutoModelForCausalLM
# Specify the GGUF repo on the Hugginface
model_name = "TheBloke/Llama-2-7B-Chat-GGUF"
# Download the the specific gguf model file from the above repo
gguf_file = "llama-2-7b-chat.Q4_0.gguf"
# make sure you are granted to access this model on the Huggingface.
tokenizer_name = "meta-llama/Llama-2-7b-chat-hf"
prompt = "Once upon a time"
tokenizer = AutoTokenizer.from_pretrained(tokenizer_name, trust_remote_code=True)
inputs = tokenizer(prompt, return_tensors="pt").input_ids
streamer = TextStreamer(tokenizer)
model = AutoModelForCausalLM.from_pretrained(model_name, gguf_file = gguf_file)
outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)
```
### PyTorch Model from Modelscope
```python
from transformers import TextStreamer
from modelscope import AutoTokenizer
from intel_extension_for_transformers.transformers import AutoModelForCausalLM
model_name = "qwen/Qwen-7B" # Modelscope model_id or local model
prompt = "Once upon a time, there existed a little girl,"
model = AutoModelForCausalLM.from_pretrained(model_name, load_in_4bit=True, model_hub="modelscope")
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
inputs = tokenizer(prompt, return_tensors="pt").input_ids
streamer = TextStreamer(tokenizer)
outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)
```
### As an Inference Backend in Neural Chat Server
`Neural Speed` can be used in [Neural Chat Server](https://github.com/intel/intel-extension-for-transformers/tree/main/intel_extension_for_transformers/neural_chat/server) of `Intel Extension for Transformers`. You can choose to enable it by adding `use_neural_speed: true` in `config.yaml`.
- add `optimization` key section to use `Neural Speed` and its RTN quantization ([example](https://github.com/intel/intel-extension-for-transformers/blob/main/intel_extension_for_transformers/neural_chat/examples/deployment/codegen/backend/pc/woq/codegen.yaml)).
```yaml
device: "cpu"
# itrex int4 llm runtime optimization
optimization:
use_neural_speed: true
optimization_type: "weight_only"
compute_dtype: "fp32"
weight_dtype: "int4"
```
- add key `use_neural_speed` and key `use_gptq` to use `Neural Speed` and load `GPT-Q` model ([example](https://github.com/intel/intel-extension-for-transformers/blob/main/intel_extension_for_transformers/neural_chat/examples/deployment/codegen/backend/pc/gptq/codegen.yaml)).
```yaml
device: "cpu"
use_neural_speed: true
use_gptq: true
```
More details please refer to [Neural Chat](https://github.com/intel/intel-extension-for-transformers/tree/main/intel_extension_for_transformers/neural_chat).
## Quick Start (llama.cpp-like usage)
### Single (One-click) Step
```
python scripts/run.py model-path --weight_dtype int4 -p "She opened the door and see"
```
### Multiple Steps
#### Convert and Quantize
```bash
# skip the step if GGUF model is from Hugging Face or generated by llama.cpp
python scripts/convert.py --outtype f32 --outfile ne-f32.bin EleutherAI/gpt-j-6b
# Using the quantize script requires a binary installation of Neural Speed
mkdir build&&cd build
cmake ..&&make -j
cd ..
python scripts/quantize.py --model_name gptj --model_file ne-f32.bin --out_file ne-q4_j.bin --build_dir ./build --weight_dtype int4 --alg sym
```
#### Inference
```bash
# Linux and WSL
OMP_NUM_THREADS= numactl -m 0 -C 0- python scripts/inference.py --model_name llama -m ne-q4_j.bin -c 512 -b 1024 -n 256 -t --color -p "She opened the door and see"
```
```bash
# Windows
python scripts/inference.py --model_name llama -m ne-q4_j.bin -c 512 -b 1024 -n 256 -t --color -p "She opened the door and see"
```
> Please refer to [Advanced Usage](./docs/advanced_usage.md) for more details.
## Advanced Topics
### New model enabling
You can consider adding your own models, please follow the document: [graph developer document](./developer_document.md).
### Performance profiling
Enable `NEURAL_SPEED_VERBOSE` environment variable for performance profiling.
Available modes:
- 0: Print full information: evaluation time and operator profiling. Need to set `NS_PROFILING` to ON and recompile.
- 1: Print evaluation time. Time taken for each evaluation.
- 2: Profile individual operator. Identify performance bottleneck within the model. Need to set `NS_PROFILING` to ON and recompile.