Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/rwkv/rwkv.cpp
INT4/INT5/INT8 and FP16 inference on CPU for RWKV language model
https://github.com/rwkv/rwkv.cpp
deep-learning ggml language-model llm machine-learning quantization rwkv
Last synced: 16 days ago
JSON representation
INT4/INT5/INT8 and FP16 inference on CPU for RWKV language model
- Host: GitHub
- URL: https://github.com/rwkv/rwkv.cpp
- Owner: RWKV
- License: mit
- Created: 2023-03-30T13:26:47.000Z (over 1 year ago)
- Default Branch: master
- Last Pushed: 2024-08-07T07:10:59.000Z (3 months ago)
- Last Synced: 2024-10-14T22:22:12.114Z (about 1 month ago)
- Topics: deep-learning, ggml, language-model, llm, machine-learning, quantization, rwkv
- Language: C++
- Homepage:
- Size: 38.1 MB
- Stars: 1,409
- Watchers: 22
- Forks: 99
- Open Issues: 29
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# rwkv.cpp
This is a port of [BlinkDL/RWKV-LM](https://github.com/BlinkDL/RWKV-LM) to [ggerganov/ggml](https://github.com/ggerganov/ggml).
Besides the usual **FP32**, it supports **FP16**, **quantized INT4, INT5 and INT8** inference. This project is **focused on CPU**, but cuBLAS is also supported.
This project provides [a C library rwkv.h](rwkv.h) and [a convinient Python wrapper](python%2Frwkv_cpp%2Frwkv_cpp_model.py) for it.
[RWKV](https://arxiv.org/abs/2305.13048) is a large language model architecture, [with the largest model in the family having 14B parameters](https://huggingface.co/BlinkDL/rwkv-4-pile-14b). In contrast to Transformer with `O(n^2)` attention, RWKV requires only state from previous step to calculate logits. This makes RWKV very CPU-friendly on large context lenghts.
[RWKV v5](https://huggingface.co/BlinkDL/rwkv-5-world) is a major upgrade to RWKV architecture, making it competitive with Transformers in quality. RWKV v5 models are supported.
[RWKV v6](https://huggingface.co/BlinkDL/rwkv-6-world) is a further improvement to RWKV architecture, with better quality. RWKV v6 models are supported.
Loading LoRA checkpoints in [Blealtan's format](https://github.com/Blealtan/RWKV-LM-LoRA) is supported through [merge_lora_into_ggml.py script](rwkv%2Fmerge_lora_into_ggml.py).
## Quality and performance
If you use `rwkv.cpp` for anything serious, please [test all available formats for perplexity and latency](rwkv%2Fmeasure_pexplexity.py) on a representative dataset, and decide which trade-off is best for you.
In general, **`RWKV v5` models are as fast as `RWKV v4` models**, with minor differencies in latency and memory consumption, and with having way higher quality than `v4`. Therefore, it is recommended to use `RWKV v5`.
Below table is for reference only. Measurements were made on 4C/8T x86 CPU with AVX2, 4 threads. The models are `RWKV v4 Pile 169M`, `RWKV v4 Pile 1.5B`.
| Format | Perplexity (169M) | Latency, ms (1.5B) | File size, GB (1.5B) |
|-----------|-------------------|--------------------|----------------------|
| `Q4_0` | 17.507 | *76* | **1.53** |
| `Q4_1` | 17.187 | **72** | 1.68 |
| `Q5_0` | 16.194 | 78 | *1.60* |
| `Q5_1` | 15.851 | 81 | 1.68 |
| `Q8_0` | *15.652* | 89 | 2.13 |
| `FP16` | **15.623** | 117 | 2.82 |
| `FP32` | **15.623** | 198 | 5.64 |
### With cuBLAS
Measurements were made on Intel i7 13700K & NVIDIA 3060 Ti 8 GB. The model is `RWKV-4-Pile-169M`, 12 layers were offloaded to GPU.
Latency per token in ms shown.
| Format | 1 thread | 2 threads | 4 threads | 8 threads | 24 threads |
|--------|----------|-----------|-----------|-----------|------------|
| `Q4_0` | 7.9 | 6.2 | 6.9 | 8.6 | 20 |
| `Q4_1` | 7.8 | 6.7 | 6.9 | 8.6 | 21 |
| `Q5_1` | 8.1 | 6.7 | 6.9 | 9.0 | 22 |
| Format | 1 thread | 2 threads | 4 threads | 8 threads | 24 threads |
|--------|----------|-----------|-----------|-----------|------------|
| `Q4_0` | 59 | 51 | 50 | 54 | 94 |
| `Q4_1` | 59 | 51 | 49 | 54 | 94 |
| `Q5_1` | 77 | 69 | 67 | 72 | 101 |
Note: since cuBLAS is supported only for `ggml_mul_mat()`, we still need to use few CPU resources to execute remaining operations.
### With hipBLAS
Measurements were made on CPU AMD Ryzen 9 5900X & GPU AMD Radeon RX 7900 XTX. The model is `RWKV-novel-4-World-7B-20230810-ctx128k`, 32 layers were offloaded to GPU.
Latency per token in ms shown.
| Format | 1 thread | 2 threads | 4 threads | 8 threads | 24 threads |
|--------|----------|-----------|-----------|-----------|------------|
| `f16` | 94 | 91 | 94 | 106 | 944 |
| `Q4_0` | 83 | 77 | 75 | 110 | 1692 |
| `Q4_1` | 85 | 80 | 85 | 93 | 1691 |
| `Q5_1` | 83 | 78 | 83 | 90 | 1115 |
Note: same as cuBLAS, hipBLAS only supports `ggml_mul_mat()`, we still need to use few CPU resources to execute remaining operations.
## How to use
### 1. Clone the repo
**Requirements**: [git](https://gitforwindows.org/).
```commandline
git clone --recursive https://github.com/saharNooby/rwkv.cpp.git
cd rwkv.cpp
```
### 2. Get the rwkv.cpp library
#### Option 2.1. Download a pre-compiled library
##### Windows / Linux / MacOS
Check out [Releases](https://github.com/saharNooby/rwkv.cpp/releases), download appropriate ZIP for your OS and CPU, extract `rwkv` library file into the repository directory.
On Windows: to check whether your CPU supports AVX2 or AVX-512, [use CPU-Z](https://www.cpuid.com/softwares/cpu-z.html).
#### Option 2.2. Build the library yourself
This option is recommended for maximum performance, because the library would be built specifically for your CPU and OS.
##### Windows
**Requirements**: [CMake](https://cmake.org/download/) or [CMake from anaconda](https://anaconda.org/conda-forge/cmake), [Build Tools for Visual Studio 2019](https://visualstudio.microsoft.com/vs/older-downloads/).
```commandline
cmake .
cmake --build . --config Release
```
If everything went OK, `bin\Release\rwkv.dll` file should appear.
##### Windows + cuBLAS
Refer to [docs/cuBLAS_on_Windows.md](docs%2FcuBLAS_on_Windows.md) for a comprehensive guide.
##### Windows + hipBLAS
Refer to [docs/hipBLAS_on_Windows.md](docs%2FhipBLAS_on_Windows.md) for a comprehensive guide.
##### Linux / MacOS
**Requirements**: CMake (Linux: `sudo apt install cmake`, MacOS: `brew install cmake`, anaconoda: [cmake package](https://anaconda.org/conda-forge/cmake)).
```commandline
cmake .
cmake --build . --config Release
```
**Anaconda & M1 users**: please verify that `CMAKE_SYSTEM_PROCESSOR: arm64` after running `cmake .` — if it detects `x86_64`, edit the `CMakeLists.txt` file under the `# Compile flags` to add `set(CMAKE_SYSTEM_PROCESSOR "arm64")`.
If everything went OK, `librwkv.so` (Linux) or `librwkv.dylib` (MacOS) file should appear in the base repo folder.
##### Linux / MacOS + cuBLAS
```commandline
cmake . -DRWKV_CUBLAS=ON
cmake --build . --config Release
```
If everything went OK, `librwkv.so` (Linux) or `librwkv.dylib` (MacOS) file should appear in the base repo folder.
### 3. Get an RWKV model
**Requirements**: Python 3.x with [PyTorch](https://pytorch.org/get-started/locally/).
**First**, download a model from [Hugging Face](https://huggingface.co/BlinkDL) like [this one](https://huggingface.co/BlinkDL/rwkv-4-pile-169m/blob/main/RWKV-4-Pile-169M-20220807-8023.pth).
**Second**, convert it into `rwkv.cpp` format using following commands:
```commandline
# Windows
python python\convert_pytorch_to_ggml.py C:\RWKV-4-Pile-169M-20220807-8023.pth C:\rwkv.cpp-169M.bin FP16
# Linux / MacOS
python python/convert_pytorch_to_ggml.py ~/Downloads/RWKV-4-Pile-169M-20220807-8023.pth ~/Downloads/rwkv.cpp-169M.bin FP16
```
**Optionally**, quantize the model into one of quantized formats from the table above:
```commandline
# Windows
python python\quantize.py C:\rwkv.cpp-169M.bin C:\rwkv.cpp-169M-Q5_1.bin Q5_1
# Linux / MacOS
python python/quantize.py ~/Downloads/rwkv.cpp-169M.bin ~/Downloads/rwkv.cpp-169M-Q5_1.bin Q5_1
```
### 4. Run the model
#### Using the command line
**Requirements**: Python 3.x with [numpy](https://numpy.org/). If using `Pile` or `Raven` models, [tokenizers](https://pypi.org/project/tokenizers/) is also required.
To generate some text, run:
```commandline
# Windows
python python\generate_completions.py C:\rwkv.cpp-169M-Q5_1.bin
# Linux / MacOS
python python/generate_completions.py ~/Downloads/rwkv.cpp-169M-Q5_1.bin
```
To chat with a bot, run:
```commandline
# Windows
python python\chat_with_bot.py C:\rwkv.cpp-169M-Q5_1.bin
# Linux / MacOS
python python/chat_with_bot.py ~/Downloads/rwkv.cpp-169M-Q5_1.bin
```
Edit [generate_completions.py](rwkv%2Fgenerate_completions.py) or [chat_with_bot.py](rwkv%2Fchat_with_bot.py) to change prompts and sampling settings.
#### Using in your own code
The short and simple script [inference_example.py](python%2Finference_example.py) demostrates the use of `rwkv.cpp` in Python.
To use `rwkv.cpp` in C/C++, include the header [rwkv.h](rwkv.h).
To use `rwkv.cpp` in any other language, see [Bindings](#Bindings) section below. If your language is missing, you can try to bind to the C API using the tooling provided by your language.
## Bindings
These projects wrap `rwkv.cpp` for easier use in other languages/frameworks.
* Golang: [seasonjs/rwkv](https://github.com/seasonjs/rwkv)
* Node.js: [Atome-FE/llama-node](https://github.com/Atome-FE/llama-node)
## Compatibility
`ggml` moves fast, and can occasionally break compatibility with older file formats.
`rwkv.cpp` will attempt it's best to explain why a model file can't be loaded and what next steps are available to the user.
For reference only, here is a list of latest versions of `rwkv.cpp` that have supported older formats. **No support will be provided for these versions**.
- `Q4_2`, old layout of quantized formats
- [commit 3ca9c7f](https://github.com/saharNooby/rwkv.cpp/commit/3ca9c7f7857a4b9f3de616ec938e71249cfb3f3f), [release with prebuilt binaries](https://github.com/saharNooby/rwkv.cpp/releases/tag/master-3ca9c7f)
- `Q4_3`, `Q4_1_O`
- [commit c736ef5](https://github.com/saharNooby/rwkv.cpp/commit/c736ef5411606b529d3a74c139ee111ef1a28bb9), [release with prebuilt binaries](https://github.com/saharNooby/rwkv.cpp/releases/tag/master-1c363e6)
See also [docs/FILE_FORMAT.md](docs/FILE_FORMAT.md) for version numbers of `rwkv.cpp` model files and their changelog.
## Contributing
Please follow the code style described in [docs/CODE_STYLE.md](docs/CODE_STYLE.md).