The RWKV Language Model

Last update: Jan 05, 2023

Related tags

Overview

RWKV-LM

We propose the RWKV language model, with alternating time-mix and channel-mix layers:

$\begin{align*} \text{Time-mix :} && \text{TM}_{t,c} &&=&&\text{sigmoid}(\text{R}_{t,c}) &&\cdot&& &&\textstyle\sum_{u} &&\textbf{W}_{t,u,c} &&\cdot&& \text{softmax}_t(\text{K}_{u,c}) &&\cdot&& \text{V}_{u,c}\\ \text{Channel-mix :} && \text{CM}_{t,c} &&=&&\text{sigmoid}(\text{R}_{t,c}) &&\cdot&& &&\textstyle\sum_d &&\textbf{W}_{c,d} &&\cdot&& \text{gelu}(\text{K}_{t,d}) &&\cdot&& \text{V}_{t,d} \end{align*}$

The R, K, V are generated by linear transforms of input, and W is parameter. The idea of RWKV is to decompose attention into R(target) * W(src, target) * K(src). So we can call R "receptance", and sigmoid means it's in 0~1 range.
The Time-mix is similar to AFT (https://arxiv.org/abs/2105.14103). There are two differences.

(1) We changed the normalization (denominator). For masked language models, we define:

$\text{softmax}_t(\text{K}_{u,c}) = \frac{\exp(\text{K}_{u,c})}{\sum_{v \leq t}\exp(\text{K}_{v,c})}$

(2) We decompose W_{t,u,c} and introduce multi-head W (here h is the corresponding head of c):

$W_{t,u,c}=f_h(t-u)\cdot \alpha_h(u) \cdot \beta_h(t)$

(3) You don't need LayerNorm for Time-mix. In fact, the model converges faster when LayerNorm is removed.

Moreover we multiply the final output of Time-mix layer by γ(t). The reason for the α β γ factors, is because the context size is smaller when t is small, and this can be compensated using the α β γ factors.

The Channel-mix is similar to GeGLU (https://arxiv.org/abs/2002.05202) with an extra R factor.
Finally, we add extra time-mixing as in (https://github.com/BlinkDL/minGPT-tuned). You can try reducing the amt of time-mixing in upper layers of deep models.

We also propose a new sampling method (as in src/utils.py):

(1) Find the max probability p_max after softmax.

(2) Remove all entries whose probability is lower than 0.02 * pow(p_max, 2)

(3) Feel free to tune the 0.02 and 2 factor.

Training loss, RWKV vs MHA+Rotary+GeGLU:

(this is character-level loss with simplebooks-92 dataset https://dldata-public.s3.us-east-2.amazonaws.com/simplebooks.zip)

Comments

Sequence to Sequence?

Hey @BlinkDL! Awesome project!

I was wondering if you have performed any Seq-2-Seq experiments with it? Any reason for going with GPT model in the first place as opposed to something like T5 (standard Transformer)? Any direction on what changes will be required to make a standard encoder-decoder architecture with RWKV?

Also, is there any report on in-context-learning/FSL capability of the latest trained model?

opened by SushantDaga 2
v4 model.py vs model_run.py

Hi, Thanks for this awesome repo! I'm trying to understand the code and found that in the v4 folder, there's this model.py and model_run.py, which contains GPT and RWKV_GPT respectively which all uses different initialization methods. Could you elaborate on when should which one be used? Thanks in advnace!

opened by jingweiz 3
RWKV-4 169m/430m in browser with ORT Web / TF.js / tfjs-tflite?

Hi, really exciting project! I'm wondering if you've published the model conversion script that you used to create the js_models files from the .pth model file? It would be awesome to see how the larger and newer models like RWKV-4 169m/430m perform in the browser! I think the inference speed of RWKV opens up many new possibilities for language models on the web.

opened by josephrocca 32

CUDA compilation error with Ctx Length>2000

Hello, I am trying out RWKV with audio modality and when I set T_MAX>>1000, it throws this error:

Emitting ninja build file /root/.cache/torch_extensions/py39_cu116/timex/build.ninja...
Building extension module timex...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
[1/2] /usr/local/cuda/bin/nvcc  -DTORCH_EXTENSION_NAME=timex -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE=\"_gcc\" -DPYBIND11_STDLIB=\"_libstdcpp\" -DPYBIND11_BUILD_ABI=\"_cxxabi1013\" -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/torch/csrc/api/include -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/TH -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/THC -isystem /usr/local/cuda/include -isystem /root/anaconda3/envs/surya-env/include/python3.9 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_80,code=sm_80 --compiler-options '-fPIC' --use_fast_math --extra-device-vectorization -DTmax=10000 -DBF=8 -DBB=2 -std=c++14 -c cuda/timex_cuda.cu -o timex_cuda.cuda.o 
FAILED: timex_cuda.cuda.o 
/usr/local/cuda/bin/nvcc  -DTORCH_EXTENSION_NAME=timex -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE=\"_gcc\" -DPYBIND11_STDLIB=\"_libstdcpp\" -DPYBIND11_BUILD_ABI=\"_cxxabi1013\" -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/torch/csrc/api/include -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/TH -isystem /root/anaconda3/envs/surya-env/lib/python3.9/site-packages/torch/include/THC -isystem /usr/local/cuda/include -isystem /root/anaconda3/envs/surya-env/include/python3.9 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_80,code=sm_80 --compiler-options '-fPIC' --use_fast_math --extra-device-vectorization -DTmax=10000 -DBF=8 -DBB=2 -std=c++14 -c cuda/timex_cuda.cu -o timex_cuda.cuda.o 
ptxas error   : Entry function '_Z15kernel_backwardIfEvPKT_S2_S2_PS0_S3_iii' uses too much shared data (0x30d40 bytes, 0xc000 max)
ptxas error   : Entry function '_Z14kernel_forwardIfEvPKT_S2_PS0_S0_iii' uses too much shared data (0x57e40 bytes, 0xc000 max)
ninja: build stopped: subcommand failed.

GPU: A100, VRAM: 42GB, CUDA 11.6

I am okay if the training takes a bit long. But I need this to work. Don't know any CUDA. Can you suggest some workarounds?

Thanks for the incredible work btw!

opened by ojus1 8

关于调用模型做分类任务

你好作者！我对此工作很感兴趣，因为我现在在用基于transformer的模型做分类任务，transformer或者RNN在分类任务里通常采用最后一个模块的每个通道的最后一个元素作为输出，并通过全连接层映射到几个类别。请问你觉得RWKV原理类似吗？依旧提取最后一个元素作为输出是否稳妥呢？希望您能给出一些建议，我将很感激！

opened by louisinhit 2

Releases(4.00)

4.00(Dec 6, 2022)

Just a stable release.
Source code(tar.gz)
Source code(zip)
2.00(Mar 25, 2022)

Attached model : ctx1024-layer6-emb512 on enwik8 with 1.65 dev perplexity (0.72 BPC)
Source code(tar.gz)
Source code(zip)
enwik8-ppl1.65-6064-1024-RWKV-6-512-2022-03-25-21-05-13.zip(95.42 MB)
0.02(Aug 25, 2021)

v0.02 with RWKV, MHA_shift, MHA_rotary, MHA_pro, and time-shift mixing.
Source code(tar.gz)
Source code(zip)
0.01(Aug 13, 2021)

first release with RWKV, MHA_rotary, MHA_pro, and time-shift mixing.
Source code(tar.gz)
Source code(zip)

Owner

PENG Bo

http://zhihu.com/people/bopengbopeng

GitHub Repository

A Neural Language Style Transfer framework to transfer natural language text smoothly between fine-grained language styles like formal/casual, active/passive, and many more. Created by Prithiviraj Damodaran. Open to pull requests and other forms of collaboration.

Styleformer A Neural Language Style Transfer framework to transfer natural language text smoothly between fine-grained language styles like formal/cas

431 Dec 19, 2022

GPT-3 command line interaction

Writer_unblock Straight-forward command line interfacing with GPT-3. Finding yourself stuck at a conceptual stage? Spinning your wheels needlessly on

6 Feb 10, 2022

Mysticbbs-rjam - rJAM splitscreen message reader for MysticBBS A46+

rJAM splitscreen message reader for MysticBBS A46+

4 Nov 22, 2022

PORORO: Platform Of neuRal mOdels for natuRal language prOcessing

PORORO: Platform Of neuRal mOdels for natuRal language prOcessing pororo performs Natural Language Processing and Speech-related tasks. It is easy to

1.2k Dec 21, 2022

Chinese Pre-Trained Language Models (CPM-LM) Version-I

CPM-Generate 为了促进中文自然语言处理研究的发展，本项目提供了 CPM-LM (2.6B) 模型的文本生成代码，可用于文本生成的本地测试，并以此为基础进一步研究零次学习/少次学习等场景。[项目首页] [模型下载] [技术报告] 若您想使用CPM-1进行推理，我们建议使用高效推理工具BMI

1.4k Jan 03, 2023

Sinkhorn Transformer - Practical implementation of Sparse Sinkhorn Attention

Sinkhorn Transformer This is a reproduction of the work outlined in Sparse Sinkhorn Attention, with additional enhancements. It includes a parameteriz

217 Nov 25, 2022

translate using your voice

speech-to-text-translator Usage translate using your voice description this project makes translating a word easy, all you have to do is speak and...

1 Oct 18, 2021

Code for Text Prior Guided Scene Text Image Super-Resolution

82 Dec 26, 2022

PatrickStar enables Larger, Faster, Greener Pretrained Models for NLP. Democratize AI for everyone.

633 Dec 28, 2022

A text augmentation tool for named entity recognition.

neraug This python library helps you with augmenting text data for named entity recognition. Augmentation Example Reference from An Analysis of Simple

48 Oct 11, 2022

GPT-2 Model for Leetcode Questions in python

Leetcode using AI 🤖 GPT-2 Model for Leetcode Questions in python New demo here: https://huggingface.co/spaces/gagan3012/project-code-py Note: the Ans

100 Dec 12, 2022

Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis (SV2TTS)

This repository is an implementation of Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis (SV2TTS) with a vocoder that works in real-time. Feel free to check my the

38.5k Jan 03, 2023

Just a basic Telegram AI chat bot written in Python using Pyrogram.

Nikko ChatBot Just a basic Telegram AI chat bot written in Python using Pyrogram. Requirements Python 3.7 or higher. A bot token. Installation $ https

2 Oct 21, 2022

Transformer - A TensorFlow Implementation of the Transformer: Attention Is All You Need

[UPDATED] A TensorFlow Implementation of Attention Is All You Need When I opened this repository in 2017, there was no official code yet. I tried to i