The RWKV Language Model

Related tags

Text Data & NLP RWKV-LM
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.

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:

RWKV-vs-MHA

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

Comments
  • Sequence to Sequence?

    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

    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?

    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

    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

    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)
Owner
PENG Bo
http://zhihu.com/people/bopengbopeng
PENG Bo
GitHub
Incorporating KenLM language model with HuggingFace implementation of Wav2Vec2CTC Model using beam search decoding

Wav2Vec2CTC With KenLM Using KenLM ARPA language model with beam search to decode audio files and show the most probable transcription. Assuming you'v

farisalasmary 65 Sep 21, 2022
A python framework to transform natural language questions to queries in a database query language.

__ _ _ _ ___ _ __ _ _ / _` | | | |/ _ \ '_ \| | | | | (_| | |_| | __/ |_) | |_| | \__, |\__,_|\___| .__/ \__, | |_| |_| |___/

Machinalis 1.2k Dec 18, 2022
A Domain Specific Language (DSL) for building language patterns. These can be later compiled into spaCy patterns, pure regex, or any other format

RITA DSL This is a language, loosely based on language Apache UIMA RUTA, focused on writing manual language rules, which compiles into either spaCy co

Šarūnas Navickas 60 Sep 26, 2022
Indobenchmark are collections of Natural Language Understanding (IndoNLU) and Natural Language Generation (IndoNLG)

Indobenchmark Toolkit Indobenchmark are collections of Natural Language Understanding (IndoNLU) and Natural Language Generation (IndoNLG) resources fo

Samuel Cahyawijaya 11 Aug 26, 2022
LegalNLP - Natural Language Processing Methods for the Brazilian Legal Language

LegalNLP - Natural Language Processing Methods for the Brazilian Legal Language ⚖️ The library of Natural Language Processing for Brazilian legal lang

Felipe Maia Polo 125 Dec 20, 2022
A design of MIDI language for music generation task, specifically for Natural Language Processing (NLP) models.

MIDI Language Introduction Reference Paper: Pop Music Transformer: Beat-based Modeling and Generation of Expressive Pop Piano Compositions: code This

Robert Bogan Kang 3 May 25, 2022
This is the Alpha of Nutte language, she is not complete yet / Essa é a Alpha da Nutte language, não está completa ainda

nutte-language This is the Alpha of Nutte language, it is not complete yet / Essa é a Alpha da Nutte language, não está completa ainda My language was

catdochrome 2 Dec 18, 2021
Simple Python script to scrape youtube channles of "Parity Technologies and Web3 Foundation" and translate them to well-known braille language or any language

Simple Python script to scrape youtube channles of "Parity Technologies and Web3 Foundation" and translate them to well-known braille language or any

Little Endian 1 Apr 28, 2022
NL. The natural language programming language.

NL A Natural-Language programming language. Built using Codex. A few examples are inside the nl_projects directory. How it works Write any code in pur

null 2 Jan 17, 2022
A model library for exploring state-of-the-art deep learning topologies and techniques for optimizing Natural Language Processing neural networks

A Deep Learning NLP/NLU library by Intel® AI Lab Overview | Models | Installation | Examples | Documentation | Tutorials | Contributing NLP Architect

Intel Labs 2.9k Jan 2, 2023
A model library for exploring state-of-the-art deep learning topologies and techniques for optimizing Natural Language Processing neural networks

A Deep Learning NLP/NLU library by Intel® AI Lab Overview | Models | Installation | Examples | Documentation | Tutorials | Contributing NLP Architect

Intel Labs 2.6k Feb 18, 2021
Implementation of COCO-LM, Correcting and Contrasting Text Sequences for Language Model Pretraining, in Pytorch

COCO LM Pretraining (wip) Implementation of COCO-LM, Correcting and Contrasting Text Sequences for Language Model Pretraining, in Pytorch. They were a

Phil Wang 44 Jul 28, 2022
MILES is a multilingual text simplifier inspired by LSBert - A BERT-based lexical simplification approach proposed in 2018. Unlike LSBert, MILES uses the bert-base-multilingual-uncased model, as well as simple language-agnostic approaches to complex word identification (CWI) and candidate ranking.

MILES Multilingual Lexical Simplifier Explore the docs » Read LSBert Paper · Report Bug · Request Feature About The Project MILES is a multilingual te

Kane 45 Oct 19, 2022
ProteinBERT is a universal protein language model pretrained on ~106M proteins from the UniRef90 dataset.

ProteinBERT is a universal protein language model pretrained on ~106M proteins from the UniRef90 dataset. Through its Python API, the pretrained model can be fine-tuned on any protein-related task in a matter of minutes. Based on our experiments with a wide range of benchmarks, ProteinBERT usually achieves state-of-the-art performance. ProteinBERT is built on TenforFlow/Keras.

null 241 Jan 4, 2023
Protein Language Model

ProteinLM We pretrain protein language model based on Megatron-LM framework, and then evaluate the pretrained model results on TAPE (Tasks Assessing P

THUDM 77 Dec 27, 2022
A simple implementation of N-gram language model.

About A simple implementation of N-gram language model. Requirements numpy Data preparation Corpus Training data for the N-gram model, a text file lik

null 4 Nov 24, 2021
PyTorch original implementation of Cross-lingual Language Model Pretraining.

XLM NEW: Added XLM-R model. PyTorch original implementation of Cross-lingual Language Model Pretraining. Includes: Monolingual language model pretrain

Facebook Research 2.7k Dec 27, 2022
PyTorch Language Model for 1-Billion Word (LM1B / GBW) Dataset

PyTorch Large-Scale Language Model A Large-Scale PyTorch Language Model trained on the 1-Billion Word (LM1B) / (GBW) dataset Latest Results 39.98 Perp

Ryan Spring 114 Nov 4, 2022
GPT-Code-Clippy (GPT-CC) is an open source version of GitHub Copilot, a language model

GPT-Code-Clippy (GPT-CC) is an open source version of GitHub Copilot, a language model -- based on GPT-3, called GPT-Codex -- that is fine-tuned on publicly available code from GitHub.

Nathan Cooper 2.3k Jan 1, 2023