Besides the usual FP32, it supports FP16, quantized INT4, INT5 and INT8 inference. This project is focused on CPU, but cuBLAS is also supported. RWKV is a novel large language model architecture, with the largest model in the family having 14B parameters. In contrast to Transformer with O(n^2) attention, RWKV requires only state from the previous step to calculate logits. This makes RWKV very CPU-friendly on large context lengths.

Features

  • Windows / Linux / MacOS
  • Build the library yourself
  • Get an RWKV model
  • Requirements: Python 3.x with PyTorch and tokenizers
  • ggml moves fast, and can occasionally break compatibility with older file formats
  • Requirements: Python 3.x with PyTorch

Project Samples

rwkv.cpp Screenshot 1

Project Activity

See All Activity >

Categories

Large Language Models (LLM), AI Models, LLM Inference

License

MIT License

Follow rwkv.cpp

rwkv.cpp Web Site

Other Useful Business Software
Custom VMs From 1 to 96 vCPUs With 99.95% Uptime Icon
Custom VMs From 1 to 96 vCPUs With 99.95% Uptime

General-purpose, compute-optimized, or GPU/TPU-accelerated. Built to your exact specs.

Live migration and automatic failover keep workloads online through maintenance. One free e2-micro VM every month.
Start Free
Rate This Project
Login To Rate This Project

User Reviews

Be the first to post a review of rwkv.cpp!

Additional Project Details

Programming Language

C++

Related Categories

C++ Large Language Models (LLM), C++ AI Models, C++ LLM Inference Tool

Registered

2023-08-25
Report inappropriate content