From 0fd2a7f53258f1ab2dd27d01ed8c60f45059670b Mon Sep 17 00:00:00 2001 From: Yaman Umuroglu <yamanu@xilinx.com> Date: Wed, 2 Oct 2019 17:07:50 +0200 Subject: [PATCH] Update index.md --- docs/index.md | 18 ++++++------------ 1 file changed, 6 insertions(+), 12 deletions(-) diff --git a/docs/index.md b/docs/index.md index 8213e6b63..d8025be0a 100644 --- a/docs/index.md +++ b/docs/index.md @@ -11,23 +11,17 @@ It is not intended to be a generic DNN accelerator like xDNN, but rather a tool for exploring the design space of DNN inference accelerators on FPGAs. <br><br> -A new, more modular version of FINN is currently under development on GitHub, and we welcome contributions from the community! -<br> +A new, more modular version of FINN is currently under development <a href="https://github.com/Xilinx/finn">on GitHub</a>, and we welcome contributions from the community! + ## Quickstart -Depending on what you would like to do, we have -different suggestions on where to get started: +Depending on what you would like to do, we have different suggestions on where to get started: -* **I want to try out premade accelerators on real hardware.** Head over to <a href="https://github.com/Xilinx/BNN-PYNQ" target="_blank">BNN-PYNQ</a> repository to try out some image +* **I want to try out prebuilt QNN accelerators on real hardware.** Head over to <a href="https://github.com/Xilinx/BNN-PYNQ" target="_blank">BNN-PYNQ</a> repository to try out some image classification accelerators, or to <a href="https://github.com/Xilinx/LSTM-PYNQ" target="_blank">LSTM-PYNQ</a> to try optical character recognition with LSTMs. -* **I want to try the full design flow.** The <a href="https://github.com/Xilinx/FINN" target="_blank">FINN</a> repository -contains the Python toolflow that goes from a trained, quantized Caffe network -to an accelerator running on real hardware. -* **I want to train new quantized networks for FINN.** Have a look <a href="https://github.com/Xilinx/BNN-PYNQ/tree/master/bnn/src/training" target="_blank">here</a>, at -[this presentation](https://drive.google.com/open?id=17oorGvtUbdFd-o1OzSuxGCSrWsvm_S2ftC1UC2FLtuE) -for an example with Fashion-MNIST, or <a href="https://github.com/Xilinx/pytorch-ocr" target="_blank">here</a> for quantized -LSTMs with PyTorch. +* **I want to train new quantized networks for FINN.** Check out <a href="https://github.com/Xilinx/brevitas">Brevitas</a>, +our PyTorch library for training quantized networks. The Brevitas-to-FINN part of the flow is coming soon! * **I want to understand how it all fits together.** Check out our [publications](#publications), particularly the <a href="https://arxiv.org/abs/1612.07119" target="_blank">FINN paper at FPGA'17</a> and the <a href="https://arxiv.org/abs/1809.04570" target="_blank">FINN-R paper in ACM TRETS</a>. -- GitLab