Note to self: #NVIDIA have an open-source inference server for machine learning models. (They mostly sell SaaS on top of it)

Supports #TensorFlow, #PyTorch, #ONNX, #TensorRT, #mxnet.

Runs on #k8s. Features queue control, monitoring.

Triton Inference Server https://github.com/triton-inference-server

MXNet doesn't look difficult...but I see two mayor issues:
a) Community yet small
b) Lack of pretrained models for transfer learning. A big hit would be support by Hugginface models and pretrained models with ImageNet.

#MXNet #DeepLearning

(1/n)𝐍𝐞𝐰 𝐫𝐞𝐥𝐞𝐚𝐬𝐞 𝐟𝐨𝐫 𝐭𝐡𝐞 𝐃𝐢𝐯𝐞 𝐢𝐧𝐭𝐨 𝐃𝐞𝐞𝐩 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐛𝐨𝐨𝐤! 🎊🎉

The 𝐃𝐢𝐯𝐞 𝐢𝐧𝐭𝐨 𝐃𝐞𝐞𝐩 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 book is an educational project by Amazon's scientists and other contributors focusing on deep learning models and their applications. It is interactive and available online for 𝐟𝐫𝐞𝐞 and implemented with multiple deep learning frameworks such as 𝐏𝐲𝐓𝐨𝐫𝐜𝐡, 𝐍𝐮𝐦𝐏𝐲/𝐌𝐗𝐍𝐞𝐭, 𝐆𝐨𝐨𝐠𝐥𝐞 𝐉𝐀𝐗, and 𝐓𝐞𝐧𝐬𝐨𝐫𝐅𝐥𝐨𝐰 🚀.

#deeplearning #neuralnetworks #machinelearning #pytorch #python #nlp #computervision #tensorflow #mxnet #jax

Alright ! I got #mxnet for #cplusplus to compile from scratch without any effort.

```
$ mkdir -p build && cd build

$ cmake -DUSE_CPP_PACKAGE=1 -DUSE_CUDA=OFF ..

$ make -j4

```
Wait almost 45 minutes and it is done.
Next step is to try out their tutorial examples.

Machine Learning with MXNet.cr

https://diode.zone/videos/watch/e2b0ef27-ca55-40c8-bd54-306c2fbc451a

cleaning up the mxnet.cr exercises.

also demonstrated above is ishi—which tries to be a mathplotlib.pyplot like wrapper on top of gnuplot albeit written in crystal—and iterm2—which outputs charts to the console.

#mxnet #ishi #iterm2