FastAI.jl: FastAI for Julia • Hacker News Board

Re: FastAI.jl: FastAI for Julia

#21

Post by xvilka » Wed, Jul 28, 2021, 10:01 AM UTC

darsnack wrote: post #16

ellisv wrote: post #2

This is interesting to me but the motivation behind this is unclear. Since FastAI.jl uses Flux, and not PyTorch, functionality has to be reimplemented. FastAI.jl has vision support but no text support yet. What does this mean for the development of fastai? What is the timeline for FastAI.jl to achieve parity? When should I choose FastAI.jl vs fastai?

I’m not the main dev on FastAI.jl, but I work on the Julia ML community team that supported this project. > Since FastAI.jl uses Flux, and not PyTorch, functionality has to be reimplemented. We are looking to offer a high level API for ML in Julia similar to fastai for PyTorch. The goal is to enrich the Flux ecosystem, so just calling into Python fastai wouldn’t be appropriate. FastAI.jl is built on top of several lo…

I didn't check the library yet, asking here in case of the quick answer - does it support GNN too via GeometricFlux.jl[1]?

[1] https://github.com/FluxML/GeometricFlux.jl

Re: FastAI.jl: FastAI for Julia

#22

Post by darsnack » Wed, Jul 28, 2021, 12:14 PM UTC

xvilka wrote: post #21

darsnack wrote: post #16

Earlier quoted context omitted.

I’m not the main dev on FastAI.jl, but I work on the Julia ML community team that supported this project. > Since FastAI.jl uses Flux, and not PyTorch, functionality has to be reimplemented. We are looking to offer a high level API for ML in Julia similar to fastai for PyTorch. The goal is to enrich the Flux ecosystem, so just calling into Python fastai wouldn’t be appropriate. FastAI.jl is built on top of several lo…

I didn't check the library yet, asking here in case of the quick answer - does it support GNN too via GeometricFlux.jl[1]? [1] https://github.com/FluxML/GeometricFlux.jl

You would have to add a learning method to tell it how to encode/decode graph data, but the framework is agnostic to the model choice. So any Flux model is supported.

Re: FastAI.jl: FastAI for Julia

#23

Post by xvilka » Wed, Jul 28, 2021, 1:17 PM UTC

darsnack wrote: post #22

xvilka wrote: post #21

Earlier quoted context omitted.

I didn't check the library yet, asking here in case of the quick answer - does it support GNN too via GeometricFlux.jl[1]? [1] https://github.com/FluxML/GeometricFlux.jl

You would have to add a learning method to tell it how to encode/decode graph data, but the framework is agnostic to the model choice. So any Flux model is supported.

Thanks! Understood.

Re: FastAI.jl: FastAI for Julia

#24

Post by BadInformatics » Wed, Jul 28, 2021, 2:58 PM UTC

WanderPanda wrote: post #19

darsnack wrote: post #16

Earlier quoted context omitted.

I’m not the main dev on FastAI.jl, but I work on the Julia ML community team that supported this project. > Since FastAI.jl uses Flux, and not PyTorch, functionality has to be reimplemented. We are looking to offer a high level API for ML in Julia similar to fastai for PyTorch. The goal is to enrich the Flux ecosystem, so just calling into Python fastai wouldn’t be appropriate. FastAI.jl is built on top of several lo…

It is not surprising to me that there is no text support yet, given the issues of integrating 3D (batch x sequence x features) RNN support (for CUDNN). This issue has prevailed so long that I came to believe it is impossible to integrate with the current Flux.jl stack.

Transformers.jl and TextAnalysis.jl already provide quite a bit of functionality for NLP, though to my knowledge neither makes use of RNNs. You may be interested in commenting on https://github.com/FluxML/Flux.jl/issues/1678.