Future MAX and DAX assets

[ckadner]

List of MAX models to be migrated into MLX:

• CodeNet code complexity analysis (currently prototype)
• ...

List of DAX datasets to be integrated into MLX:

• FinTabNet - 15 GB
• Genomics (2 datasets) - 80 GB -> need to confirm if we need this
• ...

@kmh4321

[ckadner]

@kmh4321 -- could you provide a tentative list of models and datasets you are planning to integrate into the MLX catalog?

[kmh4321]

We have all the required MAX models on MLX.

For DAX we propose the following datasets be added to the Katalog:

Must haves (either research commitments or DAX is the only home for these datasets currently):

FinTabNet
COVID Question Answers
Oil Reservoir Simulations
Split and Rephrase
Expert in the Loop AI
Taranaki Basin Curated Well Logs
D2A
Genomics BLAST Indices (can revisit with author if they can host it elsewhere as well)
PRROMenade (can revisit with author if they can host it elsewhere as well)

Nice to haves (these datasets have good numbers/metrics but can also be found outside of DAX):

Wildfire dataset ( was part of Call for Code earlier)
Fashion MNIST
Airline on time performance dataset

[ckadner]

@kmh4321 -- Regarding missing MAX models, I think we are missing the Audio Classifier still. Did you have a PR for that prepared already? If not I can create one for it?

@Tomcli @animeshsingh -- Do you recall if we deliberately chose not to include the Audio Classifier in the MLX Katalog?

FYI @michaelhind

[Tomcli]

@ckadner Audio Classifier is not in the initial 10 MAX examples that we want to maintain, and it's no long part of the maintained models?

[kmh4321]

Hi @ckadner,

I don't have that ready yet. I will be happy to put in the PR over the long weekend or else I'd be happy to review if you plan to create it.