TensorFlow Datasets

#1

Post by markerbrod » Wed, Dec 21, 2022, 12:19 PM UTC

Re: TensorFlow Datasets

#2

Post by xnx » Wed, Dec 21, 2022, 1:45 PM UTC

Great resource. My experience has been that any data project is at least 1/3 data collection/preparation, 1/3 using the right tool the right way, and 1/3 asking the right questions and interpreting the outcome.

Re: TensorFlow Datasets

#3

Post by jamesblonde » Wed, Dec 21, 2022, 2:32 PM UTC

I actually prefer Hugging Face Datasets - there's 16k+ of them today: https://huggingface.co/datasets

Re: TensorFlow Datasets

#4

Post by soraki_soladead » Wed, Dec 21, 2022, 3:05 PM UTC

jamesblonde wrote: post #3

I actually prefer Hugging Face Datasets - there's 16k+ of them today: https://huggingface.co/datasets

Quantity of datasets doesn’t seem like the right metric. The library just needs the datasets you care about and both libraries have the popular ones. What’s more important is integration and if you’re training custom TF models then tfds will generally integrate more smoothly than huggingface.

Re: TensorFlow Datasets

#5

Post by jachian » Wed, Dec 21, 2022, 4:21 PM UTC

soraki_soladead wrote: post #4

jamesblonde wrote: post #3

I actually prefer Hugging Face Datasets - there's 16k+ of them today: https://huggingface.co/datasets

Quantity of datasets doesn’t seem like the right metric. The library just needs the datasets you care about and both libraries have the popular ones. What’s more important is integration and if you’re training custom TF models then tfds will generally integrate more smoothly than huggingface.

and as well as discoverability / searchability. how easily it is to find what you're looking for

Re: TensorFlow Datasets

#6

Post by AyyWS » Wed, Dec 21, 2022, 4:29 PM UTC

Kaggle is what they talk about in my industry.

https://www.kaggle.com/datasets

Re: TensorFlow Datasets

#7

Post by yeldarb » Wed, Dec 21, 2022, 4:45 PM UTC

For computer vision, there are 100k+ open source classification, object detection, and segmentation datasets available on Roboflow Universe: https://universe.roboflow.com

Re: TensorFlow Datasets

#8

Post by dbish » Wed, Dec 21, 2022, 5:16 PM UTC

jachian wrote: post #5

soraki_soladead wrote: post #4

Earlier quoted context omitted.

Quantity of datasets doesn’t seem like the right metric. The library just needs the datasets you care about and both libraries have the popular ones. What’s more important is integration and if you’re training custom TF models then tfds will generally integrate more smoothly than huggingface.

and as well as discoverability / searchability. how easily it is to find what you're looking for

HF datasets have a much better UX for this

Re: TensorFlow Datasets

#9

Post by pj_mukh » Wed, Dec 21, 2022, 6:56 PM UTC

Direct link to the list of datasets: https://www.tensorflow.org/datasets/catalog/overview#all_dat...

Would love a direct Google Photos style search method for especially the visual datasets.

Re: TensorFlow Datasets

#10

Post by soraki_soladead » Wed, Dec 21, 2022, 7:10 PM UTC

dbish wrote: post #8

jachian wrote: post #5

Earlier quoted context omitted.

and as well as discoverability / searchability. how easily it is to find what you're looking for

HF datasets have a much better UX for this

UX preferences vary. Imo, hf is too verbose and their pages try to cram in too much information with poor information hierarchy. For example:

https://huggingface.co/datasets/glue

https://www.tensorflow.org/datasets/catalog/glue