fchollet

8 ms·

31.
by fchollet 2y ago
Hi HN, Francois here. Happy to answer any questions! Here's a start -- "Did you get poached by Anthropic/etc": No, I am starting a new company with a friend. We will announce more about it in due time! "Who uses Ker
32.
Symbolic and Imperative APIs in TensorFlow 2.0(medium.com)
2 points by fchollet 8y ago|0 comments
33.
by fchollet 8y ago
If I were to divide large tech companies between "good" and "bad", the only one I would classify as "bad" is Facebook. Most are neutral-to-good. Facebook has a huge societal and human cost and has hardly an
34.
The Bitcoin Network at Risk(medium.com)
5 points by fchollet 8y ago|0 comments
35.
by fchollet 8y ago
> there isn't anyone living today that has any idea how to get to AGI Not true.
36.
Deep Reinforcement Learning: Playing CartPole with Tf.keras and Eager Execution(medium.com)
1 points by fchollet 8y ago|0 comments
37.
Google's guide to text classification(developers.google.com)
7 points by fchollet 8y ago|0 comments
38.
by fchollet 9y ago
When someone shows you who they are, believe them the first time.
39.
by fchollet 9y ago
> IMO, AssertionErrors should indicate bugs in the called API. ValueErrors indicate bugs in the caller's use of the API. Yes, I agree with this stance. `ValueError` should be used for user-provided input validation (as well as `Type
40.
by fchollet 9y ago
To clarify: `tf.keras` is an implementation of the entire Keras API written from the ground-up in pure TensorFlow. The first benefit of that is a greater level of blending between non-Keras-TF workflows and TF-Keras workflows: for instance,
41.
by fchollet 9y ago
To play with Capsule Networks in practice, you can try this simple Keras implementation: https://github.com/XifengGuo/CapsNet-Keras
42.
by fchollet 9y ago
But that is precisely how you should be using Keras! * If you are implementing a standard model (that's 90% of industry use cases, and a large fraction of research use cases as well), Keras primitives considerably simplify your workflo
43.
A ten-minute introduction to sequence-to-sequence learning in Keras(blog.keras.io)
1 points by fchollet 9y ago|0 comments
44.
A ten-minute introduction to sequence-to-sequence learning in Keras(blog.keras.io)
6 points by fchollet 9y ago|0 comments
45.
by fchollet 9y ago
CoreML supports Keras but not TensorFlow because Keras models form a well-structured subset of all possible TensorFlow graphs. It would be quite difficult to support completely arbitrary TensorFlow graphs, but supporting every Keras layer i
46.
by fchollet 9y ago
It was always possible to train Keras models in a distributed setting (I was doing it in late 2015). And there's built-in, one-line integration with the Estimator API coming in the next version of TensorFlow.
47.
by fchollet 9y ago
Hardly a "made-up" conclusion -- just a teaser for the next post, which deals with how we can achieve "extreme generalization" via abstraction and reasoning, and how we can concretely implement those in machine learning
48.
by fchollet 9y ago
How is that the case in this specific example? It looks a lot harder to check for correctness.
49.
by fchollet 9y ago
Great initiative! Always good to see more support for open-source and for AI tooling.
50.
by fchollet 10y ago
See the release notes: https://github.com/fchollet/keras/wiki/Keras-2.0-release-not... - You can use a Keras model to compute some tensor(s), turn that into a loss, and manually add that loss to the model via
51.
by fchollet 10y ago
You can try opening an issue on Github. `val_acc` is definitely still accessible by callbacks, and the `EarlyStopping` callback, which relies on it, is fully unit-tested.
52.
by fchollet 10y ago
With the functional API of Keras, it would definitely make sense. In fact I do think that imperative model definition would be great to have at some point in the future. We'll see :)
53.
by fchollet 10y ago
To put this quote in context: this isn't specifically about PyTorch. Every couple of months since mid-2015, a new deep learning framework gets released. In the following week, someone inevitably asks "will X get added as a Keras b
54.
Introducing Keras 2.0(blog.keras.io)
12 points by fchollet 10y ago|2 comments
55.
Hack This: Make a Photo Filter with Machine Learning(motherboard.vice.com)
1 points by fchollet 10y ago|0 comments
56.
Hierarchical Object Detection with Deep Reinforcement Learning(imatge-upc.github.io)
77 points by fchollet 10y ago|5 comments
57.
by fchollet 10y ago
It would be easy to replace the input features with representations from a pixel-level convnet, to make this is a "real" self-driving car, going from pixels to commands. Anyone interested in this type of research: consider cloning
58.
by fchollet 10y ago
The equation should have per-subject weights to modulate the "how many people" part. For instance, if my app helps Mr. Musk be 10% more productive, that's more impactful than helping Mr. Smith be 10% more productive.
59.
Implementation of Neural Style Transfer in Python and Keras(github.com)
62 points by fchollet 10y ago|7 comments
60.
Keras Applications: collection of trained deep learning models(keras.io)
4 points by fchollet 10y ago|0 comments

More ›