Doc updates to switch branding to Keras 3 #1259
Conversation
mattdangerw
changed the title
| [`Models`](https://keras.io/api/models/) that receive that same level of polish | ||
| as core Keras. If you are familiar with Keras, congratulations! You already | ||
| understand most of KerasNLP. | ||
| KerasNLP is a natural language processing library that works natively |
There was a problem hiding this comment.
I removed trailing spaces, but I think that is making this diff look a lot bigger than it is.
There was a problem hiding this comment.
Yeah, there's something crazy going on with the diff here. Even identical paragraphs are marked as deleted.
| [`keras_nlp.models`](https://keras.io/api/keras_nlp/models/) API: | ||
|
|
||
| ```python | ||
| import os | ||
| os.environ["KERAS_BACKEND"] = "jax" # Or "tensorflow", or "torch". |
There was a problem hiding this comment.
All our guides and examples show setting KERAS_BACKEND, it seems consistent to do so here as well.
| using Keras Core with the backend specified (e.g., `KERAS_BACKEND=jax`). | ||
| 2. Via the `.keras/keras.json` and `.keras/keras_nlp.json` config files (which | ||
| are automatically created the first time you import KerasNLP): | ||
| - Set your backend of choice in `.keras/keras.json`; e.g., `"backend": "jax"`. |
There was a problem hiding this comment.
I've dropped any mention of .keras json files for now. There is too much to focus on!
Let's only document KERAS_BACKEND environment variable for now, and try to make that less confusing. After Keras 3 is out, we can go back to documenting persistent settings.
There was a problem hiding this comment.
This is a good direction! Some high level thoughts:
from keras_cv.backend import kerasseems less confusing that knowing toimport keras_coreand makes scripts less portable. I wonder if this remains a good intermediate approach.- Make sure to update keras.io landing pages
- Please coordinate with @ianstenbit to keep KerasCV and KerasNLP documentation aligned where possible.
| [`Models`](https://keras.io/api/models/) that receive that same level of polish | ||
| as core Keras. If you are familiar with Keras, congratulations! You already | ||
| understand most of KerasNLP. | ||
| KerasNLP is a natural language processing library that works natively |
There was a problem hiding this comment.
Yeah, there's something crazy going on with the diff here. Even identical paragraphs are marked as deleted.
This is an option! Thought about it too. But I think because of our explicit API export we would need to cut a new release for this. So slightly more work. |
I've opened a PR on KerasCV to make ~the same changes. I've updated the important bit about importing Keras to use """ Important If you want to import Keras and guarantee that you're getting the same version |
mattdangerw
force-pushed
the
doc-update
branch
from
15c15f6 to
6fab678
Compare
|
Updated this with the small comments. I think this can probably land whenever. Though #1258 is blocked on tensorflow-text fixing it's nightly build. |
Make some updates to KerasNLP to reflect our changing branding, Keras 3 not Keras Core.
View the changes rendered here:
https://github.com/mattdangerw/keras-nlp/tree/doc-update