Add example for OPT model with distribution. #1727
Draft
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
fchollet
reviewed
Jan 16, 2024
| support in the coming future. | ||
| """ | ||
|
|
||
| import os |
There was a problem hiding this comment.
Please group the imports at the top.
| print(keras.version()) | ||
| print(keras.backend.backend()) | ||
|
|
||
| keras.mixed_precision.set_global_policy("mixed_float16") |
There was a problem hiding this comment.
Please add a comment about using mixed precision.
| count other items like optimizer states, as well as forward and backward path. | ||
| """ | ||
| # model_spec = 'opt_6.7b_en' | ||
| # langauge_model = create_opt_model(model_spec) |
|
|
||
| # Create a 2D mesh for model parallel, change the mesh shape to tune the | ||
| # ratio of data/model parallelism | ||
| _BATCH_DIM_NAME = "batch" |
There was a problem hiding this comment.
No need for leading underscores
| generate function with XLA. The follow up runs will be much faster. | ||
| """ | ||
| prompt = "What is machine learning?" | ||
| print(large_model.generate(prompt)) |
There was a problem hiding this comment.
Please add a second prompt, possible with some time() calls, to demonstrate the regular (post compilation) step time
|
|
||
|
|
||
| """ | ||
| ## Introduction to KerasNLP |
There was a problem hiding this comment.
I suggest focusing the example purely on the distribution aspects, so we can replace the KerasNLP intro with ~1 sentence. Meanwhile maybe we could flesh out the distribution part, e.g. include fine-tuning or other inference performance considerations.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
@fchollet, this is the draft of the OPT model inferencing with Keras distribution API (I can add the finetune parts later). Do u still have the instructions to convert the py example to colab/MD file?
This one will require 8 V100 GPU to simulate, and I think we should have A100 to properly simulate a finetuning workflow