Mirrors_Framework/deepmind-research
1
0
Fork 0
mirror of https://github.com/google-deepmind/deepmind-research.git synced 2025-12-10 15:00:43 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Update enformer-training.ipynb to show how to restore from checkpoint.

Browse Source 
PiperOrigin-RevId: 450894644
This commit is contained in:
Kyle Taylor
2022-05-25 12:37:13 +01:00
committed by alimuldal
parent 57456a001d
commit 11c2ab53e8
2 changed files with 625 additions and 319 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
enformer/README.md
View File

@@ -120,8 +120,9 @@ df_targets.shape # (5313, 8) With rows match output shape above.
The model is implemented using [Sonnet](https://github.com/deepmind/sonnet). The The model is implemented using [Sonnet](https://github.com/deepmind/sonnet). The
full sonnet module is defined in `enformer.py` called Enformer. See full sonnet module is defined in `enformer.py` called Enformer. See
[enformer-training.ipynb](https://colab.research.google.com/github/deepmind/deepmind_research/blob/master/enformer/enformer-training.ipynb). [enformer-training.ipynb](https://colab.research.google.com/github/deepmind/deepmind_research/blob/master/enformer/enformer-training.ipynb)
on how to train the model on Basenji2 data. on how to train the model on Basenji2 data and how to load the pre-trained
weights into the Enformer module for fine-tuning.
## Colab ## Colab
@@ -143,6 +144,7 @@ This colab shows how to:
* Setup training data by directly accessing the Basenji2 data on GCS * Setup training data by directly accessing the Basenji2 data on GCS
* Train the model for a few steps on both human and mouse genomes * Train the model for a few steps on both human and mouse genomes
* Evaluate the model on human and mouse genomes * Evaluate the model on human and mouse genomes
* Restore the model from a checkpoint for fine-tuning
## Disclaimer ## Disclaimer

938
enformer/enformer-training.ipynb
View File

File diff suppressed because it is too large Load Diff