ESM2 Golden Value Testing #85
Merged
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
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
jstjohn
reviewed
Aug 9, 2024
sub-packages/bionemo-esm2/tests/bionemo/esm2/model/test_model_gv.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 9, 2024
sub-packages/bionemo-esm2/tests/bionemo/esm2/model/test_model_gv.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 9, 2024
Comment on lines
162
to
164
| tokens = tokenizer.tokenizer([row[1] for row in sample_data], return_tensors="pt", padding=True) | ||
| tokens["input_ids"] = tokens["input_ids"].to(device) | ||
| tokens["attention_mask"] = tokens["attention_mask"].to(device) |
There was a problem hiding this comment.
Would be more clear if we called this inputs or batch or something. Later when we say **tokens it looks like we're just unpacking tokens for a second. Also doesn't the return object from hugging face allow you to say batch.to(device)? It's like a special dictionary type that has a to method defined I think. If so that would be nicer to use.
jstjohn
reviewed
Aug 9, 2024
sub-packages/bionemo-esm2/tests/bionemo/esm2/model/test_model_gv.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 9, 2024
| model = esm2_650M_config_hiddens.configure_model(tokenizer).to(device) | ||
| model.eval() | ||
| hiddens = model(tokens["input_ids"], tokens["attention_mask"]) | ||
| embeddings = reduce_hiddens(torch.transpose(hiddens, 0, 1).float(), tokens["attention_mask"]) |
There was a problem hiding this comment.
Someday we should probably do this transpose inside of forward if the user wants this as their final output. Fine for now.
jstjohn
reviewed
Aug 9, 2024
sub-packages/bionemo-esm2/tests/bionemo/esm2/model/test_model_gv.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 9, 2024
sub-packages/bionemo-esm2/tests/bionemo/esm2/model/test_model_gv.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 9, 2024
Comment on lines
34
to
43
| bionemo2_root: Path = ( | ||
| # esm2 module's path is the most dependable --> don't expect this to change! | ||
| Path(esm2.__file__) | ||
| # This gets us from 'sub-packages/bionemo-esm2/src/bionemo/esm2/__init__.py' to 'sub-packages/bionemo-esm2' | ||
| .parent.parent.parent.parent | ||
| # From here, we want to get to the root of the repository: _before_ sub-packages/ | ||
| .parent.parent | ||
| ).absolute() | ||
| assert bionemo2_root != Path("/") | ||
| nemo1_checkpoint_path: Path = bionemo2_root / "models/protein/esm2nv/esm2nv_650M_converted.nemo" |
There was a problem hiding this comment.
Consider moving this into a conftest.py file in the top level esm dir. I think @skothenhill-nv did this for geneformer but my internet is acting up so I am not sure. That would allow you to use this in other tests as we grow the package.
There was a problem hiding this comment.
@jstjohn @skothenhill-nv
Geneformer is using the same pattern here. Am I looking at the right test file?
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
farhadrgh
requested review from
pstjohn,
gwarmstrong,
skothenhill-nv and
jstjohn
pstjohn
approved these changes
Aug 12, 2024
|
/build-ci |
jstjohn
reviewed
Aug 13, 2024
sub-packages/bionemo-llm/src/bionemo/llm/model/biobert/transformer_specs.py
Show resolved
Hide resolved
jstjohn
reviewed
Aug 13, 2024
sub-packages/bionemo-llm/src/bionemo/llm/model/biobert/transformer_specs.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Aug 13, 2024
jstjohn
approved these changes
Aug 13, 2024
|
/build-ci |
1 similar comment
|
/build-ci |
|
/build-ci |
|
/build-ci |
|
/build-ci |
|
/build-ci |
|
/build-ci |
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.
No description provided.