kz930 opened a new issue, #6794:
URL: https://github.com/apache/texera/issues/6794
### What happened?
`HuggingFaceSentimentAnalysisOpDesc.generatePythonCode` emits an import for
`TFAutoModelForSequenceClassification` that is never used:
`common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceSentimentAnalysisOpDesc.scala`
```python
from transformers import pipeline
from transformers import AutoModelForSequenceClassification
from transformers import TFAutoModelForSequenceClassification # <-- never
used
from transformers import AutoTokenizer, AutoConfig
```
The generated body loads the model with the PyTorch
`AutoModelForSequenceClassification` (`self.model =
AutoModelForSequenceClassification.from_pretrained(...)`) and uses `pipeline`;
`TFAutoModelForSequenceClassification` is not referenced anywhere in the
generated code.
The `TF*` classes belong to `transformers`' TensorFlow backend, so importing
this symbol forces a TensorFlow dependency. In a PyTorch-only environment
(which is what the operator actually uses), `from transformers import
TFAutoModelForSequenceClassification` raises an ImportError/RuntimeError at
module load, **crashing the whole generated script** before any inference runs.
**Expected:** the generated code does not import unused TensorFlow-only
symbols.
### How to reproduce?
**A — user-facing:** Run a **Hugging Face Sentiment Analysis** operator in
an environment where `transformers` is installed with the PyTorch backend but
without TensorFlow. The generated script fails at import on the
`TFAutoModelForSequenceClassification` line.
**B — confirmed against `main`:** in `main`'s `generatePythonCode`,
`TFAutoModelForSequenceClassification` appears only on the import line (line
64) and is never referenced in the body (the model is loaded via
`AutoModelForSequenceClassification`, line 75).
**Proposed fix:** remove the unused import:
```
from transformers import pipeline
from transformers import AutoModelForSequenceClassification
-from transformers import TFAutoModelForSequenceClassification
from transformers import AutoTokenizer, AutoConfig
```
### Version/Branch
1.3.0-incubating-SNAPSHOT (main)
### Commit Hash (Optional)
_No response_
### What browsers are you seeing the problem on?
_No response_
### Relevant log output
```shell
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
