|
--- |
|
library_name: transformers |
|
tags: |
|
- BBBP |
|
- SMILES |
|
- BPE Tokenizer |
|
- classification |
|
license: mit |
|
base_model: |
|
- mikemayuare/SMILYAPE |
|
--- |
|
|
|
# Model Card for mikemayuare/SMILY-BPE-BBBP |
|
|
|
This model is fine-tuned on the BBBP (Blood-Brain Barrier Penetration) dataset and is designed to classify chemical compounds based on their ability to penetrate the blood-brain barrier. The input to the model is in the SMILES (Simplified Molecular Input Line Entry System) molecular representation format. The model uses the BPE (Byte Pair Encoding) tokenizer for tokenizing the input. The model is intended for sequence classification tasks and should be loaded with the `AutoModelForSequenceClassification` class. Both the model and tokenizer can be loaded using the `from_pretrained` method from the Hugging Face Transformers library. |
|
|
|
## Model Details |
|
|
|
### Model Description |
|
|
|
This is a 🤗 transformers model fine-tuned on the BBBP dataset. It classifies chemical compounds as either penetrating or non-penetrating the blood-brain barrier. The model takes SMILES molecular representations as input and uses the BPE (Byte Pair Encoding) Tokenizer for tokenization. Both the model and the tokenizer can be loaded using the `from_pretrained` method from Hugging Face. |
|
|
|
- **Developed by:** Miguelangel Leon |
|
- **Funded by:** This work was supported by national funds through FCT (Fundação para a Ciência e a Tecnologia), under the project - UIDB/04152/2020 (DOI:10.54499/UIDB/04152/2020) - Centro de Investigação em Gestão de Informação (MagIC)/NOVA IMS). |
|
- **Model type:** Sequence Classification |
|
- **Language(s) (NLP):** Not applicable (SMILES molecular representation) |
|
- **License:** MIT |
|
- **Finetuned from model:** mikemayuare/SMILY-BPE-BBBP |
|
|
|
### Model Sources |
|
|
|
- **Paper :** Pending |
|
|
|
## Uses |
|
|
|
### Direct Use |
|
|
|
This model can be used directly for binary classification of chemical compounds to predict whether they penetrate the blood-brain barrier. The inputs must be formatted as SMILES strings. |
|
|
|
### Downstream Use |
|
|
|
This model can be further fine-tuned for other chemical classification tasks, particularly those that use molecular representations in SMILES format. |
|
|
|
### Out-of-Scope Use |
|
|
|
This model is not designed for tasks outside of chemical compound classification or tasks unrelated to molecular data (e.g., NLP). |
|
|
|
## Bias, Risks, and Limitations |
|
|
|
As this model is fine-tuned on the BBBP dataset, it may not generalize well to compounds outside the dataset’s chemical space. Additionally, it is not suited for use in applications outside of chemical compound classification tasks. |
|
|
|
### Recommendations |
|
|
|
Users should be cautious when applying this model to new chemical datasets that differ significantly from the BBBP dataset. Thorough evaluation on the target dataset is recommended before deployment. |
|
|
|
## How to Get Started with the Model |
|
|
|
To use the model for classification, it must be loaded with the `AutoModelForSequenceClassification` class from 🤗 transformers, and the tokenizer with the `AutoTokenizer` class from the same library. The inputs must be formatted as SMILES strings. |
|
|
|
You can load the BPE tokenizer and the model with the following steps: |
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForSequenceClassification |
|
|
|
# Load the tokenizer |
|
tokenizer = AutoTokenizer.from_pretrained("mikemayuare/SMILY-BPE-BBBP") |
|
|
|
# Load the model |
|
model = AutoModelForSequenceClassification.from_pretrained("mikemayuare/SMILY-BPE-BBBP") |
