Danroy
/

MamaFrontida

@@ -22,62 +22,8 @@ The `t5-small` model, developed by Google and fine-tuned by the Frontida team, s
 - **Paper:** "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer" (Raffel et al., 2019)
 - **Demo:** Frontida Chatbot Interface (link to demo if available)
-## Uses
-### Direct Use
-Frontida's `t5-small` model is directly used within our chatbot interface to provide immediate, contextually relevant support and video recommendations for mothers experiencing postpartum depression.
-### Downstream Use
-While primarily designed for direct interaction within Frontida, the model's applications can extend to other mental health support systems, offering a foundation for empathetic, AI-driven conversation.
-### Out-of-Scope Use
-The model is not intended for clinical diagnosis or as a substitute for professional healthcare advice.
-## Bias, Risks, and Limitations
-We acknowledge the potential for biases in AI models and have taken steps to mitigate such risks in `t5-small`. However, users should be aware of the model's limitations, particularly in understanding the full scope of an individual's emotional state.
-### Recommendations
-Users are encouraged to use Frontida as a supplementary support tool alongside traditional mental health resources. Ongoing model training and refinement are priorities to ensure the most empathetic and accurate responses.
-## How to Get Started with the Model
-To interact with Frontida's `t5-small` model, users can access our chatbot via the Frontida web application. Developers interested in exploring the model's architecture and training can visit our repository on Hugging Face.
-## Training Details
-### Training Data
-The model was fine-tuned on a curated dataset comprising diverse conversations and texts related to mental health, specifically postpartum depression, ensuring a wide range of scenarios and emotions are covered.
-### Training Procedure
-#### Preprocessing
-Text data was normalized and tokenized using standard NLP preprocessing techniques to ensure consistency and improve model understanding.
-#### Training Hyperparameters
-- Training regime details are provided in the repository, focusing on optimizing performance while maintaining the model's efficiency.
-## Evaluation
-### Testing Data, Factors & Metrics
-Evaluation was conducted using a separate test set, focusing on accuracy, empathy in responses, and relevance of video recommendations.
-## Environmental Impact
-Efforts were made to minimize the carbon footprint during training, with details on compute usage and emissions available upon request.
-## Technical Specifications
-Further details on the model's architecture, objective, and compute infrastructure are available in the Frontida repository.
-## More Information
-For additional details, including how to contribute to the model's development or integrate it into other applications, please visit the Frontida project page on Hugging Face.

 - **Paper:** "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer" (Raffel et al., 2019)
 - **Demo:** Frontida Chatbot Interface (link to demo if available)
+### Team
+- **Danroy Mwangi** - Team Lead and NLP Lead
+- **Maria Muthiore** - Backend Lead
+- **Nelson Kamau** - Frontend Lead