Instructions to use EddyIrasetsa/sti-qa-assistant with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use EddyIrasetsa/sti-qa-assistant with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("EddyIrasetsa/sti-qa-assistant") sentences = [ "That is a happy person", "That is a happy dog", "That is a very happy person", "Today is a sunny day" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4] - Notebooks
- Google Colab
- Kaggle
STI Q&A Assistant: model and answer files (student project)
Files used by the STI Q&A Assistant web app (https://github.com/Eddydev-ALU/ML_Technique_I_Summative_sti-qa-assistant).
model/: PubMedBERT fine-tuned from NeuML/pubmedbert-base-embeddings (Apache-2.0) with MultipleNegativesRankingLoss on (question, same-intent question) and (question, answer) pairs. Trained on all 2,502 questions of 109 intents for 4 epochs (batch 32, lr 2e-5, max length 256). Evaluation of a model trained the same way on the training split only: test accuracy@1 0.952, MRR 0.968.step2_model/: sentence-transformers/all-MiniLM-L6-v2 (Apache-2.0), unchanged.answer_bank.csv: the answers, and*.npz: embeddings (numbers only, no question text).
Data: Dataset Club, HASH (2026). The HASH Sexually Transmitted Infection Question and Answer Crowdsourced Dataset (Version 1). Harvard Dataverse. https://doi.org/10.7910/DVN/5SWM5U. Infectious Diseases Institute (Uganda), Makerere AI Lab and Sunbird AI. Licensed CC BY-NC 4.0, so these files are for non-commercial use only.
Educational information only, not a diagnosis. Please visit a health facility for testing and treatment.
- Downloads last month
- 22
Model tree for EddyIrasetsa/sti-qa-assistant
Finetuned
NeuML/pubmedbert-base-embeddings