app.py

from strive.reranker import Reranker, EmbeddingType, deduplicate_results
import gradio as gr

def rerank_and_deduplicate(query, corpus_text):
    corpus = [line.strip() for line in corpus_text.split("\n") if line.strip()]
    textual_reranker = Reranker(embedding_type=EmbeddingType.textual)
    semantic_reranker = Reranker(embedding_type=EmbeddingType.semantic)
    
    textual_results = textual_reranker.rerank_documents(query, corpus, top_k=len(corpus))
    semantic_results = semantic_reranker.rerank_documents(query, corpus, top_k=len(corpus))
    
    merged_results = textual_results + semantic_results
    deduplicated_results = deduplicate_results(merged_results, top_k=50)
    
    return "\n".join([f"{text} (Score: {score:.4f})" for text, score in deduplicated_results])

app = gr.Interface(
    fn=rerank_and_deduplicate,
    inputs=[
        gr.Textbox(label="Query", placeholder="Enter your query here"),
        gr.Textbox(label="Corpus", placeholder="Enter one sentence per line", lines=10)
    ],
    outputs=gr.Textbox(label="Top Ranked Results"),
    title="STRIVE: Semantic Tokenized Ranking via Vectorization & Embeddings",
    description="Enter a query and multiple sentences to test the reranking algorithm."
)

app.launch()