Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ๐ Semantic Search Engine (Nexus)
|
| 2 |
+
|
| 3 |
+
This repository contains model artifacts for a hybrid semantic search engine.
|
| 4 |
+
|
| 5 |
+
## ๐ฆ Files
|
| 6 |
+
- `faiss_index.bin` โ Dense vector index (FAISS)
|
| 7 |
+
- `corpus.pkl` โ Document corpus (text, title, URL)
|
| 8 |
+
- `bm25.pkl` โ Sparse BM25 index
|
| 9 |
+
|
| 10 |
+
## โ๏ธ Architecture
|
| 11 |
+
- Dense Retrieval: msmarco-bert-base-dot-v5
|
| 12 |
+
- Sparse Retrieval: BM25
|
| 13 |
+
- Reranker: cross-encoder/ms-marco-MiniLM-L-12-v2
|
| 14 |
+
|
| 15 |
+
## ๐ Dataset
|
| 16 |
+
- Wikipedia (100k articles)
|
| 17 |
+
|
| 18 |
+
## ๐ Use Case
|
| 19 |
+
Hybrid semantic + keyword search (Google-like engine)
|