#
miracl
Here are 7 public repositories matching this topic...
Adaptive hybrid retrieval: per-query BM25 + dense fusion via QDAP (Query-Driven Alpha Prediction). PyTorch implementation with antagonist negative sampling, evaluated on MLDR & MIRACL (29 languages).
search-engine machine-learning natural-language-processing information-retrieval deep-learning transformers pytorch question-answering semantic-search bm25 miracl neural-information-retrieval contrastive-learning multilingual-nlp hybrid-search dense-retrieval retrieval-augmented-generation mteb research-code make-2026
-
Updated
May 24, 2026 - Python
Three .NET embedding libraries measured against a sentence-transformers reference for all-MiniLM-L6-v2. All three diverge, and every divergence is in tokenization.
dotnet embeddings reproducibility tokenization onnx msmarco wordpiece miracl sentence-transformers semantic-kernel
-
Updated
Sep 23, 2026 - Python
Reproducible MIRACL lexical-search benchmarks to measure the benefits that the Bitext English Analyzer adds to Elasticsearch
search nlp elasticsearch benchmark information-retrieval german linguistics stemming lemmatization miracl hybrid-search lexical-search
-
Updated
Oct 5, 2026 - Python
Archived: merged into signal-rag/evals/retrieval/arabic.
-
Updated
Jun 29, 2026 - Python
Add this topic to your repo
To associate your repository with the miracl topic, visit your repo's landing page and select "manage topics."