Instructions to use fisherman611/vibi-appfilter-v2-c with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use fisherman611/vibi-appfilter-v2-c with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("fisherman611/vibi-appfilter-v2-c") sentences = [ "The weather is lovely today.", "It's so sunny outside!", "He drove to the stadium." ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] - Notebooks
- Google Colab
- Kaggle
vibi-appfilter-v2-c
Embedding lọc app cho LifeMate (Zalo AI Challenge 2026), fine-tune từ bkai-foundation-models/vietnamese-bi-encoder (135M). Cần pyvi để tách từ trước khi mã hoá.
Cấu hình C: B + 0.5·MNRL trong batch + 0.5·InfoNCE với 2 âm tính khó; epoch 1.
Recall (%) hybrid BM25@4 ∪ emb@5: val_in 98.2 / val_ood 98.5 / val_alias 93.6 (emb-only val_ood@5 88.8, không ổn định giữa các epoch). Chi tiết: train_log.json, train.log.
Dữ liệu: train của BTC + alias/iface/phonetic + bộ app tự viết; đã loại 5 app held-out (alarm, ewallet, flights, loyalty, podcast_radio). Không dùng nhãn hay câu của public/private test.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for fisherman611/vibi-appfilter-v2-c
Base model
bkai-foundation-models/vietnamese-bi-encoder