Pinned Loading

1. SciNER SciNER Public

   SciERC NER — comparative study of named entity recognition approaches on scientific text. Implements 10 BIO token-level models (BERT, RoBERTa, SciBERT with Linear/MLP/CRF/Concat-4 heads) and 6 span…

   Python

2. text2sql text2sql Public

   Text-to-SQL over a competition-platform database: hybrid RAG (BM25 + FAISS + cross-encoder) with FK knowledge graph expansion, LLM judge, FastAPI service, and Textual TUI.

   Python

3. credit_scoring credit_scoring Public

   This project is a ML repository focused on credit default risk prediction. The current baseline uses a tuned LightGBM model, XGBoost, CatBoost.

   Python

4. neuroRap neuroRap Public

   Automated pipeline for building a GPT-2 fine-tuning dataset by collecting lyrics from Genius and extracting numerical audio features via yt-dlp/FFmpeg.

   Python