sewoong/korean-neural-sparse-encoder

docs: use English only in README (remove Korean descriptions)

docs: add fine-tuning toolkit section to README

Add fine-tuning toolkit: CONTRIBUTING.md

Add fine-tuning toolkit: requirements.txt

Add fine-tuning toolkit: finetune_tutorial.ipynb

Add fine-tuning toolkit: val.jsonl

Add fine-tuning toolkit: train.jsonl

Add fine-tuning toolkit: finetune.yaml

Add fine-tuning toolkit: prepare_data.py

Add fine-tuning toolkit: encode.py

Add fine-tuning toolkit: finetune.py

fix: correct author name to Sewoong Kim

docs: fix README with complete code examples and training details

docs: update README for V33-ecom-v6 (e-commerce fine-tuned)

feat: update to V33-ecom-v6 (e-commerce fine-tuned)

docs: professional model card with benchmarks, training data, and citation

Korean neural sparse encoder: SPLADE-max with ModernBERT backbone

docs: update README with V28 benchmark results

Update V28 model: model.safetensors

Remove CLAUDE.md

Update to V26: Recall@1 40.7% (+44% vs V25)

Fix: Remove version from model references

v22.0: InfoNCE contrastive loss, temperature annealing, 99.87% recall

initial commit