Initial commit: caption quality test on UAV-VisLoc

Self-contained experimental track validating generated text captions
via retrieval R@1 lift on UAV-VisLoc.

Architecture: GeoRSCLIP ViT-B/32 dual encoder, 512-dim shared space.
Loss: 4-term InfoNCE (img-img + sat-cap + drone-cap + cap-cap)
      with cosine temperature decay, PALW-like curriculum.
Metric: delta R@1 (with text - without text) >= +3% => PASS.

Gin-configured (balanced / baseline_no_text / text_heavy variants).
Follows NADEZHDA code style.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.gitignore

@@ -0,0 +1,55 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+.venv/
+venv/
+env/
+
+# PyTorch / ML artefacts
+*.pt
+*.pth
+*.ckpt
+*.onnx
+*.engine
+checkpoints/
+out/
+outputs/
+runs/
+wandb/
+lightning_logs/
+
+# Data (too large / confidential — keep manifests only if < 10 MB)
+data/
+*.json.gz
+*.h5
+*.hdf5
+*.parquet
+*.npy
+*.npz
+
+# Gin / config logs
+*.gin.log
+operative_config-*.gin
+
+# Editors / IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+Thumbs.db
+
+# Secrets (never commit)
+.env
+.env.*
+*credentials*
+*secret*
+*.pem
+*.key