Initial commit: Tucano2-Commerce GRPO v3 training pipeline

Domain-specialized LLM project for Brazilian e-commerce analysis.
Includes SFT→GRPO training notebook with W&B integration, project
documentation, and markdown-to-notebook converter.

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

.gitignore

@@ -0,0 +1,52 @@
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.egg-info/
+dist/
+build/
+*.egg
+
+# Jupyter
+.ipynb_checkpoints/
+
+# Virtual environments
+.venv/
+venv/
+env/
+
+# HuggingFace / model cache
+.cache/
+
+# Model weights & checkpoints (tracked via LFS if needed)
+*.safetensors
+*.bin
+*.pt
+*.pth
+*.ckpt
+saved_model/
+
+# Training outputs
+outputs/
+runs/
+results/
+checkpoints/
+
+# Weights & Biases
+wandb/
+
+# IDE / editor
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+
+# Claude Code
+.claude/
+
+# Environment / secrets
+.env
+.env.*
+*.key