Update README.md

README.md

@@ -1,15 +1,32 @@
 ---
-title: Afrobr Langbench Eval
-emoji: 💻
+title: AfroBR-LangBench Eval
+emoji: 🌍
 colorFrom: blue
-colorTo: purple
+colorTo: indigo
 sdk: gradio
-sdk_version: 6.17.3
-python_version: '3.12'
+sdk_version: 4.44.0
+python_version: '3.10'
 app_file: app.py
 pinned: false
 license: cc-by-4.0
 short_description: 'AfroBR-LangBench: base vs adapted · sociolinguistic safety'
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# AfroBR-LangBench — Evaluation Space
+
+Side-by-side evaluation of base vs fine-tuned model on Afro-Brazilian Portuguese sociolinguistic safety.
+
+| | Base | Adapted |
+|---|---|---|
+| **Model** | Llama-4-Scout-17B-16E-Instruct | AfroBR-LangBench LoRA (AutoScientist) |
+| **Training** | — | 400 examples · 10 phenomena · Grade A |
+
+**7-dimension rubric:** phenomenon identification · non-stigmatizing explanation · respectful normalization · terminology accuracy · source grounding · meaning preservation · hallucination control
+
+**Secrets required:** `ANTHROPIC_API_KEY` · `HF_TOKEN`
+
+**Variables:** `ZEROGPU_DURATION_SECONDS=120` · `ENABLE_MODEL_B_GPU=1`
+
+- **Adapter:** [Fernandosr85/afrobr-langbench-adapter](https://huggingface.co/Fernandosr85/afrobr-langbench-adapter)
+- **Dataset:** [Fernandosr85/adaption-pt-afro-brasileiro-qa](https://huggingface.co/datasets/Fernandosr85/adaption-pt-afro-brasileiro-qa)
+- **Challenge:** [AutoScientist Challenge 2026](https://adaptionlabs.ai/blog/autoscientist-challenge)