Fix Space metadata

README.md

@@ -1,57 +1,22 @@
-# AssayLens Agent
+---
+title: AssayLens Demo
+emoji: 🧪
+colorFrom: green
+colorTo: gray
+sdk: docker
+app_port: 7860
+pinned: false
+---
 
-AssayLens models observed functional pharmacology under assay frames. It is a
-lightweight benchmark and prototype next-assay recommender for testing whether
-explicit assay-frame features improve activity, potency, quantification,
-viability-confounding, and high-frequency artifact prediction.
+# AssayLens Demo
 
-The repository is designed so full data profiling, feature extraction, training,
-evaluation, and artifact upload run on Hugging Face Jobs. Local execution is for
-code editing, tiny smoke tests, and reading compact summaries. Do not download
-or store the full EvE raw dataset locally.
+HF Spaces wrapper for the AssayLens Gradio recommender.
 
-## Local Smoke
+The container runs:
 
 ```bash
-conda env create -f environment.yml
-conda activate assaylens
-python -m pytest
-python scripts/00_check_access.py --no-write
-```
-
-## CLI and Demo UI
-
-```bash
-assaylens audit-artifacts --artifact-repo lighteternal/assaylens-artifacts
-assaylens recommend --input examples/tiny_observations.json --artifact-repo lighteternal/assaylens-artifacts
 python scripts/11_gradio_demo.py
 ```
 
-The Gradio demo uses compact model artifacts from the HF artifact repo. It does
-not require or download raw EvE locally. Optional LLM explanations require
-`OPENAI_API_KEY` in the runtime environment and summarize only structured
-AssayLens outputs.
-
-## Hugging Face Jobs
-
-```bash
-export HF_NAMESPACE="<username>"
-export ARTIFACT_REPO="<username>/assaylens-artifacts"
-# Optional; defaults to "$HF_NAMESPACE/assaylens-code-snapshot".
-export ASSAYLENS_CODE_REPO="<username>/assaylens-code-snapshot"
-
-scripts/run_hf_job.sh scripts/00_check_access.py cpu-basic
-scripts/run_hf_job.sh scripts/01_profile_eve.py cpu-upgrade
-scripts/run_hf_job.sh scripts/02_profile_hugging_science.py cpu-basic
-scripts/run_hf_job.sh scripts/03_build_features.py cpu-performance
-scripts/run_hf_job.sh scripts/04_make_splits.py cpu-upgrade
-scripts/run_hf_job.sh scripts/05_train_baselines.py cpu-performance
-scripts/run_hf_job.sh scripts/06_train_assaylens.py a10g-small
-scripts/run_hf_job.sh scripts/07_evaluate.py cpu-upgrade
-scripts/run_hf_job.sh scripts/08_next_assay_policy.py cpu-upgrade
-scripts/run_hf_job.sh scripts/09_make_report.py cpu-basic
-```
-
-The local shell does not need to expose the token if `hf auth login` is already
-configured, but HF Jobs require an account-side `HF_TOKEN` secret for gated
-dataset access inside the remote container.
+Set `OPENAI_API_KEY` as a private Space secret only if optional LLM explanations
+are needed. The app does not ask users to paste an API key.