| import streamlit as st |
| from utils.api import get_available_datasets, load_datasets |
|
|
| _LABEL_MAP = { |
| "pubmedqa": "PubMedQA", |
| "mental_health": "Mental Health Counseling", |
| "mediqa": "Medical MediQA", |
| "medqa_usmle": "MedQA-USMLE", |
| } |
|
|
|
|
| def render_dataset_panel() -> None: |
| st.markdown("### Knowledge Base") |
|
|
| try: |
| datasets = get_available_datasets() |
| except Exception: |
| st.caption("API offline — start the server to manage datasets.") |
| return |
|
|
| if not datasets: |
| st.caption("No datasets available.") |
| return |
|
|
| search = st.text_input( |
| "search_kb", |
| placeholder="Search datasets…", |
| label_visibility="collapsed", |
| key="kb_search", |
| ) |
|
|
| filtered = [ |
| ds for ds in datasets |
| if not search or search.lower() in ds["name"].lower() or search.lower() in ds["description"].lower() |
| ] |
|
|
| if not filtered: |
| st.caption("No datasets match your search.") |
| return |
|
|
| st.caption("Select datasets to load into the knowledge base:") |
|
|
| selected = [ |
| ds["name"] |
| for ds in filtered |
| if st.checkbox( |
| _LABEL_MAP.get(ds["name"], ds["name"]), |
| key=f"ds_{ds['name']}", |
| help=ds["description"], |
| ) |
| ] |
|
|
| max_samples = st.slider( |
| "Max samples per dataset", |
| min_value=100, |
| max_value=2000, |
| value=500, |
| step=100, |
| help="Higher values improve coverage but take longer to embed.", |
| ) |
|
|
| if st.button("Load Selected", disabled=not selected, use_container_width=True): |
| with st.spinner(f"Embedding {len(selected)} dataset(s)… this may take a few minutes"): |
| try: |
| response = load_datasets(selected, max_samples) |
| for name, result in response.get("results", {}).items(): |
| label = _LABEL_MAP.get(name, name) |
| if result["status"] == "success": |
| st.success(f"{label}: {result['records_upserted']} records indexed") |
| elif result["status"] == "skipped": |
| st.warning(f"{label}: skipped — {result.get('detail', '')}") |
| else: |
| st.error(f"{label}: {result.get('detail', 'failed')}") |
| except Exception as exc: |
| st.error(f"Failed to load datasets: {exc}") |
|
|
