Add multimodal LLM columns, guardrail, design layer, Safe Synthesizer

Add benefit fairness column + leaderboard dropdown filters

Add one-shot server setup script; lazy DataDesigner import; --label for external baselines

Expand baseline roster: all synthcity-available TabDiff/fair-tab-diffusion baselines + external-baseline framework

Adopt TabDiff benchmark as the evaluation standard

Add TabDiff experiment scripts (prepare + full-metric evaluate) for schizophrenia

Leverage SDGym: baseline anchors (data_identity/column/uniform) + generation-time dimension

Add lab-server batch arena runner; fix synthcity plugin ids; neutral domain branding

Loosen requests pin to >=2.33 (data-designer-config requirement)

Shorten short_description to <=60 chars (HF limit)

Add Cloudflare Worker proxy for memisislabs.ai

Add fair-tab-diffusion fairness datasets + public seed leaderboard

Rebrand to MemisisLabs (memisislabs.ai)

Memisis Arena: multi-dataset, multi-model arena + leaderboard

Restore ML efficacy via conditional target generation

Move to DataDesigner generation + SDMetrics/fairlearn tools (drop SDV/torch)

Add chat-driven SDV agent (NL -> generate + evaluate + explain)

Replace microservice with Memisis Lab (SDV x SDMetrics Streamlit app)

Revamp sidebar: Generate vs Testing sections with explanations

Remove Auto-Pilot panel

Ship pre-trained synthesizer models + metadata

Show run errors in UI and logs; pin ML deps to known-good versions

Deploy Memisis synthetic health app OpenAI LFS