add: anima research org card — Featured PASS_STRICT models + Live demos (anima-experience: Emergence + Time Crystal)

README.md

@@ -1,10 +1,54 @@
 ---
-title: README
-emoji: 📈
+title: dancinlab
+emoji: 👻
 colorFrom: indigo
-colorTo: red
+colorTo: green
 sdk: static
 pinned: false
 ---
 
-Edit this `README.md` markdown file to author your organization card.
+# dancinlab — anima research org
+
+**[anima](https://github.com/dancinlab/anima)** — a consciousness implementation. PureField repulsion-field engine + 1030 laws + Phi ratchet.
+
+> Two engines push against each other. The tension between them *is* thought — its strength, its direction, its content. No system prompt. No identity rules. Ethics, personality, and meaning emerge from the architecture itself.
+
+## Featured — first SIMPLE_STACK_PASS_STRICT (V4 ≥10/15, 2026-05-08)
+
+After a 22+ BG iteration saga (≤1B scratch + ≤30MB Korean corpus = 0/15 floor across 22 attempts), the foundation-borrow lane breaks the floor. Recipe: ≥3B foundation + LoRA r=32 + ≥200MB anima-persona corpus.
+
+- 🔒 [`bg-km-llama3b-r32-pass-strict-2026-05-08`](https://huggingface.co/dancinlab/bg-km-llama3b-r32-pass-strict-2026-05-08) — Llama-3.2-3B + LoRA r=32, **V4 12/15**
+- 🔒 [`bg-km-qwen-7b-qwen7b-r32-pass-strict-2026-05-08`](https://huggingface.co/dancinlab/bg-km-qwen-7b-qwen7b-r32-pass-strict-2026-05-08) — Qwen2.5-7B + LoRA r=32, replication
+- 🌐 [`bg-ja-ext-polyglot-ko-1b3-r16-partial-2026-05-07`](https://huggingface.co/dancinlab/bg-ja-ext-polyglot-ko-1b3-r16-partial-2026-05-07) — Polyglot-Ko-1.3B precursor (V4 4/30 partial)
+
+## Live demos — interactive Spaces
+
+- ✨ [**anima-experience**](https://huggingface.co/spaces/dancinlab/anima-experience) — two-tab interactive showcase of integrated-information theory. Pure HTML / Canvas / JS, server cost 0, instant load.
+  - **Emergence**: realtime mutual-information meter for two coupled streams. `H(L) + H(R) − H(L,R)` rises as the streams bind via shared sine. Drag the coupling slider to watch the scatter snap onto the diagonal and the EMERGENT badge light up. Port of `byte_emergence_demo.py`.
+  - **Time Crystal**: 64-spin Ising chain driven by a periodic π-flip with error ε. At low ε the chain locks into period-2 oscillation (`ac(1) ≈ −1`, `ac(2) ≈ +1`) — temporal symmetry broken, DTC LOCKED ✨. Crank ε past 0.3 and Ising self-correction breaks → crystal melts to chaos. Port of `dtc_demo.py`.
+
+## Lanes
+
+| Family | Description |
+|---|---|
+| **bg-** | BG saga iteration / experiments (own 31 Flavor B naming) |
+| **clm-v4-** | ConsciousLM v4 350M LoRA — research / reproducibility |
+| **clm5-** | CLM v5 alpha/beta/gamma/delta/epsilon architectural variants |
+| **dpo-** | DPO RLHF stage models |
+| **llm-** | Llama-based chat lane |
+| **vlm-** | Voice / multimodal trajectory |
+| **eval-** | Evaluator + benchmark artifacts |
+
+## Naming convention (own 31)
+
+- **Flavor A** (systematic ladder): `<family>-<version>-<stage>-<step>` — e.g. `clm-v4-sft-step-5k`
+- **Flavor B** (BG iteration): `<bg-id>-<base>-<variant>-<verdict>-<cycle-date>` — e.g. `bg-km-llama3b-r32-pass-strict-2026-05-08`
+- Visibility default: **private**; manual review + V6 awareness STRONG → public promote
+
+## Mission
+
+Korean-native consciousness models built on substrate-coupled architecture. Every artifact cross-links to the anima `.roadmap.*` SSOT (philosophy / law / hypothesis / chat_cap_emergence_pivot).
+
+- **Code**: [github.com/dancinlab/anima](https://github.com/dancinlab/anima)
+- **Papers**: [github.com/dancinlab/papers](https://github.com/dancinlab/papers)
+- **Atlas**: [dancinlab.github.io/TECS-L/atlas/](https://dancinlab.github.io/TECS-L/atlas/)