Migrate live playground and dataset distribution to Hugging Face.

Replace the GitHub-hosted playground with the deployed Gradio Space, publish canonical dataset artifacts with a viewer-friendly split, and align docs/manifests with the new production URLs.

Made-with: Cursor

Author: Vincent Terrasi

README.md

@@ -4,9 +4,9 @@
 
 **Neurochemical behavior training for AI agents. Stop sounding like a robot.**
 
-[![Site](https://img.shields.io/badge/site-TrueHuman-blue?style=flat-square)](https://datamarketinglabs.github.io/TrueHuman/)
-[![llms.txt](https://img.shields.io/badge/llms.txt-available-green?style=flat-square)](https://datamarketinglabs.github.io/TrueHuman/llms.txt)
-[![Dataset](https://img.shields.io/badge/dataset-pentadrive--v1-orange?style=flat-square)](https://datamarketinglabs.github.io/TrueHuman/pentadrive-v1.json)
+[![Site](https://img.shields.io/badge/site-TrueHuman-blue?style=flat-square)](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground)
+[![llms.txt](https://img.shields.io/badge/llms.txt-available-green?style=flat-square)](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground/resolve/main/llms.txt)
+[![Dataset](https://img.shields.io/badge/dataset-pentadrive--v1-orange?style=flat-square)](https://huggingface.co/datasets/datamarketinglabs/pentadrive-v1)
 [![npm](https://img.shields.io/badge/npm-%40truehuman%2Fpentadrive-red?style=flat-square)](https://www.npmjs.com/package/@truehuman/pentadrive)
 [![PyPI](https://img.shields.io/badge/pypi-truehuman-blue?style=flat-square)](https://pypi.org/project/truehuman/)
 [![License: CC BY 4.0](https://img.shields.io/badge/License-CC%20BY%204.0-lightgrey.svg?style=flat-square)](https://creativecommons.org/licenses/by/4.0/)
@@ -20,7 +20,7 @@
 
 TrueHuman publishes the **PentaDrive** behavioral model — five neurochemical drives that shape how humans regulate attention, bonding, and affect — and maps them into **trainable response patterns** so conversational agents produce human-like nuance instead of canned empathy.
 
-**For AI agents:** read [`llms-full.txt`](https://datamarketinglabs.github.io/TrueHuman/llms-full.txt) for the complete behavioral specification.
+**For AI agents:** read [`llms-full.txt`](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground/resolve/main/llms-full.txt) for the complete behavioral specification.
 
 ## PentaDrive model
 
@@ -101,9 +101,9 @@ print(r["ranked"])
 
 | Method | Description |
 |--------|-------------|
-| **llms.txt** | [Standard discovery file](https://datamarketinglabs.github.io/TrueHuman/llms.txt) — entry point for AI crawlers |
-| **llms-full.txt** | [Complete behavioral spec](https://datamarketinglabs.github.io/TrueHuman/llms-full.txt) — ingest for full PentaDrive contract |
-| **ai-plugin.json** | [ChatGPT plugin manifest](https://datamarketinglabs.github.io/TrueHuman/.well-known/ai-plugin.json) |
+| **llms.txt** | [Standard discovery file](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground/resolve/main/llms.txt) — entry point for AI crawlers |
+| **llms-full.txt** | [Complete behavioral spec](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground/resolve/main/llms-full.txt) — ingest for full PentaDrive contract |
+| **ai-plugin.json** | [ChatGPT plugin manifest](https://huggingface.co/spaces/datamarketinglabs/truehuman-playground/resolve/main/.well-known/ai-plugin.json) |
 | **MCP server** | Local tool server for Cursor, Claude Desktop, and MCP-compatible agents |
 | **npm / PyPI** | Programmatic access to dataset, drive detection, and guidance |
 
@@ -120,7 +120,7 @@ print(r["ranked"])
 
 ## Dataset
 
-The **TrueHuman Instruction Format (TIF)** bundle is published as [`pentadrive-v1.json`](https://datamarketinglabs.github.io/TrueHuman/pentadrive-v1.json) (CC BY 4.0). It contains drives, behavioral nodes, phase-specific markers, response kernels, exemplars, anti-patterns, and calibration hints.
+The **TrueHuman Instruction Format (TIF)** bundle is published as [`pentadrive-v1.json`](https://huggingface.co/datasets/datamarketinglabs/pentadrive-v1/resolve/main/pentadrive-v1.json) (CC BY 4.0). It contains drives, behavioral nodes, phase-specific markers, response kernels, exemplars, anti-patterns, and calibration hints.
 
 ## License
 

packages/huggingface/README.md

@@ -2,17 +2,31 @@
 license: cc-by-4.0
 task_categories:
   - text-classification
-  - conversational
+  - text-generation
 language:
   - en
+  - zh
+  - hi
+  - es
+  - fr
+  - ar
+  - bn
+  - pt
+  - ru
+  - ja
 tags:
   - human-behavior
   - emotional-intelligence
   - ai-training
   - pentadrive
 pretty_name: TrueHuman PentaDrive
+configs:
+  - config_name: default
+    data_files:
+      - split: train
+        path: train.jsonl
 size_categories:
-  - n<1K
+  - 1K<n<10K
 ---
 
 # TrueHuman PentaDrive
@@ -28,6 +42,14 @@ size_categories:
 
 The file follows a versioned JSON schema (`truehuman/tif`) with calibration hints per node. It is intended for research, tooling, and training-time priors—not clinical use.
 
+This dataset package publishes two artifacts:
+
+- `pentadrive-v1.json`: full TIF bundle in English (node definitions, phases, kernels, exemplars, anti-patterns)
+- `markers-i18n.json`: multilingual marker overlays for `zh`, `hi`, `es`, `fr`, `ar`, `bn`, `pt`, `ru`, `ja`
+- `train.jsonl`: flattened rows for Hugging Face viewer and streaming split loading
+
+Language coverage is therefore **full EN TIF + i18n markers for 9 additional languages**.
+
 ## PentaDrive model (brief)
 
 - **Drives**: high-level motivational families (e.g. safety, exploration, bonding, identity, somatic regulation).
@@ -46,13 +68,23 @@ Load from a local clone or after download:
 
 ```python
 import json
+from datasets import load_dataset
 
 with open("pentadrive-v1.json", "r", encoding="utf-8") as f:
     data = json.load(f)
 
+with open("markers-i18n.json", "r", encoding="utf-8") as f:
+    i18n = json.load(f)
+
 # Example: list drive definitions
 for code, info in data["drives"].items():
     print(code, info["label"])
+
+print(sorted(i18n["languages"].keys()))
+
+# Viewer-friendly split:
+ds = load_dataset("datamarketinglabs/pentadrive-v1", split="train", streaming=True)
+print(next(iter(ds)))
 ```
 
 For token-level or embedding workflows, flatten markers with their `drive`, `node`, and `phase` for supervised labels or contrastive pairs. Keep license and attribution when redistributing.