Link live Hugging Face artifacts and paper metadata

README.md

@@ -40,9 +40,11 @@ Languages: [English](README.md) | [简体中文](README.zh-CN.md) | [繁體中
 | **Evidence, including failures** | Costs, turnover, baselines, caveats, and negative results stay visible |
 | **A contribution path** | CI, tests, report templates, Colab, and newcomer-sized research tasks |
 
-Hugging Face publishing is prepared through the
-[synthetic dataset/model export](docs/huggingface_artifacts.md); the export path
-explicitly excludes real and proprietary market data.
+Try the live Hugging Face artifacts: the
+[100,000-row synthetic dataset](https://huggingface.co/datasets/dddyym/ml-quant-trading-synthetic)
+and [213-input MLP checkpoint](https://huggingface.co/dddyym/ml-quant-trading-synthetic-mlp).
+Both come from the deterministic quick start and explicitly exclude real and
+proprietary market data; see the [artifact guide](docs/huggingface_artifacts.md).
 
 ## Quick Start
 

docs/huggingface_artifacts.md

@@ -8,11 +8,13 @@ from the deterministic synthetic quick start.
 
 | Hugging Face repository | Contents | Intended use |
 |---|---|---|
-| `dddyym/ml-quant-trading-synthetic` | Viewer-ready compressed CSV, generator config, checksum manifest, dataset card | Installation, CI, teaching, and pipeline smoke tests |
-| `dddyym/ml-quant-trading-synthetic-mlp` | PyTorch checkpoint, factor order, architecture config, metrics, model card | Checkpoint loading and inference smoke tests |
+| [`dddyym/ml-quant-trading-synthetic`](https://huggingface.co/datasets/dddyym/ml-quant-trading-synthetic) | Live viewer-ready 100,000-row compressed CSV, generator config, checksum manifest, dataset card | Installation, CI, teaching, and pipeline smoke tests |
+| [`dddyym/ml-quant-trading-synthetic-mlp`](https://huggingface.co/dddyym/ml-quant-trading-synthetic-mlp) | Live PyTorch checkpoint, 213-factor order, architecture config, metrics, model card | Checkpoint loading and inference smoke tests |
 
-Neither artifact represents real instruments, proprietary data, deployable
-alpha, or evidence of live performance.
+Both public repositories were verified with anonymous downloads; the dataset
+viewer renders its schema and rows, and the checkpoint passes the documented
+loading and inference smoke test. Neither artifact represents real instruments,
+proprietary data, deployable alpha, or evidence of live performance.
 
 ## Build locally
 
@@ -50,7 +52,7 @@ artifacts/huggingface/
 The dataset gzip stream uses a fixed timestamp so the same panel produces the
 same SHA-256 digest. Manifests record the source commit and file checksums.
 
-## Upload after authentication
+## Maintainer upload after authentication
 
 Install and authenticate the Hugging Face CLI locally. Never paste a full token
 into an issue, PR, notebook, or chat transcript.
@@ -77,7 +79,7 @@ hf upload dddyym/ml-quant-trading-synthetic-mlp \
   artifacts/huggingface/ml-quant-trading-synthetic-mlp .
 ```
 
-After upload:
+After a new upload:
 
 1. Confirm the dataset viewer renders rows and column types.
 2. Run the model-card loading snippet in a clean environment.

docs/visibility_status.md

@@ -7,6 +7,8 @@ This page tracks the current public launch surface for `ml-quant-trading`.
 - Repository: <https://github.com/initial-d/ml-quant-trading>
 - Paper: <https://arxiv.org/abs/2507.07107>
 - Hugging Face paper page (verified authorship claim): <https://huggingface.co/papers/2507.07107>
+- Hugging Face synthetic dataset: <https://huggingface.co/datasets/dddyym/ml-quant-trading-synthetic>
+- Hugging Face synthetic MLP checkpoint: <https://huggingface.co/dddyym/ml-quant-trading-synthetic-mlp>
 - alphaXiv page: <https://www.alphaxiv.org/abs/2507.07107>
 - Awesome Quant listing (Factor Analysis): <https://github.com/wilsonfreitas/awesome-quant#factor-analysis>
 - ernie55ernie Awesome Quant listing (Research Frameworks): <https://github.com/ernie55ernie/awesome-quant>
@@ -182,7 +184,7 @@ Primary ask:
 - Add more public-data mini reproductions with larger or differently constructed universes.
 - Ask new users to try the Dev Container and report first-run friction.
 - Post the v0.2.1 validation entrypoint release to one relevant community at a time with a customized note.
-- Define a safe Hugging Face demo artifact or model card that uses synthetic or redistributable data only.
+- Invite reproducibility feedback through the live synthetic Hugging Face dataset and model cards.
 - Follow up on open high-relevance awesome-list pull requests after maintainers have had time to review.
 - Follow up on the auto-refreshed Awesome Quant listing PR after maintainer review.
 - Set a GitHub social preview image through repository settings when browser upload access is available.

scripts/export_huggingface_artifacts.py

@@ -131,6 +131,7 @@ def _dataset_card(
 - quantitative-trading
 - reproducibility
 - pytorch
+- arxiv:2507.07107
 configs:
 - config_name: default
   data_files:
@@ -205,6 +206,7 @@ def _model_card(
 - quantitative-trading
 - reproducibility
 - pytorch
+- arxiv:2507.07107
 ---
 
 # ml-quant-trading synthetic MLP smoke-test checkpoint

tests/test_export_huggingface_artifacts.py

@@ -61,8 +61,12 @@ def test_export_bundle_is_viewer_ready_and_explicitly_synthetic(tmp_path: Path):
     assert len(rows) == 4
     assert rows[-1]["tradable"] == "0"
     assert manifest["dataset_rows"] == 4
-    assert "no real instruments" in (dataset_dir / "README.md").read_text(encoding="utf-8")
-    assert "not as a market model" in (model_dir / "README.md").read_text(encoding="utf-8")
+    dataset_card = (dataset_dir / "README.md").read_text(encoding="utf-8")
+    model_card = (model_dir / "README.md").read_text(encoding="utf-8")
+    assert "no real instruments" in dataset_card
+    assert "not as a market model" in model_card
+    assert "arxiv:2507.07107" in dataset_card
+    assert "arxiv:2507.07107" in model_card
     assert json.loads((model_dir / "config.json").read_text())["in_dim"] == 2
     assert json.loads((model_dir / "feature_names.json").read_text()) == ["f1", "f2"]