Add public contribution templates

.github/ISSUE_TEMPLATE/benchmark_submission.yml

@@ -0,0 +1,44 @@
+name: Benchmark submission discussion
+description: Propose or discuss a model submission for the public benchmark folds.
+title: "[Submission]: "
+labels:
+  - benchmark
+body:
+  - type: input
+    id: task_id
+    attributes:
+      label: Task ID or fold package
+      placeholder: A5_skin_lomo / fold_RR-7_test
+    validations:
+      required: true
+  - type: input
+    id: model_name
+    attributes:
+      label: Model name
+      placeholder: PCA-LR, Geneformer adapter, custom baseline
+    validations:
+      required: true
+  - type: textarea
+    id: method
+    attributes:
+      label: Method summary
+      description: Describe preprocessing, feature representation, training split, and prediction output.
+    validations:
+      required: true
+  - type: textarea
+    id: validation
+    attributes:
+      label: Validation command
+      description: Include the command used to validate or evaluate the JSON submission.
+      render: bash
+    validations:
+      required: false
+  - type: checkboxes
+    id: checks
+    attributes:
+      label: Checks
+      options:
+        - label: I followed docs/submission_format.md.
+          required: true
+        - label: I did not use held-out test labels during model training or feature selection.
+          required: true

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,8 @@
+blank_issues_enabled: true
+contact_links:
+  - name: Hugging Face Dataset
+    url: https://huggingface.co/datasets/jang1563/genelab-benchmark
+    about: Download the public fold package and dataset card.
+  - name: Submission Format
+    url: https://github.com/jang1563/GeneLab_benchmark/blob/main/docs/submission_format.md
+    about: Read the public prediction-submission JSON contract.

.github/ISSUE_TEMPLATE/data_or_docs_issue.yml

@@ -0,0 +1,36 @@
+name: Data or documentation issue
+description: Report a broken path, unclear public documentation, or release-surface inconsistency.
+title: "[Data/docs]: "
+labels:
+  - documentation
+body:
+  - type: textarea
+    id: location
+    attributes:
+      label: Location
+      description: Link to the file, section, dataset path, or release asset.
+      placeholder: README.md, docs/hf_dataset_card.md, A5_skin_lomo/fold_RR-7_test/train_X.csv
+    validations:
+      required: true
+  - type: textarea
+    id: issue
+    attributes:
+      label: What is wrong or unclear?
+      description: Describe the observed issue and what you expected instead.
+    validations:
+      required: true
+  - type: textarea
+    id: reproduction
+    attributes:
+      label: Reproduction command or access path
+      description: Include a command, URL, or short sequence that reproduces the issue.
+      render: bash
+    validations:
+      required: false
+  - type: textarea
+    id: environment
+    attributes:
+      label: Environment
+      description: Include OS, Python version, package manager, or browser if relevant.
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/reproducibility_issue.yml

@@ -0,0 +1,36 @@
+name: Reproducibility issue
+description: Report a command, dependency, or release QA problem.
+title: "[Repro]: "
+labels:
+  - reproducibility
+body:
+  - type: textarea
+    id: command
+    attributes:
+      label: Command
+      description: Paste the command or workflow step that failed.
+      render: bash
+    validations:
+      required: true
+  - type: textarea
+    id: output
+    attributes:
+      label: Output
+      description: Include the relevant error or unexpected output.
+      render: text
+    validations:
+      required: true
+  - type: textarea
+    id: expected
+    attributes:
+      label: Expected behavior
+      description: Describe what you expected the command to produce.
+    validations:
+      required: true
+  - type: textarea
+    id: environment
+    attributes:
+      label: Environment
+      description: Include OS, Python version, package versions, and whether data came from GitHub or Hugging Face.
+    validations:
+      required: false

.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,15 @@
+## Summary
+
+- Briefly describe the change.
+
+## Verification
+
+- [ ] `make release-qa`
+- [ ] `make hpc-public-qa`
+- [ ] Relevant unit tests or scripts:
+
+## Public Surface Check
+
+- [ ] README/HF/CITATION/release labels remain consistent
+- [ ] No stale internal wording, placeholder paths, or obsolete release labels
+- [ ] New public artifacts are linked from the appropriate card or manifest

CONTRIBUTING.md

@@ -0,0 +1,50 @@
+# Contributing To SpaceBio-Bench
+
+Thank you for helping improve SpaceBio-Bench / GeneLab Benchmark. The most
+useful contributions are small, reproducible, and tied to the public release
+surface.
+
+## Good First Contributions
+
+- Report a broken download path, stale link, or unclear dataset-card wording.
+- Reproduce a public fold or release QA command and share the exact command and
+  output.
+- Improve documentation around task layout, submission format, or citation.
+- Submit benchmark predictions using the public JSON format.
+
+## Benchmark Submissions
+
+Prediction submissions should follow
+[docs/submission_format.md](docs/submission_format.md). Include:
+
+- task ID and fold IDs;
+- model name and short model description;
+- exact preprocessing or feature representation;
+- the command used to validate or evaluate the submission.
+
+Use a GitHub issue for discussion before opening a pull request with benchmark
+results or new public artifacts.
+
+## Pull Requests
+
+Keep pull requests focused. For documentation, metadata, or public-card changes,
+run:
+
+```bash
+make release-qa
+make hpc-public-qa
+```
+
+For code changes that touch benchmark behavior, also run the relevant unit tests
+or scripts and describe what changed in the PR body.
+
+## Public Release Boundaries
+
+Use the release labels in [README.md](README.md) and
+[release/release_manifest.json](release/release_manifest.json). In short:
+
+- v7.1 is the canonical historical result surface.
+- v7.1.2 is a public-card and metadata patch over v7.1 results.
+- v9 public bulk is a metadata catalog and baseline-summary surface.
+
+Please avoid mixing future-lane claims into the v7.1 result surface.

README.md

@@ -165,6 +165,7 @@ scripts/               Data, evaluation, upload, validation, and figure scripts
 | Release readiness | [docs/SPACEBIOBENCH_RELEASE_READINESS_CARD.md](docs/SPACEBIOBENCH_RELEASE_READINESS_CARD.md) |
 | Hugging Face dataset card source | [docs/hf_dataset_card.md](docs/hf_dataset_card.md) |
 | v9 metadata catalog card source | [docs/v9_hf_dataset_card.md](docs/v9_hf_dataset_card.md) |
+| Contributing and submissions | [CONTRIBUTING.md](CONTRIBUTING.md) and [docs/submission_format.md](docs/submission_format.md) |
 | Machine-readable release state | [release/release_manifest.json](release/release_manifest.json) |
 
 ## Data And Release Notes
@@ -179,6 +180,12 @@ Release labels are intentionally separated:
 - **v7.1**: canonical historical result surface and citation target.
 - **v9 public bulk**: metadata catalog and baseline-summary surface.
 
+## Contributing
+
+Use [CONTRIBUTING.md](CONTRIBUTING.md) for documentation fixes, data-access
+reports, reproducibility issues, and public benchmark submissions. Prediction
+submissions should follow [docs/submission_format.md](docs/submission_format.md).
+
 ## Citation
 
 Please cite the software using [CITATION.cff](CITATION.cff). GitHub renders the

scripts/validate_public_docs_consistency.py

@@ -84,6 +84,8 @@ def validate_public_docs() -> list[str]:
     require_contains(errors, "README.md", readme, "release/release_manifest.json")
     require_contains(errors, "README.md", readme, "https://huggingface.co/datasets/jang1563/genelab-benchmark")
     require_contains(errors, "README.md", readme, "docs/assets/hf_benchmark_summary.png")
+    require_contains(errors, "README.md", readme, "CONTRIBUTING.md")
+    require_contains(errors, "README.md", readme, "docs/submission_format.md")
     require_absent(errors, "README.md", readme, "Version: v7.0 (2026-04-12)")
     require_absent(errors, "README.md", readme, "Status: **v1–v7 Complete**")