README.md

---
license: mit
pretty_name: Ropedia Xperience-10M Task Suite Artifacts
tags:
  - embodied-ai
  - multimodal
  - xperience-10m
  - evaluation
  - qwen3-omni
configs:
  - config_name: episode_sample
    data_files:
      - split: public_sample
        path: viewer/episode_windows.parquet
---

# Ropedia Xperience-10M Task Suite Artifacts

![Ropedia Xperience-10M Task Suite logo](assets/brand/xperience10m-logo-social-card.png)

This dataset repository stores small derived artifacts for the Ropedia
Xperience-10M task-suite project: metrics, predictions, manifests, reports,
figures, website JSON, public-safe Qwen3-Omni diagnostic outputs, and the
Cosmos3-Nano plus Cosmos3-Super diagnostic packages.

## What To Open First

| Reader goal | Artifact |
| --- | --- |
| Trace the 128-episode source and feature map | `XPERIENCE10M_128_EPISODE_FEATURE_INDEX.md`, `docs/data/xperience10m_128_episode_feature_index.json` |
| Read the project in 8 languages | `README.md`, `README.zh.md`, `README.es.md`, `README.fr.md`, `README.de.md`, `README.ja.md`, `README.ko.md`, `README.pt.md`, `docs/data/language_versions.json` |
| Choose the right public surface | `PUBLIC_READER_MAP.md`, `docs/data/public_reader_map.json` |
| Understand current scope | `PROJECT_STATUS.md` |
| Navigate all files | `ARTIFACT_GUIDE.md` |
| Interpret task metrics | `RESEARCH_TAKEAWAYS.md` |
| Check evaluation rules | `EVALUATION_PROTOCOL.md` |
| Inspect single-episode task results | `results/episode_task_suite/summary_report.json` |
| Inspect 128-episode same-split baselines | `results/omni_finetune/multi_episode_128_task_baselines/BASELINE_ALIGNMENT_REPORT.md` |
| Inspect final Qwen3-Omni held-out diagnostic result | `docs/data/omni_finetune_verified_result.json` |
| Compare current versions and model groups | `docs/data/omni_model_comparison.json` |
| Compare Qwen3-Omni v5/v6 diagnostic runs | `docs/data/qwen3_v5_v6_comparison.json` |
| Compare Qwen3 v5/v6 diagnostic branches | `docs/data/qwen3_v5_v6_comparison.json` |
| Explain Qwen3-Omni v1-v6 run lineage | `QWEN3_OMNI_RUN_LINEAGE.md`, `docs/data/qwen3_omni_run_lineage.json` |

## 128-Episode Enhancement Pack

The no-new-episode suite push is recorded in `TASK_SUITE_ENHANCEMENT_128.md`
and `docs/data/task_suite_enhancement_128.json`. It recommends
`multiscale_20s10_40s20_80s40`, hierarchical action/subtask targets,
label-normalized scoring, and compact raw-feature shards before adding more
episodes.

## Tier-2 Extension Baselines

The public-sample task layer now includes eight Tier-2 extension baselines in
`results/episode_task_suite/tier2_task_suite/` and
`docs/data/tier2_task_suite.json`. They reuse the same 20-frame windows,
5-frame stride, feature manifest, chronological split, and minimal/neural head
pattern as the core 12 tasks.

## Unified 20-Task Suite

The public-sample task surface is now one unified 20-task suite in
`TASK_SUITE_20.md` and `docs/data/task_suite_20.json`. Tasks 1-12 are the
original sample tasks; Tasks 13-20 reuse the same 20-frame windows, 5-frame
stride, feature manifest, chronological split, and minimal/neural head pattern.
The historical `tier2_task_suite` path is retained only for stable artifact
links to tasks 13-20. The unified radar chart is published as
`docs/assets/charts/unified_task_model_radar.svg` with values in
`docs/data/unified_task_model_radar.json`; the 9-method by 20-task
completion matrix is complete at `180/180` scored method-task records and is published in `docs/data/task_method_20_result_matrix.json`, with two-line summaries in
`docs/assets/charts/two_evidence_line_map.svg`,
`docs/data/two_evidence_lines.json`, and
`docs/data/two_evidence_line_result_summary.json`, with the explicit
audit in `docs/data/task_method_20_gap_audit.json` and source-value audit in
`docs/data/task_method_20_source_audit.json`. Split radars are in
`docs/assets/charts/single_episode_task_model_radar.svg` and
`docs/assets/charts/episode128_task_model_radar.svg`.

## Public Surface Map

Use `PUBLIC_READER_MAP.md` and `docs/data/public_reader_map.json` to choose
between the GitHub repo, GitHub Pages dashboard, HF Space, artifact dataset,
baseline model repo, Qwen3/Cosmos model repos, and release-health checks without
losing the full evidence trail.

## Multilingual Entry Points

The canonical repo README now has eight public reader entry points: English,
Chinese, Spanish, French, German, Japanese, Korean, and Portuguese. The
machine-readable language map is in `docs/data/language_versions.json`; each
Hugging Face mirror carries the same translated README files so readers can
move between GitHub, the dashboard, the Space, the artifact dataset, and model
cards without losing the evidence trail.

## Dataset Boundary

This artifact bundle contains derived artifacts only. It does not redistribute
Raw Xperience-10M videos, raw `annotation.hdf5`, `.rrd` files, private gated
dataset files, full Qwen weights, LoRA adapter weights, or large checkpoints.
Use of the upstream Xperience-10M dataset remains governed by the official
Ropedia/Xperience-10M access terms.

The implemented public-sample task suite uses one public Xperience-10M sample
episode. The selected 128-episode Qwen3-Omni final diagnostic result uses a
gated local dataset copy and publishes only public-safe metrics, predictions,
manifests, reports, and audits. The Qwen3-Omni LoRA adapter weights are
published separately at `cy0307/ropedia-qwen3-omni-lora-128ep`.
Cosmos3-Nano remains an artifacts-only compatibility result. Cosmos3-Super
Forward-Dynamics LoRA has a separate weight-bearing model repo at
`cy0307/ropedia-cosmos3-super-forward-dynamics-lora-128ep`.

Source alignment is mirrored through `xperience10m_dataset_card_alignment.json`
and `source_alignment_audit.json`. The official gated dataset card records
`31.9 TB` on the live HF surface and an `about-1PB` full-scale storage
statement; the committed API-listing snapshot records `12,103 episode folders`
as upstream `metadata only`, not local raw-data possession. The public sample
remains under `cc-by-nc-4.0`, with the `HOMIE Toolkit` and `Rerun 0.29.0`
noted as source tooling. The official note that the data is `limited in diversity`
is preserved.

## Derived Artifacts

This bundle includes derived artifacts such as:

- 12-task single-episode metrics, predictions, feature manifests, and neural
  MLP result directories.
- Audio-ablation summaries and generated chart assets.
- Public website JSON and figure manifests.
- The latest verified Qwen3-Omni LoRA v6 diagnostic package for the selected
  96/16/16 episode split includes 34,269 exported windows, 4,032 held-out test
  predictions, 99.90% JSON validity, and public-safe metrics/predictions.
- Historical Qwen3-Omni packages, including the earlier v2 strict-JSON
  diagnostic, for regression and prompt-contract comparison.
- Verified Cosmos3-Nano future-window compatibility, Cosmos3-Super
  base-weight Reasoner evaluation, and Cosmos3-Super Forward-Dynamics LoRA
  public-safe packages for the same selected split family.
- 128-episode same-split simple/NN metadata baselines for the same 12 task ids,
  with unsupported markers where raw 128 sensor feature blocks are still needed.
- A model-family grouped comparison that pairs 1-episode and 128-episode entries
  for task heads, Qwen3-Omni LoRA, Cosmos3-Nano, and Cosmos3-Super without
  mixing target types.

## Related Hub Repositories

| Surface | URL |
| --- | --- |
| HF Space | https://huggingface.co/spaces/cy0307/ropedia-xperience-10m-task-suite |
| Artifact dataset | https://huggingface.co/datasets/cy0307/ropedia-xperience-10m-task-suite-artifacts |
| Baseline model repo | https://huggingface.co/cy0307/ropedia-xperience-10m-task-baselines |
| Qwen3-Omni LoRA adapter repo | https://huggingface.co/cy0307/ropedia-qwen3-omni-lora-128ep |
| Cosmos3-Super Forward-Dynamics LoRA adapter repo | https://huggingface.co/cy0307/ropedia-cosmos3-super-forward-dynamics-lora-128ep |
| GitHub repo | https://github.com/ChaoYue0307/ropedia-xperience-10m-task-suite |

## Citation

If you use these artifacts, cite this project and the upstream Ropedia
Xperience-10M dataset according to the citation guidance in `CITATION.cff` and
the official dataset card.