Publish Ropedia minimal task baseline weights

eea471e verified 28 days ago

3.33 kB

license: other
library_name: numpy
tags:
  - robotics
  - embodied-ai
  - multimodal
  - ropedia
  - xperience-10m
  - baseline
  - linear-model
  - retrieval
metrics:
  - accuracy
  - f1
  - mean-reciprocal-rank
  - mean-squared-error
model-index:
  - name: Ropedia Minimal Task Baselines
    results:
      - task:
          type: robotics
          name: Cross-modal retrieval
        dataset:
          type: ropedia-ai/xperience-10m-sample
          name: Xperience-10M public sample episode
        metrics:
          - type: top_5_accuracy
            value: 0.3764
            name: top-5 retrieval accuracy
          - type: mrr
            value: 0.2634
            name: mean reciprocal rank
      - task:
          type: robotics
          name: Transition detection
        dataset:
          type: ropedia-ai/xperience-10m-sample
          name: Xperience-10M public sample episode
        metrics:
          - type: f1
            value: 0.6552
            name: macro-F1

Ropedia Minimal Task Baselines

This repo stores the minimal baseline weights and metrics for the 12-task Ropedia episode suite.

These are intentionally small, transparent baselines:

z-score + linear softmax classifiers,
dual ridge regression/projection heads,
sigmoid multi-label logistic regression,
cosine ranking for retrieval tasks.

They are not deep robot policies or foundation models. Their purpose is to make every input/output contract auditable before scaling to many episodes.

Included

artifacts/**/model.npz: minimal baseline weights, scalers, and labels
artifacts/**/metrics.json: committed metrics
artifacts/**/feature_manifest.json: feature block boundaries where relevant
scripts/*.py: training and visualization scripts
notes/*.md: interpretation and reproducibility notes

The companion artifact dataset repo stores CSV/JSON predictions and dashboard assets:

https://huggingface.co/datasets/cy0307/ropedia-episode-task-suite-artifacts

The public visual dashboard is here:

https://huggingface.co/spaces/cy0307/ropedia-episode-task-suite

Minimal Architecture

Metrics Snapshot

Task	Minimal head	Main metric
`timeline_action`	linear softmax	0.0500 macro-F1
`timeline_subtask`	linear softmax	0.0495 macro-F1
`transition_detection`	linear softmax	0.6552 macro-F1
`next_action`	linear softmax	0.0593 macro-F1
`hand_trajectory_forecast`	ridge regression	0.8223 MPJPE
`contact_prediction`	linear softmax	1.0000 macro-F1
`object_relevance`	multi-label logistic	0.1839 micro-F1
`caption_grounding`	ridge + cosine rank	0.0172 MRR
`cross_modal_retrieval`	ridge + cosine rank	0.3764 top-5
`modality_reconstruction`	ridge regression	-0.0160 R2
`temporal_order`	binary softmax	0.5487 F1
`misalignment_detection`	binary softmax	0.4866 F1

Data Notice

This repo does not redistribute raw Ropedia videos or raw annotation.hdf5. Download the original sample from Ropedia / Hugging Face and follow the dataset terms:

Source

GitHub:

https://github.com/ChaoYue0307/ropedia-episode-task-suite