EVIDENCE_CONTRACT.md

# Evidence Contract

This project is intentionally audit-first. Every visible claim should point to a
local artifact that a reader can inspect before trusting the dashboard.

| Claim | Current evidence | Status | Boundary |
| --- | --- | --- | --- |
| The public Xperience-10M sample has been converted into aligned model windows. | `results/episode_task_suite/windows.csv`, `results/episode_task_suite/shared_windows.npz`, `results/episode_task_suite/summary_report.json` | Verified for 5,821 frames and 1,161 windows | One public sample episode only |
| The current feature contract is explicit and reviewable. | `results/episode_task_suite/feature_manifest.json`, `results/episode_task_suite/available_modalities.json` | Verified for an 8,378-d feature vector | Audio is present in MP4 streams but not yet a feature block |
| The 12 task heads are real scripts and artifacts, not presentation placeholders. | `scripts/episode_task_suite.py`, `results/episode_task_suite/*/metrics.json`, `results/episode_task_suite/*/predictions.*` | Verified for all 12 task definitions | Chronological single-episode split, not cross-episode generalization |
| Minimal and neural heads use the same task contracts. | `scripts/neural_task_models.py`, `results/episode_task_suite/neural_mlp/`, `docs/assets/task_architectures.png` | Verified for 12 minimal heads and 12 neural MLP heads | Small heads only; not a foundation model |
| Four Ropedia research directions are mapped honestly as direct, proxy, or diagnostic evidence. | `results/episode_task_suite/research_directions/research_direction_taxonomy.json`, `docs/data/research_directions.json` | Verified taxonomy | Some directions remain proxy-only |
| Four extra direction probes are coded and evaluated. | `results/episode_task_suite/research_direction_extensions/research_direction_extension_results.json`, `docs/data/research_direction_extensions.json` | Verified single-episode probes | Not full human modeling, neural rendering, intent modeling, or world modeling solutions |
| Qwen3-Omni infrastructure has passed technical smoke checks. | `results/omni_finetune/RUN_REPORT.md`, `results/omni_finetune/dataset_manifest.json`, `results/omni_finetune/metrics_eval.json` | Smoke-only evidence | One episode, 128 train windows; not a 32-episode pilot |
| The real 32-episode LoRA pilot is blocked on gated data access, not on repo presentation. | `results/omni_finetune/DATA_BLOCKER_REPORT.md`, `results/omni_finetune/A100_HF_RELAY_STATUS.md`, `results/omni_finetune/source_discovery.json` | Blocker documented | No 32-episode metric should be claimed until the gate passes |

## Review Order

1. Read `docs/assets/task_suite_infographic.png` for the high-level map.
2. Inspect `results/episode_task_suite/summary_report.json` for the task and
   metric source of truth.
3. Inspect `results/episode_task_suite/feature_manifest.json` to see which
   modalities enter the current feature vector.
4. Inspect `results/episode_task_suite/neural_mlp/` to compare minimal and
   neural heads under the same splits.
5. Inspect `results/omni_finetune/DATA_BLOCKER_REPORT.md` before interpreting
   any Qwen3-Omni artifact.