Report episode counts in the dataset overview

Replace corpus counts with episode totals and add per-collection episode counts. State that passive figures are rounded and active totals include all five iterations and AIM ablations.

README.md

@@ -56,14 +56,12 @@ trajectories, not aggregate performance or coverage.
 
 | | Passive training | Active training | Evaluation |
 | --- | ---: | ---: | ---: |
-| Corpora | 4 | 35 | 4 |
-| Archives | 32 | 280 | 4 |
-| Organization | 8 shards per corpus | 5 iterations × 8 shards per method | 1 archive per corpus |
+| Episodes | ≈1.21 million | 17,000 | 400 |
+| Compressed size | 486.249 GB | 8.901 GB | 0.239 GB |
 
-**43 corpora · 316 archives · 495.388 compressed decimal GB.** The 39 training
-corpora comprise four passive collections and 35 active-exploration iteration
-corpora. Each evaluation corpus contains 100 episodes, for 400 evaluation
-episodes total.
+Active episode counts include all five collection iterations and both AIM
+ablations. Each evaluation collection contains 100 episodes. Passive counts are
+rounded; all sizes are compressed decimal GB.
 
 The previewed training corpora use 144×144 RGB PNG observations and
 six-dimensional continuous actions and a single camera view. Trajectories are
@@ -75,26 +73,26 @@ include PNG segmentation.
 
 Passive collections represent data that can be gathered at scale:
 
-| Corpus | Behavior | Compressed size |
-| --- | --- | ---: |
-| Expert PPO | Task-directed expert trajectories | 56.976 GB |
-| Recovery | Policy rollouts with perturbations and recovery | 118.956 GB |
-| Random walk | Temporally correlated random actions | 156.312 GB |
-| Uniform random | Uniform random actions | 154.005 GB |
+| Collection | Behavior | Episodes | Compressed size |
+| --- | --- | ---: | ---: |
+| Expert PPO | Task-directed expert trajectories | ≈307,000 | 56.976 GB |
+| Recovery | Policy rollouts with perturbations and recovery | ≈300,000 | 118.956 GB |
+| Random walk | Temporally correlated random actions | ≈300,000 | 156.312 GB |
+| Uniform random | Uniform random actions | ≈300,000 | 154.005 GB |
 
 Active collections represent smaller, task-specific exploration rounds. Each
 method has five corpora, `iter-0` through `iter-4`, and each iteration has eight
 shards.
 
-| Method | Role | All five iterations |
-| --- | --- | ---: |
-| AIM | Action Influence Maximization | 2.667 GB |
-| AIM no-MI | AIM mutual-information ablation | 2.617 GB |
-| AIM no-SN | AIM spectral-normalization ablation | 2.631 GB |
-| DADS | Dynamics-Aware Discovery of Skills | 0.260 GB |
-| DIAYN | Diversity Is All You Need | 0.223 GB |
-| ICM | Intrinsic Curiosity Module | 0.244 GB |
-| METRA | Metric-aware abstraction | 0.260 GB |
+| Method | Role | Episodes, all five iterations | Compressed size |
+| --- | --- | ---: | ---: |
+| AIM | Action Influence Maximization | 5,000 | 2.667 GB |
+| AIM no-MI | AIM mutual-information ablation | 5,000 | 2.617 GB |
+| AIM no-SN | AIM spectral-normalization ablation | 5,000 | 2.631 GB |
+| DADS | Dynamics-Aware Discovery of Skills | 500 | 0.260 GB |
+| DIAYN | Diversity Is All You Need | 500 | 0.223 GB |
+| ICM | Intrinsic Curiosity Module | 500 | 0.244 GB |
+| METRA | Metric-aware abstraction | 500 | 0.260 GB |
 
 Evaluation data is separate from training data: