Ship the verified cell-type grouping so training needs no source rebuild

README.md

@@ -50,6 +50,7 @@ in the same conventions so the two compose.
 | `neurons/soma_position.i32` | 49,393 × 3 | int32 | soma location in 8 nm voxels |
 | `neurons/soma_valid.u8` | 49,393 | uint8 | 1 where a soma location exists |
 | `manifest.json` | — | — | shapes, dtypes, SHA-256 of every file, provenance |
+| `groups/groups.npz` | — | — | the cell-type grouping the trainer loads, so training needs no 1 GB source rebuild |
 
 6,061,699 edges are positive, 2,988,473 negative, none zero. 44,279 of 49,393 neurons have a
 measured soma location; the remaining 5,114 are flagged in `soma_valid` and **their
@@ -98,6 +99,19 @@ runs 18 such checks with nothing but NumPy:
 All 18 checks passed: 49,393 neurons, 9,050,172 edges, 44,279 positioned somata, 9,161 cell-type groups.
 ```
 
+## Training against it
+
+`groups/groups.npz` carries the verified neuron→cell-type grouping bound to this exact graph
+(`node_type_index`, body IDs, labels, and the frozen-buffer digests). It exists so that
+training a model on this connectome does not require re-deriving the grouping from the
+~1 GB of upstream MaleCNS feather files. The
+[fly-wordbrain pipeline](https://github.com/fernando-neto-ai/fly-wordbrain) fetches it
+straight from here:
+
+```bash
+python scripts/run_pipeline.py --quick     # one command, Apple Silicon
+```
+
 ## Provenance, and one honest gap
 
 Derived from **MaleCNS v1.0 central brain** — superclasses `cb_sensory`,