State model size as 18M parameters

README.md

@@ -15,7 +15,7 @@ metrics:
   - crps
   - mase
 model-index:
-  - name: yumoto-v0.2-q13-22m
+  - name: yumoto-v0.2-q13-18m
     results:
       - task:
           type: time-series-forecasting
@@ -76,7 +76,7 @@ it predicts a full range of outcomes rather than a single number.
 
 That range is the point. Knowing roughly where the price will land is easy;
 knowing how confident to be is what makes a forecast usable. Yumoto is built
-for the second job, and it is what lets a 17-million-parameter model beat
+for the second job, and it is what lets an 18-million-parameter model beat
 foundation models more than a hundred times its size.
 
 ## Versions
@@ -86,8 +86,8 @@ foundation models more than a hundred times its size.
 | **v0.2** | [`v0.2/`](./v0.2) | 13 (2.5% … 97.5%) | 17,759,008 | **0.6459** | **current** |
 | v0.1 | repository root | 9 (10% … 90%) | 17,691,808 | 0.7163 | previous release |
 
-Both use the `toto2-22m` model size, which is where the `22m` in the name
-comes from; the counts above are the real measured numbers. v0.2 predicts
+Both are **18M-parameter models** (measured counts above). The `22m` in the
+repository name is a legacy label from the size preset they were built on. v0.2 predicts
 four extra levels further out in the tails, so it says more about rare
 moves.