Update model card for widewin-roadmap retrain (sparse H3K27ac)

README.md

@@ -25,7 +25,7 @@ cropped for the heads (ChromBPNet-style "valid" geometry, so every output base
 has a full real-sequence receptive field). The two channels are:
 
 - **ch0 — DNase**: 5′ cut-sites (motif-sensitive, sharp).
-- **ch1 — H3K27ac**: read coverage (the histone mark's broad pile-up signal).
+- **ch1 — H3K27ac**: read signal (the active-enhancer histone mark).
 
 Each cell line gets its **own** main checkpoint (no FiLM, no cell embedding),
 paired with a per-cell **frozen 2-channel `BiasNet`** (1024-bp, run on the
@@ -47,12 +47,17 @@ bias/
   ... (11 cells)
 ```
 
+> **Shipped weights (2026-06-04): the `roadmap` retrain** — trained on Roadmap
+> DNase-summit peaks with the Roadmap-pipeline H3K27ac. Strongest per-cell DNase
+> test-r and a functional H3K27ac channel; supersedes an earlier ENCODE-IDR-peak /
+> full-coverage-H3K27ac variant.
+
 ## Training
 
 - **Window**: 2114-bp input → central 1024-bp profile crop, on roadmap DNase
   peak summits.
 - **Signals**: ch0 = per-bp 5′ DNase cut-site counts; ch1 = per-bp H3K27ac
-  read coverage.
+  signal (Roadmap v2.1 pipeline).
 - **Data**: ENCODE DNase + H3K27ac BAMs (SE/PE), fold-10 leave-chromosomes-out
   split (chr11+chr21 held out).
 - **Loss**: multinomial NLL on the per-bp 2-channel profile + MSE on log10 count