Update README.md

README.md

@@ -18,7 +18,6 @@ likelihood responses over a fixed panel of DNA sequences.
 | File | Description |
 |---|---|
 | `main_panel.parquet` | 10,000-probe biological panel (14 functional elements × 4 categories) |
-| `MLM_k1ablation_1000_main_panel.parquet` | 1,000-probe stratified subset for the k=1 vs k=6 stride PLL ablation |
 | `panel_manifest.json` | Per-element / per-dataset counts and sampling seeds |
 | `panel_summary.md` | Human-readable panel composition summary |
 
@@ -43,7 +42,7 @@ panel = pd.read_parquet(path)
 **CC-BY-NC-SA-4.0.** The panel was assembled from three upstream benchmarks:
 
 - GUE (DNABERT-2) — Apache-2.0
-- Plant Genomic Benchmark (PGB) — CC-BY-NC-SA-4.0
+- Plant Genomic Benchmark (ArgoNT) — CC-BY-NC-SA-4.0
 - DNA Foundation Benchmark (DFB) — Apache-2.0
 
 Because the panel contains 1,600 probes drawn from PGB, the combined