Add Blockwise-OAT baseline artifacts (README.md)

README.md

@@ -18,16 +18,36 @@ generation on a frozen [OAT](https://arxiv.org/abs/2602.04215) policy.
 
 ## Summary
 
-| Metric | AR baseline | Blockwise (P=4, r=1) |
-|--------|-------------|----------------------|
+Primary SR reference: **OAT8 paper** on LIBERO-10 — **56.3%** ([OAT](https://arxiv.org/abs/2602.04215), external benchmark).
+
+| Metric | AR (our eval) | Blockwise (P=4, r=1) |
+|--------|---------------|----------------------|
 | LIBERO-10 mean SR | **58.73% ± 0.18%** | **52.33% ± 1.04%** |
-| Δ vs AR | — | **-6.40 pp** |
-| Decoder speedup (bs=1) | — | **1.16×** |
-| E2E `predict_action` (bs=1) | 36.4 ms | 30.1 ms (**1.21×**) |
+| **Δ vs OAT paper (56.3%)** | **+2.43 pp** | **-3.97 pp** |
+| Paired Δ (BW − AR, same protocol) | — | **-6.40 pp** |
 | Tail train epochs | — | 15 (final CE 3.0607) |
 
-AR baseline exceeds the published OAT8 reference (~56.3%) on our cluster stack.
-Blockwise achieves **>1× decoder speedup** but **−6.4 pp** SR at 15 tail epochs (resume training planned).
+Our frozen AR checkpoint reproduces above the paper on this cluster stack (58.73% vs 56.3%).
+Blockwise trades SR for faster token generation; tail training was only 15 epochs (resume planned).
+
+### Inference speed (V100, cuda:0)
+
+**Decoder-only** — 8 action tokens after `cond` is computed (`benchmark_blockwise_vs_ar`, warmup=10, 50 repeats):
+
+| Batch | AR | Blockwise | Speedup |
+|-------|-----|-----------|---------|
+| bs=1 | 22.3 ms | 19.3 ms | **1.16×** |
+| bs=8 | 31.4 ms | 26.6 ms | **1.18×** |
+
+**End-to-end `predict_action`** — vision encoder + decoder + detokenize (warmup=20, 100 repeats):
+
+| Batch | AR | Blockwise | Speedup |
+|-------|-----|-----------|---------|
+| bs=1 | 36.4 ms | 30.1 ms | **1.21×** |
+| bs=8 | 37.0 ms | 34.8 ms | **1.06×** |
+
+Decoder speedup is modest (~14–18% faster at bs=1) because the tail module is comparable in size to the AR stack;
+e2e gain is smaller still when the vision encoder dominates latency.
 
 ## Baseline artifacts (frozen)