docs(README): blank OpenCLIP zero-shot cell; VideoMAEv2 data as 1.35M

OpenCLIP ImageNet -> – (drop zero-shot value and † footnote); VideoMAEv2 training data 0.0014B -> 1.35M; update caption + protocol note references.

README.md

@@ -96,7 +96,7 @@ Outputs (`BaseModelOutputWithPooling`): `last_hidden_state` `(B, 1+4+N, 4096)`,
 
 Compared against the strongest publicly reported self-supervised / vision backbones. Higher is
 better for every column **except KITTI depth** (lower is better). Best **comparable** value per
-column in **bold**; the footnoted figures (†, ‡) are measured under a different protocol and are
+column in **bold**; the footnoted figure (‡) is measured under a different protocol and is
 excluded from the bold ranking.
 
 DAVIS S/M/L follow the DINOv3 protocol (J&F-mean at video short side 420/480, 840/960,
@@ -109,11 +109,11 @@ single-resolution (S) figure is available.
 | DINOv3 | 1.7B | 71.1 | 79.7 | 83.3 | 88.4 | **55.9** | 87.8 | **2.3** |
 | PEcore | 5.4B | 48.2 | 53.1 | 49.8 | **89.3** | 38.9 | **87.9** | 4.1 |
 | SigLIP2 | 10B | 56.1 | 62.3 | 62.9 | 89.1 | 45.4 | 86.9 | – |
-| OpenCLIP | 2B | – | – | – | 80.1<sup>†</sup> | – | – | – |
+| OpenCLIP | 2B | – | – | – | – | – | – | – |
 | V-JEPA 2.1 | 0.022B | 69.0 | – | – | 85.5 | 47.9 | 87.7 | 3.x |
-| VideoMAEv2 | 0.0014B | – | – | – | – | – | 90.0<sup>‡</sup> | – |
+| VideoMAEv2 | 1.35M | – | – | – | – | – | 90.0<sup>‡</sup> | – |
 
-<sub>† OpenCLIP ViT-bigG/14 (LAION-2B): **zero-shot** ImageNet-1K — the authors report no linear-probe figure; shown for reference, not directly comparable to the linear-probe column.<br>‡ VideoMAEv2 ViT-g: K400 top-1 after **full fine-tuning**, not the frozen attentive-probe protocol used for the other rows.</sub>
+<sub>‡ VideoMAEv2 ViT-g: K400 top-1 after **full fine-tuning**, not the frozen attentive-probe protocol used for the other rows.</sub>
 
 - **DAVIS video segmentation (J&F)** — leads at every resolution: **74.0 (S) / 80.5 (M) /
   83.5 (L)**, surpassing DINOv3 7B (71.1 / 79.7 / 83.3) across the board, reflecting the
@@ -133,7 +133,8 @@ single-resolution (S) figure is available.
 Protocol: DINOv3-style linear/attentive probes for image tasks; V-JEPA 2-style protocol for
 video. Baseline DAVIS / ADE20K / K400 / KITTI figures are taken from the DINOv3 technical
 report's unified evaluation (Tab. 3, 5, 6) and ImageNet from Tab. 7; OpenCLIP and VideoMAEv2
-are not in that report and are cited from their own papers (see †, ‡).
+are not in that report; VideoMAEv2's K400 figure is cited from its own paper (see ‡), and
+OpenCLIP's remaining benchmarks are not reported under a comparable protocol.
 
 ## License