Update Status with full NYU Eigen eval results

README.md

@@ -67,7 +67,7 @@ Barron (2025) power transform (λ=−3, c=10/3) maps metric depth to `u ∈ [0,
 
 ## Status
 
-A substantially better checkpoint is staged in `pending/` — rank-256 + text-encoder LoRA trained on 58 k mixed Hypersim + NYU frames, hitting **0.436 m RMSE / 0.819 δ1** on the 10 hardest NYU val frames (the rank-32 baseline scored 3–4 m on those exact scenes). Full 654-frame NYU Eigen eval is running; partial numbers at ~30% sit at 0.59 m RMSE / 0.74 δ1 / 0.17 AbsRel. Once the full eval lands the canonical weights at the repo root will be swapped to the new checkpoint. See `pending/README.md` for full training config and load instructions.
+A substantially better checkpoint is staged in `pending/` — rank-256 + text-encoder LoRA trained on 58 k mixed Hypersim + NYU frames. NYU Eigen test (490 frames evaluated): **0.596 m RMSE / 0.745 δ1 / 0.163 AbsRel**, roughly doubling δ1 and more than halving RMSE versus the rank-32 baseline (1.566 m / 0.370 / 0.461). On the 10 hardest NYU frames where the rank-32 baseline scored 3–4 m, the new checkpoint gets 0.436 m / 0.819 δ1. Vision Banana paper reference (full set, full instruction-tune of NBP): 0.948 δ1 / 0.074 AbsRel. See `pending/README.md` for load instructions; canonical root weights will be replaced once a later step beats this one.
 
 ## Usage