README: clarify frames-per-ckpt mapping (blind=10.06M, sighted=5.0M)

README.md

@@ -4,19 +4,29 @@ Frozen post-training DD-PPO PointNav agents on Habitat for five visual sensor
 conditions on a shared ResNet-18 + 3-layer LSTM (512-d) backbone. Hidden
 state `h_2` (top LSTM layer) is the canonical 512-d cognitive-map readout.
 
-| folder              | encoder                                                | final ckpt |
-| ------------------- | ------------------------------------------------------ | ---------- |
-| `blind/`            | no visual encoder                                      | `ckpt.34.pth` |
-| `coarse/`           | 48 x 48 RGB, 1 x 1 encoder feature map                 | `ckpt.49.pth` |
-| `foveated/`         | 256 x 256 RGB, eccentricity Gaussian blur, 4 x 4 map   | `ckpt.49.pth` |
-| `foveated_logpolar/`| 64 x 64 log-polar resampled, ~2 x 2 map                | `ckpt.49.pth` |
-| `uniform/`          | 256 x 256 RGB, no blur, 4 x 4 map                      | `ckpt.49.pth` |
+| folder              | encoder                                                | # ckpts | frames per ckpt | converged ckpt |
+| ------------------- | ------------------------------------------------------ | ------- | --------------- | -------------- |
+| `blind/`            | no visual encoder                                      | 35 (`0..34`) | 10.06 M         | `ckpt.34.pth` (~342 M frames) |
+| `coarse/`           | 48 x 48 RGB, 1 x 1 encoder feature map                 | 50 (`0..49`) | 5.0 M           | `ckpt.49.pth` (250 M frames)  |
+| `foveated/`         | 256 x 256 RGB, eccentricity Gaussian blur, 4 x 4 map   | 50 (`0..49`) | 5.0 M           | `ckpt.49.pth` (250 M frames)  |
+| `foveated_logpolar/`| 64 x 64 log-polar resampled, ~2 x 2 map                | 50 (`0..49`) | 5.0 M           | `ckpt.49.pth` (250 M frames)  |
+| `uniform/`          | 256 x 256 RGB, no blur, 4 x 4 map                      | 50 (`0..49`) | 5.0 M           | `ckpt.49.pth` (250 M frames)  |
 
 The **full training trajectory** is included for every condition (one
-checkpoint per DD-PPO save event): sighted ckpts `0..49` (50 each),
-blind ckpts `0..34` (35). This supports per-checkpoint training-time
-analyses (subspace evolution, probe-R^2 trajectories, eigenspectrum
-emergence, etc.) at the finest available granularity.
+checkpoint per DD-PPO save event). Note that `frames per ckpt` differs
+across conditions, so to align across conditions at the same training-step
+anchor, convert ckpt index to absolute frame count first:
+
+```python
+FRAMES_PER_CKPT_M = {
+    "blind":             10.06,
+    "coarse":             5.0,
+    "foveated":           5.0,
+    "foveated_logpolar":  5.0,
+    "uniform":            5.0,
+}
+# blind ckpt.20 ~= coarse ckpt.40 (both ~200 M frames trained)
+```
 
 ## Load a checkpoint
 
@@ -42,9 +52,6 @@ Each `.pth` is a habitat-baselines checkpoint with keys `state_dict`,
 
 ```python
 from habitat_baselines.common.baseline_registry import baseline_registry
-from habitat_baselines.utils.common import get_action_space_info
-from habitat_baselines.config.default import get_config
-from habitat.config.default_structured_configs import HabitatConfigPlugin
 
 # 1. Build the same env the policy was trained on (for obs/action spaces).
 env_config = config.habitat                  # already inside ckpt