Add hardest-case training and inference dynamics GIF

Author: OptimumAF

CHANGELOG.md

@@ -59,6 +59,7 @@ for versioning even while in research-stage development.
 - Updated circadian unit tests (NumPy + Torch) with coverage for reward scaling and adaptive budget behavior.
 - Updated README, model card, and core module docs to document new circadian controls.
 - Enhanced benchmark visuals with a compact combined overview figure (static + interactive) and linked it in README/dashboard for faster comparison.
+- Added hardest-case dynamics GIF (training progression + inference decision-map evolution) and surfaced it near the top of README and docs dashboard.
 - Refreshed README benchmark section with a latest master verification run on 2026-02-28 and added raw output artifact under `docs/benchmarks/`.
 - Repositioned repository messaging to Circadian Predictive Coding as the primary focus.
 - Updated `README.md` with:

README.md

@@ -34,6 +34,10 @@ flowchart LR
     H --> A
 ```
 
+### Hardest-Case Dynamics (Train + Inference)
+
+![Hardest-Case Dynamics](docs/figures/hardest_mode_dynamics.gif)
+
 ## Core Idea
 
 The circadian algorithm models wake and sleep phases:

docs/figures/README.md

@@ -8,6 +8,7 @@ This folder stores visual assets used by `README.md`.
 - `benchmark_accuracy.png`
 - `benchmark_train_speed.png`
 - `benchmark_inference_latency_p95.png`
+- `hardest_mode_dynamics.gif`
 - `circadian_sleep_dynamics.gif`
 - `interactive_benchmark_overview.html`
 - `interactive_benchmark_accuracy.html`
@@ -20,6 +21,7 @@ Run:
 
 ```powershell
 python scripts/generate_readme_figures.py --summary-csv benchmark_multiseed_cifar100_summary.csv --output-dir docs/figures
+python scripts/generate_hardest_mode_dynamics.py --output-path docs/figures/hardest_mode_dynamics.gif
 ```
 
 The GIF is illustrative and intended for communication in the README.

docs/figures/hardest_mode_dynamics.gif

@@ -109,6 +109,20 @@
       background: #fff;
     }
 
+    .hero-visual {
+      width: 100%;
+      border-radius: 10px;
+      border: 1px solid var(--border);
+      background: #fff;
+      box-shadow: 0 4px 14px rgba(30, 36, 48, 0.05);
+    }
+
+    .hero-caption {
+      margin: 8px 8px 0;
+      font-size: 0.9rem;
+      color: var(--muted);
+    }
+
     .footer {
       margin-top: 18px;
       color: var(--muted);
@@ -149,6 +163,14 @@ <h1>Circadian Predictive Coding Benchmark Dashboard</h1>
       </article>
     </section>
 
+    <section class="section">
+      <h2>Hardest-Case Dynamics (Training + Inference)</h2>
+      <img class="hero-visual" src="figures/hardest_mode_dynamics.gif" alt="Hardest-case training and inference dynamics GIF" />
+      <p class="hero-caption">
+        Left: training progression and phase-B inference accuracy over time. Right: circadian inference decision map with correct/incorrect overlays during hardest-case drift.
+      </p>
+    </section>
+
     <section class="section">
       <h2>Compact Overview (Interactive)</h2>
       <iframe src="figures/interactive_benchmark_overview.html" title="Interactive compact benchmark overview"></iframe>