From de4f5457f0a84d8339e21ed6b519801279a039e8 Mon Sep 17 00:00:00 2001
From: Severin Magel <116261790+sevmag@users.noreply.github.com>
Date: Sun, 2 Aug 2026 13:22:28 -0400
Subject: [PATCH 1/2] README: block diagram of the components at the top

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 README.md | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/README.md b/README.md
index b7f6c66..09a5c93 100644
--- a/README.md
+++ b/README.md
@@ -9,6 +9,17 @@ The first pretext task is **CURTAIN** (occupancy / light-front forecast). The
 architecture is built so a new self-supervised method is a small plugin under
 `pretext/`, reusing the data, backbone, and training engine unchanged.
 
+```mermaid
+flowchart LR
+    R["your reader<br/>raw events"] --> T["PretextTask<br/>sample, scale, collate"]
+    G["geometry asset"] --> T
+    T -- jagged batch --> B["Backbone<br/>encoder, e.g. DeepIce"]
+    B -- token embeddings --> H["objective heads + loss"]
+    H --> M["SSLModule + fit<br/>Lightning"]
+    M -- best val export --> X["TransferCheckpoint<br/>backbone weights"]
+    X --> F["your fine tune<br/>e.g. graphnet"]
+```
+
 ## 🧭 Philosophy
 The core is framework-agnostic and fits neatly into plain PyTorch: it depends
 only on torch, pytorch-lightning and numpy. Readers are ordinary indexable

From f00022a4ed805f6f98ad6dbff8e0effdb5d4ec76 Mon Sep 17 00:00:00 2001
From: Severin Magel <116261790+sevmag@users.noreply.github.com>
Date: Sun, 2 Aug 2026 13:25:38 -0400
Subject: [PATCH 2/2] README: simpler, larger component diagram

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 README.md | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/README.md b/README.md
index 09a5c93..e9431d7 100644
--- a/README.md
+++ b/README.md
@@ -11,13 +11,10 @@ architecture is built so a new self-supervised method is a small plugin under
 
 ```mermaid
 flowchart LR
-    R["your reader<br/>raw events"] --> T["PretextTask<br/>sample, scale, collate"]
-    G["geometry asset"] --> T
-    T -- jagged batch --> B["Backbone<br/>encoder, e.g. DeepIce"]
-    B -- token embeddings --> H["objective heads + loss"]
-    H --> M["SSLModule + fit<br/>Lightning"]
-    M -- best val export --> X["TransferCheckpoint<br/>backbone weights"]
-    X --> F["your fine tune<br/>e.g. graphnet"]
+    D["your data<br/>reader + geometry"] --> T["PretextTask<br/>e.g. CURTAIN"]
+    T --> B["Backbone<br/>e.g. DeepIce"]
+    B --> H["objective heads<br/>+ loss"]
+    H --> X["pretrained backbone<br/>for your fine tune"]
 ```
 
 ## 🧭 Philosophy