README: larger section headings

README.md

@@ -11,7 +11,7 @@ tags:
 - robotwin
 ---
 
-## 📖 Overview
+# 📖 Overview
 
 **CAMP** (Compressed Action Memory Policy) gives a visuomotor policy a *behavioral memory*. A recurrent memory
 is pretrained to reconstruct a compressed (DCT) summary of the robot's own past actions, so its hidden state
@@ -26,7 +26,7 @@ policy is trained from the official RMBench demonstrations
 and evaluated with RMBench's own protocol.
 The files contain inference weights only.
 
-## 📊 Results
+# 📊 Results
 
 Success rate over **100 test episodes** per task. We follow RMBench's evaluation protocol: the `demo_clean`
 configuration, test seeds from 100000 that the scripted expert can solve, and the per-task step limit.
@@ -43,7 +43,7 @@ configuration, test seeds from 100000 that the scripted expert can solve, and th
 | [`observe_and_pickup`](#observe_and_pickup) | a past observation | 250 | 9% |
 | [`press_button`](#press_button) | counting | 1500 | 5% |
 
-## 🎬 Tasks
+# 🎬 Tasks
 
 Each preview is a **demonstration by RMBench's scripted expert**: the first of the 50 released demonstrations of
 the task, re-rendered so each task and its success condition can be seen clearly. It is filmed from the head
@@ -54,7 +54,7 @@ The full-resolution MP4 files are in [`previews/`](previews). Success rates in t
 100-episode policy evaluations.
 
 <a id="rearrange_blocks"></a>
-### rearrange_blocks: 100%
+## rearrange_blocks: 100%
 
 <p align="center"><img src="previews/rearrange_blocks.gif" width="70%"></p>
 
@@ -66,7 +66,7 @@ mat and presses the button. It then moves the second block off its mat to the sp
   between the mats. The button has been pressed exactly once and the gripper is open.
 
 <a id="blocks_ranking_try"></a>
-### blocks_ranking_try: 100%
+## blocks_ranking_try: 100%
 
 <p align="center"><img src="previews/blocks_ranking_try.gif" width="70%"></p>
 
@@ -80,7 +80,7 @@ and tests again, working through the orders until the button accepts one.
   been pressed.
 
 <a id="put_back_block"></a>
-### put_back_block: 100%
+## put_back_block: 100%
 
 <p align="center"><img src="previews/put_back_block.gif" width="70%"></p>
 
@@ -93,7 +93,7 @@ puts the block back on the mat it came from.
   3 cm of its original mat and the gripper is open.
 
 <a id="battery_try"></a>
-### battery_try: 97%
+## battery_try: 97%
 
 <p align="center"><img src="previews/battery_try.gif" width="70%"></p>
 
@@ -104,7 +104,7 @@ combination is correct. If it is not, the robot takes a battery out and re-inser
 - **Success:** both batteries are seated in the slot in the correct orientation and the dashboard turns on.
 
 <a id="swap_t"></a>
-### swap_T: 24%
+## swap_T: 24%
 
 <p align="center"><img src="previews/swap_T.gif" width="70%"></p>
 
@@ -116,7 +116,7 @@ and orientation.
   table, and both grippers are open.
 
 <a id="swap_blocks"></a>
-### swap_blocks: 19%
+## swap_blocks: 19%
 
 <p align="center"><img src="previews/swap_blocks.gif" width="70%"></p>
 
@@ -128,7 +128,7 @@ block. It swaps the two blocks using the spare tray as a buffer, then presses th
   and the gripper is open.
 
 <a id="cover_blocks"></a>
-### cover_blocks: 17%
+## cover_blocks: 17%
 
 <p align="center"><img src="previews/cover_blocks.gif" width="70%"></p>
 
@@ -140,7 +140,7 @@ blocks from left to right. Then it lifts the lids again in the order red, green,
   lifted.
 
 <a id="observe_and_pickup"></a>
-### observe_and_pickup: 9%
+## observe_and_pickup: 9%
 
 <p align="center"><img src="previews/observe_and_pickup.gif" width="70%"></p>
 
@@ -155,7 +155,7 @@ up the matching object from several distractors on the table.
   the single frame that shows the target would be skipped.
 
 <a id="press_button"></a>
-### press_button: 5%
+## press_button: 5%
 
 <p align="center"><img src="previews/press_button.gif" width="70%"></p>
 
@@ -165,7 +165,7 @@ the middle button as many times as the right card shows, then presses the right
 - **Memory:** how many presses each button has received so far. A button looks the same after every press.
 - **Success:** both press counts match the cards exactly and the confirm button has been pressed.
 
-## 📦 Files
+# 📦 Files
 
 ```
 <task>/
@@ -179,7 +179,7 @@ previews/<task>.gif | .mp4   # expert demonstrations shown above (1920×1440 MP4
 The checkpoints hold only what inference needs. There is no optimizer or scheduler state and no training
 bookkeeping.
 
-## 🧠 Training recipe
+# 🧠 Training recipe
 
 | | |
 |:--|:--|
@@ -191,7 +191,7 @@ bookkeeping.
 | Augmentation | joint noise 0.01, image noise 0.02, brightness and contrast jitter 0.15 |
 | Checkpoint | the best of the evaluated epochs per task (every 100 epochs) |
 
-## 🚀 Usage
+# 🚀 Usage
 
 Evaluation code, training launcher and full instructions are in the
 [RMBench section of the CAMP repository](https://github.com/KuanchengWang/CAMP#-rmbench). To evaluate
@@ -217,7 +217,7 @@ $E/docker_rmbench.sh python /workspace/$E/rmbench_eval.py eval  --task $T --ckpt
 To retrain a task from the official demonstrations with the same settings, run
 `bash scripts/rmbench/train_rmbench.sh <task>` inside the `camp` container.
 
-## 📝 Citation
+# 📝 Citation
 
 If you find CAMP useful, please cite: