PR-B: human-evals refactor + render pipeline + chunk fix (+ cleanup)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: m2kulkarni

.github/workflows/utest.yml

@@ -4,7 +4,6 @@ on:
   pull_request:
     branches: [ main ]
   push:
-    branches: [ main ]
 
 jobs:
   test:
@@ -21,13 +20,9 @@ jobs:
 
       - name: Free up disk space
         run: |
-          echo "Before cleanup:"
-          df -h
           sudo rm -rf /usr/share/dotnet /usr/local/lib/android /opt/ghc
           sudo apt-get clean
           sudo rm -rf ~/.cache/pip /tmp/* /var/tmp/*
-          echo "After cleanup:"
-          df -h
 
       - name: Install dependencies
         run: |
@@ -37,8 +32,7 @@ jobs:
         run: python -m pip install -U pip
 
       - name: Install pufferlib
-        run: |
-          pip install -e .[cpu] --no-cache-dir
+        run: pip install -e .[cpu] --no-cache-dir pytest
         env:
           TMPDIR: ${{ runner.temp }}/build
           PIP_NO_CACHE_DIR: 1
@@ -53,3 +47,11 @@ jobs:
 
       - name: Run python ini parsing tests
         run: python tests/test_drive_config.py
+
+      - name: Run pytest suite (conditioning, map_dir, render contracts)
+        run: |
+          pytest \
+            tests/test_drive_conditioning.py \
+            tests/test_map_dir_flow.py \
+            tests/test_render_pipeline.py \
+            -v --tb=short

docs/src/SUMMARY.md

@@ -20,6 +20,10 @@
 - [Evaluation overview](evaluation.md)
 - [WOSAC](wosac.md)
 
+# Design
+
+- [Trial mode (`goal_behavior=3`)](trial_mode.md)
+
 # Blog
 
 - [PufferDrive 2.0 release](pufferdrive-2.0.md)

docs/src/getting-started.md

@@ -25,7 +25,7 @@ python setup.py build_ext --inplace --force
 Run this with your virtual environment activated so the compiled extension links against the correct Python.
 
 ### When to rebuild the extension
-- Re-run `python setup.py build_ext --inplace --force` after changing any C/Raylib sources in `pufferlib/ocean/drive` (e.g., `drive.c`, `drive.h`, `binding.c`, `visualize.c`) or after pulling upstream changes that touch those files. This regenerates the `binding.cpython-*.so` used by `Drive`.
+- Re-run `python setup.py build_ext --inplace --force` after changing any C/Raylib sources in `pufferlib/ocean/drive` (e.g., `drive.h`, `binding.c`, `binding.h`) or after pulling upstream changes that touch those files. This regenerates the `binding.cpython-*.so` used by `Drive`.
 - Pure Python edits (training scripts, docs, data utilities) do not require a rebuild; just restart your Python process.
 
 ## Verify the setup

docs/src/simulator.md

@@ -210,15 +210,14 @@ mid_x, mid_y, length, width, dir_cos, dir_sin, type
 ## Source files
 
 ### C core
-- `drive.h`: Main simulator (stepping, observations, collisions)
-- `drive.c`: Demo and testing
-- `binding.c`: Python interface
-- `visualize.c`: Raylib renderer
-- `drivenet.h`: C inference network
+- `drive.h`: Main simulator (stepping, observations, collisions, raylib renderer)
+- `binding.c` / `binding.h`: Python interface
 
 ### Python
-- `drive.py`: Gymnasium wrapper
+- `drive.py`: Gymnasium wrapper, `Drive.render()`, `set_video_suffix()`
+- `rollout.py`: Policy-agnostic rollout loop used by both training renders and `render.py`
 - `torch.py`: Neural network (ego/partner/road encoders → actor/critic)
+- `../../../render.py`: Unified rendering CLI (replaces the old `./visualize` binary)
 
 ## Neural network