Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
159 commits
Select commit Hold shift + click to select a range
a6ee4a4
Goal behavior fixes (#124)
daphne-cornelisse Nov 11, 2025
c75b549
Update drive.h
eugenevinitsky Nov 13, 2025
d130cad
Add mode to only control the self-driving car (SDC) (#130)
daphne-cornelisse Nov 13, 2025
c26b245
Merge pull request #129 from Emerge-Lab/eugenevinitsky-patch-1
eugenevinitsky Nov 14, 2025
fecbb2d
Fix incorrect obs dim in draw_agent_obs (#109)
eugenevinitsky Nov 15, 2025
0439aa5
Replace product distribution action space with joint distribution (#104)
eugenevinitsky Nov 16, 2025
97dcb3d
Replace default ent_coef and learning_rate hparams (#134)
daphne-cornelisse Nov 16, 2025
040d39d
Add new weights binary with joint action space. (#136)
daphne-cornelisse Nov 18, 2025
058adff
Add support for logging optional evals during training (#133)
daphne-cornelisse Nov 18, 2025
7317d4f
Test for ini parsing (python and C) (#116)
aje-valeo Nov 20, 2025
c697f17
Fix missing arg (#141)
daphne-cornelisse Nov 20, 2025
2d30fa3
Add WOSAC interaction + map metrics. Switch from np -> torch. (#138)
WaelDLZ Nov 21, 2025
f8021df
Multi map render support to wandb (#143)
mpragnay Nov 24, 2025
87033d0
Add mode for controlled experiments (#144)
daphne-cornelisse Nov 24, 2025
aef6501
Little optimizations to use less memory in interaction_features.py (#…
daphne-cornelisse Nov 24, 2025
8690940
Fix broken link
daphne-cornelisse Nov 24, 2025
9d6a311
Data processing script that works decent. (#150)
daphne-cornelisse Nov 25, 2025
99060ba
Pass `map_dir` to the env via `.ini` and enable evaluation on a diffe…
daphne-cornelisse Nov 25, 2025
6eaea31
Add sprites in headless rendering (#152)
daphne-cornelisse Nov 26, 2025
a6af21c
Faster file processing (#153)
nadarenator Nov 27, 2025
6ce4879
Add link to small clean eval dataset
daphne-cornelisse Nov 27, 2025
0eab9bd
Fix link typo
daphne-cornelisse Nov 27, 2025
225ef99
Gif for readme (#155)
daphne-cornelisse Nov 27, 2025
f44573e
Fix link?
daphne-cornelisse Nov 27, 2025
b11d5e1
Fix vertical spaces.
daphne-cornelisse Nov 27, 2025
9c8b017
Update README.md
daphne-cornelisse Nov 27, 2025
eed2e39
Several small improvements for release (#159)
daphne-cornelisse Nov 30, 2025
95ceedd
WIP changes (#156)
nadarenator Dec 1, 2025
9d249b9
Releas note
daphne-cornelisse Dec 4, 2025
ff4f00c
Remove magic numbers in `drivenet.h`, set `MAX_AGENTS=32` by default …
daphne-cornelisse Dec 5, 2025
54affb0
Stable: Ensure all tests are passing (#168)
daphne-cornelisse Dec 6, 2025
a03e70a
Add option to zoom in on the map or show full map (#163)
Aditya-Gupta26 Dec 6, 2025
0cf5e3e
Add documentation (#170)
julianh65 Dec 8, 2025
1ce87a2
Add GitHub Actions workflow for docs deployment (#172)
julianh65 Dec 8, 2025
d03a69d
styling fixes (#173)
julianh65 Dec 9, 2025
217cd3b
Add clang format (#132)
aje-valeo Dec 12, 2025
9a4b09a
Add Sanity Command + Maps (#175)
julianh65 Dec 12, 2025
a8bce58
Documentation edits (#176)
daphne-cornelisse Dec 13, 2025
19b5eb6
Early environment resets based on agents' respawn status. (#167)
riccardosavorgnan Dec 13, 2025
43933b2
Speed up end-to-end training: 220K -> 320K on RTX 4080 by reducing # …
daphne-cornelisse Dec 14, 2025
0eeec10
Add pt. (#179)
daphne-cornelisse Dec 14, 2025
6fa0cbb
Docs edits (#178)
daphne-cornelisse Dec 15, 2025
712ac4f
There is a typo in torch.py
WaelDLZ Dec 15, 2025
76dcce1
Merge pull request #181 from Emerge-Lab/wbd/fix_a_typo_in_torchpy
eugenevinitsky Dec 15, 2025
57be883
Use num_maps for eval (#164)
WaelDLZ Dec 16, 2025
fae2b59
Fix small bug in `drive.c` and add binary weights cpt (#184)
daphne-cornelisse Dec 16, 2025
f349b24
Carla junction filter (#187)
mpragnay Dec 17, 2025
4900039
Working Carla Maps (#189)
mpragnay Dec 18, 2025
656d55a
collision fix (#192)
l1onh3art88 Dec 18, 2025
a033669
Fix Ego Speed Calculation (#166)
julianh65 Dec 19, 2025
273d8f2
Small bug fix that makes road edge not appear in agent view for jerk …
daphne-cornelisse Dec 19, 2025
685b3fd
add womd video (#195)
julianh65 Dec 20, 2025
6f0ca30
Add stop/remove collision behavior back (#169)
Aditya-Gupta26 Dec 20, 2025
61911a3
updated docs with multinode training cmd (#174)
mpragnay Dec 20, 2025
45588ca
Carla2d towns (#201)
mpragnay Dec 21, 2025
657281c
initial commit (#204)
julianh65 Dec 22, 2025
c0c5894
Fix goal resampling in Carla maps and make metrics suitable for resam…
daphne-cornelisse Dec 23, 2025
9a58142
Minor correction in resampling code (#183)
WaelDLZ Dec 23, 2025
b6ed82f
Allow human to drive with agents through classic and jerk dynamics mo…
daphne-cornelisse Dec 24, 2025
d4dabdb
Added WOSAC results on the 10k validation dataset (#185)
WaelDLZ Dec 25, 2025
459e875
Drive with agents in browser (#215)
daphne-cornelisse Dec 26, 2025
952069d
Fix demo (#217)
daphne-cornelisse Dec 26, 2025
7ca82c2
Do not randomly switch to another agent in FPV. (#219)
daphne-cornelisse Dec 27, 2025
38139bb
switch docs to mdbooks doc format (#218)
eugenevinitsky Dec 27, 2025
b89e157
Markdown edits and fix demo. (#221)
daphne-cornelisse Dec 28, 2025
bd20606
small fixes in the docs (#220)
eugenevinitsky Dec 28, 2025
07a7a1d
Release 2.0 (#214)
daphne-cornelisse Dec 30, 2025
cb12de1
Fix space and game files.
Dec 30, 2025
1d864db
Fix sup tags.
Dec 30, 2025
978d8a1
rebasing ada with pufferDrive 2
Jan 12, 2026
ce19279
Self play working
Jan 13, 2026
eb34531
Population play and self play rebased
Jan 14, 2026
ab2c29c
All features working
Jan 15, 2026
4d6136d
fixing co player features
Jan 15, 2026
64da48d
trying to pass tests
Jan 15, 2026
70b3148
fixing tests #2
Jan 15, 2026
299f4ff
fixing tests #3
Jan 15, 2026
ace6fee
attempting to fix tests #4
Jan 16, 2026
8f0eab0
attempting to fix tests #4
Jan 16, 2026
951062b
attempting to fix tests #4
Jan 16, 2026
ba472aa
fixing batch size > 1 bug
Jan 16, 2026
3082161
add back binary
m2kulkarni Jan 31, 2026
027bda6
Merge branch 'sync-upstream' of github.com:Emerge-Lab/Adaptive_Drivin…
m2kulkarni Jan 31, 2026
f289654
changed map dir
m2kulkarni Jan 31, 2026
4254f3f
moved maps
m2kulkarni Jan 31, 2026
1087d38
fix test API and config
m2kulkarni Jan 31, 2026
30a0636
Merge branch 'main' into sync-upstream
m2kulkarni Jan 31, 2026
0e829cf
Merge remote-tracking branch 'origin/main' into sync-upstream
m2kulkarni Jan 31, 2026
c657e25
fixing tests
m2kulkarni Feb 1, 2026
654970e
fix tests
m2kulkarni Feb 1, 2026
9403611
Transformer architecture (#23)
charliemolony Feb 16, 2026
1aff20a
Mohit/heartbeat (#25)
m2kulkarni Feb 16, 2026
a947e1e
Refactoring human evals (#27)
charliemolony Feb 20, 2026
6d8d963
fixed rendering during eval
m2kulkarni Feb 24, 2026
c2e88ec
baseline script
m2kulkarni Feb 24, 2026
60507fd
update to rendering and fixes -- works at this commit wandb run id - …
m2kulkarni Feb 26, 2026
8196587
pre-commit
m2kulkarni Feb 26, 2026
7fe4ab6
changed drive.ini to make recurrent default
m2kulkarni Feb 26, 2026
cfa46f0
womd works nuplan doesnt
m2kulkarni Feb 27, 2026
5954e39
womd works, nuplan hangs - organized scripts
m2kulkarni Feb 27, 2026
243c4ac
ada_new_coplayers from here
m2kulkarni Feb 27, 2026
9bc70ca
unified context lenght, context window etc
m2kulkarni Feb 27, 2026
be81b52
updated to 91 instead of 32
m2kulkarni Feb 27, 2026
c462d3a
pre-commit and some fixes
m2kulkarni Feb 27, 2026
69a0054
coplayer reset between trials
m2kulkarni Mar 2, 2026
3f6656d
update scenario logging
m2kulkarni Mar 12, 2026
53bb828
Bug transformer fix (#31)
charliemolony Mar 12, 2026
ddcdbb7
fixed issues
m2kulkarni Mar 12, 2026
27c7f11
new nuplan transformer coplayers
m2kulkarni Apr 1, 2026
5dd1216
nuplan transformer works
m2kulkarni Apr 4, 2026
314486f
transformer optimizeation
m2kulkarni Apr 4, 2026
d656204
cpu device
m2kulkarni Apr 4, 2026
ef4b2c5
slight human replay fix
m2kulkarni Apr 5, 2026
2076c31
Fix multiprocessing deadlock + enable compile/bfloat16 for 49% speedup
m2kulkarni Apr 8, 2026
11f35a8
Fix co-player support for Transformer + bfloat16
m2kulkarni Apr 9, 2026
f01eb80
working renders transformer
m2kulkarni Apr 17, 2026
a4550b1
Add --eval.map-dir to nuplan scripts for correct eval paths
m2kulkarni Apr 25, 2026
287ebcf
Add human_align_ablation.sh script for 6 co-player conditioning exper…
m2kulkarni Apr 25, 2026
22684c9
bad commit. the one before this is the perfect one
m2kulkarni Apr 25, 2026
c028a29
cirrect map dir for human replay
mohitmk01 Apr 25, 2026
5554d95
lane rewards
m2kulkarni Apr 25, 2026
ba32f23
lane rewards more fixes
m2kulkarni Apr 25, 2026
51c168c
fixed hardcoded values
m2kulkarni Apr 25, 2026
f8fe56a
major refactor by claude not tsted yet
m2kulkarni Apr 25, 2026
51d34b5
script changes
m2kulkarni Apr 25, 2026
9652056
sidecar autoload and tests and pre=commit
m2kulkarni Apr 25, 2026
f2968e3
minor changes
m2kulkarni Apr 25, 2026
525cf61
KV-cached forward_eval for TransformerWrapper (~9x co-player speedup)
mohitmk01 Apr 25, 2026
b5de3e9
Add PUFFER_TRANSFORMER_LEGACY_EVAL escape hatch
mohitmk01 Apr 25, 2026
071927b
Make KV-cached forward_eval compile-friendly + add real-checkpoint eq…
mohitmk01 Apr 25, 2026
81a3207
render fix
m2kulkarni Apr 26, 2026
3df8ea3
Merge branch 'mohit/cleanup-apr25' of github.com:Emerge-Lab/Adaptive_…
mohitmk01 Apr 26, 2026
0a8a580
commentsed and render
m2kulkarni Apr 26, 2026
6574160
Bump adaptive.ini default num_workers/num_envs to 32
mohitmk01 Apr 26, 2026
621ed4f
removed comments
m2kulkarni Apr 26, 2026
0f36823
nothing
mohitmk01 Apr 26, 2026
22070dd
Merge branch 'mohit/cleanup-apr25' of github.com:Emerge-Lab/Adaptive_…
mohitmk01 Apr 26, 2026
1db628d
fixed evaluations such that we now run num-rollouts and also do goal …
mohitmk01 Apr 27, 2026
5b48e72
Resume infrastructure: trainer_state load + wandb run resume + render…
mohitmk01 Apr 27, 2026
a62c346
Fix --train.cpu-offload True for adaptive training
mohitmk01 Apr 27, 2026
57312e1
Centralized co-player inference on GPU (external_co_player_actions)
mohitmk01 Apr 27, 2026
01a3d36
fixes
mohitmk01 Apr 28, 2026
b2ecee2
precommit
mohitmk01 Apr 28, 2026
e7ec7b4
Make trajectory_length parametric in process_all_maps
mohitmk01 Apr 28, 2026
61183cf
Add conditional recovery eval pipeline + 201-frame coplayer launcher
mohitmk01 Apr 28, 2026
c7f7292
Add map_rand_per_scenario + raylib survival across env reinit
mohitmk01 Apr 29, 2026
6bf11f6
Add city-adapt, diverse-partner, entropy-sweep + probes
mohitmk01 Apr 30, 2026
3507bc6
Add entropy_ub curriculum + per-resample wandb logging
mohitmk01 Apr 30, 2026
b037645
Add k_eff curriculum: anneal ego cache horizon 1 → 2 → K_max
mohitmk01 Apr 30, 2026
c0b5a19
Add entropy_curriculum_episodes_start kwarg + resume launcher
mohitmk01 May 1, 2026
f3d04e4
latest 2nd may
mohitmk01 May 1, 2026
5c1ca48
Fix critical bug: persist K/V cache across rollout steps
mohitmk01 May 2, 2026
acd3041
Reset transformer pos to 0 (not -1) on episode boundary
mohitmk01 May 2, 2026
eb2d7fe
Switch transformer PE from learnable to sinusoidal (Vaswani)
mohitmk01 May 2, 2026
1ad3857
Add nuplan_hard split + reward_only_last_scenario
mohitmk01 May 2, 2026
6ffc4e3
Add k2_201 sweep launchers (partner sweep + earlier ablations)
mohitmk01 May 3, 2026
5d9bbb4
did i finally fix things? mask still there
mohitmk01 May 4, 2026
d3b59a3
cleanup debug
mohitmk01 May 4, 2026
4dd8a01
removed some comments
mohitmk01 May 4, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .claude/scheduled_tasks.lock
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"sessionId":"7bcf951d-92e0-44b9-83b1-dc29b0d69f18","pid":8923,"procStart":"408700719","acquiredAt":1777664169801}
16 changes: 9 additions & 7 deletions .github/workflows/utest.yml
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,6 @@ on:
pull_request:
branches: [ main ]
push:
branches: [ main ]

jobs:
test:
Expand All @@ -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: |
Expand All @@ -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
Expand All @@ -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
12 changes: 10 additions & 2 deletions .gitignore
Original file line number Diff line number Diff line change
Expand Up @@ -161,13 +161,16 @@ pufferlib/ocean/impulse_wars/debug-*/
pufferlib/ocean/impulse_wars/release-*/
pufferlib/ocean/impulse_wars/benchmark/


# Ignore data files
data/
pufferlib/resources/drive/binaries/
pufferlib/resources/drive/binaries/*
pufferlib/resources/drive/binaries/training/
pufferlib/resources/drive/binaries/validation/

# But keep map_000.bin for the training test
!pufferlib/resources/drive/binaries/map_000.bin
!pufferlib/resources/drive/binaries/training/map_000.bin
pufferlib/resources/drive/sanity/sanity_binaries/

# Compiled drive binary in root
/drive
Expand All @@ -183,7 +186,12 @@ pufferlib/resources/drive/output_agent.gif pufferlib/resources/drive/output.gif
# Local artifacts and outputs
artifacts/
# Local drive renders
pufferlib/resources/drive/output*.gif
emsdk/
docs/book/*
!docs/book/assets/
pufferlib/resources/drive/output*.mp4

# Local TODO tracking
TODO.md
*.mp4
Loading
Loading