GitHub - AdarshGoel2001/orbitWars: RL self-play agent for a 1-second-per-turn CPU-budget Kaggle game. 100× inference speedup via radar + dynamic tokens + axial attention.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 17 Commits
.playwright-mcp		.playwright-mcp
checkpoints		checkpoints
data		data
kaggle_benchmarks		kaggle_benchmarks
logs		logs
markdowns		markdowns
orbit_wars		orbit_wars
runs		runs
tests		tests
.codex		.codex
.env.example		.env.example
.gitignore		.gitignore
AGENTS.md		AGENTS.md
CLAUDE.md		CLAUDE.md
GPU-CPU-plan.md		GPU-CPU-plan.md
KaggleBriefer.md		KaggleBriefer.md
LONG_TERM_FIXES.md		LONG_TERM_FIXES.md
action_space.py		action_space.py
agents.py		agents.py
agents_cpu.py		agents_cpu.py
bc_data.py		bc_data.py
bc_data_cpu.py		bc_data_cpu.py
bc_train.py		bc_train.py
bc_train_chunked_cpu.py		bc_train_chunked_cpu.py
bc_train_cpu.py		bc_train_cpu.py
bench.py		bench.py
bench_cpu.py		bench_cpu.py
chunked_bc_cpu.py		chunked_bc_cpu.py
chunked_model_cpu.py		chunked_model_cpu.py
eval_bc.py		eval_bc.py
eval_bc_cpu.py		eval_bc_cpu.py
getting-started.ipynb		getting-started.ipynb
harness.py		harness.py
harness_cpu.py		harness_cpu.py
make_chunked_submission.py		make_chunked_submission.py
make_cpu_submission.py		make_cpu_submission.py
make_nn_submission.py		make_nn_submission.py
make_submission.py		make_submission.py
model.py		model.py
model_cpu.py		model_cpu.py
radar.py		radar.py
rl_opponent_pool.py		rl_opponent_pool.py
rl_opponent_pool_cpu.py		rl_opponent_pool_cpu.py
rl_ppo.py		rl_ppo.py
rl_ppo_cpu.py		rl_ppo_cpu.py
rl_rollout.py		rl_rollout.py
rl_rollout_cpu.py		rl_rollout_cpu.py
rl_train.py		rl_train.py
rl_train_chunked_cpu.py		rl_train_chunked_cpu.py
rl_train_cpu.py		rl_train_cpu.py
server.py		server.py
server_model_eyes.py		server_model_eyes.py
submission.py		submission.py
submission_chunked.py		submission_chunked.py
submission_chunked_m2.py		submission_chunked_m2.py
submission_cpu.py		submission_cpu.py
submission_nn.tar.gz		submission_nn.tar.gz
submission_nn.zip		submission_nn.zip
targeting.py		targeting.py

About

RL self-play agent for a 1-second-per-turn CPU-budget Kaggle game. 100× inference speedup via radar + dynamic tokens + axial attention.

www.kaggle.com/competitions/orbit-wars/overview

Activity

1 star

0 watching

0 forks

Report repository

Navigation Menu

About

Releases

Packages

Contributors

Languages

Folders and files

Latest commit

History

About

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages