[WIP, CI] Pre-release submitit scripts

benchmarks/sota-check/README.md

@@ -0,0 +1,8 @@
+# SOTA Performance checks
+
+This folder contains a `submitit-release-check.sh` file that executed all
+the training scripts using `sbatch` with the default configuration and a wandb
+logger.
+
+This script is to be executed before every release to assess the performance of
+the various algorithms available in torchrl.

benchmarks/sota-check/run_a2c_atari.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=a2c_atari
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=a2c_atari_output_%j.txt
+#SBATCH --error=a2c_atari_error_%j.txt
+
+python ../../examples/a2c/a2c_atari.py

benchmarks/sota-check/run_a2c_mujoco.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=a2c_mujoco
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=a2c_mujoco_output_%j.txt
+#SBATCH --error=a2c_mujoco_error_%j.txt
+
+python ../../examples/a2c/a2c_mujoco.py

benchmarks/sota-check/run_bandits.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=bandits_dqn
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=bandits_dqn_output_%j.txt
+#SBATCH --error=bandits_dqn_error_%j.txt
+
+python ../../examples/bandits/dqn.py

benchmarks/sota-check/run_cql_offline.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=cql_offline
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=cql_offline_output_%j.txt
+#SBATCH --error=cql_offline_error_%j.txt
+
+python ../../examples/cql/cql_offline.py

benchmarks/sota-check/run_cql_online.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=cql_online
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=cql_online_output_%j.txt
+#SBATCH --error=cql_online_error_%j.txt
+
+python ../../examples/cql/cql_online.py

benchmarks/sota-check/run_ddpg.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=ddpg
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=ddpg_output_%j.txt
+#SBATCH --error=ddpg_error_%j.txt
+
+python ../../examples/ddpg/ddpg.py

benchmarks/sota-check/run_discrete_sac.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=discrete_sac
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=discrete_sac_output_%j.txt
+#SBATCH --error=discrete_sac_error_%j.txt
+
+python ../../examples/discrete_sac/discrete_sac.py

benchmarks/sota-check/run_dqn_atari.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=dqn_atari
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=dqn_atari_output_%j.txt
+#SBATCH --error=dqn_atari_error_%j.txt
+
+python ../../examples/dqn/dqn_atari.py

benchmarks/sota-check/run_dqn_cartpole.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=dqn_cartpole
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=dqn_cartpole_output_%j.txt
+#SBATCH --error=dqn_cartpole_error_%j.txt
+
+python ../../examples/dqn/dqn_cartpole.py

benchmarks/sota-check/run_dreamer.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=dreamer
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=dreamer_output_%j.txt
+#SBATCH --error=dreamer_error_%j.txt
+
+python ../../examples/dreamer/dreamer.py

benchmarks/sota-check/run_dt_offline.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=dt_offline
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=dt_offline_output_%j.txt
+#SBATCH --error=dt_offline_error_%j.txt
+
+python ../../examples/dt/dt_offline.py

benchmarks/sota-check/run_dt_online.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=dt_online
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=dt_online_output_%j.txt
+#SBATCH --error=dt_online_error_%j.txt
+
+python ../../examples/dt/dt_online.py

benchmarks/sota-check/run_impala.sh

@@ -0,0 +1,11 @@
+Shell Script
+
+#!/bin/bash
+#SBATCH --job-name=a2c_job
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=a2c_output_%j.txt
+#SBATCH --error=a2c_error_%j.txt
+module load python/3.7.4
+python a2c.py

benchmarks/sota-check/run_iql_offline.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=iql_offline
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=iql_offline_output_%j.txt
+#SBATCH --error=iql_offline_error_%j.txt
+
+python ../../examples/iql/iql_offline.py

benchmarks/sota-check/run_iql_online.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=iql_online
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=iql_online_output_%j.txt
+#SBATCH --error=iql_online_error_%j.txt
+
+python ../../examples/iql/iql_online.py

benchmarks/sota-check/run_multiagent_iddpg.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=marl_iddpg
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=marl_iddpg_output_%j.txt
+#SBATCH --error=marl_iddpg_error_%j.txt
+
+python ../../examples/multiagent/maddpg_iddpg.py

benchmarks/sota-check/run_multiagent_ippo.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=marl_ippo
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=marl_ippo_output_%j.txt
+#SBATCH --error=marl_ippo_error_%j.txt
+
+python ../../examples/multiagent/maddpg_ippo.py

benchmarks/sota-check/run_multiagent_iql.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=qmix_vdn
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=qmix_vdn_output_%j.txt
+#SBATCH --error=qmix_vdn_error_%j.txt
+
+python ../../examples/multiagent/qmix_vdn.py

benchmarks/sota-check/run_multiagent_qmix.sh

@@ -0,0 +1,11 @@
+Shell Script
+
+#!/bin/bash
+#SBATCH --job-name=a2c_job
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=a2c_output_%j.txt
+#SBATCH --error=a2c_error_%j.txt
+module load python/3.7.4
+python a2c.py

benchmarks/sota-check/run_multiagent_sac.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=marl_sac
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=marl_sac_output_%j.txt
+#SBATCH --error=marl_sac_error_%j.txt
+
+python ../../examples/multiagent/sac.py

benchmarks/sota-check/run_ppo_atari.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=ppo_atari
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=ppo_atari_output_%j.txt
+#SBATCH --error=ppo_atari_error_%j.txt
+
+python ../../examples/ppo/ppo_atari.py

benchmarks/sota-check/run_ppo_mujoco.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=ppo_mujoco
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=ppo_mujoco_output_%j.txt
+#SBATCH --error=ppo_mujoco_error_%j.txt
+
+python ../../examples/ppo/ppo_mujoco.py

benchmarks/sota-check/run_redq.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=redq
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=redq_output_%j.txt
+#SBATCH --error=redq_error_%j.txt
+
+python ../../examples/redq/redq.py

benchmarks/sota-check/run_rlhf.sh

@@ -0,0 +1,11 @@
+Shell Script
+
+#!/bin/bash
+#SBATCH --job-name=a2c_job
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=a2c_output_%j.txt
+#SBATCH --error=a2c_error_%j.txt
+module load python/3.7.4
+python a2c.py

benchmarks/sota-check/run_sac.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=sac
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=sac_output_%j.txt
+#SBATCH --error=sac_error_%j.txt
+
+python ../../examples/sac/sac.py

benchmarks/sota-check/run_td3.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+#SBATCH --job-name=td3
+#SBATCH --ntasks=32
+#SBATCH --cpus-per-task=1
+#SBATCH --gres=gpu:1
+#SBATCH --output=td3_output_%j.txt
+#SBATCH --error=td3_error_%j.txt
+
+python ../../examples/td3/td3.py

benchmarks/sota-check/submitit-release-check.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+
+sbatch run_a2c_atari.sh
+sbatch run_a2c_mujoco.sh
+sbatch run_bandits.sh
+sbatch run_cql_offline.sh
+sbatch run_cql_online.sh
+sbatch run_ddpg.sh
+sbatch run_discrete_sac.sh
+sbatch run_dqn_atari.sh
+sbatch run_dqn_cartpole.sh
+sbatch run_dreamer.sh
+sbatch run_dt_offline.sh
+sbatch run_dt_online.sh
+sbatch run_impala.sh
+sbatch run_iql_offline.sh
+sbatch run_iql_online.sh
+sbatch run_multiagent_iddpg.sh
+sbatch run_multiagent_ippo.sh
+sbatch run_multiagent_iql.sh
+sbatch run_multiagent_qmix.sh
+sbatch run_multiagent_sac.sh
+sbatch run_ppo_atari.sh
+sbatch run_ppo_mujoco.sh
+sbatch run_redq.sh
+sbatch run_rlhf.sh
+sbatch run_sac.sh
+sbatch run_td3.sh