"--task",
type=str,
default="twotargets",
- help="byheart, learnop, guessop, twotargets, addition, picoclvr, mnist, maze, snake, stack, expr, rpl",
+ help="byheart, learnop, guessop, twocuts, twotargets, addition, picoclvr, mnist, maze, snake, stack, expr, rpl, grid, qmlp",
)
parser.add_argument("--log_filename", type=str, default="train.log", help=" ")
########################################
-parser.add_argument("--nb_epochs", type=int, default=None)
+parser.add_argument("--nb_epochs", type=int, default=25)
parser.add_argument("--batch_size", type=int, default=None)
parser.add_argument("--rpl_no_prog", action="store_true", default=False)
+##############################
+# grid options
+
+parser.add_argument("--grid_size", type=int, default=6)
+
##############################
# picoclvr options
parser.add_argument("--expr_input_file", type=str, default=None)
##############################
-# World options
+# Misc
-parser.add_argument("--world_vqae_nb_epochs", type=int, default=25)
+parser.add_argument("--twocuts_no_global", action="store_true", default=False)
######################################################################
######################################################################
default_task_args = {
+ "addition": {
+ "model": "352M",
+ "batch_size": 25,
+ "nb_train_samples": 250000,
+ "nb_test_samples": 10000,
+ },
"byheart": {
"model": "37M",
- "nb_epochs": 2,
"batch_size": 25,
"nb_train_samples": 50000,
"nb_test_samples": 10000,
},
- "learnop": {
+ "expr": {
+ "model": "352M",
+ "batch_size": 25,
+ "nb_train_samples": 2500000,
+ "nb_test_samples": 10000,
+ },
+ "grid": {
"model": "37M",
- "nb_epochs": 15,
"batch_size": 25,
- "nb_train_samples": 50000,
+ "nb_train_samples": 250000,
"nb_test_samples": 10000,
},
+ "qmlp": {
+ "model": "37M",
+ "batch_size": 10,
+ "nb_train_samples": 100000,
+ "nb_test_samples": 1000,
+ },
"guessop": {
"model": "352M",
- "nb_epochs": 5,
"batch_size": 25,
"nb_train_samples": 1000000,
"nb_test_samples": 10000,
},
- "twotargets": {
+ "learnop": {
"model": "37M",
- "nb_epochs": 10,
"batch_size": 25,
"nb_train_samples": 50000,
"nb_test_samples": 10000,
},
- "addition": {
- "model": "352M",
- "nb_epochs": 50,
- "batch_size": 25,
- "nb_train_samples": 250000,
+ "maze": {
+ "model": "37M",
+ "batch_size": 5,
+ "nb_train_samples": 100000,
"nb_test_samples": 10000,
},
"picoclvr": {
"model": "37M",
- "nb_epochs": 25,
"batch_size": 25,
"nb_train_samples": 250000,
"nb_test_samples": 10000,
},
- "mnist": {
- "model": "37M",
- "nb_epochs": 25,
- "batch_size": 10,
- "nb_train_samples": 60000,
- "nb_test_samples": 10000,
- },
- "maze": {
- "model": "37M",
- "nb_epochs": 25,
+ "rpl": {
+ "model": "352M",
"batch_size": 5,
- "nb_train_samples": 100000,
+ "nb_train_samples": 2500000,
"nb_test_samples": 10000,
},
"snake": {
"model": "37M",
- "nb_epochs": 5,
"batch_size": 25,
"nb_train_samples": 250000,
"nb_test_samples": 10000,
},
"stack": {
"model": "37M",
- "nb_epochs": 15,
"batch_size": 25,
"nb_train_samples": 100000,
"nb_test_samples": 1000,
},
- "expr": {
- "model": "352M",
- "nb_epochs": 25,
+ "twotargets": {
+ "model": "37M",
"batch_size": 25,
- "nb_train_samples": 2500000,
+ "nb_train_samples": 50000,
"nb_test_samples": 10000,
},
- "rpl": {
- "model": "122M",
- "nb_epochs": 50,
- "batch_size": 5,
- "nb_train_samples": 1000000,
+ "twocuts": {
+ "model": "37M",
+ "batch_size": 25,
+ "nb_train_samples": 100000,
"nb_test_samples": 10000,
},
- "world": {
+ "mnist": {
"model": "37M",
- "nb_epochs": 10,
- "batch_size": 25,
- "nb_train_samples": 25000,
- "nb_test_samples": 1000,
+ "batch_size": 10,
+ "nb_train_samples": 60000,
+ "nb_test_samples": 10000,
},
}
device=device,
)
+elif args.task == "twocuts":
+ task = tasks.SandBox(
+ problem=problems.ProblemTwoCuts(global_constraint = not args.twocuts_no_global),
+ nb_train_samples=args.nb_train_samples,
+ nb_test_samples=args.nb_test_samples,
+ batch_size=args.batch_size,
+ logger=log_string,
+ device=device,
+ )
+
elif args.task == "addition":
task = tasks.SandBox(
problem=problems.ProblemAddition(),
device=device,
)
-elif args.task == "world":
- task = tasks.World(
+elif args.task == "grid":
+ task = tasks.Grid(
nb_train_samples=args.nb_train_samples,
nb_test_samples=args.nb_test_samples,
batch_size=args.batch_size,
- vqae_nb_epochs=args.world_vqae_nb_epochs,
+ size=args.grid_size,
+ logger=log_string,
+ device=device,
+ )
+
+elif args.task == "qmlp":
+ task = tasks.QMLP(
+ nb_train_samples=args.nb_train_samples,
+ nb_test_samples=args.nb_test_samples,
+ batch_size=args.batch_size,
+ result_dir=args.result_dir,
logger=log_string,
device=device,
)