Update.

diff --git a/evasion.py b/escape.py
similarity index 97%
rename from evasion.py
rename to escape.py
index 9f37494..7fcafaa 100755 (executable)
--- a/evasion.py
+++ b/escape.py
@@ -1,5 +1,10 @@
 #!/usr/bin/env python
 
+# Any copyright is dedicated to the Public Domain.
+# https://creativecommons.org/publicdomain/zero/1.0/
+
+# Written by Francois Fleuret <francois@fleuret.org>
+
 import torch
 
 from torch.nn import functional as F

diff --git a/main.py b/main.py
index 958dfc7..b4f831f 100755 (executable)
--- a/main.py
+++ b/main.py
@@ -33,7 +33,7 @@ parser.add_argument(
     "--task",
     type=str,
     default="twotargets",
-    help="file, byheart, learnop, guessop, mixing, memory, twotargets, addition, picoclvr, mnist, maze, snake, stack, expr, rpl, grid, qmlp",
+    help="file, byheart, learnop, guessop, mixing, memory, twotargets, addition, picoclvr, mnist, maze, snake, stack, expr, rpl, grid, qmlp, escape",
 )
 
 parser.add_argument("--log_filename", type=str, default="train.log", help=" ")
@@ -175,6 +175,15 @@ parser.add_argument("--mixing_hard", action="store_true", default=False)
 
 parser.add_argument("--mixing_deterministic_start", action="store_true", default=False)
 
+##############################
+# escape options
+
+parser.add_argument("--escape_height", type=int, default=4)
+
+parser.add_argument("--escape_width", type=int, default=6)
+
+parser.add_argument("--escape_T", type=int, default=20)
+
 ######################################################################
 
 args = parser.parse_args()
@@ -289,6 +298,12 @@ default_task_args = {
         "nb_train_samples": 60000,
         "nb_test_samples": 10000,
     },
+    "escape": {
+        "model": "37M",
+        "batch_size": 25,
+        "nb_train_samples": 25000,
+        "nb_test_samples": 10000,
+    },
 }
 
 if args.task in default_task_args:
@@ -599,6 +614,18 @@ elif args.task == "qmlp":
         device=device,
     )
 
+elif args.task == "escape":
+    task = tasks.Escape(
+        nb_train_samples=args.nb_train_samples,
+        nb_test_samples=args.nb_test_samples,
+        batch_size=args.batch_size,
+        height=args.escape_height,
+        width=args.escape_width,
+        T=args.escape_T,
+        logger=log_string,
+        device=device,
+    )
+
 else:
     raise ValueError(f"Unknown task {args.task}")
 

diff --git a/tasks.py b/tasks.py
index d680951..8b22cad 100755 (executable)
--- a/tasks.py
+++ b/tasks.py
@@ -1861,3 +1861,61 @@ class QMLP(Task):
 
 
 ######################################################################
+
+import escape
+
+
+class Escape(Task):
+    def __init__(
+        self,
+        nb_train_samples,
+        nb_test_samples,
+        batch_size,
+        height,
+        width,
+        T,
+        logger=None,
+        device=torch.device("cpu"),
+    ):
+        super().__init__()
+
+        self.batch_size = batch_size
+        self.device = device
+
+        states, actions, rewards = escape.generate_episodes(
+            nb_train_samples + nb_test_samples, height, width, T
+        )
+        seq = escape.episodes2seq(states, actions, rewards)
+        self.train_input = seq[:nb_train_samples]
+        self.test_input = seq[nb_train_samples:]
+
+        self.nb_codes = max(self.train_input.max(), self.test_input.max()) + 1
+
+        # if logger is not None:
+        # for s, a in zip(self.train_input[:100], self.train_ar_mask[:100]):
+        # logger(f"train_sequences {self.problem.seq2str(s)}")
+        # a = "".join(["01"[x.item()] for x in a])
+        # logger(f"                {a}")
+
+    def batches(self, split="train", nb_to_use=-1, desc=None):
+        assert split in {"train", "test"}
+        input = self.train_input if split == "train" else self.test_input
+        if nb_to_use > 0:
+            input = input[:nb_to_use]
+        if desc is None:
+            desc = f"epoch-{split}"
+        for batch in tqdm.tqdm(
+            input.split(self.batch_size), dynamic_ncols=True, desc=desc
+        ):
+            yield batch
+
+    def vocabulary_size(self):
+        return self.nb_codes
+
+    def produce_results(
+        self, n_epoch, model, result_dir, logger, deterministic_synthesis, nmax=1000
+    ):
+        pass
+
+
+######################################################################