Update.

[picoclvr.git] / tasks.py

diff --git a/tasks.py b/tasks.py
index 0f80d4f..2f3db6a 100755 (executable)
--- a/tasks.py
+++ b/tasks.py
@@ -1885,20 +1885,15 @@ class Escape(Task):
         self.width = width
 
         states, actions, rewards = escape.generate_episodes(
-            nb_train_samples + nb_test_samples, height, width, T
+            nb_train_samples + nb_test_samples, height, width, 3 * T
         )
-        seq = escape.episodes2seq(states, actions, rewards, lookahead_delta=5)
+        seq = escape.episodes2seq(states, actions, rewards, lookahead_delta=T)
+        seq = seq[:, seq.size(1) // 3 : 2 * seq.size(1) // 3]
         self.train_input = seq[:nb_train_samples].to(self.device)
         self.test_input = seq[nb_train_samples:].to(self.device)
 
         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
@@ -1918,6 +1913,23 @@ class Escape(Task):
         self, n_epoch, model, result_dir, logger, deterministic_synthesis, nmax=1000
     ):
         result = self.test_input[:100].clone()
+
+        # Saving the ground truth
+
+        s, a, r, lr = escape.seq2episodes(
+            result, self.height, self.width, lookahead=True
+        )
+        str = escape.episodes2str(
+            s, a, r, lookahead_rewards=lr, unicode=True, ansi_colors=True
+        )
+
+        filename = os.path.join(result_dir, f"test_true_seq_{n_epoch:04d}.txt")
+        with open(filename, "w") as f:
+            f.write(str)
+            logger(f"wrote {filename}")
+
+        # Re-generating from the first frame
+
         ar_mask = (
             torch.arange(result.size(1), device=result.device)
             > self.height * self.width + 2
@@ -1934,6 +1946,8 @@ class Escape(Task):
             device=self.device,
         )
 
+        # Saving the generated sequences
+
         s, a, r, lr = escape.seq2episodes(
             result, self.height, self.width, lookahead=True
         )