X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=tasks.py;h=ee06c25e1923ce1a69880384526b1a1636064a5f;hb=9707563cb32ed2335dc4a6edddaa0ebe9cfd1243;hp=5edb472024342565801fee1f7de48f55bb17756a;hpb=702e672dcf9ebcfad11ae4034e64117f2c67ead5;p=culture.git

diff --git a/tasks.py b/tasks.py
index 5edb472..ee06c25 100755
--- a/tasks.py
+++ b/tasks.py
@@ -22,6 +22,7 @@ def masked_inplace_autoregression(
     batch_size,
     input,
     ar_mask,
+    summed_logits,
     temperature,
     deterministic_synthesis,
     forbidden_tokens=None,
@@ -45,12 +46,11 @@ def masked_inplace_autoregression(
         t = model.training
         model.eval()
 
-        sum_logits = 0
-
         for input, ar_mask in batches:
-            sum_logits += model.masked_inplace_autoregression(
+            model.masked_inplace_autoregression(
                 input=input,
                 ar_mask=ar_mask,
+                summed_logits=summed_logits,
                 temperature=temperature,
                 deterministic_synthesis=deterministic_synthesis,
                 forbidden_tokens=forbidden_tokens,
@@ -59,8 +59,6 @@ def masked_inplace_autoregression(
 
         model.train(t)
 
-        return sum_logits
-
 
 ######################################################################
 
@@ -154,6 +152,9 @@ class World(Task):
             self.nb_batch_samples_world = input.size(0)
             self.nb_batch_samples_quizzes = 0
 
+        # Shuffle
+        input = input[torch.randperm(input.size(0))]
+
         if desc is None:
             desc = f"epoch-{split}"
         for batch in tqdm.tqdm(
@@ -177,6 +178,7 @@ class World(Task):
                 batch_size=self.batch_size,
                 input=result,
                 ar_mask=ar_mask,
+                summed_logits=None,
                 temperature=1.0,
                 deterministic_synthesis=deterministic_synthesis,
                 progress_bar_desc=None,
@@ -216,6 +218,7 @@ class World(Task):
             batch_size=self.batch_size,
             input=result,
             ar_mask=ar_mask,
+            summed_logits=None,
             temperature=1.0,
             deterministic_synthesis=deterministic_synthesis,
             progress_bar_desc=None,
@@ -261,41 +264,49 @@ class World(Task):
         quizzes = torch.empty(
             nb, self.height * self.width * 2 + 1, device=self.device, dtype=torch.int64
         )
-        ar_mask = torch.full(quizzes.size(), 1, device=self.device)
 
-        sum_logits = masked_inplace_autoregression(
-            model=model,
-            batch_size=self.batch_size,
-            input=quizzes,
-            ar_mask=ar_mask,
-            temperature=1.0,
-            deterministic_synthesis=False,
-            progress_bar_desc="creating quizzes",
-            device=self.device,
-        )
-
-        # Should not be necessary though, the autoregression is done
-        # in eval mode
-        sum_logits = sum_logits.detach()
+        ar_mask = torch.full(quizzes.size(), 1, device=self.device)
+        summed_logits = torch.empty(nb, device=self.device)
 
-        average_logits = sum_logits / quizzes.numel()
+        temperature = 1
+        d_temperature = 1
 
-        # It's a bit brutal to do it twice, we should probably have a
-        # moving average and apply it right away
+        while True:
+            summed_logits[...] = 0
 
-        if desired_average_logits is not None:
-            temperature = average_logits / desired_average_logits
             masked_inplace_autoregression(
                 model=model,
                 batch_size=self.batch_size,
                 input=quizzes,
                 ar_mask=ar_mask,
+                summed_logits=summed_logits,
                 temperature=temperature,
                 deterministic_synthesis=False,
                 progress_bar_desc="creating quizzes",
                 device=self.device,
             )
 
+            average_logits = summed_logits.mean()
+
+            logger(f"{average_logits=} {desired_average_logits=}")
+
+            if desired_average_logits is None:
+                break
+
+            # Oh man that's ugly
+            if average_logits < desired_average_logits * 1.1:
+                if d_temperature > 0:
+                    d_temperature *= -0.5
+                temperature += d_temperature
+            elif average_logits > desired_average_logits:
+                if d_temperature < 0:
+                    d_temperature *= -0.5
+                temperature += d_temperature
+            else:
+                break
+
+            logger(f"chaging temperature to {temperature}")
+
         ###############################################################
         # Create the reverse quizzes
 
@@ -324,6 +335,7 @@ class World(Task):
                 batch_size=self.batch_size,
                 input=result,
                 ar_mask=ar_mask,
+                summed_logits=None,
                 temperature=1.0,
                 deterministic_synthesis=True,
                 progress_bar_desc="solving quizzes",
@@ -339,6 +351,7 @@ class World(Task):
                 batch_size=self.batch_size,
                 input=reverse_result,
                 ar_mask=ar_mask,
+                summed_logits=None,
                 temperature=1.0,
                 deterministic_synthesis=True,
                 progress_bar_desc="solving reversed quizzes",
@@ -353,9 +366,9 @@ class World(Task):
 
         nb_correct = torch.cat(nb_correct, dim=0)
 
-        filename = os.path.join(result_dir, "correct_{n_epoch:04d}.dat")
-        with open(filename, "w") as f:
-            for k in nb_correct:
-                f.write(f"{k}\n")
+        # filename = os.path.join(result_dir, "correct_{n_epoch:04d}.dat")
+        # with open(filename, "w") as f:
+        # for k in nb_correct:
+        # f.write(f"{k}\n")
 
-        return quizzes, nb_correct.sum(dim=0), average_logits
+        return quizzes, nb_correct.sum(dim=0), summed_logits.mean()