Update.

[culture.git] / quizz_machine.py

diff --git a/quizz_machine.py b/quizz_machine.py
index 1a20563..470b095 100755 (executable)
--- a/quizz_machine.py
+++ b/quizz_machine.py
@@ -29,8 +29,6 @@ def one_batch_masked_inplace_autoregression(
     seq_logproba,
     temperature=1.0,
     deterministic_synthesis=False,
-    forbidden_tokens=None,
-    forced_biases=None,
 ):
     to_generate = (ar_mask.sum(0) > 0).nonzero()
 
@@ -45,12 +43,6 @@ def one_batch_masked_inplace_autoregression(
 
         logits = (logits / temperature).log_softmax(dim=-1)
 
-        if forbidden_tokens is not None:
-            logits = logits.masked_fill(forbidden_tokens, float("-inf"))
-
-        if forced_biases is not None:
-            logits = logits + forced_biases[None, :]
-
         if deterministic_synthesis:
             t_next = logits.argmax(-1)
         else:
@@ -104,8 +96,6 @@ def masked_inplace_autoregression(
                 seq_logproba=seq_logproba,
                 temperature=temperature,
                 deterministic_synthesis=deterministic_synthesis,
-                forbidden_tokens=forbidden_tokens,
-                forced_biases=logit_biases,
             )
 
         model.train(t)
@@ -170,7 +160,6 @@ class QuizzMachine:
             )
 
     def save_quizzes(self, result_dir, filename_prefix, quizzes, prediction=False):
-        print(f"DEBUG {quizzes.size()=}")
         l = (quizzes.size(1) - 1) // 2
         forward = (quizzes[:, 0] == self.token_forward).long()
         backward = (quizzes[:, 0] == self.token_backward).long()
@@ -333,12 +322,16 @@ class QuizzMachine:
         )
 
     def compute_correctness(
-        self, c_quizzes, models_for_validation, both_directions=True
+        self, c_quizzes, models_for_validation, both_directions=False
     ):
         reversed_c_quizzes = self.reverse_time(c_quizzes)
 
         ar_mask = self.make_ar_mask(c_quizzes)
-        seq_logproba = torch.empty(ar_mask.size(0), device=self.device)
+        seq_logproba = torch.zeros(
+            c_quizzes.size(0),
+            max([m.id for m in models_for_validation]) + 1,
+            device=self.device,
+        )
 
         # Check how many of models can solve the quizzes in both directions
 
@@ -347,12 +340,14 @@ class QuizzMachine:
         for model in models_for_validation:
             result = c_quizzes.clone()
 
+            seq_logproba[...] = 0.0
+
             masked_inplace_autoregression(
                 model=model,
                 batch_size=self.batch_size,
                 input=result,
                 ar_mask=ar_mask,
-                seq_logproba=seq_logproba,
+                seq_logproba=seq_logproba[:, model.id],
                 temperature=1.0,
                 deterministic_synthesis=True,
                 # progress_bar_desc="solving c_quizzes",
@@ -369,7 +364,7 @@ class QuizzMachine:
                     batch_size=self.batch_size,
                     input=reversed_result,
                     ar_mask=ar_mask,
-                    seq_logproba=seq_logproba,
+                    seq_logproba=seq_logproba[:, model.id],
                     temperature=1.0,
                     deterministic_synthesis=True,
                     # progress_bar_desc="solving reversed c_quizzes",
@@ -386,36 +381,28 @@ class QuizzMachine:
 
             nb_correct += correct
 
-        return nb_correct
+        return nb_correct, seq_logproba
 
     ###############################################################
 
-    def generate_quizzes(self, nb, model_for_generation, reverse_cleanup=False):
+    def generate_quizzes(self, nb, model_for_generation):
         c_quizzes = torch.empty(
             nb, self.train_w_quizzes.size(1), device=self.device, dtype=torch.int64
         )
 
-        c_quizzes[:, 0] = self.token_forward
-
         ar_mask_first = torch.zeros(c_quizzes.size(), device=self.device)
         ar_mask_first[:, : ar_mask_first.size(1) // 2 + 1] = 1
         ar_mask_second = 1 - ar_mask_first
         ar_mask_first[:, 0] = 0
         ar_mask_second[:, 0] = 0
 
-        seq_logproba = torch.empty(ar_mask_first.size(0), device=self.device)
+        seq_logproba = torch.zeros(ar_mask_first.size(0), device=self.device)
 
-        if reverse_cleanup:
-            warnings.warn("very high temperature with reversed cleanup", RuntimeWarning)
-            temperature = 10.0
-        else:
-            temperature = 1.0
+        temperature = 10.0
 
-        # warnings.warn("noise injection", RuntimeWarning)
-        # noise_std = torch.rand(1).item()
-        # self.logger(f"{noise_std=}")
+        # First, we generate the answer at high temperature
 
-        # mygpt.set_noise_injection(model_for_generation, noise_std)
+        c_quizzes[:, 0] = self.token_backward
 
         masked_inplace_autoregression(
             model=model_for_generation,
@@ -428,9 +415,7 @@ class QuizzMachine:
             device=self.device,
         )
 
-        # mygpt.set_noise_injection(model_for_generation, 0.0)
-
-        ave_seq_logproba = seq_logproba.mean()
+        # Then, we generate the prompt deterministically
 
         masked_inplace_autoregression(
             model=model_for_generation,
@@ -438,36 +423,25 @@ class QuizzMachine:
             input=c_quizzes,
             ar_mask=ar_mask_second,
             seq_logproba=seq_logproba,
-            temperature=temperature,
+            temperature=1.0,
             deterministic_synthesis=True,
             device=self.device,
         )
 
-        if reverse_cleanup:
-            c_quizzes = self.reverse_time(c_quizzes)
+        # Then we return the quizz, and re-generate the response, now
+        # deterministically
 
-            masked_inplace_autoregression(
-                model=model_for_generation,
-                batch_size=self.batch_size,
-                input=c_quizzes,
-                ar_mask=ar_mask_second,
-                seq_logproba=seq_logproba,
-                temperature=temperature,
-                deterministic_synthesis=True,
-                device=self.device,
-            )
+        c_quizzes = self.reverse_time(c_quizzes)
 
-            c_quizzes = self.reverse_time(c_quizzes)
-
-            masked_inplace_autoregression(
-                model=model_for_generation,
-                batch_size=self.batch_size,
-                input=c_quizzes,
-                ar_mask=ar_mask_second,
-                seq_logproba=seq_logproba,
-                temperature=temperature,
-                deterministic_synthesis=True,
-                device=self.device,
-            )
+        masked_inplace_autoregression(
+            model=model_for_generation,
+            batch_size=self.batch_size,
+            input=c_quizzes,
+            ar_mask=ar_mask_second,
+            seq_logproba=seq_logproba,
+            temperature=temperature,
+            deterministic_synthesis=True,
+            device=self.device,
+        )
 
-        return c_quizzes, seq_logproba.mean()
+        return c_quizzes