fix: changing the dtype in random_utils to uint32

algorithmic_efficiency/random_utils.py

@@ -18,30 +18,30 @@
 
 # Annoyingly, RandomState(seed) requires seed to be in [0, 2 ** 32 - 1] (an
 # unsigned int), while RandomState.randint only accepts and returns signed ints.
-MAX_INT32 = 2**31
-MIN_INT32 = -MAX_INT32
+MAX_UINT32 = 2**32 - 1
+MIN_UINT32 = 0
 
 SeedType = Union[int, list, np.ndarray]
 
 
 def _signed_to_unsigned(seed: SeedType) -> SeedType:
   if isinstance(seed, int):
-    return seed % 2**32
+    return seed % MAX_UINT32
   if isinstance(seed, list):
-    return [s % 2**32 for s in seed]
+    return [s % MAX_UINT32 for s in seed]
   if isinstance(seed, np.ndarray):
-    return np.array([s % 2**32 for s in seed.tolist()])
+    return np.array([s % MAX_UINT32 for s in seed.tolist()])
 
 
 def _fold_in(seed: SeedType, data: Any) -> List[Union[SeedType, Any]]:
   rng = np.random.RandomState(seed=_signed_to_unsigned(seed))
-  new_seed = rng.randint(MIN_INT32, MAX_INT32, dtype=np.int32)
+  new_seed = rng.randint(MIN_UINT32, MAX_UINT32, dtype=np.uint32)
   return [new_seed, data]
 
 
 def _split(seed: SeedType, num: int = 2) -> SeedType:
   rng = np.random.RandomState(seed=_signed_to_unsigned(seed))
-  return rng.randint(MIN_INT32, MAX_INT32, dtype=np.int32, size=[num, 2])
+  return rng.randint(MIN_UINT32, MAX_UINT32, dtype=np.uint32, size=[num, 2])
 
 
 def _PRNGKey(seed: SeedType) -> SeedType:  # pylint: disable=invalid-name

scoring/performance_profile.py

@@ -274,7 +274,8 @@ def compute_performance_profiles(submissions,
                                  scale='linear',
                                  verbosity=0,
                                  strict=False,
-                                 self_tuning_ruleset=False):
+                                 self_tuning_ruleset=False,
+                                 output_dir=None):
   """Compute performance profiles for a set of submission by some time column.
 
   Args:
@@ -321,6 +322,8 @@ def compute_performance_profiles(submissions,
   # Sort workloads alphabetically (for better display)
   df = df.reindex(sorted(df.columns), axis=1)
 
+  # Save time to target dataframe
+  df.to_csv(os.path.join(output_dir, 'time_to_targets.csv'))
   # For each held-out workload set to inf if the base workload is inf or nan
   for workload in df.keys():
     if workload not in BASE_WORKLOADS:

scoring/score_submissions.py

@@ -210,7 +210,9 @@ def main(_):
         scale='linear',
         verbosity=0,
         self_tuning_ruleset=FLAGS.self_tuning_ruleset,
-        strict=FLAGS.strict)
+        strict=FLAGS.strict,
+        output_dir=FLAGS.output_dir,
+    )
     if not os.path.exists(FLAGS.output_dir):
       os.mkdir(FLAGS.output_dir)
     performance_profile.plot_performance_profiles(

setup.cfg

@@ -121,7 +121,6 @@ jax_core_deps =
   chex==0.1.7
   ml_dtypes==0.2.0
   protobuf==4.25.3
-  scipy==1.11.4
 
 
 # JAX CPU