Hopefully more compact keys for known seeds

JasonGross · JasonGross · commit df927056c3c6 · 2024-11-06T13:17:35.000-08:00
diff --git a/notebooks_jason/max_of_K_all_models.py b/notebooks_jason/max_of_K_all_models.py
@@ -3266,7 +3266,7 @@ def do_linear_regression(X, Y):
         ]
         print(f"len after: {len(df)}")
 
-    sorted_known_seeds = tuple(sorted(df["seed"].values))
+    sorted_known_seeds = np.unique(np.array(df["seed"].values), return_counts=True)
     # Group by 'attention_error_handling' and calculate the max 'normalized-accuracy-bound' for sorting groups
     df = df[
         [

Original file line number	Diff line number	Diff line change
`@@ -3266,7 +3266,7 @@ def do_linear_regression(X, Y):`
`3266`	`3266`	`]`
`3267`	`3267`	`print(f"len after: {len(df)}")`
`3268`	`3268`
`3269`		`- sorted_known_seeds = tuple(sorted(df["seed"].values))`
	`3269`	`+ sorted_known_seeds = np.unique(np.array(df["seed"].values), return_counts=True)`
`3270`	`3270`	`# Group by 'attention_error_handling' and calculate the max 'normalized-accuracy-bound' for sorting groups`
`3271`	`3271`	`df = df[`
`3272`	`3272`	`[`