subset2evaluate 1.0.20__tar.gz → 1.0.22__tar.gz

{subset2evaluate-1.0.20 → subset2evaluate-1.0.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: subset2evaluate
-Version: 1.0.20
+Version: 1.0.22
 Summary: Find informative examples to efficiently (human-)evaluate NLG models.
 Author-email: Vilém Zouhar <vilem.zouhar@gmail.com>
 License: MIT

{subset2evaluate-1.0.20 → subset2evaluate-1.0.22}/pyproject.toml

@@ -11,7 +11,7 @@ packages = ["subset2evaluate"]
 
 [project]
 name = "subset2evaluate"
-version = "1.0.20"
+version = "1.0.22"
 description = "Find informative examples to efficiently (human-)evaluate NLG models."
 license = {text = "MIT"}
 readme = "README.md"

{subset2evaluate-1.0.20 → subset2evaluate-1.0.22}/subset2evaluate/utils.py

@@ -16,6 +16,8 @@ def _data_minmax_normalize(data):
     for line in data:
         for met_all in line["scores"].values():
             for met_k, met_v in met_all.items():
+                if met_v is None:
+                    continue
                 data_flat[met_k].append(met_v)
 
     # normalize
@@ -24,10 +26,15 @@ def _data_minmax_normalize(data):
     for line in data:
         for model, met_all in line["scores"].items():
             for met_k, met_v in met_all.items():
-                # (x-min)/(max-min) normalize
-                line["scores"][model][met_k] = (met_v - data_flat[met_k][0]) / (
-                    data_flat[met_k][1] - data_flat[met_k][0]
-                )
+                if met_v is None:
+                    continue
+                if data_flat[met_k][1] - data_flat[met_k][0] == 0:
+                    line["scores"][model][met_k] = 0
+                else:
+                    # (x-min)/(max-min) normalize
+                    line["scores"][model][met_k] = (met_v - data_flat[met_k][0]) / (
+                        data_flat[met_k][1] - data_flat[met_k][0]
+                    )
 
 
 def confidence_interval(data, confidence=0.95):
@@ -844,18 +851,21 @@ def load_data_wmt(  # noqa: C901
         humscores = [
             model_v["human"] for line in data for model_v in line["scores"].values()
         ]
-        if all(x <= 0 for x in humscores):
+        if all(x <= 0 for x in humscores if x is not None):
             for line in data:
                 for model_v in line["scores"].values():
-                    model_v["human"] = max(0, min(100, 100 + 4 * model_v["human"]))
-        elif all(x >= 0 and x <= 1 for x in humscores):
+                    if model_v["human"] is not None:
+                        model_v["human"] = max(0, min(100, 100 + 4 * model_v["human"]))
+        elif all(x >= 0 and x <= 1 for x in humscores if x is not None):
             for line in data:
                 for model_v in line["scores"].values():
-                    model_v["human"] = max(0, min(100, model_v["human"] * 100))
+                    if model_v["human"] is not None:
+                        model_v["human"] = max(0, min(100, model_v["human"] * 100))
         else:
             for line in data:
                 for model_v in line["scores"].values():
-                    model_v["human"] = max(0, min(100, model_v["human"]))
+                    if model_v["human"] is not None:
+                        model_v["human"] = max(0, min(100, model_v["human"]))
 
         # this is min-max normalization
         if normalize and not binarize:

{subset2evaluate-1.0.20 → subset2evaluate-1.0.22}/subset2evaluate.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: subset2evaluate
-Version: 1.0.20
+Version: 1.0.22
 Summary: Find informative examples to efficiently (human-)evaluate NLG models.
 Author-email: Vilém Zouhar <vilem.zouhar@gmail.com>
 License: MIT