add new evaluation module

app.py

@@ -0,0 +1,6 @@
+import evaluate
+from evaluate.utils import launch_gradio_widget
+
+
+module = evaluate.load("simondoebele/autologiccreateworld")
+launch_gradio_widget(module)

autologiccreateworld.py

@@ -0,0 +1,166 @@
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""This metric module is designed to solve the task of creating a world model and corresponding keys, 
+given a formula and whether that formula needs to be satiesfied. Please read my M.Sc. thesis for details."""
+
+import evaluate
+import datasets
+from Parser import parse_LLM_output
+from nltk.sem.logic import *
+import nltk
+from nltk.sem.logic import LogicParser
+from nltk.sem.evaluate import Valuation, Model
+
+
+# TODO: Add BibTeX citation
+_CITATION = """\
+@InProceedings{huggingface:module,
+title = {Teaching LLMs predicate Logic (M.Sc. Thesis)},
+authors={Simon Döbele},
+year={upcoming}
+}
+"""
+
+# TODO: Add description of the module here
+_DESCRIPTION = """\
+This metric module is designed to solve the task of creating a world model and corresponding keys, 
+given a formula and whether that formula needs to be satiesfied. Please read my M.Sc. thesis for details.
+
+In summary, my compute function behaves like it usually does,
+but in order to be able to compare references and predictions, I need to
+pass the predictions to a parser. Hence, why I rewrote just the compute function.
+"""
+
+
+# TODO: Add description of the arguments of the module here
+_KWARGS_DESCRIPTION = """
+Calculates how good are predictions given some references, using certain scores
+Args:
+    predictions: list of predictions to score. Each predictions
+        should be a string with tokens separated by spaces.
+    references: list of references, one for each prediction. Each
+        reference should be a string with tokens separated by spaces.
+    formulas: list of strings, needed for the parser to decide 
+        whether the overall construction is valid (satisfied or unsatisfied).
+Returns:
+    accuracy: description of the first score,
+    another_score: description of the second score,
+Examples:
+    Examples should be written in doctest format, and should illustrate how
+    to use the function.
+
+    >>> my_new_module = evaluate.load("my_new_module")
+    >>> results = my_new_module.compute(references=[0, 1], predictions=[0, 1])
+    >>> print(results)
+    {'accuracy': 1.0}
+"""
+
+# TODO: Define external resources urls if needed (once thesis is published)
+# BAD_WORDS_URL = "http://url/to/external/resource/bad_words.txt"
+
+
+# def convert_valuation_back(valuation):
+#     # this is necessary, as jsonl could not serialize sets, but nltk expects sets for predicates.
+#     return [(v[0], set(v[1])) if v[0].isupper() else v for v in valuation]
+
+# def eval_task1(dataset):
+    #     results = []
+    #     for generated_output, target, valuation in zip(dataset["Predictions"], dataset["Target-sat"], dataset["Valuation"]):
+    #         try:
+    #             generated_formula = Expression.fromstring(generated_output)
+    #             #print("Parsed output:" + generated_formula)
+    #             valuation = convert_valuation_back(valuation)
+    #             #print("Valuation:" + valuation)
+    #             val = Valuation(valuation)
+    #             dom = val.domain
+    #             m = nltk.sem.evaluate.Model(dom, val)
+    #             g = nltk.sem.Assignment(dom)
+    #             sat = m.evaluate(generated_formula, g)
+    #             if sat == True:
+    #                 prediction = "satisfied"
+    #             elif sat == False:
+    #                 prediction = "unsatisfied"   
+    #         except:
+    #             prediction = "undefined"
+            
+    #         #print("Output:" + prediction + "-----Target:" + target)
+    #         results.append(prediction==target)
+
+    #     accuracy = sum(results)/len(results)
+    #     return accuracy
+
+
+@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
+class AutoLogicCreateWorld(evaluate.Metric):
+    """NOTE: only the compute function changes, compared to a normal evaluateMetric.compute()
+    as well as the datasets.Value (we are working with strings)."""
+
+    def _info(self):
+        # TODO: Specifies the evaluate.EvaluationModuleInfo object
+        return evaluate.MetricInfo(
+            # This is the description that will appear on the modules page.
+            module_type="metric",
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            inputs_description=_KWARGS_DESCRIPTION,
+            # This defines the format of each prediction and reference
+            features=datasets.Features({
+                'predictions': datasets.Value('string'),
+                'references': datasets.Value('string'),
+                'formulas': datasets.Value('string')
+            }),
+            # Homepage of the module for documentation
+            homepage="http://module.homepage", # TODO: change, once thesis is published
+            # Additional links to the codebase or references
+            codebase_urls=["http://github.com/path/to/codebase/of/new_module"], # TODO: change, once thesis is published
+            reference_urls=["http://path.to.reference.url/new_module"] # TODO: change, once thesis is published
+        )
+
+    def _download_and_prepare(self, dl_manager):
+        """Optional: download external resources useful to compute the scores"""
+        # TODO: Download external resources if needed
+        pass
+
+
+    def _compute(self, predictions, references, formulas):
+        """Returns the accuracy, given the parsed output."""
+
+        results = []
+        for generated_output, target, formula in zip(predictions, references, formulas):
+            v = parse_LLM_output(generated_output)
+            #print("Output:" + generated_output)
+            #print("Parsed output:")
+            #print(len(v))
+            val = Valuation(v)
+            dom = val.domain
+            m = nltk.sem.evaluate.Model(dom, val)
+            g = nltk.sem.Assignment(dom)
+            sat = m.evaluate(formula, g)
+            
+            if len(v) == 0:
+                prediction = "undefined"
+            elif sat == True:
+                prediction = "satisfied"
+            elif sat == False:
+                prediction = "unsatisfied"
+            else:
+                prediction = "undefined"
+            
+            #print("Output:" + prediction + "-----Target:" + target)
+            results.append(prediction==target)
+
+        accuracy = sum(results)/len(results)
+        return {
+            "accuracy": accuracy,
+        }

tests.py

@@ -0,0 +1,17 @@
+test_cases = [
+    {
+        "predictions": [0, 0],
+        "references": [1, 1],
+        "result": {"metric_score": 0}
+    },
+    {
+        "predictions": [1, 1],
+        "references": [1, 1],
+        "result": {"metric_score": 1}
+    },
+    {
+        "predictions": [1, 0],
+        "references": [1, 1],
+        "result": {"metric_score": 0.5}
+    }
+]