Adding MMLU and Winogrande human-translated into 11 African languages (…

…#3237)

src/helm/benchmark/run_specs/mmlu_clinical_afr_run_specs.py

@@ -0,0 +1,65 @@
+"""Run spec functions for three clinical sections of MMLU human-translated into 11 African languages
+
+Available subjects: "clinical_knowledge", "college_medicine", "virology"
+Available langs: "af", "zu", "xh", "am", "bm", "ig", "nso", "sn", "st", "tn", "ts" (see lang_map below for language code mapping to language name, or here for ISO code reference: https://huggingface.co/languages)
+"""
+
+from helm.benchmark.adaptation.adapter_spec import (
+    ADAPT_GENERATION,
+    ADAPT_MULTIPLE_CHOICE_JOINT,
+    AdapterSpec,
+)
+from helm.benchmark.adaptation.common_adapter_specs import (
+    get_generation_adapter_spec,
+    get_machine_translation_adapter_spec,
+    get_multiple_choice_adapter_spec,
+)
+from helm.benchmark.metrics.common_metric_specs import (
+    get_basic_generation_metric_specs,
+    get_basic_metric_specs,
+    get_exact_match_metric_specs,
+    get_f1_metric_specs,
+    get_generative_harms_metric_specs,
+    get_generic_metric_specs,
+    get_open_ended_generation_metric_specs,
+)
+from helm.benchmark.run_spec import RunSpec, run_spec_function
+from helm.benchmark.runner import get_benchmark_output_path
+from helm.benchmark.scenarios.scenario import ScenarioSpec, get_scenario_cache_path
+
+
+@run_spec_function("mmlu_clinical_afr")
+def get_mmlu_clinical_afr_spec(subject: str, lang: str, method: str = ADAPT_MULTIPLE_CHOICE_JOINT) -> RunSpec:
+    scenario_spec = ScenarioSpec(
+        class_name="helm.benchmark.scenarios.mmlu_clinical_afr_scenario.MMLU_Clinical_Afr_Scenario", args={"subject": subject, "lang": lang}
+    )
+
+    lang_map = {
+        'af': 'Afrikaans',
+        'zu': 'Zulu',
+        'xh': 'Xhosa',
+        'am': 'Amharic',
+        'bm': 'Bambara',
+        'ig': 'Igbo',
+        'nso': 'Sepedi',
+        'sn': 'Shona',
+        'st': 'Sesotho',
+        'tn': 'Setswana',
+        'ts': 'Tsonga',
+    }
+
+    adapter_spec = get_multiple_choice_adapter_spec(
+        method=method,
+        instructions=f"The following are multiple choice questions (with answers) about {subject.replace('_', ' ')} "
+                     f"in {lang_map[lang]}.",
+        input_noun="Question",
+        output_noun="Answer",
+    )
+
+    return RunSpec(
+        name=f"mmlu_clinical_afr:subject={subject},lang={lang},method={method}",
+        scenario_spec=scenario_spec,
+        adapter_spec=adapter_spec,
+        metric_specs=get_exact_match_metric_specs(),
+        groups=["low_resource_languages"],
+    )

src/helm/benchmark/run_specs/winogrande_afr_run_specs.py

@@ -0,0 +1,64 @@
+"""Run spec functions for Winogrande human-translated into 11 African languages
+
+Available langs: "af", "zu", "xh", "am", "bm", "ig", "nso", "sn", "st", "tn", "ts" (see lang_map below for language code mapping to language name, or here for ISO code reference: https://huggingface.co/languages)
+"""
+
+from helm.benchmark.adaptation.adapter_spec import (
+    ADAPT_GENERATION,
+    ADAPT_MULTIPLE_CHOICE_JOINT,
+    AdapterSpec,
+)
+from helm.benchmark.adaptation.common_adapter_specs import (
+    get_generation_adapter_spec,
+    get_machine_translation_adapter_spec,
+    get_multiple_choice_adapter_spec,
+)
+from helm.benchmark.metrics.common_metric_specs import (
+    get_basic_generation_metric_specs,
+    get_basic_metric_specs,
+    get_exact_match_metric_specs,
+    get_f1_metric_specs,
+    get_generative_harms_metric_specs,
+    get_generic_metric_specs,
+    get_open_ended_generation_metric_specs,
+)
+from helm.benchmark.run_spec import RunSpec, run_spec_function
+from helm.benchmark.runner import get_benchmark_output_path
+from helm.benchmark.scenarios.scenario import ScenarioSpec, get_scenario_cache_path
+
+
+@run_spec_function("winogrande_afr")
+def get_winogrande_afr_spec(lang: str, method: str = ADAPT_MULTIPLE_CHOICE_JOINT) -> RunSpec:
+    scenario_spec = ScenarioSpec(
+        class_name="helm.benchmark.scenarios.winogrande_afr_scenario.Winogrande_Afr_Scenario", args={"lang": lang}
+    )
+
+    lang_map = {
+        'af': 'Afrikaans',
+        'zu': 'Zulu',
+        'xh': 'Xhosa',
+        'am': 'Amharic',
+        'bm': 'Bambara',
+        'ig': 'Igbo',
+        'nso': 'Sepedi',
+        'sn': 'Shona',
+        'st': 'Sesotho',
+        'tn': 'Setswana',
+        'ts': 'Tsonga',
+    }
+
+    adapter_spec = get_multiple_choice_adapter_spec(
+        method=method,
+        instructions=f"The following are binary choice fill-in-the-blank sentences (with answers), requiring common sense reasoning "
+                     f"in {lang_map[lang]}.",
+        input_noun="Question",
+        output_noun="Answer",
+    )
+
+    return RunSpec(
+        name=f"winogrande_afr:lang={lang},method={method}",
+        scenario_spec=scenario_spec,
+        adapter_spec=adapter_spec,
+        metric_specs=get_exact_match_metric_specs(),
+        groups=["low_resource_languages"],
+    )

src/helm/benchmark/scenarios/mmlu_clinical_afr_scenario.py

@@ -0,0 +1,74 @@
+import csv
+import os
+from typing import Dict, List
+
+from helm.common.general import ensure_file_downloaded
+from helm.common.hierarchical_logger import hlog
+from .scenario import Scenario, Instance, Reference, TRAIN_SPLIT, VALID_SPLIT, TEST_SPLIT, CORRECT_TAG, Input, Output
+
+
+class MMLU_Clinical_Afr_Scenario(Scenario):
+    """
+    https://github.com/InstituteforDiseaseModeling/Bridging-the-Gap-Low-Resource-African-Languages
+    """
+
+    name = "mmlu_clinical_afr"
+    description = "Massive Multitask Language Understanding (MMLU) translated into 11 African low-resource languages"
+    tags = ["knowledge", "multiple_choice", "low_resource_languages"]
+
+    def __init__(self, subject: str = "clinical_knowledge", lang: str = "af"):
+        super().__init__()
+        self.subject: str = subject
+        self.lang: str = lang
+
+    def download_mmlu_clinical_afr(self, path: str):
+        ensure_file_downloaded(
+            source_url="https://github.com/InstituteforDiseaseModeling/Bridging-the-Gap-Low-Resource-African-Languages/raw/refs/heads/main/data/evaluation_benchmarks_afr_release.zip",
+            target_path=path,
+            unpack=True,
+            unpack_type='unzip'
+        )
+
+    def process_csv(self, csv_path: str, split: str) -> List[Instance]:
+        instances: List[Instance] = []
+        hlog(f"Reading {csv_path}")
+        with open(csv_path) as f:
+            reader = csv.reader(f, delimiter=",")
+            for row in reader:
+
+                question, answers, correct_choice = row[0], row[1:-1], row[-1]
+                answers_dict = dict(zip(["A", "B", "C", "D"], answers))
+                correct_answer: str = answers_dict[correct_choice]
+
+                def answer_to_reference(answer: str) -> Reference:
+                    return Reference(Output(text=answer), tags=[CORRECT_TAG] if answer == correct_answer else [])
+
+                instance = Instance(
+                    input=Input(text=question),
+                    references=list(map(answer_to_reference, answers)),
+                    split=split,
+                )
+                instances.append(instance)
+        return instances
+
+    def get_instances(self, output_path: str) -> List[Instance]:
+        # Download the raw data
+        desired_dir = 'mmlu_cm_ck_vir'
+        data_path: str = os.path.join(output_path, desired_dir)
+        self.download_mmlu_clinical_afr(data_path)
+
+        # Read all the instances
+        instances: List[Instance] = []
+        splits: Dict[str, str] = {
+            "dev": TRAIN_SPLIT,
+            "val": VALID_SPLIT,
+            "test": TEST_SPLIT,
+        }
+        for split in splits:
+            csv_path: str = os.path.join(data_path, desired_dir, f"{self.subject}_{split}_{self.lang}.csv")
+            if not os.path.exists(csv_path):
+                hlog(f"{csv_path} doesn't exist, skipping")
+                continue
+            instances.extend(self.process_csv(csv_path, splits[split]))
+
+        return instances

src/helm/benchmark/scenarios/test_mmlu_clinical_afr_scenario.py

@@ -0,0 +1,21 @@
+import pytest
+from tempfile import TemporaryDirectory
+
+from helm.benchmark.scenarios.mmlu_clinical_afr_scenario import MMLU_Clinical_Afr_Scenario
+from helm.benchmark.scenarios.scenario import CORRECT_TAG, Input, Output, Reference
+
+
+@pytest.mark.scenarios
+def test_mmlu_clinical_afr_scenario():
+    with TemporaryDirectory() as tmpdir:
+        scenario = MMLU_Clinical_Afr_Scenario(subject="clinical_knowledge", lang="am")
+        instances = scenario.get_instances(tmpdir)
+        assert len(instances) == 299
+        assert instances[0].input == Input(text="ለሁሉም የጡንቻ መኮማተር ዓይነቶች የሚያስፈልገው ኢኔርጅ የሚቀርበው ከሚከተሉት ነው፦")
+        assert instances[0].references == [
+            Reference(output=Output(text="ATP።"), tags=[CORRECT_TAG]),
+            Reference(output=Output(text="ADP።"), tags=[]),
+            Reference(output=Output(text="ፎስፎክሬቲን።"), tags=[]),
+            Reference(output=Output(text="ኦክስዳቲቪ ፎስፎሪሌሽን።"), tags=[]),
+        ]
+        assert instances[0].split == "train"

src/helm/benchmark/scenarios/test_winogrande_afr_scenario.py

@@ -0,0 +1,19 @@
+import pytest
+from tempfile import TemporaryDirectory
+
+from helm.benchmark.scenarios.winogrande_afr_scenario import Winogrande_Afr_Scenario
+from helm.benchmark.scenarios.scenario import CORRECT_TAG, Input, Output, Reference
+
+
+@pytest.mark.scenarios
+def test_winogrande_afr_scenario():
+    with TemporaryDirectory() as tmpdir:
+        scenario = Winogrande_Afr_Scenario(lang="am")
+        instances = scenario.get_instances(tmpdir)
+        assert len(instances) == 3674
+        assert instances[0].input == Input(text="ሳራ ከማሪያ በጣም የተሻለች የቀዶ ጥገና ሐኪም ስለነበረች ሁልጊዜ _ ቀላል ህመሞችን ታክማለች.")
+        assert instances[0].references == [
+            Reference(output=Output(text="ሳራ"), tags=[]),
+            Reference(output=Output(text="ማሪያ"), tags=[CORRECT_TAG]),
+        ]
+        assert instances[0].split == "train"

src/helm/benchmark/scenarios/winogrande_afr_scenario.py

@@ -0,0 +1,78 @@
+import csv
+import os
+from typing import Dict, List
+
+from helm.common.general import ensure_file_downloaded
+from helm.common.hierarchical_logger import hlog
+from .scenario import Scenario, Instance, Reference, TRAIN_SPLIT, VALID_SPLIT, TEST_SPLIT, CORRECT_TAG, Input, Output
+
+
+class Winogrande_Afr_Scenario(Scenario):
+    """
+    https://github.com/InstituteforDiseaseModeling/Bridging-the-Gap-Low-Resource-African-Languages
+    """
+
+    name = "winogrande_afr"
+    description = "Winogrande (S) translated into 11 African low-resource languages"
+    tags = ["knowledge", "multiple_choice", "low_resource_languages"]
+
+    def __init__(self, lang: str = "af"):
+        super().__init__()
+        self.lang: str = lang
+
+    def download_winogrande_afr(self, path: str):
+        ensure_file_downloaded(
+            source_url="https://github.com/InstituteforDiseaseModeling/Bridging-the-Gap-Low-Resource-African-Languages/raw/refs/heads/main/data/evaluation_benchmarks_afr_release.zip",
+            target_path=path,
+            unpack=True,
+            unpack_type='unzip'
+        )
+
+    def process_csv(self, csv_path: str, split: str, pseudo_split: str) -> List[Instance]:
+        # Match naming in Winogrande
+        if pseudo_split == 'val':
+            pseudo_split = 'train_s'
+        instances: List[Instance] = []
+        hlog(f"Reading {csv_path}")
+        with open(csv_path) as f:
+            reader = csv.reader(f, delimiter=",")
+            next(reader, None)  # skip the header
+            for row in reader:
+                if row[-1] != pseudo_split:  # ensure correct split is taken
+                    continue
+                question, answers, correct_choice = row[-5], row[-4:-2], row[-2]
+                answers_dict = dict(zip(["1", "2"], answers))
+                correct_answer: str = answers_dict[correct_choice]
+
+                def answer_to_reference(answer: str) -> Reference:
+                    return Reference(Output(text=answer), tags=[CORRECT_TAG] if answer == correct_answer else [])
+
+                instance = Instance(
+                    input=Input(text=question),
+                    references=list(map(answer_to_reference, answers)),
+                    split=split,
+                )
+                instances.append(instance)
+        return instances
+
+    def get_instances(self, output_path: str) -> List[Instance]:
+        # Download the raw data
+        desired_dir = 'winogrande_s'
+        data_path: str = os.path.join(output_path, desired_dir)
+        self.download_winogrande_afr(data_path)
+
+        # Read all the instances
+        instances: List[Instance] = []
+        splits: Dict[str, str] = {
+            "dev": TRAIN_SPLIT,
+            "val": VALID_SPLIT,
+            "test": TEST_SPLIT,
+        }
+        for split in splits:
+            csv_path: str = os.path.join(data_path, desired_dir, f"winogrande_{self.lang}.csv")
+            if not os.path.exists(csv_path):
+                hlog(f"{csv_path} doesn't exist, skipping")
+                continue
+            instances.extend(self.process_csv(csv_path, splits[split], split))
+
+        return instances