add Darija (Moroccan dialects) tasks including darijammlu. darijahellaswag and darija_bench (#2521)

* add Darija tasks

* fix multiple groups issue in darijammlu

* add MT to the description of the Darija tasks

* Update README.md

nit

* fix the recursion error caused by the darija_summarization task

* use a custom filter instead of the decorator for the strip function

---------
Co-authored-by: Baber Abbasi <92168766+baberabb@users.noreply.github.com>

lm_eval/tasks/darijammlu/darijammlu_political_science.yaml

+"dataset_name": "political_science"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_ar_mmlu_tasks"
+"task": "darijammlu_political_science"
+"task_alias": "political science"

lm_eval/tasks/darijammlu/darijammlu_professional_law.yaml

+"dataset_name": "professional_law"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_professional_law"
+"task_alias": "professional law"

lm_eval/tasks/darijammlu/darijammlu_professional_psychology.yaml

+"dataset_name": "professional_psychology"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_professional_psychology"
+"task_alias": "professional psychology"

lm_eval/tasks/darijammlu/darijammlu_public_relations.yaml

+"dataset_name": "public_relations"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_public_relations"
+"task_alias": "public relations"

lm_eval/tasks/darijammlu/darijammlu_security_studies.yaml

+"dataset_name": "security_studies"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_security_studies"
+"task_alias": "security studies"

lm_eval/tasks/darijammlu/darijammlu_social_science.yaml

+"dataset_name": "social_science"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_ar_mmlu_tasks"
+"task": "darijammlu_social_science"
+"task_alias": "social science"

lm_eval/tasks/darijammlu/darijammlu_sociology.yaml

+"dataset_name": "sociology"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_sociology"
+"task_alias": "sociology"

lm_eval/tasks/darijammlu/darijammlu_world_religions.yaml

+"dataset_name": "world_religions"
+"include": "_default_darijammlu_template_yaml"
+"tag":
+- "darijammlu_mmlu_tasks"
+"task": "darijammlu_world_religions"
+"task_alias": "world religions"

lm_eval/tasks/darijammlu/utils.py

+PROMPT = "هادا سؤال متعدد الخيارات (مع الجواب ديالو) على {}\n\n{}\n{}\nالجواب:"
+
+
+alpha = ["A.", "B.", "C.", "D.", "E."]
+
+
+def doc_to_text(doc):
+
+    subject = doc["subject_darija"]
+    question = (
+        doc["question"]
+        if doc["context"] == ""
+        else f"{doc['context']}\n\n{doc['question']}"
+    )
+
+    options = []
+    for i, opt in enumerate(doc["choices"]):
+        options.append(f"{alpha[i]} {opt}")
+
+    doc_text = PROMPT.format(subject, question, "\n".join(options))
+
+    return doc_text
+
+
+def doc_to_choice(doc):
+    return [alpha[i][0] for i in range(len(doc['choices']))]