Merge branch 'main' into longcxt

lm_eval/tasks/mlqa/mlqa_ar_en.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_ar_en
+dataset_name: mlqa.ar.en
+process_results: !function utils.process_results_ar

lm_eval/tasks/mlqa/mlqa_ar_es.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_ar_es
+dataset_name: mlqa.ar.es
+process_results: !function utils.process_results_ar

lm_eval/tasks/mlqa/mlqa_ar_hi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_ar_hi
+dataset_name: mlqa.ar.hi
+process_results: !function utils.process_results_ar

lm_eval/tasks/mlqa/mlqa_ar_vi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_ar_vi
+dataset_name: mlqa.ar.vi
+process_results: !function utils.process_results_ar

lm_eval/tasks/mlqa/mlqa_ar_zh.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_ar_zh
+dataset_name: mlqa.ar.zh
+process_results: !function utils.process_results_ar

lm_eval/tasks/mlqa/mlqa_common_yaml

+dataset_path: facebook/mlqa
+dataset_kwargs:
+  trust_remote_code: true
+test_split: test
+validation_split: validation
+output_type: generate_until
+doc_to_text: "Context: {{context}}\n\nQuestion: {{question}}\n\nAnswer:"
+doc_to_target: "{{answers}}"
+process_docs: !function utils.process_docs
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+  - metric: f1
+    aggregation: mean
+    higher_is_better: true
+generation_kwargs:
+  until:
+    - "\n"
+  do_sample: false
+metadata:
+  version: 0.0

lm_eval/tasks/mlqa/mlqa_de_ar.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_ar
+dataset_name: mlqa.de.ar
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_de.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_de
+dataset_name: mlqa.de.de
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_en.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_en
+dataset_name: mlqa.de.en
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_es.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_es
+dataset_name: mlqa.de.es
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_hi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_hi
+dataset_name: mlqa.de.hi
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_vi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_vi
+dataset_name: mlqa.de.vi
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_de_zh.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_de_zh
+dataset_name: mlqa.de.zh
+process_results: !function utils.process_results_de

lm_eval/tasks/mlqa/mlqa_en_ar.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_ar
+dataset_name: mlqa.en.ar
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_de.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_de
+dataset_name: mlqa.en.de
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_en.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_en
+dataset_name: mlqa.en.en
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_es.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_es
+dataset_name: mlqa.en.es
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_hi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_hi
+dataset_name: mlqa.en.hi
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_vi.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_vi
+dataset_name: mlqa.en.vi
+process_results: !function utils.process_results_en

lm_eval/tasks/mlqa/mlqa_en_zh.yaml

+# Generated by generate_tasks.py
+include: mlqa_common_yaml
+task: mlqa_en_zh
+dataset_name: mlqa.en.zh
+process_results: !function utils.process_results_en