Merge branch 'main' into humaneval

# Conflicts:
#	lm_eval/api/task.py

lm_eval/tasks/afrimgsm/run.sh

+lm_eval --model hf   \
+        --model_args pretrained="google/gemma-7b"  --tasks afrimgsm_en_cot_eng,mgsm_en_cot_en,afrimgsm_native_cot_eng,mgsm_native_cot_en,afrimgsm_direct_eng,mgsm_direct_en,afrimgsm_direct_native_eng  \
+        --device cuda:0     \
+        --batch_size 1  \
+        --verbosity DEBUG \
+        --limit 5

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_amh.yaml

+# Generated by utils.py
+dataset_name: amh
+doc_to_target: '{% if answer is not none %}{{answer[15:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_amh

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_eng.yaml

+# Generated by utils.py
+dataset_name: eng
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_eng

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_ewe.yaml

+# Generated by utils.py
+dataset_name: ewe
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_ewe

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_fra.yaml

+# Generated by utils.py
+dataset_name: fra
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_fra

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_hau.yaml

+# Generated by utils.py
+dataset_name: hau
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_hau

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_ibo.yaml

+# Generated by utils.py
+dataset_name: ibo
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_ibo

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_kin.yaml

+# Generated by utils.py
+dataset_name: kin
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_kin

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_lin.yaml

+# Generated by utils.py
+dataset_name: lin
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_lin

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_lug.yaml

+# Generated by utils.py
+dataset_name: lug
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_lug

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_orm.yaml

+# Generated by utils.py
+dataset_name: orm
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_orm

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_sna.yaml

+# Generated by utils.py
+dataset_name: sna
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_sna

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_sot.yaml

+# Generated by utils.py
+dataset_name: sot
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_sot

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_swa.yaml

+# Generated by utils.py
+dataset_name: swa
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_swa

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_twi.yaml

+# Generated by utils.py
+dataset_name: twi
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_twi

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_wol.yaml

+# Generated by utils.py
+dataset_name: wol
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_wol

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_xho.yaml

+# Generated by utils.py
+dataset_name: xho
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_xho

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_yor.yaml

+# Generated by utils.py
+dataset_name: yor
+doc_to_target: '{% if answer is not none %}{{answer[16:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_yor

lm_eval/tasks/afrimgsm/translate/afrimgsm_translate_zul.yaml

+# Generated by utils.py
+dataset_name: zul
+doc_to_target: '{% if answer is not none %}{{answer[21:]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
+generation_kwargs:
+  do_sample: false
+  until:
+  - 'Question:'
+  - </s>
+  - <|im_end|>
+include: translate_direct_yaml
+task: afrimgsm_translate_direct_zul

lm_eval/tasks/afrimgsm/translate/translate_direct_yaml

+# This file will be included in the generated language-specific task configs.
+# It doesn't have a yaml file extension as it is not meant to be imported directly
+# by the harness.
+group:
+    - afrimgsm
+    - afrimgsm_translate
+dataset_path: masakhane/afrimgsm-translate-test
+dataset_name: null  # Overridden by language-specific config.
+output_type: generate_until
+test_split: test
+generation_kwargs:
+  until:
+    - "\n\n"
+    - "\n"
+  do_sample: false
+  temperature: 0.0
+target_delimiter: " "
+filter_list:
+  - name: remove_whitespace
+    filter:
+      - function: remove_whitespace
+      - function: take_first
+  - filter:
+    - function: regex
+      group_select: -1
+      regex_pattern: (-?[$0-9.,]{2,})|(-?[0-9]+)
+    - function: take_first
+    name: flexible-extract
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+    ignore_case: true
+    ignore_punctuation: true
+metadata:
+  version: 2.0