Merge branch 'main' into mathvista

# Conflicts:
#	lm_eval/models/hf_vlms.py

lm_eval/tasks/basque_bench/mgsm_direct_eu.yaml

+task: mgsm_direct_eu
+dataset_path: HiTZ/MGSM-eu
+dataset_name: null
+doc_to_target: '{{answer_number|string}}'
+doc_to_text: '{% if answer is not none %}{{question+"\nErantzuna:"}}{% else %}{{"Galdera: "+question+"\nErantzuna:"}}{% endif %}'
+output_type: generate_until
+training_split: train
+test_split: test
+target_delimiter: " "
+generation_kwargs:
+  until:
+    - "\n\n"
+    - "\n"
+    - "Galdera:"
+    - </s>
+    - <|im_end|>
+  do_sample: false
+  temperature: 0.0
+filter_list:
+  - name: remove_whitespace
+    filter:
+      - function: remove_whitespace
+      - function: take_first
+  - name: flexible-extract
+    filter:
+    - function: regex
+      group_select: -1
+      regex_pattern: (-?[0-9]+([ .,][0-9.,]+)?)
+    - function: take_first
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+    ignore_case: true
+    ignore_punctuation: true
+    regexes_to_ignore:
+      - " "
+metadata:
+  version: 1.0

lm_eval/tasks/basque_bench/utils.py

+from functools import partial
+
+
+# ~~~~~~~~~~~ XCOPA ~~~~~~~~~~~ #
+
+xcopa_connectors = {"cause": " Izan ere,", "effect": " Beraz,"}
+
+
+def xcopa_doc_to_text(doc):
+    conn = xcopa_connectors[doc["question"]]
+    return doc["premise"].strip() + f"{conn}"
+
+
+def xcopa_doc_to_choice(doc):
+    def convert_choice(choice):
+        return choice[0].lower() + choice[1:]
+
+    return [convert_choice(doc["choice1"]), convert_choice(doc["choice2"])]
+
+
+# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ #

lm_eval/tasks/basque_bench/wnli_eu.yaml

+task: wnli_eu
+dataset_path: HiTZ/wnli-eu
+dataset_name: null
+output_type: multiple_choice
+training_split: null
+validation_split: validation
+test_split: null
+doc_to_text: "{{sentence1}}\nGaldera: {{sentence2}} Egia edo Gezurra?\nErantzuna:"
+doc_to_target: label
+doc_to_choice: ["Gezurra", "Egia"]
+metric_list:
+  - metric: acc
+metadata:
+  version: 1.0

lm_eval/tasks/basque_bench/xcopa_eu.yaml

+task: xcopa_eu
+dataset_path: HiTZ/XCOPA-eu
+dataset_name: null
+output_type: multiple_choice
+training_split: null
+validation_split: validation
+test_split: test
+doc_to_text: !function utils.xcopa_doc_to_text
+doc_to_target: label
+doc_to_choice: !function utils.xcopa_doc_to_choice
+metric_list:
+  - metric: acc
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/README.md

+# CatalanBench
+
+### Paper
+
+CatalanBench is a benchmark for evaluating language models in Catalan tasks. This is, it evaluates the ability of a language model to understand and generate Catalan text. CatalanBench offers a combination of pre-existing, open datasets and datasets developed exclusivelly for this benchmark. All the details of CatalanBench will be published in a paper soon.
+
+The new evaluation datasets included in CatalanBench are:
+| Task          | Category       | Homepage  |
+|:-------------:|:-----:|:-----:|
+| ARC_ca | Question Answering | https://huggingface.co/datasets/projecte-aina/arc_ca |
+| MGSM_ca | Math | https://huggingface.co/datasets/projecte-aina/mgsm_ca |
+| OpenBookQA_ca | Question Answering | https://huggingface.co/datasets/projecte-aina/openbookqa_ca |
+| Parafraseja | Paraphrasing | https://huggingface.co/datasets/projecte-aina/Parafraseja |
+| PIQA_ca | Question Answering | https://huggingface.co/datasets/projecte-aina/piqa_ca |
+| SIQA_ca | Question Answering | https://huggingface.co/datasets/projecte-aina/siqa_ca |
+| XStoryCloze_ca | Commonsense Reasoning | https://huggingface.co/datasets/projecte-aina/xstorycloze_ca |
+
+The datasets included in CatalanBench that have been made public in previous pubications are:
+
+| Task          | Category       | Paper title          | Homepage  |
+|:-------------:|:-----:|:-------------:|:-----:|
+| Belebele_ca | Reading Comprehension | [The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants](https://arxiv.org/abs/2308.16884) | https://huggingface.co/datasets/facebook/belebele |
+| caBREU | Summarization | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/caBreu |
+| CatalanQA | Question Answering | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/catalanqa |
+| CatCoLA | Linguistic Acceptability | CatCoLA: Catalan Corpus of Linguistic Acceptability | https://huggingface.co/datasets/nbel/CatCoLA |
+| COPA-ca | Commonsense Reasoning | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/COPA-ca |
+| CoQCat | Question Answering | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/CoQCat |
+| FLORES_ca | Translation | [The FLORES-101  Evaluation Benchmark for Low-Resource and Multilingual Machine Translation](https://arxiv.org/abs/2106.03193) | https://huggingface.co/datasets/facebook/flores |
+| PAWS-ca | Paraphrasing | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/PAWS-ca |
+| TE-ca | Natural Language Inference | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/teca |
+| VeritasQA_ca | Truthfulness | VeritasQA: A Truthfulness Benchmark Aimed at Multilingual Transferability | TBA |
+| WNLI-ca | Natural Language Inference | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/wnli-ca |
+| XNLI-ca | Natural Language Inference | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/xnli-ca |
+| XQuAD-ca | Question Answering | [Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan](https://aclanthology.org/2024.lrec-main.231/) | https://huggingface.co/datasets/projecte-aina/xquad-ca |
+
+
+### Citation
+Paper for CatalanBench coming soon.
+
+<!--```bibtex
+@inproceedings{baucells-2024-iberobench,
+    title = "IberoBench: A Benchmark for LLM Evaluation in Iberian Languages",
+    author = "Baucells, Irene  and
+      AUTHORS, ADD",
+    booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
+    year = "2024",
+    publisher = "Association for Computational Linguistics",
+}
+```
+-->
+
+### Groups and Tasks
+
+#### Groups
+
+- `catalan_bench`: All tasks included in CatalanBench.
+- `flores_ca`: All FLORES translation tasks from or to Catalan.
+
+#### Tags
+- `cabreu`: Three CaBREU tasks for each type of summary (extractive, abstractive and extreme).
+- `phrases_va`: Two Phrases_va tasks for language adaptation between Catalan and Valencian.
+
+#### Tasks
+
+The following tasks evaluate tasks on CatalanBench dataset using various scoring methods.
+  - `arc_ca_challenge`
+  - `arc_ca_easy`
+  - `belebele_cat_Latn`
+  - `cabreu`
+  - `catalanqa`
+  - `catcola`
+  - `copa_ca`
+  - `coqcat`
+  - `flores_ca`
+  - `flores_ca-de`
+  - `flores_ca-en`
+  - `flores_ca-es`
+  - `flores_ca-eu`
+  - `flores_ca-fr`
+  - `flores_ca-gl`
+  - `flores_ca-it`
+  - `flores_ca-pt`
+  - `flores_de-ca`
+  - `flores_en-ca`
+  - `flores_es-ca`
+  - `flores_eu-ca`
+  - `flores_fr-ca`
+  - `flores_gl-ca`
+  - `flores_it-ca`
+  - `flores_pt-ca`
+  - `mgsm_direct_ca`
+  - `openbookqa_ca`
+  - `parafraseja`
+  - `paws_ca`
+  - `phrases_ca`
+  - `piqa_ca`
+  - `siqa_ca`
+  - `teca`
+  - `veritasqa_gen_ca`
+  - `veritasqa_mc1_ca`
+  - `veritasqa_mc2_ca`
+  - `wnli_ca`
+  - `xnli_ca`
+  - `xquad_ca`
+  - `xstorycloze_ca`
+
+Some of these tasks are taken from benchmarks already available in LM Evaluation Harness. These are:
+- `belebele_cat_Latn`: Belebele Catalan
+
+
+### Checklist
+
+* [x] Is the task an existing benchmark in the literature?
+  * [ ] Have you referenced the original paper that introduced the task?
+  * [ ] If yes, does the original paper provide a reference implementation?
+    * [ ] Yes, original implementation contributed by author of the benchmark
+
+If other tasks on this dataset are already supported:
+* [ ] Is the "Main" variant of this task clearly denoted?
+* [ ] Have you provided a short sentence in a README on what each new variant adds / evaluates?
+* [ ] Have you noted which, if any, published evaluation setups are matched by this variant?

lm_eval/tasks/catalan_bench/_arc_ca_common_yaml

+tag: arc_ca
+dataset_path: projecte-aina/arc_ca
+output_type: multiple_choice
+training_split: null
+validation_split: validation
+test_split: test
+doc_to_text: "Pregunta: {{question}}\nResposta:"
+doc_to_target: "{{choices.label.index(answerKey)}}"
+doc_to_choice: "{{choices.text}}"
+should_decontaminate: true
+doc_to_decontamination_query: "Pregunta: {{question}}\nResposta:"
+metric_list:
+  - metric: acc
+    aggregation: mean
+    higher_is_better: true
+  - metric: acc_norm
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/_cabreu_common_yaml

+tag: cabreu
+dataset_path: projecte-aina/caBreu
+dataset_name: null
+output_type: generate_until
+test_split: test
+training_split: train
+validation_split: validation
+process_docs: !function utils.process_doc_cabreu
+metric_list:
+  - metric: bleu
+    aggregation: bleu
+    higher_is_better: true
+  - metric: !function utils.rouge1
+    aggregation: !function utils.rouge1_agg
+    higher_is_better: true
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/arc_ca_challenge.yaml

+task: arc_ca_challenge
+dataset_name: ARC-Challenge
+include: _arc_ca_common_yaml

lm_eval/tasks/catalan_bench/arc_ca_easy.yaml

+task: arc_ca_easy
+dataset_name: ARC-Easy
+include: _arc_ca_common_yaml

lm_eval/tasks/catalan_bench/cabreu_abstractive.yaml

+include: _cabreu_common_yaml
+task: cabreu_abstractive
+description: "Examina el text següent i genera'n un resum abstractiu, expressant el significat del text original d'una manera més natural i concisa.\n"
+doc_to_text: >-
+  Text: {{content}}
+
+  Resum:
+doc_to_target: '{{summaries["abstractive"]["a1"]}}'

lm_eval/tasks/catalan_bench/cabreu_extractive.yaml

+include: _cabreu_common_yaml
+task: cabreu_extractive
+description: "Examina el text següent i genera'n un resum extractiu, utilitzant les frases o oracions més rellevants del text original.\n"
+doc_to_text: >-
+  Text: {{content}}
+
+  Resum:
+doc_to_target: '{{summaries["extractive"]["a1"]}}'

lm_eval/tasks/catalan_bench/cabreu_extreme.yaml

+include: _cabreu_common_yaml
+task: cabreu_extreme
+description: "Examina el text següent i genera'n un resum que sigui el més concís possible i que preservi el significat del text original.\n"
+doc_to_text: >-
+  Text: {{content}}
+
+  Resum:
+doc_to_target: '{{summaries["extreme"]["a1"]}}'

lm_eval/tasks/catalan_bench/catalan_bench.yaml

+group: catalan_bench
+task:
+    - belebele_cat_Latn
+    - xnli_ca
+    - catcola
+    - copa_ca
+    - openbookqa_ca
+    - parafraseja
+    - paws_ca
+    - piqa_ca
+    - siqa_ca
+    - teca
+    - wnli_ca
+    - arc_ca_easy
+    - arc_ca_challenge
+    - xstorycloze_ca
+    - xquad_ca
+    - catalanqa
+    - coqcat
+    - flores_ca
+    - cabreu
+    - mgsm_direct_ca
+    - phrases_va
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/catalanqa.yaml

+task: catalanqa
+dataset_path: projecte-aina/catalanqa
+dataset_name: null
+output_type: generate_until
+training_split: train
+validation_split: validation
+test_split: test
+doc_to_text: "Context: {{context}}\n\nPregunta: {{question}}\n\nResposta:"
+doc_to_target: '{{answers[0]["text"]}}'
+target_delimiter: ' '
+process_results: !function utils.process_results_qa
+generation_kwargs:
+  until:
+    - "\n"
+  do_sample: false
+  temperature: 0.0
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+  - metric: f1
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/catcola.yaml

+task: catcola
+dataset_path: nbel/CatCoLA
+output_type: multiple_choice
+training_split: train
+validation_split: validation
+test_split: null
+doc_to_text: "{{Sentence}}\nPregunta: Té sentit aquesta frase?\nResposta:"
+doc_to_target: label
+doc_to_choice: ["no", "sí"]
+metric_list:
+  - metric: mcc
+  - metric: acc
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/copa_ca.yaml

+task: copa_ca
+dataset_path: projecte-aina/COPA-ca
+dataset_name: null
+output_type: multiple_choice
+training_split: train
+validation_split: validation
+test_split: test
+process_docs: !function utils.process_docs_copa_ca
+doc_to_text: '{{premise[:-1].strip() + " " + {"cause": "perquè", "effect": "i per tant"}[question]}}'
+doc_to_target: '{{choice1 if label == 0 else choice2}}'
+doc_to_choice: '{{[choice1, choice2]}}'
+metric_list:
+  - metric: acc
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/coqcat.yaml

+task: coqcat
+dataset_path: projecte-aina/CoQCat
+output_type: generate_until
+training_split: train
+validation_split: validation
+test_split: test
+doc_to_text: '{{story+"\n\n"}}{% for i in range(questions|length-1) %}{{"Q: "+questions[i]+"\n\n"+"A: "+answers["input_text"][i]+"\n\n"}}{% endfor %}{{"Q: "+questions[-1]+"\n\n"+"A:"}}'
+doc_to_target: '{{ answers["input_text"][questions|length - 1] }}'
+process_results: !function utils.process_results_coqcat
+should_decontaminate: true
+doc_to_decontamination_query: "{{story}} {{question.input_text|join('\n')}}"
+generation_kwargs:
+  until:
+    - "\nQ:"
+metric_list:
+  - metric: "em"
+    aggregation: mean
+    higher_is_better: true
+  - metric: "f1"
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 1.0

lm_eval/tasks/catalan_bench/flores_ca/_flores_common_yaml

+dataset_path: facebook/flores
+dataset_name: all
+output_type: generate_until
+training_split: dev
+validation_split: dev
+test_split: devtest
+fewshot_split: dev
+target_delimiter: ''
+generation_kwargs:
+  until:
+    - "\n"
+metric_list:
+  - metric: bleu
+    aggregation: bleu
+    higher_is_better: true
+  - metric: ter
+    aggregation: ter
+    higher_is_better: false
+  - metric: chrf
+    aggregation: chrf
+    higher_is_better: true
+metadata:
+  version: 1.0
+dataset_kwargs:
+  trust_remote_code: true

lm_eval/tasks/catalan_bench/flores_ca/create_yamls_flores_ca.py

+"""
+Script to generate task YAMLs for the FLORES-200 dataset.
+Based on `tasks/translation/utils.py`.
+"""
+
+import argparse
+
+import yaml
+from langcodes import Language
+
+
+# constants
+_LANGUAGES = [
+    "ace_Arab",
+    "bam_Latn",
+    "dzo_Tibt",
+    "hin_Deva",
+    "khm_Khmr",
+    "mag_Deva",
+    "pap_Latn",
+    "sot_Latn",
+    "tur_Latn",
+    "ace_Latn",
+    "ban_Latn",
+    "ell_Grek",
+    "hne_Deva",
+    "kik_Latn",
+    "mai_Deva",
+    "pbt_Arab",
+    "spa_Latn",
+    "twi_Latn",
+    "acm_Arab",
+    "bel_Cyrl",
+    "eng_Latn",
+    "hrv_Latn",
+    "kin_Latn",
+    "mal_Mlym",
+    "pes_Arab",
+    "srd_Latn",
+    "tzm_Tfng",
+    "acq_Arab",
+    "bem_Latn",
+    "epo_Latn",
+    "hun_Latn",
+    "kir_Cyrl",
+    "mar_Deva",
+    "plt_Latn",
+    "srp_Cyrl",
+    "uig_Arab",
+    "aeb_Arab",
+    "ben_Beng",
+    "est_Latn",
+    "hye_Armn",
+    "kmb_Latn",
+    "min_Arab",
+    "pol_Latn",
+    "ssw_Latn",
+    "ukr_Cyrl",
+    "afr_Latn",
+    "bho_Deva",
+    "eus_Latn",
+    "ibo_Latn",
+    "kmr_Latn",
+    "min_Latn",
+    "por_Latn",
+    "sun_Latn",
+    "umb_Latn",
+    "ajp_Arab",
+    "bjn_Arab",
+    "ewe_Latn",
+    "ilo_Latn",
+    "knc_Arab",
+    "mkd_Cyrl",
+    "prs_Arab",
+    "swe_Latn",
+    "urd_Arab",
+    "aka_Latn",
+    "bjn_Latn",
+    "fao_Latn",
+    "ind_Latn",
+    "knc_Latn",
+    "mlt_Latn",
+    "quy_Latn",
+    "swh_Latn",
+    "uzn_Latn",
+    "als_Latn",
+    "bod_Tibt",
+    "fij_Latn",
+    "isl_Latn",
+    "kon_Latn",
+    "mni_Beng",
+    "ron_Latn",
+    "szl_Latn",
+    "vec_Latn",
+    "amh_Ethi",
+    "bos_Latn",
+    "fin_Latn",
+    "ita_Latn",
+    "kor_Hang",
+    "mos_Latn",
+    "run_Latn",
+    "tam_Taml",
+    "vie_Latn",
+    "apc_Arab",
+    "bug_Latn",
+    "fon_Latn",
+    "jav_Latn",
+    "lao_Laoo",
+    "mri_Latn",
+    "rus_Cyrl",
+    "taq_Latn",
+    "war_Latn",
+    "arb_Arab",
+    "bul_Cyrl",
+    "fra_Latn",
+    "jpn_Jpan",
+    "lij_Latn",
+    "mya_Mymr",
+    "sag_Latn",
+    "taq_Tfng",
+    "wol_Latn",
+    "arb_Latn",
+    "cat_Latn",
+    "fur_Latn",
+    "kab_Latn",
+    "lim_Latn",
+    "nld_Latn",
+    "san_Deva",
+    "tat_Cyrl",
+    "xho_Latn",
+    "ars_Arab",
+    "ceb_Latn",
+    "fuv_Latn",
+    "kac_Latn",
+    "lin_Latn",
+    "nno_Latn",
+    "sat_Olck",
+    "tel_Telu",
+    "ydd_Hebr",
+    "ary_Arab",
+    "ces_Latn",
+    "gaz_Latn",
+    "kam_Latn",
+    "lit_Latn",
+    "nob_Latn",
+    "scn_Latn",
+    "tgk_Cyrl",
+    "yor_Latn",
+    "arz_Arab",
+    "cjk_Latn",
+    "gla_Latn",
+    "kan_Knda",
+    "lmo_Latn",
+    "npi_Deva",
+    "shn_Mymr",
+    "tgl_Latn",
+    "yue_Hant",
+    "asm_Beng",
+    "ckb_Arab",
+    "gle_Latn",
+    "kas_Arab",
+    "ltg_Latn",
+    "nso_Latn",
+    "sin_Sinh",
+    "tha_Thai",
+    "zho_Hans",
+    "ast_Latn",
+    "crh_Latn",
+    "glg_Latn",
+    "kas_Deva",
+    "ltz_Latn",
+    "nus_Latn",
+    "slk_Latn",
+    "tir_Ethi",
+    "zho_Hant",
+    "awa_Deva",
+    "cym_Latn",
+    "grn_Latn",
+    "kat_Geor",
+    "lua_Latn",
+    "nya_Latn",
+    "slv_Latn",
+    "tpi_Latn",
+    "zsm_Latn",
+    "ayr_Latn",
+    "dan_Latn",
+    "guj_Gujr",
+    "kaz_Cyrl",
+    "lug_Latn",
+    "oci_Latn",
+    "smo_Latn",
+    "tsn_Latn",
+    "zul_Latn",
+    "azb_Arab",
+    "deu_Latn",
+    "hat_Latn",
+    "kbp_Latn",
+    "luo_Latn",
+    "ory_Orya",
+    "sna_Latn",
+    "tso_Latn",
+    "azj_Latn",
+    "dik_Latn",
+    "hau_Latn",
+    "kea_Latn",
+    "lus_Latn",
+    "pag_Latn",
+    "snd_Arab",
+    "tuk_Latn",
+    "bak_Cyrl",
+    "dyu_Latn",
+    "heb_Hebr",
+    "khk_Cyrl",
+    "lvs_Latn",
+    "pan_Guru",
+    "som_Latn",
+    "tum_Latn",
+]
+LANGUAGE_PAIRS = [
+    (a, b) for idx, a in enumerate(_LANGUAGES) for b in _LANGUAGES[idx + 1 :]
+]
+
+LANGUAGES_OF_INTEREST = [
+    "cat_Latn",
+    "spa_Latn",
+    "eng_Latn",
+    "glg_Latn",
+    "eus_Latn",
+    "ita_Latn",
+    "deu_Latn",
+    "por_Latn",
+    "fra_Latn",
+]
+MAIN_LANG = "cat_Latn"
+LANGUAGE_PAIRS = [
+    (a, b)
+    for (a, b) in LANGUAGE_PAIRS
+    if a in LANGUAGES_OF_INTEREST
+    and b in LANGUAGES_OF_INTEREST
+    and "cat_Latn" in (a, b)
+]
+
+# auxiliary functions
+
+
+def code_to_language_name(code):
+    return Language.make(language=Language.get(code)["language"]).display_name()
+
+
+def code_to_short_name(code):
+    return Language.get(code)["language"]
+
+
+def jinja_var(s):
+    return "{{" + s + "}}"
+
+
+def doc_to_text(src: str, tgt: str) -> str:
+    src_name, tgt_name = map(code_to_language_name, [src, tgt])
+
+    return f"""\
+{src_name} sentence: {jinja_var('sentence_' + src)}
+{tgt_name} sentence:"""
+
+
+def doc_to_target(tgt: str) -> str:
+    return f"{jinja_var('sentence_' + tgt)}"
+
+
+# main function
+
+
+def gen_lang_yamls(output_dir: str, overwrite: bool) -> None:
+    """
+    Generate a YAML file for each translation direction.
+    """
+
+    err = []
+    for src, tgt in LANGUAGE_PAIRS:
+        # do both translation directions for each lang pair
+        for src, tgt in [(src, tgt), (tgt, src)]:
+            lang_pair_name = f"{code_to_short_name(src)}-{code_to_short_name(tgt)}"
+            yaml_file_name = f"flores_{lang_pair_name}.yaml"
+
+            try:
+                with open(
+                    f"{output_dir}/{yaml_file_name}",
+                    "w" if overwrite else "x",
+                    encoding="utf-8",
+                ) as outfile:
+                    print(f"Creating {yaml_file_name}...")
+                    outfile.write("# File generated by `create-yamls.py`\n")
+                    yaml.dump(
+                        {
+                            #                             "group": [f"{BENCH_NAME}_bench", f"{BENCH_NAME}_bench_flores"],
+                            #                            "group": "flores_ca",
+                            "include": "_flores_common_yaml",
+                            "task": f"flores_{lang_pair_name}",
+                            "doc_to_text": doc_to_text(src, tgt),
+                            "doc_to_target": doc_to_target(tgt),
+                        },
+                        outfile,
+                        sort_keys=False,
+                    )
+
+            except FileExistsError:
+                err.append(yaml_file_name)
+
+    if len(err) > 0:
+        raise FileExistsError(
+            "Files were not created because they already exist:"
+            f" {', '.join(err)}"
+            "\nUse flag --overwrite to overwrite them."
+        )
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--overwrite",
+        default=False,
+        action="store_true",
+        help="Overwrite files if they already exist",
+    )
+    parser.add_argument(
+        "--output-dir", default=".", help="Directory to write yaml files to"
+    )
+    args = parser.parse_args()
+
+    gen_lang_yamls(output_dir=args.output_dir, overwrite=args.overwrite)
+
+
+if __name__ == "__main__":
+    main()

lm_eval/tasks/catalan_bench/flores_ca/flores_ca-de.yaml

+# File generated by `create-yamls.py`
+include: _flores_common_yaml
+task: flores_ca-de
+doc_to_text: 'Catalan sentence: {{sentence_cat_Latn}}
+
+  German sentence:'
+doc_to_target: '{{sentence_deu_Latn}}'