sample_id stringlengths 23 23 | category stringclasses 6
values | task_name stringclasses 21
values | source_dataset stringclasses 15
values | original_split stringclasses 3
values | messages listlengths 3 3 |
|---|---|---|---|---|---|
benchmark_full_00000000 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Choose the correct label for evidence presence in the ESG passage.\nUse exactly one of: Yes, No.\nReturn JSON ... |
benchmark_full_00000001 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_parent_category_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB-aligned ESG sentence classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Select the high-level ESG classification for this sentence.\nLabels: Business Model & Innovation, Envir... |
benchmark_full_00000002 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Review the evidence and output the correct fact-checking label for the climate claim.... |
benchmark_full_00000003 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the verdict for the climate claim using only explicit information from the evi... |
benchmark_full_00000004 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_parent_category_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB-aligned ESG sentence classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign one SASB parent category to the sentence.\nValid labels: Business Model & Innovation, Environmen... |
benchmark_full_00000005 | esg_relevance_detection | environmental_claims__environmental_claims_classification | environmental_claims | test | [
{
"role": "system",
"content": "You are an environmental claims classification assistant. Use only the provided sentence. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify whether the sentence refers to an environmental claim.\nValid labels: ... |
benchmark_full_00000006 | esg_commitment_analysis | promise_eval_en__promise_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG promise analysis assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify whether this report excerpt includes an ESG target, commitment, or promise.\nAllowed labels: Yes, No.\n... |
benchmark_full_00000007 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Read the climate claim and evidence set, then return the correct verification label.\... |
benchmark_full_00000008 | esg_topic_identification | esg_ftse_corpus__ftse_esg_primary_category_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Categorize this article for ESG multiclass classification.\nLabels: Environmental, Social, Governance, Irrelevant.\nReturn ... |
benchmark_full_00000009 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine the final label for the climate claim from the evidence passages without us... |
benchmark_full_00000010 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Evaluate the climate claim against the evidence and return one valid label in JSON.\n... |
benchmark_full_00000011 | esg_impact_analysis | ml_esg3_en__mlesg3_impact_level_classification | ML-ESG3 | test | [
{
"role": "system",
"content": "You are an ESG impact severity classification assistant. Use only the provided English text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Analyze the provided ESG news text and choose its impact severity.\nAllowed labels: Low, Medium, High.\nReturn J... |
benchmark_full_00000012 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the verdict for the climate claim using only explicit information from the evi... |
benchmark_full_00000013 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign the evidence status for the ESG report excerpt.\nValid labels: Yes, No.\nReturn JSON with this schema: ... |
benchmark_full_00000014 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign the sentence to the correct fine-grained SASB issue.\nUse exactly one of: Access & Affordability, Air Qua... |
benchmark_full_00000015 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which of the following best captures the implication of reducing the discount rate from 5% to 2% on the cost-effective trajectory for climate mitig... |
benchmark_full_00000016 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify the detailed sustainability topic discussed in the sentence.\nValid answers: Access & Affordability, Ai... |
benchmark_full_00000017 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign the article to the ESG issue category that best describes its content.\nCategory list: Acc... |
benchmark_full_00000018 | esg_commitment_analysis | netzero_reduction_data__netzero_reduction_target_classification | netzero_reduction_data | test | [
{
"role": "system",
"content": "You are a climate target classification assistant. Use only the provided text. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Evaluate whether the text refers to net-zero, reduction, or none.\nValid labels: net-zero,... |
benchmark_full_00000019 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign the sentence to the correct fine-grained SASB issue.\nUse exactly one of: Access & Affordability, Air Qua... |
benchmark_full_00000020 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Read the title and article excerpt, then assign the correct ESG issue category.\nValid categories... |
benchmark_full_00000021 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine which specificity category best fits the paragraph.\nValid ... |
benchmark_full_00000022 | esg_relevance_detection | esg_ftse_corpus__ftse_esg_relevance_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine whether the article is ESG-related.\nValid labels: ESG-related, Not ESG-related.\nReturn JSON with this schema: {... |
benchmark_full_00000023 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Label the sentence with the appropriate SASB issue category.\nValid options: Access & Affordability, Air Quality... |
benchmark_full_00000024 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify the fine-grained ESG category for this article.\nOnly use one of these categories: Acces... |
benchmark_full_00000025 | esg_commitment_analysis | promise_eval_en__promise_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG promise analysis assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Label the passage as containing an ESG promise or not.\nValid options: Yes, No.\nReturn JSON with this schema: {... |
benchmark_full_00000026 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Analyze the ESG report text and classify evidence presence.\nAllowed labels: Yes, No.\nReturn JSON with this s... |
benchmark_full_00000027 | esg_impact_analysis | climate_sentiment__climate_sentiment_classification | climate_sentiment | test | [
{
"role": "system",
"content": "You are a climate disclosure sentiment classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Judge whether the paragraph expresses climate risk, climate opportunity... |
benchmark_full_00000028 | esg_claim_verification | promise_eval_en__promise_evidence_quality_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence quality assessment assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the evidence as clear, not clear, misleading, or N/A.\nReturn JSON with this schema: {\"evid... |
benchmark_full_00000029 | esg_relevance_detection | esg_ftse_corpus__ftse_esg_relevance_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Review the article and judge whether it is ESG-related.\nValid labels: ESG-related, Not ESG-related.\nReturn JSON with this... |
benchmark_full_00000030 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Evaluate the climate claim against the evidence and return one valid label in JSON.\n... |
benchmark_full_00000031 | esg_relevance_detection | esg_ftse_corpus__ftse_esg_relevance_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify whether ESG content is present in the article.\nValid labels: ESG-related, Not ESG-related.\nReturn JSON with this... |
benchmark_full_00000032 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_parent_category_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB-aligned ESG sentence classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the sentence into one SASB parent category.\nValid labels: Business Model & Innovation, Enviro... |
benchmark_full_00000033 | esg_topic_identification | tcfd_recommendations__tcfd_recommendations_classification | tcfd_recommendations | test | [
{
"role": "system",
"content": "You are a TCFD recommendation classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the TCFD category label for the paragraph.\nValid labels: none, metrics, ... |
benchmark_full_00000034 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign the sentence to the correct fine-grained SASB issue.\nUse exactly one of: Access & Affordability, Air Qua... |
benchmark_full_00000035 | esg_relevance_detection | esg_ftse_corpus__ftse_esg_relevance_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Is this article relevant to ESG analysis?\nAnswer with one valid label: ESG-related, Not ESG-related.\nReturn JSON with thi... |
benchmark_full_00000036 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Perform SASB issue-topic classification.\nValid labels: Access & Affordability, Air Quality, Business Ethics, Bu... |
benchmark_full_00000037 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Judge the climate claim against the evidence and return one valid label.\nValid label... |
benchmark_full_00000038 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine the most appropriate ESG issue label for the news text.\nChoose one valid category from... |
benchmark_full_00000039 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide the correct specificity label for the paragraph.\nValid labels... |
benchmark_full_00000040 | esg_topic_identification | esg_ftse_corpus__ftse_esg_primary_category_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the primary ESG category for this article.\nValid categories: Environmental, Social, Governance, Irrelevant.\nReturn... |
benchmark_full_00000041 | esg_impact_analysis | ml_esg3_en__mlesg3_impact_duration_classification | ML-ESG3 | test | [
{
"role": "system",
"content": "You are an ESG impact duration classification assistant. Use only the provided English text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Analyze the ESG content and select its impact duration.\nAllowed labels: Less than 2 years, 2 to 5 years, More t... |
benchmark_full_00000042 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify whether this report excerpt contains evidence relevant to an ESG commitment or claim.\nAllowed labels... |
benchmark_full_00000043 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which factor is most directly responsible for the limited adoption of ESCO business models despite their potential to mitigate financial risks and ... |
benchmark_full_00000044 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide if the passage contains evidence that substantiates the ESG promise or statement.\nValid labels: Yes, N... |
benchmark_full_00000045 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify whether the climate-related paragraph is specific.\nValid la... |
benchmark_full_00000046 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign one SASB child issue label to the sentence.\nValid labels: Access & Affordability, Air Quality, Business ... |
benchmark_full_00000047 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the verdict for the climate claim using only explicit information from the evi... |
benchmark_full_00000048 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "According to CDP Full Corporate Questionnaire April 2025 – Module 8-13, when reporting water intensity metrics for products, which of the following... |
benchmark_full_00000049 | esg_topic_identification | tcfd_recommendations__tcfd_recommendations_classification | tcfd_recommendations | test | [
{
"role": "system",
"content": "You are a TCFD recommendation classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Inspect the paragraph and classify its TCFD recommendation category.\nValid labe... |
benchmark_full_00000050 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Judge the climate claim against the evidence and return one valid label.\nValid label... |
benchmark_full_00000051 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the paragraph as specific or non-specific in climate-related... |
benchmark_full_00000052 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which of the following scenarios would most directly challenge an entity's ability to comply with future regulatory obligations regarding fuel econ... |
benchmark_full_00000053 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Review the evidence and output the correct fact-checking label for the climate claim.... |
benchmark_full_00000054 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return whether evidence is present in the ESG passage.\nValid labels: Yes, No.\nReturn JSON with this schema: ... |
benchmark_full_00000055 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Read the evidence carefully and determine the correct verdict for the climate claim.\... |
benchmark_full_00000056 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which scenario would invalidate an entity's claim of using renewable energy under the disclosure requirements of the Leisure Facilities – Sustainab... |
benchmark_full_00000057 | esg_commitment_analysis | netzero_reduction_data__netzero_reduction_target_classification | netzero_reduction_data | test | [
{
"role": "system",
"content": "You are a climate target classification assistant. Use only the provided text. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the text as net-zero, reduction, or none.\nValid labels: net-zero, reduction, non... |
benchmark_full_00000058 | esg_claim_verification | promise_eval_en__promise_evidence_quality_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence quality assessment assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the evidence quality for the ESG claim or promise described below.\nValid labels: Clear, Not C... |
benchmark_full_00000059 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Select one verification label for the climate claim from SUPPORTS, REFUTES, NOT_ENOUG... |
benchmark_full_00000060 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Choose the most specific ESG issue category supported by the article.\nValid categories: Access t... |
benchmark_full_00000061 | esg_commitment_analysis | climate_commitments_actions__climate_commitments_actions_classification | climate_commitments_actions | test | [
{
"role": "system",
"content": "You are a climate disclosure classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine if the paragraph is focused on climate commitments and actions.\nValid ... |
benchmark_full_00000062 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which statement accurately captures the relationship between agricultural intensification, deforestation control mechanisms, and their broader impl... |
benchmark_full_00000063 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which combination of considerations ensures the most comprehensive evaluation of a company's climate transition risk strategy, given the uncertaint... |
benchmark_full_00000064 | esg_commitment_analysis | netzero_reduction_data__netzero_reduction_target_classification | netzero_reduction_data | test | [
{
"role": "system",
"content": "You are a climate target classification assistant. Use only the provided text. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the text as about net-zero, reduction, or none.\nValid labels: net-zero, reductio... |
benchmark_full_00000065 | esg_relevance_detection | climate_detection__climate_detection_classification | climate_detection | test | [
{
"role": "system",
"content": "You are a climate disclosure classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Check whether the paragraph is specifically climate-related.\nValid labels: yes, ... |
benchmark_full_00000066 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_issue_topic_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB issue topic classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign one SASB child issue label to the sentence.\nValid labels: Access & Affordability, Air Quality, Business ... |
benchmark_full_00000067 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine the final label for the climate claim from the evidence passages without us... |
benchmark_full_00000068 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Use the evidence passages to decide whether the climate claim is supported, refuted, ... |
benchmark_full_00000069 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_parent_category_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB-aligned ESG sentence classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify this English sentence into a SASB parent class.\nClasses: Business Model & Innovation, Environ... |
benchmark_full_00000070 | esg_relevance_detection | esg_ftse_corpus__ftse_esg_relevance_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine whether the article is ESG-related.\nValid labels: ESG-related, Not ESG-related.\nReturn JSON with this schema: {... |
benchmark_full_00000071 | esg_impact_analysis | climate_sentiment__climate_sentiment_classification | climate_sentiment | test | [
{
"role": "system",
"content": "You are a climate disclosure sentiment classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Judge whether the paragraph expresses climate risk, climate opportunity... |
benchmark_full_00000072 | esg_topic_identification | tcfd_recommendations__tcfd_recommendations_classification | tcfd_recommendations | test | [
{
"role": "system",
"content": "You are a TCFD recommendation classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Read the paragraph and decide whether its TCFD category is none, metrics, strate... |
benchmark_full_00000073 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine whether the text includes concrete evidence or no evidence.\nReturn JSON with this schema: {\"eviden... |
benchmark_full_00000074 | esg_commitment_analysis | promise_eval_en__promise_verification_timeline_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG promise timeline classification assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Perform ESG promise timeline classification.\nUse exactly one label from: Already, Less than 2 ye... |
benchmark_full_00000075 | esg_topic_identification | esg_ftse_corpus__ftse_esg_primary_category_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify the article by ESG topic area.\nUse Environmental for environment, Social for people/stakeholders, Governance for ... |
benchmark_full_00000076 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide which specificity category the climate-related paragraph belon... |
benchmark_full_00000077 | esg_commitment_analysis | promise_eval_en__promise_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG promise analysis assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify whether this passage includes an ESG promise or commitment.\nAllowed labels: Yes, No.\nReturn JSON with... |
benchmark_full_00000078 | esg_relevance_detection | environmental_claims__environmental_claims_classification | environmental_claims | test | [
{
"role": "system",
"content": "You are an environmental claims classification assistant. Use only the provided sentence. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Review the sentence and return whether it is an environmental claim.\nValid lab... |
benchmark_full_00000079 | esg_impact_analysis | ml_esg2_en__mlesg2_impact_type_classification | ML-ESG2 | test | [
{
"role": "system",
"content": "You are an ESG impact classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Label the article as either an ESG opportunity or an ESG risk.\nAllowed labels: Opportunity, Ri... |
benchmark_full_00000080 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "When an entity is required by a jurisdictional authority or exchange to use a different method for measuring greenhouse gas emissions, under which ... |
benchmark_full_00000081 | esg_relevance_detection | environmental_claims__environmental_claims_classification | environmental_claims | test | [
{
"role": "system",
"content": "You are an environmental claims classification assistant. Use only the provided sentence. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide if the provided sentence discusses an environmental claim.\nValid labels... |
benchmark_full_00000082 | esg_relevance_detection | climate_detection__climate_detection_classification | climate_detection | test | [
{
"role": "system",
"content": "You are a climate disclosure classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Classify if the paragraph mentions climate-related issues.\nValid labels: yes, no... |
benchmark_full_00000083 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine the final verdict for the climate claim using only the provided evidence.\n... |
benchmark_full_00000084 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Determine if the company gives verifiable evidence in this passage.\nValid labels: Yes, No.\nReturn JSON with ... |
benchmark_full_00000085 | esg_topic_identification | sasb_aligned_esg_sentences__sasb_parent_category_classification | SASB-Aligned ESG Sentences | test | [
{
"role": "system",
"content": "You are a SASB-aligned ESG sentence classification assistant. Use only the provided English sentence. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify the broad sustainability domain discussed in the sentence.\nValid answers: Business Model & In... |
benchmark_full_00000086 | esg_impact_analysis | ml_esg3_en__mlesg3_impact_level_classification | ML-ESG3 | test | [
{
"role": "system",
"content": "You are an ESG impact severity classification assistant. Use only the provided English text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Analyze the provided ESG news text and choose its impact severity.\nAllowed labels: Low, Medium, High.\nReturn J... |
benchmark_full_00000087 | esg_impact_analysis | climate_sentiment__climate_sentiment_classification | climate_sentiment | test | [
{
"role": "system",
"content": "You are a climate disclosure sentiment classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the climate sentiment label for the paragraph.\nValid labels: ri... |
benchmark_full_00000088 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "What is the primary reason the ISSB rejected the hierarchical structure approach despite its potential to enhance comparability of digital informat... |
benchmark_full_00000089 | esg_commitment_analysis | netzero_reduction_data__netzero_reduction_target_classification | netzero_reduction_data | test | [
{
"role": "system",
"content": "You are a climate target classification assistant. Use only the provided text. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Using only the text, determine whether it covers a net-zero target, a reduction target, or... |
benchmark_full_00000090 | esg_qa | esgenius__esg_qa | esgenius | benchmark_only | [
{
"role": "system",
"content": "You are a helpful assistant specialized in ESG and sustainability benchmark tasks."
},
{
"role": "user",
"content": "Which inference about the geographic distribution of asset managers and asset owners can be drawn from their respective headquarters locations as d... |
benchmark_full_00000091 | esg_commitment_analysis | climate_specificity__climate_specificity_classification | climate_specificity | test | [
{
"role": "system",
"content": "You are a climate disclosure specificity classification assistant. Use only the provided paragraph. Do not infer facts beyond the text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Check whether the paragraph is specifically climate-related or remain... |
benchmark_full_00000092 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Identify the ESG issue category discussed in the news article.\nValid categories: Access to Commu... |
benchmark_full_00000093 | esg_claim_verification | promise_eval_en__promise_evidence_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG evidence detection assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Label the passage as containing ESG evidence or not.\nValid options: Yes, No.\nReturn JSON with this schema: {... |
benchmark_full_00000094 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide whether the evidence supports, refutes, disputes, or leaves unresolved the cli... |
benchmark_full_00000095 | esg_topic_identification | ml_esg1_en__mlesg1_esg_issue_classification | ML-ESG1 | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided English news title and article excerpt. Return valid JSON."
},
{
"role": "user",
"content": "Task: Assign a single ESG issue label to the news article.\nValid labels: Access to Communications, Acc... |
benchmark_full_00000096 | esg_commitment_analysis | promise_eval_en__promise_status_classification | PromiseEval | test | [
{
"role": "system",
"content": "You are an ESG promise analysis assistant. Use only the provided English ESG report passage. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide if the passage states a concrete ESG promise, target, or future action.\nValid labels: Yes, No.\nReturn J... |
benchmark_full_00000097 | esg_claim_verification | climate_fever__claim_verification | climate_fever | all | [
{
"role": "system",
"content": "You are a climate claim verification assistant. Use only the provided claim and evidence. Do not use outside knowledge. Return valid JSON."
},
{
"role": "user",
"content": "Task: Decide whether the evidence supports, refutes, disputes, or leaves unresolved the cli... |
benchmark_full_00000098 | esg_topic_identification | esg_ftse_corpus__ftse_esg_primary_category_classification | ESG-FTSE-Corpus | test | [
{
"role": "system",
"content": "You are an ESG news classification assistant. Use only the provided article text. Return valid JSON."
},
{
"role": "user",
"content": "Task: Return the primary ESG category for this article.\nValid categories: Environmental, Social, Governance, Irrelevant.\nReturn... |
End of preview. Expand in Data Studio
ESGU-Bench
ESGU-Bench is an English multi-task benchmark for evaluating language models on ESG and sustainability understanding. It contains classification, claim verification, and multiple-choice question-answering examples drawn from human-annotated or expert-validated sources. Every example uses a three-message system/user/assistant conversation so the same file can be used for prompted evaluation across tasks.
Dataset file
| File | Rows | Description |
|---|---|---|
esgu_bench.jsonl |
9,514 | Full public benchmark across 21 detail tasks |
Internal construction artifacts, task-wise duplicate views, and analysis files are not included.
Loading
from datasets import load_dataset
benchmark = load_dataset("nowzero/ESGU-Bench", split="train")
messages = benchmark[0]["messages"]
prompt_messages = messages[:-1]
reference_answer = messages[-1]["content"]
Hugging Face names the single published file train for loading purposes; train here denotes the dataset configuration, not a model-training recommendation. ESGU-Bench should be kept held out for evaluation.
To load a downloaded copy directly:
benchmark = load_dataset(
"json",
data_files="esgu_bench.jsonl",
split="train",
)
Main schema
sample_id: unique ESGU-Bench identifier used to join predictionscategory,task_name: unified category and task identifierssource_dataset,original_split: source provenancemessages:system,user, andassistantmessages; the final assistant message contains the reference answer
Composition
| Category | Rows | Detail tasks |
|---|---|---|
| ESG relevance detection | 1,059 | 3 |
| ESG topic identification | 2,390 | 5 |
| ESG commitment analysis | 1,784 | 5 |
| ESG impact analysis | 810 | 4 |
| ESG claim verification | 2,335 | 3 |
| ESG question answering | 1,136 | 1 |
| Total | 9,514 | 21 |
ESGU-Bench contains 21 tasks across six ESG understanding areas. These tasks cover ESG relevance, issue and taxonomy classification, commitments and targets, sentiment and impact analysis, evidence and claim verification, and ESG multiple-choice reasoning.
Evaluation
The released benchmark normalizes examples from the source datasets into a unified messages schema rather than reproducing their original records verbatim. For most tasks, prompt the model with messages[:-1] and compare its output with messages[-1]. For the ESG question-answering task, the evaluation code extracts the question and options from the unified user message and excludes the added Context section, matching the context-free multiple-choice setup used in our experiments. The released protocol uses strict task-specific parsing and reports classification F1/accuracy or ESG QA accuracy at task and category level. Use deterministic decoding and report the aggregation method with any overall score.
Intended use and limitations
ESGU-Bench supports reproducible comparison of ESG-capable language models and should remain held out when reporting benchmark results. It inherits the scope, annotation choices, and possible errors of its upstream datasets; exact-format scoring can also penalize malformed but semantically plausible answers. Results do not establish factual reliability, calibration, or fitness for financial, legal, regulatory, investment, or compliance decisions.
Citation
If you use ESGU-Bench, please cite:
Jaeyoung Lee and Misuk Kim. “ESGPT: A Data-Centric ESG Language Model with Unified Instruction and Benchmark Resources.” In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026.
@inproceedings{lee2026esgpt,
title = {{ESGPT}: A Data-Centric {ESG} Language Model with Unified Instruction and Benchmark Resources},
author = {Lee, Jaeyoung and Kim, Misuk},
booktitle = {Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing},
year = {2026}
}
Code: https://github.com/nowzer0/ESGPT
License
ESGU-Bench is released under the Creative Commons Attribution-NonCommercial 4.0 International license (CC BY-NC 4.0). See LICENSE for details.
- Downloads last month
- 53