data_source stringclasses 1
value | prompt listlengths 2 2 | ability stringclasses 1
value | reward_model dict | extra_info dict |
|---|---|---|---|---|
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 716.8219136074913, \"nll_base\": 6.574827432632446, \"eval_ref\": \"global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__... | {
"index": "global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb131072__v000",
"uid": "global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb131072__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 716.8219136074913, \"nll_base\": 6.574827432632446, \"eval_ref\": \"global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb... | {
"index": "global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb65536__v002",
"uid": "global_mmlu_full_en_clinical_knowledge__d78fe3f0__s000__tb65536__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_econometrics__5d236586__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 586.7302208621402, \"nll_base\": 6.374565124511719, \"eval_ref\": \"global_mmlu_full_en_econometrics__5d236586__s000__tb65536__v000\... | {
"index": "global_mmlu_full_en_econometrics__5d236586__s000__tb65536__v000",
"uid": "global_mmlu_full_en_econometrics__5d236586__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_electrical_engineering__ad0de0d3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 719.1943849130886, \"nll_base\": 6.578131675720215, \"eval_ref\": \"global_mmlu_full_en_electrical_engineering__ad0de0d3... | {
"index": "global_mmlu_full_en_electrical_engineering__ad0de0d3__s000__tb131072__v000",
"uid": "global_mmlu_full_en_electrical_engineering__ad0de0d3__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_geography__9504da39__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1094.880701782392, \"nll_base\": 6.998400688171387, \"eval_ref\": \"global_mmlu_full_en_high_school_geography__9504da39__s0... | {
"index": "global_mmlu_full_en_high_school_geography__9504da39__s000__tb65536__v000",
"uid": "global_mmlu_full_en_high_school_geography__9504da39__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_government_and_politics__79d61e17__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1032.6959147834157, \"nll_base\": 6.93992805480957, \"eval_ref\": \"global_mmlu_full_en_high_school_governmen... | {
"index": "global_mmlu_full_en_high_school_government_and_politics__79d61e17__s000__tb65536__v000",
"uid": "global_mmlu_full_en_high_school_government_and_politics__79d61e17__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_microeconomics__5b48f924__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 568.1610074519392, \"nll_base\": 6.342404842376709, \"eval_ref\": \"global_mmlu_full_en_high_school_microeconomics__... | {
"index": "global_mmlu_full_en_high_school_microeconomics__5b48f924__s000__tb131072__v000",
"uid": "global_mmlu_full_en_high_school_microeconomics__5b48f924__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_psychology__02780070__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 826.496555173162, \"nll_base\": 6.717195749282837, \"eval_ref\": \"global_mmlu_full_en_high_school_psychology__02780070_... | {
"index": "global_mmlu_full_en_high_school_psychology__02780070__s000__tb131072__v000",
"uid": "global_mmlu_full_en_high_school_psychology__02780070__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_statistics__f4517813__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1245.350717575934, \"nll_base\": 7.127172470092773, \"eval_ref\": \"global_mmlu_full_en_high_school_statistics__f4517813... | {
"index": "global_mmlu_full_en_high_school_statistics__f4517813__s000__tb262144__v000",
"uid": "global_mmlu_full_en_high_school_statistics__f4517813__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_high_school_us_history__09100d56__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1168.8975387217429, \"nll_base\": 7.06381630897522, \"eval_ref\": \"global_mmlu_full_en_high_school_us_history__09100d56... | {
"index": "global_mmlu_full_en_high_school_us_history__09100d56__s000__tb131072__v000",
"uid": "global_mmlu_full_en_high_school_us_history__09100d56__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_jurisprudence__80bf630c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1020.3675614453726, \"nll_base\": 6.927918195724487, \"eval_ref\": \"global_mmlu_full_en_jurisprudence__80bf630c__s000__tb65536__v0... | {
"index": "global_mmlu_full_en_jurisprudence__80bf630c__s000__tb65536__v000",
"uid": "global_mmlu_full_en_jurisprudence__80bf630c__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_management__0eff5433__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1618.3023983371727, \"nll_base\": 7.389132976531982, \"eval_ref\": \"global_mmlu_full_en_management__0eff5433__s000__tb65536__v000\", ... | {
"index": "global_mmlu_full_en_management__0eff5433__s000__tb65536__v000",
"uid": "global_mmlu_full_en_management__0eff5433__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 912.506754823979, \"nll_base\": 6.816195487976074, \"eval_ref\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb131... | {
"index": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb131072__v002",
"uid": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 912.506754823979, \"nll_base\": 6.816195487976074, \"eval_ref\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb262... | {
"index": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb262144__v000",
"uid": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 912.506754823979, \"nll_base\": 6.816195487976074, \"eval_ref\": \"global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb65536... | {
"index": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb65536__v001",
"uid": "global_mmlu_full_en_medical_genetics__1dcc1466__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_en_public_relations__fcf06fd7__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1421.1091589837463, \"nll_base\": 7.259192943572998, \"eval_ref\": \"global_mmlu_full_en_public_relations__fcf06fd7__s000__tb1... | {
"index": "global_mmlu_full_en_public_relations__fcf06fd7__s000__tb131072__v000",
"uid": "global_mmlu_full_en_public_relations__fcf06fd7__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1045.309786915157, \"nll_base\": 6.952068567276001, \"eval_ref\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__... | {
"index": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb131072__v002",
"uid": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1045.309786915157, \"nll_base\": 6.952068567276001, \"eval_ref\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__... | {
"index": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb262144__v000",
"uid": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1045.309786915157, \"nll_base\": 6.952068567276001, \"eval_ref\": \"global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb... | {
"index": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb65536__v001",
"uid": "global_mmlu_full_es_clinical_knowledge__e54858bb__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_formal_logic__065d2c00__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 583.757894079285, \"nll_base\": 6.369486331939697, \"eval_ref\": \"global_mmlu_full_es_formal_logic__065d2c00__s000__tb65536__v000\"... | {
"index": "global_mmlu_full_es_formal_logic__065d2c00__s000__tb65536__v000",
"uid": "global_mmlu_full_es_formal_logic__065d2c00__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_high_school_geography__1e9b7e72__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1407.4040427262591, \"nll_base\": 7.249502182006836, \"eval_ref\": \"global_mmlu_full_es_high_school_geography__1e9b7e72__s... | {
"index": "global_mmlu_full_es_high_school_geography__1e9b7e72__s000__tb65536__v000",
"uid": "global_mmlu_full_es_high_school_geography__1e9b7e72__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_high_school_macroeconomics__3e7421c6__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 838.1718125567312, \"nll_base\": 6.731223106384277, \"eval_ref\": \"global_mmlu_full_es_high_school_macroeconomics__... | {
"index": "global_mmlu_full_es_high_school_macroeconomics__3e7421c6__s000__tb262144__v000",
"uid": "global_mmlu_full_es_high_school_macroeconomics__3e7421c6__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_human_sexuality__cb1ac2e3__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1030.9886032380994, \"nll_base\": 6.9382734298706055, \"eval_ref\": \"global_mmlu_full_es_human_sexuality__cb1ac2e3__s000__tb26... | {
"index": "global_mmlu_full_es_human_sexuality__cb1ac2e3__s000__tb262144__v000",
"uid": "global_mmlu_full_es_human_sexuality__cb1ac2e3__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_logical_fallacies__510bf269__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 673.8274193978662, \"nll_base\": 6.51297402381897, \"eval_ref\": \"global_mmlu_full_es_logical_fallacies__510bf269__s000__tb1... | {
"index": "global_mmlu_full_es_logical_fallacies__510bf269__s000__tb131072__v000",
"uid": "global_mmlu_full_es_logical_fallacies__510bf269__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_management__c100168c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1241.8896501291692, \"nll_base\": 7.124389410018921, \"eval_ref\": \"global_mmlu_full_es_management__c100168c__s000__tb262144__v000\... | {
"index": "global_mmlu_full_es_management__c100168c__s000__tb262144__v000",
"uid": "global_mmlu_full_es_management__c100168c__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1243.4711824088254, \"nll_base\": 7.125662088394165, \"eval_ref\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb1... | {
"index": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb131072__v002",
"uid": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1243.4711824088254, \"nll_base\": 7.125662088394165, \"eval_ref\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb2... | {
"index": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb262144__v000",
"uid": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1243.4711824088254, \"nll_base\": 7.125662088394165, \"eval_ref\": \"global_mmlu_full_es_medical_genetics__00556fcf__s000__tb655... | {
"index": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb65536__v001",
"uid": "global_mmlu_full_es_medical_genetics__00556fcf__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_nutrition__109e6b1c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 828.8564643916623, \"nll_base\": 6.7200469970703125, \"eval_ref\": \"global_mmlu_full_es_nutrition__109e6b1c__s000__tb262144__v000\",... | {
"index": "global_mmlu_full_es_nutrition__109e6b1c__s000__tb262144__v000",
"uid": "global_mmlu_full_es_nutrition__109e6b1c__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_prehistory__f1329c20__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 508.1234016519944, \"nll_base\": 6.230724334716797, \"eval_ref\": \"global_mmlu_full_es_prehistory__f1329c20__s000__tb131072__v000\"... | {
"index": "global_mmlu_full_es_prehistory__f1329c20__s000__tb131072__v000",
"uid": "global_mmlu_full_es_prehistory__f1329c20__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_es_public_relations__0fa42132__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1154.8556453707752, \"nll_base\": 7.051730632781982, \"eval_ref\": \"global_mmlu_full_es_public_relations__0fa42132__s000__tb1... | {
"index": "global_mmlu_full_es_public_relations__0fa42132__s000__tb131072__v000",
"uid": "global_mmlu_full_es_public_relations__0fa42132__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_clinical_knowledge__94e45224__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 531.5541087676096, \"nll_base\": 6.2758049964904785, \"eval_ref\": \"global_mmlu_full_fa_clinical_knowledge__94e45224__s000__t... | {
"index": "global_mmlu_full_fa_clinical_knowledge__94e45224__s000__tb65536__v001",
"uid": "global_mmlu_full_fa_clinical_knowledge__94e45224__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_college_chemistry__febbd5cf__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 533.2142285156009, \"nll_base\": 6.278923273086548, \"eval_ref\": \"global_mmlu_full_fa_college_chemistry__febbd5cf__s000__tb... | {
"index": "global_mmlu_full_fa_college_chemistry__febbd5cf__s000__tb262144__v000",
"uid": "global_mmlu_full_fa_college_chemistry__febbd5cf__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_high_school_microeconomics__ceda58bf__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 362.86013535406937, \"nll_base\": 5.894017457962036, \"eval_ref\": \"global_mmlu_full_fa_high_school_microeconomics_... | {
"index": "global_mmlu_full_fa_high_school_microeconomics__ceda58bf__s000__tb131072__v000",
"uid": "global_mmlu_full_fa_high_school_microeconomics__ceda58bf__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_jurisprudence__c7f17db1__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 500.1691143766362, \"nll_base\": 6.214946269989014, \"eval_ref\": \"global_mmlu_full_fa_jurisprudence__c7f17db1__s000__tb131072__... | {
"index": "global_mmlu_full_fa_jurisprudence__c7f17db1__s000__tb131072__v000",
"uid": "global_mmlu_full_fa_jurisprudence__c7f17db1__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_machine_learning__62ffa6e1__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 358.2076384539313, \"nll_base\": 5.881112813949585, \"eval_ref\": \"global_mmlu_full_fa_machine_learning__62ffa6e1__s000__tb6553... | {
"index": "global_mmlu_full_fa_machine_learning__62ffa6e1__s000__tb65536__v000",
"uid": "global_mmlu_full_fa_machine_learning__62ffa6e1__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fa_medical_genetics__008c72f2__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 702.0207387287634, \"nll_base\": 6.55396294593811, \"eval_ref\": \"global_mmlu_full_fa_medical_genetics__008c72f2__s000__tb65536... | {
"index": "global_mmlu_full_fa_medical_genetics__008c72f2__s000__tb65536__v001",
"uid": "global_mmlu_full_fa_medical_genetics__008c72f2__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1189.7751322208637, \"nll_base\": 7.081519603729248, \"eval_ref\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s00... | {
"index": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb131072__v002",
"uid": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1189.7751322208637, \"nll_base\": 7.081519603729248, \"eval_ref\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s00... | {
"index": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb262144__v000",
"uid": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1189.7751322208637, \"nll_base\": 7.081519603729248, \"eval_ref\": \"global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000_... | {
"index": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb65536__v001",
"uid": "global_mmlu_full_fil_clinical_knowledge__a2c93d3c__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_global_facts__8ac6e866__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1848.4131729235714, \"nll_base\": 7.522082805633545, \"eval_ref\": \"global_mmlu_full_fil_global_facts__8ac6e866__s000__tb262144_... | {
"index": "global_mmlu_full_fil_global_facts__8ac6e866__s000__tb262144__v000",
"uid": "global_mmlu_full_fil_global_facts__8ac6e866__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_human_sexuality__35dfc9ab__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1775.3906472465796, \"nll_base\": 7.481775760650635, \"eval_ref\": \"global_mmlu_full_fil_human_sexuality__35dfc9ab__s000__tb655... | {
"index": "global_mmlu_full_fil_human_sexuality__35dfc9ab__s000__tb65536__v000",
"uid": "global_mmlu_full_fil_human_sexuality__35dfc9ab__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_logical_fallacies__49561e1c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1529.6112805151843, \"nll_base\": 7.33276891708374, \"eval_ref\": \"global_mmlu_full_fil_logical_fallacies__49561e1c__s000__... | {
"index": "global_mmlu_full_fil_logical_fallacies__49561e1c__s000__tb262144__v000",
"uid": "global_mmlu_full_fil_logical_fallacies__49561e1c__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fil_us_foreign_policy__b1ff9cd1__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1787.2720801314038, \"nll_base\": 7.48844575881958, \"eval_ref\": \"global_mmlu_full_fil_us_foreign_policy__b1ff9cd1__s000__... | {
"index": "global_mmlu_full_fil_us_foreign_policy__b1ff9cd1__s000__tb131072__v000",
"uid": "global_mmlu_full_fil_us_foreign_policy__b1ff9cd1__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 675.6115794414841, \"nll_base\": 6.515618324279785, \"eval_ref\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__... | {
"index": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb131072__v001",
"uid": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb131072__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 675.6115794414841, \"nll_base\": 6.515618324279785, \"eval_ref\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__... | {
"index": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb262144__v002",
"uid": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb262144__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 675.6115794414841, \"nll_base\": 6.515618324279785, \"eval_ref\": \"global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb... | {
"index": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb65536__v000",
"uid": "global_mmlu_full_fr_clinical_knowledge__06fb2ca8__s000__tb65536__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_college_medicine__7d1b5bdf__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 793.0928289038991, \"nll_base\": 6.675940275192261, \"eval_ref\": \"global_mmlu_full_fr_college_medicine__7d1b5bdf__s000__tb6553... | {
"index": "global_mmlu_full_fr_college_medicine__7d1b5bdf__s000__tb65536__v000",
"uid": "global_mmlu_full_fr_college_medicine__7d1b5bdf__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_college_physics__dda15cdc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 634.5925509796762, \"nll_base\": 6.452983140945435, \"eval_ref\": \"global_mmlu_full_fr_college_physics__dda15cdc__s000__tb65536_... | {
"index": "global_mmlu_full_fr_college_physics__dda15cdc__s000__tb65536__v000",
"uid": "global_mmlu_full_fr_college_physics__dda15cdc__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_conceptual_physics__e60ff7c1__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 781.9394109337986, \"nll_base\": 6.6617772579193115, \"eval_ref\": \"global_mmlu_full_fr_conceptual_physics__e60ff7c1__s000_... | {
"index": "global_mmlu_full_fr_conceptual_physics__e60ff7c1__s000__tb262144__v000",
"uid": "global_mmlu_full_fr_conceptual_physics__e60ff7c1__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_high_school_psychology__decd9f5b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 854.9616987438866, \"nll_base\": 6.751056671142578, \"eval_ref\": \"global_mmlu_full_fr_high_school_psychology__decd9f5b... | {
"index": "global_mmlu_full_fr_high_school_psychology__decd9f5b__s000__tb131072__v000",
"uid": "global_mmlu_full_fr_high_school_psychology__decd9f5b__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_high_school_statistics__50cfc6c8__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 768.5227877961776, \"nll_base\": 6.64447021484375, \"eval_ref\": \"global_mmlu_full_fr_high_school_statistics__50cfc6c8_... | {
"index": "global_mmlu_full_fr_high_school_statistics__50cfc6c8__s000__tb262144__v000",
"uid": "global_mmlu_full_fr_high_school_statistics__50cfc6c8__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_marketing__38b9ba25__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 643.5240991625191, \"nll_base\": 6.466959476470947, \"eval_ref\": \"global_mmlu_full_fr_marketing__38b9ba25__s000__tb65536__v000\", \"o... | {
"index": "global_mmlu_full_fr_marketing__38b9ba25__s000__tb65536__v000",
"uid": "global_mmlu_full_fr_marketing__38b9ba25__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 972.6837501653375, \"nll_base\": 6.880059003829956, \"eval_ref\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb13... | {
"index": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb131072__v000",
"uid": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb131072__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 972.6837501653375, \"nll_base\": 6.880059003829956, \"eval_ref\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb26... | {
"index": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb262144__v001",
"uid": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb262144__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 972.6837501653375, \"nll_base\": 6.880059003829956, \"eval_ref\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb6553... | {
"index": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb65536__v002",
"uid": "global_mmlu_full_fr_medical_genetics__30b6be47__s000__tb65536__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 863.457848519984, \"nll_base\": 6.760945081710815, \"eval_ref\": \"global_mmlu_full_fr_medical_genetics__30b6be47__s001__tb131... | {
"index": "global_mmlu_full_fr_medical_genetics__30b6be47__s001__tb131072__v003",
"uid": "global_mmlu_full_fr_medical_genetics__30b6be47__s001__tb131072__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_fr_moral_disputes__c6579b01__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 694.7069982580227, \"nll_base\": 6.543490171432495, \"eval_ref\": \"global_mmlu_full_fr_moral_disputes__c6579b01__s000__tb131072... | {
"index": "global_mmlu_full_fr_moral_disputes__c6579b01__s000__tb131072__v000",
"uid": "global_mmlu_full_fr_moral_disputes__c6579b01__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_college_mathematics__fd8c3a85__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 831.8178616713027, \"nll_base\": 6.723613500595093, \"eval_ref\": \"global_mmlu_full_ha_college_mathematics__fd8c3a85__s000... | {
"index": "global_mmlu_full_ha_college_mathematics__fd8c3a85__s000__tb262144__v000",
"uid": "global_mmlu_full_ha_college_mathematics__fd8c3a85__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_high_school_chemistry__d4668596__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 880.7057474369483, \"nll_base\": 6.780723571777344, \"eval_ref\": \"global_mmlu_full_ha_high_school_chemistry__d4668596__... | {
"index": "global_mmlu_full_ha_high_school_chemistry__d4668596__s000__tb262144__v000",
"uid": "global_mmlu_full_ha_high_school_chemistry__d4668596__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1462.5226860581145, \"nll_base\": 7.2879180908203125, \"eval_ref\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb... | {
"index": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb131072__v002",
"uid": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1462.5226860581145, \"nll_base\": 7.2879180908203125, \"eval_ref\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb... | {
"index": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb262144__v000",
"uid": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1462.5226860581145, \"nll_base\": 7.2879180908203125, \"eval_ref\": \"global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb65... | {
"index": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb65536__v001",
"uid": "global_mmlu_full_ha_medical_genetics__9cdaa3af__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_moral_disputes__fab5937e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1013.4652287330758, \"nll_base\": 6.921130657196045, \"eval_ref\": \"global_mmlu_full_ha_moral_disputes__fab5937e__s000__tb26214... | {
"index": "global_mmlu_full_ha_moral_disputes__fab5937e__s000__tb262144__v000",
"uid": "global_mmlu_full_ha_moral_disputes__fab5937e__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ha_philosophy__74fdb545__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 865.5620779361844, \"nll_base\": 6.763379096984863, \"eval_ref\": \"global_mmlu_full_ha_philosophy__74fdb545__s000__tb262144__v000\"... | {
"index": "global_mmlu_full_ha_philosophy__74fdb545__s000__tb262144__v000",
"uid": "global_mmlu_full_ha_philosophy__74fdb545__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_he_high_school_us_history__c26ed361__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 316.6040831688351, \"nll_base\": 5.757652044296265, \"eval_ref\": \"global_mmlu_full_he_high_school_us_history__c26ed361... | {
"index": "global_mmlu_full_he_high_school_us_history__c26ed361__s000__tb262144__v000",
"uid": "global_mmlu_full_he_high_school_us_history__c26ed361__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_abstract_algebra__617049eb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 879.0786736625216, \"nll_base\": 6.778874397277832, \"eval_ref\": \"global_mmlu_full_hi_abstract_algebra__617049eb__s000__tb6553... | {
"index": "global_mmlu_full_hi_abstract_algebra__617049eb__s000__tb65536__v000",
"uid": "global_mmlu_full_hi_abstract_algebra__617049eb__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 766.3266206994052, \"nll_base\": 6.641608476638794, \"eval_ref\": \"global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__... | {
"index": "global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb131072__v002",
"uid": "global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 766.3266206994052, \"nll_base\": 6.641608476638794, \"eval_ref\": \"global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb... | {
"index": "global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb65536__v001",
"uid": "global_mmlu_full_hi_clinical_knowledge__19f8f4c8__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_econometrics__d4248930__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 586.19651359671, \"nll_base\": 6.373655080795288, \"eval_ref\": \"global_mmlu_full_hi_econometrics__d4248930__s000__tb65536__v000\",... | {
"index": "global_mmlu_full_hi_econometrics__d4248930__s000__tb65536__v000",
"uid": "global_mmlu_full_hi_econometrics__d4248930__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_high_school_psychology__3744596a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 696.7700088143503, \"nll_base\": 6.546455383300781, \"eval_ref\": \"global_mmlu_full_hi_high_school_psychology__3744596a... | {
"index": "global_mmlu_full_hi_high_school_psychology__3744596a__s000__tb131072__v000",
"uid": "global_mmlu_full_hi_high_school_psychology__3744596a__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_international_law__05693e59__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 441.7703918976858, \"nll_base\": 6.090790271759033, \"eval_ref\": \"global_mmlu_full_hi_international_law__05693e59__s000__tb... | {
"index": "global_mmlu_full_hi_international_law__05693e59__s000__tb131072__v000",
"uid": "global_mmlu_full_hi_international_law__05693e59__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_logical_fallacies__3cffe8dd__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 558.7404333508514, \"nll_base\": 6.325685024261475, \"eval_ref\": \"global_mmlu_full_hi_logical_fallacies__3cffe8dd__s000__tb65... | {
"index": "global_mmlu_full_hi_logical_fallacies__3cffe8dd__s000__tb65536__v000",
"uid": "global_mmlu_full_hi_logical_fallacies__3cffe8dd__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1150.510729119424, \"nll_base\": 7.047961235046387, \"eval_ref\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb13... | {
"index": "global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb131072__v001",
"uid": "global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb131072__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1150.510729119424, \"nll_base\": 7.047961235046387, \"eval_ref\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb6553... | {
"index": "global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb65536__v000",
"uid": "global_mmlu_full_hi_medical_genetics__9b74f343__s000__tb65536__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 852.5884427423505, \"nll_base\": 6.748276948928833, \"eval_ref\": \"global_mmlu_full_hi_medical_genetics__9b74f343__s001__tb6553... | {
"index": "global_mmlu_full_hi_medical_genetics__9b74f343__s001__tb65536__v003",
"uid": "global_mmlu_full_hi_medical_genetics__9b74f343__s001__tb65536__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_miscellaneous__64a4cae2__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1214.7522174384346, \"nll_base\": 7.102295398712158, \"eval_ref\": \"global_mmlu_full_hi_miscellaneous__64a4cae2__s000__tb262144_... | {
"index": "global_mmlu_full_hi_miscellaneous__64a4cae2__s000__tb262144__v000",
"uid": "global_mmlu_full_hi_miscellaneous__64a4cae2__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_hi_public_relations__37dec4ea__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 743.5339539112779, \"nll_base\": 6.611414432525635, \"eval_ref\": \"global_mmlu_full_hi_public_relations__37dec4ea__s000__tb6553... | {
"index": "global_mmlu_full_hi_public_relations__37dec4ea__s000__tb65536__v000",
"uid": "global_mmlu_full_hi_public_relations__37dec4ea__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_anatomy__e1e402aa__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 955.5141884470511, \"nll_base\": 6.8622496128082275, \"eval_ref\": \"global_mmlu_full_id_anatomy__e1e402aa__s000__tb131072__v000\", \"o... | {
"index": "global_mmlu_full_id_anatomy__e1e402aa__s000__tb131072__v000",
"uid": "global_mmlu_full_id_anatomy__e1e402aa__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1727.4205536036302, \"nll_base\": 7.4543845653533936, \"eval_ref\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000... | {
"index": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb131072__v001",
"uid": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb131072__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1727.4205536036302, \"nll_base\": 7.4543845653533936, \"eval_ref\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000... | {
"index": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb262144__v002",
"uid": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb262144__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1727.4205536036302, \"nll_base\": 7.4543845653533936, \"eval_ref\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s000__... | {
"index": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb65536__v000",
"uid": "global_mmlu_full_id_clinical_knowledge__671e1794__s000__tb65536__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1290.0490651291082, \"nll_base\": 7.162435531616211, \"eval_ref\": \"global_mmlu_full_id_clinical_knowledge__671e1794__s001__t... | {
"index": "global_mmlu_full_id_clinical_knowledge__671e1794__s001__tb65536__v003",
"uid": "global_mmlu_full_id_clinical_knowledge__671e1794__s001__tb65536__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_college_mathematics__cab438ee__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1028.6367062053741, \"nll_base\": 6.935989618301392, \"eval_ref\": \"global_mmlu_full_id_college_mathematics__cab438ee__s000_... | {
"index": "global_mmlu_full_id_college_mathematics__cab438ee__s000__tb65536__v000",
"uid": "global_mmlu_full_id_college_mathematics__cab438ee__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_elementary_mathematics__4b5c657f__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1719.9899366829534, \"nll_base\": 7.450073719024658, \"eval_ref\": \"global_mmlu_full_id_elementary_mathematics__4b5c657... | {
"index": "global_mmlu_full_id_elementary_mathematics__4b5c657f__s000__tb262144__v000",
"uid": "global_mmlu_full_id_elementary_mathematics__4b5c657f__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_high_school_computer_science__6b27aebc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1075.326463517705, \"nll_base\": 6.980379581451416, \"eval_ref\": \"global_mmlu_full_id_high_school_computer_science... | {
"index": "global_mmlu_full_id_high_school_computer_science__6b27aebc__s000__tb65536__v000",
"uid": "global_mmlu_full_id_high_school_computer_science__6b27aebc__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_high_school_government_and_politics__d5cf65c2__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1482.5557443957916, \"nll_base\": 7.301522731781006, \"eval_ref\": \"global_mmlu_full_id_high_school_govern... | {
"index": "global_mmlu_full_id_high_school_government_and_politics__d5cf65c2__s000__tb262144__v000",
"uid": "global_mmlu_full_id_high_school_government_and_politics__d5cf65c2__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_high_school_mathematics__49034f7b__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1436.7579958198985, \"nll_base\": 7.270144462585449, \"eval_ref\": \"global_mmlu_full_id_high_school_mathematics__49034... | {
"index": "global_mmlu_full_id_high_school_mathematics__49034f7b__s000__tb262144__v000",
"uid": "global_mmlu_full_id_high_school_mathematics__49034f7b__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_machine_learning__e7495e6c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 812.7189315374304, \"nll_base\": 6.700385332107544, \"eval_ref\": \"global_mmlu_full_id_machine_learning__e7495e6c__s000__tb26... | {
"index": "global_mmlu_full_id_machine_learning__e7495e6c__s000__tb262144__v000",
"uid": "global_mmlu_full_id_machine_learning__e7495e6c__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_moral_scenarios__dd29b1b3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1198.7653509888885, \"nll_base\": 7.089047431945801, \"eval_ref\": \"global_mmlu_full_id_moral_scenarios__dd29b1b3__s000__tb131... | {
"index": "global_mmlu_full_id_moral_scenarios__dd29b1b3__s000__tb131072__v000",
"uid": "global_mmlu_full_id_moral_scenarios__dd29b1b3__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_professional_medicine__ea70c963__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1882.8648552351387, \"nll_base\": 7.5405497550964355, \"eval_ref\": \"global_mmlu_full_id_professional_medicine__ea70c963... | {
"index": "global_mmlu_full_id_professional_medicine__ea70c963__s000__tb131072__v000",
"uid": "global_mmlu_full_id_professional_medicine__ea70c963__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_id_security_studies__00065761__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1469.449427090177, \"nll_base\": 7.292643070220947, \"eval_ref\": \"global_mmlu_full_id_security_studies__00065761__s000__tb13... | {
"index": "global_mmlu_full_id_security_studies__00065761__s000__tb131072__v000",
"uid": "global_mmlu_full_id_security_studies__00065761__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 832.8964340555725, \"nll_base\": 6.72490930557251, \"eval_ref\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__t... | {
"index": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb131072__v002",
"uid": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb131072__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 832.8964340555725, \"nll_base\": 6.72490930557251, \"eval_ref\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__t... | {
"index": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb262144__v000",
"uid": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb262144__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 832.8964340555725, \"nll_base\": 6.72490930557251, \"eval_ref\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb6... | {
"index": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb65536__v001",
"uid": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s000__tb65536__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s001__tb262144__v003\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 893.5414449815302, \"nll_base\": 6.795192718505859, \"eval_ref\": \"global_mmlu_full_ig_clinical_knowledge__04c17c97__s001__... | {
"index": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s001__tb262144__v003",
"uid": "global_mmlu_full_ig_clinical_knowledge__04c17c97__s001__tb262144__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_elementary_mathematics__649acf2e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1237.8466285656998, \"nll_base\": 7.121128559112549, \"eval_ref\": \"global_mmlu_full_ig_elementary_mathematics__649acf2e_... | {
"index": "global_mmlu_full_ig_elementary_mathematics__649acf2e__s000__tb65536__v000",
"uid": "global_mmlu_full_ig_elementary_mathematics__649acf2e__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_formal_logic__fc43b09a__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 381.58842763882006, \"nll_base\": 5.944342613220215, \"eval_ref\": \"global_mmlu_full_ig_formal_logic__fc43b09a__s000__tb65536__v000... | {
"index": "global_mmlu_full_ig_formal_logic__fc43b09a__s000__tb65536__v000",
"uid": "global_mmlu_full_ig_formal_logic__fc43b09a__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_high_school_chemistry__bfb02173__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 919.6535478736565, \"nll_base\": 6.8239970207214355, \"eval_ref\": \"global_mmlu_full_ig_high_school_chemistry__bfb02173__s... | {
"index": "global_mmlu_full_ig_high_school_chemistry__bfb02173__s000__tb65536__v000",
"uid": "global_mmlu_full_ig_high_school_chemistry__bfb02173__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_mmlu_full_ig_high_school_statistics__52be17af__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 895.3094806550349, \"nll_base\": 6.79716944694519, \"eval_ref\": \"global_mmlu_full_ig_high_school_statistics__52be17af__s... | {
"index": "global_mmlu_full_ig_high_school_statistics__52be17af__s000__tb65536__v000",
"uid": "global_mmlu_full_ig_high_school_statistics__52be17af__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.