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\": \"mmmlu_de_de_business_ethics__49d7d8ac__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1349.4231832626406, \"nll_base\": 7.207432508468628, \"eval_ref\": \"mmmlu_de_de_business_ethics__49d7d8ac__s000__tb262144__v000\", \"o... | {
"index": "mmmlu_de_de_business_ethics__49d7d8ac__s000__tb262144__v000",
"uid": "mmmlu_de_de_business_ethics__49d7d8ac__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\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1142.0394957640103, \"nll_base\": 7.040570974349976, \"eval_ref\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb131072__v000\... | {
"index": "mmmlu_de_de_clinical_knowledge__3c063000__s000__tb131072__v000",
"uid": "mmmlu_de_de_clinical_knowledge__3c063000__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\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1142.0394957640103, \"nll_base\": 7.040570974349976, \"eval_ref\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb262144__v001\... | {
"index": "mmmlu_de_de_clinical_knowledge__3c063000__s000__tb262144__v001",
"uid": "mmmlu_de_de_clinical_knowledge__3c063000__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\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1142.0394957640103, \"nll_base\": 7.040570974349976, \"eval_ref\": \"mmmlu_de_de_clinical_knowledge__3c063000__s000__tb65536__v002\", ... | {
"index": "mmmlu_de_de_clinical_knowledge__3c063000__s000__tb65536__v002",
"uid": "mmmlu_de_de_clinical_knowledge__3c063000__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\": \"mmmlu_de_de_electrical_engineering__028d8e42__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1034.623596769837, \"nll_base\": 6.941792964935303, \"eval_ref\": \"mmmlu_de_de_electrical_engineering__028d8e42__s000__tb65536__v... | {
"index": "mmmlu_de_de_electrical_engineering__028d8e42__s000__tb65536__v000",
"uid": "mmmlu_de_de_electrical_engineering__028d8e42__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\": \"mmmlu_de_de_elementary_mathematics__d3fcd457__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2658.7548828899453, \"nll_base\": 7.885613203048706, \"eval_ref\": \"mmmlu_de_de_elementary_mathematics__d3fcd457__s000__tb65536__... | {
"index": "mmmlu_de_de_elementary_mathematics__d3fcd457__s000__tb65536__v000",
"uid": "mmmlu_de_de_elementary_mathematics__d3fcd457__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\": \"mmmlu_de_de_high_school_biology__79eb87d8__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1275.0140665014014, \"nll_base\": 7.150712490081787, \"eval_ref\": \"mmmlu_de_de_high_school_biology__79eb87d8__s000__tb262144__v00... | {
"index": "mmmlu_de_de_high_school_biology__79eb87d8__s000__tb262144__v000",
"uid": "mmmlu_de_de_high_school_biology__79eb87d8__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\": \"mmmlu_de_de_high_school_psychology__61aa72aa__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1484.5371997459713, \"nll_base\": 7.302858352661133, \"eval_ref\": \"mmmlu_de_de_high_school_psychology__61aa72aa__s000__tb65536__... | {
"index": "mmmlu_de_de_high_school_psychology__61aa72aa__s000__tb65536__v000",
"uid": "mmmlu_de_de_high_school_psychology__61aa72aa__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\": \"mmmlu_de_de_machine_learning__889bc414__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 956.9639457996631, \"nll_base\": 6.863765716552734, \"eval_ref\": \"mmmlu_de_de_machine_learning__889bc414__s000__tb131072__v000\", \"... | {
"index": "mmmlu_de_de_machine_learning__889bc414__s000__tb131072__v000",
"uid": "mmmlu_de_de_machine_learning__889bc414__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\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1609.8339842358191, \"nll_base\": 7.383886337280273, \"eval_ref\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb131072__v000\", \... | {
"index": "mmmlu_de_de_medical_genetics__997f584d__s000__tb131072__v000",
"uid": "mmmlu_de_de_medical_genetics__997f584d__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\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1609.8339842358191, \"nll_base\": 7.383886337280273, \"eval_ref\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb262144__v001\", \... | {
"index": "mmmlu_de_de_medical_genetics__997f584d__s000__tb262144__v001",
"uid": "mmmlu_de_de_medical_genetics__997f584d__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\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1609.8339842358191, \"nll_base\": 7.383886337280273, \"eval_ref\": \"mmmlu_de_de_medical_genetics__997f584d__s000__tb65536__v002\", \"or... | {
"index": "mmmlu_de_de_medical_genetics__997f584d__s000__tb65536__v002",
"uid": "mmmlu_de_de_medical_genetics__997f584d__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\": \"mmmlu_de_de_professional_medicine__fecaf2e9__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1376.8251148839247, \"nll_base\": 7.2275354862213135, \"eval_ref\": \"mmmlu_de_de_professional_medicine__fecaf2e9__s000__tb262144... | {
"index": "mmmlu_de_de_professional_medicine__fecaf2e9__s000__tb262144__v000",
"uid": "mmmlu_de_de_professional_medicine__fecaf2e9__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\": \"mmmlu_de_de_security_studies__113d4965__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1641.372606569487, \"nll_base\": 7.403288125991821, \"eval_ref\": \"mmmlu_de_de_security_studies__113d4965__s000__tb262144__v000\", \"... | {
"index": "mmmlu_de_de_security_studies__113d4965__s000__tb262144__v000",
"uid": "mmmlu_de_de_security_studies__113d4965__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\": \"mmmlu_de_de_us_foreign_policy__08b21b71__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1363.4046879073767, \"nll_base\": 7.217740297317505, \"eval_ref\": \"mmmlu_de_de_us_foreign_policy__08b21b71__s000__tb131072__v000\",... | {
"index": "mmmlu_de_de_us_foreign_policy__08b21b71__s000__tb131072__v000",
"uid": "mmmlu_de_de_us_foreign_policy__08b21b71__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\": \"mmmlu_es_la_abstract_algebra__7c04648b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 847.0199539023289, \"nll_base\": 6.741724252700806, \"eval_ref\": \"mmmlu_es_la_abstract_algebra__7c04648b__s000__tb131072__v000\", \"... | {
"index": "mmmlu_es_la_abstract_algebra__7c04648b__s000__tb131072__v000",
"uid": "mmmlu_es_la_abstract_algebra__7c04648b__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\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 751.7236822136956, \"nll_base\": 6.622368812561035, \"eval_ref\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb131072__v001\"... | {
"index": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb131072__v001",
"uid": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__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\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 751.7236822136956, \"nll_base\": 6.622368812561035, \"eval_ref\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb262144__v002\"... | {
"index": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb262144__v002",
"uid": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__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\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 751.7236822136956, \"nll_base\": 6.622368812561035, \"eval_ref\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb65536__v000\", \... | {
"index": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__s000__tb65536__v000",
"uid": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__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\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 809.3676790947818, \"nll_base\": 6.696253299713135, \"eval_ref\": \"mmmlu_es_la_clinical_knowledge__a5cd6a8b__s001__tb65536__v003\", \... | {
"index": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__s001__tb65536__v003",
"uid": "mmmlu_es_la_clinical_knowledge__a5cd6a8b__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\": \"mmmlu_es_la_college_chemistry__a1eb14df__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1217.9067872812984, \"nll_base\": 7.104888916015625, \"eval_ref\": \"mmmlu_es_la_college_chemistry__a1eb14df__s000__tb262144__v000\",... | {
"index": "mmmlu_es_la_college_chemistry__a1eb14df__s000__tb262144__v000",
"uid": "mmmlu_es_la_college_chemistry__a1eb14df__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\": \"mmmlu_es_la_electrical_engineering__9e0ba8a9__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 601.0701689412074, \"nll_base\": 6.398711681365967, \"eval_ref\": \"mmmlu_es_la_electrical_engineering__9e0ba8a9__s000__tb65536__v... | {
"index": "mmmlu_es_la_electrical_engineering__9e0ba8a9__s000__tb65536__v000",
"uid": "mmmlu_es_la_electrical_engineering__9e0ba8a9__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\": \"mmmlu_es_la_high_school_computer_science__a1750518__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 684.6581734207656, \"nll_base\": 6.528919696807861, \"eval_ref\": \"mmmlu_es_la_high_school_computer_science__a1750518__s0... | {
"index": "mmmlu_es_la_high_school_computer_science__a1750518__s000__tb131072__v000",
"uid": "mmmlu_es_la_high_school_computer_science__a1750518__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\": \"mmmlu_es_la_professional_accounting__2a864846__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 608.4364601846976, \"nll_base\": 6.410892486572266, \"eval_ref\": \"mmmlu_es_la_professional_accounting__2a864846__s000__tb1310... | {
"index": "mmmlu_es_la_professional_accounting__2a864846__s000__tb131072__v000",
"uid": "mmmlu_es_la_professional_accounting__2a864846__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\": \"mmmlu_fr_fr_college_computer_science__60940368__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 823.8918948351037, \"nll_base\": 6.714039325714111, \"eval_ref\": \"mmmlu_fr_fr_college_computer_science__60940368__s000__tb6553... | {
"index": "mmmlu_fr_fr_college_computer_science__60940368__s000__tb65536__v000",
"uid": "mmmlu_fr_fr_college_computer_science__60940368__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\": \"mmmlu_fr_fr_conceptual_physics__c4aa6de8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 670.2452563424704, \"nll_base\": 6.507643699645996, \"eval_ref\": \"mmmlu_fr_fr_conceptual_physics__c4aa6de8__s000__tb65536__v000\", \... | {
"index": "mmmlu_fr_fr_conceptual_physics__c4aa6de8__s000__tb65536__v000",
"uid": "mmmlu_fr_fr_conceptual_physics__c4aa6de8__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\": \"mmmlu_fr_fr_econometrics__3a9e4e71__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 617.7978430975339, \"nll_base\": 6.426161289215088, \"eval_ref\": \"mmmlu_fr_fr_econometrics__3a9e4e71__s000__tb262144__v000\", \"oracle_d... | {
"index": "mmmlu_fr_fr_econometrics__3a9e4e71__s000__tb262144__v000",
"uid": "mmmlu_fr_fr_econometrics__3a9e4e71__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\": \"mmmlu_fr_fr_high_school_biology__09b166f3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 764.5629776095145, \"nll_base\": 6.6393043994903564, \"eval_ref\": \"mmmlu_fr_fr_high_school_biology__09b166f3__s000__tb131072__v00... | {
"index": "mmmlu_fr_fr_high_school_biology__09b166f3__s000__tb131072__v000",
"uid": "mmmlu_fr_fr_high_school_biology__09b166f3__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\": \"mmmlu_fr_fr_high_school_macroeconomics__86ef6641__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 738.2808744054864, \"nll_base\": 6.6043243408203125, \"eval_ref\": \"mmmlu_fr_fr_high_school_macroeconomics__86ef6641__s000_... | {
"index": "mmmlu_fr_fr_high_school_macroeconomics__86ef6641__s000__tb131072__v000",
"uid": "mmmlu_fr_fr_high_school_macroeconomics__86ef6641__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\": \"mmmlu_fr_fr_human_aging__cb4a4703__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 842.5624727645267, \"nll_base\": 6.736447811126709, \"eval_ref\": \"mmmlu_fr_fr_human_aging__cb4a4703__s000__tb131072__v000\", \"oracle_del... | {
"index": "mmmlu_fr_fr_human_aging__cb4a4703__s000__tb131072__v000",
"uid": "mmmlu_fr_fr_human_aging__cb4a4703__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\": \"mmmlu_fr_fr_human_sexuality__2ef1f3f3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1286.9082819537775, \"nll_base\": 7.159997940063477, \"eval_ref\": \"mmmlu_fr_fr_human_sexuality__2ef1f3f3__s000__tb65536__v000\", \"orac... | {
"index": "mmmlu_fr_fr_human_sexuality__2ef1f3f3__s000__tb65536__v000",
"uid": "mmmlu_fr_fr_human_sexuality__2ef1f3f3__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\": \"mmmlu_fr_fr_moral_disputes__8be110c0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 735.7059798115142, \"nll_base\": 6.600830554962158, \"eval_ref\": \"mmmlu_fr_fr_moral_disputes__8be110c0__s000__tb65536__v000\", \"oracle_... | {
"index": "mmmlu_fr_fr_moral_disputes__8be110c0__s000__tb65536__v000",
"uid": "mmmlu_fr_fr_moral_disputes__8be110c0__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\": \"mmmlu_fr_fr_nutrition__ae1615cc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 804.5409404120927, \"nll_base\": 6.690271854400635, \"eval_ref\": \"mmmlu_fr_fr_nutrition__ae1615cc__s000__tb65536__v000\", \"oracle_delta\": 0... | {
"index": "mmmlu_fr_fr_nutrition__ae1615cc__s000__tb65536__v000",
"uid": "mmmlu_fr_fr_nutrition__ae1615cc__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\": \"mmmlu_fr_fr_security_studies__90b11dd5__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 867.1161698236838, \"nll_base\": 6.765172958374023, \"eval_ref\": \"mmmlu_fr_fr_security_studies__90b11dd5__s000__tb131072__v000\", \"... | {
"index": "mmmlu_fr_fr_security_studies__90b11dd5__s000__tb131072__v000",
"uid": "mmmlu_fr_fr_security_studies__90b11dd5__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\": \"mmmlu_fr_fr_sociology__8e52f3e7__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 724.0176283111261, \"nll_base\": 6.584815740585327, \"eval_ref\": \"mmmlu_fr_fr_sociology__8e52f3e7__s000__tb131072__v000\", \"oracle_delta\"... | {
"index": "mmmlu_fr_fr_sociology__8e52f3e7__s000__tb131072__v000",
"uid": "mmmlu_fr_fr_sociology__8e52f3e7__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\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 864.1106853186026, \"nll_base\": 6.761700868606567, \"eval_ref\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb131072__v000\"... | {
"index": "mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb131072__v000",
"uid": "mmmlu_hi_in_clinical_knowledge__9df5f582__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\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 864.1106853186026, \"nll_base\": 6.761700868606567, \"eval_ref\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb65536__v002\", \... | {
"index": "mmmlu_hi_in_clinical_knowledge__9df5f582__s000__tb65536__v002",
"uid": "mmmlu_hi_in_clinical_knowledge__9df5f582__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\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 672.2356250052737, \"nll_base\": 6.510608911514282, \"eval_ref\": \"mmmlu_hi_in_clinical_knowledge__9df5f582__s001__tb131072__v003\"... | {
"index": "mmmlu_hi_in_clinical_knowledge__9df5f582__s001__tb131072__v003",
"uid": "mmmlu_hi_in_clinical_knowledge__9df5f582__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\": \"mmmlu_hi_in_college_computer_science__8f29240c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 488.08253652398844, \"nll_base\": 6.190484523773193, \"eval_ref\": \"mmmlu_hi_in_college_computer_science__8f29240c__s000__tb655... | {
"index": "mmmlu_hi_in_college_computer_science__8f29240c__s000__tb65536__v000",
"uid": "mmmlu_hi_in_college_computer_science__8f29240c__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\": \"mmmlu_hi_in_computer_security__631a658b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 983.7250670627716, \"nll_base\": 6.891346454620361, \"eval_ref\": \"mmmlu_hi_in_computer_security__631a658b__s000__tb131072__v000\", ... | {
"index": "mmmlu_hi_in_computer_security__631a658b__s000__tb131072__v000",
"uid": "mmmlu_hi_in_computer_security__631a658b__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\": \"mmmlu_hi_in_high_school_microeconomics__bfa6d3c4__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 692.4717669608627, \"nll_base\": 6.540267467498779, \"eval_ref\": \"mmmlu_hi_in_high_school_microeconomics__bfa6d3c4__s000__tb... | {
"index": "mmmlu_hi_in_high_school_microeconomics__bfa6d3c4__s000__tb65536__v000",
"uid": "mmmlu_hi_in_high_school_microeconomics__bfa6d3c4__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\": \"mmmlu_hi_in_logical_fallacies__8a481c96__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 639.8749179324976, \"nll_base\": 6.461272716522217, \"eval_ref\": \"mmmlu_hi_in_logical_fallacies__8a481c96__s000__tb65536__v000\", \"o... | {
"index": "mmmlu_hi_in_logical_fallacies__8a481c96__s000__tb65536__v000",
"uid": "mmmlu_hi_in_logical_fallacies__8a481c96__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\": \"mmmlu_hi_in_professional_psychology__448e2360__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 523.9144767996185, \"nll_base\": 6.261328458786011, \"eval_ref\": \"mmmlu_hi_in_professional_psychology__448e2360__s000__tb1310... | {
"index": "mmmlu_hi_in_professional_psychology__448e2360__s000__tb131072__v000",
"uid": "mmmlu_hi_in_professional_psychology__448e2360__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\": \"mmmlu_hi_in_virology__c04597c3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 658.5250965439201, \"nll_base\": 6.490002632141113, \"eval_ref\": \"mmmlu_hi_in_virology__c04597c3__s000__tb65536__v000\", \"oracle_delta\": 0.1... | {
"index": "mmmlu_hi_in_virology__c04597c3__s000__tb65536__v000",
"uid": "mmmlu_hi_in_virology__c04597c3__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\": \"mmmlu_id_id_business_ethics__677ad76e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1012.0377247326161, \"nll_base\": 6.9197211265563965, \"eval_ref\": \"mmmlu_id_id_business_ethics__677ad76e__s000__tb262144__v000\", \"... | {
"index": "mmmlu_id_id_business_ethics__677ad76e__s000__tb262144__v000",
"uid": "mmmlu_id_id_business_ethics__677ad76e__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\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1231.2904240728465, \"nll_base\": 7.115818023681641, \"eval_ref\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb131072__v001\... | {
"index": "mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb131072__v001",
"uid": "mmmlu_id_id_clinical_knowledge__c5eda9c3__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\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1231.2904240728465, \"nll_base\": 7.115818023681641, \"eval_ref\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb262144__v002\... | {
"index": "mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb262144__v002",
"uid": "mmmlu_id_id_clinical_knowledge__c5eda9c3__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\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1231.2904240728465, \"nll_base\": 7.115818023681641, \"eval_ref\": \"mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb65536__v000\", ... | {
"index": "mmmlu_id_id_clinical_knowledge__c5eda9c3__s000__tb65536__v000",
"uid": "mmmlu_id_id_clinical_knowledge__c5eda9c3__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\": \"mmmlu_id_id_college_medicine__2718ac60__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1327.7869937493447, \"nll_base\": 7.1912689208984375, \"eval_ref\": \"mmmlu_id_id_college_medicine__2718ac60__s000__tb65536__v000\", \"o... | {
"index": "mmmlu_id_id_college_medicine__2718ac60__s000__tb65536__v000",
"uid": "mmmlu_id_id_college_medicine__2718ac60__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\": \"mmmlu_id_id_high_school_government_and_politics__228dba28__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1462.7835313669946, \"nll_base\": 7.2880964279174805, \"eval_ref\": \"mmmlu_id_id_high_school_government_and_politi... | {
"index": "mmmlu_id_id_high_school_government_and_politics__228dba28__s000__tb131072__v000",
"uid": "mmmlu_id_id_high_school_government_and_politics__228dba28__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\": \"mmmlu_id_id_high_school_macroeconomics__7982d68b__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1555.689269350241, \"nll_base\": 7.3496739864349365, \"eval_ref\": \"mmmlu_id_id_high_school_macroeconomics__7982d68b__s000__t... | {
"index": "mmmlu_id_id_high_school_macroeconomics__7982d68b__s000__tb65536__v000",
"uid": "mmmlu_id_id_high_school_macroeconomics__7982d68b__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\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1479.004461533021, \"nll_base\": 7.299124479293823, \"eval_ref\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb131072__v002\", \"... | {
"index": "mmmlu_id_id_medical_genetics__c889cca1__s000__tb131072__v002",
"uid": "mmmlu_id_id_medical_genetics__c889cca1__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\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1479.004461533021, \"nll_base\": 7.299124479293823, \"eval_ref\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb262144__v000\", \"... | {
"index": "mmmlu_id_id_medical_genetics__c889cca1__s000__tb262144__v000",
"uid": "mmmlu_id_id_medical_genetics__c889cca1__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\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1479.004461533021, \"nll_base\": 7.299124479293823, \"eval_ref\": \"mmmlu_id_id_medical_genetics__c889cca1__s000__tb65536__v001\", \"ora... | {
"index": "mmmlu_id_id_medical_genetics__c889cca1__s000__tb65536__v001",
"uid": "mmmlu_id_id_medical_genetics__c889cca1__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\": \"mmmlu_id_id_medical_genetics__c889cca1__s001__tb262144__v003\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1106.45909702215, \"nll_base\": 7.008920192718506, \"eval_ref\": \"mmmlu_id_id_medical_genetics__c889cca1__s001__tb262144__v003\", \"o... | {
"index": "mmmlu_id_id_medical_genetics__c889cca1__s001__tb262144__v003",
"uid": "mmmlu_id_id_medical_genetics__c889cca1__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\": \"mmmlu_id_id_nutrition__c8424460__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1377.3038021608384, \"nll_base\": 7.2278831005096436, \"eval_ref\": \"mmmlu_id_id_nutrition__c8424460__s000__tb65536__v000\", \"oracle_delta\":... | {
"index": "mmmlu_id_id_nutrition__c8424460__s000__tb65536__v000",
"uid": "mmmlu_id_id_nutrition__c8424460__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\": \"mmmlu_id_id_professional_accounting__ac097366__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1231.1782882992502, \"nll_base\": 7.115726947784424, \"eval_ref\": \"mmmlu_id_id_professional_accounting__ac097366__s000__tb262... | {
"index": "mmmlu_id_id_professional_accounting__ac097366__s000__tb262144__v000",
"uid": "mmmlu_id_id_professional_accounting__ac097366__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\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 890.0483077908044, \"nll_base\": 6.7912757396698, \"eval_ref\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb131072__v002\", ... | {
"index": "mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb131072__v002",
"uid": "mmmlu_it_it_clinical_knowledge__bc6a44c1__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\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 890.0483077908044, \"nll_base\": 6.7912757396698, \"eval_ref\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb262144__v000\", ... | {
"index": "mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb262144__v000",
"uid": "mmmlu_it_it_clinical_knowledge__bc6a44c1__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\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 890.0483077908044, \"nll_base\": 6.7912757396698, \"eval_ref\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb65536__v001\", \"o... | {
"index": "mmmlu_it_it_clinical_knowledge__bc6a44c1__s000__tb65536__v001",
"uid": "mmmlu_it_it_clinical_knowledge__bc6a44c1__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\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s001__tb262144__v003\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1026.9720977800912, \"nll_base\": 6.934370040893555, \"eval_ref\": \"mmmlu_it_it_clinical_knowledge__bc6a44c1__s001__tb262144__v003\... | {
"index": "mmmlu_it_it_clinical_knowledge__bc6a44c1__s001__tb262144__v003",
"uid": "mmmlu_it_it_clinical_knowledge__bc6a44c1__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\": \"mmmlu_it_it_college_biology__a7483b3a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 779.0560247073515, \"nll_base\": 6.658082962036133, \"eval_ref\": \"mmmlu_it_it_college_biology__a7483b3a__s000__tb262144__v000\", \"or... | {
"index": "mmmlu_it_it_college_biology__a7483b3a__s000__tb262144__v000",
"uid": "mmmlu_it_it_college_biology__a7483b3a__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\": \"mmmlu_it_it_college_medicine__9981669f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 727.8970727510115, \"nll_base\": 6.59015965461731, \"eval_ref\": \"mmmlu_it_it_college_medicine__9981669f__s000__tb65536__v000\", \"orac... | {
"index": "mmmlu_it_it_college_medicine__9981669f__s000__tb65536__v000",
"uid": "mmmlu_it_it_college_medicine__9981669f__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\": \"mmmlu_it_it_conceptual_physics__2fc63319__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 744.3583723159252, \"nll_base\": 6.612522602081299, \"eval_ref\": \"mmmlu_it_it_conceptual_physics__2fc63319__s000__tb262144__v000\"... | {
"index": "mmmlu_it_it_conceptual_physics__2fc63319__s000__tb262144__v000",
"uid": "mmmlu_it_it_conceptual_physics__2fc63319__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\": \"mmmlu_it_it_econometrics__8cc8314e__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 783.4897970774687, \"nll_base\": 6.663758039474487, \"eval_ref\": \"mmmlu_it_it_econometrics__8cc8314e__s000__tb131072__v000\", \"oracle_d... | {
"index": "mmmlu_it_it_econometrics__8cc8314e__s000__tb131072__v000",
"uid": "mmmlu_it_it_econometrics__8cc8314e__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\": \"mmmlu_it_it_high_school_computer_science__0735ec89__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 800.0467871414676, \"nll_base\": 6.6846702098846436, \"eval_ref\": \"mmmlu_it_it_high_school_computer_science__0735ec89__s... | {
"index": "mmmlu_it_it_high_school_computer_science__0735ec89__s000__tb262144__v000",
"uid": "mmmlu_it_it_high_school_computer_science__0735ec89__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\": \"mmmlu_it_it_high_school_physics__bdb28c94__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 708.2833860753534, \"nll_base\": 6.562844276428223, \"eval_ref\": \"mmmlu_it_it_high_school_physics__bdb28c94__s000__tb262144__v000... | {
"index": "mmmlu_it_it_high_school_physics__bdb28c94__s000__tb262144__v000",
"uid": "mmmlu_it_it_high_school_physics__bdb28c94__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\": \"mmmlu_it_it_high_school_statistics__f82038f9__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 977.999125351756, \"nll_base\": 6.88550877571106, \"eval_ref\": \"mmmlu_it_it_high_school_statistics__f82038f9__s000__tb262144__... | {
"index": "mmmlu_it_it_high_school_statistics__f82038f9__s000__tb262144__v000",
"uid": "mmmlu_it_it_high_school_statistics__f82038f9__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\": \"mmmlu_it_it_philosophy__b97ebe09__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 776.8906940269292, \"nll_base\": 6.655299663543701, \"eval_ref\": \"mmmlu_it_it_philosophy__b97ebe09__s000__tb131072__v000\", \"oracle_delta... | {
"index": "mmmlu_it_it_philosophy__b97ebe09__s000__tb131072__v000",
"uid": "mmmlu_it_it_philosophy__b97ebe09__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\": \"mmmlu_it_it_professional_accounting__14c45364__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 682.232886033485, \"nll_base\": 6.525371074676514, \"eval_ref\": \"mmmlu_it_it_professional_accounting__14c45364__s000__tb26214... | {
"index": "mmmlu_it_it_professional_accounting__14c45364__s000__tb262144__v000",
"uid": "mmmlu_it_it_professional_accounting__14c45364__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\": \"mmmlu_it_it_professional_law__dbbf01c3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 431.8216753086253, \"nll_base\": 6.068012714385986, \"eval_ref\": \"mmmlu_it_it_professional_law__dbbf01c3__s000__tb131072__v000\", \"... | {
"index": "mmmlu_it_it_professional_law__dbbf01c3__s000__tb131072__v000",
"uid": "mmmlu_it_it_professional_law__dbbf01c3__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\": \"mmmlu_it_it_virology__fb4dc921__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 858.7957116890503, \"nll_base\": 6.755531072616577, \"eval_ref\": \"mmmlu_it_it_virology__fb4dc921__s000__tb262144__v000\", \"oracle_delta\": ... | {
"index": "mmmlu_it_it_virology__fb4dc921__s000__tb262144__v000",
"uid": "mmmlu_it_it_virology__fb4dc921__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\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1090.0805506559761, \"nll_base\": 6.994006872177124, \"eval_ref\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb131072__v001\... | {
"index": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb131072__v001",
"uid": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__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\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1090.0805506559761, \"nll_base\": 6.994006872177124, \"eval_ref\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb262144__v002\... | {
"index": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb262144__v002",
"uid": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__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\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1090.0805506559761, \"nll_base\": 6.994006872177124, \"eval_ref\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb65536__v000\", ... | {
"index": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s000__tb65536__v000",
"uid": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__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\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1192.439735547466, \"nll_base\": 7.083756685256958, \"eval_ref\": \"mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s001__tb65536__v003\", \... | {
"index": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__s001__tb65536__v003",
"uid": "mmmlu_ja_jp_clinical_knowledge__7e6eeceb__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\": \"mmmlu_ja_jp_medical_genetics__c6df534c__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 788.1331167991171, \"nll_base\": 6.66966700553894, \"eval_ref\": \"mmmlu_ja_jp_medical_genetics__c6df534c__s000__tb131072__v002\", \"o... | {
"index": "mmmlu_ja_jp_medical_genetics__c6df534c__s000__tb131072__v002",
"uid": "mmmlu_ja_jp_medical_genetics__c6df534c__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\": \"mmmlu_ja_jp_public_relations__13c93233__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 917.9880051635895, \"nll_base\": 6.822184324264526, \"eval_ref\": \"mmmlu_ja_jp_public_relations__13c93233__s000__tb131072__v000\", \"... | {
"index": "mmmlu_ja_jp_public_relations__13c93233__s000__tb131072__v000",
"uid": "mmmlu_ja_jp_public_relations__13c93233__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\": \"mmmlu_ko_kr_clinical_knowledge__7470b193__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 853.8490560951991, \"nll_base\": 6.749754428863525, \"eval_ref\": \"mmmlu_ko_kr_clinical_knowledge__7470b193__s000__tb65536__v001\", \... | {
"index": "mmmlu_ko_kr_clinical_knowledge__7470b193__s000__tb65536__v001",
"uid": "mmmlu_ko_kr_clinical_knowledge__7470b193__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\": \"mmmlu_ko_kr_college_computer_science__916e0d54__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 811.9637839072934, \"nll_base\": 6.699455738067627, \"eval_ref\": \"mmmlu_ko_kr_college_computer_science__916e0d54__s000__tb13... | {
"index": "mmmlu_ko_kr_college_computer_science__916e0d54__s000__tb131072__v000",
"uid": "mmmlu_ko_kr_college_computer_science__916e0d54__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\": \"mmmlu_ko_kr_elementary_mathematics__6a724bde__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2220.404297677337, \"nll_base\": 7.705444574356079, \"eval_ref\": \"mmmlu_ko_kr_elementary_mathematics__6a724bde__s000__tb131072... | {
"index": "mmmlu_ko_kr_elementary_mathematics__6a724bde__s000__tb131072__v000",
"uid": "mmmlu_ko_kr_elementary_mathematics__6a724bde__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\": \"mmmlu_ko_kr_high_school_computer_science__939f2847__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 729.7509725410417, \"nll_base\": 6.592703342437744, \"eval_ref\": \"mmmlu_ko_kr_high_school_computer_science__939f2847__s000... | {
"index": "mmmlu_ko_kr_high_school_computer_science__939f2847__s000__tb65536__v000",
"uid": "mmmlu_ko_kr_high_school_computer_science__939f2847__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\": \"mmmlu_ko_kr_moral_scenarios__373b2b4b__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1759.781574441107, \"nll_base\": 7.472944974899292, \"eval_ref\": \"mmmlu_ko_kr_moral_scenarios__373b2b4b__s000__tb65536__v000\", \"oracl... | {
"index": "mmmlu_ko_kr_moral_scenarios__373b2b4b__s000__tb65536__v000",
"uid": "mmmlu_ko_kr_moral_scenarios__373b2b4b__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\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 848.8475081189536, \"nll_base\": 6.743879556655884, \"eval_ref\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb131072__v001\"... | {
"index": "mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb131072__v001",
"uid": "mmmlu_pt_br_clinical_knowledge__0217fd19__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\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 848.8475081189536, \"nll_base\": 6.743879556655884, \"eval_ref\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb262144__v002\"... | {
"index": "mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb262144__v002",
"uid": "mmmlu_pt_br_clinical_knowledge__0217fd19__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\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 848.8475081189536, \"nll_base\": 6.743879556655884, \"eval_ref\": \"mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb65536__v000\", \... | {
"index": "mmmlu_pt_br_clinical_knowledge__0217fd19__s000__tb65536__v000",
"uid": "mmmlu_pt_br_clinical_knowledge__0217fd19__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\": \"mmmlu_pt_br_computer_security__807bedf3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 615.1751032971408, \"nll_base\": 6.4219069480896, \"eval_ref\": \"mmmlu_pt_br_computer_security__807bedf3__s000__tb131072__v000\", \"... | {
"index": "mmmlu_pt_br_computer_security__807bedf3__s000__tb131072__v000",
"uid": "mmmlu_pt_br_computer_security__807bedf3__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\": \"mmmlu_pt_br_machine_learning__a3fc5ab2__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 790.7962822350811, \"nll_base\": 6.673040390014648, \"eval_ref\": \"mmmlu_pt_br_machine_learning__a3fc5ab2__s000__tb131072__v000\", \"... | {
"index": "mmmlu_pt_br_machine_learning__a3fc5ab2__s000__tb131072__v000",
"uid": "mmmlu_pt_br_machine_learning__a3fc5ab2__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\": \"mmmlu_sw_ke_abstract_algebra__c99c716a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 903.6014399324415, \"nll_base\": 6.8063883781433105, \"eval_ref\": \"mmmlu_sw_ke_abstract_algebra__c99c716a__s000__tb131072__v000\", \... | {
"index": "mmmlu_sw_ke_abstract_algebra__c99c716a__s000__tb131072__v000",
"uid": "mmmlu_sw_ke_abstract_algebra__c99c716a__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\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 993.8684867626231, \"nll_base\": 6.901604890823364, \"eval_ref\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb131072__v001\"... | {
"index": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb131072__v001",
"uid": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__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\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 993.8684867626231, \"nll_base\": 6.901604890823364, \"eval_ref\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb262144__v002\"... | {
"index": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb262144__v002",
"uid": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__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\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 993.8684867626231, \"nll_base\": 6.901604890823364, \"eval_ref\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb65536__v000\", \... | {
"index": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s000__tb65536__v000",
"uid": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__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\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 944.2184351080928, \"nll_base\": 6.850357532501221, \"eval_ref\": \"mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s001__tb65536__v003\", \... | {
"index": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__s001__tb65536__v003",
"uid": "mmmlu_sw_ke_clinical_knowledge__ca09ce1d__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\": \"mmmlu_sw_ke_college_biology__2c4d8ecb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 977.7014085479075, \"nll_base\": 6.885204315185547, \"eval_ref\": \"mmmlu_sw_ke_college_biology__2c4d8ecb__s000__tb65536__v000\", \"oracl... | {
"index": "mmmlu_sw_ke_college_biology__2c4d8ecb__s000__tb65536__v000",
"uid": "mmmlu_sw_ke_college_biology__2c4d8ecb__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\": \"mmmlu_sw_ke_college_chemistry__68ccb541__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1386.31056295222, \"nll_base\": 7.234401226043701, \"eval_ref\": \"mmmlu_sw_ke_college_chemistry__68ccb541__s000__tb262144__v000\", \... | {
"index": "mmmlu_sw_ke_college_chemistry__68ccb541__s000__tb262144__v000",
"uid": "mmmlu_sw_ke_college_chemistry__68ccb541__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\": \"mmmlu_sw_ke_computer_security__ead6d0bc__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 977.0494061445673, \"nll_base\": 6.884537220001221, \"eval_ref\": \"mmmlu_sw_ke_computer_security__ead6d0bc__s000__tb262144__v000\", ... | {
"index": "mmmlu_sw_ke_computer_security__ead6d0bc__s000__tb262144__v000",
"uid": "mmmlu_sw_ke_computer_security__ead6d0bc__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\": \"mmmlu_sw_ke_conceptual_physics__953a6f32__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1145.509017455405, \"nll_base\": 7.043604373931885, \"eval_ref\": \"mmmlu_sw_ke_conceptual_physics__953a6f32__s000__tb65536__v000\", \... | {
"index": "mmmlu_sw_ke_conceptual_physics__953a6f32__s000__tb65536__v000",
"uid": "mmmlu_sw_ke_conceptual_physics__953a6f32__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\": \"mmmlu_sw_ke_high_school_european_history__2e17b906__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 801.6906060416674, \"nll_base\": 6.686722755432129, \"eval_ref\": \"mmmlu_sw_ke_high_school_european_history__2e17b906__s0... | {
"index": "mmmlu_sw_ke_high_school_european_history__2e17b906__s000__tb131072__v000",
"uid": "mmmlu_sw_ke_high_school_european_history__2e17b906__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\": \"mmmlu_sw_ke_high_school_microeconomics__a66752c7__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 955.2319706495446, \"nll_base\": 6.861954212188721, \"eval_ref\": \"mmmlu_sw_ke_high_school_microeconomics__a66752c7__s000__tb... | {
"index": "mmmlu_sw_ke_high_school_microeconomics__a66752c7__s000__tb65536__v000",
"uid": "mmmlu_sw_ke_high_school_microeconomics__a66752c7__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\": \"mmmlu_sw_ke_high_school_psychology__f4c1f4ec__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1089.160389533722, \"nll_base\": 6.993162393569946, \"eval_ref\": \"mmmlu_sw_ke_high_school_psychology__f4c1f4ec__s000__tb65536__v... | {
"index": "mmmlu_sw_ke_high_school_psychology__f4c1f4ec__s000__tb65536__v000",
"uid": "mmmlu_sw_ke_high_school_psychology__f4c1f4ec__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.