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_ig_medical_genetics__e9e40079__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 864.8957720867103, \"nll_base\": 6.762609004974365, \"eval_ref\": \"global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb13... | {
"index": "global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb131072__v001",
"uid": "global_mmlu_full_ig_medical_genetics__e9e40079__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_ig_medical_genetics__e9e40079__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 864.8957720867103, \"nll_base\": 6.762609004974365, \"eval_ref\": \"global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb26... | {
"index": "global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb262144__v002",
"uid": "global_mmlu_full_ig_medical_genetics__e9e40079__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_ig_medical_genetics__e9e40079__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 864.8957720867103, \"nll_base\": 6.762609004974365, \"eval_ref\": \"global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb6553... | {
"index": "global_mmlu_full_ig_medical_genetics__e9e40079__s000__tb65536__v000",
"uid": "global_mmlu_full_ig_medical_genetics__e9e40079__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_ig_medical_genetics__e9e40079__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 984.6185945882806, \"nll_base\": 6.89225435256958, \"eval_ref\": \"global_mmlu_full_ig_medical_genetics__e9e40079__s001__tb65536... | {
"index": "global_mmlu_full_ig_medical_genetics__e9e40079__s001__tb65536__v003",
"uid": "global_mmlu_full_ig_medical_genetics__e9e40079__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_it_college_biology__1d1bc788__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 901.547447339986, \"nll_base\": 6.804112672805786, \"eval_ref\": \"global_mmlu_full_it_college_biology__1d1bc788__s000__tb65536__... | {
"index": "global_mmlu_full_it_college_biology__1d1bc788__s000__tb65536__v000",
"uid": "global_mmlu_full_it_college_biology__1d1bc788__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_it_elementary_mathematics__20db560c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1006.2702923064869, \"nll_base\": 6.914005994796753, \"eval_ref\": \"global_mmlu_full_it_elementary_mathematics__20db560c_... | {
"index": "global_mmlu_full_it_elementary_mathematics__20db560c__s000__tb65536__v000",
"uid": "global_mmlu_full_it_elementary_mathematics__20db560c__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_it_high_school_world_history__018a90bb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 771.1871639776663, \"nll_base\": 6.647931098937988, \"eval_ref\": \"global_mmlu_full_it_high_school_world_history__018a... | {
"index": "global_mmlu_full_it_high_school_world_history__018a90bb__s000__tb65536__v000",
"uid": "global_mmlu_full_it_high_school_world_history__018a90bb__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_it_human_sexuality__356d3a93__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1169.064205363049, \"nll_base\": 7.0639588832855225, \"eval_ref\": \"global_mmlu_full_it_human_sexuality__356d3a93__s000__tb262... | {
"index": "global_mmlu_full_it_human_sexuality__356d3a93__s000__tb262144__v000",
"uid": "global_mmlu_full_it_human_sexuality__356d3a93__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_it_medical_genetics__b92f21ad__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 835.8812336105459, \"nll_base\": 6.72848653793335, \"eval_ref\": \"global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb131... | {
"index": "global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb131072__v000",
"uid": "global_mmlu_full_it_medical_genetics__b92f21ad__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_it_medical_genetics__b92f21ad__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 835.8812336105459, \"nll_base\": 6.72848653793335, \"eval_ref\": \"global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb262... | {
"index": "global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb262144__v001",
"uid": "global_mmlu_full_it_medical_genetics__b92f21ad__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_it_medical_genetics__b92f21ad__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 835.8812336105459, \"nll_base\": 6.72848653793335, \"eval_ref\": \"global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb65536... | {
"index": "global_mmlu_full_it_medical_genetics__b92f21ad__s000__tb65536__v002",
"uid": "global_mmlu_full_it_medical_genetics__b92f21ad__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_it_moral_scenarios__88039af5__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1296.2491829715573, \"nll_base\": 7.167230129241943, \"eval_ref\": \"global_mmlu_full_it_moral_scenarios__88039af5__s000__tb65536... | {
"index": "global_mmlu_full_it_moral_scenarios__88039af5__s000__tb65536__v000",
"uid": "global_mmlu_full_it_moral_scenarios__88039af5__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_it_professional_law__e0f28202__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 481.74796976168074, \"nll_base\": 6.1774210929870605, \"eval_ref\": \"global_mmlu_full_it_professional_law__e0f28202__s000__tb65... | {
"index": "global_mmlu_full_it_professional_law__e0f28202__s000__tb65536__v000",
"uid": "global_mmlu_full_it_professional_law__e0f28202__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_it_security_studies__6b276488__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1155.7139205978951, \"nll_base\": 7.052473545074463, \"eval_ref\": \"global_mmlu_full_it_security_studies__6b276488__s000__tb2... | {
"index": "global_mmlu_full_it_security_studies__6b276488__s000__tb262144__v000",
"uid": "global_mmlu_full_it_security_studies__6b276488__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_ja_clinical_knowledge__a9631e45__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1082.5099577627789, \"nll_base\": 6.987037658691406, \"eval_ref\": \"global_mmlu_full_ja_clinical_knowledge__a9631e45__s000_... | {
"index": "global_mmlu_full_ja_clinical_knowledge__a9631e45__s000__tb131072__v002",
"uid": "global_mmlu_full_ja_clinical_knowledge__a9631e45__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_ja_clinical_knowledge__a9631e45__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1082.5099577627789, \"nll_base\": 6.987037658691406, \"eval_ref\": \"global_mmlu_full_ja_clinical_knowledge__a9631e45__s000__t... | {
"index": "global_mmlu_full_ja_clinical_knowledge__a9631e45__s000__tb65536__v001",
"uid": "global_mmlu_full_ja_clinical_knowledge__a9631e45__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_ja_elementary_mathematics__7aa6cf76__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1682.419627662913, \"nll_base\": 7.427988290786743, \"eval_ref\": \"global_mmlu_full_ja_elementary_mathematics__7aa6cf76... | {
"index": "global_mmlu_full_ja_elementary_mathematics__7aa6cf76__s000__tb131072__v000",
"uid": "global_mmlu_full_ja_elementary_mathematics__7aa6cf76__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_ja_marketing__19dee382__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 613.7400085872965, \"nll_base\": 6.419571399688721, \"eval_ref\": \"global_mmlu_full_ja_marketing__19dee382__s000__tb65536__v000\", \"o... | {
"index": "global_mmlu_full_ja_marketing__19dee382__s000__tb65536__v000",
"uid": "global_mmlu_full_ja_marketing__19dee382__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_ja_medical_genetics__c3dd62ae__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 761.1046540915584, \"nll_base\": 6.63477087020874, \"eval_ref\": \"global_mmlu_full_ja_medical_genetics__c3dd62ae__s000__tb65536... | {
"index": "global_mmlu_full_ja_medical_genetics__c3dd62ae__s000__tb65536__v002",
"uid": "global_mmlu_full_ja_medical_genetics__c3dd62ae__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_ja_world_religions__64d3823e__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 665.9308884185353, \"nll_base\": 6.501185894012451, \"eval_ref\": \"global_mmlu_full_ja_world_religions__64d3823e__s000__tb1310... | {
"index": "global_mmlu_full_ja_world_religions__64d3823e__s000__tb131072__v000",
"uid": "global_mmlu_full_ja_world_religions__64d3823e__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_ko_clinical_knowledge__c2b402fc__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1016.7244752320431, \"nll_base\": 6.924341440200806, \"eval_ref\": \"global_mmlu_full_ko_clinical_knowledge__c2b402fc__s000__t... | {
"index": "global_mmlu_full_ko_clinical_knowledge__c2b402fc__s000__tb65536__v002",
"uid": "global_mmlu_full_ko_clinical_knowledge__c2b402fc__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_ko_medical_genetics__94aa114e__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1079.2716460768727, \"nll_base\": 6.984041690826416, \"eval_ref\": \"global_mmlu_full_ko_medical_genetics__94aa114e__s000__tb655... | {
"index": "global_mmlu_full_ko_medical_genetics__94aa114e__s000__tb65536__v001",
"uid": "global_mmlu_full_ko_medical_genetics__94aa114e__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_ky_high_school_geography__97b979aa__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1918.2975380249516, \"nll_base\": 7.55919337272644, \"eval_ref\": \"global_mmlu_full_ky_high_school_geography__97b979aa__... | {
"index": "global_mmlu_full_ky_high_school_geography__97b979aa__s000__tb131072__v000",
"uid": "global_mmlu_full_ky_high_school_geography__97b979aa__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_ky_medical_genetics__ee64e441__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1457.2529969937136, \"nll_base\": 7.284308433532715, \"eval_ref\": \"global_mmlu_full_ky_medical_genetics__ee64e441__s000__tb1... | {
"index": "global_mmlu_full_ky_medical_genetics__ee64e441__s000__tb131072__v000",
"uid": "global_mmlu_full_ky_medical_genetics__ee64e441__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_ky_medical_genetics__ee64e441__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1457.2529969937136, \"nll_base\": 7.284308433532715, \"eval_ref\": \"global_mmlu_full_ky_medical_genetics__ee64e441__s000__tb655... | {
"index": "global_mmlu_full_ky_medical_genetics__ee64e441__s000__tb65536__v002",
"uid": "global_mmlu_full_ky_medical_genetics__ee64e441__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_ky_medical_genetics__ee64e441__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1506.642753516751, \"nll_base\": 7.317639112472534, \"eval_ref\": \"global_mmlu_full_ky_medical_genetics__ee64e441__s001__tb13... | {
"index": "global_mmlu_full_ky_medical_genetics__ee64e441__s001__tb131072__v003",
"uid": "global_mmlu_full_ky_medical_genetics__ee64e441__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_lt_clinical_knowledge__18aa41b6__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1031.7331750756987, \"nll_base\": 6.938995361328125, \"eval_ref\": \"global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000_... | {
"index": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000__tb131072__v001",
"uid": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__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_lt_clinical_knowledge__18aa41b6__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1031.7331750756987, \"nll_base\": 6.938995361328125, \"eval_ref\": \"global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000_... | {
"index": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000__tb262144__v002",
"uid": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__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_lt_clinical_knowledge__18aa41b6__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1031.7331750756987, \"nll_base\": 6.938995361328125, \"eval_ref\": \"global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000__t... | {
"index": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__s000__tb65536__v000",
"uid": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__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_lt_clinical_knowledge__18aa41b6__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 762.8825094421609, \"nll_base\": 6.637104034423828, \"eval_ref\": \"global_mmlu_full_lt_clinical_knowledge__18aa41b6__s001__tb... | {
"index": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__s001__tb65536__v003",
"uid": "global_mmlu_full_lt_clinical_knowledge__18aa41b6__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_lt_computer_security__dd7d23ba__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 855.9700717294647, \"nll_base\": 6.752235412597656, \"eval_ref\": \"global_mmlu_full_lt_computer_security__dd7d23ba__s000__tb65... | {
"index": "global_mmlu_full_lt_computer_security__dd7d23ba__s000__tb65536__v000",
"uid": "global_mmlu_full_lt_computer_security__dd7d23ba__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_lt_high_school_biology__3031cca7__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 719.7189292379347, \"nll_base\": 6.578860759735107, \"eval_ref\": \"global_mmlu_full_lt_high_school_biology__3031cca7__s000... | {
"index": "global_mmlu_full_lt_high_school_biology__3031cca7__s000__tb262144__v000",
"uid": "global_mmlu_full_lt_high_school_biology__3031cca7__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_lt_medical_genetics__2401eff6__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 830.1577901367607, \"nll_base\": 6.721615791320801, \"eval_ref\": \"global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb13... | {
"index": "global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb131072__v000",
"uid": "global_mmlu_full_lt_medical_genetics__2401eff6__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_lt_medical_genetics__2401eff6__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 830.1577901367607, \"nll_base\": 6.721615791320801, \"eval_ref\": \"global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb26... | {
"index": "global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb262144__v001",
"uid": "global_mmlu_full_lt_medical_genetics__2401eff6__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_lt_medical_genetics__2401eff6__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 830.1577901367607, \"nll_base\": 6.721615791320801, \"eval_ref\": \"global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb6553... | {
"index": "global_mmlu_full_lt_medical_genetics__2401eff6__s000__tb65536__v002",
"uid": "global_mmlu_full_lt_medical_genetics__2401eff6__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_lt_medical_genetics__2401eff6__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 989.121767372326, \"nll_base\": 6.896817445755005, \"eval_ref\": \"global_mmlu_full_lt_medical_genetics__2401eff6__s001__tb131... | {
"index": "global_mmlu_full_lt_medical_genetics__2401eff6__s001__tb131072__v003",
"uid": "global_mmlu_full_lt_medical_genetics__2401eff6__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_lt_miscellaneous__b3ff566d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1556.5574297276178, \"nll_base\": 7.350231885910034, \"eval_ref\": \"global_mmlu_full_lt_miscellaneous__b3ff566d__s000__tb131072_... | {
"index": "global_mmlu_full_lt_miscellaneous__b3ff566d__s000__tb131072__v000",
"uid": "global_mmlu_full_lt_miscellaneous__b3ff566d__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_mg_clinical_knowledge__34d79f3e__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1165.513613792623, \"nll_base\": 7.060917139053345, \"eval_ref\": \"global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__... | {
"index": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__tb131072__v002",
"uid": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__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_mg_clinical_knowledge__34d79f3e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1165.513613792623, \"nll_base\": 7.060917139053345, \"eval_ref\": \"global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__... | {
"index": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__tb262144__v000",
"uid": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__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_mg_clinical_knowledge__34d79f3e__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1165.513613792623, \"nll_base\": 7.060917139053345, \"eval_ref\": \"global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__tb... | {
"index": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__s000__tb65536__v001",
"uid": "global_mmlu_full_mg_clinical_knowledge__34d79f3e__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_mg_elementary_mathematics__c0ea2c05__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1372.5633506455217, \"nll_base\": 7.224435329437256, \"eval_ref\": \"global_mmlu_full_mg_elementary_mathematics__c0ea2c0... | {
"index": "global_mmlu_full_mg_elementary_mathematics__c0ea2c05__s000__tb131072__v000",
"uid": "global_mmlu_full_mg_elementary_mathematics__c0ea2c05__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_mg_global_facts__b3a2e622__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1478.4248643281608, \"nll_base\": 7.29873251914978, \"eval_ref\": \"global_mmlu_full_mg_global_facts__b3a2e622__s000__tb65536__v000\... | {
"index": "global_mmlu_full_mg_global_facts__b3a2e622__s000__tb65536__v000",
"uid": "global_mmlu_full_mg_global_facts__b3a2e622__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_mg_high_school_government_and_politics__dc2e5d65__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 919.6535478736565, \"nll_base\": 6.8239970207214355, \"eval_ref\": \"global_mmlu_full_mg_high_school_govern... | {
"index": "global_mmlu_full_mg_high_school_government_and_politics__dc2e5d65__s000__tb262144__v000",
"uid": "global_mmlu_full_mg_high_school_government_and_politics__dc2e5d65__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_mg_high_school_us_history__5c681504__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 786.9325638573473, \"nll_base\": 6.668142557144165, \"eval_ref\": \"global_mmlu_full_mg_high_school_us_history__5c681504... | {
"index": "global_mmlu_full_mg_high_school_us_history__5c681504__s000__tb131072__v000",
"uid": "global_mmlu_full_mg_high_school_us_history__5c681504__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_mg_jurisprudence__3ecf6850__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1006.6729477544195, \"nll_base\": 6.914406061172485, \"eval_ref\": \"global_mmlu_full_mg_jurisprudence__3ecf6850__s000__tb65536__v0... | {
"index": "global_mmlu_full_mg_jurisprudence__3ecf6850__s000__tb65536__v000",
"uid": "global_mmlu_full_mg_jurisprudence__3ecf6850__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_mg_machine_learning__dc4be3fa__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 675.1676323464701, \"nll_base\": 6.514961004257202, \"eval_ref\": \"global_mmlu_full_mg_machine_learning__dc4be3fa__s000__tb13... | {
"index": "global_mmlu_full_mg_machine_learning__dc4be3fa__s000__tb131072__v000",
"uid": "global_mmlu_full_mg_machine_learning__dc4be3fa__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_mg_medical_genetics__b41e4284__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 813.6412054525251, \"nll_base\": 6.70151948928833, \"eval_ref\": \"global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb131... | {
"index": "global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb131072__v001",
"uid": "global_mmlu_full_mg_medical_genetics__b41e4284__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_mg_medical_genetics__b41e4284__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 813.6412054525251, \"nll_base\": 6.70151948928833, \"eval_ref\": \"global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb262... | {
"index": "global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb262144__v002",
"uid": "global_mmlu_full_mg_medical_genetics__b41e4284__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_mg_medical_genetics__b41e4284__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 813.6412054525251, \"nll_base\": 6.70151948928833, \"eval_ref\": \"global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb65536... | {
"index": "global_mmlu_full_mg_medical_genetics__b41e4284__s000__tb65536__v000",
"uid": "global_mmlu_full_mg_medical_genetics__b41e4284__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_mg_prehistory__63046133__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 962.7948756268913, \"nll_base\": 6.869840383529663, \"eval_ref\": \"global_mmlu_full_mg_prehistory__63046133__s000__tb131072__v000\"... | {
"index": "global_mmlu_full_mg_prehistory__63046133__s000__tb131072__v000",
"uid": "global_mmlu_full_mg_prehistory__63046133__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_mg_virology__d4f13836__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 889.1449894331246, \"nll_base\": 6.790260314941406, \"eval_ref\": \"global_mmlu_full_mg_virology__d4f13836__s000__tb131072__v000\", \"... | {
"index": "global_mmlu_full_mg_virology__d4f13836__s000__tb131072__v000",
"uid": "global_mmlu_full_mg_virology__d4f13836__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_ms_clinical_knowledge__ffdd4faf__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 950.5666559009227, \"nll_base\": 6.85705828666687, \"eval_ref\": \"global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__t... | {
"index": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__tb131072__v001",
"uid": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__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_ms_clinical_knowledge__ffdd4faf__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 950.5666559009227, \"nll_base\": 6.85705828666687, \"eval_ref\": \"global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__t... | {
"index": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__tb262144__v002",
"uid": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__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_ms_clinical_knowledge__ffdd4faf__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 950.5666559009227, \"nll_base\": 6.85705828666687, \"eval_ref\": \"global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__tb6... | {
"index": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s000__tb65536__v000",
"uid": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__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_ms_clinical_knowledge__ffdd4faf__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1133.6282037632486, \"nll_base\": 7.0331785678863525, \"eval_ref\": \"global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s001__... | {
"index": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__s001__tb65536__v003",
"uid": "global_mmlu_full_ms_clinical_knowledge__ffdd4faf__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_ms_college_biology__2a11fb4b__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1007.9328254967953, \"nll_base\": 6.915656805038452, \"eval_ref\": \"global_mmlu_full_ms_college_biology__2a11fb4b__s000__tb65536... | {
"index": "global_mmlu_full_ms_college_biology__2a11fb4b__s000__tb65536__v000",
"uid": "global_mmlu_full_ms_college_biology__2a11fb4b__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_ms_formal_logic__5123317c__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 792.7861873571348, \"nll_base\": 6.675553560256958, \"eval_ref\": \"global_mmlu_full_ms_formal_logic__5123317c__s000__tb131072__v0... | {
"index": "global_mmlu_full_ms_formal_logic__5123317c__s000__tb131072__v000",
"uid": "global_mmlu_full_ms_formal_logic__5123317c__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_ms_high_school_chemistry__56fbd01d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 910.9946716493582, \"nll_base\": 6.814537048339844, \"eval_ref\": \"global_mmlu_full_ms_high_school_chemistry__56fbd01d__... | {
"index": "global_mmlu_full_ms_high_school_chemistry__56fbd01d__s000__tb262144__v000",
"uid": "global_mmlu_full_ms_high_school_chemistry__56fbd01d__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_ms_high_school_government_and_politics__781a91ba__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1279.9149167792127, \"nll_base\": 7.15454888343811, \"eval_ref\": \"global_mmlu_full_ms_high_school_governm... | {
"index": "global_mmlu_full_ms_high_school_government_and_politics__781a91ba__s000__tb262144__v000",
"uid": "global_mmlu_full_ms_high_school_government_and_politics__781a91ba__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_ms_high_school_microeconomics__0b175ff6__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 985.401566789652, \"nll_base\": 6.893049240112305, \"eval_ref\": \"global_mmlu_full_ms_high_school_microeconomics__0... | {
"index": "global_mmlu_full_ms_high_school_microeconomics__0b175ff6__s000__tb131072__v000",
"uid": "global_mmlu_full_ms_high_school_microeconomics__0b175ff6__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_ms_high_school_psychology__d5154287__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1880.5745639420973, \"nll_base\": 7.539332628250122, \"eval_ref\": \"global_mmlu_full_ms_high_school_psychology__d515428... | {
"index": "global_mmlu_full_ms_high_school_psychology__d5154287__s000__tb262144__v000",
"uid": "global_mmlu_full_ms_high_school_psychology__d5154287__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_ms_human_aging__81837b71__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 924.984299556568, \"nll_base\": 6.829776763916016, \"eval_ref\": \"global_mmlu_full_ms_human_aging__81837b71__s000__tb262144__v000\... | {
"index": "global_mmlu_full_ms_human_aging__81837b71__s000__tb262144__v000",
"uid": "global_mmlu_full_ms_human_aging__81837b71__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_ms_medical_genetics__954a39b5__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1154.9206272128386, \"nll_base\": 7.05178689956665, \"eval_ref\": \"global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb13... | {
"index": "global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb131072__v001",
"uid": "global_mmlu_full_ms_medical_genetics__954a39b5__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_ms_medical_genetics__954a39b5__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1154.9206272128386, \"nll_base\": 7.05178689956665, \"eval_ref\": \"global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb26... | {
"index": "global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb262144__v002",
"uid": "global_mmlu_full_ms_medical_genetics__954a39b5__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_ms_medical_genetics__954a39b5__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1154.9206272128386, \"nll_base\": 7.05178689956665, \"eval_ref\": \"global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb6553... | {
"index": "global_mmlu_full_ms_medical_genetics__954a39b5__s000__tb65536__v000",
"uid": "global_mmlu_full_ms_medical_genetics__954a39b5__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_ms_moral_scenarios__862102ce__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 888.9434110471024, \"nll_base\": 6.790033578872681, \"eval_ref\": \"global_mmlu_full_ms_moral_scenarios__862102ce__s000__tb65536_... | {
"index": "global_mmlu_full_ms_moral_scenarios__862102ce__s000__tb65536__v000",
"uid": "global_mmlu_full_ms_moral_scenarios__862102ce__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_ms_philosophy__a1fae2f3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1209.9220146808775, \"nll_base\": 7.098311185836792, \"eval_ref\": \"global_mmlu_full_ms_philosophy__a1fae2f3__s000__tb131072__v000\... | {
"index": "global_mmlu_full_ms_philosophy__a1fae2f3__s000__tb131072__v000",
"uid": "global_mmlu_full_ms_philosophy__a1fae2f3__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_ms_professional_medicine__623d022e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1441.3087682178766, \"nll_base\": 7.273306846618652, \"eval_ref\": \"global_mmlu_full_ms_professional_medicine__623d022e_... | {
"index": "global_mmlu_full_ms_professional_medicine__623d022e__s000__tb262144__v000",
"uid": "global_mmlu_full_ms_professional_medicine__623d022e__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_ne_clinical_knowledge__b3a0e08d__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 880.6068540049978, \"nll_base\": 6.780611276626587, \"eval_ref\": \"global_mmlu_full_ne_clinical_knowledge__b3a0e08d__s000__... | {
"index": "global_mmlu_full_ne_clinical_knowledge__b3a0e08d__s000__tb131072__v002",
"uid": "global_mmlu_full_ne_clinical_knowledge__b3a0e08d__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_ne_clinical_knowledge__b3a0e08d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 880.6068540049978, \"nll_base\": 6.780611276626587, \"eval_ref\": \"global_mmlu_full_ne_clinical_knowledge__b3a0e08d__s000__tb... | {
"index": "global_mmlu_full_ne_clinical_knowledge__b3a0e08d__s000__tb65536__v001",
"uid": "global_mmlu_full_ne_clinical_knowledge__b3a0e08d__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_ne_college_computer_science__cebfe0d4__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 524.1372403598605, \"nll_base\": 6.261753559112549, \"eval_ref\": \"global_mmlu_full_ne_college_computer_science__cebfe0... | {
"index": "global_mmlu_full_ne_college_computer_science__cebfe0d4__s000__tb65536__v000",
"uid": "global_mmlu_full_ne_college_computer_science__cebfe0d4__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_ne_formal_logic__b9655690__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 352.36965015123866, \"nll_base\": 5.864680767059326, \"eval_ref\": \"global_mmlu_full_ne_formal_logic__b9655690__s000__tb131072__v... | {
"index": "global_mmlu_full_ne_formal_logic__b9655690__s000__tb131072__v000",
"uid": "global_mmlu_full_ne_formal_logic__b9655690__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_ne_high_school_psychology__dde8f5e2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 877.1471335590475, \"nll_base\": 6.776674747467041, \"eval_ref\": \"global_mmlu_full_ne_high_school_psychology__dde8f5e2__... | {
"index": "global_mmlu_full_ne_high_school_psychology__dde8f5e2__s000__tb65536__v000",
"uid": "global_mmlu_full_ne_high_school_psychology__dde8f5e2__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_ne_jurisprudence__e3aa439d__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 620.1500363941544, \"nll_base\": 6.429961442947388, \"eval_ref\": \"global_mmlu_full_ne_jurisprudence__e3aa439d__s000__tb65536__v00... | {
"index": "global_mmlu_full_ne_jurisprudence__e3aa439d__s000__tb65536__v000",
"uid": "global_mmlu_full_ne_jurisprudence__e3aa439d__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_ne_medical_genetics__ce6832d9__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 802.1465996502026, \"nll_base\": 6.687291383743286, \"eval_ref\": \"global_mmlu_full_ne_medical_genetics__ce6832d9__s000__tb6553... | {
"index": "global_mmlu_full_ne_medical_genetics__ce6832d9__s000__tb65536__v002",
"uid": "global_mmlu_full_ne_medical_genetics__ce6832d9__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_ne_medical_genetics__ce6832d9__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 845.1687278898978, \"nll_base\": 6.739536285400391, \"eval_ref\": \"global_mmlu_full_ne_medical_genetics__ce6832d9__s001__tb13... | {
"index": "global_mmlu_full_ne_medical_genetics__ce6832d9__s001__tb131072__v003",
"uid": "global_mmlu_full_ne_medical_genetics__ce6832d9__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_ne_security_studies__f3b71525__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 633.9506130953647, \"nll_base\": 6.451971054077148, \"eval_ref\": \"global_mmlu_full_ne_security_studies__f3b71525__s000__tb13... | {
"index": "global_mmlu_full_ne_security_studies__f3b71525__s000__tb131072__v000",
"uid": "global_mmlu_full_ne_security_studies__f3b71525__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_nl_clinical_knowledge__04d5f213__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 983.2214077029045, \"nll_base\": 6.890834331512451, \"eval_ref\": \"global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__... | {
"index": "global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__tb131072__v001",
"uid": "global_mmlu_full_nl_clinical_knowledge__04d5f213__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_nl_clinical_knowledge__04d5f213__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 983.2214077029045, \"nll_base\": 6.890834331512451, \"eval_ref\": \"global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__... | {
"index": "global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__tb262144__v002",
"uid": "global_mmlu_full_nl_clinical_knowledge__04d5f213__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_nl_clinical_knowledge__04d5f213__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 983.2214077029045, \"nll_base\": 6.890834331512451, \"eval_ref\": \"global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__tb... | {
"index": "global_mmlu_full_nl_clinical_knowledge__04d5f213__s000__tb65536__v000",
"uid": "global_mmlu_full_nl_clinical_knowledge__04d5f213__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_nl_computer_security__61472772__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 975.9561004494009, \"nll_base\": 6.88341760635376, \"eval_ref\": \"global_mmlu_full_nl_computer_security__61472772__s000__tb1... | {
"index": "global_mmlu_full_nl_computer_security__61472772__s000__tb131072__v000",
"uid": "global_mmlu_full_nl_computer_security__61472772__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_nl_econometrics__7e6946fa__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 687.3758695919497, \"nll_base\": 6.532881259918213, \"eval_ref\": \"global_mmlu_full_nl_econometrics__7e6946fa__s000__tb131072__v0... | {
"index": "global_mmlu_full_nl_econometrics__7e6946fa__s000__tb131072__v000",
"uid": "global_mmlu_full_nl_econometrics__7e6946fa__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_nl_high_school_statistics__56734853__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 795.9503579044605, \"nll_base\": 6.679536819458008, \"eval_ref\": \"global_mmlu_full_nl_high_school_statistics__56734853... | {
"index": "global_mmlu_full_nl_high_school_statistics__56734853__s000__tb262144__v000",
"uid": "global_mmlu_full_nl_high_school_statistics__56734853__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_nl_jurisprudence__f63c07ad__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1138.612243687632, \"nll_base\": 7.037565469741821, \"eval_ref\": \"global_mmlu_full_nl_jurisprudence__f63c07ad__s000__tb262144__... | {
"index": "global_mmlu_full_nl_jurisprudence__f63c07ad__s000__tb262144__v000",
"uid": "global_mmlu_full_nl_jurisprudence__f63c07ad__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_nl_medical_genetics__17e51ba3__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 980.3072277147824, \"nll_base\": 6.887866020202637, \"eval_ref\": \"global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb13... | {
"index": "global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb131072__v001",
"uid": "global_mmlu_full_nl_medical_genetics__17e51ba3__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_nl_medical_genetics__17e51ba3__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 980.3072277147824, \"nll_base\": 6.887866020202637, \"eval_ref\": \"global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb26... | {
"index": "global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb262144__v002",
"uid": "global_mmlu_full_nl_medical_genetics__17e51ba3__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_nl_medical_genetics__17e51ba3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 980.3072277147824, \"nll_base\": 6.887866020202637, \"eval_ref\": \"global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb6553... | {
"index": "global_mmlu_full_nl_medical_genetics__17e51ba3__s000__tb65536__v000",
"uid": "global_mmlu_full_nl_medical_genetics__17e51ba3__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_nl_medical_genetics__17e51ba3__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 900.74068756595, \"nll_base\": 6.80321741104126, \"eval_ref\": \"global_mmlu_full_nl_medical_genetics__17e51ba3__s001__tb65536__... | {
"index": "global_mmlu_full_nl_medical_genetics__17e51ba3__s001__tb65536__v003",
"uid": "global_mmlu_full_nl_medical_genetics__17e51ba3__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_nl_nutrition__c2a4b99c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1097.1261169937616, \"nll_base\": 7.0004494190216064, \"eval_ref\": \"global_mmlu_full_nl_nutrition__c2a4b99c__s000__tb262144__v000\"... | {
"index": "global_mmlu_full_nl_nutrition__c2a4b99c__s000__tb262144__v000",
"uid": "global_mmlu_full_nl_nutrition__c2a4b99c__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_nl_prehistory__89cb2ff6__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 939.1124046032513, \"nll_base\": 6.844935178756714, \"eval_ref\": \"global_mmlu_full_nl_prehistory__89cb2ff6__s000__tb65536__v000\", \... | {
"index": "global_mmlu_full_nl_prehistory__89cb2ff6__s000__tb65536__v000",
"uid": "global_mmlu_full_nl_prehistory__89cb2ff6__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_nl_public_relations__82a23736__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1058.2576117955914, \"nll_base\": 6.964379072189331, \"eval_ref\": \"global_mmlu_full_nl_public_relations__82a23736__s000__tb1... | {
"index": "global_mmlu_full_nl_public_relations__82a23736__s000__tb131072__v000",
"uid": "global_mmlu_full_nl_public_relations__82a23736__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_nl_us_foreign_policy__c5863d80__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1018.5167061624688, \"nll_base\": 6.926102638244629, \"eval_ref\": \"global_mmlu_full_nl_us_foreign_policy__c5863d80__s000__t... | {
"index": "global_mmlu_full_nl_us_foreign_policy__c5863d80__s000__tb262144__v000",
"uid": "global_mmlu_full_nl_us_foreign_policy__c5863d80__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_ny_clinical_knowledge__d4310c1c__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1131.9253809990814, \"nll_base\": 7.031675338745117, \"eval_ref\": \"global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000_... | {
"index": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000__tb131072__v002",
"uid": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__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_ny_clinical_knowledge__d4310c1c__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1131.9253809990814, \"nll_base\": 7.031675338745117, \"eval_ref\": \"global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000_... | {
"index": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000__tb262144__v000",
"uid": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__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_ny_clinical_knowledge__d4310c1c__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1131.9253809990814, \"nll_base\": 7.031675338745117, \"eval_ref\": \"global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000__t... | {
"index": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__s000__tb65536__v001",
"uid": "global_mmlu_full_ny_clinical_knowledge__d4310c1c__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_ny_conceptual_physics__89a8efd0__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1025.1943189614985, \"nll_base\": 6.932637453079224, \"eval_ref\": \"global_mmlu_full_ny_conceptual_physics__89a8efd0__s000_... | {
"index": "global_mmlu_full_ny_conceptual_physics__89a8efd0__s000__tb131072__v000",
"uid": "global_mmlu_full_ny_conceptual_physics__89a8efd0__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_ny_high_school_government_and_politics__caf0ba70__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 673.0588122950859, \"nll_base\": 6.5118327140808105, \"eval_ref\": \"global_mmlu_full_ny_high_school_govern... | {
"index": "global_mmlu_full_ny_high_school_government_and_politics__caf0ba70__s000__tb262144__v000",
"uid": "global_mmlu_full_ny_high_school_government_and_politics__caf0ba70__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_ny_high_school_psychology__690b0091__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1142.15440515367, \"nll_base\": 7.0406715869903564, \"eval_ref\": \"global_mmlu_full_ny_high_school_psychology__690b0091... | {
"index": "global_mmlu_full_ny_high_school_psychology__690b0091__s000__tb131072__v000",
"uid": "global_mmlu_full_ny_high_school_psychology__690b0091__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_ny_medical_genetics__5c80ef46__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1038.6524452057367, \"nll_base\": 6.945679426193237, \"eval_ref\": \"global_mmlu_full_ny_medical_genetics__5c80ef46__s000__tb1... | {
"index": "global_mmlu_full_ny_medical_genetics__5c80ef46__s000__tb131072__v001",
"uid": "global_mmlu_full_ny_medical_genetics__5c80ef46__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_ny_medical_genetics__5c80ef46__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1038.6524452057367, \"nll_base\": 6.945679426193237, \"eval_ref\": \"global_mmlu_full_ny_medical_genetics__5c80ef46__s000__tb2... | {
"index": "global_mmlu_full_ny_medical_genetics__5c80ef46__s000__tb262144__v002",
"uid": "global_mmlu_full_ny_medical_genetics__5c80ef46__s000__tb262144__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.