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\": \"include_base_44_japanese_few_shot_en_medical_license__c6b37752__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 439.7366032033586, \"nll_base\": 6.086175918579102, \"eval_ref\": \"include_base_44_japanese_few_shot_en_medical... | {
"index": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s001__tb65536__v003",
"uid": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__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\": \"include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 534.138997118782, \"nll_base\": 6.280656099319458, \"eval_ref\": \"include_base_44_japanese_few_shot_og_medica... | {
"index": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb131072__v000",
"uid": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__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\": \"include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 534.138997118782, \"nll_base\": 6.280656099319458, \"eval_ref\": \"include_base_44_japanese_few_shot_og_medica... | {
"index": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb262144__v001",
"uid": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__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\": \"include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 534.138997118782, \"nll_base\": 6.280656099319458, \"eval_ref\": \"include_base_44_japanese_few_shot_og_medical_... | {
"index": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s000__tb65536__v002",
"uid": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__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\": \"include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 411.2277337139789, \"nll_base\": 6.019147157669067, \"eval_ref\": \"include_base_44_japanese_few_shot_og_medic... | {
"index": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__s001__tb131072__v003",
"uid": "include_base_44_japanese_few_shot_og_medical_license__7b4b269d__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\": \"include_base_44_japanese_medical_license__a338b091__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 380.06749812889984, \"nll_base\": 5.940348863601685, \"eval_ref\": \"include_base_44_japanese_medical_license__a338b091__s... | {
"index": "include_base_44_japanese_medical_license__a338b091__s000__tb131072__v002",
"uid": "include_base_44_japanese_medical_license__a338b091__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\": \"include_base_44_japanese_medical_license__a338b091__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 380.06749812889984, \"nll_base\": 5.940348863601685, \"eval_ref\": \"include_base_44_japanese_medical_license__a338b091__s... | {
"index": "include_base_44_japanese_medical_license__a338b091__s000__tb262144__v000",
"uid": "include_base_44_japanese_medical_license__a338b091__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\": \"include_base_44_japanese_medical_license__a338b091__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 380.06749812889984, \"nll_base\": 5.940348863601685, \"eval_ref\": \"include_base_44_japanese_medical_license__a338b091__s00... | {
"index": "include_base_44_japanese_medical_license__a338b091__s000__tb65536__v001",
"uid": "include_base_44_japanese_medical_license__a338b091__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\": \"include_base_44_korean_few_shot_og_social_science__d1743b1d__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1360.3402112372019, \"nll_base\": 7.215490102767944, \"eval_ref\": \"include_base_44_korean_few_shot_og_social_scie... | {
"index": "include_base_44_korean_few_shot_og_social_science__d1743b1d__s000__tb65536__v000",
"uid": "include_base_44_korean_few_shot_og_social_science__d1743b1d__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\": \"include_base_44_lithuanian_business_commerce__8f74aeff__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 862.3701291617969, \"nll_base\": 6.7596845626831055, \"eval_ref\": \"include_base_44_lithuanian_business_commerce__8f7... | {
"index": "include_base_44_lithuanian_business_commerce__8f74aeff__s000__tb131072__v000",
"uid": "include_base_44_lithuanian_business_commerce__8f74aeff__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\": \"include_base_44_lithuanian_few_shot_og_stem__cbbac354__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1057.9057002754528, \"nll_base\": 6.964046478271484, \"eval_ref\": \"include_base_44_lithuanian_few_shot_og_stem__cbbac... | {
"index": "include_base_44_lithuanian_few_shot_og_stem__cbbac354__s000__tb262144__v000",
"uid": "include_base_44_lithuanian_few_shot_og_stem__cbbac354__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\": \"include_base_44_malay_few_shot_en_business_commerce__c9dd80e9__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1497.0633481021875, \"nll_base\": 7.31126070022583, \"eval_ref\": \"include_base_44_malay_few_shot_en_business_... | {
"index": "include_base_44_malay_few_shot_en_business_commerce__c9dd80e9__s000__tb262144__v000",
"uid": "include_base_44_malay_few_shot_en_business_commerce__c9dd80e9__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\": \"include_base_44_nepali_professional_certification__c0e69bcd__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 920.828667918395, \"nll_base\": 6.8252739906311035, \"eval_ref\": \"include_base_44_nepali_professional_certifica... | {
"index": "include_base_44_nepali_professional_certification__c0e69bcd__s000__tb131072__v000",
"uid": "include_base_44_nepali_professional_certification__c0e69bcd__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\": \"include_base_44_persian_few_shot_en_arts_humanities__efc62838__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 349.64282592972626, \"nll_base\": 5.856912136077881, \"eval_ref\": \"include_base_44_persian_few_shot_en_arts_hum... | {
"index": "include_base_44_persian_few_shot_en_arts_humanities__efc62838__s000__tb65536__v000",
"uid": "include_base_44_persian_few_shot_en_arts_humanities__efc62838__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\": \"include_base_44_persian_social_science__25b43c44__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 398.59155144830567, \"nll_base\": 5.9879372119903564, \"eval_ref\": \"include_base_44_persian_social_science__25b43c44__s000__... | {
"index": "include_base_44_persian_social_science__25b43c44__s000__tb65536__v000",
"uid": "include_base_44_persian_social_science__25b43c44__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\": \"include_base_44_polish_few_shot_en_stem__422e30b2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 451.1529389040615, \"nll_base\": 6.111806392669678, \"eval_ref\": \"include_base_44_polish_few_shot_en_stem__422e30b2__s000__... | {
"index": "include_base_44_polish_few_shot_en_stem__422e30b2__s000__tb65536__v000",
"uid": "include_base_44_polish_few_shot_en_stem__422e30b2__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\": \"include_base_44_russian_few_shot_en_stem__e3e06e7b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1102.9497973442235, \"nll_base\": 7.005743503570557, \"eval_ref\": \"include_base_44_russian_few_shot_en_stem__e3e06e7b__s... | {
"index": "include_base_44_russian_few_shot_en_stem__e3e06e7b__s000__tb131072__v000",
"uid": "include_base_44_russian_few_shot_en_stem__e3e06e7b__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\": \"include_base_44_russian_few_shot_og_social_science__5d6fe4c0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1398.225902288161, \"nll_base\": 7.242959499359131, \"eval_ref\": \"include_base_44_russian_few_shot_og_social_sci... | {
"index": "include_base_44_russian_few_shot_og_social_science__5d6fe4c0__s000__tb65536__v000",
"uid": "include_base_44_russian_few_shot_og_social_science__5d6fe4c0__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\": \"include_base_44_russian_marine_license__76e53231__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1086.8633834806212, \"nll_base\": 6.991051197052002, \"eval_ref\": \"include_base_44_russian_marine_license__76e53231__s000__t... | {
"index": "include_base_44_russian_marine_license__76e53231__s000__tb65536__v000",
"uid": "include_base_44_russian_marine_license__76e53231__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\": \"include_base_44_tamil_stem__e45de233__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 507.15588235207133, \"nll_base\": 6.228818416595459, \"eval_ref\": \"include_base_44_tamil_stem__e45de233__s000__tb131072__v000\", \"ora... | {
"index": "include_base_44_tamil_stem__e45de233__s000__tb131072__v000",
"uid": "include_base_44_tamil_stem__e45de233__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\": \"include_base_44_turkish_stem__696910b3__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 924.6152002350178, \"nll_base\": 6.8293776512146, \"eval_ref\": \"include_base_44_turkish_stem__696910b3__s000__tb131072__v000\", \"or... | {
"index": "include_base_44_turkish_stem__696910b3__s000__tb131072__v000",
"uid": "include_base_44_turkish_stem__696910b3__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\": \"include_base_44_ukrainian_few_shot_og_arts_humanities__7becc909__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1476.8398892224673, \"nll_base\": 7.297659873962402, \"eval_ref\": \"include_base_44_ukrainian_few_shot_og_arts... | {
"index": "include_base_44_ukrainian_few_shot_og_arts_humanities__7becc909__s000__tb65536__v000",
"uid": "include_base_44_ukrainian_few_shot_og_arts_humanities__7becc909__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\": \"include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1165.915497688236, \"nll_base\": 7.061261892318726, \"eval_ref\": \"include_base_44_uzbek_few_shot_en_medical_lic... | {
"index": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb131072__v002",
"uid": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__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\": \"include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1165.915497688236, \"nll_base\": 7.061261892318726, \"eval_ref\": \"include_base_44_uzbek_few_shot_en_medical_lic... | {
"index": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb262144__v000",
"uid": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__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\": \"include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1165.915497688236, \"nll_base\": 7.061261892318726, \"eval_ref\": \"include_base_44_uzbek_few_shot_en_medical_licen... | {
"index": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__s000__tb65536__v001",
"uid": "include_base_44_uzbek_few_shot_en_medical_license__3e43ce5a__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\": \"include_base_44_uzbek_few_shot_en_social_science__219ceb56__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1176.8526352586741, \"nll_base\": 7.070598895733173, \"eval_ref\": \"include_base_44_uzbek_few_shot_en_social_scie... | {
"index": "include_base_44_uzbek_few_shot_en_social_science__219ceb56__s000__tb131072__v000",
"uid": "include_base_44_uzbek_few_shot_en_social_science__219ceb56__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\": \"include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1014.4143080654493, \"nll_base\": 6.922066688537598, \"eval_ref\": \"include_base_44_uzbek_few_shot_og_medical_li... | {
"index": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb131072__v002",
"uid": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__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\": \"include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1014.4143080654493, \"nll_base\": 6.922066688537598, \"eval_ref\": \"include_base_44_uzbek_few_shot_og_medical_li... | {
"index": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb262144__v000",
"uid": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__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\": \"include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1014.4143080654493, \"nll_base\": 6.922066688537598, \"eval_ref\": \"include_base_44_uzbek_few_shot_og_medical_lice... | {
"index": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s000__tb65536__v001",
"uid": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__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\": \"include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s001__tb262144__v003\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 875.4245654041395, \"nll_base\": 6.774708986282349, \"eval_ref\": \"include_base_44_uzbek_few_shot_og_medical_lic... | {
"index": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s001__tb262144__v003",
"uid": "include_base_44_uzbek_few_shot_og_medical_license__8f16a43d__s001__tb262144__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_uzbek_medical_license__d5dea5e1__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1039.9873068186373, \"nll_base\": 6.946963787078857, \"eval_ref\": \"include_base_44_uzbek_medical_license__d5dea5e1__s000__t... | {
"index": "include_base_44_uzbek_medical_license__d5dea5e1__s000__tb131072__v002",
"uid": "include_base_44_uzbek_medical_license__d5dea5e1__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\": \"include_base_44_uzbek_medical_license__d5dea5e1__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1039.9873068186373, \"nll_base\": 6.946963787078857, \"eval_ref\": \"include_base_44_uzbek_medical_license__d5dea5e1__s000__tb6... | {
"index": "include_base_44_uzbek_medical_license__d5dea5e1__s000__tb65536__v001",
"uid": "include_base_44_uzbek_medical_license__d5dea5e1__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\": \"include_base_44_uzbek_social_science__f8ab3b6b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 883.5261371461836, \"nll_base\": 6.78392087496244, \"eval_ref\": \"include_base_44_uzbek_social_science__f8ab3b6b__s000__tb131... | {
"index": "include_base_44_uzbek_social_science__f8ab3b6b__s000__tb131072__v000",
"uid": "include_base_44_uzbek_social_science__f8ab3b6b__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\": \"include_base_44_vietnamese_few_shot_en_social_science__2e8ca690__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 651.6626347270168, \"nll_base\": 6.479526996612549, \"eval_ref\": \"include_base_44_vietnamese_few_shot_en_so... | {
"index": "include_base_44_vietnamese_few_shot_en_social_science__2e8ca690__s000__tb131072__v000",
"uid": "include_base_44_vietnamese_few_shot_en_social_science__2e8ca690__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\": \"include_base_44_vietnamese_few_shot_og_stem__97733abd__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 649.2661430013044, \"nll_base\": 6.475842714309692, \"eval_ref\": \"include_base_44_vietnamese_few_shot_og_stem__97733a... | {
"index": "include_base_44_vietnamese_few_shot_og_stem__97733abd__s000__tb262144__v000",
"uid": "include_base_44_vietnamese_few_shot_og_stem__97733abd__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\": \"include_base_44_vietnamese_social_science__c942669f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 654.8361250574385, \"nll_base\": 6.484385013580322, \"eval_ref\": \"include_base_44_vietnamese_social_science__c942669f__s0... | {
"index": "include_base_44_vietnamese_social_science__c942669f__s000__tb65536__v000",
"uid": "include_base_44_vietnamese_social_science__c942669f__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\": \"indicparam_bodo__c51dbb5d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1327.7401423508645, \"nll_base\": 7.1912336349487305, \"eval_ref\": \"indicparam_bodo__c51dbb5d__s000__tb131072__v000\", \"oracle_delta\": 0.489161... | {
"index": "indicparam_bodo__c51dbb5d__s000__tb131072__v000",
"uid": "indicparam_bodo__c51dbb5d__s000__tb131072__v000",
"axis": "instruction",
"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\": \"indicparam_marathi__fc3ba376__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1655.9357474567837, \"nll_base\": 7.412121534347534, \"eval_ref\": \"indicparam_marathi__fc3ba376__s000__tb65536__v000\", \"oracle_delta\": 0.1451... | {
"index": "indicparam_marathi__fc3ba376__s000__tb65536__v000",
"uid": "indicparam_marathi__fc3ba376__s000__tb65536__v000",
"axis": "instruction",
"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\": \"indicparam_santali__fa0e11b3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1507.2571306256227, \"nll_base\": 7.318046808242798, \"eval_ref\": \"indicparam_santali__fa0e11b3__s000__tb65536__v000\", \"oracle_delta\": 0.1403... | {
"index": "indicparam_santali__fa0e11b3__s000__tb65536__v000",
"uid": "indicparam_santali__fa0e11b3__s000__tb65536__v000",
"axis": "instruction",
"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\": \"injongointent_eng_prompt_2__0b38aa35__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 411.4244249864197, \"nll_base\": 6.019625345865886, \"eval_ref\": \"injongointent_eng_prompt_2__0b38aa35__s000__tb131072__v000\", \"orac... | {
"index": "injongointent_eng_prompt_2__0b38aa35__s000__tb131072__v000",
"uid": "injongointent_eng_prompt_2__0b38aa35__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\": \"injongointent_eng_prompt_3__a03a6eff__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 476.6588282639688, \"nll_base\": 6.166800990249172, \"eval_ref\": \"injongointent_eng_prompt_3__a03a6eff__s000__tb65536__v000\", \"oracle_... | {
"index": "injongointent_eng_prompt_3__a03a6eff__s000__tb65536__v000",
"uid": "injongointent_eng_prompt_3__a03a6eff__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\": \"injongointent_hau_prompt_5__71c0070e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 558.400259692145, \"nll_base\": 6.325076016512784, \"eval_ref\": \"injongointent_hau_prompt_5__71c0070e__s000__tb262144__v000\", \"oracl... | {
"index": "injongointent_hau_prompt_5__71c0070e__s000__tb262144__v000",
"uid": "injongointent_hau_prompt_5__71c0070e__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\": \"injongointent_ibo_prompt_2__ad00fe30__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 503.39538221679754, \"nll_base\": 6.22137590947042, \"eval_ref\": \"injongointent_ibo_prompt_2__ad00fe30__s000__tb262144__v000\", \"orac... | {
"index": "injongointent_ibo_prompt_2__ad00fe30__s000__tb262144__v000",
"uid": "injongointent_ibo_prompt_2__ad00fe30__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\": \"injongointent_ibo_prompt_3__3a5abbfd__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 620.4338402010998, \"nll_base\": 6.430418975593508, \"eval_ref\": \"injongointent_ibo_prompt_3__3a5abbfd__s000__tb262144__v000\", \"orac... | {
"index": "injongointent_ibo_prompt_3__3a5abbfd__s000__tb262144__v000",
"uid": "injongointent_ibo_prompt_3__3a5abbfd__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\": \"injongointent_kin_prompt_1__e1c60289__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 459.7695026072027, \"nll_base\": 6.130725282611269, \"eval_ref\": \"injongointent_kin_prompt_1__e1c60289__s000__tb131072__v000\", \"orac... | {
"index": "injongointent_kin_prompt_1__e1c60289__s000__tb131072__v000",
"uid": "injongointent_kin_prompt_1__e1c60289__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\": \"injongointent_kin_prompt_3__60cc02d4__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 480.01691521677657, \"nll_base\": 6.173821343315972, \"eval_ref\": \"injongointent_kin_prompt_3__60cc02d4__s000__tb262144__v000\", \"ora... | {
"index": "injongointent_kin_prompt_3__60cc02d4__s000__tb262144__v000",
"uid": "injongointent_kin_prompt_3__60cc02d4__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\": \"injongointent_lin_prompt_1__160851b9__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 555.4493849398713, \"nll_base\": 6.319777488708496, \"eval_ref\": \"injongointent_lin_prompt_1__160851b9__s000__tb65536__v000\", \"oracle_... | {
"index": "injongointent_lin_prompt_1__160851b9__s000__tb65536__v000",
"uid": "injongointent_lin_prompt_1__160851b9__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\": \"injongointent_lin_prompt_2__67edf467__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 502.40858304003757, \"nll_base\": 6.219413699084566, \"eval_ref\": \"injongointent_lin_prompt_2__67edf467__s000__tb65536__v000\", \"oracle... | {
"index": "injongointent_lin_prompt_2__67edf467__s000__tb65536__v000",
"uid": "injongointent_lin_prompt_2__67edf467__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\": \"injongointent_lin_prompt_4__09f5ac17__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 584.7055016761234, \"nll_base\": 6.37110830453726, \"eval_ref\": \"injongointent_lin_prompt_4__09f5ac17__s000__tb262144__v000\", \"oracl... | {
"index": "injongointent_lin_prompt_4__09f5ac17__s000__tb262144__v000",
"uid": "injongointent_lin_prompt_4__09f5ac17__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\": \"injongointent_lug_prompt_2__1501ac99__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 991.0415641092515, \"nll_base\": 6.898756475033967, \"eval_ref\": \"injongointent_lug_prompt_2__1501ac99__s000__tb65536__v000\", \"oracle_... | {
"index": "injongointent_lug_prompt_2__1501ac99__s000__tb65536__v000",
"uid": "injongointent_lug_prompt_2__1501ac99__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\": \"injongointent_lug_prompt_4__52156d06__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 765.9694773699142, \"nll_base\": 6.6411423221711186, \"eval_ref\": \"injongointent_lug_prompt_4__52156d06__s000__tb65536__v000\", \"oracle... | {
"index": "injongointent_lug_prompt_4__52156d06__s000__tb65536__v000",
"uid": "injongointent_lug_prompt_4__52156d06__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\": \"injongointent_orm_prompt_1__549ce617__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 646.0299473009486, \"nll_base\": 6.470845860759105, \"eval_ref\": \"injongointent_orm_prompt_1__549ce617__s000__tb131072__v000\", \"orac... | {
"index": "injongointent_orm_prompt_1__549ce617__s000__tb131072__v000",
"uid": "injongointent_orm_prompt_1__549ce617__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\": \"injongointent_orm_prompt_2__9cc0d8ca__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 511.34102418056256, \"nll_base\": 6.2370367339163115, \"eval_ref\": \"injongointent_orm_prompt_2__9cc0d8ca__s000__tb262144__v000\", \"or... | {
"index": "injongointent_orm_prompt_2__9cc0d8ca__s000__tb262144__v000",
"uid": "injongointent_orm_prompt_2__9cc0d8ca__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\": \"injongointent_orm_prompt_4__1cd14fbc__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 470.507479195728, \"nll_base\": 6.153811855170563, \"eval_ref\": \"injongointent_orm_prompt_4__1cd14fbc__s000__tb262144__v000\", \"oracl... | {
"index": "injongointent_orm_prompt_4__1cd14fbc__s000__tb262144__v000",
"uid": "injongointent_orm_prompt_4__1cd14fbc__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\": \"injongointent_sna_prompt_4__e9cd5f3a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 590.027158976727, \"nll_base\": 6.380168568004262, \"eval_ref\": \"injongointent_sna_prompt_4__e9cd5f3a__s000__tb262144__v000\", \"oracl... | {
"index": "injongointent_sna_prompt_4__e9cd5f3a__s000__tb262144__v000",
"uid": "injongointent_sna_prompt_4__e9cd5f3a__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\": \"injongointent_wol_prompt_2__b9f30ac2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 433.60523025206265, \"nll_base\": 6.07213451244213, \"eval_ref\": \"injongointent_wol_prompt_2__b9f30ac2__s000__tb65536__v000\", \"oracle_... | {
"index": "injongointent_wol_prompt_2__b9f30ac2__s000__tb65536__v000",
"uid": "injongointent_wol_prompt_2__b9f30ac2__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\": \"injongointent_xho_prompt_3__53930fbb__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 741.9005591795501, \"nll_base\": 6.609215216908028, \"eval_ref\": \"injongointent_xho_prompt_3__53930fbb__s000__tb262144__v000\", \"orac... | {
"index": "injongointent_xho_prompt_3__53930fbb__s000__tb262144__v000",
"uid": "injongointent_xho_prompt_3__53930fbb__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\": \"injongointent_xho_prompt_5__bc1b5fac__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 820.6516820409402, \"nll_base\": 6.710098758820565, \"eval_ref\": \"injongointent_xho_prompt_5__bc1b5fac__s000__tb65536__v000\", \"oracle_... | {
"index": "injongointent_xho_prompt_5__bc1b5fac__s000__tb65536__v000",
"uid": "injongointent_xho_prompt_5__bc1b5fac__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\": \"injongointent_zul_prompt_3__d93fc592__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 403.76116436050904, \"nll_base\": 6.0008235258214615, \"eval_ref\": \"injongointent_zul_prompt_3__d93fc592__s000__tb65536__v000\", \"oracl... | {
"index": "injongointent_zul_prompt_3__d93fc592__s000__tb65536__v000",
"uid": "injongointent_zul_prompt_3__d93fc592__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\": \"inverse_scaling_memo_trap__345058c2__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1572.8036400078793, \"nll_base\": 7.360615063739079, \"eval_ref\": \"inverse_scaling_memo_trap__345058c2__s000__tb131072__v001\", \"oracl... | {
"index": "inverse_scaling_memo_trap__345058c2__s000__tb131072__v001",
"uid": "inverse_scaling_memo_trap__345058c2__s000__tb131072__v001",
"axis": "reasoning",
"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\": \"inverse_scaling_memo_trap__345058c2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1572.8036400078793, \"nll_base\": 7.360615063739079, \"eval_ref\": \"inverse_scaling_memo_trap__345058c2__s000__tb65536__v000\", \"oracle_d... | {
"index": "inverse_scaling_memo_trap__345058c2__s000__tb65536__v000",
"uid": "inverse_scaling_memo_trap__345058c2__s000__tb65536__v000",
"axis": "reasoning",
"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\": \"inverse_scaling_modus_tollens__3e32f142__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 6603.520408253261, \"nll_base\": 8.795358180999756, \"eval_ref\": \"inverse_scaling_modus_tollens__3e32f142__s000__tb65536__v001\", \"o... | {
"index": "inverse_scaling_modus_tollens__3e32f142__s000__tb65536__v001",
"uid": "inverse_scaling_modus_tollens__3e32f142__s000__tb65536__v001",
"axis": "reasoning",
"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\": \"inverse_scaling_neqa__f78aba16__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 117797.3549295537, \"nll_base\": 11.676721096038818, \"eval_ref\": \"inverse_scaling_neqa__f78aba16__s000__tb131072__v002\", \"oracle_delta\":... | {
"index": "inverse_scaling_neqa__f78aba16__s000__tb131072__v002",
"uid": "inverse_scaling_neqa__f78aba16__s000__tb131072__v002",
"axis": "reasoning",
"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\": \"inverse_scaling_neqa__f78aba16__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 117797.3549295537, \"nll_base\": 11.676721096038818, \"eval_ref\": \"inverse_scaling_neqa__f78aba16__s000__tb262144__v000\", \"oracle_delta\":... | {
"index": "inverse_scaling_neqa__f78aba16__s000__tb262144__v000",
"uid": "inverse_scaling_neqa__f78aba16__s000__tb262144__v000",
"axis": "reasoning",
"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\": \"inverse_scaling_neqa__f78aba16__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 117797.3549295537, \"nll_base\": 11.676721096038818, \"eval_ref\": \"inverse_scaling_neqa__f78aba16__s000__tb65536__v001\", \"oracle_delta\": 0.... | {
"index": "inverse_scaling_neqa__f78aba16__s000__tb65536__v001",
"uid": "inverse_scaling_neqa__f78aba16__s000__tb65536__v001",
"axis": "reasoning",
"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\": \"inverse_scaling_quote_repetition__35a0cf43__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1493.503555273754, \"nll_base\": 7.308880018151325, \"eval_ref\": \"inverse_scaling_quote_repetition__35a0cf43__s000__tb65536__v001\... | {
"index": "inverse_scaling_quote_repetition__35a0cf43__s000__tb65536__v001",
"uid": "inverse_scaling_quote_repetition__35a0cf43__s000__tb65536__v001",
"axis": "reasoning",
"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\": \"inverse_scaling_redefine_math__c1401777__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2162.761945869405, \"nll_base\": 7.679141362508138, \"eval_ref\": \"inverse_scaling_redefine_math__c1401777__s000__tb131072__v001\", ... | {
"index": "inverse_scaling_redefine_math__c1401777__s000__tb131072__v001",
"uid": "inverse_scaling_redefine_math__c1401777__s000__tb131072__v001",
"axis": "reasoning",
"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\": \"inverse_scaling_redefine_math__c1401777__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2162.761945869405, \"nll_base\": 7.679141362508138, \"eval_ref\": \"inverse_scaling_redefine_math__c1401777__s000__tb65536__v000\", \"o... | {
"index": "inverse_scaling_redefine_math__c1401777__s000__tb65536__v000",
"uid": "inverse_scaling_redefine_math__c1401777__s000__tb65536__v000",
"axis": "reasoning",
"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\": \"inverse_scaling_winobias_antistereotype__e783279b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 18.10949902023195, \"nll_base\": 2.896436608355978, \"eval_ref\": \"inverse_scaling_winobias_antistereotype__e783279b__s000__... | {
"index": "inverse_scaling_winobias_antistereotype__e783279b__s000__tb65536__v001",
"uid": "inverse_scaling_winobias_antistereotype__e783279b__s000__tb65536__v001",
"axis": "reasoning",
"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\": \"jfinqa_numerical__d3c4aa3c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 33.608209833158135, \"nll_base\": 3.5147703773954992, \"eval_ref\": \"jfinqa_numerical__d3c4aa3c__s000__tb65536__v000\", \"oracle_delta\": 0.0945709... | {
"index": "jfinqa_numerical__d3c4aa3c__s000__tb65536__v000",
"uid": "jfinqa_numerical__d3c4aa3c__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\": \"jfinqa_temporal__24b0e909__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 34.20952092784509, \"nll_base\": 3.532503995028409, \"eval_ref\": \"jfinqa_temporal__24b0e909__s000__tb131072__v001\", \"oracle_delta\": 0.22837250... | {
"index": "jfinqa_temporal__24b0e909__s000__tb131072__v001",
"uid": "jfinqa_temporal__24b0e909__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\": \"jfinqa_temporal__24b0e909__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 34.20952092784509, \"nll_base\": 3.532503995028409, \"eval_ref\": \"jfinqa_temporal__24b0e909__s000__tb262144__v002\", \"oracle_delta\": 0.31047115... | {
"index": "jfinqa_temporal__24b0e909__s000__tb262144__v002",
"uid": "jfinqa_temporal__24b0e909__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\": \"jfinqa_temporal__24b0e909__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 34.20952092784509, \"nll_base\": 3.532503995028409, \"eval_ref\": \"jfinqa_temporal__24b0e909__s000__tb65536__v000\", \"oracle_delta\": 0.13875706310... | {
"index": "jfinqa_temporal__24b0e909__s000__tb65536__v000",
"uid": "jfinqa_temporal__24b0e909__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\": \"jfinqa_temporal__24b0e909__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 32.47387060185295, \"nll_base\": 3.480435784487801, \"eval_ref\": \"jfinqa_temporal__24b0e909__s001__tb65536__v003\", \"oracle_delta\": 0.11917897596... | {
"index": "jfinqa_temporal__24b0e909__s001__tb65536__v003",
"uid": "jfinqa_temporal__24b0e909__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\": \"kbl_bar_exam_em_criminal_2020__d0729a12__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 703.4992085106462, \"nll_base\": 6.5560667514801025, \"eval_ref\": \"kbl_bar_exam_em_criminal_2020__d0729a12__s000__tb65536__v000\", \"... | {
"index": "kbl_bar_exam_em_criminal_2020__d0729a12__s000__tb65536__v000",
"uid": "kbl_bar_exam_em_criminal_2020__d0729a12__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\": \"kbl_bar_exam_em_criminal_2024__1ea0c8fb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 782.2869916010961, \"nll_base\": 6.662221670150757, \"eval_ref\": \"kbl_bar_exam_em_criminal_2024__1ea0c8fb__s000__tb65536__v000\", \"o... | {
"index": "kbl_bar_exam_em_criminal_2024__1ea0c8fb__s000__tb65536__v000",
"uid": "kbl_bar_exam_em_criminal_2024__1ea0c8fb__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\": \"kbl_legal_concept_qa_em__0e1f20d5__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3561.2094227292487, \"nll_base\": 8.177855491638184, \"eval_ref\": \"kbl_legal_concept_qa_em__0e1f20d5__s000__tb131072__v002\", \"oracle_de... | {
"index": "kbl_legal_concept_qa_em__0e1f20d5__s000__tb131072__v002",
"uid": "kbl_legal_concept_qa_em__0e1f20d5__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\": \"kbl_legal_concept_qa_em__0e1f20d5__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3561.2094227292487, \"nll_base\": 8.177855491638184, \"eval_ref\": \"kbl_legal_concept_qa_em__0e1f20d5__s000__tb65536__v001\", \"oracle_delta... | {
"index": "kbl_legal_concept_qa_em__0e1f20d5__s000__tb65536__v001",
"uid": "kbl_legal_concept_qa_em__0e1f20d5__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\": \"kobest_boolq__7ad501c4__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 6.529061671810297, \"nll_base\": 1.8762632379877753, \"eval_ref\": \"kobest_boolq__7ad501c4__s000__tb131072__v000\", \"oracle_delta\": 0.1673372100671... | {
"index": "kobest_boolq__7ad501c4__s000__tb131072__v000",
"uid": "kobest_boolq__7ad501c4__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\": \"kormedmcqa_doctor__421f20b9__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 480.19610586698803, \"nll_base\": 6.174194574356079, \"eval_ref\": \"kormedmcqa_doctor__421f20b9__s000__tb131072__v001\", \"oracle_delta\": 0.278... | {
"index": "kormedmcqa_doctor__421f20b9__s000__tb131072__v001",
"uid": "kormedmcqa_doctor__421f20b9__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\": \"kormedmcqa_doctor__421f20b9__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 480.19610586698803, \"nll_base\": 6.174194574356079, \"eval_ref\": \"kormedmcqa_doctor__421f20b9__s000__tb65536__v000\", \"oracle_delta\": 0.103222... | {
"index": "kormedmcqa_doctor__421f20b9__s000__tb65536__v000",
"uid": "kormedmcqa_doctor__421f20b9__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\": \"kormedmcqa_doctor__421f20b9__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 667.5254192210824, \"nll_base\": 6.503577470779419, \"eval_ref\": \"kormedmcqa_doctor__421f20b9__s001__tb65536__v003\", \"oracle_delta\": 0.1256601... | {
"index": "kormedmcqa_doctor__421f20b9__s001__tb65536__v003",
"uid": "kormedmcqa_doctor__421f20b9__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\": \"lambada_openai__4ad5aa20__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 78617.44453112964, \"nll_base\": 11.27234889439174, \"eval_ref\": \"lambada_openai__4ad5aa20__s000__tb262144__v002\", \"oracle_delta\": 1.1319004603... | {
"index": "lambada_openai__4ad5aa20__s000__tb262144__v002",
"uid": "lambada_openai__4ad5aa20__s000__tb262144__v002",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s001__tb131072__v004\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 46942.5403565442, \"nll_base\": 10.756679587168236, \"eval_ref\": \"lambada_openai__4ad5aa20__s001__tb131072__v004\", \"oracle_delta\": 0.2848418771... | {
"index": "lambada_openai__4ad5aa20__s001__tb131072__v004",
"uid": "lambada_openai__4ad5aa20__s001__tb131072__v004",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s001__tb262144__v005\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 46942.5403565442, \"nll_base\": 10.756679587168236, \"eval_ref\": \"lambada_openai__4ad5aa20__s001__tb262144__v005\", \"oracle_delta\": 0.9139512989... | {
"index": "lambada_openai__4ad5aa20__s001__tb262144__v005",
"uid": "lambada_openai__4ad5aa20__s001__tb262144__v005",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s004__tb262144__v014\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 42390.740906891035, \"nll_base\": 10.654685242535317, \"eval_ref\": \"lambada_openai__4ad5aa20__s004__tb262144__v014\", \"oracle_delta\": 0.77032846... | {
"index": "lambada_openai__4ad5aa20__s004__tb262144__v014",
"uid": "lambada_openai__4ad5aa20__s004__tb262144__v014",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s006__tb131072__v019\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 58064.29404578835, \"nll_base\": 10.969306193606954, \"eval_ref\": \"lambada_openai__4ad5aa20__s006__tb131072__v019\", \"oracle_delta\": 0.420150112... | {
"index": "lambada_openai__4ad5aa20__s006__tb131072__v019",
"uid": "lambada_openai__4ad5aa20__s006__tb131072__v019",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s006__tb65536__v018\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 58064.29404578835, \"nll_base\": 10.969306193606954, \"eval_ref\": \"lambada_openai__4ad5aa20__s006__tb65536__v018\", \"oracle_delta\": 0.128155345648... | {
"index": "lambada_openai__4ad5aa20__s006__tb65536__v018",
"uid": "lambada_openai__4ad5aa20__s006__tb65536__v018",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s007__tb131072__v022\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 15852.224869509755, \"nll_base\": 9.671065139770509, \"eval_ref\": \"lambada_openai__4ad5aa20__s007__tb131072__v022\", \"oracle_delta\": 0.623996734... | {
"index": "lambada_openai__4ad5aa20__s007__tb131072__v022",
"uid": "lambada_openai__4ad5aa20__s007__tb131072__v022",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s007__tb262144__v023\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 15852.224869509755, \"nll_base\": 9.671065139770509, \"eval_ref\": \"lambada_openai__4ad5aa20__s007__tb262144__v023\", \"oracle_delta\": 1.218675613... | {
"index": "lambada_openai__4ad5aa20__s007__tb262144__v023",
"uid": "lambada_openai__4ad5aa20__s007__tb262144__v023",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s007__tb65536__v021\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 15852.224869509755, \"nll_base\": 9.671065139770509, \"eval_ref\": \"lambada_openai__4ad5aa20__s007__tb65536__v021\", \"oracle_delta\": 0.281407165527... | {
"index": "lambada_openai__4ad5aa20__s007__tb65536__v021",
"uid": "lambada_openai__4ad5aa20__s007__tb65536__v021",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s008__tb131072__v025\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 26586.192841356293, \"nll_base\": 10.188147293893914, \"eval_ref\": \"lambada_openai__4ad5aa20__s008__tb131072__v025\", \"oracle_delta\": 0.70443524... | {
"index": "lambada_openai__4ad5aa20__s008__tb131072__v025",
"uid": "lambada_openai__4ad5aa20__s008__tb131072__v025",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s008__tb262144__v026\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 26586.192841356293, \"nll_base\": 10.188147293893914, \"eval_ref\": \"lambada_openai__4ad5aa20__s008__tb262144__v026\", \"oracle_delta\": 1.21653586... | {
"index": "lambada_openai__4ad5aa20__s008__tb262144__v026",
"uid": "lambada_openai__4ad5aa20__s008__tb262144__v026",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s008__tb65536__v024\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 26586.192841356293, \"nll_base\": 10.188147293893914, \"eval_ref\": \"lambada_openai__4ad5aa20__s008__tb65536__v024\", \"oracle_delta\": 0.26505038612... | {
"index": "lambada_openai__4ad5aa20__s008__tb65536__v024",
"uid": "lambada_openai__4ad5aa20__s008__tb65536__v024",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s009__tb131072__v028\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 37104.8218795084, \"nll_base\": 10.521502209948254, \"eval_ref\": \"lambada_openai__4ad5aa20__s009__tb131072__v028\", \"oracle_delta\": 0.2888579925... | {
"index": "lambada_openai__4ad5aa20__s009__tb131072__v028",
"uid": "lambada_openai__4ad5aa20__s009__tb131072__v028",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s009__tb262144__v029\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 37104.8218795084, \"nll_base\": 10.521502209948254, \"eval_ref\": \"lambada_openai__4ad5aa20__s009__tb262144__v029\", \"oracle_delta\": 0.9222748991... | {
"index": "lambada_openai__4ad5aa20__s009__tb262144__v029",
"uid": "lambada_openai__4ad5aa20__s009__tb262144__v029",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s012__tb131072__v037\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 24160.18061336233, \"nll_base\": 10.092461127739448, \"eval_ref\": \"lambada_openai__4ad5aa20__s012__tb131072__v037\", \"oracle_delta\": 0.588380491... | {
"index": "lambada_openai__4ad5aa20__s012__tb131072__v037",
"uid": "lambada_openai__4ad5aa20__s012__tb131072__v037",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s012__tb262144__v038\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 24160.18061336233, \"nll_base\": 10.092461127739448, \"eval_ref\": \"lambada_openai__4ad5aa20__s012__tb262144__v038\", \"oracle_delta\": 1.288690740... | {
"index": "lambada_openai__4ad5aa20__s012__tb262144__v038",
"uid": "lambada_openai__4ad5aa20__s012__tb262144__v038",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s012__tb65536__v036\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 24160.18061336233, \"nll_base\": 10.092461127739448, \"eval_ref\": \"lambada_openai__4ad5aa20__s012__tb65536__v036\", \"oracle_delta\": 0.256405124416... | {
"index": "lambada_openai__4ad5aa20__s012__tb65536__v036",
"uid": "lambada_openai__4ad5aa20__s012__tb65536__v036",
"axis": "domain_ppl",
"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\": \"lambada_openai__4ad5aa20__s013__tb131072__v040\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 24919.386365166323, \"nll_base\": 10.123401348407452, \"eval_ref\": \"lambada_openai__4ad5aa20__s013__tb131072__v040\", \"oracle_delta\": 0.51909813... | {
"index": "lambada_openai__4ad5aa20__s013__tb131072__v040",
"uid": "lambada_openai__4ad5aa20__s013__tb131072__v040",
"axis": "domain_ppl",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.