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\": \"mmlu_es_llama_medical_genetics__5c7807be__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 223.6127864377181, \"nll_base\": 5.409915924072266, \"eval_ref\": \"mmlu_es_llama_medical_genetics__5c7807be__s000__tb131072__v002\"... | {
"index": "mmlu_es_llama_medical_genetics__5c7807be__s000__tb131072__v002",
"uid": "mmlu_es_llama_medical_genetics__5c7807be__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\": \"mmlu_es_llama_medical_genetics__5c7807be__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 223.6127864377181, \"nll_base\": 5.409915924072266, \"eval_ref\": \"mmlu_es_llama_medical_genetics__5c7807be__s000__tb65536__v001\", \... | {
"index": "mmlu_es_llama_medical_genetics__5c7807be__s000__tb65536__v001",
"uid": "mmlu_es_llama_medical_genetics__5c7807be__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\": \"mmlu_es_llama_miscellaneous__4c9488ba__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 278.31856579937147, \"nll_base\": 5.628766377766927, \"eval_ref\": \"mmlu_es_llama_miscellaneous__4c9488ba__s000__tb131072__v000\", \"o... | {
"index": "mmlu_es_llama_miscellaneous__4c9488ba__s000__tb131072__v000",
"uid": "mmlu_es_llama_miscellaneous__4c9488ba__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\": \"mmlu_es_llama_moral_scenarios__ab4da792__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 307.08186078086936, \"nll_base\": 5.727114359537761, \"eval_ref\": \"mmlu_es_llama_moral_scenarios__ab4da792__s000__tb131072__v000\",... | {
"index": "mmlu_es_llama_moral_scenarios__ab4da792__s000__tb131072__v000",
"uid": "mmlu_es_llama_moral_scenarios__ab4da792__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\": \"mmlu_es_llama_nutrition__1d3d3ea1__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 276.3102075612115, \"nll_base\": 5.621524175008138, \"eval_ref\": \"mmlu_es_llama_nutrition__1d3d3ea1__s000__tb131072__v000\", \"oracle_del... | {
"index": "mmlu_es_llama_nutrition__1d3d3ea1__s000__tb131072__v000",
"uid": "mmlu_es_llama_nutrition__1d3d3ea1__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\": \"mmlu_es_llama_professional_accounting__f3a97d01__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 237.7639817940787, \"nll_base\": 5.4712785085042315, \"eval_ref\": \"mmlu_es_llama_professional_accounting__f3a97d01__s000__t... | {
"index": "mmlu_es_llama_professional_accounting__f3a97d01__s000__tb262144__v000",
"uid": "mmlu_es_llama_professional_accounting__f3a97d01__s000__tb262144__v000",
"axis": "instruction",
"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\": \"mmlu_es_llama_professional_law__ef1a261b__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 178.67721134513957, \"nll_base\": 5.185580889383952, \"eval_ref\": \"mmlu_es_llama_professional_law__ef1a261b__s000__tb262144__v000\... | {
"index": "mmlu_es_llama_professional_law__ef1a261b__s000__tb262144__v000",
"uid": "mmlu_es_llama_professional_law__ef1a261b__s000__tb262144__v000",
"axis": "instruction",
"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\": \"mmlu_es_llama_professional_medicine__4215f025__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 281.31334272218817, \"nll_base\": 5.639469146728516, \"eval_ref\": \"mmlu_es_llama_professional_medicine__4215f025__s000__tb262... | {
"index": "mmlu_es_llama_professional_medicine__4215f025__s000__tb262144__v000",
"uid": "mmlu_es_llama_professional_medicine__4215f025__s000__tb262144__v000",
"axis": "instruction",
"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\": \"mmlu_es_llama_professional_psychology__1008a613__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 273.57090029833427, \"nll_base\": 5.611560821533203, \"eval_ref\": \"mmlu_es_llama_professional_psychology__1008a613__s000__tb6... | {
"index": "mmlu_es_llama_professional_psychology__1008a613__s000__tb65536__v000",
"uid": "mmlu_es_llama_professional_psychology__1008a613__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\": \"mmlu_es_llama_security_studies__a5420c06__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 182.9513172514484, \"nll_base\": 5.209220091501872, \"eval_ref\": \"mmlu_es_llama_security_studies__a5420c06__s000__tb65536__v000\", \... | {
"index": "mmlu_es_llama_security_studies__a5420c06__s000__tb65536__v000",
"uid": "mmlu_es_llama_security_studies__a5420c06__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\": \"mmlu_es_llama_sociology__7f4f3044__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 323.71227863816677, \"nll_base\": 5.779855092366536, \"eval_ref\": \"mmlu_es_llama_sociology__7f4f3044__s000__tb65536__v000\", \"oracle_delta... | {
"index": "mmlu_es_llama_sociology__7f4f3044__s000__tb65536__v000",
"uid": "mmlu_es_llama_sociology__7f4f3044__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\": \"mmlu_es_llama_world_religions__1b0c27eb__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 292.43526302751843, \"nll_base\": 5.678243319193522, \"eval_ref\": \"mmlu_es_llama_world_religions__1b0c27eb__s000__tb131072__v000\",... | {
"index": "mmlu_es_llama_world_religions__1b0c27eb__s000__tb131072__v000",
"uid": "mmlu_es_llama_world_religions__1b0c27eb__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\": \"mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 67.49596663134892, \"nll_base\": 4.2120678424835205, \"eval_ref\": \"mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb131072__v001\",... | {
"index": "mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 67.49596663134892, \"nll_base\": 4.2120678424835205, \"eval_ref\": \"mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb65536__v000\", \"... | {
"index": "mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_anatomy__79e40ffc__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_astronomy__4641d9f2__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 81.02366696664602, \"nll_base\": 4.3947412967681885, \"eval_ref\": \"mmlu_flan_cot_fewshot_astronomy__4641d9f2__s000__tb131072__v00... | {
"index": "mmlu_flan_cot_fewshot_astronomy__4641d9f2__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_astronomy__4641d9f2__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 85.07897239751283, \"nll_base\": 4.443579912185669, \"eval_ref\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s0... | {
"index": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb131072__v002",
"uid": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__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\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 85.07897239751283, \"nll_base\": 4.443579912185669, \"eval_ref\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s0... | {
"index": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__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\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 85.07897239751283, \"nll_base\": 4.443579912185669, \"eval_ref\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000... | {
"index": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__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\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s001__tb262144__v003\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 93.65267397496707, \"nll_base\": 4.539592981338501, \"eval_ref\": \"mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s0... | {
"index": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__s001__tb262144__v003",
"uid": "mmlu_flan_cot_fewshot_clinical_knowledge__16a22714__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\": \"mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 77.30814364027137, \"nll_base\": 4.347799301147461, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb... | {
"index": "mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 77.30814364027137, \"nll_base\": 4.347799301147461, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb65... | {
"index": "mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_college_biology__90fedfbc__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 73.4909518374028, \"nll_base\": 4.297162294387817, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000_... | {
"index": "mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 73.4909518374028, \"nll_base\": 4.297162294387817, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__t... | {
"index": "mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_college_chemistry__d4448e23__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_mathematics__77237305__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 70.61605040783161, \"nll_base\": 4.257257461547852, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_mathematics__77237305__... | {
"index": "mmlu_flan_cot_fewshot_college_mathematics__77237305__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_college_mathematics__77237305__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_medicine__74412a8a__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 79.37405053976946, \"nll_base\": 4.374171495437622, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_medicine__74412a8a__s000__tb... | {
"index": "mmlu_flan_cot_fewshot_college_medicine__74412a8a__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_college_medicine__74412a8a__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_college_physics__f6008c19__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 76.94301042087984, \"nll_base\": 4.343065023422241, \"eval_ref\": \"mmlu_flan_cot_fewshot_college_physics__f6008c19__s000__tb... | {
"index": "mmlu_flan_cot_fewshot_college_physics__f6008c19__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_college_physics__f6008c19__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 88.02569036910543, \"nll_base\": 4.477628707885742, \"eval_ref\": \"mmlu_flan_cot_fewshot_computer_security__c47d0162__s000... | {
"index": "mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 88.02569036910543, \"nll_base\": 4.477628707885742, \"eval_ref\": \"mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__... | {
"index": "mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_computer_security__c47d0162__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_econometrics__2941e1ec__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 93.59381212713883, \"nll_base\": 4.53896427154541, \"eval_ref\": \"mmlu_flan_cot_fewshot_econometrics__2941e1ec__s000__tb65536__v0... | {
"index": "mmlu_flan_cot_fewshot_econometrics__2941e1ec__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_econometrics__2941e1ec__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_electrical_engineering__8a2d7d3d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 68.69608568578896, \"nll_base\": 4.229692220687866, \"eval_ref\": \"mmlu_flan_cot_fewshot_electrical_engineering__8a2d7d... | {
"index": "mmlu_flan_cot_fewshot_electrical_engineering__8a2d7d3d__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_electrical_engineering__8a2d7d3d__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_elementary_mathematics__529ea1fe__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 84.61992151471348, \"nll_base\": 4.438169717788696, \"eval_ref\": \"mmlu_flan_cot_fewshot_elementary_mathematics__529ea1... | {
"index": "mmlu_flan_cot_fewshot_elementary_mathematics__529ea1fe__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_elementary_mathematics__529ea1fe__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 71.2893535953784, \"nll_base\": 4.266746997833252, \"eval_ref\": \"mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb131072_... | {
"index": "mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 71.2893535953784, \"nll_base\": 4.266746997833252, \"eval_ref\": \"mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb65536__v0... | {
"index": "mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_formal_logic__dda47d14__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_global_facts__a2208502__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 107.37820937747861, \"nll_base\": 4.676357269287109, \"eval_ref\": \"mmlu_flan_cot_fewshot_global_facts__a2208502__s000__tb65536__... | {
"index": "mmlu_flan_cot_fewshot_global_facts__a2208502__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_global_facts__a2208502__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 75.82306388273848, \"nll_base\": 4.328402519226074, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__... | {
"index": "mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 75.82306388273848, \"nll_base\": 4.328402519226074, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s0... | {
"index": "mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_biology__be5a12f3__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_european_history__07849b79__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 90.92981697456379, \"nll_base\": 4.510087966918945, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_european_h... | {
"index": "mmlu_flan_cot_fewshot_high_school_european_history__07849b79__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_european_history__07849b79__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_geography__2a01a0ab__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 73.21770347723175, \"nll_base\": 4.293437242507935, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_geography__2a01a0... | {
"index": "mmlu_flan_cot_fewshot_high_school_geography__2a01a0ab__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_geography__2a01a0ab__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_government_and_politics__85ec7914__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 67.89418186576172, \"nll_base\": 4.217950344085693, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_gov... | {
"index": "mmlu_flan_cot_fewshot_high_school_government_and_politics__85ec7914__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_government_and_politics__85ec7914__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_macroeconomics__a47625ef__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 84.20822725438748, \"nll_base\": 4.433292627334595, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_macroeconomi... | {
"index": "mmlu_flan_cot_fewshot_high_school_macroeconomics__a47625ef__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_macroeconomics__a47625ef__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 73.48709718466692, \"nll_base\": 4.297109842300415, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_mathematics__20... | {
"index": "mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 73.48709718466692, \"nll_base\": 4.297109842300415, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_mathematics__20fd... | {
"index": "mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_mathematics__20fd10af__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_microeconomics__9bb96453__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 82.64697300906386, \"nll_base\": 4.414578199386597, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_microeconomi... | {
"index": "mmlu_flan_cot_fewshot_high_school_microeconomics__9bb96453__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_microeconomics__9bb96453__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_psychology__cb69aefa__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 79.62215590517386, \"nll_base\": 4.3772923946380615, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_psychology__cb6... | {
"index": "mmlu_flan_cot_fewshot_high_school_psychology__cb69aefa__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_psychology__cb69aefa__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 84.72315919222183, \"nll_base\": 4.439388990402222, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_statistics__44f6... | {
"index": "mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 84.72315919222183, \"nll_base\": 4.439388990402222, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_statistics__44f680... | {
"index": "mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_statistics__44f68013__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 75.7491443653664, \"nll_base\": 4.32742714881897, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_us_history__f4965d... | {
"index": "mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 75.7491443653664, \"nll_base\": 4.32742714881897, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_us_history__f4965d... | {
"index": "mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb262144__v001",
"uid": "mmlu_flan_cot_fewshot_high_school_us_history__f4965d13__s000__tb262144__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 86.31003085046784, \"nll_base\": 4.457945823669434, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_world_history... | {
"index": "mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 86.31003085046784, \"nll_base\": 4.457945823669434, \"eval_ref\": \"mmlu_flan_cot_fewshot_high_school_world_history__... | {
"index": "mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_high_school_world_history__868a59d9__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_human_aging__81b14ea0__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 86.15883345238223, \"nll_base\": 4.456192493438721, \"eval_ref\": \"mmlu_flan_cot_fewshot_human_aging__81b14ea0__s000__tb131072__... | {
"index": "mmlu_flan_cot_fewshot_human_aging__81b14ea0__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_human_aging__81b14ea0__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_human_sexuality__5c288e17__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 90.3937768230748, \"nll_base\": 4.504175424575806, \"eval_ref\": \"mmlu_flan_cot_fewshot_human_sexuality__5c288e17__s000__tb655... | {
"index": "mmlu_flan_cot_fewshot_human_sexuality__5c288e17__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_human_sexuality__5c288e17__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_international_law__a1fd15af__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 87.1461432941564, \"nll_base\": 4.467586517333984, \"eval_ref\": \"mmlu_flan_cot_fewshot_international_law__a1fd15af__s000__t... | {
"index": "mmlu_flan_cot_fewshot_international_law__a1fd15af__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_international_law__a1fd15af__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 75.21897873408804, \"nll_base\": 4.320403575897217, \"eval_ref\": \"mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb1310... | {
"index": "mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 75.21897873408804, \"nll_base\": 4.320403575897217, \"eval_ref\": \"mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb65536_... | {
"index": "mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_jurisprudence__b878c901__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_logical_fallacies__67c3d99a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 79.05802190282883, \"nll_base\": 4.370182037353516, \"eval_ref\": \"mmlu_flan_cot_fewshot_logical_fallacies__67c3d99a__s000... | {
"index": "mmlu_flan_cot_fewshot_logical_fallacies__67c3d99a__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_logical_fallacies__67c3d99a__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_management__c4d4c6da__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 80.36684266653586, \"nll_base\": 4.386601686477661, \"eval_ref\": \"mmlu_flan_cot_fewshot_management__c4d4c6da__s000__tb65536__v001\... | {
"index": "mmlu_flan_cot_fewshot_management__c4d4c6da__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_management__c4d4c6da__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 77.73861192043196, \"nll_base\": 4.353352069854736, \"eval_ref\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__... | {
"index": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb131072__v002",
"uid": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__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\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 77.73861192043196, \"nll_base\": 4.353352069854736, \"eval_ref\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__... | {
"index": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__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\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 77.73861192043196, \"nll_base\": 4.353352069854736, \"eval_ref\": \"mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb... | {
"index": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_medical_genetics__5ff1a07d__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\": \"mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 89.9190616097546, \"nll_base\": 4.498909950256348, \"eval_ref\": \"mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb26214... | {
"index": "mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 89.9190616097546, \"nll_base\": 4.498909950256348, \"eval_ref\": \"mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb65536__... | {
"index": "mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_miscellaneous__c303bf2f__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_moral_disputes__60b2e229__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 80.02581360589733, \"nll_base\": 4.382349252700806, \"eval_ref\": \"mmlu_flan_cot_fewshot_moral_disputes__60b2e229__s000__tb6553... | {
"index": "mmlu_flan_cot_fewshot_moral_disputes__60b2e229__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_moral_disputes__60b2e229__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_nutrition__38cc14bb__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 99.32191688343337, \"nll_base\": 4.5983662605285645, \"eval_ref\": \"mmlu_flan_cot_fewshot_nutrition__38cc14bb__s000__tb262144__v00... | {
"index": "mmlu_flan_cot_fewshot_nutrition__38cc14bb__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_nutrition__38cc14bb__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 83.83318694578789, \"nll_base\": 4.428828954696655, \"eval_ref\": \"mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb131072__v0... | {
"index": "mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 83.83318694578789, \"nll_base\": 4.428828954696655, \"eval_ref\": \"mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb65536__v000\... | {
"index": "mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_philosophy__71fbb6d3__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 68.5584321119084, \"nll_base\": 4.227686405181885, \"eval_ref\": \"mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb131072__v00... | {
"index": "mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 68.5584321119084, \"nll_base\": 4.227686405181885, \"eval_ref\": \"mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb262144__v00... | {
"index": "mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb262144__v001",
"uid": "mmlu_flan_cot_fewshot_prehistory__05fe319d__s000__tb262144__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_professional_accounting__f460b832__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 78.61486805205, \"nll_base\": 4.364560842514038, \"eval_ref\": \"mmlu_flan_cot_fewshot_professional_accounting__f460b... | {
"index": "mmlu_flan_cot_fewshot_professional_accounting__f460b832__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_professional_accounting__f460b832__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 104.55470390641608, \"nll_base\": 4.649710416793823, \"eval_ref\": \"mmlu_flan_cot_fewshot_professional_medicine__1ad06... | {
"index": "mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb131072__v001",
"uid": "mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 104.55470390641608, \"nll_base\": 4.649710416793823, \"eval_ref\": \"mmlu_flan_cot_fewshot_professional_medicine__1ad06fd... | {
"index": "mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb65536__v000",
"uid": "mmlu_flan_cot_fewshot_professional_medicine__1ad06fdc__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_sociology__1b130e4e__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 83.13090994667829, \"nll_base\": 4.420416593551636, \"eval_ref\": \"mmlu_flan_cot_fewshot_sociology__1b130e4e__s000__tb65536__v001\",... | {
"index": "mmlu_flan_cot_fewshot_sociology__1b130e4e__s000__tb65536__v001",
"uid": "mmlu_flan_cot_fewshot_sociology__1b130e4e__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_virology__fe4c8163__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 89.45581803544556, \"nll_base\": 4.493744850158691, \"eval_ref\": \"mmlu_flan_cot_fewshot_virology__fe4c8163__s000__tb262144__v000\"... | {
"index": "mmlu_flan_cot_fewshot_virology__fe4c8163__s000__tb262144__v000",
"uid": "mmlu_flan_cot_fewshot_virology__fe4c8163__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_fewshot_world_religions__8f73e7d8__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 89.6103442246332, \"nll_base\": 4.495470762252808, \"eval_ref\": \"mmlu_flan_cot_fewshot_world_religions__8f73e7d8__s000__tb1... | {
"index": "mmlu_flan_cot_fewshot_world_religions__8f73e7d8__s000__tb131072__v000",
"uid": "mmlu_flan_cot_fewshot_world_religions__8f73e7d8__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 68.9927754471425, \"nll_base\": 4.234001795450847, \"eval_ref\": \"mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb131072__v001\",... | {
"index": "mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 68.9927754471425, \"nll_base\": 4.234001795450847, \"eval_ref\": \"mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb65536__v000\", \"... | {
"index": "mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_anatomy__340f169b__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_astronomy__bc2f1c10__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 73.3180945193846, \"nll_base\": 4.294807434082031, \"eval_ref\": \"mmlu_flan_cot_zeroshot_astronomy__bc2f1c10__s000__tb131072__v00... | {
"index": "mmlu_flan_cot_zeroshot_astronomy__bc2f1c10__s000__tb131072__v000",
"uid": "mmlu_flan_cot_zeroshot_astronomy__bc2f1c10__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 91.40374032397787, \"nll_base\": 4.515286400204613, \"eval_ref\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__... | {
"index": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__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\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 91.40374032397787, \"nll_base\": 4.515286400204613, \"eval_ref\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__... | {
"index": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb262144__v002",
"uid": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__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\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 91.40374032397787, \"nll_base\": 4.515286400204613, \"eval_ref\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s0... | {
"index": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__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\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 90.54393615296468, \"nll_base\": 4.505835215250651, \"eval_ref\": \"mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s0... | {
"index": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__s001__tb65536__v003",
"uid": "mmlu_flan_cot_zeroshot_clinical_knowledge__2d3fbbc9__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\": \"mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 57.94650426083341, \"nll_base\": 4.059520244598389, \"eval_ref\": \"mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__... | {
"index": "mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 57.94650426083341, \"nll_base\": 4.059520244598389, \"eval_ref\": \"mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb... | {
"index": "mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_college_biology__db26a45c__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_college_mathematics__82de7b2c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 70.05976338483867, \"nll_base\": 4.2493486404418945, \"eval_ref\": \"mmlu_flan_cot_zeroshot_college_mathematics__82de7b2c_... | {
"index": "mmlu_flan_cot_zeroshot_college_mathematics__82de7b2c__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_college_mathematics__82de7b2c__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_college_medicine__f003e23a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 89.66325596048449, \"nll_base\": 4.496061052594866, \"eval_ref\": \"mmlu_flan_cot_zeroshot_college_medicine__f003e23a__s000... | {
"index": "mmlu_flan_cot_zeroshot_college_medicine__f003e23a__s000__tb262144__v000",
"uid": "mmlu_flan_cot_zeroshot_college_medicine__f003e23a__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_college_physics__48d7ab73__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 81.34774707329386, \"nll_base\": 4.398733139038086, \"eval_ref\": \"mmlu_flan_cot_zeroshot_college_physics__48d7ab73__s000__tb... | {
"index": "mmlu_flan_cot_zeroshot_college_physics__48d7ab73__s000__tb65536__v001",
"uid": "mmlu_flan_cot_zeroshot_college_physics__48d7ab73__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_computer_security__0ba4c1c0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 108.31489716652489, \"nll_base\": 4.68504269917806, \"eval_ref\": \"mmlu_flan_cot_zeroshot_computer_security__0ba4c1c0__s0... | {
"index": "mmlu_flan_cot_zeroshot_computer_security__0ba4c1c0__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_computer_security__0ba4c1c0__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 82.95171296469474, \"nll_base\": 4.4182586669921875, \"eval_ref\": \"mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb131... | {
"index": "mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 82.95171296469474, \"nll_base\": 4.4182586669921875, \"eval_ref\": \"mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb65536... | {
"index": "mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_econometrics__d5d313c2__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 68.05189893452675, \"nll_base\": 4.22027063369751, \"eval_ref\": \"mmlu_flan_cot_zeroshot_electrical_engineering__908... | {
"index": "mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb131072__v001",
"uid": "mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb131072__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 68.05189893452675, \"nll_base\": 4.22027063369751, \"eval_ref\": \"mmlu_flan_cot_zeroshot_electrical_engineering__908c0... | {
"index": "mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_electrical_engineering__908c093d__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_elementary_mathematics__07fc9a41__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 72.0592336285056, \"nll_base\": 4.277488470077515, \"eval_ref\": \"mmlu_flan_cot_zeroshot_elementary_mathematics__07fc9... | {
"index": "mmlu_flan_cot_zeroshot_elementary_mathematics__07fc9a41__s000__tb65536__v000",
"uid": "mmlu_flan_cot_zeroshot_elementary_mathematics__07fc9a41__s000__tb65536__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 119.8153268384347, \"nll_base\": 4.785951614379883, \"eval_ref\": \"mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb2621... | {
"index": "mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb262144__v000",
"uid": "mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 119.8153268384347, \"nll_base\": 4.785951614379883, \"eval_ref\": \"mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb65536_... | {
"index": "mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb65536__v001",
"uid": "mmlu_flan_cot_zeroshot_global_facts__e9d120da__s000__tb65536__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 81.65885142735617, \"nll_base\": 4.402550220489502, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_biology__2739396e... | {
"index": "mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb131072__v000",
"uid": "mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 81.65885142735617, \"nll_base\": 4.402550220489502, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_biology__2739396e... | {
"index": "mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb262144__v001",
"uid": "mmlu_flan_cot_zeroshot_high_school_biology__2739396e__s000__tb262144__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_european_history__162950a5__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 71.5941696625759, \"nll_base\": 4.271013641357422, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_european_... | {
"index": "mmlu_flan_cot_zeroshot_high_school_european_history__162950a5__s000__tb262144__v000",
"uid": "mmlu_flan_cot_zeroshot_high_school_european_history__162950a5__s000__tb262144__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 73.90010225151099, \"nll_base\": 4.3027142116001675, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_geography__03b... | {
"index": "mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb131072__v000",
"uid": "mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb131072__v000",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 73.90010225151099, \"nll_base\": 4.3027142116001675, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_geography__03b... | {
"index": "mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb262144__v001",
"uid": "mmlu_flan_cot_zeroshot_high_school_geography__03b2553d__s000__tb262144__v001",
"axis": "knowledge",
"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\": \"mmlu_flan_cot_zeroshot_high_school_government_and_politics__23c593f1__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 65.9069284984172, \"nll_base\": 4.188243572528545, \"eval_ref\": \"mmlu_flan_cot_zeroshot_high_school_go... | {
"index": "mmlu_flan_cot_zeroshot_high_school_government_and_politics__23c593f1__s000__tb131072__v000",
"uid": "mmlu_flan_cot_zeroshot_high_school_government_and_politics__23c593f1__s000__tb131072__v000",
"axis": "knowledge",
"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.