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\": \"bigbench_general_knowledge_multiple_choice__29420800__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1759.0900239270036, \"nll_base\": 7.472551922465479, \"eval_ref\": \"bigbench_general_knowledge_multiple_choice__2942080... | {
"index": "bigbench_general_knowledge_multiple_choice__29420800__s000__tb262144__v000",
"uid": "bigbench_general_knowledge_multiple_choice__29420800__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_general_knowledge_multiple_choice__29420800__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1759.0900239270036, \"nll_base\": 7.472551922465479, \"eval_ref\": \"bigbench_general_knowledge_multiple_choice__29420800_... | {
"index": "bigbench_general_knowledge_multiple_choice__29420800__s000__tb65536__v001",
"uid": "bigbench_general_knowledge_multiple_choice__29420800__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_goal_step_wikihow_multiple_choice__5edf3035__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 8.244464814417617, \"nll_base\": 2.109542043585526, \"eval_ref\": \"bigbench_goal_step_wikihow_multiple_choice__5edf3035__... | {
"index": "bigbench_goal_step_wikihow_multiple_choice__5edf3035__s000__tb65536__v001",
"uid": "bigbench_goal_step_wikihow_multiple_choice__5edf3035__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3.711286762868488, \"nll_base\": 1.311378652838212, \"eval_ref\": \"bigbench_gre_reading_comprehension_generate_u... | {
"index": "bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb131072__v001",
"uid": "bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3.711286762868488, \"nll_base\": 1.311378652838212, \"eval_ref\": \"bigbench_gre_reading_comprehension_generate_unt... | {
"index": "bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb65536__v000",
"uid": "bigbench_gre_reading_comprehension_generate_until__8ff14492__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3.640061955629225, \"nll_base\": 1.2920007022810571, \"eval_ref\": \"bigbench_gre_reading_comprehension_multiple... | {
"index": "bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb262144__v000",
"uid": "bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3.640061955629225, \"nll_base\": 1.2920007022810571, \"eval_ref\": \"bigbench_gre_reading_comprehension_multiple_c... | {
"index": "bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb65536__v001",
"uid": "bigbench_gre_reading_comprehension_multiple_choice__03b90b49__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hhh_alignment_generate_until__6916704f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 13.439641491757746, \"nll_base\": 2.598208660018821, \"eval_ref\": \"bigbench_hhh_alignment_generate_until__6916704f__s000__t... | {
"index": "bigbench_hhh_alignment_generate_until__6916704f__s000__tb131072__v001",
"uid": "bigbench_hhh_alignment_generate_until__6916704f__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hhh_alignment_generate_until__6916704f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 13.439641491757746, \"nll_base\": 2.598208660018821, \"eval_ref\": \"bigbench_hhh_alignment_generate_until__6916704f__s000__tb6... | {
"index": "bigbench_hhh_alignment_generate_until__6916704f__s000__tb65536__v000",
"uid": "bigbench_hhh_alignment_generate_until__6916704f__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3.5072038375148438, \"nll_base\": 1.25481909251283, \"eval_ref\": \"bigbench_hindi_question_answering_generate_unt... | {
"index": "bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb262144__v000",
"uid": "bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3.5072038375148438, \"nll_base\": 1.25481909251283, \"eval_ref\": \"bigbench_hindi_question_answering_generate_until... | {
"index": "bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb65536__v001",
"uid": "bigbench_hindi_question_answering_generate_until__1569a9a2__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 6703.23583821943, \"nll_base\": 8.810345649719238, \"eval_ref\": \"bigbench_hinglish_toxicity_generate_until__35c5e1db__s... | {
"index": "bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb131072__v001",
"uid": "bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 6703.23583821943, \"nll_base\": 8.810345649719238, \"eval_ref\": \"bigbench_hinglish_toxicity_generate_until__35c5e1db__s00... | {
"index": "bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb65536__v000",
"uid": "bigbench_hinglish_toxicity_generate_until__35c5e1db__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hinglish_toxicity_multiple_choice__57f3358d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2661.1248099501927, \"nll_base\": 7.886504173278809, \"eval_ref\": \"bigbench_hinglish_toxicity_multiple_choice__57f3358... | {
"index": "bigbench_hinglish_toxicity_multiple_choice__57f3358d__s000__tb131072__v000",
"uid": "bigbench_hinglish_toxicity_multiple_choice__57f3358d__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_human_organs_senses_generate_until__9d714ede__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3343.0488526337617, \"nll_base\": 8.11463849972456, \"eval_ref\": \"bigbench_human_organs_senses_generate_until__9d714e... | {
"index": "bigbench_human_organs_senses_generate_until__9d714ede__s000__tb131072__v001",
"uid": "bigbench_human_organs_senses_generate_until__9d714ede__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_human_organs_senses_generate_until__9d714ede__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3343.0488526337617, \"nll_base\": 8.11463849972456, \"eval_ref\": \"bigbench_human_organs_senses_generate_until__9d714ede... | {
"index": "bigbench_human_organs_senses_generate_until__9d714ede__s000__tb65536__v000",
"uid": "bigbench_human_organs_senses_generate_until__9d714ede__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2879.591960651525, \"nll_base\": 7.965403882763054, \"eval_ref\": \"bigbench_human_organs_senses_multiple_choice__9f35... | {
"index": "bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb131072__v001",
"uid": "bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2879.591960651525, \"nll_base\": 7.965403882763054, \"eval_ref\": \"bigbench_human_organs_senses_multiple_choice__9f35d8... | {
"index": "bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb65536__v000",
"uid": "bigbench_human_organs_senses_multiple_choice__9f35d8ca__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1752.909552225219, \"nll_base\": 7.469032287597656, \"eval_ref\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb1310... | {
"index": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb131072__v002",
"uid": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb131072__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1752.909552225219, \"nll_base\": 7.469032287597656, \"eval_ref\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb2621... | {
"index": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb262144__v000",
"uid": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1752.909552225219, \"nll_base\": 7.469032287597656, \"eval_ref\": \"bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb65536_... | {
"index": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb65536__v001",
"uid": "bigbench_hyperbaton_multiple_choice__a4a31964__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_identify_odd_metaphor_generate_until__b53d8f63__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 7.427042174618931, \"nll_base\": 2.0051276872248174, \"eval_ref\": \"bigbench_identify_odd_metaphor_generate_until__b... | {
"index": "bigbench_identify_odd_metaphor_generate_until__b53d8f63__s000__tb131072__v000",
"uid": "bigbench_identify_odd_metaphor_generate_until__b53d8f63__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_implicatures_generate_until__59734d12__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 12652.469116711061, \"nll_base\": 9.445607662200928, \"eval_ref\": \"bigbench_implicatures_generate_until__59734d12__s000__tb1... | {
"index": "bigbench_implicatures_generate_until__59734d12__s000__tb131072__v001",
"uid": "bigbench_implicatures_generate_until__59734d12__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_implicatures_generate_until__59734d12__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 12652.469116711061, \"nll_base\": 9.445607662200928, \"eval_ref\": \"bigbench_implicatures_generate_until__59734d12__s000__tb655... | {
"index": "bigbench_implicatures_generate_until__59734d12__s000__tb65536__v000",
"uid": "bigbench_implicatures_generate_until__59734d12__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_implicatures_multiple_choice__d5db3da0__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 12373.096559938043, \"nll_base\": 9.423279762268066, \"eval_ref\": \"bigbench_implicatures_multiple_choice__d5db3da0__s000__t... | {
"index": "bigbench_implicatures_multiple_choice__d5db3da0__s000__tb262144__v000",
"uid": "bigbench_implicatures_multiple_choice__d5db3da0__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_implicatures_multiple_choice__d5db3da0__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 12373.096559938043, \"nll_base\": 9.423279762268066, \"eval_ref\": \"bigbench_implicatures_multiple_choice__d5db3da0__s000__tb6... | {
"index": "bigbench_implicatures_multiple_choice__d5db3da0__s000__tb65536__v001",
"uid": "bigbench_implicatures_multiple_choice__d5db3da0__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_implicit_relations_multiple_choice__52d50fbc__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 187.86755907138576, \"nll_base\": 5.2357372415476835, \"eval_ref\": \"bigbench_implicit_relations_multiple_choice__52d5... | {
"index": "bigbench_implicit_relations_multiple_choice__52d50fbc__s000__tb131072__v001",
"uid": "bigbench_implicit_relations_multiple_choice__52d50fbc__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_intent_recognition_generate_until__f0a51cd5__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 34.904707123467325, \"nll_base\": 3.5526216947115383, \"eval_ref\": \"bigbench_intent_recognition_generate_until__f0a51c... | {
"index": "bigbench_intent_recognition_generate_until__f0a51cd5__s000__tb131072__v001",
"uid": "bigbench_intent_recognition_generate_until__f0a51cd5__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 565.3272331944629, \"nll_base\": 6.3374047372855395, \"eval_ref\": \"bigbench_international_phonetic_al... | {
"index": "bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb131072__v000",
"uid": "bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 565.3272331944629, \"nll_base\": 6.3374047372855395, \"eval_ref\": \"bigbench_international_phonetic_al... | {
"index": "bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb262144__v001",
"uid": "bigbench_international_phonetic_alphabet_nli_generate_until__1579067d__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 378.05819494616713, \"nll_base\": 5.935048138653791, \"eval_ref\": \"bigbench_international_phonetic_a... | {
"index": "bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb131072__v001",
"uid": "bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 378.05819494616713, \"nll_base\": 5.935048138653791, \"eval_ref\": \"bigbench_international_phonetic_alp... | {
"index": "bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb65536__v000",
"uid": "bigbench_international_phonetic_alphabet_nli_multiple_choice__995b52f0__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_irony_identification_generate_until__a23e35a6__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 411.23992407326654, \"nll_base\": 6.019176801045735, \"eval_ref\": \"bigbench_irony_identification_generate_until__a23... | {
"index": "bigbench_irony_identification_generate_until__a23e35a6__s000__tb131072__v000",
"uid": "bigbench_irony_identification_generate_until__a23e35a6__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_irony_identification_multiple_choice__b9008888__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 538.7094141064199, \"nll_base\": 6.289176305135091, \"eval_ref\": \"bigbench_irony_identification_multiple_choice__b900... | {
"index": "bigbench_irony_identification_multiple_choice__b9008888__s000__tb65536__v001",
"uid": "bigbench_irony_identification_multiple_choice__b9008888__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_kanji_ascii_generate_until__e8770fbd__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3025.1089556665816, \"nll_base\": 8.01470238821847, \"eval_ref\": \"bigbench_kanji_ascii_generate_until__e8770fbd__s000__tb1310... | {
"index": "bigbench_kanji_ascii_generate_until__e8770fbd__s000__tb131072__v001",
"uid": "bigbench_kanji_ascii_generate_until__e8770fbd__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 8.618593546378177, \"nll_base\": 2.1539219096902227, \"eval_ref\": \"bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb1... | {
"index": "bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb131072__v001",
"uid": "bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 8.618593546378177, \"nll_base\": 2.1539219096902227, \"eval_ref\": \"bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb655... | {
"index": "bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb65536__v000",
"uid": "bigbench_kanji_ascii_multiple_choice__ae597571__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 301.95587433714866, \"nll_base\": 5.710280895233154, \"eval_ref\": \"bigbench_key_value_maps_multiple_choice__5fc976f8__s00... | {
"index": "bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb131072__v001",
"uid": "bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 301.95587433714866, \"nll_base\": 5.710280895233154, \"eval_ref\": \"bigbench_key_value_maps_multiple_choice__5fc976f8__s000_... | {
"index": "bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb65536__v000",
"uid": "bigbench_key_value_maps_multiple_choice__5fc976f8__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_known_unknowns_generate_until__b998eac0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 165.19895751910843, \"nll_base\": 5.107150550649948, \"eval_ref\": \"bigbench_known_unknowns_generate_until__b998eac0__s000_... | {
"index": "bigbench_known_unknowns_generate_until__b998eac0__s000__tb131072__v001",
"uid": "bigbench_known_unknowns_generate_until__b998eac0__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_known_unknowns_generate_until__b998eac0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 165.19895751910843, \"nll_base\": 5.107150550649948, \"eval_ref\": \"bigbench_known_unknowns_generate_until__b998eac0__s000__t... | {
"index": "bigbench_known_unknowns_generate_until__b998eac0__s000__tb65536__v000",
"uid": "bigbench_known_unknowns_generate_until__b998eac0__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_known_unknowns_multiple_choice__23de36df__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 72.14767637101382, \"nll_base\": 4.278715079171317, \"eval_ref\": \"bigbench_known_unknowns_multiple_choice__23de36df__s000... | {
"index": "bigbench_known_unknowns_multiple_choice__23de36df__s000__tb262144__v000",
"uid": "bigbench_known_unknowns_multiple_choice__23de36df__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_known_unknowns_multiple_choice__23de36df__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 72.14767637101382, \"nll_base\": 4.278715079171317, \"eval_ref\": \"bigbench_known_unknowns_multiple_choice__23de36df__s000__... | {
"index": "bigbench_known_unknowns_multiple_choice__23de36df__s000__tb65536__v001",
"uid": "bigbench_known_unknowns_multiple_choice__23de36df__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_language_games_generate_until__fc881c09__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 295.5996334518714, \"nll_base\": 5.689005949059311, \"eval_ref\": \"bigbench_language_games_generate_until__fc881c09__s000__... | {
"index": "bigbench_language_games_generate_until__fc881c09__s000__tb131072__v001",
"uid": "bigbench_language_games_generate_until__fc881c09__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_language_games_generate_until__fc881c09__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 295.5996334518714, \"nll_base\": 5.689005949059311, \"eval_ref\": \"bigbench_language_games_generate_until__fc881c09__s000__tb... | {
"index": "bigbench_language_games_generate_until__fc881c09__s000__tb65536__v000",
"uid": "bigbench_language_games_generate_until__fc881c09__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 26.14335289815886, \"nll_base\": 3.263594966923858, \"eval_ref\": \"bigbench_linguistics_puzzles_generate_until__ae2a65... | {
"index": "bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb131072__v000",
"uid": "bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 26.14335289815886, \"nll_base\": 3.263594966923858, \"eval_ref\": \"bigbench_linguistics_puzzles_generate_until__ae2a65... | {
"index": "bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb262144__v001",
"uid": "bigbench_linguistics_puzzles_generate_until__ae2a65fc__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logic_grid_puzzle_multiple_choice__6def4829__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1328.4088810282285, \"nll_base\": 7.191737174987793, \"eval_ref\": \"bigbench_logic_grid_puzzle_multiple_choice__6def482... | {
"index": "bigbench_logic_grid_puzzle_multiple_choice__6def4829__s000__tb131072__v001",
"uid": "bigbench_logic_grid_puzzle_multiple_choice__6def4829__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_args_generate_until__1dd3be2b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2.8638154354666097, \"nll_base\": 1.052154804098195, \"eval_ref\": \"bigbench_logical_args_generate_until__1dd3be2b__s000__tb655... | {
"index": "bigbench_logical_args_generate_until__1dd3be2b__s000__tb65536__v001",
"uid": "bigbench_logical_args_generate_until__1dd3be2b__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_args_multiple_choice__fbd3e678__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2.9583640548648042, \"nll_base\": 1.0846364313824812, \"eval_ref\": \"bigbench_logical_args_multiple_choice__fbd3e678__s000__tb... | {
"index": "bigbench_logical_args_multiple_choice__fbd3e678__s000__tb65536__v001",
"uid": "bigbench_logical_args_multiple_choice__fbd3e678__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_deduction_multiple_choice__02d13266__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4.839624219394029, \"nll_base\": 1.576837077093283, \"eval_ref\": \"bigbench_logical_deduction_multiple_choice__02d13266... | {
"index": "bigbench_logical_deduction_multiple_choice__02d13266__s000__tb131072__v001",
"uid": "bigbench_logical_deduction_multiple_choice__02d13266__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 7855.149024197088, \"nll_base\": 8.968924522399902, \"eval_ref\": \"bigbench_logical_fallacy_detection_generate_u... | {
"index": "bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb131072__v002",
"uid": "bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb131072__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 7855.149024197088, \"nll_base\": 8.968924522399902, \"eval_ref\": \"bigbench_logical_fallacy_detection_generate_unt... | {
"index": "bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb65536__v001",
"uid": "bigbench_logical_fallacy_detection_generate_until__449cbc65__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_sequence_generate_until__df7235fa__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4.15527155033159, \"nll_base\": 1.42437778124565, \"eval_ref\": \"bigbench_logical_sequence_generate_until__df7235fa__s000... | {
"index": "bigbench_logical_sequence_generate_until__df7235fa__s000__tb131072__v001",
"uid": "bigbench_logical_sequence_generate_until__df7235fa__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_logical_sequence_generate_until__df7235fa__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4.15527155033159, \"nll_base\": 1.42437778124565, \"eval_ref\": \"bigbench_logical_sequence_generate_until__df7235fa__s000__... | {
"index": "bigbench_logical_sequence_generate_until__df7235fa__s000__tb65536__v000",
"uid": "bigbench_logical_sequence_generate_until__df7235fa__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 732.427426637355, \"nll_base\": 6.596364259719849, \"eval_ref\": \"bigbench_mathematical_induction_generate_until__3... | {
"index": "bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb131072__v001",
"uid": "bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 732.427426637355, \"nll_base\": 6.596364259719849, \"eval_ref\": \"bigbench_mathematical_induction_generate_until__3e6... | {
"index": "bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb65536__v000",
"uid": "bigbench_mathematical_induction_generate_until__3e6865b9__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 825.6500588860711, \"nll_base\": 6.716171026229858, \"eval_ref\": \"bigbench_mathematical_induction_multiple_choice... | {
"index": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb131072__v000",
"uid": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 825.6500588860711, \"nll_base\": 6.716171026229858, \"eval_ref\": \"bigbench_mathematical_induction_multiple_choice... | {
"index": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb262144__v001",
"uid": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 825.6500588860711, \"nll_base\": 6.716171026229858, \"eval_ref\": \"bigbench_mathematical_induction_multiple_choice__... | {
"index": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb65536__v002",
"uid": "bigbench_mathematical_induction_multiple_choice__1e058345__s000__tb65536__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_matrixshapes_generate_until__74d0e60c__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 21.534591406882157, \"nll_base\": 3.0696605448820153, \"eval_ref\": \"bigbench_matrixshapes_generate_until__74d0e60c__s000__tb... | {
"index": "bigbench_matrixshapes_generate_until__74d0e60c__s000__tb131072__v002",
"uid": "bigbench_matrixshapes_generate_until__74d0e60c__s000__tb131072__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_matrixshapes_generate_until__74d0e60c__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 21.534591406882157, \"nll_base\": 3.0696605448820153, \"eval_ref\": \"bigbench_matrixshapes_generate_until__74d0e60c__s000__tb65... | {
"index": "bigbench_matrixshapes_generate_until__74d0e60c__s000__tb65536__v001",
"uid": "bigbench_matrixshapes_generate_until__74d0e60c__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1081.7625277790019, \"nll_base\": 6.986346960067749, \"eval_ref\": \"bigbench_metaphor_boolean_generate_until__b55a9307__s... | {
"index": "bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb262144__v000",
"uid": "bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1081.7625277790019, \"nll_base\": 6.986346960067749, \"eval_ref\": \"bigbench_metaphor_boolean_generate_until__b55a9307__s00... | {
"index": "bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb65536__v001",
"uid": "bigbench_metaphor_boolean_generate_until__b55a9307__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1083.222780184991, \"nll_base\": 6.987695932388306, \"eval_ref\": \"bigbench_metaphor_boolean_multiple_choice__cc612668__... | {
"index": "bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb131072__v001",
"uid": "bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1083.222780184991, \"nll_base\": 6.987695932388306, \"eval_ref\": \"bigbench_metaphor_boolean_multiple_choice__cc612668__s0... | {
"index": "bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb65536__v000",
"uid": "bigbench_metaphor_boolean_multiple_choice__cc612668__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3.6426660477627704, \"nll_base\": 1.292715844256239, \"eval_ref\": \"bigbench_metaphor_understanding_generate_until_... | {
"index": "bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb131072__v000",
"uid": "bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3.6426660477627704, \"nll_base\": 1.292715844256239, \"eval_ref\": \"bigbench_metaphor_understanding_generate_until__0... | {
"index": "bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb65536__v002",
"uid": "bigbench_metaphor_understanding_generate_until__021d1c41__s000__tb65536__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_misconceptions_generate_until__a546f5db__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 680.8305992979004, \"nll_base\": 6.523313522338867, \"eval_ref\": \"bigbench_misconceptions_generate_until__a546f5db__s000__... | {
"index": "bigbench_misconceptions_generate_until__a546f5db__s000__tb131072__v000",
"uid": "bigbench_misconceptions_generate_until__a546f5db__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_misconceptions_generate_until__a546f5db__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 680.8305992979004, \"nll_base\": 6.523313522338867, \"eval_ref\": \"bigbench_misconceptions_generate_until__a546f5db__s000__... | {
"index": "bigbench_misconceptions_generate_until__a546f5db__s000__tb262144__v001",
"uid": "bigbench_misconceptions_generate_until__a546f5db__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_misconceptions_generate_until__a546f5db__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 680.8305992979004, \"nll_base\": 6.523313522338867, \"eval_ref\": \"bigbench_misconceptions_generate_until__a546f5db__s000__tb... | {
"index": "bigbench_misconceptions_generate_until__a546f5db__s000__tb65536__v002",
"uid": "bigbench_misconceptions_generate_until__a546f5db__s000__tb65536__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mnist_ascii_generate_until__da2817fe__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2757.1543476952465, \"nll_base\": 7.921954393386841, \"eval_ref\": \"bigbench_mnist_ascii_generate_until__da2817fe__s000__tb65536... | {
"index": "bigbench_mnist_ascii_generate_until__da2817fe__s000__tb65536__v001",
"uid": "bigbench_mnist_ascii_generate_until__da2817fe__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mnist_ascii_multiple_choice__f9b207ef__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3559.1247454839345, \"nll_base\": 8.17726993560791, \"eval_ref\": \"bigbench_mnist_ascii_multiple_choice__f9b207ef__s000__tb6553... | {
"index": "bigbench_mnist_ascii_multiple_choice__f9b207ef__s000__tb65536__v001",
"uid": "bigbench_mnist_ascii_multiple_choice__f9b207ef__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_moral_permissibility_generate_until__0ff7942a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2784.63977243394, \"nll_base\": 7.931873798370361, \"eval_ref\": \"bigbench_moral_permissibility_generate_until__0ff79... | {
"index": "bigbench_moral_permissibility_generate_until__0ff7942a__s000__tb131072__v000",
"uid": "bigbench_moral_permissibility_generate_until__0ff7942a__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1553.078435308379, \"nll_base\": 7.347994327545166, \"eval_ref\": \"bigbench_moral_permissibility_multiple_choice__a4... | {
"index": "bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb131072__v001",
"uid": "bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1553.078435308379, \"nll_base\": 7.347994327545166, \"eval_ref\": \"bigbench_moral_permissibility_multiple_choice__a4e7... | {
"index": "bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb65536__v000",
"uid": "bigbench_moral_permissibility_multiple_choice__a4e74a57__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb131072__v002\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 48983.5182678777, \"nll_base\": 10.799239158630371, \"eval_ref\": \"bigbench_movie_dialog_same_or_different_... | {
"index": "bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb131072__v002",
"uid": "bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb131072__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 48983.5182678777, \"nll_base\": 10.799239158630371, \"eval_ref\": \"bigbench_movie_dialog_same_or_different_ge... | {
"index": "bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb65536__v001",
"uid": "bigbench_movie_dialog_same_or_different_generate_until__7e222565__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 19.12458460072518, \"nll_base\": 2.950974659253192, \"eval_ref\": \"bigbench_movie_recommendation_generate_until__486c... | {
"index": "bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb131072__v001",
"uid": "bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 19.12458460072518, \"nll_base\": 2.950974659253192, \"eval_ref\": \"bigbench_movie_recommendation_generate_until__486cf8... | {
"index": "bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb65536__v000",
"uid": "bigbench_movie_recommendation_generate_until__486cf8d0__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 13.72217758966342, \"nll_base\": 2.6190133261513875, \"eval_ref\": \"bigbench_mult_data_wrangling_generate_until__74d1d... | {
"index": "bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb131072__v000",
"uid": "bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 13.72217758966342, \"nll_base\": 2.6190133261513875, \"eval_ref\": \"bigbench_mult_data_wrangling_generate_until__74d1d... | {
"index": "bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb262144__v001",
"uid": "bigbench_mult_data_wrangling_generate_until__74d1da4f__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_multiemo_generate_until__63ee29d0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 18002.032875691835, \"nll_base\": 9.798239968039773, \"eval_ref\": \"bigbench_multiemo_generate_until__63ee29d0__s000__tb131072__v... | {
"index": "bigbench_multiemo_generate_until__63ee29d0__s000__tb131072__v001",
"uid": "bigbench_multiemo_generate_until__63ee29d0__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_multiemo_generate_until__63ee29d0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 18002.032875691835, \"nll_base\": 9.798239968039773, \"eval_ref\": \"bigbench_multiemo_generate_until__63ee29d0__s000__tb65536__v000... | {
"index": "bigbench_multiemo_generate_until__63ee29d0__s000__tb65536__v000",
"uid": "bigbench_multiemo_generate_until__63ee29d0__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 17331.99323871674, \"nll_base\": 9.760309392755682, \"eval_ref\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb131072__... | {
"index": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb131072__v001",
"uid": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 17331.99323871674, \"nll_base\": 9.760309392755682, \"eval_ref\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb262144__... | {
"index": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb262144__v002",
"uid": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb262144__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 17331.99323871674, \"nll_base\": 9.760309392755682, \"eval_ref\": \"bigbench_multiemo_multiple_choice__30803dd8__s000__tb65536__v00... | {
"index": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb65536__v000",
"uid": "bigbench_multiemo_multiple_choice__30803dd8__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 23.113713035146947, \"nll_base\": 3.1404260792796306, \"eval_ref\": \"bigbench_natural_instructions_generate_until__6f... | {
"index": "bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb131072__v001",
"uid": "bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 23.113713035146947, \"nll_base\": 3.1404260792796306, \"eval_ref\": \"bigbench_natural_instructions_generate_until__6f23... | {
"index": "bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb65536__v000",
"uid": "bigbench_natural_instructions_generate_until__6f23bdcd__s000__tb65536__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2581.1090136443904, \"nll_base\": 7.855974435806274, \"eval_ref\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb131072__v... | {
"index": "bigbench_navigate_generate_until__2f1bd585__s000__tb131072__v000",
"uid": "bigbench_navigate_generate_until__2f1bd585__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2581.1090136443904, \"nll_base\": 7.855974435806274, \"eval_ref\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb262144__v... | {
"index": "bigbench_navigate_generate_until__2f1bd585__s000__tb262144__v001",
"uid": "bigbench_navigate_generate_until__2f1bd585__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2581.1090136443904, \"nll_base\": 7.855974435806274, \"eval_ref\": \"bigbench_navigate_generate_until__2f1bd585__s000__tb65536__v002... | {
"index": "bigbench_navigate_generate_until__2f1bd585__s000__tb65536__v002",
"uid": "bigbench_navigate_generate_until__2f1bd585__s000__tb65536__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_navigate_multiple_choice__4c849480__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 457.8407191996278, \"nll_base\": 6.126521348953247, \"eval_ref\": \"bigbench_navigate_multiple_choice__4c849480__s000__tb131072__... | {
"index": "bigbench_navigate_multiple_choice__4c849480__s000__tb131072__v000",
"uid": "bigbench_navigate_multiple_choice__4c849480__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_navigate_multiple_choice__4c849480__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 457.8407191996278, \"nll_base\": 6.126521348953247, \"eval_ref\": \"bigbench_navigate_multiple_choice__4c849480__s000__tb65536__v00... | {
"index": "bigbench_navigate_multiple_choice__4c849480__s000__tb65536__v002",
"uid": "bigbench_navigate_multiple_choice__4c849480__s000__tb65536__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 338.6121549397133, \"nll_base\": 5.824855366691214, \"eval_ref\": \"bigbench_nonsense_words_grammar_generate_until__... | {
"index": "bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb131072__v000",
"uid": "bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb131072__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 338.6121549397133, \"nll_base\": 5.824855366691214, \"eval_ref\": \"bigbench_nonsense_words_grammar_generate_until__... | {
"index": "bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb262144__v001",
"uid": "bigbench_nonsense_words_grammar_generate_until__8cc31fd4__s000__tb262144__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3508.3024719227387, \"nll_base\": 8.162887573242188, \"eval_ref\": \"bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb262... | {
"index": "bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb262144__v000",
"uid": "bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb262144__v000",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3508.3024719227387, \"nll_base\": 8.162887573242188, \"eval_ref\": \"bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb65536... | {
"index": "bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb65536__v001",
"uid": "bigbench_odd_one_out_generate_until__f40c2fd1__s000__tb65536__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 12967.5265860091, \"nll_base\": 9.470203556426585, \"eval_ref\": \"bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb131... | {
"index": "bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb131072__v001",
"uid": "bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb131072__v001",
"axis": "reasoning",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 12967.5265860091, \"nll_base\": 9.470203556426585, \"eval_ref\": \"bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb262... | {
"index": "bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb262144__v002",
"uid": "bigbench_odd_one_out_multiple_choice__b4135a10__s000__tb262144__v002",
"axis": "reasoning",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.