data_source stringclasses 1
value | prompt listlengths 2 2 | ability stringclasses 1
value | reward_model dict | extra_info dict |
|---|---|---|---|---|
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb262144__v012\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3702.1926333986967, \"nll_base\": 8.216680526733398, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb262144__v012",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb262144__v012",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb65536__v013\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3702.1926333986967, \"nll_base\": 8.216680526733398, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb65536__v013",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s004__tb65536__v013",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb131072__v017\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3405.8709059678104, \"nll_base\": 8.133255958557129, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb131072__v017",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb131072__v017",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb262144__v015\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3405.8709059678104, \"nll_base\": 8.133255958557129, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb262144__v015",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb262144__v015",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb65536__v016\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3405.8709059678104, \"nll_base\": 8.133255958557129, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb65536__v016",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s005__tb65536__v016",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb131072__v020\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3439.00567361536, \"nll_base\": 8.142937660217285, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb131072__v020",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb131072__v020",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb262144__v018\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3439.00567361536, \"nll_base\": 8.142937660217285, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb262144__v018",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb262144__v018",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb65536__v019\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3439.00567361536, \"nll_base\": 8.142937660217285, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s006_... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb65536__v019",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s006__tb65536__v019",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb131072__v023\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3605.0662422422424, \"nll_base\": 8.1900954246521, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb131072__v023",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb131072__v023",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb262144__v021\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3605.0662422422424, \"nll_base\": 8.1900954246521, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb262144__v021",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb262144__v021",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb65536__v022\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3605.0662422422424, \"nll_base\": 8.1900954246521, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s007_... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb65536__v022",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s007__tb65536__v022",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb131072__v026\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3527.5397129790176, \"nll_base\": 8.168355941772461, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb131072__v026",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb131072__v026",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb262144__v024\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3527.5397129790176, \"nll_base\": 8.168355941772461, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb262144__v024",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb262144__v024",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb65536__v025\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3527.5397129790176, \"nll_base\": 8.168355941772461, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s00... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb65536__v025",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s008__tb65536__v025",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb131072__v029\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3758.204537380276, \"nll_base\": 8.231696605682373, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb131072__v029",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb131072__v029",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb262144__v027\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3758.204537380276, \"nll_base\": 8.231696605682373, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb262144__v027",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb262144__v027",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb65536__v028\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3758.204537380276, \"nll_base\": 8.231696605682373, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s009... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb65536__v028",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s009__tb65536__v028",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb131072__v032\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3666.585003937374, \"nll_base\": 8.207015991210938, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb131072__v032",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb131072__v032",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb262144__v030\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3666.585003937374, \"nll_base\": 8.207015991210938, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb262144__v030",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb262144__v030",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb65536__v031\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3666.585003937374, \"nll_base\": 8.207015991210938, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s010... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb65536__v031",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s010__tb65536__v031",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb131072__v035\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3432.023748290662, \"nll_base\": 8.140905380249023, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb131072__v035",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb131072__v035",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb262144__v033\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3432.023748290662, \"nll_base\": 8.140905380249023, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb262144__v033",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb262144__v033",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb65536__v034\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3432.023748290662, \"nll_base\": 8.140905380249023, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s011... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb65536__v034",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s011__tb65536__v034",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb131072__v038\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4049.396154996185, \"nll_base\": 8.306323051452637, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb131072__v038",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb131072__v038",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb262144__v036\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4049.396154996185, \"nll_base\": 8.306323051452637, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s0... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb262144__v036",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb262144__v036",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb65536__v037\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4049.396154996185, \"nll_base\": 8.306323051452637, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s012... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb65536__v037",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s012__tb65536__v037",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb131072__v041\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3449.146643327246, \"nll_base\": 8.14588212966919, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s01... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb131072__v041",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb131072__v041",
"axis": "code",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb262144__v039\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3449.146643327246, \"nll_base\": 8.14588212966919, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s01... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb262144__v039",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb262144__v039",
"axis": "code",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb65536__v040\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3449.146643327246, \"nll_base\": 8.14588212966919, \"eval_ref\": \"global_piqa_parallel_generation_swe_latn__d9d0dde4__s013_... | {
"index": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb65536__v040",
"uid": "global_piqa_parallel_generation_swe_latn__d9d0dde4__s013__tb65536__v040",
"axis": "code",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"global_piqa_parallel_generation_uzn_latn__8462785a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2943.8077715856534, \"nll_base\": 7.987459182739258, \"eval_ref\": \"global_piqa_parallel_generation_uzn_latn__8462785a__s... | {
"index": "global_piqa_parallel_generation_uzn_latn__8462785a__s000__tb131072__v000",
"uid": "global_piqa_parallel_generation_uzn_latn__8462785a__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"groundcocoa__c219114d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 9.473006239862016, \"nll_base\": 2.248446305592855, \"eval_ref\": \"groundcocoa__c219114d__s000__tb262144__v000\", \"oracle_delta\": 0.3746191660563152... | {
"index": "groundcocoa__c219114d__s000__tb262144__v000",
"uid": "groundcocoa__c219114d__s000__tb262144__v000",
"axis": "instruction",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"gsm8k_platinum__56dd7e4b__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4.825732269226175, \"nll_base\": 1.5739624889184398, \"eval_ref\": \"gsm8k_platinum__56dd7e4b__s000__tb262144__v000\", \"oracle_delta\": 0.057549102... | {
"index": "gsm8k_platinum__56dd7e4b__s000__tb262144__v000",
"uid": "gsm8k_platinum__56dd7e4b__s000__tb262144__v000",
"axis": "math",
"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\": \"gsm8k_platinum__56dd7e4b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4.825732269226175, \"nll_base\": 1.5739624889184398, \"eval_ref\": \"gsm8k_platinum__56dd7e4b__s000__tb65536__v001\", \"oracle_delta\": 0.049708836950... | {
"index": "gsm8k_platinum__56dd7e4b__s000__tb65536__v001",
"uid": "gsm8k_platinum__56dd7e4b__s000__tb65536__v001",
"axis": "math",
"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\": \"gsm8k_platinum_cot__b89b2c6b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 271.234909729201, \"nll_base\": 5.602985270972391, \"eval_ref\": \"gsm8k_platinum_cot__b89b2c6b__s000__tb131072__v000\", \"oracle_delta\": 0.675... | {
"index": "gsm8k_platinum_cot__b89b2c6b__s000__tb131072__v000",
"uid": "gsm8k_platinum_cot__b89b2c6b__s000__tb131072__v000",
"axis": "math",
"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\": \"gsm8k_platinum_cot_llama__61a2c18c__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 252.17621668641135, \"nll_base\": 5.530128115699405, \"eval_ref\": \"gsm8k_platinum_cot_llama__61a2c18c__s000__tb131072__v001\", \"oracle_... | {
"index": "gsm8k_platinum_cot_llama__61a2c18c__s000__tb131072__v001",
"uid": "gsm8k_platinum_cot_llama__61a2c18c__s000__tb131072__v001",
"axis": "math",
"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\": \"gsm8k_platinum_cot_llama__61a2c18c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 252.17621668641135, \"nll_base\": 5.530128115699405, \"eval_ref\": \"gsm8k_platinum_cot_llama__61a2c18c__s000__tb65536__v000\", \"oracle_del... | {
"index": "gsm8k_platinum_cot_llama__61a2c18c__s000__tb65536__v000",
"uid": "gsm8k_platinum_cot_llama__61a2c18c__s000__tb65536__v000",
"axis": "math",
"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\": \"gsm8k_platinum_cot_self_consistency__a3122000__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 246.59616274985262, \"nll_base\": 5.507752029984085, \"eval_ref\": \"gsm8k_platinum_cot_self_consistency__a3122000__s000__tb131... | {
"index": "gsm8k_platinum_cot_self_consistency__a3122000__s000__tb131072__v001",
"uid": "gsm8k_platinum_cot_self_consistency__a3122000__s000__tb131072__v001",
"axis": "math",
"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\": \"gsm8k_platinum_cot_self_consistency__a3122000__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 246.59616274985262, \"nll_base\": 5.507752029984085, \"eval_ref\": \"gsm8k_platinum_cot_self_consistency__a3122000__s000__tb65536... | {
"index": "gsm8k_platinum_cot_self_consistency__a3122000__s000__tb65536__v000",
"uid": "gsm8k_platinum_cot_self_consistency__a3122000__s000__tb65536__v000",
"axis": "math",
"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\": \"gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4.854452185246225, \"nll_base\": 1.5798962602062159, \"eval_ref\": \"gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb131072__v001\", \"o... | {
"index": "gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb131072__v001",
"uid": "gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb131072__v001",
"axis": "math",
"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\": \"gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4.854452185246225, \"nll_base\": 1.5798962602062159, \"eval_ref\": \"gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb65536__v000\", \"orac... | {
"index": "gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb65536__v000",
"uid": "gsm8k_platinum_cot_zeroshot__51ad749f__s000__tb65536__v000",
"axis": "math",
"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\": \"gsm_plus__cad90933__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4.499616657719155, \"nll_base\": 1.5039922059741198, \"eval_ref\": \"gsm_plus__cad90933__s000__tb131072__v000\", \"oracle_delta\": 0.05461629766322895, \"... | {
"index": "gsm_plus__cad90933__s000__tb131072__v000",
"uid": "gsm_plus__cad90933__s000__tb131072__v000",
"axis": "math",
"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\": \"gsm_plus__cad90933__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4.499616657719155, \"nll_base\": 1.5039922059741198, \"eval_ref\": \"gsm_plus__cad90933__s000__tb262144__v001\", \"oracle_delta\": 0.05592468871108114, \"... | {
"index": "gsm_plus__cad90933__s000__tb262144__v001",
"uid": "gsm_plus__cad90933__s000__tb262144__v001",
"axis": "math",
"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\": \"gsm_plus_mini__29ef6b94__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 5.273620719999376, \"nll_base\": 1.6627171702899783, \"eval_ref\": \"gsm_plus_mini__29ef6b94__s000__tb131072__v000\", \"oracle_delta\": 0.04319489712... | {
"index": "gsm_plus_mini__29ef6b94__s000__tb131072__v000",
"uid": "gsm_plus_mini__29ef6b94__s000__tb131072__v000",
"axis": "math",
"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\": \"gsm_plus_mini__29ef6b94__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 5.273620719999376, \"nll_base\": 1.6627171702899783, \"eval_ref\": \"gsm_plus_mini__29ef6b94__s000__tb262144__v001\", \"oracle_delta\": 0.05045856209... | {
"index": "gsm_plus_mini__29ef6b94__s000__tb262144__v001",
"uid": "gsm_plus_mini__29ef6b94__s000__tb262144__v001",
"axis": "math",
"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\": \"headqa_en__59986e36__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 5.943016395035706, \"nll_base\": 1.7822168150944497, \"eval_ref\": \"headqa_en__59986e36__s000__tb131072__v001\", \"oracle_delta\": 0.11994738222947765, ... | {
"index": "headqa_en__59986e36__s000__tb131072__v001",
"uid": "headqa_en__59986e36__s000__tb131072__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"headqa_en__59986e36__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 5.943016395035706, \"nll_base\": 1.7822168150944497, \"eval_ref\": \"headqa_en__59986e36__s000__tb65536__v000\", \"oracle_delta\": 0.052007965543376944, \"... | {
"index": "headqa_en__59986e36__s000__tb65536__v000",
"uid": "headqa_en__59986e36__s000__tb65536__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"headqa_es__489447b4__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3.8441818702095367, \"nll_base\": 1.3465608027384526, \"eval_ref\": \"headqa_es__489447b4__s000__tb131072__v000\", \"oracle_delta\": 0.16583041706762147,... | {
"index": "headqa_es__489447b4__s000__tb131072__v000",
"uid": "headqa_es__489447b4__s000__tb131072__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"headqa_es__489447b4__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3.8441818702095367, \"nll_base\": 1.3465608027384526, \"eval_ref\": \"headqa_es__489447b4__s000__tb262144__v001\", \"oracle_delta\": 0.24264870406690386,... | {
"index": "headqa_es__489447b4__s000__tb262144__v001",
"uid": "headqa_es__489447b4__s000__tb262144__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"headqa_es__489447b4__s000__tb65536__v002\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3.8441818702095367, \"nll_base\": 1.3465608027384526, \"eval_ref\": \"headqa_es__489447b4__s000__tb65536__v002\", \"oracle_delta\": 0.024344044090063965, \... | {
"index": "headqa_es__489447b4__s000__tb65536__v002",
"uid": "headqa_es__489447b4__s000__tb65536__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"headqa_es__489447b4__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2.4045689018120466, \"nll_base\": 0.8773706366845769, \"eval_ref\": \"headqa_es__489447b4__s001__tb131072__v003\", \"oracle_delta\": 0.11839641504300558,... | {
"index": "headqa_es__489447b4__s001__tb131072__v003",
"uid": "headqa_es__489447b4__s001__tb131072__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"hellaswag__9af89211__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2.037593923008016, \"nll_base\": 0.7117696622522866, \"eval_ref\": \"hellaswag__9af89211__s000__tb131072__v001\", \"oracle_delta\": 0.07112566562624267, ... | {
"index": "hellaswag__9af89211__s000__tb131072__v001",
"uid": "hellaswag__9af89211__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\": \"hellaswag__9af89211__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2.037593923008016, \"nll_base\": 0.7117696622522866, \"eval_ref\": \"hellaswag__9af89211__s000__tb65536__v000\", \"oracle_delta\": 0.03037097724599014, \"r... | {
"index": "hellaswag__9af89211__s000__tb65536__v000",
"uid": "hellaswag__9af89211__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\": \"hendrycks_math500__9dc6bbe9__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 55.26032557088112, \"nll_base\": 4.01205521109719, \"eval_ref\": \"hendrycks_math500__9dc6bbe9__s000__tb131072__v000\", \"oracle_delta\": 0.33552... | {
"index": "hendrycks_math500__9dc6bbe9__s000__tb131072__v000",
"uid": "hendrycks_math500__9dc6bbe9__s000__tb131072__v000",
"axis": "math",
"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\": \"hendrycks_math500__9dc6bbe9__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 55.26032557088112, \"nll_base\": 4.01205521109719, \"eval_ref\": \"hendrycks_math500__9dc6bbe9__s000__tb262144__v001\", \"oracle_delta\": 0.38369... | {
"index": "hendrycks_math500__9dc6bbe9__s000__tb262144__v001",
"uid": "hendrycks_math500__9dc6bbe9__s000__tb262144__v001",
"axis": "math",
"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\": \"hendrycks_math_algebra__2ba16a7f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 140.73136878246464, \"nll_base\": 4.946852887263064, \"eval_ref\": \"hendrycks_math_algebra__2ba16a7f__s000__tb131072__v001\", \"oracle_delt... | {
"index": "hendrycks_math_algebra__2ba16a7f__s000__tb131072__v001",
"uid": "hendrycks_math_algebra__2ba16a7f__s000__tb131072__v001",
"axis": "math",
"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\": \"hendrycks_math_algebra__2ba16a7f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 140.73136878246464, \"nll_base\": 4.946852887263064, \"eval_ref\": \"hendrycks_math_algebra__2ba16a7f__s000__tb65536__v000\", \"oracle_delta\"... | {
"index": "hendrycks_math_algebra__2ba16a7f__s000__tb65536__v000",
"uid": "hendrycks_math_algebra__2ba16a7f__s000__tb65536__v000",
"axis": "math",
"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\": \"hendrycks_math_counting_and_prob__d96ed8b8__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 46.511038776843606, \"nll_base\": 3.839689677470439, \"eval_ref\": \"hendrycks_math_counting_and_prob__d96ed8b8__s000__tb262144__v... | {
"index": "hendrycks_math_counting_and_prob__d96ed8b8__s000__tb262144__v000",
"uid": "hendrycks_math_counting_and_prob__d96ed8b8__s000__tb262144__v000",
"axis": "math",
"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\": \"hendrycks_math_counting_and_prob__d96ed8b8__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 46.511038776843606, \"nll_base\": 3.839689677470439, \"eval_ref\": \"hendrycks_math_counting_and_prob__d96ed8b8__s000__tb65536__v001... | {
"index": "hendrycks_math_counting_and_prob__d96ed8b8__s000__tb65536__v001",
"uid": "hendrycks_math_counting_and_prob__d96ed8b8__s000__tb65536__v001",
"axis": "math",
"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\": \"hendrycks_math_geometry__84e902b4__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 56.54157114722433, \"nll_base\": 4.034976140121741, \"eval_ref\": \"hendrycks_math_geometry__84e902b4__s000__tb131072__v001\", \"oracle_del... | {
"index": "hendrycks_math_geometry__84e902b4__s000__tb131072__v001",
"uid": "hendrycks_math_geometry__84e902b4__s000__tb131072__v001",
"axis": "math",
"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\": \"hendrycks_math_geometry__84e902b4__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 56.54157114722433, \"nll_base\": 4.034976140121741, \"eval_ref\": \"hendrycks_math_geometry__84e902b4__s000__tb65536__v000\", \"oracle_delta\... | {
"index": "hendrycks_math_geometry__84e902b4__s000__tb65536__v000",
"uid": "hendrycks_math_geometry__84e902b4__s000__tb65536__v000",
"axis": "math",
"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\": \"hendrycks_math_num_theory__685c8d61__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 161.76747377751406, \"nll_base\": 5.086159957082648, \"eval_ref\": \"hendrycks_math_num_theory__685c8d61__s000__tb131072__v000\", \"oracl... | {
"index": "hendrycks_math_num_theory__685c8d61__s000__tb131072__v000",
"uid": "hendrycks_math_num_theory__685c8d61__s000__tb131072__v000",
"axis": "math",
"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\": \"hendrycks_math_prealgebra__29395b9f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 103.98957977790042, \"nll_base\": 4.644290699678309, \"eval_ref\": \"hendrycks_math_prealgebra__29395b9f__s000__tb131072__v001\", \"oracl... | {
"index": "hendrycks_math_prealgebra__29395b9f__s000__tb131072__v001",
"uid": "hendrycks_math_prealgebra__29395b9f__s000__tb131072__v001",
"axis": "math",
"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\": \"hendrycks_math_prealgebra__29395b9f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 103.98957977790042, \"nll_base\": 4.644290699678309, \"eval_ref\": \"hendrycks_math_prealgebra__29395b9f__s000__tb65536__v000\", \"oracle_d... | {
"index": "hendrycks_math_prealgebra__29395b9f__s000__tb65536__v000",
"uid": "hendrycks_math_prealgebra__29395b9f__s000__tb65536__v000",
"axis": "math",
"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\": \"hendrycks_math_precalc__e92ca789__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 17.702301952934835, \"nll_base\": 2.8736946849616425, \"eval_ref\": \"hendrycks_math_precalc__e92ca789__s000__tb131072__v001\", \"oracle_del... | {
"index": "hendrycks_math_precalc__e92ca789__s000__tb131072__v001",
"uid": "hendrycks_math_precalc__e92ca789__s000__tb131072__v001",
"axis": "math",
"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\": \"hendrycks_math_precalc__e92ca789__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 17.702301952934835, \"nll_base\": 2.8736946849616425, \"eval_ref\": \"hendrycks_math_precalc__e92ca789__s000__tb65536__v000\", \"oracle_delta\... | {
"index": "hendrycks_math_precalc__e92ca789__s000__tb65536__v000",
"uid": "hendrycks_math_precalc__e92ca789__s000__tb65536__v000",
"axis": "math",
"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\": \"hrm8k_gsm8k__44dcf9ff__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 518.7823332208548, \"nll_base\": 6.251484398702973, \"eval_ref\": \"hrm8k_gsm8k__44dcf9ff__s000__tb65536__v000\", \"oracle_delta\": 0.2678273025068263, \... | {
"index": "hrm8k_gsm8k__44dcf9ff__s000__tb65536__v000",
"uid": "hrm8k_gsm8k__44dcf9ff__s000__tb65536__v000",
"axis": "math",
"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\": \"hrm8k_gsm8k_en__c2f4f840__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 268.076982306793, \"nll_base\": 5.591274186676624, \"eval_ref\": \"hrm8k_gsm8k_en__c2f4f840__s000__tb131072__v000\", \"oracle_delta\": 0.43285713943... | {
"index": "hrm8k_gsm8k_en__c2f4f840__s000__tb131072__v000",
"uid": "hrm8k_gsm8k_en__c2f4f840__s000__tb131072__v000",
"axis": "math",
"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\": \"hrm8k_gsm8k_en__c2f4f840__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 268.076982306793, \"nll_base\": 5.591274186676624, \"eval_ref\": \"hrm8k_gsm8k_en__c2f4f840__s000__tb262144__v001\", \"oracle_delta\": 0.53286997477... | {
"index": "hrm8k_gsm8k_en__c2f4f840__s000__tb262144__v001",
"uid": "hrm8k_gsm8k_en__c2f4f840__s000__tb262144__v001",
"axis": "math",
"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\": \"hrm8k_ksm__5b634d43__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 335.277091407606, \"nll_base\": 5.814957328464674, \"eval_ref\": \"hrm8k_ksm__5b634d43__s000__tb65536__v000\", \"oracle_delta\": 0.21454069718070645, \"ran... | {
"index": "hrm8k_ksm__5b634d43__s000__tb65536__v000",
"uid": "hrm8k_ksm__5b634d43__s000__tb65536__v000",
"axis": "math",
"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\": \"hrm8k_math__dc1c6d32__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 510.5113605329386, \"nll_base\": 6.2354128910945015, \"eval_ref\": \"hrm8k_math__dc1c6d32__s000__tb65536__v001\", \"oracle_delta\": 0.2832932105431194, \"... | {
"index": "hrm8k_math__dc1c6d32__s000__tb65536__v001",
"uid": "hrm8k_math__dc1c6d32__s000__tb65536__v001",
"axis": "math",
"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\": \"hrm8k_math_en__dc2ebb6e__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 307.9424283534122, \"nll_base\": 5.7299128445712, \"eval_ref\": \"hrm8k_math_en__dc2ebb6e__s000__tb131072__v001\", \"oracle_delta\": 0.37214834039861... | {
"index": "hrm8k_math_en__dc2ebb6e__s000__tb131072__v001",
"uid": "hrm8k_math_en__dc2ebb6e__s000__tb131072__v001",
"axis": "math",
"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\": \"hrm8k_math_en__dc2ebb6e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 307.9424283534122, \"nll_base\": 5.7299128445712, \"eval_ref\": \"hrm8k_math_en__dc2ebb6e__s000__tb65536__v000\", \"oracle_delta\": 0.18074625188654103... | {
"index": "hrm8k_math_en__dc2ebb6e__s000__tb65536__v000",
"uid": "hrm8k_math_en__dc2ebb6e__s000__tb65536__v000",
"axis": "math",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_albanian_few_shot_en_health_oriented_education__bc0404bb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1122.2927997791423, \"nll_base\": 7.023129014407887, \"eval_ref\": \"include_base_44_albanian_few_shot... | {
"index": "include_base_44_albanian_few_shot_en_health_oriented_education__bc0404bb__s000__tb65536__v000",
"uid": "include_base_44_albanian_few_shot_en_health_oriented_education__bc0404bb__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_albanian_few_shot_og_social_science__a7a3f1a8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 621.4223037477204, \"nll_base\": 6.432010889053345, \"eval_ref\": \"include_base_44_albanian_few_shot_og_social_s... | {
"index": "include_base_44_albanian_few_shot_og_social_science__a7a3f1a8__s000__tb65536__v000",
"uid": "include_base_44_albanian_few_shot_og_social_science__a7a3f1a8__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_azerbaijani_few_shot_en_applied_science__613fd60d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1084.4432312407764, \"nll_base\": 6.988821983337402, \"eval_ref\": \"include_base_44_azerbaijani_few_shot_e... | {
"index": "include_base_44_azerbaijani_few_shot_en_applied_science__613fd60d__s000__tb262144__v000",
"uid": "include_base_44_azerbaijani_few_shot_en_applied_science__613fd60d__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_belarusian_few_shot_og_stem__1e0dceb6__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1006.1153200985779, \"nll_base\": 6.913851976394653, \"eval_ref\": \"include_base_44_belarusian_few_shot_og_stem__1e0dc... | {
"index": "include_base_44_belarusian_few_shot_og_stem__1e0dceb6__s000__tb131072__v000",
"uid": "include_base_44_belarusian_few_shot_og_stem__1e0dceb6__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_bengali_arts_humanities__a81bcfab__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1000.8122408772598, \"nll_base\": 6.908567190170288, \"eval_ref\": \"include_base_44_bengali_arts_humanities__a81bcfab__s000_... | {
"index": "include_base_44_bengali_arts_humanities__a81bcfab__s000__tb65536__v000",
"uid": "include_base_44_bengali_arts_humanities__a81bcfab__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_dutch_few_shot_og_social_science__82223b77__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 818.0037979797511, \"nll_base\": 6.706866979598999, \"eval_ref\": \"include_base_44_dutch_few_shot_og_social_scien... | {
"index": "include_base_44_dutch_few_shot_og_social_science__82223b77__s000__tb262144__v000",
"uid": "include_base_44_dutch_few_shot_og_social_science__82223b77__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_dutch_few_shot_og_stem__190a3d3c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 799.3073150522772, \"nll_base\": 6.683745496413287, \"eval_ref\": \"include_base_44_dutch_few_shot_og_stem__190a3d3c__s000__tb... | {
"index": "include_base_44_dutch_few_shot_og_stem__190a3d3c__s000__tb65536__v000",
"uid": "include_base_44_dutch_few_shot_og_stem__190a3d3c__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_dutch_stem__e4082437__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 867.5504010625682, \"nll_base\": 6.765673609340892, \"eval_ref\": \"include_base_44_dutch_stem__e4082437__s000__tb65536__v000\", \"oracle_... | {
"index": "include_base_44_dutch_stem__e4082437__s000__tb65536__v000",
"uid": "include_base_44_dutch_stem__e4082437__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_finnish_few_shot_en_health_oriented_education__8d29ba39__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1164.628346288231, \"nll_base\": 7.060157299041748, \"eval_ref\": \"include_base_44_finnish_few_shot_... | {
"index": "include_base_44_finnish_few_shot_en_health_oriented_education__8d29ba39__s000__tb131072__v000",
"uid": "include_base_44_finnish_few_shot_en_health_oriented_education__8d29ba39__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_finnish_few_shot_og_stem__1d1ac9eb__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 859.2388027309863, \"nll_base\": 6.756046884200153, \"eval_ref\": \"include_base_44_finnish_few_shot_og_stem__1d1ac9eb__s0... | {
"index": "include_base_44_finnish_few_shot_og_stem__1d1ac9eb__s000__tb131072__v000",
"uid": "include_base_44_finnish_few_shot_og_stem__1d1ac9eb__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_german_few_shot_og_stem__e423fd46__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 753.5330757961767, \"nll_base\": 6.6247729133157165, \"eval_ref\": \"include_base_44_german_few_shot_og_stem__e423fd46__s00... | {
"index": "include_base_44_german_few_shot_og_stem__e423fd46__s000__tb262144__v000",
"uid": "include_base_44_german_few_shot_og_stem__e423fd46__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_greek_medical_license__8da76180__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 786.3299737391648, \"nll_base\": 6.667376518249512, \"eval_ref\": \"include_base_44_greek_medical_license__8da76180__s000__tb... | {
"index": "include_base_44_greek_medical_license__8da76180__s000__tb131072__v000",
"uid": "include_base_44_greek_medical_license__8da76180__s000__tb131072__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_greek_medical_license__8da76180__s001__tb131072__v003\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 715.0040984425162, \"nll_base\": 6.572288274765015, \"eval_ref\": \"include_base_44_greek_medical_license__8da76180__s001__tb... | {
"index": "include_base_44_greek_medical_license__8da76180__s001__tb131072__v003",
"uid": "include_base_44_greek_medical_license__8da76180__s001__tb131072__v003",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hindi_few_shot_en_social_science__db748bfc__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 542.707650400149, \"nll_base\": 6.296570777893066, \"eval_ref\": \"include_base_44_hindi_few_shot_en_social_scienc... | {
"index": "include_base_44_hindi_few_shot_en_social_science__db748bfc__s000__tb131072__v000",
"uid": "include_base_44_hindi_few_shot_en_social_science__db748bfc__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hindi_few_shot_og_applied_science__2fc8edae__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 522.3389754031365, \"nll_base\": 6.2583167552948, \"eval_ref\": \"include_base_44_hindi_few_shot_og_applied_science... | {
"index": "include_base_44_hindi_few_shot_og_applied_science__2fc8edae__s000__tb65536__v000",
"uid": "include_base_44_hindi_few_shot_og_applied_science__2fc8edae__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hindi_few_shot_og_arts_humanities__de4e1c31__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 654.4939882520173, \"nll_base\": 6.483862400054932, \"eval_ref\": \"include_base_44_hindi_few_shot_og_arts_humaniti... | {
"index": "include_base_44_hindi_few_shot_og_arts_humanities__de4e1c31__s000__tb65536__v000",
"uid": "include_base_44_hindi_few_shot_og_arts_humanities__de4e1c31__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hungarian_few_shot_en_stem__9936a40a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 673.5317410610378, \"nll_base\": 6.5125351232640885, \"eval_ref\": \"include_base_44_hungarian_few_shot_en_stem__9936a40... | {
"index": "include_base_44_hungarian_few_shot_en_stem__9936a40a__s000__tb262144__v000",
"uid": "include_base_44_hungarian_few_shot_en_stem__9936a40a__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hungarian_few_shot_og_social_science__b02224ca__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 545.2595401993376, \"nll_base\": 6.301261901855469, \"eval_ref\": \"include_base_44_hungarian_few_shot_og_soci... | {
"index": "include_base_44_hungarian_few_shot_og_social_science__b02224ca__s000__tb262144__v000",
"uid": "include_base_44_hungarian_few_shot_og_social_science__b02224ca__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_hungarian_stem__7360be9d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 752.5774247340707, \"nll_base\": 6.623503881342271, \"eval_ref\": \"include_base_44_hungarian_stem__7360be9d__s000__tb262144__v000\"... | {
"index": "include_base_44_hungarian_stem__7360be9d__s000__tb262144__v000",
"uid": "include_base_44_hungarian_stem__7360be9d__s000__tb262144__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_indonesian_few_shot_en_stem__9c33c850__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1409.1546949640494, \"nll_base\": 7.2507452964782715, \"eval_ref\": \"include_base_44_indonesian_few_shot_en_stem__9c33c8... | {
"index": "include_base_44_indonesian_few_shot_en_stem__9c33c850__s000__tb65536__v000",
"uid": "include_base_44_indonesian_few_shot_en_stem__9c33c850__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_indonesian_few_shot_og_stem__e7d36ffc__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1230.6638275965515, \"nll_base\": 7.115309000015259, \"eval_ref\": \"include_base_44_indonesian_few_shot_og_stem__e7d36... | {
"index": "include_base_44_indonesian_few_shot_og_stem__e7d36ffc__s000__tb131072__v000",
"uid": "include_base_44_indonesian_few_shot_og_stem__e7d36ffc__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_italian_few_shot_en_stem__8972de1a__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1696.2615531058987, \"nll_base\": 7.436182022094727, \"eval_ref\": \"include_base_44_italian_few_shot_en_stem__8972de1a__s00... | {
"index": "include_base_44_italian_few_shot_en_stem__8972de1a__s000__tb65536__v000",
"uid": "include_base_44_italian_few_shot_en_stem__8972de1a__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_italian_few_shot_og_applied_science__2361cd00__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 503.7131220609985, \"nll_base\": 6.22200690375434, \"eval_ref\": \"include_base_44_italian_few_shot_og_applied_... | {
"index": "include_base_44_italian_few_shot_og_applied_science__2361cd00__s000__tb131072__v000",
"uid": "include_base_44_italian_few_shot_og_applied_science__2361cd00__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_italian_few_shot_og_professional_certification__b1ebda14__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 710.1286411785868, \"nll_base\": 6.565446138381958, \"eval_ref\": \"include_base_44_italian_few_shot... | {
"index": "include_base_44_italian_few_shot_og_professional_certification__b1ebda14__s000__tb131072__v000",
"uid": "include_base_44_italian_few_shot_og_professional_certification__b1ebda14__s000__tb131072__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_italian_stem__290bcc3e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1023.9663286510158, \"nll_base\": 6.93143892288208, \"eval_ref\": \"include_base_44_italian_stem__290bcc3e__s000__tb65536__v000\", \"ora... | {
"index": "include_base_44_italian_stem__290bcc3e__s000__tb65536__v000",
"uid": "include_base_44_italian_stem__290bcc3e__s000__tb65536__v000",
"axis": "multilingual",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 440.2996470254708, \"nll_base\": 6.08745551109314, \"eval_ref\": \"include_base_44_japanese_few_shot_en_medica... | {
"index": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb131072__v001",
"uid": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb131072__v001",
"axis": "domain",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 440.2996470254708, \"nll_base\": 6.08745551109314, \"eval_ref\": \"include_base_44_japanese_few_shot_en_medica... | {
"index": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb262144__v002",
"uid": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb262144__v002",
"axis": "domain",
"budget": 2048,
"token_budget": 262144,
"variant": "filter"
} |
datacuration | [
{
"role": "system",
"content": "You are a **training-data curator**. A small base language model needs to do better on a **target task**, and you must choose which examples from a large pool of candidate training examples it should be fine-tuned on.\n\nThe base model will be LoRA-fine-tuned on exactly the e... | datacuration_selection | {
"style": "datacuration",
"ground_truth": "{\"task_id\": \"include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 440.2996470254708, \"nll_base\": 6.08745551109314, \"eval_ref\": \"include_base_44_japanese_few_shot_en_medical_... | {
"index": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb65536__v000",
"uid": "include_base_44_japanese_few_shot_en_medical_license__c6b37752__s000__tb65536__v000",
"axis": "domain",
"budget": 2048,
"token_budget": 65536,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.