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\": \"afrimgsm_lin_prompt_4__5b669bea__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 119.36625822490349, \"nll_base\": 4.782196566589877, \"eval_ref\": \"afrimgsm_lin_prompt_4__5b669bea__s000__tb65536__v001\", \"oracle_delta\": ... | {
"index": "afrimgsm_lin_prompt_4__5b669bea__s000__tb65536__v001",
"uid": "afrimgsm_lin_prompt_4__5b669bea__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\": \"afrimgsm_lin_prompt_5__a90a1360__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 231.3412388636341, \"nll_base\": 5.443893845016892, \"eval_ref\": \"afrimgsm_lin_prompt_5__a90a1360__s000__tb262144__v000\", \"oracle_delta\"... | {
"index": "afrimgsm_lin_prompt_5__a90a1360__s000__tb262144__v000",
"uid": "afrimgsm_lin_prompt_5__a90a1360__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\": \"afrimgsm_lin_prompt_5__a90a1360__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 231.3412388636341, \"nll_base\": 5.443893845016892, \"eval_ref\": \"afrimgsm_lin_prompt_5__a90a1360__s000__tb65536__v001\", \"oracle_delta\": 0... | {
"index": "afrimgsm_lin_prompt_5__a90a1360__s000__tb65536__v001",
"uid": "afrimgsm_lin_prompt_5__a90a1360__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\": \"afrimgsm_lug_prompt_1__5ce46c37__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 375.45174410559315, \"nll_base\": 5.928129951908903, \"eval_ref\": \"afrimgsm_lug_prompt_1__5ce46c37__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_lug_prompt_1__5ce46c37__s000__tb131072__v001",
"uid": "afrimgsm_lug_prompt_1__5ce46c37__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\": \"afrimgsm_lug_prompt_1__5ce46c37__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 375.45174410559315, \"nll_base\": 5.928129951908903, \"eval_ref\": \"afrimgsm_lug_prompt_1__5ce46c37__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_lug_prompt_1__5ce46c37__s000__tb65536__v000",
"uid": "afrimgsm_lug_prompt_1__5ce46c37__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\": \"afrimgsm_lug_prompt_2__77212532__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 348.89519005008844, \"nll_base\": 5.854771562056108, \"eval_ref\": \"afrimgsm_lug_prompt_2__77212532__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_lug_prompt_2__77212532__s000__tb131072__v001",
"uid": "afrimgsm_lug_prompt_2__77212532__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\": \"afrimgsm_lug_prompt_2__77212532__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 348.89519005008844, \"nll_base\": 5.854771562056108, \"eval_ref\": \"afrimgsm_lug_prompt_2__77212532__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_lug_prompt_2__77212532__s000__tb65536__v000",
"uid": "afrimgsm_lug_prompt_2__77212532__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\": \"afrimgsm_lug_prompt_3__7966bea6__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 233.86139536441553, \"nll_base\": 5.454728612359965, \"eval_ref\": \"afrimgsm_lug_prompt_3__7966bea6__s000__tb262144__v000\", \"oracle_delta\... | {
"index": "afrimgsm_lug_prompt_3__7966bea6__s000__tb262144__v000",
"uid": "afrimgsm_lug_prompt_3__7966bea6__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\": \"afrimgsm_lug_prompt_3__7966bea6__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 233.86139536441553, \"nll_base\": 5.454728612359965, \"eval_ref\": \"afrimgsm_lug_prompt_3__7966bea6__s000__tb65536__v001\", \"oracle_delta\": ... | {
"index": "afrimgsm_lug_prompt_3__7966bea6__s000__tb65536__v001",
"uid": "afrimgsm_lug_prompt_3__7966bea6__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\": \"afrimgsm_lug_prompt_4__d477cceb__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 193.1420587301583, \"nll_base\": 5.263425973745493, \"eval_ref\": \"afrimgsm_lug_prompt_4__d477cceb__s000__tb131072__v001\", \"oracle_delta\"... | {
"index": "afrimgsm_lug_prompt_4__d477cceb__s000__tb131072__v001",
"uid": "afrimgsm_lug_prompt_4__d477cceb__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\": \"afrimgsm_lug_prompt_4__d477cceb__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 193.1420587301583, \"nll_base\": 5.263425973745493, \"eval_ref\": \"afrimgsm_lug_prompt_4__d477cceb__s000__tb65536__v000\", \"oracle_delta\": 0... | {
"index": "afrimgsm_lug_prompt_4__d477cceb__s000__tb65536__v000",
"uid": "afrimgsm_lug_prompt_4__d477cceb__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\": \"afrimgsm_lug_prompt_5__c64c345c__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 342.96843720725883, \"nll_base\": 5.837638423127948, \"eval_ref\": \"afrimgsm_lug_prompt_5__c64c345c__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_lug_prompt_5__c64c345c__s000__tb131072__v001",
"uid": "afrimgsm_lug_prompt_5__c64c345c__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\": \"afrimgsm_lug_prompt_5__c64c345c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 342.96843720725883, \"nll_base\": 5.837638423127948, \"eval_ref\": \"afrimgsm_lug_prompt_5__c64c345c__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_lug_prompt_5__c64c345c__s000__tb65536__v000",
"uid": "afrimgsm_lug_prompt_5__c64c345c__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\": \"afrimgsm_orm_prompt_1__069f4e02__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 494.1300162861968, \"nll_base\": 6.20279867342203, \"eval_ref\": \"afrimgsm_orm_prompt_1__069f4e02__s000__tb262144__v000\", \"oracle_delta\":... | {
"index": "afrimgsm_orm_prompt_1__069f4e02__s000__tb262144__v000",
"uid": "afrimgsm_orm_prompt_1__069f4e02__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\": \"afrimgsm_orm_prompt_1__069f4e02__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 494.1300162861968, \"nll_base\": 6.20279867342203, \"eval_ref\": \"afrimgsm_orm_prompt_1__069f4e02__s000__tb65536__v001\", \"oracle_delta\": 0.... | {
"index": "afrimgsm_orm_prompt_1__069f4e02__s000__tb65536__v001",
"uid": "afrimgsm_orm_prompt_1__069f4e02__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\": \"afrimgsm_orm_prompt_2__7ed75868__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 331.71150800312677, \"nll_base\": 5.804265639361213, \"eval_ref\": \"afrimgsm_orm_prompt_2__7ed75868__s000__tb131072__v000\", \"oracle_delta\... | {
"index": "afrimgsm_orm_prompt_2__7ed75868__s000__tb131072__v000",
"uid": "afrimgsm_orm_prompt_2__7ed75868__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\": \"afrimgsm_orm_prompt_2__7ed75868__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 331.71150800312677, \"nll_base\": 5.804265639361213, \"eval_ref\": \"afrimgsm_orm_prompt_2__7ed75868__s000__tb262144__v001\", \"oracle_delta\... | {
"index": "afrimgsm_orm_prompt_2__7ed75868__s000__tb262144__v001",
"uid": "afrimgsm_orm_prompt_2__7ed75868__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\": \"afrimgsm_orm_prompt_3__b90c1f2f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 233.88311609583815, \"nll_base\": 5.4548214867001485, \"eval_ref\": \"afrimgsm_orm_prompt_3__b90c1f2f__s000__tb131072__v001\", \"oracle_delta... | {
"index": "afrimgsm_orm_prompt_3__b90c1f2f__s000__tb131072__v001",
"uid": "afrimgsm_orm_prompt_3__b90c1f2f__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\": \"afrimgsm_orm_prompt_3__b90c1f2f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 233.88311609583815, \"nll_base\": 5.4548214867001485, \"eval_ref\": \"afrimgsm_orm_prompt_3__b90c1f2f__s000__tb65536__v000\", \"oracle_delta\":... | {
"index": "afrimgsm_orm_prompt_3__b90c1f2f__s000__tb65536__v000",
"uid": "afrimgsm_orm_prompt_3__b90c1f2f__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\": \"afrimgsm_orm_prompt_4__24537c0a__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 211.07651160675798, \"nll_base\": 5.352220681997446, \"eval_ref\": \"afrimgsm_orm_prompt_4__24537c0a__s000__tb262144__v000\", \"oracle_delta\... | {
"index": "afrimgsm_orm_prompt_4__24537c0a__s000__tb262144__v000",
"uid": "afrimgsm_orm_prompt_4__24537c0a__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\": \"afrimgsm_orm_prompt_4__24537c0a__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 211.07651160675798, \"nll_base\": 5.352220681997446, \"eval_ref\": \"afrimgsm_orm_prompt_4__24537c0a__s000__tb65536__v001\", \"oracle_delta\": ... | {
"index": "afrimgsm_orm_prompt_4__24537c0a__s000__tb65536__v001",
"uid": "afrimgsm_orm_prompt_4__24537c0a__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\": \"afrimgsm_orm_prompt_5__6ca05787__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 356.091880952706, \"nll_base\": 5.875188790115655, \"eval_ref\": \"afrimgsm_orm_prompt_5__6ca05787__s000__tb262144__v000\", \"oracle_delta\":... | {
"index": "afrimgsm_orm_prompt_5__6ca05787__s000__tb262144__v000",
"uid": "afrimgsm_orm_prompt_5__6ca05787__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\": \"afrimgsm_orm_prompt_5__6ca05787__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 356.091880952706, \"nll_base\": 5.875188790115655, \"eval_ref\": \"afrimgsm_orm_prompt_5__6ca05787__s000__tb65536__v001\", \"oracle_delta\": 0.... | {
"index": "afrimgsm_orm_prompt_5__6ca05787__s000__tb65536__v001",
"uid": "afrimgsm_orm_prompt_5__6ca05787__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\": \"afrimgsm_sna_prompt_1__7a5e7157__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 392.3008461345469, \"nll_base\": 5.972029010069023, \"eval_ref\": \"afrimgsm_sna_prompt_1__7a5e7157__s000__tb131072__v001\", \"oracle_delta\"... | {
"index": "afrimgsm_sna_prompt_1__7a5e7157__s000__tb131072__v001",
"uid": "afrimgsm_sna_prompt_1__7a5e7157__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\": \"afrimgsm_sna_prompt_1__7a5e7157__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 392.3008461345469, \"nll_base\": 5.972029010069023, \"eval_ref\": \"afrimgsm_sna_prompt_1__7a5e7157__s000__tb65536__v000\", \"oracle_delta\": 0... | {
"index": "afrimgsm_sna_prompt_1__7a5e7157__s000__tb65536__v000",
"uid": "afrimgsm_sna_prompt_1__7a5e7157__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\": \"afrimgsm_sna_prompt_2__a407e9c0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 264.586692232879, \"nll_base\": 5.578168956511611, \"eval_ref\": \"afrimgsm_sna_prompt_2__a407e9c0__s000__tb131072__v001\", \"oracle_delta\":... | {
"index": "afrimgsm_sna_prompt_2__a407e9c0__s000__tb131072__v001",
"uid": "afrimgsm_sna_prompt_2__a407e9c0__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\": \"afrimgsm_sna_prompt_2__a407e9c0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 264.586692232879, \"nll_base\": 5.578168956511611, \"eval_ref\": \"afrimgsm_sna_prompt_2__a407e9c0__s000__tb65536__v000\", \"oracle_delta\": 0.... | {
"index": "afrimgsm_sna_prompt_2__a407e9c0__s000__tb65536__v000",
"uid": "afrimgsm_sna_prompt_2__a407e9c0__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\": \"afrimgsm_sna_prompt_3__74e57f72__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 208.86552822736965, \"nll_base\": 5.341690639279923, \"eval_ref\": \"afrimgsm_sna_prompt_3__74e57f72__s000__tb131072__v000\", \"oracle_delta\... | {
"index": "afrimgsm_sna_prompt_3__74e57f72__s000__tb131072__v000",
"uid": "afrimgsm_sna_prompt_3__74e57f72__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\": \"afrimgsm_sna_prompt_3__74e57f72__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 208.86552822736965, \"nll_base\": 5.341690639279923, \"eval_ref\": \"afrimgsm_sna_prompt_3__74e57f72__s000__tb262144__v001\", \"oracle_delta\... | {
"index": "afrimgsm_sna_prompt_3__74e57f72__s000__tb262144__v001",
"uid": "afrimgsm_sna_prompt_3__74e57f72__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\": \"afrimgsm_sna_prompt_4__64422b55__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 276.1917413242032, \"nll_base\": 5.621095339457194, \"eval_ref\": \"afrimgsm_sna_prompt_4__64422b55__s000__tb262144__v000\", \"oracle_delta\"... | {
"index": "afrimgsm_sna_prompt_4__64422b55__s000__tb262144__v000",
"uid": "afrimgsm_sna_prompt_4__64422b55__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\": \"afrimgsm_sna_prompt_4__64422b55__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 276.1917413242032, \"nll_base\": 5.621095339457194, \"eval_ref\": \"afrimgsm_sna_prompt_4__64422b55__s000__tb65536__v001\", \"oracle_delta\": 0... | {
"index": "afrimgsm_sna_prompt_4__64422b55__s000__tb65536__v001",
"uid": "afrimgsm_sna_prompt_4__64422b55__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\": \"afrimgsm_sna_prompt_5__8a1f5313__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 317.4582528780853, \"nll_base\": 5.760346322689417, \"eval_ref\": \"afrimgsm_sna_prompt_5__8a1f5313__s000__tb262144__v000\", \"oracle_delta\"... | {
"index": "afrimgsm_sna_prompt_5__8a1f5313__s000__tb262144__v000",
"uid": "afrimgsm_sna_prompt_5__8a1f5313__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\": \"afrimgsm_sot_prompt_1__bc1452e1__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 310.64896725027904, \"nll_base\": 5.738663552024148, \"eval_ref\": \"afrimgsm_sot_prompt_1__bc1452e1__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_sot_prompt_1__bc1452e1__s000__tb131072__v001",
"uid": "afrimgsm_sot_prompt_1__bc1452e1__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\": \"afrimgsm_sot_prompt_1__bc1452e1__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 310.64896725027904, \"nll_base\": 5.738663552024148, \"eval_ref\": \"afrimgsm_sot_prompt_1__bc1452e1__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_sot_prompt_1__bc1452e1__s000__tb65536__v000",
"uid": "afrimgsm_sot_prompt_1__bc1452e1__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\": \"afrimgsm_sot_prompt_2__665e6766__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 260.6751323775808, \"nll_base\": 5.5632749285016745, \"eval_ref\": \"afrimgsm_sot_prompt_2__665e6766__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_sot_prompt_2__665e6766__s000__tb131072__v001",
"uid": "afrimgsm_sot_prompt_2__665e6766__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\": \"afrimgsm_sot_prompt_2__665e6766__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 260.6751323775808, \"nll_base\": 5.5632749285016745, \"eval_ref\": \"afrimgsm_sot_prompt_2__665e6766__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_sot_prompt_2__665e6766__s000__tb65536__v000",
"uid": "afrimgsm_sot_prompt_2__665e6766__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\": \"afrimgsm_sot_prompt_3__4ef35869__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 230.86196197851194, \"nll_base\": 5.44181996471477, \"eval_ref\": \"afrimgsm_sot_prompt_3__4ef35869__s000__tb131072__v000\", \"oracle_delta\"... | {
"index": "afrimgsm_sot_prompt_3__4ef35869__s000__tb131072__v000",
"uid": "afrimgsm_sot_prompt_3__4ef35869__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\": \"afrimgsm_sot_prompt_3__4ef35869__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 230.86196197851194, \"nll_base\": 5.44181996471477, \"eval_ref\": \"afrimgsm_sot_prompt_3__4ef35869__s000__tb262144__v001\", \"oracle_delta\"... | {
"index": "afrimgsm_sot_prompt_3__4ef35869__s000__tb262144__v001",
"uid": "afrimgsm_sot_prompt_3__4ef35869__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\": \"afrimgsm_sot_prompt_4__00caa9ec__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 225.91642620100004, \"nll_base\": 5.4201651353102465, \"eval_ref\": \"afrimgsm_sot_prompt_4__00caa9ec__s000__tb65536__v001\", \"oracle_delta\":... | {
"index": "afrimgsm_sot_prompt_4__00caa9ec__s000__tb65536__v001",
"uid": "afrimgsm_sot_prompt_4__00caa9ec__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\": \"afrimgsm_sot_prompt_5__95dcf269__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 252.65751871200965, \"nll_base\": 5.532034890693531, \"eval_ref\": \"afrimgsm_sot_prompt_5__95dcf269__s000__tb131072__v001\", \"oracle_delta\... | {
"index": "afrimgsm_sot_prompt_5__95dcf269__s000__tb131072__v001",
"uid": "afrimgsm_sot_prompt_5__95dcf269__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\": \"afrimgsm_sot_prompt_5__95dcf269__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 252.65751871200965, \"nll_base\": 5.532034890693531, \"eval_ref\": \"afrimgsm_sot_prompt_5__95dcf269__s000__tb65536__v000\", \"oracle_delta\": ... | {
"index": "afrimgsm_sot_prompt_5__95dcf269__s000__tb65536__v000",
"uid": "afrimgsm_sot_prompt_5__95dcf269__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\": \"afrimgsm_translate_amh_prompt_1__dcfeed04__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 247.54606346151056, \"nll_base\": 5.5115966796875, \"eval_ref\": \"afrimgsm_translate_amh_prompt_1__dcfeed04__s000__tb131072__v000\... | {
"index": "afrimgsm_translate_amh_prompt_1__dcfeed04__s000__tb131072__v000",
"uid": "afrimgsm_translate_amh_prompt_1__dcfeed04__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\": \"afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 191.19958222653352, \"nll_base\": 5.253317815607244, \"eval_ref\": \"afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb131072__v00... | {
"index": "afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb131072__v001",
"uid": "afrimgsm_translate_amh_prompt_2__e9ee9bd1__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\": \"afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 191.19958222653352, \"nll_base\": 5.253317815607244, \"eval_ref\": \"afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_amh_prompt_2__e9ee9bd1__s000__tb65536__v000",
"uid": "afrimgsm_translate_amh_prompt_2__e9ee9bd1__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\": \"afrimgsm_translate_amh_prompt_3__09c87868__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 219.17421043226207, \"nll_base\": 5.389866895014697, \"eval_ref\": \"afrimgsm_translate_amh_prompt_3__09c87868__s000__tb131072__v00... | {
"index": "afrimgsm_translate_amh_prompt_3__09c87868__s000__tb131072__v001",
"uid": "afrimgsm_translate_amh_prompt_3__09c87868__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\": \"afrimgsm_translate_amh_prompt_3__09c87868__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 219.17421043226207, \"nll_base\": 5.389866895014697, \"eval_ref\": \"afrimgsm_translate_amh_prompt_3__09c87868__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_amh_prompt_3__09c87868__s000__tb65536__v000",
"uid": "afrimgsm_translate_amh_prompt_3__09c87868__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\": \"afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 148.86552560738443, \"nll_base\": 5.0030433857335455, \"eval_ref\": \"afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb131072__v0... | {
"index": "afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb131072__v000",
"uid": "afrimgsm_translate_amh_prompt_4__48a6d4df__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\": \"afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 148.86552560738443, \"nll_base\": 5.0030433857335455, \"eval_ref\": \"afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb262144__v0... | {
"index": "afrimgsm_translate_amh_prompt_4__48a6d4df__s000__tb262144__v001",
"uid": "afrimgsm_translate_amh_prompt_4__48a6d4df__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\": \"afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 274.94800224003944, \"nll_base\": 5.616581997024679, \"eval_ref\": \"afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb262144__v00... | {
"index": "afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb262144__v000",
"uid": "afrimgsm_translate_amh_prompt_5__407f22b8__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\": \"afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 274.94800224003944, \"nll_base\": 5.616581997024679, \"eval_ref\": \"afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb65536__v001\"... | {
"index": "afrimgsm_translate_amh_prompt_5__407f22b8__s000__tb65536__v001",
"uid": "afrimgsm_translate_amh_prompt_5__407f22b8__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\": \"afrimgsm_translate_fra_prompt_2__9c546049__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 215.11820123426577, \"nll_base\": 5.371187650240385, \"eval_ref\": \"afrimgsm_translate_fra_prompt_2__9c546049__s000__tb131072__v00... | {
"index": "afrimgsm_translate_fra_prompt_2__9c546049__s000__tb131072__v000",
"uid": "afrimgsm_translate_fra_prompt_2__9c546049__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\": \"afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 170.5346091060284, \"nll_base\": 5.13893826208382, \"eval_ref\": \"afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb262144__v000\... | {
"index": "afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb262144__v000",
"uid": "afrimgsm_translate_fra_prompt_3__b0fcb4eb__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\": \"afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 170.5346091060284, \"nll_base\": 5.13893826208382, \"eval_ref\": \"afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb65536__v001\", ... | {
"index": "afrimgsm_translate_fra_prompt_3__b0fcb4eb__s000__tb65536__v001",
"uid": "afrimgsm_translate_fra_prompt_3__b0fcb4eb__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\": \"afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 140.32404078924253, \"nll_base\": 4.943954325167932, \"eval_ref\": \"afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb131072__v00... | {
"index": "afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb131072__v001",
"uid": "afrimgsm_translate_fra_prompt_4__7f7bdbca__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\": \"afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 140.32404078924253, \"nll_base\": 4.943954325167932, \"eval_ref\": \"afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_fra_prompt_4__7f7bdbca__s000__tb65536__v000",
"uid": "afrimgsm_translate_fra_prompt_4__7f7bdbca__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\": \"afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 265.2197441432548, \"nll_base\": 5.580558705552716, \"eval_ref\": \"afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb262144__v000... | {
"index": "afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb262144__v000",
"uid": "afrimgsm_translate_fra_prompt_5__cbe42846__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\": \"afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 265.2197441432548, \"nll_base\": 5.580558705552716, \"eval_ref\": \"afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb65536__v001\",... | {
"index": "afrimgsm_translate_fra_prompt_5__cbe42846__s000__tb65536__v001",
"uid": "afrimgsm_translate_fra_prompt_5__cbe42846__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\": \"afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 219.9308203105525, \"nll_base\": 5.393313043767756, \"eval_ref\": \"afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb131072__v001... | {
"index": "afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb131072__v001",
"uid": "afrimgsm_translate_hau_prompt_1__b0888d36__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\": \"afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 219.9308203105525, \"nll_base\": 5.393313043767756, \"eval_ref\": \"afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb65536__v000\",... | {
"index": "afrimgsm_translate_hau_prompt_1__b0888d36__s000__tb65536__v000",
"uid": "afrimgsm_translate_hau_prompt_1__b0888d36__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\": \"afrimgsm_translate_hau_prompt_2__ed589d4b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 271.1159979370778, \"nll_base\": 5.602546765966323, \"eval_ref\": \"afrimgsm_translate_hau_prompt_2__ed589d4b__s000__tb65536__v001\",... | {
"index": "afrimgsm_translate_hau_prompt_2__ed589d4b__s000__tb65536__v001",
"uid": "afrimgsm_translate_hau_prompt_2__ed589d4b__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\": \"afrimgsm_translate_hau_prompt_3__5bb5ba9a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 135.64205347453705, \"nll_base\": 4.910019456294545, \"eval_ref\": \"afrimgsm_translate_hau_prompt_3__5bb5ba9a__s000__tb131072__v00... | {
"index": "afrimgsm_translate_hau_prompt_3__5bb5ba9a__s000__tb131072__v000",
"uid": "afrimgsm_translate_hau_prompt_3__5bb5ba9a__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\": \"afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 139.85869239969674, \"nll_base\": 4.940632572880498, \"eval_ref\": \"afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb131072__v00... | {
"index": "afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb131072__v000",
"uid": "afrimgsm_translate_hau_prompt_4__8b55726e__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\": \"afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 139.85869239969674, \"nll_base\": 4.940632572880498, \"eval_ref\": \"afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb262144__v00... | {
"index": "afrimgsm_translate_hau_prompt_4__8b55726e__s000__tb262144__v001",
"uid": "afrimgsm_translate_hau_prompt_4__8b55726e__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\": \"afrimgsm_translate_hau_prompt_5__3c1757f2__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 206.08044706104653, \"nll_base\": 5.328266612270422, \"eval_ref\": \"afrimgsm_translate_hau_prompt_5__3c1757f2__s000__tb131072__v00... | {
"index": "afrimgsm_translate_hau_prompt_5__3c1757f2__s000__tb131072__v000",
"uid": "afrimgsm_translate_hau_prompt_5__3c1757f2__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\": \"afrimgsm_translate_kin_prompt_1__9d112457__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 246.5890882089138, \"nll_base\": 5.5077233408011645, \"eval_ref\": \"afrimgsm_translate_kin_prompt_1__9d112457__s000__tb131072__v00... | {
"index": "afrimgsm_translate_kin_prompt_1__9d112457__s000__tb131072__v001",
"uid": "afrimgsm_translate_kin_prompt_1__9d112457__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\": \"afrimgsm_translate_kin_prompt_1__9d112457__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 246.5890882089138, \"nll_base\": 5.5077233408011645, \"eval_ref\": \"afrimgsm_translate_kin_prompt_1__9d112457__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_kin_prompt_1__9d112457__s000__tb65536__v000",
"uid": "afrimgsm_translate_kin_prompt_1__9d112457__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\": \"afrimgsm_translate_kin_prompt_2__e3832710__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 211.55867010168726, \"nll_base\": 5.354502360026042, \"eval_ref\": \"afrimgsm_translate_kin_prompt_2__e3832710__s000__tb262144__v00... | {
"index": "afrimgsm_translate_kin_prompt_2__e3832710__s000__tb262144__v000",
"uid": "afrimgsm_translate_kin_prompt_2__e3832710__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\": \"afrimgsm_translate_kin_prompt_2__e3832710__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 211.55867010168726, \"nll_base\": 5.354502360026042, \"eval_ref\": \"afrimgsm_translate_kin_prompt_2__e3832710__s000__tb65536__v001\"... | {
"index": "afrimgsm_translate_kin_prompt_2__e3832710__s000__tb65536__v001",
"uid": "afrimgsm_translate_kin_prompt_2__e3832710__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\": \"afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 166.53472686321228, \"nll_base\": 5.115203857421875, \"eval_ref\": \"afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb131072__v00... | {
"index": "afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb131072__v001",
"uid": "afrimgsm_translate_kin_prompt_3__8382bc1e__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\": \"afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 166.53472686321228, \"nll_base\": 5.115203857421875, \"eval_ref\": \"afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_kin_prompt_3__8382bc1e__s000__tb65536__v000",
"uid": "afrimgsm_translate_kin_prompt_3__8382bc1e__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\": \"afrimgsm_translate_kin_prompt_5__dda64983__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 266.996100290376, \"nll_base\": 5.587234052639563, \"eval_ref\": \"afrimgsm_translate_kin_prompt_5__dda64983__s000__tb131072__v000\... | {
"index": "afrimgsm_translate_kin_prompt_5__dda64983__s000__tb131072__v000",
"uid": "afrimgsm_translate_kin_prompt_5__dda64983__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\": \"afrimgsm_translate_kin_prompt_5__dda64983__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 266.996100290376, \"nll_base\": 5.587234052639563, \"eval_ref\": \"afrimgsm_translate_kin_prompt_5__dda64983__s000__tb262144__v001\... | {
"index": "afrimgsm_translate_kin_prompt_5__dda64983__s000__tb262144__v001",
"uid": "afrimgsm_translate_kin_prompt_5__dda64983__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\": \"afrimgsm_translate_lin_prompt_1__933816d3__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 275.5503841291807, \"nll_base\": 5.618770494373567, \"eval_ref\": \"afrimgsm_translate_lin_prompt_1__933816d3__s000__tb65536__v001\",... | {
"index": "afrimgsm_translate_lin_prompt_1__933816d3__s000__tb65536__v001",
"uid": "afrimgsm_translate_lin_prompt_1__933816d3__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\": \"afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 161.4133478395509, \"nll_base\": 5.083968452785326, \"eval_ref\": \"afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb131072__v001... | {
"index": "afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb131072__v001",
"uid": "afrimgsm_translate_lin_prompt_2__5bf40427__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\": \"afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 161.4133478395509, \"nll_base\": 5.083968452785326, \"eval_ref\": \"afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb65536__v000\",... | {
"index": "afrimgsm_translate_lin_prompt_2__5bf40427__s000__tb65536__v000",
"uid": "afrimgsm_translate_lin_prompt_2__5bf40427__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\": \"afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 171.0062143240069, \"nll_base\": 5.141699896918403, \"eval_ref\": \"afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb131072__v001... | {
"index": "afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb131072__v001",
"uid": "afrimgsm_translate_lin_prompt_3__a3239d82__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\": \"afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 171.0062143240069, \"nll_base\": 5.141699896918403, \"eval_ref\": \"afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb65536__v000\",... | {
"index": "afrimgsm_translate_lin_prompt_3__a3239d82__s000__tb65536__v000",
"uid": "afrimgsm_translate_lin_prompt_3__a3239d82__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\": \"afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 133.3675798513996, \"nll_base\": 4.893109074345341, \"eval_ref\": \"afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb131072__v001... | {
"index": "afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb131072__v001",
"uid": "afrimgsm_translate_lin_prompt_4__5cc5aab4__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\": \"afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 133.3675798513996, \"nll_base\": 4.893109074345341, \"eval_ref\": \"afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb65536__v000\",... | {
"index": "afrimgsm_translate_lin_prompt_4__5cc5aab4__s000__tb65536__v000",
"uid": "afrimgsm_translate_lin_prompt_4__5cc5aab4__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\": \"afrimgsm_translate_lin_prompt_5__2b09663b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 221.50573775550973, \"nll_base\": 5.400448493237765, \"eval_ref\": \"afrimgsm_translate_lin_prompt_5__2b09663b__s000__tb65536__v001\"... | {
"index": "afrimgsm_translate_lin_prompt_5__2b09663b__s000__tb65536__v001",
"uid": "afrimgsm_translate_lin_prompt_5__2b09663b__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\": \"afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 211.1674478380482, \"nll_base\": 5.35265141039823, \"eval_ref\": \"afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb131072__v001\... | {
"index": "afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb131072__v001",
"uid": "afrimgsm_translate_lug_prompt_1__687af7bf__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\": \"afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 211.1674478380482, \"nll_base\": 5.35265141039823, \"eval_ref\": \"afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb65536__v000\", ... | {
"index": "afrimgsm_translate_lug_prompt_1__687af7bf__s000__tb65536__v000",
"uid": "afrimgsm_translate_lug_prompt_1__687af7bf__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\": \"afrimgsm_translate_lug_prompt_2__073a2ceb__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 192.2069749135808, \"nll_base\": 5.25857278575068, \"eval_ref\": \"afrimgsm_translate_lug_prompt_2__073a2ceb__s000__tb65536__v001\", ... | {
"index": "afrimgsm_translate_lug_prompt_2__073a2ceb__s000__tb65536__v001",
"uid": "afrimgsm_translate_lug_prompt_2__073a2ceb__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\": \"afrimgsm_translate_lug_prompt_3__a116ec77__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 194.76127648380336, \"nll_base\": 5.271774585430439, \"eval_ref\": \"afrimgsm_translate_lug_prompt_3__a116ec77__s000__tb131072__v00... | {
"index": "afrimgsm_translate_lug_prompt_3__a116ec77__s000__tb131072__v000",
"uid": "afrimgsm_translate_lug_prompt_3__a116ec77__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\": \"afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 113.80545278527319, \"nll_base\": 4.734490436056386, \"eval_ref\": \"afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb131072__v00... | {
"index": "afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb131072__v001",
"uid": "afrimgsm_translate_lug_prompt_4__c97e326c__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\": \"afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 113.80545278527319, \"nll_base\": 4.734490436056386, \"eval_ref\": \"afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_lug_prompt_4__c97e326c__s000__tb65536__v000",
"uid": "afrimgsm_translate_lug_prompt_4__c97e326c__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\": \"afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 259.25299539041225, \"nll_base\": 5.557804401104267, \"eval_ref\": \"afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb262144__v00... | {
"index": "afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb262144__v000",
"uid": "afrimgsm_translate_lug_prompt_5__1423aff8__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\": \"afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 259.25299539041225, \"nll_base\": 5.557804401104267, \"eval_ref\": \"afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb65536__v001\"... | {
"index": "afrimgsm_translate_lug_prompt_5__1423aff8__s000__tb65536__v001",
"uid": "afrimgsm_translate_lug_prompt_5__1423aff8__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\": \"afrimgsm_translate_orm_prompt_1__4e141dd8__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 268.06649379691834, \"nll_base\": 5.5912350609188985, \"eval_ref\": \"afrimgsm_translate_orm_prompt_1__4e141dd8__s000__tb131072__v0... | {
"index": "afrimgsm_translate_orm_prompt_1__4e141dd8__s000__tb131072__v001",
"uid": "afrimgsm_translate_orm_prompt_1__4e141dd8__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\": \"afrimgsm_translate_orm_prompt_2__940c7494__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 208.2420423329954, \"nll_base\": 5.338701068230395, \"eval_ref\": \"afrimgsm_translate_orm_prompt_2__940c7494__s000__tb262144__v000... | {
"index": "afrimgsm_translate_orm_prompt_2__940c7494__s000__tb262144__v000",
"uid": "afrimgsm_translate_orm_prompt_2__940c7494__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\": \"afrimgsm_translate_orm_prompt_2__940c7494__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 208.2420423329954, \"nll_base\": 5.338701068230395, \"eval_ref\": \"afrimgsm_translate_orm_prompt_2__940c7494__s000__tb65536__v001\",... | {
"index": "afrimgsm_translate_orm_prompt_2__940c7494__s000__tb65536__v001",
"uid": "afrimgsm_translate_orm_prompt_2__940c7494__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\": \"afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 173.49123545076236, \"nll_base\": 5.156127081976996, \"eval_ref\": \"afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb131072__v00... | {
"index": "afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb131072__v001",
"uid": "afrimgsm_translate_orm_prompt_3__1fd8a5ca__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\": \"afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 173.49123545076236, \"nll_base\": 5.156127081976996, \"eval_ref\": \"afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb65536__v000\"... | {
"index": "afrimgsm_translate_orm_prompt_3__1fd8a5ca__s000__tb65536__v000",
"uid": "afrimgsm_translate_orm_prompt_3__1fd8a5ca__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\": \"afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 119.2205425279212, \"nll_base\": 4.78097507642663, \"eval_ref\": \"afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb262144__v000\... | {
"index": "afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb262144__v000",
"uid": "afrimgsm_translate_orm_prompt_4__4a34d76d__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\": \"afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 119.2205425279212, \"nll_base\": 4.78097507642663, \"eval_ref\": \"afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb65536__v001\", ... | {
"index": "afrimgsm_translate_orm_prompt_4__4a34d76d__s000__tb65536__v001",
"uid": "afrimgsm_translate_orm_prompt_4__4a34d76d__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\": \"afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 361.0016996591929, \"nll_base\": 5.888882666519008, \"eval_ref\": \"afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb131072__v001... | {
"index": "afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb131072__v001",
"uid": "afrimgsm_translate_orm_prompt_5__b6d70d60__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\": \"afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 361.0016996591929, \"nll_base\": 5.888882666519008, \"eval_ref\": \"afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb65536__v000\",... | {
"index": "afrimgsm_translate_orm_prompt_5__b6d70d60__s000__tb65536__v000",
"uid": "afrimgsm_translate_orm_prompt_5__b6d70d60__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\": \"afrimgsm_translate_sot_prompt_1__46c6e539__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 156.9329286442698, \"nll_base\": 5.055818507982337, \"eval_ref\": \"afrimgsm_translate_sot_prompt_1__46c6e539__s000__tb65536__v001\",... | {
"index": "afrimgsm_translate_sot_prompt_1__46c6e539__s000__tb65536__v001",
"uid": "afrimgsm_translate_sot_prompt_1__46c6e539__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\": \"afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 200.60298197661717, \"nll_base\": 5.301327740704572, \"eval_ref\": \"afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb262144__v00... | {
"index": "afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb262144__v000",
"uid": "afrimgsm_translate_sot_prompt_2__d74acd13__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\": \"afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 200.60298197661717, \"nll_base\": 5.301327740704572, \"eval_ref\": \"afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb65536__v001\"... | {
"index": "afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb65536__v001",
"uid": "afrimgsm_translate_sot_prompt_2__d74acd13__s000__tb65536__v001",
"axis": "math",
"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.