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_cot_sna_prompt_2__9e4b147c__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 338.8160491098862, \"nll_base\": 5.825457332110164, \"eval_ref\": \"afrimgsm_cot_sna_prompt_2__9e4b147c__s000__tb65536__v000\", \"oracle_de... | {
"index": "afrimgsm_cot_sna_prompt_2__9e4b147c__s000__tb65536__v000",
"uid": "afrimgsm_cot_sna_prompt_2__9e4b147c__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_cot_sna_prompt_3__62fae6a0__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 211.877101765037, \"nll_base\": 5.356006397920496, \"eval_ref\": \"afrimgsm_cot_sna_prompt_3__62fae6a0__s000__tb131072__v001\", \"oracle_... | {
"index": "afrimgsm_cot_sna_prompt_3__62fae6a0__s000__tb131072__v001",
"uid": "afrimgsm_cot_sna_prompt_3__62fae6a0__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_cot_sna_prompt_3__62fae6a0__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 211.877101765037, \"nll_base\": 5.356006397920496, \"eval_ref\": \"afrimgsm_cot_sna_prompt_3__62fae6a0__s000__tb65536__v000\", \"oracle_del... | {
"index": "afrimgsm_cot_sna_prompt_3__62fae6a0__s000__tb65536__v000",
"uid": "afrimgsm_cot_sna_prompt_3__62fae6a0__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_cot_sna_prompt_4__0357372e__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 141.9092417772185, \"nll_base\": 4.955187710848722, \"eval_ref\": \"afrimgsm_cot_sna_prompt_4__0357372e__s000__tb131072__v001\", \"oracle... | {
"index": "afrimgsm_cot_sna_prompt_4__0357372e__s000__tb131072__v001",
"uid": "afrimgsm_cot_sna_prompt_4__0357372e__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_cot_sna_prompt_4__0357372e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 141.9092417772185, \"nll_base\": 4.955187710848722, \"eval_ref\": \"afrimgsm_cot_sna_prompt_4__0357372e__s000__tb65536__v000\", \"oracle_de... | {
"index": "afrimgsm_cot_sna_prompt_4__0357372e__s000__tb65536__v000",
"uid": "afrimgsm_cot_sna_prompt_4__0357372e__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_cot_sna_prompt_5__740f0414__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 219.59622213923092, \"nll_base\": 5.3917905061141305, \"eval_ref\": \"afrimgsm_cot_sna_prompt_5__740f0414__s000__tb262144__v000\", \"orac... | {
"index": "afrimgsm_cot_sna_prompt_5__740f0414__s000__tb262144__v000",
"uid": "afrimgsm_cot_sna_prompt_5__740f0414__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_cot_sot_prompt_1__0b4fc1bc__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 293.4948089110946, \"nll_base\": 5.681859952267085, \"eval_ref\": \"afrimgsm_cot_sot_prompt_1__0b4fc1bc__s000__tb262144__v000\", \"oracle... | {
"index": "afrimgsm_cot_sot_prompt_1__0b4fc1bc__s000__tb262144__v000",
"uid": "afrimgsm_cot_sot_prompt_1__0b4fc1bc__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_cot_sot_prompt_1__0b4fc1bc__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 293.4948089110946, \"nll_base\": 5.681859952267085, \"eval_ref\": \"afrimgsm_cot_sot_prompt_1__0b4fc1bc__s000__tb65536__v001\", \"oracle_de... | {
"index": "afrimgsm_cot_sot_prompt_1__0b4fc1bc__s000__tb65536__v001",
"uid": "afrimgsm_cot_sot_prompt_1__0b4fc1bc__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_cot_sot_prompt_2__1934d90e__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 312.3197193994792, \"nll_base\": 5.7440274047851565, \"eval_ref\": \"afrimgsm_cot_sot_prompt_2__1934d90e__s000__tb65536__v001\", \"oracle_d... | {
"index": "afrimgsm_cot_sot_prompt_2__1934d90e__s000__tb65536__v001",
"uid": "afrimgsm_cot_sot_prompt_2__1934d90e__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_cot_sot_prompt_3__82b4fd0a__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 156.10850864166574, \"nll_base\": 5.050551333669889, \"eval_ref\": \"afrimgsm_cot_sot_prompt_3__82b4fd0a__s000__tb131072__v001\", \"oracl... | {
"index": "afrimgsm_cot_sot_prompt_3__82b4fd0a__s000__tb131072__v001",
"uid": "afrimgsm_cot_sot_prompt_3__82b4fd0a__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_cot_sot_prompt_3__82b4fd0a__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 156.10850864166574, \"nll_base\": 5.050551333669889, \"eval_ref\": \"afrimgsm_cot_sot_prompt_3__82b4fd0a__s000__tb65536__v000\", \"oracle_d... | {
"index": "afrimgsm_cot_sot_prompt_3__82b4fd0a__s000__tb65536__v000",
"uid": "afrimgsm_cot_sot_prompt_3__82b4fd0a__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_cot_sot_prompt_4__06122f91__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 141.39666298798375, \"nll_base\": 4.9515691533818975, \"eval_ref\": \"afrimgsm_cot_sot_prompt_4__06122f91__s000__tb262144__v000\", \"orac... | {
"index": "afrimgsm_cot_sot_prompt_4__06122f91__s000__tb262144__v000",
"uid": "afrimgsm_cot_sot_prompt_4__06122f91__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_cot_sot_prompt_4__06122f91__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 141.39666298798375, \"nll_base\": 4.9515691533818975, \"eval_ref\": \"afrimgsm_cot_sot_prompt_4__06122f91__s000__tb65536__v001\", \"oracle_... | {
"index": "afrimgsm_cot_sot_prompt_4__06122f91__s000__tb65536__v001",
"uid": "afrimgsm_cot_sot_prompt_4__06122f91__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_cot_sot_prompt_5__6afcac59__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 240.89177870477207, \"nll_base\": 5.48434778161951, \"eval_ref\": \"afrimgsm_cot_sot_prompt_5__6afcac59__s000__tb131072__v001\", \"oracle... | {
"index": "afrimgsm_cot_sot_prompt_5__6afcac59__s000__tb131072__v001",
"uid": "afrimgsm_cot_sot_prompt_5__6afcac59__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_cot_sot_prompt_5__6afcac59__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 240.89177870477207, \"nll_base\": 5.48434778161951, \"eval_ref\": \"afrimgsm_cot_sot_prompt_5__6afcac59__s000__tb65536__v000\", \"oracle_de... | {
"index": "afrimgsm_cot_sot_prompt_5__6afcac59__s000__tb65536__v000",
"uid": "afrimgsm_cot_sot_prompt_5__6afcac59__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_cot_translate_amh_prompt_1__73f3be2e__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 253.0128596463545, \"nll_base\": 5.533440316077506, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_1__73f3be2e__s000__tb2621... | {
"index": "afrimgsm_cot_translate_amh_prompt_1__73f3be2e__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_amh_prompt_1__73f3be2e__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_cot_translate_amh_prompt_1__73f3be2e__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 253.0128596463545, \"nll_base\": 5.533440316077506, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_1__73f3be2e__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_amh_prompt_1__73f3be2e__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_amh_prompt_1__73f3be2e__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_cot_translate_amh_prompt_2__66e24a03__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 248.33514552986847, \"nll_base\": 5.514779227120536, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_2__66e24a03__s000__tb131... | {
"index": "afrimgsm_cot_translate_amh_prompt_2__66e24a03__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_amh_prompt_2__66e24a03__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_cot_translate_amh_prompt_2__66e24a03__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 248.33514552986847, \"nll_base\": 5.514779227120536, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_2__66e24a03__s000__tb65536... | {
"index": "afrimgsm_cot_translate_amh_prompt_2__66e24a03__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_amh_prompt_2__66e24a03__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_cot_translate_amh_prompt_3__716d7f46__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 152.580380915761, \"nll_base\": 5.027691545157597, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_3__716d7f46__s000__tb65536__... | {
"index": "afrimgsm_cot_translate_amh_prompt_3__716d7f46__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_amh_prompt_3__716d7f46__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_cot_translate_amh_prompt_4__997bb219__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 163.43915339855587, \"nll_base\": 5.096440770604589, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_4__997bb219__s000__tb131... | {
"index": "afrimgsm_cot_translate_amh_prompt_4__997bb219__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_amh_prompt_4__997bb219__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_cot_translate_amh_prompt_4__997bb219__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 163.43915339855587, \"nll_base\": 5.096440770604589, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_4__997bb219__s000__tb262... | {
"index": "afrimgsm_cot_translate_amh_prompt_4__997bb219__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_amh_prompt_4__997bb219__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_cot_translate_amh_prompt_5__34b6a69d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 365.3051619260538, \"nll_base\": 5.900733064350329, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_5__34b6a69d__s000__tb1310... | {
"index": "afrimgsm_cot_translate_amh_prompt_5__34b6a69d__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_amh_prompt_5__34b6a69d__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_cot_translate_amh_prompt_5__34b6a69d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 365.3051619260538, \"nll_base\": 5.900733064350329, \"eval_ref\": \"afrimgsm_cot_translate_amh_prompt_5__34b6a69d__s000__tb2621... | {
"index": "afrimgsm_cot_translate_amh_prompt_5__34b6a69d__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_amh_prompt_5__34b6a69d__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_cot_translate_fra_prompt_1__ab77e1f4__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 215.33450810900405, \"nll_base\": 5.37219267071418, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_1__ab77e1f4__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_fra_prompt_1__ab77e1f4__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_fra_prompt_1__ab77e1f4__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_cot_translate_fra_prompt_2__8c24c011__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 154.17937856600454, \"nll_base\": 5.03811672044837, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_2__8c24c011__s000__tb1310... | {
"index": "afrimgsm_cot_translate_fra_prompt_2__8c24c011__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_fra_prompt_2__8c24c011__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_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 179.76356841221065, \"nll_base\": 5.191642478660301, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb131... | {
"index": "afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__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_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 179.76356841221065, \"nll_base\": 5.191642478660301, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb65536... | {
"index": "afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_fra_prompt_3__d1f1d3f7__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_cot_translate_fra_prompt_4__4eb61a28__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 166.67699432862398, \"nll_base\": 5.1160577738060145, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_4__4eb61a28__s000__tb26... | {
"index": "afrimgsm_cot_translate_fra_prompt_4__4eb61a28__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_fra_prompt_4__4eb61a28__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_cot_translate_fra_prompt_5__a0610e6d__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 203.95521543764994, \"nll_base\": 5.31790043757512, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_5__a0610e6d__s000__tb2621... | {
"index": "afrimgsm_cot_translate_fra_prompt_5__a0610e6d__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_fra_prompt_5__a0610e6d__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_cot_translate_fra_prompt_5__a0610e6d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 203.95521543764994, \"nll_base\": 5.31790043757512, \"eval_ref\": \"afrimgsm_cot_translate_fra_prompt_5__a0610e6d__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_fra_prompt_5__a0610e6d__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_fra_prompt_5__a0610e6d__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_cot_translate_hau_prompt_1__953c621d__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 200.56687522756368, \"nll_base\": 5.301147733415876, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_1__953c621d__s000__tb131... | {
"index": "afrimgsm_cot_translate_hau_prompt_1__953c621d__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_hau_prompt_1__953c621d__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_cot_translate_hau_prompt_1__953c621d__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 200.56687522756368, \"nll_base\": 5.301147733415876, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_1__953c621d__s000__tb65536... | {
"index": "afrimgsm_cot_translate_hau_prompt_1__953c621d__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_hau_prompt_1__953c621d__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_cot_translate_hau_prompt_2__e24d57b5__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 185.52719914097477, \"nll_base\": 5.223201497395833, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_2__e24d57b5__s000__tb131... | {
"index": "afrimgsm_cot_translate_hau_prompt_2__e24d57b5__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_hau_prompt_2__e24d57b5__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_cot_translate_hau_prompt_2__e24d57b5__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 185.52719914097477, \"nll_base\": 5.223201497395833, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_2__e24d57b5__s000__tb65536... | {
"index": "afrimgsm_cot_translate_hau_prompt_2__e24d57b5__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_hau_prompt_2__e24d57b5__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_cot_translate_hau_prompt_3__a83e9dff__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 206.31245235640802, \"nll_base\": 5.32939177868413, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_3__a83e9dff__s000__tb1310... | {
"index": "afrimgsm_cot_translate_hau_prompt_3__a83e9dff__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_hau_prompt_3__a83e9dff__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_cot_translate_hau_prompt_3__a83e9dff__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 206.31245235640802, \"nll_base\": 5.32939177868413, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_3__a83e9dff__s000__tb2621... | {
"index": "afrimgsm_cot_translate_hau_prompt_3__a83e9dff__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_hau_prompt_3__a83e9dff__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_cot_translate_hau_prompt_4__691881ca__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 139.8194194511826, \"nll_base\": 4.940351728959517, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_4__691881ca__s000__tb1310... | {
"index": "afrimgsm_cot_translate_hau_prompt_4__691881ca__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_hau_prompt_4__691881ca__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_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 281.10009067438995, \"nll_base\": 5.638710800482302, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb262... | {
"index": "afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__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_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 281.10009067438995, \"nll_base\": 5.638710800482302, \"eval_ref\": \"afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb65536... | {
"index": "afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_hau_prompt_5__4ca5ccc2__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_cot_translate_kin_prompt_1__5363370d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 230.63221547917692, \"nll_base\": 5.440824300935953, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_1__5363370d__s000__tb131... | {
"index": "afrimgsm_cot_translate_kin_prompt_1__5363370d__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_kin_prompt_1__5363370d__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_cot_translate_kin_prompt_1__5363370d__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 230.63221547917692, \"nll_base\": 5.440824300935953, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_1__5363370d__s000__tb262... | {
"index": "afrimgsm_cot_translate_kin_prompt_1__5363370d__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_kin_prompt_1__5363370d__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_cot_translate_kin_prompt_2__7e192208__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 254.03038289383747, \"nll_base\": 5.537453877557184, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_2__7e192208__s000__tb65536... | {
"index": "afrimgsm_cot_translate_kin_prompt_2__7e192208__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_kin_prompt_2__7e192208__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_cot_translate_kin_prompt_3__ed234644__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 150.06752681121523, \"nll_base\": 5.011085371537642, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_3__ed234644__s000__tb131... | {
"index": "afrimgsm_cot_translate_kin_prompt_3__ed234644__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_kin_prompt_3__ed234644__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_cot_translate_kin_prompt_3__ed234644__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 150.06752681121523, \"nll_base\": 5.011085371537642, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_3__ed234644__s000__tb65536... | {
"index": "afrimgsm_cot_translate_kin_prompt_3__ed234644__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_kin_prompt_3__ed234644__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_cot_translate_kin_prompt_4__25c29890__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 153.10093535899958, \"nll_base\": 5.031097412109375, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_4__25c29890__s000__tb131... | {
"index": "afrimgsm_cot_translate_kin_prompt_4__25c29890__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_kin_prompt_4__25c29890__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_cot_translate_kin_prompt_4__25c29890__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 153.10093535899958, \"nll_base\": 5.031097412109375, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_4__25c29890__s000__tb262... | {
"index": "afrimgsm_cot_translate_kin_prompt_4__25c29890__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_kin_prompt_4__25c29890__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_cot_translate_kin_prompt_5__6442a079__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 239.29745565549104, \"nll_base\": 5.477707362411046, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_5__6442a079__s000__tb131... | {
"index": "afrimgsm_cot_translate_kin_prompt_5__6442a079__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_kin_prompt_5__6442a079__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_cot_translate_kin_prompt_5__6442a079__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 239.29745565549104, \"nll_base\": 5.477707362411046, \"eval_ref\": \"afrimgsm_cot_translate_kin_prompt_5__6442a079__s000__tb262... | {
"index": "afrimgsm_cot_translate_kin_prompt_5__6442a079__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_kin_prompt_5__6442a079__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_cot_translate_lin_prompt_1__20b5f033__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 189.17597917574827, \"nll_base\": 5.242677688598633, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_1__20b5f033__s000__tb131... | {
"index": "afrimgsm_cot_translate_lin_prompt_1__20b5f033__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_lin_prompt_1__20b5f033__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_cot_translate_lin_prompt_1__20b5f033__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 189.17597917574827, \"nll_base\": 5.242677688598633, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_1__20b5f033__s000__tb65536... | {
"index": "afrimgsm_cot_translate_lin_prompt_1__20b5f033__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_lin_prompt_1__20b5f033__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_cot_translate_lin_prompt_2__3e4059bd__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 181.8496455476413, \"nll_base\": 5.203180222284226, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_2__3e4059bd__s000__tb1310... | {
"index": "afrimgsm_cot_translate_lin_prompt_2__3e4059bd__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_lin_prompt_2__3e4059bd__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_cot_translate_lin_prompt_2__3e4059bd__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 181.8496455476413, \"nll_base\": 5.203180222284226, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_2__3e4059bd__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_lin_prompt_2__3e4059bd__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_lin_prompt_2__3e4059bd__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_cot_translate_lin_prompt_3__05385f84__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 200.669572619218, \"nll_base\": 5.301659638026975, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_3__05385f84__s000__tb13107... | {
"index": "afrimgsm_cot_translate_lin_prompt_3__05385f84__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_lin_prompt_3__05385f84__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_cot_translate_lin_prompt_4__56710fc8__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 198.1855665096523, \"nll_base\": 5.289203796386719, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_4__56710fc8__s000__tb1310... | {
"index": "afrimgsm_cot_translate_lin_prompt_4__56710fc8__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_lin_prompt_4__56710fc8__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_cot_translate_lin_prompt_4__56710fc8__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 198.1855665096523, \"nll_base\": 5.289203796386719, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_4__56710fc8__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_lin_prompt_4__56710fc8__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_lin_prompt_4__56710fc8__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_cot_translate_lin_prompt_5__74b85159__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 200.0857777805969, \"nll_base\": 5.298746163504465, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_5__74b85159__s000__tb2621... | {
"index": "afrimgsm_cot_translate_lin_prompt_5__74b85159__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_lin_prompt_5__74b85159__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_cot_translate_lin_prompt_5__74b85159__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 200.0857777805969, \"nll_base\": 5.298746163504465, \"eval_ref\": \"afrimgsm_cot_translate_lin_prompt_5__74b85159__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_lin_prompt_5__74b85159__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_lin_prompt_5__74b85159__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_cot_translate_lug_prompt_1__d2d8595d__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 262.20162218219775, \"nll_base\": 5.56911375812281, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_1__d2d8595d__s000__tb1310... | {
"index": "afrimgsm_cot_translate_lug_prompt_1__d2d8595d__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_lug_prompt_1__d2d8595d__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_cot_translate_lug_prompt_2__35e7fdff__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 174.01459542343682, \"nll_base\": 5.159139177440542, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_2__35e7fdff__s000__tb131... | {
"index": "afrimgsm_cot_translate_lug_prompt_2__35e7fdff__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_lug_prompt_2__35e7fdff__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_cot_translate_lug_prompt_3__3b10b042__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 176.92255518342424, \"nll_base\": 5.175712095488102, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_3__3b10b042__s000__tb65536... | {
"index": "afrimgsm_cot_translate_lug_prompt_3__3b10b042__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_lug_prompt_3__3b10b042__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_cot_translate_lug_prompt_4__43f30777__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 167.19075288595585, \"nll_base\": 5.119135393415179, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_4__43f30777__s000__tb131... | {
"index": "afrimgsm_cot_translate_lug_prompt_4__43f30777__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_lug_prompt_4__43f30777__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_cot_translate_lug_prompt_4__43f30777__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 167.19075288595585, \"nll_base\": 5.119135393415179, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_4__43f30777__s000__tb262... | {
"index": "afrimgsm_cot_translate_lug_prompt_4__43f30777__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_lug_prompt_4__43f30777__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_cot_translate_lug_prompt_5__ed061306__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 217.18306901161205, \"nll_base\": 5.380740633877841, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_5__ed061306__s000__tb262... | {
"index": "afrimgsm_cot_translate_lug_prompt_5__ed061306__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_lug_prompt_5__ed061306__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_cot_translate_lug_prompt_5__ed061306__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 217.18306901161205, \"nll_base\": 5.380740633877841, \"eval_ref\": \"afrimgsm_cot_translate_lug_prompt_5__ed061306__s000__tb65536... | {
"index": "afrimgsm_cot_translate_lug_prompt_5__ed061306__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_lug_prompt_5__ed061306__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_cot_translate_orm_prompt_1__0771190a__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 184.530727461933, \"nll_base\": 5.217815994122707, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_1__0771190a__s000__tb13107... | {
"index": "afrimgsm_cot_translate_orm_prompt_1__0771190a__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_orm_prompt_1__0771190a__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_cot_translate_orm_prompt_2__0f4ce59b__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 248.12026650926754, \"nll_base\": 5.51391357421875, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_2__0f4ce59b__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_orm_prompt_2__0f4ce59b__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_orm_prompt_2__0f4ce59b__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_cot_translate_orm_prompt_3__6d2eea25__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 152.13315191257286, \"nll_base\": 5.024756136807528, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_3__6d2eea25__s000__tb65536... | {
"index": "afrimgsm_cot_translate_orm_prompt_3__6d2eea25__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_orm_prompt_3__6d2eea25__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_cot_translate_orm_prompt_4__e020dc9b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 146.81953808383432, \"nll_base\": 4.989204200538429, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_4__e020dc9b__s000__tb131... | {
"index": "afrimgsm_cot_translate_orm_prompt_4__e020dc9b__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_orm_prompt_4__e020dc9b__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_cot_translate_orm_prompt_4__e020dc9b__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 146.81953808383432, \"nll_base\": 4.989204200538429, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_4__e020dc9b__s000__tb262... | {
"index": "afrimgsm_cot_translate_orm_prompt_4__e020dc9b__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_orm_prompt_4__e020dc9b__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_cot_translate_orm_prompt_5__a6421558__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 189.363694680608, \"nll_base\": 5.243669476425438, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_5__a6421558__s000__tb13107... | {
"index": "afrimgsm_cot_translate_orm_prompt_5__a6421558__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_orm_prompt_5__a6421558__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_cot_translate_orm_prompt_5__a6421558__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 189.363694680608, \"nll_base\": 5.243669476425438, \"eval_ref\": \"afrimgsm_cot_translate_orm_prompt_5__a6421558__s000__tb65536__... | {
"index": "afrimgsm_cot_translate_orm_prompt_5__a6421558__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_orm_prompt_5__a6421558__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_cot_translate_sot_prompt_1__a491b01e__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 187.93330170118915, \"nll_base\": 5.236087121696116, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_1__a491b01e__s000__tb131... | {
"index": "afrimgsm_cot_translate_sot_prompt_1__a491b01e__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_sot_prompt_1__a491b01e__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_cot_translate_sot_prompt_1__a491b01e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 187.93330170118915, \"nll_base\": 5.236087121696116, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_1__a491b01e__s000__tb65536... | {
"index": "afrimgsm_cot_translate_sot_prompt_1__a491b01e__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_sot_prompt_1__a491b01e__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_cot_translate_sot_prompt_2__e5ed3612__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 160.32586372970766, \"nll_base\": 5.077208392387997, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_2__e5ed3612__s000__tb65536... | {
"index": "afrimgsm_cot_translate_sot_prompt_2__e5ed3612__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_sot_prompt_2__e5ed3612__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_cot_translate_sot_prompt_3__a99c9efd__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 205.56619605048638, \"nll_base\": 5.325768103966346, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_3__a99c9efd__s000__tb131... | {
"index": "afrimgsm_cot_translate_sot_prompt_3__a99c9efd__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_sot_prompt_3__a99c9efd__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_cot_translate_sot_prompt_3__a99c9efd__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 205.56619605048638, \"nll_base\": 5.325768103966346, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_3__a99c9efd__s000__tb65536... | {
"index": "afrimgsm_cot_translate_sot_prompt_3__a99c9efd__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_sot_prompt_3__a99c9efd__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_cot_translate_sot_prompt_4__48f7327b__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 153.98650279698296, \"nll_base\": 5.036864954397219, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_4__48f7327b__s000__tb131... | {
"index": "afrimgsm_cot_translate_sot_prompt_4__48f7327b__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_sot_prompt_4__48f7327b__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_cot_translate_sot_prompt_4__48f7327b__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 153.98650279698296, \"nll_base\": 5.036864954397219, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_4__48f7327b__s000__tb262... | {
"index": "afrimgsm_cot_translate_sot_prompt_4__48f7327b__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_sot_prompt_4__48f7327b__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_cot_translate_sot_prompt_5__fec6f964__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 202.6093183884094, \"nll_base\": 5.31127958477668, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_5__fec6f964__s000__tb26214... | {
"index": "afrimgsm_cot_translate_sot_prompt_5__fec6f964__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_sot_prompt_5__fec6f964__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_cot_translate_sot_prompt_5__fec6f964__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 202.6093183884094, \"nll_base\": 5.31127958477668, \"eval_ref\": \"afrimgsm_cot_translate_sot_prompt_5__fec6f964__s000__tb65536__... | {
"index": "afrimgsm_cot_translate_sot_prompt_5__fec6f964__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_sot_prompt_5__fec6f964__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_cot_translate_swa_prompt_1__3830e25d__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 210.5770366026187, \"nll_base\": 5.349851555780533, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_1__3830e25d__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_swa_prompt_1__3830e25d__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_swa_prompt_1__3830e25d__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_cot_translate_swa_prompt_2__db3d1cc2__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 202.5081761562514, \"nll_base\": 5.310780261811756, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__s000__tb1310... | {
"index": "afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__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_cot_translate_swa_prompt_2__db3d1cc2__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 202.5081761562514, \"nll_base\": 5.310780261811756, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__s000__tb65536_... | {
"index": "afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_swa_prompt_2__db3d1cc2__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_cot_translate_swa_prompt_3__b9362078__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 188.38981473305836, \"nll_base\": 5.238513298754422, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_3__b9362078__s000__tb131... | {
"index": "afrimgsm_cot_translate_swa_prompt_3__b9362078__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_swa_prompt_3__b9362078__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_cot_translate_swa_prompt_4__c5fb5d21__s000__tb262144__v000\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 233.78657036813027, \"nll_base\": 5.4544086067044, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__s000__tb26214... | {
"index": "afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__s000__tb262144__v000",
"uid": "afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__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_cot_translate_swa_prompt_4__c5fb5d21__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 233.78657036813027, \"nll_base\": 5.4544086067044, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__s000__tb65536__... | {
"index": "afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_swa_prompt_4__c5fb5d21__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_cot_translate_swa_prompt_5__486d336e__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 151.0265718505082, \"nll_base\": 5.0174557938534985, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_5__486d336e__s000__tb131... | {
"index": "afrimgsm_cot_translate_swa_prompt_5__486d336e__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_swa_prompt_5__486d336e__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_cot_translate_swa_prompt_5__486d336e__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 151.0265718505082, \"nll_base\": 5.0174557938534985, \"eval_ref\": \"afrimgsm_cot_translate_swa_prompt_5__486d336e__s000__tb65536... | {
"index": "afrimgsm_cot_translate_swa_prompt_5__486d336e__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_swa_prompt_5__486d336e__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_cot_translate_twi_prompt_1__8f86f22f__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 264.26913944039165, \"nll_base\": 5.576968051419399, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_1__8f86f22f__s000__tb131... | {
"index": "afrimgsm_cot_translate_twi_prompt_1__8f86f22f__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_twi_prompt_1__8f86f22f__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_cot_translate_twi_prompt_1__8f86f22f__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 264.26913944039165, \"nll_base\": 5.576968051419399, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_1__8f86f22f__s000__tb65536... | {
"index": "afrimgsm_cot_translate_twi_prompt_1__8f86f22f__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_twi_prompt_1__8f86f22f__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_cot_translate_twi_prompt_2__208d5f06__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 181.87835110785778, \"nll_base\": 5.203338063091313, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_2__208d5f06__s000__tb131... | {
"index": "afrimgsm_cot_translate_twi_prompt_2__208d5f06__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_twi_prompt_2__208d5f06__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_cot_translate_twi_prompt_3__edaadf3e__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 203.7218750826147, \"nll_base\": 5.316755706188726, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_3__edaadf3e__s000__tb1310... | {
"index": "afrimgsm_cot_translate_twi_prompt_3__edaadf3e__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_twi_prompt_3__edaadf3e__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_cot_translate_twi_prompt_4__20af3cc1__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 150.32324993237825, \"nll_base\": 5.012787974964489, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_4__20af3cc1__s000__tb131... | {
"index": "afrimgsm_cot_translate_twi_prompt_4__20af3cc1__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_twi_prompt_4__20af3cc1__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_cot_translate_twi_prompt_5__62687798__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 254.97334238748448, \"nll_base\": 5.541159000036852, \"eval_ref\": \"afrimgsm_cot_translate_twi_prompt_5__62687798__s000__tb65536... | {
"index": "afrimgsm_cot_translate_twi_prompt_5__62687798__s000__tb65536__v000",
"uid": "afrimgsm_cot_translate_twi_prompt_5__62687798__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_cot_translate_wol_prompt_1__28c6cd64__s000__tb131072__v000\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 187.00269042364536, \"nll_base\": 5.231123004043312, \"eval_ref\": \"afrimgsm_cot_translate_wol_prompt_1__28c6cd64__s000__tb131... | {
"index": "afrimgsm_cot_translate_wol_prompt_1__28c6cd64__s000__tb131072__v000",
"uid": "afrimgsm_cot_translate_wol_prompt_1__28c6cd64__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_cot_translate_wol_prompt_1__28c6cd64__s000__tb262144__v001\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 187.00269042364536, \"nll_base\": 5.231123004043312, \"eval_ref\": \"afrimgsm_cot_translate_wol_prompt_1__28c6cd64__s000__tb262... | {
"index": "afrimgsm_cot_translate_wol_prompt_1__28c6cd64__s000__tb262144__v001",
"uid": "afrimgsm_cot_translate_wol_prompt_1__28c6cd64__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_cot_translate_wol_prompt_3__eb0fb150__s000__tb65536__v001\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 163.93385095985423, \"nll_base\": 5.099462998651825, \"eval_ref\": \"afrimgsm_cot_translate_wol_prompt_3__eb0fb150__s000__tb65536... | {
"index": "afrimgsm_cot_translate_wol_prompt_3__eb0fb150__s000__tb65536__v001",
"uid": "afrimgsm_cot_translate_wol_prompt_3__eb0fb150__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_cot_translate_wol_prompt_4__ae54e6e2__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 159.08405682711904, \"nll_base\": 5.069432721819196, \"eval_ref\": \"afrimgsm_cot_translate_wol_prompt_4__ae54e6e2__s000__tb131... | {
"index": "afrimgsm_cot_translate_wol_prompt_4__ae54e6e2__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_wol_prompt_4__ae54e6e2__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_cot_translate_wol_prompt_5__df4ace54__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 221.6577181612342, \"nll_base\": 5.401134381975447, \"eval_ref\": \"afrimgsm_cot_translate_wol_prompt_5__df4ace54__s000__tb1310... | {
"index": "afrimgsm_cot_translate_wol_prompt_5__df4ace54__s000__tb131072__v001",
"uid": "afrimgsm_cot_translate_wol_prompt_5__df4ace54__s000__tb131072__v001",
"axis": "math",
"budget": 2048,
"token_budget": 131072,
"variant": "filter"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.