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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb131072__v004\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4494.842106154521, \"nll_base\": 8.41068581973805, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb131072__v004\", \...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb131072__v004", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb131072__v004", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb262144__v005\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4494.842106154521, \"nll_base\": 8.41068581973805, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb262144__v005\", \...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb262144__v005", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb262144__v005", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4494.842106154521, \"nll_base\": 8.41068581973805, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb65536__v003\", \"or...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb65536__v003", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s001__tb65536__v003", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb131072__v007\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3583.851160475203, \"nll_base\": 8.184193244347206, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb131072__v007\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb131072__v007", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb131072__v007", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb262144__v008\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3583.851160475203, \"nll_base\": 8.184193244347206, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb262144__v008\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb262144__v008", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb262144__v008", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb65536__v006\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3583.851160475203, \"nll_base\": 8.184193244347206, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb65536__v006\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb65536__v006", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s002__tb65536__v006", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb131072__v010\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2672.903845730397, \"nll_base\": 7.890920743075284, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb131072__v010\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb131072__v010", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb131072__v010", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb262144__v011\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2672.903845730397, \"nll_base\": 7.890920743075284, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb262144__v011\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb262144__v011", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb262144__v011", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb65536__v009\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2672.903845730397, \"nll_base\": 7.890920743075284, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb65536__v009\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb65536__v009", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s003__tb65536__v009", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb131072__v013\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4467.750662473236, \"nll_base\": 8.404640353455836, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb131072__v013\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb131072__v013", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb131072__v013", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb262144__v014\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4467.750662473236, \"nll_base\": 8.404640353455836, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb262144__v014\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb262144__v014", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb262144__v014", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb65536__v012\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4467.750662473236, \"nll_base\": 8.404640353455836, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb65536__v012\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb65536__v012", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s004__tb65536__v012", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb131072__v016\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 10050.04471732911, \"nll_base\": 9.215332362962807, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb131072__v016\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb131072__v016", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb131072__v016", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb262144__v017\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 10050.04471732911, \"nll_base\": 9.215332362962807, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb262144__v017\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb262144__v017", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb262144__v017", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb65536__v015\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 10050.04471732911, \"nll_base\": 9.215332362962807, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb65536__v015\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb65536__v015", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s005__tb65536__v015", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb131072__v019\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3762.6854000117437, \"nll_base\": 8.23288818359375, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb131072__v019\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb131072__v019", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb131072__v019", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb262144__v020\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3762.6854000117437, \"nll_base\": 8.23288818359375, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb262144__v020\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb262144__v020", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb262144__v020", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb65536__v018\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3762.6854000117437, \"nll_base\": 8.23288818359375, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb65536__v018\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb65536__v018", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s006__tb65536__v018", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb131072__v022\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3756.9490310040815, \"nll_base\": 8.231362479073661, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb131072__v022\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb131072__v022", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb131072__v022", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb262144__v023\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3756.9490310040815, \"nll_base\": 8.231362479073661, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb262144__v023\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb262144__v023", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb262144__v023", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb65536__v021\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3756.9490310040815, \"nll_base\": 8.231362479073661, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb65536__v021\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb65536__v021", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s007__tb65536__v021", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb131072__v025\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1849.0820540275474, \"nll_base\": 7.522444607918723, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb131072__v025\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb131072__v025", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb131072__v025", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb262144__v026\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1849.0820540275474, \"nll_base\": 7.522444607918723, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb262144__v026\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb262144__v026", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb262144__v026", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb65536__v024\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1849.0820540275474, \"nll_base\": 7.522444607918723, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb65536__v024\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb65536__v024", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s008__tb65536__v024", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb131072__v028\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1907.9809812229507, \"nll_base\": 7.553800884046052, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb131072__v028\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb131072__v028", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb131072__v028", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb262144__v029\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1907.9809812229507, \"nll_base\": 7.553800884046052, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb262144__v029\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb262144__v029", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb262144__v029", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb65536__v027\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1907.9809812229507, \"nll_base\": 7.553800884046052, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb65536__v027\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb65536__v027", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s009__tb65536__v027", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb131072__v031\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2848.7758719028197, \"nll_base\": 7.954644662362558, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb131072__v031\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb131072__v031", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb131072__v031", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb262144__v032\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2848.7758719028197, \"nll_base\": 7.954644662362558, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb262144__v032\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb262144__v032", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb262144__v032", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb65536__v030\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2848.7758719028197, \"nll_base\": 7.954644662362558, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb65536__v030\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb65536__v030", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s010__tb65536__v030", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb131072__v034\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2506.6805657026594, \"nll_base\": 7.826714673088592, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb131072__v034\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb131072__v034", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb131072__v034", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb262144__v035\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2506.6805657026594, \"nll_base\": 7.826714673088592, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb262144__v035\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb262144__v035", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb262144__v035", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb65536__v033\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2506.6805657026594, \"nll_base\": 7.826714673088592, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb65536__v033\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb65536__v033", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s011__tb65536__v033", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb131072__v037\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1931.3641097569675, \"nll_base\": 7.5659818248588495, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb131072__v037\"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb131072__v037", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb131072__v037", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb262144__v038\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1931.3641097569675, \"nll_base\": 7.5659818248588495, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb262144__v038\"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb262144__v038", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb262144__v038", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb65536__v036\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1931.3641097569675, \"nll_base\": 7.5659818248588495, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb65536__v036\", \...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb65536__v036", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s012__tb65536__v036", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb131072__v040\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2954.2660717694366, \"nll_base\": 7.991005530724158, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb131072__v040\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb131072__v040", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb131072__v040", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb262144__v041\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2954.2660717694366, \"nll_base\": 7.991005530724158, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb262144__v041\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb262144__v041", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb262144__v041", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb65536__v039\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2954.2660717694366, \"nll_base\": 7.991005530724158, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb65536__v039\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb65536__v039", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s013__tb65536__v039", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb131072__v043\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4005.248037638981, \"nll_base\": 8.295360789579505, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb131072__v043\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb131072__v043", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb131072__v043", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb262144__v044\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4005.248037638981, \"nll_base\": 8.295360789579505, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb262144__v044\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb262144__v044", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb262144__v044", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb65536__v042\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4005.248037638981, \"nll_base\": 8.295360789579505, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb65536__v042\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb65536__v042", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s014__tb65536__v042", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb131072__v046\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1787.5701332469002, \"nll_base\": 7.4886125091813565, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb131072__v046\"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb131072__v046", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb131072__v046", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb262144__v047\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1787.5701332469002, \"nll_base\": 7.4886125091813565, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb262144__v047\"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb262144__v047", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb262144__v047", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb65536__v045\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1787.5701332469002, \"nll_base\": 7.4886125091813565, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb65536__v045\", \...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb65536__v045", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s015__tb65536__v045", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb131072__v049\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4301.080568492511, \"nll_base\": 8.366621565110613, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb131072__v049\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb131072__v049", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb131072__v049", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb262144__v050\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4301.080568492511, \"nll_base\": 8.366621565110613, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb262144__v050\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb262144__v050", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb262144__v050", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb65536__v048\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4301.080568492511, \"nll_base\": 8.366621565110613, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb65536__v048\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb65536__v048", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s016__tb65536__v048", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb131072__v052\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2780.9650729735786, \"nll_base\": 7.930553294994213, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb131072__v052\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb131072__v052", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb131072__v052", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb262144__v053\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2780.9650729735786, \"nll_base\": 7.930553294994213, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb262144__v053\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb262144__v053", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb262144__v053", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb65536__v051\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2780.9650729735786, \"nll_base\": 7.930553294994213, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb65536__v051\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb65536__v051", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s017__tb65536__v051", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb131072__v055\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4389.194297450948, \"nll_base\": 8.386900957892923, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb131072__v055\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb131072__v055", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb131072__v055", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb262144__v056\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4389.194297450948, \"nll_base\": 8.386900957892923, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb262144__v056\", ...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb262144__v056", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb262144__v056", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb65536__v054\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4389.194297450948, \"nll_base\": 8.386900957892923, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb65536__v054\", \"o...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb65536__v054", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s018__tb65536__v054", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb131072__v058\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1395.9315132722368, \"nll_base\": 7.241317222858298, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb131072__v058\",...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb131072__v058", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb131072__v058", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb65536__v057\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1395.9315132722368, \"nll_base\": 7.241317222858298, \"eval_ref\": \"lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb65536__v057\", \"...
{ "index": "lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb65536__v057", "uid": "lambada_openai_mt_stablelm_fr__ddbab25c__s019__tb65536__v057", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb131072__v001\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2816.2321175190095, \"nll_base\": 7.943155141977163, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb131072__v001\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb131072__v001", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb131072__v001", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb262144__v002\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2816.2321175190095, \"nll_base\": 7.943155141977163, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb262144__v002\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb262144__v002", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb262144__v002", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb65536__v000\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2816.2321175190095, \"nll_base\": 7.943155141977163, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s000__tb65536__v000\", \"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb65536__v000", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s000__tb65536__v000", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb131072__v004\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1638.228906483673, \"nll_base\": 7.401371002197266, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb131072__v004\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb131072__v004", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb131072__v004", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb262144__v005\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1638.228906483673, \"nll_base\": 7.401371002197266, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb262144__v005\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb262144__v005", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb262144__v005", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb65536__v003\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1638.228906483673, \"nll_base\": 7.401371002197266, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s001__tb65536__v003\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb65536__v003", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s001__tb65536__v003", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb131072__v007\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2866.363573676768, \"nll_base\": 7.960799457870912, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb131072__v007\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb131072__v007", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb131072__v007", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb262144__v008\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2866.363573676768, \"nll_base\": 7.960799457870912, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb262144__v008\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb262144__v008", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb262144__v008", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb65536__v006\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2866.363573676768, \"nll_base\": 7.960799457870912, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s002__tb65536__v006\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb65536__v006", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s002__tb65536__v006", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb131072__v010\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1930.1791520309632, \"nll_base\": 7.5653681024774775, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb131072__v010\"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb131072__v010", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb131072__v010", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb262144__v011\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1930.1791520309632, \"nll_base\": 7.5653681024774775, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb262144__v011\"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb262144__v011", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb262144__v011", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb65536__v009\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1930.1791520309632, \"nll_base\": 7.5653681024774775, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s003__tb65536__v009\", \...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb65536__v009", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s003__tb65536__v009", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb131072__v013\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2221.754800920891, \"nll_base\": 7.706052613489836, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb131072__v013\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb131072__v013", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb131072__v013", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb262144__v014\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2221.754800920891, \"nll_base\": 7.706052613489836, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb262144__v014\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb262144__v014", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb262144__v014", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb65536__v012\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2221.754800920891, \"nll_base\": 7.706052613489836, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s004__tb65536__v012\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb65536__v012", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s004__tb65536__v012", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb131072__v016\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 3148.4220228702934, \"nll_base\": 8.054656661184211, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb131072__v016\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb131072__v016", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb131072__v016", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb262144__v017\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 3148.4220228702934, \"nll_base\": 8.054656661184211, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb262144__v017\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb262144__v017", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb262144__v017", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb65536__v015\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 3148.4220228702934, \"nll_base\": 8.054656661184211, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s005__tb65536__v015\", \"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb65536__v015", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s005__tb65536__v015", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb131072__v019\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4581.298066583714, \"nll_base\": 8.429737657603651, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb131072__v019\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb131072__v019", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb131072__v019", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb262144__v020\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4581.298066583714, \"nll_base\": 8.429737657603651, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb262144__v020\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb262144__v020", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb262144__v020", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb65536__v018\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4581.298066583714, \"nll_base\": 8.429737657603651, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s006__tb65536__v018\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb65536__v018", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s006__tb65536__v018", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb131072__v022\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4759.252410226889, \"nll_base\": 8.467845878215751, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb131072__v022\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb131072__v022", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb131072__v022", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb262144__v023\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4759.252410226889, \"nll_base\": 8.467845878215751, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb262144__v023\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb262144__v023", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb262144__v023", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb65536__v021\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4759.252410226889, \"nll_base\": 8.467845878215751, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s007__tb65536__v021\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb65536__v021", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s007__tb65536__v021", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb131072__v025\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 1932.5052519204594, \"nll_base\": 7.566572498139881, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb131072__v025\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb131072__v025", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb131072__v025", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb262144__v026\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 1932.5052519204594, \"nll_base\": 7.566572498139881, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb262144__v026\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb262144__v026", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb262144__v026", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb65536__v024\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 1932.5052519204594, \"nll_base\": 7.566572498139881, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s008__tb65536__v024\", \"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb65536__v024", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s008__tb65536__v024", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb131072__v028\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2006.9405486216913, \"nll_base\": 7.604366726345486, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb131072__v028\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb131072__v028", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb131072__v028", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb262144__v029\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2006.9405486216913, \"nll_base\": 7.604366726345486, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb262144__v029\",...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb262144__v029", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb262144__v029", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb65536__v027\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2006.9405486216913, \"nll_base\": 7.604366726345486, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s009__tb65536__v027\", \"...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb65536__v027", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s009__tb65536__v027", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb131072__v031\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 5508.726042739464, \"nll_base\": 8.614088667199967, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb131072__v031\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb131072__v031", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb131072__v031", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb262144__v032\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 5508.726042739464, \"nll_base\": 8.614088667199967, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb262144__v032\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb262144__v032", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb262144__v032", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb65536__v030\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 5508.726042739464, \"nll_base\": 8.614088667199967, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s010__tb65536__v030\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb65536__v030", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s010__tb65536__v030", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb131072__v034\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2234.697191881553, \"nll_base\": 7.711861013252045, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb131072__v034\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb131072__v034", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb131072__v034", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb262144__v035\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2234.697191881553, \"nll_base\": 7.711861013252045, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb262144__v035\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb262144__v035", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb262144__v035", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb65536__v033\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2234.697191881553, \"nll_base\": 7.711861013252045, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s011__tb65536__v033\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb65536__v033", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s011__tb65536__v033", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb131072__v037\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 2138.932310879119, \"nll_base\": 7.668062063363882, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb131072__v037\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb131072__v037", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb131072__v037", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb262144__v038\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 2138.932310879119, \"nll_base\": 7.668062063363882, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb262144__v038\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb262144__v038", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb262144__v038", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb65536__v036\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 2138.932310879119, \"nll_base\": 7.668062063363882, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s012__tb65536__v036\", \"o...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb65536__v036", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s012__tb65536__v036", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb131072__v040\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 4465.943761997418, \"nll_base\": 8.40423583984375, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb131072__v040\", \...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb131072__v040", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb131072__v040", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb262144__v041\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 4465.943761997418, \"nll_base\": 8.40423583984375, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb262144__v041\", \...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb262144__v041", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb262144__v041", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb65536__v039\", \"budget\": 2048, \"token_budget\": 65536, \"ppl_base\": 4465.943761997418, \"nll_base\": 8.40423583984375, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s013__tb65536__v039\", \"or...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb65536__v039", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s013__tb65536__v039", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s014__tb131072__v043\", \"budget\": 2048, \"token_budget\": 131072, \"ppl_base\": 8759.700694663114, \"nll_base\": 9.077917016070822, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s014__tb131072__v043\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s014__tb131072__v043", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s014__tb131072__v043", "axis": "multilingual_ppl", "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\": \"lambada_openai_mt_stablelm_pt__e348cb41__s014__tb262144__v044\", \"budget\": 2048, \"token_budget\": 262144, \"ppl_base\": 8759.700694663114, \"nll_base\": 9.077917016070822, \"eval_ref\": \"lambada_openai_mt_stablelm_pt__e348cb41__s014__tb262144__v044\", ...
{ "index": "lambada_openai_mt_stablelm_pt__e348cb41__s014__tb262144__v044", "uid": "lambada_openai_mt_stablelm_pt__e348cb41__s014__tb262144__v044", "axis": "multilingual_ppl", "budget": 2048, "token_budget": 262144, "variant": "filter" }