Datasetpublic

tytodd/gpt-oss-120b-10k

gpt-oss-120b-10k Repo: tytodd/gpt-oss-120b-10k Config: /Users/tytodd/Desktop/Modaic/code/core/probe-lab/configs/datasets/10k/10k.yaml Model: openai/gpt-oss-120b Runtime: Modal local vLLM on localhost benchmark train val ood all customer_support_tickets_gorkem 1.70% 0.00% 1.55% mfrc 0.00% 0.00% 0.00% go_emotions 11.56% 6.90% 11.15% customer_support_tickets_en 30.27% 20.69% 29.41% aes2_essay_scoring 26.87% 20.69% 26.32% ultrafeedback 35.71% 48.28%… See the full description on the dataset page: https://huggingface.co/datasets/tytodd/gpt-oss-120b-10k.

sourceHugging Faceupdated 6mo agoView on Hugging Face

0likes278downloads

Dataset Card

gpt-oss-120b-10k

—Repo: tytodd/gpt-oss-120b-10k
—Config: /Users/tytodd/Desktop/Modaic/code/core/probe-lab/configs/datasets/10k/10k.yaml
—Model: openai/gpt-oss-120b
—Runtime: Modal local vLLM on localhost

benchmark	train	val	ood	all
customersupporttickets_gorkem	1.70%	0.00%		1.55%
mfrc	0.00%	0.00%		0.00%
go_emotions	11.56%	6.90%		11.15%
customersupporttickets_en	30.27%	20.69%		29.41%
aes2essayscoring	26.87%	20.69%		26.32%
ultrafeedback	35.71%	48.28%		36.84%
smollm_corpus	50.34%	51.72%		50.46%
orbench80k	43.54%	68.97%		45.82%
lexgluescotus	56.12%	61.60%		60.36%
medical_abstracts	65.31%	65.52%		65.33%
seekbench_evidence	63.27%	82.76%		65.02%
yelp	63.27%	58.62%		62.85%
tweetevalsentiment	62.93%	82.76%		64.71%
hh_rlhf	62.59%	68.97%		63.16%
tweetevalstance_hillary	66.33%	62.07%		65.94%
tweetevalstance_atheism	81.63%	58.62%		79.57%
rewardbench2	84.35%	93.10%		85.14%
semeval2010task8	54.08%	55.17%		54.18%
seekbench	81.29%	86.21%		81.73%
tweetevalirony	74.49%	72.41%		74.30%
tweetevaloffensive	74.15%	86.21%		75.23%
lexgluecase_hold	63.95%	79.31%		65.33%
tweetevalhate	72.79%	65.52%		72.14%
orbenchhard_1k	49.66%	89.66%		53.25%
tweetevalstance_climate	73.47%	65.52%		72.76%
snli	75.17%	65.52%		74.30%
tweetevalstance_feminist	79.25%	72.41%		78.64%
tweetevalstance_abortion	77.89%	79.31%		78.02%
toxigen_data	85.03%	82.76%		84.83%
tweetevalemotion	80.95%	82.76%		81.11%
civil_comments	90.82%	89.66%		90.71%
projudgebench	82.31%	100.00%		83.90%
colberthumordetection	88.10%	79.31%		87.31%
supportticketsalpha	94.90%	89.66%		94.43%
argumentqualityranking			10.00%	10.00%
rod101essayscoring			27.50%	27.50%
orbenchtoxic			45.00%	45.00%
judge_bench			63.75%	63.75%
musrteamallocation			72.50%	72.50%
musrobjectplacements			50.00%	50.00%
musrmurdermysteries			75.00%	75.00%
halueval_summarization			75.00%	75.00%
bbeh			60.00%	60.00%
codejudgebench			70.00%	70.00%
mmlu_pro			77.50%	77.50%
bbh_snarks			78.75%	78.75%
gpqa_diamond			68.75%	68.75%
mmlu			92.50%	92.50%
arc_challenge			96.25%	96.25%
all	61.88%	63.06%	64.17%	62.27%