Hummingbird-V2 / evaluation /arithmark_3.json
juinron's picture
Publish Hummingbird-V2 10B base model and model card
ebd2f40 verified
Raw
History Blame Contribute Delete
4.81 kB
{
"acc": 35.699999999999996,
"acc_norm": 36.0,
"categories": {
"elementary_school_math_continuation::addition::grades_1_2::easy": {
"acc": 28.125,
"acc_norm": 28.125,
"norm_correct": 36,
"raw_correct": 36,
"total": 128
},
"elementary_school_math_continuation::comparison::grades_2_3::medium": {
"acc": 27.27272727272727,
"acc_norm": 27.27272727272727,
"norm_correct": 12,
"raw_correct": 12,
"total": 44
},
"elementary_school_math_continuation::comparison_difference::grades_2_3::medium": {
"acc": 25.0,
"acc_norm": 25.0,
"norm_correct": 12,
"raw_correct": 12,
"total": 48
},
"elementary_school_math_continuation::data::grades_2_3::easy": {
"acc": 23.25581395348837,
"acc_norm": 25.581395348837212,
"norm_correct": 11,
"raw_correct": 10,
"total": 43
},
"elementary_school_math_continuation::division::grades_3_4::medium": {
"acc": 25.925925925925924,
"acc_norm": 24.074074074074073,
"norm_correct": 13,
"raw_correct": 14,
"total": 54
},
"elementary_school_math_continuation::fractions_counting::grades_3_4::medium": {
"acc": 20.0,
"acc_norm": 20.0,
"norm_correct": 10,
"raw_correct": 10,
"total": 50
},
"elementary_school_math_continuation::geometry_area::grades_4_5::medium": {
"acc": 51.92307692307693,
"acc_norm": 53.84615384615385,
"norm_correct": 28,
"raw_correct": 27,
"total": 52
},
"elementary_school_math_continuation::geometry_perimeter::grades_4_5::medium": {
"acc": 62.22222222222222,
"acc_norm": 62.22222222222222,
"norm_correct": 28,
"raw_correct": 28,
"total": 45
},
"elementary_school_math_continuation::measurement::grades_2_3::easy": {
"acc": 32.89473684210527,
"acc_norm": 32.89473684210527,
"norm_correct": 25,
"raw_correct": 25,
"total": 76
},
"elementary_school_math_continuation::money::grades_3_4::medium": {
"acc": 29.6875,
"acc_norm": 29.6875,
"norm_correct": 19,
"raw_correct": 19,
"total": 64
},
"elementary_school_math_continuation::multiplication::grades_3_4::medium": {
"acc": 51.35135135135135,
"acc_norm": 52.702702702702695,
"norm_correct": 39,
"raw_correct": 38,
"total": 74
},
"elementary_school_math_continuation::patterns::grades_3_4::medium": {
"acc": 24.528301886792452,
"acc_norm": 24.528301886792452,
"norm_correct": 13,
"raw_correct": 13,
"total": 53
},
"elementary_school_math_continuation::subtraction::grades_1_2::easy": {
"acc": 30.76923076923077,
"acc_norm": 31.62393162393162,
"norm_correct": 37,
"raw_correct": 36,
"total": 117
},
"elementary_school_math_continuation::time::grades_2_3::easy": {
"acc": 100.0,
"acc_norm": 100.0,
"norm_correct": 55,
"raw_correct": 55,
"total": 55
},
"elementary_school_math_continuation::two_step_add_subtract::grades_2_3::medium": {
"acc": 21.73913043478261,
"acc_norm": 21.73913043478261,
"norm_correct": 10,
"raw_correct": 10,
"total": 46
},
"elementary_school_math_continuation::two_step_addition::grades_2_3::medium": {
"acc": 15.789473684210526,
"acc_norm": 15.789473684210526,
"norm_correct": 3,
"raw_correct": 3,
"total": 19
},
"elementary_school_math_continuation::two_step_subtraction::grades_2_3::medium": {
"acc": 28.125,
"acc_norm": 28.125,
"norm_correct": 9,
"raw_correct": 9,
"total": 32
}
},
"frost_evaluation": {
"checkpoint": "D:\\llm\\hummingbird\\artifacts\\runs\\flightmix_10b_balanced_10m_muon_lrfix_20260921\\checkpoint-tokens-10000000000",
"data_sha256": "bf8ab1a5193d52cdf0e05ff0b3ca226bdfcf416cb6e75562dcbe72e7e4559435",
"dataset": "AxiomicLabs/Arithmark-3.0",
"dataset_revision": "6f6e59dd9b7e2c63455f7af7f838f9ecc3d0a746",
"device": "cuda",
"evaluator_sha256": "09667e1ea05ecd8b91dbfea14a47365b555c72622afed9fc284c127ecd5e1bcd",
"limit": null,
"loop_count": 1,
"numeric_serialization": "none",
"precision": "bfloat16",
"qk_norm_position": null,
"tokenizer": "D:\\llm\\hummingbird\\artifacts\\runs\\e3_tokenizers\\tok_4k_digit"
},
"norm_correct": 360,
"raw_correct": 357,
"timing": {
"evaluation_seconds": 0.8658304999989923,
"examples_per_second": 1154.9604685919055,
"tokenization_seconds": 0.30124549998436123
},
"total": 1000
}