LightDec / falcondec_report.json
RealFalconsAI's picture
Upload 14 files
3e2212d verified
Raw History Blame Contribute Delete
24.2 kB
Invalid JSON:Unexpected token 'N', ..." (card)": NaN, "... is not valid JSON
{
"version": "1.0.1",
"notebook_version": "V2",
"mode": "finetune",
"parent_version": "1.0.0",
"lineage": [
"jhu-clsp/ettin-encoder-150m",
"FalconDec-v1.0.0"
],
"config": {
"MODE": "finetune",
"FINETUNE_FROM": "falcondec_runs/FalconDec-v1.0.0",
"MODEL_NAME": "LightDec",
"BACKBONE": "jhu-clsp/ettin-encoder-150m",
"BACKBONE_INIT": "pretrained",
"PRESET": "full",
"CUSTOM_DATA_JSONL": "",
"CUSTOM_WEIGHT": 3.0,
"INCLUDE_BUILDERS": [],
"EXCLUDE_BUILDERS": [
"mind2web"
],
"MAX_LEN": 512,
"LONG_MAX_LEN": 2048,
"LONG_OPTS_THRESHOLD": 24,
"HEAD_MAX_LEN": 192,
"MAX_TOK_PER_OPT": 24,
"MAX_OPTS_SINGLE_PASS": 96,
"INTERACT_LAYERS": 2,
"INTERACT_HEADS": 8,
"BATCH_SIZE": 32,
"TOKENS_PER_BATCH": 16384,
"GRAD_ACCUM": 1,
"LR_ENCODER": 4e-05,
"LR_HEAD": 0.0003,
"LLRD": 0.9,
"WEIGHT_DECAY": 0.01,
"WARMUP_FRAC": 0.06,
"EPOCHS": 0,
"EMA_DECAY": 0.999,
"GRAD_CLIP": 1.0,
"FINETUNE_LR_SCALE": 0.5,
"SPHERICAL_W": 0.5,
"BRIER_W": 0.0,
"RPS_W": 1.0,
"TASK_SAMPLING_ALPHA": 0.5,
"NOTA_PROB": 0.08,
"RLCD_EPOCHS": 0.0,
"RLCD_SAMPLES": 4,
"RLCD_SIGMA": 0.3,
"RLCD_CE_ANCHOR": 0.5,
"DEVICE": "cuda",
"SEED": 42,
"NUM_WORKERS": 0,
"USE_COMPILE": false,
"ATTN_IMPL": "auto",
"OUTPUT_ROOT": "./falcondec_runs",
"EXPORT_INT8": true,
"SIZE_TARGET_MB": 250,
"DEFER_THRESHOLD": 0.7,
"EVAL_BASELINE_PROOF_V2": true,
"BASELINE_MAX_PER_TASK": 120,
"RUN_CPU_BENCH": true,
"PUSH_TO_HUB": false,
"HUB_REPO": ""
},
"epochs": 2,
"best_epoch": 1,
"rlcd_kept": false,
"history": [
{
"epoch": 1,
"train_loss": 0.47463753718974483,
"train_acc": 0.8307763076996187,
"val_macro": 0.8253118682436162,
"val_micro": 0.834065625876414,
"val_nll": 0.4055366800702441
},
{
"epoch": 2,
"train_loss": 0.36265113689414885,
"train_acc": 0.8738680252304237,
"val_macro": 0.8252539304231531,
"val_micro": 0.8348602411891184,
"val_nll": 0.4680047751477799
}
],
"train_minutes": 172.17712485790253,
"temperatures": [
[
1.3723914623260498,
1.3805755376815796,
1.5033340454101562,
1.320592999458313
],
[
1.3903899192810059,
1.3903899192810059,
1.3903899192810059,
1.3903899192810059
],
[
1.4738539457321167,
1.4738539457321167,
1.4738539457321167,
1.4738539457321167
]
],
"data_counts": {
"train": 946028,
"validation": 21394,
"test": 26597
},
"skipped_builders": [],
"test_overall": {
"n": 26597.0,
"micro_acc": 0.7810655337068091,
"macro_acc": 0.7694407230754166,
"nll": 0.5288964978022598,
"brier": 0.2897912847111225,
"ece": 0.01969622960708085,
"aurc": 0.06403486126686972,
"score_mae": 0.4928175845338412,
"coverage@0.7": 0.6595104711057638,
"acc_on_covered@0.7": 0.9167094236360527
},
"test_per_domain": {
"agentic": 0.7781070271610528,
"classification": 0.615,
"code": 0.8717717677968562,
"guardrails": 0.7951432854293641,
"intents": 0.8278333333333333,
"policy": 0.7873333333333333,
"reasoning": 0.7006267469170804,
"support": 1.0,
"workflows": 0.755
},
"test_per_task": [
{
"task": "ag_news/topic",
"n": 500,
"acc": 0.888,
"chance": 0.25,
"nll": 0.28043824258210953,
"brier": 0.154232526432949,
"heldout": false,
"domain": "classification",
"ece": 0.03968544936180117,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "agentharm/refuse",
"n": 416,
"acc": 0.7427884615384616,
"chance": 0.5,
"nll": 0.5331332034928402,
"brier": 0.35988097897984717,
"heldout": true,
"domain": "guardrails",
"ece": 0.06706968007179408,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "anli/nli",
"n": 498,
"acc": 0.41566265060240964,
"chance": 0.33333333333333326,
"nll": 1.1498472186171984,
"brier": 0.7016487136039561,
"heldout": false,
"domain": "reasoning",
"ece": 0.18407419437624845,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "aqua_rat/math",
"n": 247,
"acc": 0.27530364372469635,
"chance": 0.2,
"nll": 1.6108178135746645,
"brier": 0.7991190068538744,
"heldout": false,
"domain": "reasoning",
"ece": 0.04157990880823329,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "arc_challenge/mcq",
"n": 500,
"acc": 0.46,
"chance": 0.25006666666666666,
"nll": 1.2087727874126286,
"brier": 0.653150732152712,
"heldout": true,
"domain": "reasoning",
"ece": 0.08824656188488007,
"proof_v2 (card)": 0.308,
"\u0394 vs proof_v2": 0.15200000000000002
},
{
"task": "arc_easy/mcq",
"n": 500,
"acc": 0.594,
"chance": 0.24980000000000002,
"nll": 0.9366558539052494,
"brier": 0.5142523023860749,
"heldout": true,
"domain": "reasoning",
"ece": 0.0377397992014885,
"proof_v2 (card)": 0.425,
"\u0394 vs proof_v2": 0.16899999999999998
},
{
"task": "banking77/intent",
"n": 500,
"acc": 0.93,
"chance": 0.3214666666666666,
"nll": 0.21486007268911725,
"brier": 0.11047492184269238,
"heldout": true,
"domain": "intents",
"ece": 0.04031616294384006,
"proof_v2 (card)": 0.883,
"\u0394 vs proof_v2": 0.04700000000000004
},
{
"task": "banking77/intent_77",
"n": 300,
"acc": 0.5333333333333333,
"chance": 0.012987012987012986,
"nll": 2.063840282613334,
"brier": 0.6459762259581862,
"heldout": true,
"domain": "intents",
"ece": 0.14717036068439482,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "bigclonebench/clone",
"n": 500,
"acc": 0.952,
"chance": 0.5,
"nll": 0.14842138955372502,
"brier": 0.07646761075289837,
"heldout": false,
"domain": "code",
"ece": 0.015586107730865495,
"proof_v2 (card)": 0.383,
"\u0394 vs proof_v2": 0.569
},
{
"task": "bitext/category",
"n": 500,
"acc": 1.0,
"chance": 0.16128124098124094,
"nll": 0.0018036427325441764,
"brier": 0.000553129204988908,
"heldout": false,
"domain": "support",
"ece": 0.0015759592056274496,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "bitext/route",
"n": 500,
"acc": 1.0,
"chance": 0.19016313131313134,
"nll": 0.0011070070102664432,
"brier": 7.369536871875006e-05,
"heldout": false,
"domain": "support",
"ece": 0.0010853933095931502,
"proof_v2 (card)": 0.958,
"\u0394 vs proof_v2": 0.04200000000000004
},
{
"task": "boolq/yes_no",
"n": 500,
"acc": 0.816,
"chance": 0.5,
"nll": 0.4159143784940243,
"brier": 0.26228454396614875,
"heldout": false,
"domain": "reasoning",
"ece": 0.02717309403419496,
"proof_v2 (card)": 0.717,
"\u0394 vs proof_v2": 0.09899999999999998
},
{
"task": "civil_comments/toxic",
"n": 500,
"acc": 0.906,
"chance": 0.5,
"nll": 0.2400463573904126,
"brier": 0.13785060332637392,
"heldout": false,
"domain": "guardrails",
"ece": 0.035369446873664855,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "clinc150/intent",
"n": 500,
"acc": 0.942,
"chance": 0.12404474984829318,
"nll": 0.16339450041492637,
"brier": 0.08445263661020003,
"heldout": false,
"domain": "intents",
"ece": 0.036179060459136984,
"proof_v2 (card)": 0.85,
"\u0394 vs proof_v2": 0.09199999999999997
},
{
"task": "codexglue/code_to_doc",
"n": 504,
"acc": 0.9841269841269841,
"chance": 0.26233465608465617,
"nll": 0.04367021227094591,
"brier": 0.02423435580505018,
"heldout": false,
"domain": "code",
"ece": 0.011909238107147682,
"proof_v2 (card)": 0.969,
"\u0394 vs proof_v2": 0.0151269841269841
},
{
"task": "codexglue/doc_to_code",
"n": 504,
"acc": 0.9761904761904762,
"chance": 0.2786044973544974,
"nll": 0.05729491385996503,
"brier": 0.030591094765289834,
"heldout": false,
"domain": "code",
"ece": 0.01021853999959103,
"proof_v2 (card)": 0.961,
"\u0394 vs proof_v2": 0.015190476190476199
},
{
"task": "codexglue/func_name",
"n": 467,
"acc": 0.9528907922912205,
"chance": 0.25499643112062814,
"nll": 0.17043053883545994,
"brier": 0.08725536960997705,
"heldout": false,
"domain": "code",
"ece": 0.023257157468183163,
"proof_v2 (card)": 0.901,
"\u0394 vs proof_v2": 0.05189079229122051
},
{
"task": "codexglue/lang_id",
"n": 504,
"acc": 1.0,
"chance": 0.2354166666666667,
"nll": 0.0005584346231148719,
"brier": 3.0963766792822474e-06,
"heldout": false,
"domain": "code",
"ece": 0.0005574606004214999,
"proof_v2 (card)": 0.997,
"\u0394 vs proof_v2": 0.0030000000000000027
},
{
"task": "commonsense_qa/mcq",
"n": 493,
"acc": 0.6389452332657201,
"chance": 0.19999999999999998,
"nll": 0.904400206106629,
"brier": 0.47107552904904443,
"heldout": false,
"domain": "reasoning",
"ece": 0.04405802657589711,
"proof_v2 (card)": 0.442,
"\u0394 vs proof_v2": 0.19694523326572005
},
{
"task": "counsel/critique_quality",
"n": 201,
"acc": 0.5920398009950248,
"chance": 0.3333333333333333,
"nll": 1.3054839486733487,
"brier": 0.6194979652829024,
"heldout": false,
"domain": "agentic",
"ece": 0.22704330650135063,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "counsel/step_has_error",
"n": 201,
"acc": 0.8009950248756219,
"chance": 0.5,
"nll": 0.8227317674037697,
"brier": 0.3473574496250388,
"heldout": false,
"domain": "agentic",
"ece": 0.16924872979595879,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "devign/vulnerability",
"n": 500,
"acc": 0.596,
"chance": 0.5,
"nll": 0.6676419580578804,
"brier": 0.4753553111644763,
"heldout": false,
"domain": "code",
"ece": 0.05009425449371341,
"proof_v2 (card)": 0.542,
"\u0394 vs proof_v2": 0.05399999999999994
},
{
"task": "emotion/6way",
"n": 500,
"acc": 0.502,
"chance": 0.16666666666666663,
"nll": 1.3070907976925372,
"brier": 0.6468766489597455,
"heldout": true,
"domain": "classification",
"ece": 0.08215402710437775,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "gsm8k/math",
"n": 500,
"acc": 0.662,
"chance": 0.25,
"nll": 0.7718028125888668,
"brier": 0.4427847443219045,
"heldout": false,
"domain": "reasoning",
"ece": 0.06086966800689696,
"proof_v2 (card)": 0.275,
"\u0394 vs proof_v2": 0.387
},
{
"task": "hellaswag/continuation",
"n": 500,
"acc": 0.58,
"chance": 0.25,
"nll": 0.9953804229423404,
"brier": 0.5371978334623486,
"heldout": false,
"domain": "reasoning",
"ece": 0.07333891946077346,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "hotpotqa/comparison_yes_no",
"n": 26,
"acc": 0.8461538461538461,
"chance": 0.5,
"nll": 0.43042818936877525,
"brier": 0.2727966735853732,
"heldout": false,
"domain": "agentic",
"ece": 0.15112700829139125,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "hotpotqa/retrieve",
"n": 497,
"acc": 0.8732394366197183,
"chance": 0.16741720162243304,
"nll": 0.4343729186731075,
"brier": 0.20390003436957638,
"heldout": false,
"domain": "agentic",
"ece": 0.08316291382974779,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "humaneval/completion",
"n": 119,
"acc": 0.6722689075630253,
"chance": 0.39355742296918783,
"nll": 0.7120104984726616,
"brier": 0.44585727429903876,
"heldout": true,
"domain": "code",
"ece": 0.10659772783768277,
"proof_v2 (card)": 0.575,
"\u0394 vs proof_v2": 0.09726890756302531
},
{
"task": "jailbreak/detect",
"n": 262,
"acc": 0.9541984732824428,
"chance": 0.5,
"nll": 0.10692274935157883,
"brier": 0.06256811361157136,
"heldout": false,
"domain": "guardrails",
"ece": 0.0232666851455019,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "massive_en/intent",
"n": 500,
"acc": 0.906,
"chance": 0.12958727789085603,
"nll": 0.26250843075065133,
"brier": 0.12679194471370903,
"heldout": false,
"domain": "intents",
"ece": 0.022001883208751682,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "mbpp/bugspot",
"n": 256,
"acc": 0.73046875,
"chance": 0.40559895833333326,
"nll": 0.5201096873639273,
"brier": 0.3294220878204956,
"heldout": false,
"domain": "code",
"ece": 0.1015084749087691,
"proof_v2 (card)": 0.475,
"\u0394 vs proof_v2": 0.25546875
},
{
"task": "mbpp/solution",
"n": 500,
"acc": 0.982,
"chance": 0.25,
"nll": 0.06299449819558049,
"brier": 0.03283427225700136,
"heldout": false,
"domain": "code",
"ece": 0.014863664209842696,
"proof_v2 (card)": 0.992,
"\u0394 vs proof_v2": -0.010000000000000009
},
{
"task": "mmlu/mcq",
"n": 500,
"acc": 0.356,
"chance": 0.25,
"nll": 1.370702707901597,
"brier": 0.738678903692766,
"heldout": true,
"domain": "reasoning",
"ece": 0.10921649205684662,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "mnli/claim",
"n": 500,
"acc": 0.838,
"chance": 0.33333333333333326,
"nll": 0.41442326370440424,
"brier": 0.23035331540149503,
"heldout": false,
"domain": "reasoning",
"ece": 0.07134601444005968,
"proof_v2 (card)": 0.492,
"\u0394 vs proof_v2": 0.346
},
{
"task": "openbookqa/mcq",
"n": 500,
"acc": 0.61,
"chance": 0.25,
"nll": 1.024268753977958,
"brier": 0.5355113333378367,
"heldout": false,
"domain": "reasoning",
"ece": 0.09224175179004669,
"proof_v2 (card)": 0.292,
"\u0394 vs proof_v2": 0.318
},
{
"task": "policy/access_control_transfer",
"n": 500,
"acc": 1.0,
"chance": 0.33333333333333326,
"nll": 0.0002645651292309594,
"brier": 1.1635768837216181e-06,
"heldout": false,
"domain": "policy",
"ece": 0.00026422095298772597,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/count_threshold_transfer",
"n": 500,
"acc": 0.846,
"chance": 0.17848809523809525,
"nll": 0.4453474920627887,
"brier": 0.239936219050859,
"heldout": false,
"domain": "policy",
"ece": 0.10926523756980895,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/free_shipping_transfer",
"n": 500,
"acc": 0.766,
"chance": 0.5,
"nll": 0.39174148760084065,
"brier": 0.2747179012451008,
"heldout": false,
"domain": "policy",
"ece": 0.10137953925132755,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/invoice_overdue_transfer",
"n": 500,
"acc": 0.6,
"chance": 0.5,
"nll": 0.8301094107693061,
"brier": 0.5458620193712137,
"heldout": false,
"domain": "policy",
"ece": 0.20100946557521818,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/invoice_total_transfer",
"n": 500,
"acc": 0.486,
"chance": 0.5,
"nll": 0.69399339812994,
"brier": 0.5008197939882997,
"heldout": false,
"domain": "policy",
"ece": 0.03210310280323031,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/refund_approval_transfer",
"n": 500,
"acc": 0.78,
"chance": 0.33333333333333326,
"nll": 0.8478962361293161,
"brier": 0.38071118861218445,
"heldout": false,
"domain": "policy",
"ece": 0.18587953794002532,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/return_window_transfer",
"n": 500,
"acc": 0.998,
"chance": 0.33333333333333326,
"nll": 0.02459742584443666,
"brier": 0.005749234483560448,
"heldout": false,
"domain": "policy",
"ece": 0.019777906179428068,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/sla_urgency_transfer",
"n": 500,
"acc": 0.698,
"chance": 0.25,
"nll": 0.9843494065143168,
"brier": 0.49182705060356263,
"heldout": false,
"domain": "policy",
"ece": 0.21234694278240196,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "policy/table_extreme_transfer",
"n": 500,
"acc": 0.912,
"chance": 0.23716666666666666,
"nll": 0.22187845180515342,
"brier": 0.12271458379265106,
"heldout": false,
"domain": "policy",
"ece": 0.013898970842361484,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "prompt_injections/detect",
"n": 116,
"acc": 0.5775862068965517,
"chance": 0.5,
"nll": 1.0671347150446622,
"brier": 0.628204618498277,
"heldout": true,
"domain": "guardrails",
"ece": 0.2859799378904803,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "qasc/mcq",
"n": 500,
"acc": 0.99,
"chance": 0.125,
"nll": 0.0315895584258742,
"brier": 0.01700795004425532,
"heldout": false,
"domain": "reasoning",
"ece": 0.006339998722076429,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "sciq/mcq",
"n": 498,
"acc": 0.9538152610441767,
"chance": 0.25,
"nll": 0.11897318925779031,
"brier": 0.06399753046801272,
"heldout": false,
"domain": "reasoning",
"ece": 0.03059945360244999,
"proof_v2 (card)": 0.692,
"\u0394 vs proof_v2": 0.26181526104417674
},
{
"task": "scitail/support",
"n": 500,
"acc": 0.964,
"chance": 0.5,
"nll": 0.10391226119734347,
"brier": 0.05616228693429717,
"heldout": false,
"domain": "reasoning",
"ece": 0.020327572941780014,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "snli/contradicts",
"n": 500,
"acc": 0.986,
"chance": 0.33333333333333326,
"nll": 0.048629490838502536,
"brier": 0.024212671137030826,
"heldout": false,
"domain": "reasoning",
"ece": 0.01413622707128527,
"proof_v2 (card)": 0.892,
"\u0394 vs proof_v2": 0.09399999999999997
},
{
"task": "snli/must_be_true",
"n": 500,
"acc": 0.98,
"chance": 0.33333333333333326,
"nll": 0.060363937412155795,
"brier": 0.02947848703942969,
"heldout": false,
"domain": "reasoning",
"ece": 0.02905823123455048,
"proof_v2 (card)": 0.908,
"\u0394 vs proof_v2": 0.07199999999999995
},
{
"task": "snli/nli",
"n": 988,
"acc": 0.8795546558704453,
"chance": 0.33333333333333326,
"nll": 0.3359528171680472,
"brier": 0.18385021007121183,
"heldout": false,
"domain": "reasoning",
"ece": 0.04961462647688053,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "sst5/score",
"n": 500,
"acc": 0.416,
"chance": 0.2,
"nll": 1.246531030535698,
"brier": 0.6734807272262489,
"heldout": true,
"domain": "classification",
"ece": 0.04567561271786689,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "typed_decisions/agent_trace_observability",
"n": 500,
"acc": 0.712,
"chance": 0.3,
"nll": 0.7305186161920428,
"brier": 0.42510022971515116,
"heldout": false,
"domain": "workflows",
"ece": 0.11957757359743115,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "typed_decisions/customer_service",
"n": 500,
"acc": 0.74,
"chance": 0.28,
"nll": 0.6714302012249828,
"brier": 0.3723593006177565,
"heldout": false,
"domain": "workflows",
"ece": 0.08548467075824738,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "typed_decisions/invoice_processing",
"n": 500,
"acc": 0.838,
"chance": 0.35,
"nll": 0.5010932807754725,
"brier": 0.26905854083557984,
"heldout": false,
"domain": "workflows",
"ece": 0.12220623338222505,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "typed_decisions/security_incidents",
"n": 500,
"acc": 0.73,
"chance": 0.34,
"nll": 0.6859722163826227,
"brier": 0.3929527039537006,
"heldout": false,
"domain": "workflows",
"ece": 0.143579844892025,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "winogrande/blank",
"n": 500,
"acc": 0.612,
"chance": 0.5,
"nll": 0.7296091562472283,
"brier": 0.508893263835874,
"heldout": false,
"domain": "reasoning",
"ece": 0.11118709647655486,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
},
{
"task": "yelp/score",
"n": 500,
"acc": 0.654,
"chance": 0.2,
"nll": 0.7885563414096832,
"brier": 0.4562026964016218,
"heldout": false,
"domain": "classification",
"ece": 0.06915212672948838,
"proof_v2 (card)": NaN,
"\u0394 vs proof_v2": NaN
}
],
"head_to_head": null,
"env": {
"torch": "2.11.0+cu128",
"transformers": "5.17.0",
"python": "3.14.4",
"gpu": "NVIDIA GeForce RTX 5090 Laptop GPU"
}
}