加入实验输出(批次 12/14)

This commit is contained in:
2026-09-23 23:37:16 +08:00
parent d4c11396e0
commit 0c9c62de03
121 changed files with 6082 additions and 0 deletions
@@ -0,0 +1,41 @@
method,seed,fold,accuracy,macro_f1,mae,pearson,n_train,n_validation
M1,fixed,1,0.8,0.736111111111111,0.4612815961241722,0.3908651518384776,80,20
M2,fixed,1,0.75,0.393939393939394,0.4419183313846588,0.4236518393081548,80,20
M3,42,1,0.8,0.736111111111111,0.4484931230545044,0.41279811854541476,80,20
M4,42,1,0.8,0.6344086021505376,0.4766388975083828,0.43187935519491855,80,20
M3,3407,1,0.8,0.736111111111111,0.44865339174866675,0.4171944975077341,80,20
M4,3407,1,0.8,0.6344086021505376,0.47647997066378595,0.4320761354143285,80,20
M3,2026,1,0.8,0.736111111111111,0.4473679281771183,0.419287895205426,80,20
M4,2026,1,0.8,0.6344086021505376,0.47662248760461806,0.43167719260246934,80,20
M1,fixed,2,0.6,0.44523809523809527,0.7316692218184471,0.6480682206513965,80,20
M2,fixed,2,0.6,0.44523809523809527,0.7364562056958676,0.6535275734065676,80,20
M3,42,2,0.7,0.5598290598290597,0.7191767528653145,0.6829254472780363,80,20
M4,42,2,0.6,0.48939393939393944,0.8361533507704735,0.5952034035922986,80,20
M3,3407,2,0.65,0.4764957264957264,0.7146183483302593,0.6888039663406401,80,20
M4,3407,2,0.6,0.48939393939393944,0.8360007748007774,0.595177745117279,80,20
M3,2026,2,0.65,0.47252747252747246,0.7159126169979573,0.689385710098449,80,20
M4,2026,2,0.6,0.48939393939393944,0.8361966282129287,0.5951825310175703,80,20
M1,fixed,3,0.7,0.5731922398589065,0.6056819573044777,0.5391663374393444,80,20
M2,fixed,3,0.65,0.48412698412698413,0.6305533096194267,0.4961935349272063,80,20
M3,42,3,0.65,0.5421245421245421,0.6378875352442265,0.5325578720674686,80,20
M4,42,3,0.55,0.34215167548500885,0.6873359516263008,0.5315284769823068,80,20
M3,3407,3,0.65,0.5421245421245421,0.626622511446476,0.5478317984415519,80,20
M4,3407,3,0.55,0.34215167548500885,0.6872267439961434,0.5313424475222814,80,20
M3,2026,3,0.7,0.5731922398589065,0.6307285942137242,0.5464569461654115,80,20
M4,2026,3,0.55,0.34215167548500885,0.6871870808303356,0.5319799133727069,80,20
M1,fixed,4,0.4,0.19047619047619047,0.5336167700588703,0.5772197323434819,80,20
M2,fixed,4,0.3,0.15384615384615385,0.5659813567996025,0.5500284823994642,80,20
M3,42,4,0.3,0.16,0.5302071489393712,0.5725429629109834,80,20
M4,42,4,0.35,0.2711111111111111,0.5166212148964405,0.6177149095726567,80,20
M3,3407,4,0.3,0.15384615384615385,0.5506807617843151,0.5860130646535472,80,20
M4,3407,4,0.35,0.2711111111111111,0.5164451241493225,0.6181052902824066,80,20
M3,2026,4,0.3,0.16,0.531651595979929,0.5564093983759777,80,20
M4,2026,4,0.35,0.2711111111111111,0.5163851425051689,0.617812107663041,80,20
M1,fixed,5,0.45,0.24,0.4284325443208218,0.49726227320567845,80,20
M2,fixed,5,0.45,0.24,0.47108725607395174,0.4178180837146751,80,20
M3,42,5,0.45,0.24,0.44473609402775766,0.4783453741103445,80,20
M4,42,5,0.5,0.3352380952380953,0.5995716892182827,0.35723746565175724,80,20
M3,3407,5,0.45,0.24,0.4421664297580719,0.4940687599587745,80,20
M4,3407,5,0.5,0.3352380952380953,0.5993113562464714,0.35800549475908866,80,20
M3,2026,5,0.45,0.24,0.4126793794333935,0.5612399837857652,80,20
M4,2026,5,0.5,0.3352380952380953,0.5996754877269268,0.3568218465158214,80,20
1 method seed fold accuracy macro_f1 mae pearson n_train n_validation
2 M1 fixed 1 0.8 0.736111111111111 0.4612815961241722 0.3908651518384776 80 20
3 M2 fixed 1 0.75 0.393939393939394 0.4419183313846588 0.4236518393081548 80 20
4 M3 42 1 0.8 0.736111111111111 0.4484931230545044 0.41279811854541476 80 20
5 M4 42 1 0.8 0.6344086021505376 0.4766388975083828 0.43187935519491855 80 20
6 M3 3407 1 0.8 0.736111111111111 0.44865339174866675 0.4171944975077341 80 20
7 M4 3407 1 0.8 0.6344086021505376 0.47647997066378595 0.4320761354143285 80 20
8 M3 2026 1 0.8 0.736111111111111 0.4473679281771183 0.419287895205426 80 20
9 M4 2026 1 0.8 0.6344086021505376 0.47662248760461806 0.43167719260246934 80 20
10 M1 fixed 2 0.6 0.44523809523809527 0.7316692218184471 0.6480682206513965 80 20
11 M2 fixed 2 0.6 0.44523809523809527 0.7364562056958676 0.6535275734065676 80 20
12 M3 42 2 0.7 0.5598290598290597 0.7191767528653145 0.6829254472780363 80 20
13 M4 42 2 0.6 0.48939393939393944 0.8361533507704735 0.5952034035922986 80 20
14 M3 3407 2 0.65 0.4764957264957264 0.7146183483302593 0.6888039663406401 80 20
15 M4 3407 2 0.6 0.48939393939393944 0.8360007748007774 0.595177745117279 80 20
16 M3 2026 2 0.65 0.47252747252747246 0.7159126169979573 0.689385710098449 80 20
17 M4 2026 2 0.6 0.48939393939393944 0.8361966282129287 0.5951825310175703 80 20
18 M1 fixed 3 0.7 0.5731922398589065 0.6056819573044777 0.5391663374393444 80 20
19 M2 fixed 3 0.65 0.48412698412698413 0.6305533096194267 0.4961935349272063 80 20
20 M3 42 3 0.65 0.5421245421245421 0.6378875352442265 0.5325578720674686 80 20
21 M4 42 3 0.55 0.34215167548500885 0.6873359516263008 0.5315284769823068 80 20
22 M3 3407 3 0.65 0.5421245421245421 0.626622511446476 0.5478317984415519 80 20
23 M4 3407 3 0.55 0.34215167548500885 0.6872267439961434 0.5313424475222814 80 20
24 M3 2026 3 0.7 0.5731922398589065 0.6307285942137242 0.5464569461654115 80 20
25 M4 2026 3 0.55 0.34215167548500885 0.6871870808303356 0.5319799133727069 80 20
26 M1 fixed 4 0.4 0.19047619047619047 0.5336167700588703 0.5772197323434819 80 20
27 M2 fixed 4 0.3 0.15384615384615385 0.5659813567996025 0.5500284823994642 80 20
28 M3 42 4 0.3 0.16 0.5302071489393712 0.5725429629109834 80 20
29 M4 42 4 0.35 0.2711111111111111 0.5166212148964405 0.6177149095726567 80 20
30 M3 3407 4 0.3 0.15384615384615385 0.5506807617843151 0.5860130646535472 80 20
31 M4 3407 4 0.35 0.2711111111111111 0.5164451241493225 0.6181052902824066 80 20
32 M3 2026 4 0.3 0.16 0.531651595979929 0.5564093983759777 80 20
33 M4 2026 4 0.35 0.2711111111111111 0.5163851425051689 0.617812107663041 80 20
34 M1 fixed 5 0.45 0.24 0.4284325443208218 0.49726227320567845 80 20
35 M2 fixed 5 0.45 0.24 0.47108725607395174 0.4178180837146751 80 20
36 M3 42 5 0.45 0.24 0.44473609402775766 0.4783453741103445 80 20
37 M4 42 5 0.5 0.3352380952380953 0.5995716892182827 0.35723746565175724 80 20
38 M3 3407 5 0.45 0.24 0.4421664297580719 0.4940687599587745 80 20
39 M4 3407 5 0.5 0.3352380952380953 0.5993113562464714 0.35800549475908866 80 20
40 M3 2026 5 0.45 0.24 0.4126793794333935 0.5612399837857652 80 20
41 M4 2026 5 0.5 0.3352380952380953 0.5996754877269268 0.3568218465158214 80 20
@@ -0,0 +1,121 @@
method,seed,fold,target_modality,mask_ratio,mae_standardized,smooth_l1_standardized,masked_values
M1,fixed,1,text,0.2,0.7864587306976318,0.41599002480506897,153600
M1,fixed,1,audio,0.2,0.6423318386077881,0.2985542118549347,5000
M1,fixed,1,vision,0.2,0.8732286691665649,0.49052613973617554,38400
M2,fixed,1,text,0.2,0.7657629251480103,0.3995133638381958,153600
M2,fixed,1,audio,0.2,0.6078358292579651,0.27118614315986633,5000
M2,fixed,1,vision,0.2,0.8754072189331055,0.49298685789108276,38400
M3,42,1,text,0.2,0.7869179248809814,0.4176875948905945,153600
M3,42,1,audio,0.2,0.3639732003211975,0.1077285036444664,5000
M3,42,1,vision,0.2,0.8297806978225708,0.4545438885688782,38400
M4,42,1,text,0.2,0.4791339337825775,0.17556655406951904,153600
M4,42,1,audio,0.2,0.31638050079345703,0.07683445513248444,5000
M4,42,1,vision,0.2,0.8184546828269958,0.44288504123687744,38400
M3,3407,1,text,0.2,0.7894418239593506,0.4194052815437317,153600
M3,3407,1,audio,0.2,0.3394777178764343,0.09327597916126251,5000
M3,3407,1,vision,0.2,0.8291587233543396,0.4534059762954712,38400
M4,3407,1,text,0.2,0.4733425974845886,0.17110656201839447,153600
M4,3407,1,audio,0.2,0.33666083216667175,0.08810637891292572,5000
M4,3407,1,vision,0.2,0.817406415939331,0.44359228014945984,38400
M3,2026,1,text,0.2,0.7847438454627991,0.41536271572113037,153600
M3,2026,1,audio,0.2,0.35511481761932373,0.10340271890163422,5000
M3,2026,1,vision,0.2,0.8283053636550903,0.45340079069137573,38400
M4,2026,1,text,0.2,0.47494950890541077,0.1725955754518509,153600
M4,2026,1,audio,0.2,0.32631760835647583,0.08271428942680359,5000
M4,2026,1,vision,0.2,0.8221946954727173,0.4466344714164734,38400
M1,fixed,2,text,0.2,0.8431306481361389,0.46495506167411804,153600
M1,fixed,2,audio,0.2,0.6558716893196106,0.31570741534233093,5000
M1,fixed,2,vision,0.2,0.8846356272697449,0.5022978782653809,38400
M2,fixed,2,text,0.2,0.8220721483230591,0.4483771324157715,153600
M2,fixed,2,audio,0.2,0.6195802688598633,0.28658542037010193,5000
M2,fixed,2,vision,0.2,0.8831602931022644,0.5007410049438477,38400
M3,42,2,text,0.2,0.8472743034362793,0.4686535894870758,153600
M3,42,2,audio,0.2,0.34122517704963684,0.11146467179059982,5000
M3,42,2,vision,0.2,0.8413732051849365,0.4642646312713623,38400
M4,42,2,text,0.2,0.5017852187156677,0.19247683882713318,153600
M4,42,2,audio,0.2,0.34099525213241577,0.1067037358880043,5000
M4,42,2,vision,0.2,0.8211991190910339,0.4461081624031067,38400
M3,3407,2,text,0.2,0.8380216360092163,0.4611693322658539,153600
M3,3407,2,audio,0.2,0.3393073081970215,0.10981127619743347,5000
M3,3407,2,vision,0.2,0.8233676552772522,0.44843095541000366,38400
M4,3407,2,text,0.2,0.49964815378189087,0.1912464201450348,153600
M4,3407,2,audio,0.2,0.3402937948703766,0.11066138744354248,5000
M4,3407,2,vision,0.2,0.8373452425003052,0.46078431606292725,38400
M3,2026,2,text,0.2,0.8452751636505127,0.4673306345939636,153600
M3,2026,2,audio,0.2,0.3425142467021942,0.1117163822054863,5000
M3,2026,2,vision,0.2,0.8416237831115723,0.4639897644519806,38400
M4,2026,2,text,0.2,0.5024572610855103,0.19227026402950287,153600
M4,2026,2,audio,0.2,0.3313565254211426,0.10437174141407013,5000
M4,2026,2,vision,0.2,0.8250595331192017,0.45077505707740784,38400
M1,fixed,3,text,0.2,0.8600677847862244,0.48109835386276245,153600
M1,fixed,3,audio,0.2,0.6301218867301941,0.28287407755851746,5000
M1,fixed,3,vision,0.2,0.9313846826553345,0.543266236782074,38400
M2,fixed,3,text,0.2,0.8661688566207886,0.4862343370914459,153600
M2,fixed,3,audio,0.2,0.6363305449485779,0.295511931180954,5000
M2,fixed,3,vision,0.2,0.9410251379013062,0.5511942505836487,38400
M3,42,3,text,0.2,0.8575569987297058,0.4784933626651764,153600
M3,42,3,audio,0.2,0.35879722237586975,0.10602780431509018,5000
M3,42,3,vision,0.2,0.848150908946991,0.4709389805793762,38400
M4,42,3,text,0.2,0.5080369114875793,0.1971929669380188,153600
M4,42,3,audio,0.2,0.35229456424713135,0.09507919102907181,5000
M4,42,3,vision,0.2,0.8596658706665039,0.4819534420967102,38400
M3,3407,3,text,0.2,0.8525339365005493,0.4742853045463562,153600
M3,3407,3,audio,0.2,0.3356912434101105,0.09289933741092682,5000
M3,3407,3,vision,0.2,0.852375328540802,0.4742961823940277,38400
M4,3407,3,text,0.2,0.5013428926467896,0.1922634392976761,153600
M4,3407,3,audio,0.2,0.3563744127750397,0.09956291317939758,5000
M4,3407,3,vision,0.2,0.8604409098625183,0.48151978850364685,38400
M3,2026,3,text,0.2,0.8605049848556519,0.48118850588798523,153600
M3,2026,3,audio,0.2,0.36211082339286804,0.10407590121030807,5000
M3,2026,3,vision,0.2,0.859515905380249,0.4795020520687103,38400
M4,2026,3,text,0.2,0.5066819787025452,0.1963006854057312,153600
M4,2026,3,audio,0.2,0.3566948175430298,0.09866075217723846,5000
M4,2026,3,vision,0.2,0.8549585342407227,0.4780130684375763,38400
M1,fixed,4,text,0.2,0.8199025392532349,0.44484803080558777,153600
M1,fixed,4,audio,0.2,0.7768001556396484,0.4197060465812683,5000
M1,fixed,4,vision,0.2,0.9551154971122742,0.5669697523117065,38400
M2,fixed,4,text,0.2,0.8191158175468445,0.4448056221008301,153600
M2,fixed,4,audio,0.2,0.7202740907669067,0.3684014081954956,5000
M2,fixed,4,vision,0.2,0.9408901929855347,0.5558907389640808,38400
M3,42,4,text,0.2,0.8254697918891907,0.4495771527290344,153600
M3,42,4,audio,0.2,0.417472243309021,0.15880697965621948,5000
M3,42,4,vision,0.2,0.9076664447784424,0.5263801217079163,38400
M4,42,4,text,0.2,0.4919026792049408,0.1856606900691986,153600
M4,42,4,audio,0.2,0.4052186906337738,0.15105251967906952,5000
M4,42,4,vision,0.2,0.8914499878883362,0.5127754807472229,38400
M3,3407,4,text,0.2,0.8279706835746765,0.4522389769554138,153600
M3,3407,4,audio,0.2,0.39376136660575867,0.14974132180213928,5000
M3,3407,4,vision,0.2,0.9084181785583496,0.5274486541748047,38400
M4,3407,4,text,0.2,0.4878115952014923,0.18323160707950592,153600
M4,3407,4,audio,0.2,0.40141838788986206,0.15048016607761383,5000
M4,3407,4,vision,0.2,0.8952069282531738,0.5167335867881775,38400
M3,2026,4,text,0.2,0.8241562843322754,0.4491640329360962,153600
M3,2026,4,audio,0.2,0.40933120250701904,0.1570354402065277,5000
M3,2026,4,vision,0.2,0.9041170477867126,0.5228397846221924,38400
M4,2026,4,text,0.2,0.49553707242012024,0.18893885612487793,153600
M4,2026,4,audio,0.2,0.39918839931488037,0.14919213950634003,5000
M4,2026,4,vision,0.2,0.8958951830863953,0.5186710953712463,38400
M1,fixed,5,text,0.2,0.8444457650184631,0.4662260413169861,153600
M1,fixed,5,audio,0.2,0.7492777109146118,0.41332030296325684,5000
M1,fixed,5,vision,0.2,0.9604420065879822,0.5708574652671814,38400
M2,fixed,5,text,0.2,0.8222289085388184,0.44781213998794556,153600
M2,fixed,5,audio,0.2,0.7265015840530396,0.3929823040962219,5000
M2,fixed,5,vision,0.2,0.9572133421897888,0.5693861842155457,38400
M3,42,5,text,0.2,0.847065269947052,0.4688800573348999,153600
M3,42,5,audio,0.2,0.47439461946487427,0.2379949688911438,5000
M3,42,5,vision,0.2,0.91214519739151,0.527710497379303,38400
M4,42,5,text,0.2,0.49351584911346436,0.18649226427078247,153600
M4,42,5,audio,0.2,0.46833816170692444,0.23857423663139343,5000
M4,42,5,vision,0.2,0.9112464785575867,0.5267350673675537,38400
M3,3407,5,text,0.2,0.8362759947776794,0.4592466354370117,153600
M3,3407,5,audio,0.2,0.4684489071369171,0.23598544299602509,5000
M3,3407,5,vision,0.2,0.9186314344406128,0.5335334539413452,38400
M4,3407,5,text,0.2,0.48672324419021606,0.18171463906764984,153600
M4,3407,5,audio,0.2,0.4682857394218445,0.23653720319271088,5000
M4,3407,5,vision,0.2,0.9220170378684998,0.5346263647079468,38400
M3,2026,5,text,0.2,0.8434508442878723,0.46527090668678284,153600
M3,2026,5,audio,0.2,0.4658173620700836,0.23616507649421692,5000
M3,2026,5,vision,0.2,0.918639063835144,0.5329208374023438,38400
M4,2026,5,text,0.2,0.4841386675834656,0.18039487302303314,153600
M4,2026,5,audio,0.2,0.47616907954216003,0.2406836897134781,5000
M4,2026,5,vision,0.2,0.9134588837623596,0.5281268358230591,38400
1 method seed fold target_modality mask_ratio mae_standardized smooth_l1_standardized masked_values
2 M1 fixed 1 text 0.2 0.7864587306976318 0.41599002480506897 153600
3 M1 fixed 1 audio 0.2 0.6423318386077881 0.2985542118549347 5000
4 M1 fixed 1 vision 0.2 0.8732286691665649 0.49052613973617554 38400
5 M2 fixed 1 text 0.2 0.7657629251480103 0.3995133638381958 153600
6 M2 fixed 1 audio 0.2 0.6078358292579651 0.27118614315986633 5000
7 M2 fixed 1 vision 0.2 0.8754072189331055 0.49298685789108276 38400
8 M3 42 1 text 0.2 0.7869179248809814 0.4176875948905945 153600
9 M3 42 1 audio 0.2 0.3639732003211975 0.1077285036444664 5000
10 M3 42 1 vision 0.2 0.8297806978225708 0.4545438885688782 38400
11 M4 42 1 text 0.2 0.4791339337825775 0.17556655406951904 153600
12 M4 42 1 audio 0.2 0.31638050079345703 0.07683445513248444 5000
13 M4 42 1 vision 0.2 0.8184546828269958 0.44288504123687744 38400
14 M3 3407 1 text 0.2 0.7894418239593506 0.4194052815437317 153600
15 M3 3407 1 audio 0.2 0.3394777178764343 0.09327597916126251 5000
16 M3 3407 1 vision 0.2 0.8291587233543396 0.4534059762954712 38400
17 M4 3407 1 text 0.2 0.4733425974845886 0.17110656201839447 153600
18 M4 3407 1 audio 0.2 0.33666083216667175 0.08810637891292572 5000
19 M4 3407 1 vision 0.2 0.817406415939331 0.44359228014945984 38400
20 M3 2026 1 text 0.2 0.7847438454627991 0.41536271572113037 153600
21 M3 2026 1 audio 0.2 0.35511481761932373 0.10340271890163422 5000
22 M3 2026 1 vision 0.2 0.8283053636550903 0.45340079069137573 38400
23 M4 2026 1 text 0.2 0.47494950890541077 0.1725955754518509 153600
24 M4 2026 1 audio 0.2 0.32631760835647583 0.08271428942680359 5000
25 M4 2026 1 vision 0.2 0.8221946954727173 0.4466344714164734 38400
26 M1 fixed 2 text 0.2 0.8431306481361389 0.46495506167411804 153600
27 M1 fixed 2 audio 0.2 0.6558716893196106 0.31570741534233093 5000
28 M1 fixed 2 vision 0.2 0.8846356272697449 0.5022978782653809 38400
29 M2 fixed 2 text 0.2 0.8220721483230591 0.4483771324157715 153600
30 M2 fixed 2 audio 0.2 0.6195802688598633 0.28658542037010193 5000
31 M2 fixed 2 vision 0.2 0.8831602931022644 0.5007410049438477 38400
32 M3 42 2 text 0.2 0.8472743034362793 0.4686535894870758 153600
33 M3 42 2 audio 0.2 0.34122517704963684 0.11146467179059982 5000
34 M3 42 2 vision 0.2 0.8413732051849365 0.4642646312713623 38400
35 M4 42 2 text 0.2 0.5017852187156677 0.19247683882713318 153600
36 M4 42 2 audio 0.2 0.34099525213241577 0.1067037358880043 5000
37 M4 42 2 vision 0.2 0.8211991190910339 0.4461081624031067 38400
38 M3 3407 2 text 0.2 0.8380216360092163 0.4611693322658539 153600
39 M3 3407 2 audio 0.2 0.3393073081970215 0.10981127619743347 5000
40 M3 3407 2 vision 0.2 0.8233676552772522 0.44843095541000366 38400
41 M4 3407 2 text 0.2 0.49964815378189087 0.1912464201450348 153600
42 M4 3407 2 audio 0.2 0.3402937948703766 0.11066138744354248 5000
43 M4 3407 2 vision 0.2 0.8373452425003052 0.46078431606292725 38400
44 M3 2026 2 text 0.2 0.8452751636505127 0.4673306345939636 153600
45 M3 2026 2 audio 0.2 0.3425142467021942 0.1117163822054863 5000
46 M3 2026 2 vision 0.2 0.8416237831115723 0.4639897644519806 38400
47 M4 2026 2 text 0.2 0.5024572610855103 0.19227026402950287 153600
48 M4 2026 2 audio 0.2 0.3313565254211426 0.10437174141407013 5000
49 M4 2026 2 vision 0.2 0.8250595331192017 0.45077505707740784 38400
50 M1 fixed 3 text 0.2 0.8600677847862244 0.48109835386276245 153600
51 M1 fixed 3 audio 0.2 0.6301218867301941 0.28287407755851746 5000
52 M1 fixed 3 vision 0.2 0.9313846826553345 0.543266236782074 38400
53 M2 fixed 3 text 0.2 0.8661688566207886 0.4862343370914459 153600
54 M2 fixed 3 audio 0.2 0.6363305449485779 0.295511931180954 5000
55 M2 fixed 3 vision 0.2 0.9410251379013062 0.5511942505836487 38400
56 M3 42 3 text 0.2 0.8575569987297058 0.4784933626651764 153600
57 M3 42 3 audio 0.2 0.35879722237586975 0.10602780431509018 5000
58 M3 42 3 vision 0.2 0.848150908946991 0.4709389805793762 38400
59 M4 42 3 text 0.2 0.5080369114875793 0.1971929669380188 153600
60 M4 42 3 audio 0.2 0.35229456424713135 0.09507919102907181 5000
61 M4 42 3 vision 0.2 0.8596658706665039 0.4819534420967102 38400
62 M3 3407 3 text 0.2 0.8525339365005493 0.4742853045463562 153600
63 M3 3407 3 audio 0.2 0.3356912434101105 0.09289933741092682 5000
64 M3 3407 3 vision 0.2 0.852375328540802 0.4742961823940277 38400
65 M4 3407 3 text 0.2 0.5013428926467896 0.1922634392976761 153600
66 M4 3407 3 audio 0.2 0.3563744127750397 0.09956291317939758 5000
67 M4 3407 3 vision 0.2 0.8604409098625183 0.48151978850364685 38400
68 M3 2026 3 text 0.2 0.8605049848556519 0.48118850588798523 153600
69 M3 2026 3 audio 0.2 0.36211082339286804 0.10407590121030807 5000
70 M3 2026 3 vision 0.2 0.859515905380249 0.4795020520687103 38400
71 M4 2026 3 text 0.2 0.5066819787025452 0.1963006854057312 153600
72 M4 2026 3 audio 0.2 0.3566948175430298 0.09866075217723846 5000
73 M4 2026 3 vision 0.2 0.8549585342407227 0.4780130684375763 38400
74 M1 fixed 4 text 0.2 0.8199025392532349 0.44484803080558777 153600
75 M1 fixed 4 audio 0.2 0.7768001556396484 0.4197060465812683 5000
76 M1 fixed 4 vision 0.2 0.9551154971122742 0.5669697523117065 38400
77 M2 fixed 4 text 0.2 0.8191158175468445 0.4448056221008301 153600
78 M2 fixed 4 audio 0.2 0.7202740907669067 0.3684014081954956 5000
79 M2 fixed 4 vision 0.2 0.9408901929855347 0.5558907389640808 38400
80 M3 42 4 text 0.2 0.8254697918891907 0.4495771527290344 153600
81 M3 42 4 audio 0.2 0.417472243309021 0.15880697965621948 5000
82 M3 42 4 vision 0.2 0.9076664447784424 0.5263801217079163 38400
83 M4 42 4 text 0.2 0.4919026792049408 0.1856606900691986 153600
84 M4 42 4 audio 0.2 0.4052186906337738 0.15105251967906952 5000
85 M4 42 4 vision 0.2 0.8914499878883362 0.5127754807472229 38400
86 M3 3407 4 text 0.2 0.8279706835746765 0.4522389769554138 153600
87 M3 3407 4 audio 0.2 0.39376136660575867 0.14974132180213928 5000
88 M3 3407 4 vision 0.2 0.9084181785583496 0.5274486541748047 38400
89 M4 3407 4 text 0.2 0.4878115952014923 0.18323160707950592 153600
90 M4 3407 4 audio 0.2 0.40141838788986206 0.15048016607761383 5000
91 M4 3407 4 vision 0.2 0.8952069282531738 0.5167335867881775 38400
92 M3 2026 4 text 0.2 0.8241562843322754 0.4491640329360962 153600
93 M3 2026 4 audio 0.2 0.40933120250701904 0.1570354402065277 5000
94 M3 2026 4 vision 0.2 0.9041170477867126 0.5228397846221924 38400
95 M4 2026 4 text 0.2 0.49553707242012024 0.18893885612487793 153600
96 M4 2026 4 audio 0.2 0.39918839931488037 0.14919213950634003 5000
97 M4 2026 4 vision 0.2 0.8958951830863953 0.5186710953712463 38400
98 M1 fixed 5 text 0.2 0.8444457650184631 0.4662260413169861 153600
99 M1 fixed 5 audio 0.2 0.7492777109146118 0.41332030296325684 5000
100 M1 fixed 5 vision 0.2 0.9604420065879822 0.5708574652671814 38400
101 M2 fixed 5 text 0.2 0.8222289085388184 0.44781213998794556 153600
102 M2 fixed 5 audio 0.2 0.7265015840530396 0.3929823040962219 5000
103 M2 fixed 5 vision 0.2 0.9572133421897888 0.5693861842155457 38400
104 M3 42 5 text 0.2 0.847065269947052 0.4688800573348999 153600
105 M3 42 5 audio 0.2 0.47439461946487427 0.2379949688911438 5000
106 M3 42 5 vision 0.2 0.91214519739151 0.527710497379303 38400
107 M4 42 5 text 0.2 0.49351584911346436 0.18649226427078247 153600
108 M4 42 5 audio 0.2 0.46833816170692444 0.23857423663139343 5000
109 M4 42 5 vision 0.2 0.9112464785575867 0.5267350673675537 38400
110 M3 3407 5 text 0.2 0.8362759947776794 0.4592466354370117 153600
111 M3 3407 5 audio 0.2 0.4684489071369171 0.23598544299602509 5000
112 M3 3407 5 vision 0.2 0.9186314344406128 0.5335334539413452 38400
113 M4 3407 5 text 0.2 0.48672324419021606 0.18171463906764984 153600
114 M4 3407 5 audio 0.2 0.4682857394218445 0.23653720319271088 5000
115 M4 3407 5 vision 0.2 0.9220170378684998 0.5346263647079468 38400
116 M3 2026 5 text 0.2 0.8434508442878723 0.46527090668678284 153600
117 M3 2026 5 audio 0.2 0.4658173620700836 0.23616507649421692 5000
118 M3 2026 5 vision 0.2 0.918639063835144 0.5329208374023438 38400
119 M4 2026 5 text 0.2 0.4841386675834656 0.18039487302303314 153600
120 M4 2026 5 audio 0.2 0.47616907954216003 0.2406836897134781 5000
121 M4 2026 5 vision 0.2 0.9134588837623596 0.5281268358230591 38400
@@ -0,0 +1,13 @@
method,target_modality,n,mae_standardized_mean,mae_standardized_std,smooth_l1_standardized_mean,smooth_l1_standardized_std
M1,text,5,0.8308010935783386,0.028634584948895196,0.4546235024929047,0.025151457325080215
M1,audio,5,0.6908806562423706,0.06720606353302543,0.34603241086006165,0.06541831705153776
M1,vision,5,0.9209612965583801,0.04009937912435952,0.5347834944725036,0.036821434614539225
M2,text,5,0.8190697312355042,0.03563334190353864,0.44534851908683776,0.030773561031959877
M2,audio,5,0.6621044635772705,0.05689547728439605,0.32293344140052793,0.05414120404134669
M2,vision,5,0.9195392370223999,0.03744298345037278,0.5340398073196411,0.0346960350528028
M3,text,15,0.8311106324195862,0.025205971351551344,0.45519693891207375,0.021717713583393992
M3,audio,15,0.38449583053588865,0.050754623523549675,0.14107545365889868,0.05381054674326515
M3,vision,15,0.868217929204305,0.03802189028391118,0.48890710473060606,0.03450529817617854
M4,text,15,0.4924671709537506,0.011169848357006705,0.18583014905452727,0.008262501869688806
M4,audio,15,0.37839911778767904,0.05503994105535867,0.13528098662694296,0.05838370221384358
M4,vision,15,0.8630666335423788,0.03867536107911695,0.48466227054595945,0.03512809212998346
1 method target_modality n mae_standardized_mean mae_standardized_std smooth_l1_standardized_mean smooth_l1_standardized_std
2 M1 text 5 0.8308010935783386 0.028634584948895196 0.4546235024929047 0.025151457325080215
3 M1 audio 5 0.6908806562423706 0.06720606353302543 0.34603241086006165 0.06541831705153776
4 M1 vision 5 0.9209612965583801 0.04009937912435952 0.5347834944725036 0.036821434614539225
5 M2 text 5 0.8190697312355042 0.03563334190353864 0.44534851908683776 0.030773561031959877
6 M2 audio 5 0.6621044635772705 0.05689547728439605 0.32293344140052793 0.05414120404134669
7 M2 vision 5 0.9195392370223999 0.03744298345037278 0.5340398073196411 0.0346960350528028
8 M3 text 15 0.8311106324195862 0.025205971351551344 0.45519693891207375 0.021717713583393992
9 M3 audio 15 0.38449583053588865 0.050754623523549675 0.14107545365889868 0.05381054674326515
10 M3 vision 15 0.868217929204305 0.03802189028391118 0.48890710473060606 0.03450529817617854
11 M4 text 15 0.4924671709537506 0.011169848357006705 0.18583014905452727 0.008262501869688806
12 M4 audio 15 0.37839911778767904 0.05503994105535867 0.13528098662694296 0.05838370221384358
13 M4 vision 15 0.8630666335423788 0.03867536107911695 0.48466227054595945 0.03512809212998346
@@ -0,0 +1,3 @@
# M1-M4 comparison report bundle
This compact bundle contains grouped-CV metrics, fold assignments, run manifest, log, and example figures. The full local method-comparison directory also keeps 30 fold checkpoints and all out-of-fold alignment matrices; those bulky files are omitted here. Alignment claims remain limited by the absence of human event-time labels.
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,13 @@
method,modality,n,mvr_mean,mvr_std,normalized_entropy_mean,normalized_entropy_std,width80_source_positions_mean,width80_source_positions_std,expected_time_start_s_mean,expected_time_start_s_std,expected_time_end_s_mean,expected_time_end_s_std,trajectory_span_fraction_mean,trajectory_span_fraction_std
M1,text,100,0.0,0.0,0.0008249471336603165,0.005233484169432736,1.004799998998642,0.0312849358458409,0.373650000790672,0.28760723340035366,7.3229000100512325,4.567867965796829,0.8612861529795919,0.07875035859337948
M1,audio,100,0.0,0.0,0.35824004098773005,0.05432258900057586,12.056200053691864,7.307551875172266,0.3177500062027157,0.2794568898376313,7.441361117547106,4.533387053269832,0.8937758523714728,0.06338502509117576
M1,vision,100,0.0,0.0,0.02091189981205389,0.03423249778642653,1.144999988079071,0.27983220085257804,0.30618889875375566,0.2956530091396046,7.43663269007148,4.535796288658362,0.8956127409730107,0.0656268624353691
M2,text,100,0.0,0.0,0.003971234038472176,0.010095580232451877,1.0215999960899353,0.060615033764357146,0.3753062509072731,0.2858492171110105,7.3229000100512325,4.567867965796829,0.8611038638930768,0.07866203063075565
M2,audio,100,0.0,0.0,0.39199826538562776,0.04856653297051989,12.777399990558624,7.437441491467199,0.08135000219190694,0.046356409936463026,7.716148159798975,4.594836781525238,0.968455815287307,0.009448427161791137
M2,vision,100,0.0,0.0,0.015714890849776567,0.04120196183036711,1.108399999141693,0.3072087474461631,0.024000000567226686,0.0514830577428781,7.7077327980137875,4.617512750249008,0.9744537002407014,0.016307502074103464
M3,text,300,0.0,0.0,0.0004981429129838944,0.004964740878146852,1.0029999995231629,0.02989949245257891,0.37365000079067195,0.28664372237282515,7.320650003693404,4.545633324884766,0.8611844960658346,0.07838045740132776
M3,audio,300,0.01857142824679613,0.028241165293696132,0.992265993754069,0.00445452263162519,607.8159977213542,365.5015286317014,3.883805515007902,2.328319755482711,3.8684255829138965,2.2993181536785716,-0.001562372346719104,0.016018112979249148
M3,vision,300,0.06843537279715141,0.06722724739800441,0.9827406370639801,0.021948411615917433,31.175333293279014,18.019629268682312,3.8615792202155004,2.266734235588232,3.834497800808173,2.259828297195092,-0.004611149926980335,0.0433981312652511
M4,text,300,0.0,0.0,0.9999777108430863,4.937972466967483e-06,15.891133365631104,8.431963588888667,3.83314954836363,2.345445678904365,3.8348252384682473,2.3463801144674252,0.00020046850045522113,0.0014442569299196843
M4,audio,300,0.0,0.0,0.9999900078773498,3.4153207662384735e-06,618.1359326680501,369.4452964156476,3.867822667326802,2.309874647070878,3.868319556290074,2.3104860442974764,4.607309897740703e-05,0.0006428369960135046
M4,vision,300,0.0,0.0,0.9999960112571716,4.416330070963206e-06,32.02326666514079,18.480166711480305,3.8529649544991638,2.3080516925693866,3.8542076376005214,2.3106234287813843,0.00010548800230026034,0.001267114074095688
1 method modality n mvr_mean mvr_std normalized_entropy_mean normalized_entropy_std width80_source_positions_mean width80_source_positions_std expected_time_start_s_mean expected_time_start_s_std expected_time_end_s_mean expected_time_end_s_std trajectory_span_fraction_mean trajectory_span_fraction_std
2 M1 text 100 0.0 0.0 0.0008249471336603165 0.005233484169432736 1.004799998998642 0.0312849358458409 0.373650000790672 0.28760723340035366 7.3229000100512325 4.567867965796829 0.8612861529795919 0.07875035859337948
3 M1 audio 100 0.0 0.0 0.35824004098773005 0.05432258900057586 12.056200053691864 7.307551875172266 0.3177500062027157 0.2794568898376313 7.441361117547106 4.533387053269832 0.8937758523714728 0.06338502509117576
4 M1 vision 100 0.0 0.0 0.02091189981205389 0.03423249778642653 1.144999988079071 0.27983220085257804 0.30618889875375566 0.2956530091396046 7.43663269007148 4.535796288658362 0.8956127409730107 0.0656268624353691
5 M2 text 100 0.0 0.0 0.003971234038472176 0.010095580232451877 1.0215999960899353 0.060615033764357146 0.3753062509072731 0.2858492171110105 7.3229000100512325 4.567867965796829 0.8611038638930768 0.07866203063075565
6 M2 audio 100 0.0 0.0 0.39199826538562776 0.04856653297051989 12.777399990558624 7.437441491467199 0.08135000219190694 0.046356409936463026 7.716148159798975 4.594836781525238 0.968455815287307 0.009448427161791137
7 M2 vision 100 0.0 0.0 0.015714890849776567 0.04120196183036711 1.108399999141693 0.3072087474461631 0.024000000567226686 0.0514830577428781 7.7077327980137875 4.617512750249008 0.9744537002407014 0.016307502074103464
8 M3 text 300 0.0 0.0 0.0004981429129838944 0.004964740878146852 1.0029999995231629 0.02989949245257891 0.37365000079067195 0.28664372237282515 7.320650003693404 4.545633324884766 0.8611844960658346 0.07838045740132776
9 M3 audio 300 0.01857142824679613 0.028241165293696132 0.992265993754069 0.00445452263162519 607.8159977213542 365.5015286317014 3.883805515007902 2.328319755482711 3.8684255829138965 2.2993181536785716 -0.001562372346719104 0.016018112979249148
10 M3 vision 300 0.06843537279715141 0.06722724739800441 0.9827406370639801 0.021948411615917433 31.175333293279014 18.019629268682312 3.8615792202155004 2.266734235588232 3.834497800808173 2.259828297195092 -0.004611149926980335 0.0433981312652511
11 M4 text 300 0.0 0.0 0.9999777108430863 4.937972466967483e-06 15.891133365631104 8.431963588888667 3.83314954836363 2.345445678904365 3.8348252384682473 2.3463801144674252 0.00020046850045522113 0.0014442569299196843
12 M4 audio 300 0.0 0.0 0.9999900078773498 3.4153207662384735e-06 618.1359326680501 369.4452964156476 3.867822667326802 2.309874647070878 3.868319556290074 2.3104860442974764 4.607309897740703e-05 0.0006428369960135046
13 M4 vision 300 0.0 0.0 0.9999960112571716 4.416330070963206e-06 32.02326666514079 18.480166711480305 3.8529649544991638 2.3080516925693866 3.8542076376005214 2.3106234287813843 0.00010548800230026034 0.001267114074095688
@@ -0,0 +1,5 @@
method,mvr_text,entropy_text,trajectory_span_text,mvr_audio,entropy_audio,trajectory_span_audio,mvr_vision,entropy_vision,trajectory_span_vision,r_at_1_text_to_audio,r_at_5_text_to_audio,r_at_1_text_to_vision,r_at_5_text_to_vision,reconstruction_mae_text,reconstruction_mae_audio,reconstruction_mae_vision,emotion_accuracy,emotion_macro_f1,emotion_mae,emotion_pearson
M1,0.0,0.0008249471336603165,0.8612861529795919,0.0,0.35824004098773005,0.8937758523714728,0.0,0.02091189981205389,0.8956127409730107,0.0012000000569969416,0.007000000309199094,0.0020000000949949026,0.007000000402331352,0.8308010935783386,0.6908806562423706,0.9209612965583801,0.5900000000000001,0.43700352733686065,0.5521364179253578,0.5305163430956757
M2,0.0,0.003971234038472176,0.8611038638930768,0.0,0.39199826538562776,0.968455815287307,0.0,0.015714890849776567,0.9744537002407014,0.0014000000432133675,0.009200000390410423,0.003600000194273889,0.008800000487826764,0.8190697312355042,0.6621044635772705,0.9195392370223999,0.55,0.3434301254301254,0.5691992919147014,0.5082439027512137
M3,0.0,0.0004981429129838944,0.8611844960658346,0.01857142824679613,0.992265993754069,-0.001562372346719104,0.06843537279715141,0.9827406370639801,-0.004611149926980335,0.01966666733690848,0.043000001646578315,0.0062666668944681685,0.013800000709791978,0.8311106324195862,0.38449583053588865,0.868217929204305,0.5766666666666667,0.43789820467598245,0.5534388141334057,0.5457241196963684
M4,0.0,0.9999777108430863,0.00020046850045522113,0.0,0.9999900078773498,4.607309897740703e-05,0.0,0.9999960112571716,0.00010548800230026034,0.007066666963510215,0.01226666709408164,0.0016000000604738792,0.00760000046963493,0.4924671709537506,0.37839911778767904,0.8630666335423788,0.56,0.41446068467573843,0.6231901267170906,0.5067829543507287
1 method mvr_text entropy_text trajectory_span_text mvr_audio entropy_audio trajectory_span_audio mvr_vision entropy_vision trajectory_span_vision r_at_1_text_to_audio r_at_5_text_to_audio r_at_1_text_to_vision r_at_5_text_to_vision reconstruction_mae_text reconstruction_mae_audio reconstruction_mae_vision emotion_accuracy emotion_macro_f1 emotion_mae emotion_pearson
2 M1 0.0 0.0008249471336603165 0.8612861529795919 0.0 0.35824004098773005 0.8937758523714728 0.0 0.02091189981205389 0.8956127409730107 0.0012000000569969416 0.007000000309199094 0.0020000000949949026 0.007000000402331352 0.8308010935783386 0.6908806562423706 0.9209612965583801 0.5900000000000001 0.43700352733686065 0.5521364179253578 0.5305163430956757
3 M2 0.0 0.003971234038472176 0.8611038638930768 0.0 0.39199826538562776 0.968455815287307 0.0 0.015714890849776567 0.9744537002407014 0.0014000000432133675 0.009200000390410423 0.003600000194273889 0.008800000487826764 0.8190697312355042 0.6621044635772705 0.9195392370223999 0.55 0.3434301254301254 0.5691992919147014 0.5082439027512137
4 M3 0.0 0.0004981429129838944 0.8611844960658346 0.01857142824679613 0.992265993754069 -0.001562372346719104 0.06843537279715141 0.9827406370639801 -0.004611149926980335 0.01966666733690848 0.043000001646578315 0.0062666668944681685 0.013800000709791978 0.8311106324195862 0.38449583053588865 0.868217929204305 0.5766666666666667 0.43789820467598245 0.5534388141334057 0.5457241196963684
5 M4 0.0 0.9999777108430863 0.00020046850045522113 0.0 0.9999900078773498 4.607309897740703e-05 0.0 0.9999960112571716 0.00010548800230026034 0.007066666963510215 0.01226666709408164 0.0016000000604738792 0.00760000046963493 0.4924671709537506 0.37839911778767904 0.8630666335423788 0.56 0.41446068467573843 0.6231901267170906 0.5067829543507287
@@ -0,0 +1,5 @@
method,n,accuracy_mean,accuracy_std,macro_f1_mean,macro_f1_std,mae_mean,mae_std,pearson_mean,pearson_std
M1,5,0.5900000000000001,0.1673320053068151,0.43700352733686065,0.22785227549938045,0.5521364179253578,0.12144893660249381,0.5305163430956757,0.09576372130151391
M2,5,0.55,0.1767766952966369,0.3434301254301254,0.14084701638388486,0.5691992919147014,0.11997495602498047,0.5082439027512137,0.09789171743073
M3,15,0.5766666666666667,0.18503538915054937,0.43789820467598245,0.21990438278507196,0.5534388141334057,0.11230839313313754,0.5457241196963684,0.09265676731571848
M4,15,0.56,0.15142183273415838,0.41446068467573843,0.13578816810911878,0.6231901267170906,0.13326862372482587,0.5067829543507287,0.1021700904550588
1 method n accuracy_mean accuracy_std macro_f1_mean macro_f1_std mae_mean mae_std pearson_mean pearson_std
2 M1 5 0.5900000000000001 0.1673320053068151 0.43700352733686065 0.22785227549938045 0.5521364179253578 0.12144893660249381 0.5305163430956757 0.09576372130151391
3 M2 5 0.55 0.1767766952966369 0.3434301254301254 0.14084701638388486 0.5691992919147014 0.11997495602498047 0.5082439027512137 0.09789171743073
4 M3 15 0.5766666666666667 0.18503538915054937 0.43789820467598245 0.21990438278507196 0.5534388141334057 0.11230839313313754 0.5457241196963684 0.09265676731571848
5 M4 15 0.56 0.15142183273415838 0.41446068467573843 0.13578816810911878 0.6231901267170906 0.13326862372482587 0.5067829543507287 0.1021700904550588
@@ -0,0 +1,129 @@
[start] samples=100 groups=37 folds=5 seeds=[42, 3407, 2026] device=cuda
[fold 1/5] train=80 validation=20 train_video_ids=31 validation_video_ids=6
[fold 1] evaluate M1 and train identical probes
[fold 1] evaluate M2 and train identical probes
[fold 1] train M3, seed=42
In file included from /usr/include/python3.14/pyconfig.h:6,
from /usr/include/python3.14/Python.h:14,
from /home/gloamxun/modeling_zhaocui/Q1/.venv/lib/python3.14/site-packages/triton/backends/nvidia/driver.c:9:
/usr/include/python3.14/pyconfig-64.h:2031:9: warning: ‘_POSIX_C_SOURCE’ redefined
2031 | #define _POSIX_C_SOURCE 200809L
| ^~~~~~~~~~~~~~~
In file included from /usr/include/bits/libc-header-start.h:33,
from /usr/include/stdlib.h:26,
from /home/gloamxun/modeling_zhaocui/Q1/.venv/lib/python3.14/site-packages/triton/backends/nvidia/include/cuda.h:56,
from /home/gloamxun/modeling_zhaocui/Q1/.venv/lib/python3.14/site-packages/triton/backends/nvidia/driver.c:1:
/usr/include/features.h:319:10: note: this is the location of the previous definition
319 | # define _POSIX_C_SOURCE 202405L
| ^~~~~~~~~~~~~~~
[fold 1] M3, seed=42 best_epoch=50 val_objective=4.93766; running frozen probes
[fold 1] train M4, seed=42
[fold 1] M4, seed=42 best_epoch=4 val_objective=6.07687; running frozen probes
[fold 1] train M3, seed=3407
[fold 1] M3, seed=3407 best_epoch=50 val_objective=4.98259; running frozen probes
[fold 1] train M4, seed=3407
[fold 1] M4, seed=3407 best_epoch=4 val_objective=5.87185; running frozen probes
[fold 1] train M3, seed=2026
[fold 1] M3, seed=2026 best_epoch=50 val_objective=5.10248; running frozen probes
[fold 1] train M4, seed=2026
[fold 1] M4, seed=2026 best_epoch=3 val_objective=6.09901; running frozen probes
[fold 2/5] train=80 validation=20 train_video_ids=30 validation_video_ids=7
[fold 2] evaluate M1 and train identical probes
[fold 2] evaluate M2 and train identical probes
[fold 2] train M3, seed=42
[fold 2] M3, seed=42 best_epoch=1 val_objective=6.12588; running frozen probes
[fold 2] train M4, seed=42
[fold 2] M4, seed=42 best_epoch=1 val_objective=5.94268; running frozen probes
[fold 2] train M3, seed=3407
[fold 2] M3, seed=3407 best_epoch=50 val_objective=5.14454; running frozen probes
[fold 2] train M4, seed=3407
[fold 2] M4, seed=3407 best_epoch=3 val_objective=6.06774; running frozen probes
[fold 2] train M3, seed=2026
[fold 2] M3, seed=2026 best_epoch=50 val_objective=5.17416; running frozen probes
[fold 2] train M4, seed=2026
[fold 2] M4, seed=2026 best_epoch=1 val_objective=5.96431; running frozen probes
[fold 3/5] train=80 validation=20 train_video_ids=29 validation_video_ids=8
[fold 3] evaluate M1 and train identical probes
[fold 3] evaluate M2 and train identical probes
[fold 3] train M3, seed=42
[fold 3] M3, seed=42 best_epoch=50 val_objective=4.97643; running frozen probes
[fold 3] train M4, seed=42
[fold 3] M4, seed=42 best_epoch=3 val_objective=6.19582; running frozen probes
[fold 3] train M3, seed=3407
[fold 3] M3, seed=3407 best_epoch=50 val_objective=4.65949; running frozen probes
[fold 3] train M4, seed=3407
[fold 3] M4, seed=3407 best_epoch=9 val_objective=5.91066; running frozen probes
[fold 3] train M3, seed=2026
[fold 3] M3, seed=2026 best_epoch=50 val_objective=4.99993; running frozen probes
[fold 3] train M4, seed=2026
[fold 3] M4, seed=2026 best_epoch=3 val_objective=6.02409; running frozen probes
[fold 4/5] train=80 validation=20 train_video_ids=29 validation_video_ids=8
[fold 4] evaluate M1 and train identical probes
[fold 4] evaluate M2 and train identical probes
[fold 4] train M3, seed=42
[fold 4] M3, seed=42 best_epoch=50 val_objective=5.15230; running frozen probes
[fold 4] train M4, seed=42
[fold 4] M4, seed=42 best_epoch=1 val_objective=5.99857; running frozen probes
[fold 4] train M3, seed=3407
[fold 4] M3, seed=3407 best_epoch=50 val_objective=5.09800; running frozen probes
[fold 4] train M4, seed=3407
[fold 4] M4, seed=3407 best_epoch=2 val_objective=6.03456; running frozen probes
[fold 4] train M3, seed=2026
[fold 4] M3, seed=2026 best_epoch=50 val_objective=5.26026; running frozen probes
[fold 4] train M4, seed=2026
[fold 4] M4, seed=2026 best_epoch=2 val_objective=6.08121; running frozen probes
[fold 5/5] train=80 validation=20 train_video_ids=29 validation_video_ids=8
[fold 5] evaluate M1 and train identical probes
[fold 5] evaluate M2 and train identical probes
[fold 5] train M3, seed=42
[fold 5] M3, seed=42 best_epoch=50 val_objective=5.29874; running frozen probes
[fold 5] train M4, seed=42
[fold 5] M4, seed=42 best_epoch=4 val_objective=6.02189; running frozen probes
[fold 5] train M3, seed=3407
[fold 5] M3, seed=3407 best_epoch=50 val_objective=4.95355; running frozen probes
[fold 5] train M4, seed=3407
[fold 5] M4, seed=3407 best_epoch=8 val_objective=6.05872; running frozen probes
[fold 5] train M3, seed=2026
[fold 5] M3, seed=2026 best_epoch=50 val_objective=5.13369; running frozen probes
[fold 5] train M4, seed=2026
[fold 5] M4, seed=2026 best_epoch=1 val_objective=5.88269; running frozen probes
[done] elapsed_seconds=516.3 output=/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison
{
"created_utc": "2026-09-23T09:07:49.626404+00:00",
"sample_count": 100,
"group_count": 37,
"folds": 5,
"seeds": [
42,
3407,
2026
],
"device": "cuda",
"gpu_name": "NVIDIA GeForce RTX 5070 Ti",
"python": "3.14.7",
"torch": "2.14.0+cu130",
"parameters": {
"grid_size": 50,
"hidden_size": 128,
"heads": 4,
"dropout": 0.1,
"batch_size": 8,
"epochs_max": 50,
"early_stopping_patience": 8,
"learning_rate": 0.0001,
"mask_ratio": 0.2,
"retrieval_probe_epochs": 20,
"reconstruction_probe_epochs": 25,
"mvr_epsilon": 0.02
},
"objective": "masked reconstruction + cross-modal contrastive + temporal monotonicity; emotion labels unused",
"split_rule": "GroupKFold by group_id/video_id",
"example_sample_id": "-tPCytz4rww/12",
"elapsed_seconds": 516.2944533824921,
"interpretation_limits": [
"Grid-index retrieval is a representation-consistency probe, not independent temporal ground truth.",
"Masked reconstruction uses a decoder trained on the training fold and reports standardized-feature errors.",
"The emotion probe is a small-sample downstream utility check, not a claim of generalization to MOSEI.",
"No human event timestamps are available, so human IoU/MATE is not reported."
]
}
@@ -0,0 +1,41 @@
method,seed,fold,accuracy,macro_f1,mae,pearson,n_train,n_validation
M1,fixed,1,0.8,0.736111111111111,0.4612815961241722,0.3908651518384776,80,20
M2,fixed,1,0.75,0.393939393939394,0.4419183313846588,0.4236518393081548,80,20
M3,42,1,0.8,0.736111111111111,0.4484931230545044,0.41279811854541476,80,20
M4,42,1,0.8,0.6344086021505376,0.4766388975083828,0.43187935519491855,80,20
M3,3407,1,0.8,0.736111111111111,0.44865339174866675,0.4171944975077341,80,20
M4,3407,1,0.8,0.6344086021505376,0.47647997066378595,0.4320761354143285,80,20
M3,2026,1,0.8,0.736111111111111,0.4473679281771183,0.419287895205426,80,20
M4,2026,1,0.8,0.6344086021505376,0.47662248760461806,0.43167719260246934,80,20
M1,fixed,2,0.6,0.44523809523809527,0.7316692218184471,0.6480682206513965,80,20
M2,fixed,2,0.6,0.44523809523809527,0.7364562056958676,0.6535275734065676,80,20
M3,42,2,0.7,0.5598290598290597,0.7191767528653145,0.6829254472780363,80,20
M4,42,2,0.6,0.48939393939393944,0.8361533507704735,0.5952034035922986,80,20
M3,3407,2,0.65,0.4764957264957264,0.7146183483302593,0.6888039663406401,80,20
M4,3407,2,0.6,0.48939393939393944,0.8360007748007774,0.595177745117279,80,20
M3,2026,2,0.65,0.47252747252747246,0.7159126169979573,0.689385710098449,80,20
M4,2026,2,0.6,0.48939393939393944,0.8361966282129287,0.5951825310175703,80,20
M1,fixed,3,0.7,0.5731922398589065,0.6056819573044777,0.5391663374393444,80,20
M2,fixed,3,0.65,0.48412698412698413,0.6305533096194267,0.4961935349272063,80,20
M3,42,3,0.65,0.5421245421245421,0.6378875352442265,0.5325578720674686,80,20
M4,42,3,0.55,0.34215167548500885,0.6873359516263008,0.5315284769823068,80,20
M3,3407,3,0.65,0.5421245421245421,0.626622511446476,0.5478317984415519,80,20
M4,3407,3,0.55,0.34215167548500885,0.6872267439961434,0.5313424475222814,80,20
M3,2026,3,0.7,0.5731922398589065,0.6307285942137242,0.5464569461654115,80,20
M4,2026,3,0.55,0.34215167548500885,0.6871870808303356,0.5319799133727069,80,20
M1,fixed,4,0.4,0.19047619047619047,0.5336167700588703,0.5772197323434819,80,20
M2,fixed,4,0.3,0.15384615384615385,0.5659813567996025,0.5500284823994642,80,20
M3,42,4,0.3,0.16,0.5302071489393712,0.5725429629109834,80,20
M4,42,4,0.35,0.2711111111111111,0.5166212148964405,0.6177149095726567,80,20
M3,3407,4,0.3,0.15384615384615385,0.5506807617843151,0.5860130646535472,80,20
M4,3407,4,0.35,0.2711111111111111,0.5164451241493225,0.6181052902824066,80,20
M3,2026,4,0.3,0.16,0.531651595979929,0.5564093983759777,80,20
M4,2026,4,0.35,0.2711111111111111,0.5163851425051689,0.617812107663041,80,20
M1,fixed,5,0.45,0.24,0.4284325443208218,0.49726227320567845,80,20
M2,fixed,5,0.45,0.24,0.47108725607395174,0.4178180837146751,80,20
M3,42,5,0.45,0.24,0.44473609402775766,0.4783453741103445,80,20
M4,42,5,0.5,0.3352380952380953,0.5995716892182827,0.35723746565175724,80,20
M3,3407,5,0.45,0.24,0.4421664297580719,0.4940687599587745,80,20
M4,3407,5,0.5,0.3352380952380953,0.5993113562464714,0.35800549475908866,80,20
M3,2026,5,0.45,0.24,0.4126793794333935,0.5612399837857652,80,20
M4,2026,5,0.5,0.3352380952380953,0.5996754877269268,0.3568218465158214,80,20
1 method seed fold accuracy macro_f1 mae pearson n_train n_validation
2 M1 fixed 1 0.8 0.736111111111111 0.4612815961241722 0.3908651518384776 80 20
3 M2 fixed 1 0.75 0.393939393939394 0.4419183313846588 0.4236518393081548 80 20
4 M3 42 1 0.8 0.736111111111111 0.4484931230545044 0.41279811854541476 80 20
5 M4 42 1 0.8 0.6344086021505376 0.4766388975083828 0.43187935519491855 80 20
6 M3 3407 1 0.8 0.736111111111111 0.44865339174866675 0.4171944975077341 80 20
7 M4 3407 1 0.8 0.6344086021505376 0.47647997066378595 0.4320761354143285 80 20
8 M3 2026 1 0.8 0.736111111111111 0.4473679281771183 0.419287895205426 80 20
9 M4 2026 1 0.8 0.6344086021505376 0.47662248760461806 0.43167719260246934 80 20
10 M1 fixed 2 0.6 0.44523809523809527 0.7316692218184471 0.6480682206513965 80 20
11 M2 fixed 2 0.6 0.44523809523809527 0.7364562056958676 0.6535275734065676 80 20
12 M3 42 2 0.7 0.5598290598290597 0.7191767528653145 0.6829254472780363 80 20
13 M4 42 2 0.6 0.48939393939393944 0.8361533507704735 0.5952034035922986 80 20
14 M3 3407 2 0.65 0.4764957264957264 0.7146183483302593 0.6888039663406401 80 20
15 M4 3407 2 0.6 0.48939393939393944 0.8360007748007774 0.595177745117279 80 20
16 M3 2026 2 0.65 0.47252747252747246 0.7159126169979573 0.689385710098449 80 20
17 M4 2026 2 0.6 0.48939393939393944 0.8361966282129287 0.5951825310175703 80 20
18 M1 fixed 3 0.7 0.5731922398589065 0.6056819573044777 0.5391663374393444 80 20
19 M2 fixed 3 0.65 0.48412698412698413 0.6305533096194267 0.4961935349272063 80 20
20 M3 42 3 0.65 0.5421245421245421 0.6378875352442265 0.5325578720674686 80 20
21 M4 42 3 0.55 0.34215167548500885 0.6873359516263008 0.5315284769823068 80 20
22 M3 3407 3 0.65 0.5421245421245421 0.626622511446476 0.5478317984415519 80 20
23 M4 3407 3 0.55 0.34215167548500885 0.6872267439961434 0.5313424475222814 80 20
24 M3 2026 3 0.7 0.5731922398589065 0.6307285942137242 0.5464569461654115 80 20
25 M4 2026 3 0.55 0.34215167548500885 0.6871870808303356 0.5319799133727069 80 20
26 M1 fixed 4 0.4 0.19047619047619047 0.5336167700588703 0.5772197323434819 80 20
27 M2 fixed 4 0.3 0.15384615384615385 0.5659813567996025 0.5500284823994642 80 20
28 M3 42 4 0.3 0.16 0.5302071489393712 0.5725429629109834 80 20
29 M4 42 4 0.35 0.2711111111111111 0.5166212148964405 0.6177149095726567 80 20
30 M3 3407 4 0.3 0.15384615384615385 0.5506807617843151 0.5860130646535472 80 20
31 M4 3407 4 0.35 0.2711111111111111 0.5164451241493225 0.6181052902824066 80 20
32 M3 2026 4 0.3 0.16 0.531651595979929 0.5564093983759777 80 20
33 M4 2026 4 0.35 0.2711111111111111 0.5163851425051689 0.617812107663041 80 20
34 M1 fixed 5 0.45 0.24 0.4284325443208218 0.49726227320567845 80 20
35 M2 fixed 5 0.45 0.24 0.47108725607395174 0.4178180837146751 80 20
36 M3 42 5 0.45 0.24 0.44473609402775766 0.4783453741103445 80 20
37 M4 42 5 0.5 0.3352380952380953 0.5995716892182827 0.35723746565175724 80 20
38 M3 3407 5 0.45 0.24 0.4421664297580719 0.4940687599587745 80 20
39 M4 3407 5 0.5 0.3352380952380953 0.5993113562464714 0.35800549475908866 80 20
40 M3 2026 5 0.45 0.24 0.4126793794333935 0.5612399837857652 80 20
41 M4 2026 5 0.5 0.3352380952380953 0.5996754877269268 0.3568218465158214 80 20
@@ -0,0 +1,121 @@
method,seed,fold,target_modality,mask_ratio,mae_standardized,smooth_l1_standardized,masked_values
M1,fixed,1,text,0.2,0.7864587306976318,0.41599002480506897,153600
M1,fixed,1,audio,0.2,0.6423318386077881,0.2985542118549347,5000
M1,fixed,1,vision,0.2,0.8732286691665649,0.49052613973617554,38400
M2,fixed,1,text,0.2,0.7657629251480103,0.3995133638381958,153600
M2,fixed,1,audio,0.2,0.6078358292579651,0.27118614315986633,5000
M2,fixed,1,vision,0.2,0.8754072189331055,0.49298685789108276,38400
M3,42,1,text,0.2,0.7869179248809814,0.4176875948905945,153600
M3,42,1,audio,0.2,0.3639732003211975,0.1077285036444664,5000
M3,42,1,vision,0.2,0.8297806978225708,0.4545438885688782,38400
M4,42,1,text,0.2,0.4791339337825775,0.17556655406951904,153600
M4,42,1,audio,0.2,0.31638050079345703,0.07683445513248444,5000
M4,42,1,vision,0.2,0.8184546828269958,0.44288504123687744,38400
M3,3407,1,text,0.2,0.7894418239593506,0.4194052815437317,153600
M3,3407,1,audio,0.2,0.3394777178764343,0.09327597916126251,5000
M3,3407,1,vision,0.2,0.8291587233543396,0.4534059762954712,38400
M4,3407,1,text,0.2,0.4733425974845886,0.17110656201839447,153600
M4,3407,1,audio,0.2,0.33666083216667175,0.08810637891292572,5000
M4,3407,1,vision,0.2,0.817406415939331,0.44359228014945984,38400
M3,2026,1,text,0.2,0.7847438454627991,0.41536271572113037,153600
M3,2026,1,audio,0.2,0.35511481761932373,0.10340271890163422,5000
M3,2026,1,vision,0.2,0.8283053636550903,0.45340079069137573,38400
M4,2026,1,text,0.2,0.47494950890541077,0.1725955754518509,153600
M4,2026,1,audio,0.2,0.32631760835647583,0.08271428942680359,5000
M4,2026,1,vision,0.2,0.8221946954727173,0.4466344714164734,38400
M1,fixed,2,text,0.2,0.8431306481361389,0.46495506167411804,153600
M1,fixed,2,audio,0.2,0.6558716893196106,0.31570741534233093,5000
M1,fixed,2,vision,0.2,0.8846356272697449,0.5022978782653809,38400
M2,fixed,2,text,0.2,0.8220721483230591,0.4483771324157715,153600
M2,fixed,2,audio,0.2,0.6195802688598633,0.28658542037010193,5000
M2,fixed,2,vision,0.2,0.8831602931022644,0.5007410049438477,38400
M3,42,2,text,0.2,0.8472743034362793,0.4686535894870758,153600
M3,42,2,audio,0.2,0.34122517704963684,0.11146467179059982,5000
M3,42,2,vision,0.2,0.8413732051849365,0.4642646312713623,38400
M4,42,2,text,0.2,0.5017852187156677,0.19247683882713318,153600
M4,42,2,audio,0.2,0.34099525213241577,0.1067037358880043,5000
M4,42,2,vision,0.2,0.8211991190910339,0.4461081624031067,38400
M3,3407,2,text,0.2,0.8380216360092163,0.4611693322658539,153600
M3,3407,2,audio,0.2,0.3393073081970215,0.10981127619743347,5000
M3,3407,2,vision,0.2,0.8233676552772522,0.44843095541000366,38400
M4,3407,2,text,0.2,0.49964815378189087,0.1912464201450348,153600
M4,3407,2,audio,0.2,0.3402937948703766,0.11066138744354248,5000
M4,3407,2,vision,0.2,0.8373452425003052,0.46078431606292725,38400
M3,2026,2,text,0.2,0.8452751636505127,0.4673306345939636,153600
M3,2026,2,audio,0.2,0.3425142467021942,0.1117163822054863,5000
M3,2026,2,vision,0.2,0.8416237831115723,0.4639897644519806,38400
M4,2026,2,text,0.2,0.5024572610855103,0.19227026402950287,153600
M4,2026,2,audio,0.2,0.3313565254211426,0.10437174141407013,5000
M4,2026,2,vision,0.2,0.8250595331192017,0.45077505707740784,38400
M1,fixed,3,text,0.2,0.8600677847862244,0.48109835386276245,153600
M1,fixed,3,audio,0.2,0.6301218867301941,0.28287407755851746,5000
M1,fixed,3,vision,0.2,0.9313846826553345,0.543266236782074,38400
M2,fixed,3,text,0.2,0.8661688566207886,0.4862343370914459,153600
M2,fixed,3,audio,0.2,0.6363305449485779,0.295511931180954,5000
M2,fixed,3,vision,0.2,0.9410251379013062,0.5511942505836487,38400
M3,42,3,text,0.2,0.8575569987297058,0.4784933626651764,153600
M3,42,3,audio,0.2,0.35879722237586975,0.10602780431509018,5000
M3,42,3,vision,0.2,0.848150908946991,0.4709389805793762,38400
M4,42,3,text,0.2,0.5080369114875793,0.1971929669380188,153600
M4,42,3,audio,0.2,0.35229456424713135,0.09507919102907181,5000
M4,42,3,vision,0.2,0.8596658706665039,0.4819534420967102,38400
M3,3407,3,text,0.2,0.8525339365005493,0.4742853045463562,153600
M3,3407,3,audio,0.2,0.3356912434101105,0.09289933741092682,5000
M3,3407,3,vision,0.2,0.852375328540802,0.4742961823940277,38400
M4,3407,3,text,0.2,0.5013428926467896,0.1922634392976761,153600
M4,3407,3,audio,0.2,0.3563744127750397,0.09956291317939758,5000
M4,3407,3,vision,0.2,0.8604409098625183,0.48151978850364685,38400
M3,2026,3,text,0.2,0.8605049848556519,0.48118850588798523,153600
M3,2026,3,audio,0.2,0.36211082339286804,0.10407590121030807,5000
M3,2026,3,vision,0.2,0.859515905380249,0.4795020520687103,38400
M4,2026,3,text,0.2,0.5066819787025452,0.1963006854057312,153600
M4,2026,3,audio,0.2,0.3566948175430298,0.09866075217723846,5000
M4,2026,3,vision,0.2,0.8549585342407227,0.4780130684375763,38400
M1,fixed,4,text,0.2,0.8199025392532349,0.44484803080558777,153600
M1,fixed,4,audio,0.2,0.7768001556396484,0.4197060465812683,5000
M1,fixed,4,vision,0.2,0.9551154971122742,0.5669697523117065,38400
M2,fixed,4,text,0.2,0.8191158175468445,0.4448056221008301,153600
M2,fixed,4,audio,0.2,0.7202740907669067,0.3684014081954956,5000
M2,fixed,4,vision,0.2,0.9408901929855347,0.5558907389640808,38400
M3,42,4,text,0.2,0.8254697918891907,0.4495771527290344,153600
M3,42,4,audio,0.2,0.417472243309021,0.15880697965621948,5000
M3,42,4,vision,0.2,0.9076664447784424,0.5263801217079163,38400
M4,42,4,text,0.2,0.4919026792049408,0.1856606900691986,153600
M4,42,4,audio,0.2,0.4052186906337738,0.15105251967906952,5000
M4,42,4,vision,0.2,0.8914499878883362,0.5127754807472229,38400
M3,3407,4,text,0.2,0.8279706835746765,0.4522389769554138,153600
M3,3407,4,audio,0.2,0.39376136660575867,0.14974132180213928,5000
M3,3407,4,vision,0.2,0.9084181785583496,0.5274486541748047,38400
M4,3407,4,text,0.2,0.4878115952014923,0.18323160707950592,153600
M4,3407,4,audio,0.2,0.40141838788986206,0.15048016607761383,5000
M4,3407,4,vision,0.2,0.8952069282531738,0.5167335867881775,38400
M3,2026,4,text,0.2,0.8241562843322754,0.4491640329360962,153600
M3,2026,4,audio,0.2,0.40933120250701904,0.1570354402065277,5000
M3,2026,4,vision,0.2,0.9041170477867126,0.5228397846221924,38400
M4,2026,4,text,0.2,0.49553707242012024,0.18893885612487793,153600
M4,2026,4,audio,0.2,0.39918839931488037,0.14919213950634003,5000
M4,2026,4,vision,0.2,0.8958951830863953,0.5186710953712463,38400
M1,fixed,5,text,0.2,0.8444457650184631,0.4662260413169861,153600
M1,fixed,5,audio,0.2,0.7492777109146118,0.41332030296325684,5000
M1,fixed,5,vision,0.2,0.9604420065879822,0.5708574652671814,38400
M2,fixed,5,text,0.2,0.8222289085388184,0.44781213998794556,153600
M2,fixed,5,audio,0.2,0.7265015840530396,0.3929823040962219,5000
M2,fixed,5,vision,0.2,0.9572133421897888,0.5693861842155457,38400
M3,42,5,text,0.2,0.847065269947052,0.4688800573348999,153600
M3,42,5,audio,0.2,0.47439461946487427,0.2379949688911438,5000
M3,42,5,vision,0.2,0.91214519739151,0.527710497379303,38400
M4,42,5,text,0.2,0.49351584911346436,0.18649226427078247,153600
M4,42,5,audio,0.2,0.46833816170692444,0.23857423663139343,5000
M4,42,5,vision,0.2,0.9112464785575867,0.5267350673675537,38400
M3,3407,5,text,0.2,0.8362759947776794,0.4592466354370117,153600
M3,3407,5,audio,0.2,0.4684489071369171,0.23598544299602509,5000
M3,3407,5,vision,0.2,0.9186314344406128,0.5335334539413452,38400
M4,3407,5,text,0.2,0.48672324419021606,0.18171463906764984,153600
M4,3407,5,audio,0.2,0.4682857394218445,0.23653720319271088,5000
M4,3407,5,vision,0.2,0.9220170378684998,0.5346263647079468,38400
M3,2026,5,text,0.2,0.8434508442878723,0.46527090668678284,153600
M3,2026,5,audio,0.2,0.4658173620700836,0.23616507649421692,5000
M3,2026,5,vision,0.2,0.918639063835144,0.5329208374023438,38400
M4,2026,5,text,0.2,0.4841386675834656,0.18039487302303314,153600
M4,2026,5,audio,0.2,0.47616907954216003,0.2406836897134781,5000
M4,2026,5,vision,0.2,0.9134588837623596,0.5281268358230591,38400
1 method seed fold target_modality mask_ratio mae_standardized smooth_l1_standardized masked_values
2 M1 fixed 1 text 0.2 0.7864587306976318 0.41599002480506897 153600
3 M1 fixed 1 audio 0.2 0.6423318386077881 0.2985542118549347 5000
4 M1 fixed 1 vision 0.2 0.8732286691665649 0.49052613973617554 38400
5 M2 fixed 1 text 0.2 0.7657629251480103 0.3995133638381958 153600
6 M2 fixed 1 audio 0.2 0.6078358292579651 0.27118614315986633 5000
7 M2 fixed 1 vision 0.2 0.8754072189331055 0.49298685789108276 38400
8 M3 42 1 text 0.2 0.7869179248809814 0.4176875948905945 153600
9 M3 42 1 audio 0.2 0.3639732003211975 0.1077285036444664 5000
10 M3 42 1 vision 0.2 0.8297806978225708 0.4545438885688782 38400
11 M4 42 1 text 0.2 0.4791339337825775 0.17556655406951904 153600
12 M4 42 1 audio 0.2 0.31638050079345703 0.07683445513248444 5000
13 M4 42 1 vision 0.2 0.8184546828269958 0.44288504123687744 38400
14 M3 3407 1 text 0.2 0.7894418239593506 0.4194052815437317 153600
15 M3 3407 1 audio 0.2 0.3394777178764343 0.09327597916126251 5000
16 M3 3407 1 vision 0.2 0.8291587233543396 0.4534059762954712 38400
17 M4 3407 1 text 0.2 0.4733425974845886 0.17110656201839447 153600
18 M4 3407 1 audio 0.2 0.33666083216667175 0.08810637891292572 5000
19 M4 3407 1 vision 0.2 0.817406415939331 0.44359228014945984 38400
20 M3 2026 1 text 0.2 0.7847438454627991 0.41536271572113037 153600
21 M3 2026 1 audio 0.2 0.35511481761932373 0.10340271890163422 5000
22 M3 2026 1 vision 0.2 0.8283053636550903 0.45340079069137573 38400
23 M4 2026 1 text 0.2 0.47494950890541077 0.1725955754518509 153600
24 M4 2026 1 audio 0.2 0.32631760835647583 0.08271428942680359 5000
25 M4 2026 1 vision 0.2 0.8221946954727173 0.4466344714164734 38400
26 M1 fixed 2 text 0.2 0.8431306481361389 0.46495506167411804 153600
27 M1 fixed 2 audio 0.2 0.6558716893196106 0.31570741534233093 5000
28 M1 fixed 2 vision 0.2 0.8846356272697449 0.5022978782653809 38400
29 M2 fixed 2 text 0.2 0.8220721483230591 0.4483771324157715 153600
30 M2 fixed 2 audio 0.2 0.6195802688598633 0.28658542037010193 5000
31 M2 fixed 2 vision 0.2 0.8831602931022644 0.5007410049438477 38400
32 M3 42 2 text 0.2 0.8472743034362793 0.4686535894870758 153600
33 M3 42 2 audio 0.2 0.34122517704963684 0.11146467179059982 5000
34 M3 42 2 vision 0.2 0.8413732051849365 0.4642646312713623 38400
35 M4 42 2 text 0.2 0.5017852187156677 0.19247683882713318 153600
36 M4 42 2 audio 0.2 0.34099525213241577 0.1067037358880043 5000
37 M4 42 2 vision 0.2 0.8211991190910339 0.4461081624031067 38400
38 M3 3407 2 text 0.2 0.8380216360092163 0.4611693322658539 153600
39 M3 3407 2 audio 0.2 0.3393073081970215 0.10981127619743347 5000
40 M3 3407 2 vision 0.2 0.8233676552772522 0.44843095541000366 38400
41 M4 3407 2 text 0.2 0.49964815378189087 0.1912464201450348 153600
42 M4 3407 2 audio 0.2 0.3402937948703766 0.11066138744354248 5000
43 M4 3407 2 vision 0.2 0.8373452425003052 0.46078431606292725 38400
44 M3 2026 2 text 0.2 0.8452751636505127 0.4673306345939636 153600
45 M3 2026 2 audio 0.2 0.3425142467021942 0.1117163822054863 5000
46 M3 2026 2 vision 0.2 0.8416237831115723 0.4639897644519806 38400
47 M4 2026 2 text 0.2 0.5024572610855103 0.19227026402950287 153600
48 M4 2026 2 audio 0.2 0.3313565254211426 0.10437174141407013 5000
49 M4 2026 2 vision 0.2 0.8250595331192017 0.45077505707740784 38400
50 M1 fixed 3 text 0.2 0.8600677847862244 0.48109835386276245 153600
51 M1 fixed 3 audio 0.2 0.6301218867301941 0.28287407755851746 5000
52 M1 fixed 3 vision 0.2 0.9313846826553345 0.543266236782074 38400
53 M2 fixed 3 text 0.2 0.8661688566207886 0.4862343370914459 153600
54 M2 fixed 3 audio 0.2 0.6363305449485779 0.295511931180954 5000
55 M2 fixed 3 vision 0.2 0.9410251379013062 0.5511942505836487 38400
56 M3 42 3 text 0.2 0.8575569987297058 0.4784933626651764 153600
57 M3 42 3 audio 0.2 0.35879722237586975 0.10602780431509018 5000
58 M3 42 3 vision 0.2 0.848150908946991 0.4709389805793762 38400
59 M4 42 3 text 0.2 0.5080369114875793 0.1971929669380188 153600
60 M4 42 3 audio 0.2 0.35229456424713135 0.09507919102907181 5000
61 M4 42 3 vision 0.2 0.8596658706665039 0.4819534420967102 38400
62 M3 3407 3 text 0.2 0.8525339365005493 0.4742853045463562 153600
63 M3 3407 3 audio 0.2 0.3356912434101105 0.09289933741092682 5000
64 M3 3407 3 vision 0.2 0.852375328540802 0.4742961823940277 38400
65 M4 3407 3 text 0.2 0.5013428926467896 0.1922634392976761 153600
66 M4 3407 3 audio 0.2 0.3563744127750397 0.09956291317939758 5000
67 M4 3407 3 vision 0.2 0.8604409098625183 0.48151978850364685 38400
68 M3 2026 3 text 0.2 0.8605049848556519 0.48118850588798523 153600
69 M3 2026 3 audio 0.2 0.36211082339286804 0.10407590121030807 5000
70 M3 2026 3 vision 0.2 0.859515905380249 0.4795020520687103 38400
71 M4 2026 3 text 0.2 0.5066819787025452 0.1963006854057312 153600
72 M4 2026 3 audio 0.2 0.3566948175430298 0.09866075217723846 5000
73 M4 2026 3 vision 0.2 0.8549585342407227 0.4780130684375763 38400
74 M1 fixed 4 text 0.2 0.8199025392532349 0.44484803080558777 153600
75 M1 fixed 4 audio 0.2 0.7768001556396484 0.4197060465812683 5000
76 M1 fixed 4 vision 0.2 0.9551154971122742 0.5669697523117065 38400
77 M2 fixed 4 text 0.2 0.8191158175468445 0.4448056221008301 153600
78 M2 fixed 4 audio 0.2 0.7202740907669067 0.3684014081954956 5000
79 M2 fixed 4 vision 0.2 0.9408901929855347 0.5558907389640808 38400
80 M3 42 4 text 0.2 0.8254697918891907 0.4495771527290344 153600
81 M3 42 4 audio 0.2 0.417472243309021 0.15880697965621948 5000
82 M3 42 4 vision 0.2 0.9076664447784424 0.5263801217079163 38400
83 M4 42 4 text 0.2 0.4919026792049408 0.1856606900691986 153600
84 M4 42 4 audio 0.2 0.4052186906337738 0.15105251967906952 5000
85 M4 42 4 vision 0.2 0.8914499878883362 0.5127754807472229 38400
86 M3 3407 4 text 0.2 0.8279706835746765 0.4522389769554138 153600
87 M3 3407 4 audio 0.2 0.39376136660575867 0.14974132180213928 5000
88 M3 3407 4 vision 0.2 0.9084181785583496 0.5274486541748047 38400
89 M4 3407 4 text 0.2 0.4878115952014923 0.18323160707950592 153600
90 M4 3407 4 audio 0.2 0.40141838788986206 0.15048016607761383 5000
91 M4 3407 4 vision 0.2 0.8952069282531738 0.5167335867881775 38400
92 M3 2026 4 text 0.2 0.8241562843322754 0.4491640329360962 153600
93 M3 2026 4 audio 0.2 0.40933120250701904 0.1570354402065277 5000
94 M3 2026 4 vision 0.2 0.9041170477867126 0.5228397846221924 38400
95 M4 2026 4 text 0.2 0.49553707242012024 0.18893885612487793 153600
96 M4 2026 4 audio 0.2 0.39918839931488037 0.14919213950634003 5000
97 M4 2026 4 vision 0.2 0.8958951830863953 0.5186710953712463 38400
98 M1 fixed 5 text 0.2 0.8444457650184631 0.4662260413169861 153600
99 M1 fixed 5 audio 0.2 0.7492777109146118 0.41332030296325684 5000
100 M1 fixed 5 vision 0.2 0.9604420065879822 0.5708574652671814 38400
101 M2 fixed 5 text 0.2 0.8222289085388184 0.44781213998794556 153600
102 M2 fixed 5 audio 0.2 0.7265015840530396 0.3929823040962219 5000
103 M2 fixed 5 vision 0.2 0.9572133421897888 0.5693861842155457 38400
104 M3 42 5 text 0.2 0.847065269947052 0.4688800573348999 153600
105 M3 42 5 audio 0.2 0.47439461946487427 0.2379949688911438 5000
106 M3 42 5 vision 0.2 0.91214519739151 0.527710497379303 38400
107 M4 42 5 text 0.2 0.49351584911346436 0.18649226427078247 153600
108 M4 42 5 audio 0.2 0.46833816170692444 0.23857423663139343 5000
109 M4 42 5 vision 0.2 0.9112464785575867 0.5267350673675537 38400
110 M3 3407 5 text 0.2 0.8362759947776794 0.4592466354370117 153600
111 M3 3407 5 audio 0.2 0.4684489071369171 0.23598544299602509 5000
112 M3 3407 5 vision 0.2 0.9186314344406128 0.5335334539413452 38400
113 M4 3407 5 text 0.2 0.48672324419021606 0.18171463906764984 153600
114 M4 3407 5 audio 0.2 0.4682857394218445 0.23653720319271088 5000
115 M4 3407 5 vision 0.2 0.9220170378684998 0.5346263647079468 38400
116 M3 2026 5 text 0.2 0.8434508442878723 0.46527090668678284 153600
117 M3 2026 5 audio 0.2 0.4658173620700836 0.23616507649421692 5000
118 M3 2026 5 vision 0.2 0.918639063835144 0.5329208374023438 38400
119 M4 2026 5 text 0.2 0.4841386675834656 0.18039487302303314 153600
120 M4 2026 5 audio 0.2 0.47616907954216003 0.2406836897134781 5000
121 M4 2026 5 vision 0.2 0.9134588837623596 0.5281268358230591 38400
@@ -0,0 +1,13 @@
method,target_modality,n,mae_standardized_mean,mae_standardized_std,smooth_l1_standardized_mean,smooth_l1_standardized_std
M1,text,5,0.8308010935783386,0.028634584948895196,0.4546235024929047,0.025151457325080215
M1,audio,5,0.6908806562423706,0.06720606353302543,0.34603241086006165,0.06541831705153776
M1,vision,5,0.9209612965583801,0.04009937912435952,0.5347834944725036,0.036821434614539225
M2,text,5,0.8190697312355042,0.03563334190353864,0.44534851908683776,0.030773561031959877
M2,audio,5,0.6621044635772705,0.05689547728439605,0.32293344140052793,0.05414120404134669
M2,vision,5,0.9195392370223999,0.03744298345037278,0.5340398073196411,0.0346960350528028
M3,text,15,0.8311106324195862,0.025205971351551344,0.45519693891207375,0.021717713583393992
M3,audio,15,0.38449583053588865,0.050754623523549675,0.14107545365889868,0.05381054674326515
M3,vision,15,0.868217929204305,0.03802189028391118,0.48890710473060606,0.03450529817617854
M4,text,15,0.4924671709537506,0.011169848357006705,0.18583014905452727,0.008262501869688806
M4,audio,15,0.37839911778767904,0.05503994105535867,0.13528098662694296,0.05838370221384358
M4,vision,15,0.8630666335423788,0.03867536107911695,0.48466227054595945,0.03512809212998346
1 method target_modality n mae_standardized_mean mae_standardized_std smooth_l1_standardized_mean smooth_l1_standardized_std
2 M1 text 5 0.8308010935783386 0.028634584948895196 0.4546235024929047 0.025151457325080215
3 M1 audio 5 0.6908806562423706 0.06720606353302543 0.34603241086006165 0.06541831705153776
4 M1 vision 5 0.9209612965583801 0.04009937912435952 0.5347834944725036 0.036821434614539225
5 M2 text 5 0.8190697312355042 0.03563334190353864 0.44534851908683776 0.030773561031959877
6 M2 audio 5 0.6621044635772705 0.05689547728439605 0.32293344140052793 0.05414120404134669
7 M2 vision 5 0.9195392370223999 0.03744298345037278 0.5340398073196411 0.0346960350528028
8 M3 text 15 0.8311106324195862 0.025205971351551344 0.45519693891207375 0.021717713583393992
9 M3 audio 15 0.38449583053588865 0.050754623523549675 0.14107545365889868 0.05381054674326515
10 M3 vision 15 0.868217929204305 0.03802189028391118 0.48890710473060606 0.03450529817617854
11 M4 text 15 0.4924671709537506 0.011169848357006705 0.18583014905452727 0.008262501869688806
12 M4 audio 15 0.37839911778767904 0.05503994105535867 0.13528098662694296 0.05838370221384358
13 M4 vision 15 0.8630666335423788 0.03867536107911695 0.48466227054595945 0.03512809212998346
@@ -0,0 +1,241 @@
method,seed,fold,direction,r_at_1,r_at_5,mrr,queries
M1,fixed,1,text_to_audio,0.0010000000474974513,0.00800000037997961,0.010086040943861008,1000.0
M1,fixed,1,audio_to_text,0.003000000026077032,0.00800000037997961,0.009918988682329655,1000.0
M1,fixed,1,text_to_vision,0.003000000026077032,0.009000000543892384,0.012364907190203667,1000.0
M1,fixed,1,vision_to_text,0.0,0.0,0.006473104935139418,1000.0
M1,fixed,1,audio_to_vision,0.0,0.00800000037997961,0.009185504168272018,1000.0
M1,fixed,1,vision_to_audio,0.0010000000474974513,0.006000000052154064,0.008802860975265503,1000.0
M2,fixed,1,text_to_audio,0.003000000026077032,0.013000000268220901,0.012891415506601334,1000.0
M2,fixed,1,audio_to_text,0.004000000189989805,0.01100000087171793,0.012235527858138084,1000.0
M2,fixed,1,text_to_vision,0.00800000037997961,0.014000000432133675,0.017583860084414482,1000.0
M2,fixed,1,vision_to_text,0.005000000353902578,0.00800000037997961,0.013148980215191841,1000.0
M2,fixed,1,audio_to_vision,0.0010000000474974513,0.00800000037997961,0.009523994289338589,1000.0
M2,fixed,1,vision_to_audio,0.0010000000474974513,0.006000000052154064,0.009459399618208408,1000.0
M3,42,1,text_to_audio,0.021000001579523087,0.03700000047683716,0.03344433382153511,1000.0
M3,42,1,audio_to_text,0.00800000037997961,0.014000000432133675,0.01682252623140812,1000.0
M3,42,1,text_to_vision,0.006000000052154064,0.027000000700354576,0.018166253343224525,1000.0
M3,42,1,vision_to_text,0.010000000707805157,0.02200000174343586,0.018923873081803322,1000.0
M3,42,1,audio_to_vision,0.003000000026077032,0.003000000026077032,0.010045807808637619,1000.0
M3,42,1,vision_to_audio,0.009000000543892384,0.013000000268220901,0.020716793835163116,1000.0
M4,42,1,text_to_audio,0.0,0.0,0.002745515899732709,1000.0
M4,42,1,audio_to_text,0.0010000000474974513,0.005000000353902578,0.006256598513573408,1000.0
M4,42,1,text_to_vision,0.0010000000474974513,0.005000000353902578,0.007646685931831598,1000.0
M4,42,1,vision_to_text,0.0020000000949949026,0.010000000707805157,0.011485339142382145,1000.0
M4,42,1,audio_to_vision,0.0020000000949949026,0.01100000087171793,0.011041815392673016,1000.0
M4,42,1,vision_to_audio,0.0010000000474974513,0.003000000026077032,0.005618138238787651,1000.0
M3,3407,1,text_to_audio,0.009000000543892384,0.017000000923871994,0.024300672113895416,1000.0
M3,3407,1,audio_to_text,0.005000000353902578,0.00800000037997961,0.016388019546866417,1000.0
M3,3407,1,text_to_vision,0.007000000216066837,0.01100000087171793,0.014694208279252052,1000.0
M3,3407,1,vision_to_text,0.013000000268220901,0.018000001087784767,0.019234079867601395,1000.0
M3,3407,1,audio_to_vision,0.007000000216066837,0.010000000707805157,0.01576545275747776,1000.0
M3,3407,1,vision_to_audio,0.005000000353902578,0.005000000353902578,0.010637721046805382,1000.0
M4,3407,1,text_to_audio,0.0010000000474974513,0.005000000353902578,0.006559826899319887,1000.0
M4,3407,1,audio_to_text,0.0010000000474974513,0.005000000353902578,0.006270123180001974,1000.0
M4,3407,1,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011605769395828247,1000.0
M4,3407,1,vision_to_text,0.003000000026077032,0.015000000596046448,0.016500672325491905,1000.0
M4,3407,1,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.00809124018996954,1000.0
M4,3407,1,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012351044453680515,1000.0
M3,2026,1,text_to_audio,0.017000000923871994,0.02500000037252903,0.026665309444069862,1000.0
M3,2026,1,audio_to_text,0.0010000000474974513,0.01100000087171793,0.011807681992650032,1000.0
M3,2026,1,text_to_vision,0.006000000052154064,0.012000000104308128,0.01321154274046421,1000.0
M3,2026,1,vision_to_text,0.0,0.006000000052154064,0.007229108829051256,1000.0
M3,2026,1,audio_to_vision,0.004000000189989805,0.010000000707805157,0.012219780124723911,1000.0
M3,2026,1,vision_to_audio,0.0,0.0,0.00736302649602294,1000.0
M4,2026,1,text_to_audio,0.0,0.0,0.00321578630246222,1000.0
M4,2026,1,audio_to_text,0.0,0.0,0.0029425874818116426,1000.0
M4,2026,1,text_to_vision,0.0010000000474974513,0.005000000353902578,0.00850450899451971,1000.0
M4,2026,1,vision_to_text,0.0020000000949949026,0.010000000707805157,0.01312504056841135,1000.0
M4,2026,1,audio_to_vision,0.0020000000949949026,0.010000000707805157,0.01287905778735876,1000.0
M4,2026,1,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012796711176633835,1000.0
M1,fixed,2,text_to_audio,0.004000000189989805,0.009000000543892384,0.010721729137003422,1000.0
M1,fixed,2,audio_to_text,0.0020000000949949026,0.005000000353902578,0.008698307909071445,1000.0
M1,fixed,2,text_to_vision,0.0010000000474974513,0.003000000026077032,0.007055815774947405,1000.0
M1,fixed,2,vision_to_text,0.006000000052154064,0.006000000052154064,0.011396339163184166,1000.0
M1,fixed,2,audio_to_vision,0.0010000000474974513,0.003000000026077032,0.006511436775326729,1000.0
M1,fixed,2,vision_to_audio,0.003000000026077032,0.00800000037997961,0.008912489749491215,1000.0
M2,fixed,2,text_to_audio,0.0020000000949949026,0.007000000216066837,0.008748839609324932,1000.0
M2,fixed,2,audio_to_text,0.003000000026077032,0.005000000353902578,0.008852123282849789,1000.0
M2,fixed,2,text_to_vision,0.004000000189989805,0.010000000707805157,0.01082402840256691,1000.0
M2,fixed,2,vision_to_text,0.007000000216066837,0.01100000087171793,0.014397574588656425,1000.0
M2,fixed,2,audio_to_vision,0.0010000000474974513,0.004000000189989805,0.007378880865871906,1000.0
M2,fixed,2,vision_to_audio,0.005000000353902578,0.006000000052154064,0.010349743068218231,1000.0
M3,42,2,text_to_audio,0.0010000000474974513,0.009000000543892384,0.008156482130289078,1000.0
M3,42,2,audio_to_text,0.0,0.0010000000474974513,0.004495462402701378,1000.0
M3,42,2,text_to_vision,0.004000000189989805,0.01100000087171793,0.012809447944164276,1000.0
M3,42,2,vision_to_text,0.00800000037997961,0.009000000543892384,0.013898583129048347,1000.0
M3,42,2,audio_to_vision,0.003000000026077032,0.00800000037997961,0.009653640910983086,1000.0
M3,42,2,vision_to_audio,0.0,0.0010000000474974513,0.0033326284028589725,1000.0
M4,42,2,text_to_audio,0.0,0.0,0.002928863512352109,1000.0
M4,42,2,audio_to_text,0.0,0.0,0.0023521804250776768,1000.0
M4,42,2,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011110157705843449,1000.0
M4,42,2,vision_to_text,0.0010000000474974513,0.005000000353902578,0.006895906291902065,1000.0
M4,42,2,audio_to_vision,0.003000000026077032,0.015000000596046448,0.015711786225438118,1000.0
M4,42,2,vision_to_audio,0.0020000000949949026,0.012000000104308128,0.012692151591181755,1000.0
M3,3407,2,text_to_audio,0.018000001087784767,0.05300000309944153,0.041355911642313004,1000.0
M3,3407,2,audio_to_text,0.007000000216066837,0.03100000135600567,0.023398615419864655,1000.0
M3,3407,2,text_to_vision,0.0010000000474974513,0.005000000353902578,0.009130735881626606,1000.0
M3,3407,2,vision_to_text,0.0,0.006000000052154064,0.008097999729216099,1000.0
M3,3407,2,audio_to_vision,0.006000000052154064,0.010000000707805157,0.013130152598023415,1000.0
M3,3407,2,vision_to_audio,0.0020000000949949026,0.009000000543892384,0.008952965028584003,1000.0
M4,3407,2,text_to_audio,0.0,0.0,0.003208547830581665,1000.0
M4,3407,2,audio_to_text,0.0,0.0,0.0023785873781889677,1000.0
M4,3407,2,text_to_vision,0.0,0.0,0.0033639008179306984,1000.0
M4,3407,2,vision_to_text,0.0020000000949949026,0.010000000707805157,0.01099696196615696,1000.0
M4,3407,2,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.006608974188566208,1000.0
M4,3407,2,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011376568116247654,1000.0
M3,2026,2,text_to_audio,0.024000000208616257,0.04800000041723251,0.04059242084622383,1000.0
M3,2026,2,audio_to_text,0.0010000000474974513,0.013000000268220901,0.014835278503596783,1000.0
M3,2026,2,text_to_vision,0.014000000432133675,0.02200000174343586,0.02588539943099022,1000.0
M3,2026,2,vision_to_text,0.007000000216066837,0.018000001087784767,0.016461974009871483,1000.0
M3,2026,2,audio_to_vision,0.012000000104308128,0.012000000104308128,0.018243461847305298,1000.0
M3,2026,2,vision_to_audio,0.006000000052154064,0.015000000596046448,0.015246634371578693,1000.0
M4,2026,2,text_to_audio,0.0010000000474974513,0.005000000353902578,0.006106700282543898,1000.0
M4,2026,2,audio_to_text,0.0,0.0,0.0024518424179404974,1000.0
M4,2026,2,text_to_vision,0.0,0.0,0.003404181217774749,1000.0
M4,2026,2,vision_to_text,0.0,0.0,0.004238351248204708,1000.0
M4,2026,2,audio_to_vision,0.003000000026077032,0.015000000596046448,0.016335317865014076,1000.0
M4,2026,2,vision_to_audio,0.0010000000474974513,0.005000000353902578,0.007721395697444677,1000.0
M1,fixed,3,text_to_audio,0.0010000000474974513,0.005000000353902578,0.008115584030747414,1000.0
M1,fixed,3,audio_to_text,0.003000000026077032,0.006000000052154064,0.010013893246650696,1000.0
M1,fixed,3,text_to_vision,0.005000000353902578,0.01100000087171793,0.014654486440122128,1000.0
M1,fixed,3,vision_to_text,0.00800000037997961,0.01100000087171793,0.015268602408468723,1000.0
M1,fixed,3,audio_to_vision,0.0,0.007000000216066837,0.0085630863904953,1000.0
M1,fixed,3,vision_to_audio,0.0010000000474974513,0.003000000026077032,0.007109377533197403,1000.0
M2,fixed,3,text_to_audio,0.0010000000474974513,0.00800000037997961,0.009955259971320629,1000.0
M2,fixed,3,audio_to_text,0.006000000052154064,0.013000000268220901,0.013718895614147186,1000.0
M2,fixed,3,text_to_vision,0.005000000353902578,0.010000000707805157,0.014849589206278324,1000.0
M2,fixed,3,vision_to_text,0.006000000052154064,0.00800000037997961,0.014947512187063694,1000.0
M2,fixed,3,audio_to_vision,0.0020000000949949026,0.006000000052154064,0.009267324581742287,1000.0
M2,fixed,3,vision_to_audio,0.0,0.0020000000949949026,0.005884611047804356,1000.0
M3,42,3,text_to_audio,0.014000000432133675,0.0430000014603138,0.03258649632334709,1000.0
M3,42,3,audio_to_text,0.006000000052154064,0.010000000707805157,0.01843179017305374,1000.0
M3,42,3,text_to_vision,0.013000000268220901,0.024000000208616257,0.023365069180727005,1000.0
M3,42,3,vision_to_text,0.006000000052154064,0.015000000596046448,0.015362463891506195,1000.0
M3,42,3,audio_to_vision,0.007000000216066837,0.01900000125169754,0.016374722123146057,1000.0
M3,42,3,vision_to_audio,0.003000000026077032,0.006000000052154064,0.008777403272688389,1000.0
M4,42,3,text_to_audio,0.0,0.0,0.002880551852285862,1000.0
M4,42,3,audio_to_text,0.0,0.0,0.003347013145685196,1000.0
M4,42,3,text_to_vision,0.004000000189989805,0.020000001415610313,0.021214386448264122,1000.0
M4,42,3,vision_to_text,0.003000000026077032,0.015000000596046448,0.016206368803977966,1000.0
M4,42,3,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.00828514900058508,1000.0
M4,42,3,vision_to_audio,0.0010000000474974513,0.005000000353902578,0.007783014792948961,1000.0
M3,3407,3,text_to_audio,0.039000000804662704,0.055000003427267075,0.05350850895047188,1000.0
M3,3407,3,audio_to_text,0.03800000250339508,0.04400000348687172,0.04879831150174141,1000.0
M3,3407,3,text_to_vision,0.0020000000949949026,0.009000000543892384,0.011307715438306332,1000.0
M3,3407,3,vision_to_text,0.003000000026077032,0.013000000268220901,0.011990648694336414,1000.0
M3,3407,3,audio_to_vision,0.007000000216066837,0.02500000037252903,0.022219380363821983,1000.0
M3,3407,3,vision_to_audio,0.003000000026077032,0.006000000052154064,0.009232758544385433,1000.0
M4,3407,3,text_to_audio,0.0,0.0,0.0037817470729351044,1000.0
M4,3407,3,audio_to_text,0.0,0.0,0.004079542122781277,1000.0
M4,3407,3,text_to_vision,0.003000000026077032,0.014000000432133675,0.015179671347141266,1000.0
M4,3407,3,vision_to_text,0.0,0.0,0.004454099107533693,1000.0
M4,3407,3,audio_to_vision,0.003000000026077032,0.015000000596046448,0.016372714191675186,1000.0
M4,3407,3,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011763687245547771,1000.0
M3,2026,3,text_to_audio,0.012000000104308128,0.02800000086426735,0.029318703338503838,1000.0
M3,2026,3,audio_to_text,0.012000000104308128,0.021000001579523087,0.027406470850110054,1000.0
M3,2026,3,text_to_vision,0.005000000353902578,0.01100000087171793,0.01579951122403145,1000.0
M3,2026,3,vision_to_text,0.006000000052154064,0.014000000432133675,0.015296950936317444,1000.0
M3,2026,3,audio_to_vision,0.006000000052154064,0.018000001087784767,0.016852334141731262,1000.0
M3,2026,3,vision_to_audio,0.005000000353902578,0.007000000216066837,0.01105457916855812,1000.0
M4,2026,3,text_to_audio,0.0010000000474974513,0.005000000353902578,0.007778078783303499,1000.0
M4,2026,3,audio_to_text,0.0010000000474974513,0.004000000189989805,0.007144550327211618,1000.0
M4,2026,3,text_to_vision,0.003000000026077032,0.015000000596046448,0.016557173803448677,1000.0
M4,2026,3,vision_to_text,0.0010000000474974513,0.005000000353902578,0.008665882982313633,1000.0
M4,2026,3,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.008280493319034576,1000.0
M4,2026,3,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011866509914398193,1000.0
M1,fixed,4,text_to_audio,0.0,0.006000000052154064,0.006950765382498503,1000.0
M1,fixed,4,audio_to_text,0.005000000353902578,0.00800000037997961,0.010673508048057556,1000.0
M1,fixed,4,text_to_vision,0.0010000000474974513,0.007000000216066837,0.007159821223467588,1000.0
M1,fixed,4,vision_to_text,0.009000000543892384,0.009000000543892384,0.01242038607597351,1000.0
M1,fixed,4,audio_to_vision,0.0020000000949949026,0.013000000268220901,0.01288546621799469,1000.0
M1,fixed,4,vision_to_audio,0.0020000000949949026,0.01100000087171793,0.012257258407771587,1000.0
M2,fixed,4,text_to_audio,0.0010000000474974513,0.009000000543892384,0.009888535365462303,1000.0
M2,fixed,4,audio_to_text,0.005000000353902578,0.00800000037997961,0.011750306934118271,1000.0
M2,fixed,4,text_to_vision,0.0010000000474974513,0.0010000000474974513,0.005150882992893457,1000.0
M2,fixed,4,vision_to_text,0.010000000707805157,0.010000000707805157,0.013726433739066124,1000.0
M2,fixed,4,audio_to_vision,0.0010000000474974513,0.004000000189989805,0.00889560766518116,1000.0
M2,fixed,4,vision_to_audio,0.0020000000949949026,0.01100000087171793,0.011912405490875244,1000.0
M3,42,4,text_to_audio,0.02500000037252903,0.036000002175569534,0.038420792669057846,1000.0
M3,42,4,audio_to_text,0.00800000037997961,0.023000001907348633,0.02173793874680996,1000.0
M3,42,4,text_to_vision,0.0,0.005000000353902578,0.006278287619352341,1000.0
M3,42,4,vision_to_text,0.0020000000949949026,0.003000000026077032,0.007351293694227934,1000.0
M3,42,4,audio_to_vision,0.004000000189989805,0.01100000087171793,0.010880847461521626,1000.0
M3,42,4,vision_to_audio,0.006000000052154064,0.013000000268220901,0.015541859902441502,1000.0
M4,42,4,text_to_audio,0.004000000189989805,0.01900000125169754,0.020215420052409172,1000.0
M4,42,4,audio_to_text,0.0,0.0,0.004233467858284712,1000.0
M4,42,4,text_to_vision,0.0020000000949949026,0.010000000707805157,0.012029506266117096,1000.0
M4,42,4,vision_to_text,0.0,0.0,0.002725329017266631,1000.0
M4,42,4,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.009229524992406368,1000.0
M4,42,4,vision_to_audio,0.003000000026077032,0.015000000596046448,0.017202889546751976,1000.0
M3,3407,4,text_to_audio,0.013000000268220901,0.032999999821186066,0.02899651788175106,1000.0
M3,3407,4,audio_to_text,0.007000000216066837,0.013000000268220901,0.01741517148911953,1000.0
M3,3407,4,text_to_vision,0.003000000026077032,0.006000000052154064,0.009169278666377068,1000.0
M3,3407,4,vision_to_text,0.006000000052154064,0.006000000052154064,0.011985262855887413,1000.0
M3,3407,4,audio_to_vision,0.0,0.004000000189989805,0.008569329045712948,1000.0
M3,3407,4,vision_to_audio,0.007000000216066837,0.010000000707805157,0.01606379821896553,1000.0
M4,3407,4,text_to_audio,0.003000000026077032,0.015000000596046448,0.015794366598129272,1000.0
M4,3407,4,audio_to_text,0.0,0.0,0.004071671981364489,1000.0
M4,3407,4,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011711233295500278,1000.0
M4,3407,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.00810206588357687,1000.0
M4,3407,4,audio_to_vision,0.0020000000949949026,0.010000000707805157,0.013361790217459202,1000.0
M4,3407,4,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012590405531227589,1000.0
M3,2026,4,text_to_audio,0.024000000208616257,0.04000000283122063,0.039450231939554214,1000.0
M3,2026,4,audio_to_text,0.006000000052154064,0.015000000596046448,0.01841675490140915,1000.0
M3,2026,4,text_to_vision,0.01100000087171793,0.01600000075995922,0.01930876635015011,1000.0
M3,2026,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.007208418566733599,1000.0
M3,2026,4,audio_to_vision,0.00800000037997961,0.017000000923871994,0.01701623573899269,1000.0
M3,2026,4,vision_to_audio,0.006000000052154064,0.014000000432133675,0.015153588727116585,1000.0
M4,2026,4,text_to_audio,0.0020000000949949026,0.010000000707805157,0.011758673004806042,1000.0
M4,2026,4,audio_to_text,0.0010000000474974513,0.005000000353902578,0.008434664458036423,1000.0
M4,2026,4,text_to_vision,0.0,0.0,0.0034921157639473677,1000.0
M4,2026,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.007859558798372746,1000.0
M4,2026,4,audio_to_vision,0.003000000026077032,0.014000000432133675,0.01731833629310131,1000.0
M4,2026,4,vision_to_audio,0.003000000026077032,0.015000000596046448,0.016523126512765884,1000.0
M1,fixed,5,text_to_audio,0.0,0.007000000216066837,0.008882568217813969,1000.0
M1,fixed,5,audio_to_text,0.0010000000474974513,0.003000000026077032,0.008665341883897781,1000.0
M1,fixed,5,text_to_vision,0.0,0.005000000353902578,0.00567932054400444,1000.0
M1,fixed,5,vision_to_text,0.0020000000949949026,0.009000000543892384,0.00881400890648365,1000.0
M1,fixed,5,audio_to_vision,0.0020000000949949026,0.012000000104308128,0.012908315286040306,1000.0
M1,fixed,5,vision_to_audio,0.003000000026077032,0.012000000104308128,0.0124351242557168,1000.0
M2,fixed,5,text_to_audio,0.0,0.009000000543892384,0.009715890511870384,1000.0
M2,fixed,5,audio_to_text,0.0020000000949949026,0.009000000543892384,0.010659169405698776,1000.0
M2,fixed,5,text_to_vision,0.0,0.009000000543892384,0.008167226798832417,1000.0
M2,fixed,5,vision_to_text,0.006000000052154064,0.009000000543892384,0.013293917290866375,1000.0
M2,fixed,5,audio_to_vision,0.0020000000949949026,0.012000000104308128,0.012093348428606987,1000.0
M2,fixed,5,vision_to_audio,0.0020000000949949026,0.007000000216066837,0.010753415524959564,1000.0
M3,42,5,text_to_audio,0.01600000075995922,0.06400000303983688,0.04854178801178932,1000.0
M3,42,5,audio_to_text,0.014000000432133675,0.03700000047683716,0.034194398671388626,1000.0
M3,42,5,text_to_vision,0.006000000052154064,0.006000000052154064,0.011698092333972454,1000.0
M3,42,5,vision_to_text,0.005000000353902578,0.012000000104308128,0.012927225790917873,1000.0
M3,42,5,audio_to_vision,0.013000000268220901,0.026000000536441803,0.02460731379687786,1000.0
M3,42,5,vision_to_audio,0.004000000189989805,0.007000000216066837,0.012679936364293098,1000.0
M4,42,5,text_to_audio,0.04000000283122063,0.052000001072883606,0.05324704200029373,1000.0
M4,42,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.008529611863195896,1000.0
M4,42,5,text_to_vision,0.0010000000474974513,0.005000000353902578,0.00876129325479269,1000.0
M4,42,5,vision_to_text,0.0020000000949949026,0.010000000707805157,0.011993298307061195,1000.0
M4,42,5,audio_to_vision,0.003000000026077032,0.015000000596046448,0.017660561949014664,1000.0
M4,42,5,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.014293942600488663,1000.0
M3,3407,5,text_to_audio,0.03100000135600567,0.07200000435113907,0.061651285737752914,1000.0
M3,3407,5,audio_to_text,0.018000001087784767,0.04000000283122063,0.042421478778123856,1000.0
M3,3407,5,text_to_vision,0.009000000543892384,0.021000001579523087,0.020427849143743515,1000.0
M3,3407,5,vision_to_text,0.004000000189989805,0.013000000268220901,0.012980340048670769,1000.0
M3,3407,5,audio_to_vision,0.007000000216066837,0.018000001087784767,0.021291999146342278,1000.0
M3,3407,5,vision_to_audio,0.004000000189989805,0.01600000075995922,0.016669930890202522,1000.0
M4,3407,5,text_to_audio,0.0020000000949949026,0.010000000707805157,0.0127327349036932,1000.0
M4,3407,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.00789211317896843,1000.0
M4,3407,5,text_to_vision,0.0020000000949949026,0.005000000353902578,0.00888831913471222,1000.0
M4,3407,5,vision_to_text,0.0010000000474974513,0.005000000353902578,0.008448818698525429,1000.0
M4,3407,5,audio_to_vision,0.003000000026077032,0.013000000268220901,0.016633378341794014,1000.0
M4,3407,5,vision_to_audio,0.003000000026077032,0.015000000596046448,0.01952446810901165,1000.0
M3,2026,5,text_to_audio,0.03100000135600567,0.08500000089406967,0.06357216835021973,1000.0
M3,2026,5,audio_to_text,0.005000000353902578,0.032999999821186066,0.030199285596609116,1000.0
M3,2026,5,text_to_vision,0.007000000216066837,0.021000001579523087,0.015703614801168442,1000.0
M3,2026,5,vision_to_text,0.005000000353902578,0.009000000543892384,0.01244888361543417,1000.0
M3,2026,5,audio_to_vision,0.009000000543892384,0.023000001907348633,0.02339860051870346,1000.0
M3,2026,5,vision_to_audio,0.004000000189989805,0.015000000596046448,0.01597834937274456,1000.0
M4,2026,5,text_to_audio,0.052000001072883606,0.06300000101327896,0.06472568958997726,1000.0
M4,2026,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.007973534055054188,1000.0
M4,2026,5,text_to_vision,0.0010000000474974513,0.005000000353902578,0.007530428934842348,1000.0
M4,2026,5,vision_to_text,0.003000000026077032,0.015000000596046448,0.016602179035544395,1000.0
M4,2026,5,audio_to_vision,0.003000000026077032,0.015000000596046448,0.017103707417845726,1000.0
M4,2026,5,vision_to_audio,0.003000000026077032,0.014000000432133675,0.01859205961227417,1000.0
1 method seed fold direction r_at_1 r_at_5 mrr queries
2 M1 fixed 1 text_to_audio 0.0010000000474974513 0.00800000037997961 0.010086040943861008 1000.0
3 M1 fixed 1 audio_to_text 0.003000000026077032 0.00800000037997961 0.009918988682329655 1000.0
4 M1 fixed 1 text_to_vision 0.003000000026077032 0.009000000543892384 0.012364907190203667 1000.0
5 M1 fixed 1 vision_to_text 0.0 0.0 0.006473104935139418 1000.0
6 M1 fixed 1 audio_to_vision 0.0 0.00800000037997961 0.009185504168272018 1000.0
7 M1 fixed 1 vision_to_audio 0.0010000000474974513 0.006000000052154064 0.008802860975265503 1000.0
8 M2 fixed 1 text_to_audio 0.003000000026077032 0.013000000268220901 0.012891415506601334 1000.0
9 M2 fixed 1 audio_to_text 0.004000000189989805 0.01100000087171793 0.012235527858138084 1000.0
10 M2 fixed 1 text_to_vision 0.00800000037997961 0.014000000432133675 0.017583860084414482 1000.0
11 M2 fixed 1 vision_to_text 0.005000000353902578 0.00800000037997961 0.013148980215191841 1000.0
12 M2 fixed 1 audio_to_vision 0.0010000000474974513 0.00800000037997961 0.009523994289338589 1000.0
13 M2 fixed 1 vision_to_audio 0.0010000000474974513 0.006000000052154064 0.009459399618208408 1000.0
14 M3 42 1 text_to_audio 0.021000001579523087 0.03700000047683716 0.03344433382153511 1000.0
15 M3 42 1 audio_to_text 0.00800000037997961 0.014000000432133675 0.01682252623140812 1000.0
16 M3 42 1 text_to_vision 0.006000000052154064 0.027000000700354576 0.018166253343224525 1000.0
17 M3 42 1 vision_to_text 0.010000000707805157 0.02200000174343586 0.018923873081803322 1000.0
18 M3 42 1 audio_to_vision 0.003000000026077032 0.003000000026077032 0.010045807808637619 1000.0
19 M3 42 1 vision_to_audio 0.009000000543892384 0.013000000268220901 0.020716793835163116 1000.0
20 M4 42 1 text_to_audio 0.0 0.0 0.002745515899732709 1000.0
21 M4 42 1 audio_to_text 0.0010000000474974513 0.005000000353902578 0.006256598513573408 1000.0
22 M4 42 1 text_to_vision 0.0010000000474974513 0.005000000353902578 0.007646685931831598 1000.0
23 M4 42 1 vision_to_text 0.0020000000949949026 0.010000000707805157 0.011485339142382145 1000.0
24 M4 42 1 audio_to_vision 0.0020000000949949026 0.01100000087171793 0.011041815392673016 1000.0
25 M4 42 1 vision_to_audio 0.0010000000474974513 0.003000000026077032 0.005618138238787651 1000.0
26 M3 3407 1 text_to_audio 0.009000000543892384 0.017000000923871994 0.024300672113895416 1000.0
27 M3 3407 1 audio_to_text 0.005000000353902578 0.00800000037997961 0.016388019546866417 1000.0
28 M3 3407 1 text_to_vision 0.007000000216066837 0.01100000087171793 0.014694208279252052 1000.0
29 M3 3407 1 vision_to_text 0.013000000268220901 0.018000001087784767 0.019234079867601395 1000.0
30 M3 3407 1 audio_to_vision 0.007000000216066837 0.010000000707805157 0.01576545275747776 1000.0
31 M3 3407 1 vision_to_audio 0.005000000353902578 0.005000000353902578 0.010637721046805382 1000.0
32 M4 3407 1 text_to_audio 0.0010000000474974513 0.005000000353902578 0.006559826899319887 1000.0
33 M4 3407 1 audio_to_text 0.0010000000474974513 0.005000000353902578 0.006270123180001974 1000.0
34 M4 3407 1 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011605769395828247 1000.0
35 M4 3407 1 vision_to_text 0.003000000026077032 0.015000000596046448 0.016500672325491905 1000.0
36 M4 3407 1 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.00809124018996954 1000.0
37 M4 3407 1 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012351044453680515 1000.0
38 M3 2026 1 text_to_audio 0.017000000923871994 0.02500000037252903 0.026665309444069862 1000.0
39 M3 2026 1 audio_to_text 0.0010000000474974513 0.01100000087171793 0.011807681992650032 1000.0
40 M3 2026 1 text_to_vision 0.006000000052154064 0.012000000104308128 0.01321154274046421 1000.0
41 M3 2026 1 vision_to_text 0.0 0.006000000052154064 0.007229108829051256 1000.0
42 M3 2026 1 audio_to_vision 0.004000000189989805 0.010000000707805157 0.012219780124723911 1000.0
43 M3 2026 1 vision_to_audio 0.0 0.0 0.00736302649602294 1000.0
44 M4 2026 1 text_to_audio 0.0 0.0 0.00321578630246222 1000.0
45 M4 2026 1 audio_to_text 0.0 0.0 0.0029425874818116426 1000.0
46 M4 2026 1 text_to_vision 0.0010000000474974513 0.005000000353902578 0.00850450899451971 1000.0
47 M4 2026 1 vision_to_text 0.0020000000949949026 0.010000000707805157 0.01312504056841135 1000.0
48 M4 2026 1 audio_to_vision 0.0020000000949949026 0.010000000707805157 0.01287905778735876 1000.0
49 M4 2026 1 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012796711176633835 1000.0
50 M1 fixed 2 text_to_audio 0.004000000189989805 0.009000000543892384 0.010721729137003422 1000.0
51 M1 fixed 2 audio_to_text 0.0020000000949949026 0.005000000353902578 0.008698307909071445 1000.0
52 M1 fixed 2 text_to_vision 0.0010000000474974513 0.003000000026077032 0.007055815774947405 1000.0
53 M1 fixed 2 vision_to_text 0.006000000052154064 0.006000000052154064 0.011396339163184166 1000.0
54 M1 fixed 2 audio_to_vision 0.0010000000474974513 0.003000000026077032 0.006511436775326729 1000.0
55 M1 fixed 2 vision_to_audio 0.003000000026077032 0.00800000037997961 0.008912489749491215 1000.0
56 M2 fixed 2 text_to_audio 0.0020000000949949026 0.007000000216066837 0.008748839609324932 1000.0
57 M2 fixed 2 audio_to_text 0.003000000026077032 0.005000000353902578 0.008852123282849789 1000.0
58 M2 fixed 2 text_to_vision 0.004000000189989805 0.010000000707805157 0.01082402840256691 1000.0
59 M2 fixed 2 vision_to_text 0.007000000216066837 0.01100000087171793 0.014397574588656425 1000.0
60 M2 fixed 2 audio_to_vision 0.0010000000474974513 0.004000000189989805 0.007378880865871906 1000.0
61 M2 fixed 2 vision_to_audio 0.005000000353902578 0.006000000052154064 0.010349743068218231 1000.0
62 M3 42 2 text_to_audio 0.0010000000474974513 0.009000000543892384 0.008156482130289078 1000.0
63 M3 42 2 audio_to_text 0.0 0.0010000000474974513 0.004495462402701378 1000.0
64 M3 42 2 text_to_vision 0.004000000189989805 0.01100000087171793 0.012809447944164276 1000.0
65 M3 42 2 vision_to_text 0.00800000037997961 0.009000000543892384 0.013898583129048347 1000.0
66 M3 42 2 audio_to_vision 0.003000000026077032 0.00800000037997961 0.009653640910983086 1000.0
67 M3 42 2 vision_to_audio 0.0 0.0010000000474974513 0.0033326284028589725 1000.0
68 M4 42 2 text_to_audio 0.0 0.0 0.002928863512352109 1000.0
69 M4 42 2 audio_to_text 0.0 0.0 0.0023521804250776768 1000.0
70 M4 42 2 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011110157705843449 1000.0
71 M4 42 2 vision_to_text 0.0010000000474974513 0.005000000353902578 0.006895906291902065 1000.0
72 M4 42 2 audio_to_vision 0.003000000026077032 0.015000000596046448 0.015711786225438118 1000.0
73 M4 42 2 vision_to_audio 0.0020000000949949026 0.012000000104308128 0.012692151591181755 1000.0
74 M3 3407 2 text_to_audio 0.018000001087784767 0.05300000309944153 0.041355911642313004 1000.0
75 M3 3407 2 audio_to_text 0.007000000216066837 0.03100000135600567 0.023398615419864655 1000.0
76 M3 3407 2 text_to_vision 0.0010000000474974513 0.005000000353902578 0.009130735881626606 1000.0
77 M3 3407 2 vision_to_text 0.0 0.006000000052154064 0.008097999729216099 1000.0
78 M3 3407 2 audio_to_vision 0.006000000052154064 0.010000000707805157 0.013130152598023415 1000.0
79 M3 3407 2 vision_to_audio 0.0020000000949949026 0.009000000543892384 0.008952965028584003 1000.0
80 M4 3407 2 text_to_audio 0.0 0.0 0.003208547830581665 1000.0
81 M4 3407 2 audio_to_text 0.0 0.0 0.0023785873781889677 1000.0
82 M4 3407 2 text_to_vision 0.0 0.0 0.0033639008179306984 1000.0
83 M4 3407 2 vision_to_text 0.0020000000949949026 0.010000000707805157 0.01099696196615696 1000.0
84 M4 3407 2 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.006608974188566208 1000.0
85 M4 3407 2 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011376568116247654 1000.0
86 M3 2026 2 text_to_audio 0.024000000208616257 0.04800000041723251 0.04059242084622383 1000.0
87 M3 2026 2 audio_to_text 0.0010000000474974513 0.013000000268220901 0.014835278503596783 1000.0
88 M3 2026 2 text_to_vision 0.014000000432133675 0.02200000174343586 0.02588539943099022 1000.0
89 M3 2026 2 vision_to_text 0.007000000216066837 0.018000001087784767 0.016461974009871483 1000.0
90 M3 2026 2 audio_to_vision 0.012000000104308128 0.012000000104308128 0.018243461847305298 1000.0
91 M3 2026 2 vision_to_audio 0.006000000052154064 0.015000000596046448 0.015246634371578693 1000.0
92 M4 2026 2 text_to_audio 0.0010000000474974513 0.005000000353902578 0.006106700282543898 1000.0
93 M4 2026 2 audio_to_text 0.0 0.0 0.0024518424179404974 1000.0
94 M4 2026 2 text_to_vision 0.0 0.0 0.003404181217774749 1000.0
95 M4 2026 2 vision_to_text 0.0 0.0 0.004238351248204708 1000.0
96 M4 2026 2 audio_to_vision 0.003000000026077032 0.015000000596046448 0.016335317865014076 1000.0
97 M4 2026 2 vision_to_audio 0.0010000000474974513 0.005000000353902578 0.007721395697444677 1000.0
98 M1 fixed 3 text_to_audio 0.0010000000474974513 0.005000000353902578 0.008115584030747414 1000.0
99 M1 fixed 3 audio_to_text 0.003000000026077032 0.006000000052154064 0.010013893246650696 1000.0
100 M1 fixed 3 text_to_vision 0.005000000353902578 0.01100000087171793 0.014654486440122128 1000.0
101 M1 fixed 3 vision_to_text 0.00800000037997961 0.01100000087171793 0.015268602408468723 1000.0
102 M1 fixed 3 audio_to_vision 0.0 0.007000000216066837 0.0085630863904953 1000.0
103 M1 fixed 3 vision_to_audio 0.0010000000474974513 0.003000000026077032 0.007109377533197403 1000.0
104 M2 fixed 3 text_to_audio 0.0010000000474974513 0.00800000037997961 0.009955259971320629 1000.0
105 M2 fixed 3 audio_to_text 0.006000000052154064 0.013000000268220901 0.013718895614147186 1000.0
106 M2 fixed 3 text_to_vision 0.005000000353902578 0.010000000707805157 0.014849589206278324 1000.0
107 M2 fixed 3 vision_to_text 0.006000000052154064 0.00800000037997961 0.014947512187063694 1000.0
108 M2 fixed 3 audio_to_vision 0.0020000000949949026 0.006000000052154064 0.009267324581742287 1000.0
109 M2 fixed 3 vision_to_audio 0.0 0.0020000000949949026 0.005884611047804356 1000.0
110 M3 42 3 text_to_audio 0.014000000432133675 0.0430000014603138 0.03258649632334709 1000.0
111 M3 42 3 audio_to_text 0.006000000052154064 0.010000000707805157 0.01843179017305374 1000.0
112 M3 42 3 text_to_vision 0.013000000268220901 0.024000000208616257 0.023365069180727005 1000.0
113 M3 42 3 vision_to_text 0.006000000052154064 0.015000000596046448 0.015362463891506195 1000.0
114 M3 42 3 audio_to_vision 0.007000000216066837 0.01900000125169754 0.016374722123146057 1000.0
115 M3 42 3 vision_to_audio 0.003000000026077032 0.006000000052154064 0.008777403272688389 1000.0
116 M4 42 3 text_to_audio 0.0 0.0 0.002880551852285862 1000.0
117 M4 42 3 audio_to_text 0.0 0.0 0.003347013145685196 1000.0
118 M4 42 3 text_to_vision 0.004000000189989805 0.020000001415610313 0.021214386448264122 1000.0
119 M4 42 3 vision_to_text 0.003000000026077032 0.015000000596046448 0.016206368803977966 1000.0
120 M4 42 3 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.00828514900058508 1000.0
121 M4 42 3 vision_to_audio 0.0010000000474974513 0.005000000353902578 0.007783014792948961 1000.0
122 M3 3407 3 text_to_audio 0.039000000804662704 0.055000003427267075 0.05350850895047188 1000.0
123 M3 3407 3 audio_to_text 0.03800000250339508 0.04400000348687172 0.04879831150174141 1000.0
124 M3 3407 3 text_to_vision 0.0020000000949949026 0.009000000543892384 0.011307715438306332 1000.0
125 M3 3407 3 vision_to_text 0.003000000026077032 0.013000000268220901 0.011990648694336414 1000.0
126 M3 3407 3 audio_to_vision 0.007000000216066837 0.02500000037252903 0.022219380363821983 1000.0
127 M3 3407 3 vision_to_audio 0.003000000026077032 0.006000000052154064 0.009232758544385433 1000.0
128 M4 3407 3 text_to_audio 0.0 0.0 0.0037817470729351044 1000.0
129 M4 3407 3 audio_to_text 0.0 0.0 0.004079542122781277 1000.0
130 M4 3407 3 text_to_vision 0.003000000026077032 0.014000000432133675 0.015179671347141266 1000.0
131 M4 3407 3 vision_to_text 0.0 0.0 0.004454099107533693 1000.0
132 M4 3407 3 audio_to_vision 0.003000000026077032 0.015000000596046448 0.016372714191675186 1000.0
133 M4 3407 3 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011763687245547771 1000.0
134 M3 2026 3 text_to_audio 0.012000000104308128 0.02800000086426735 0.029318703338503838 1000.0
135 M3 2026 3 audio_to_text 0.012000000104308128 0.021000001579523087 0.027406470850110054 1000.0
136 M3 2026 3 text_to_vision 0.005000000353902578 0.01100000087171793 0.01579951122403145 1000.0
137 M3 2026 3 vision_to_text 0.006000000052154064 0.014000000432133675 0.015296950936317444 1000.0
138 M3 2026 3 audio_to_vision 0.006000000052154064 0.018000001087784767 0.016852334141731262 1000.0
139 M3 2026 3 vision_to_audio 0.005000000353902578 0.007000000216066837 0.01105457916855812 1000.0
140 M4 2026 3 text_to_audio 0.0010000000474974513 0.005000000353902578 0.007778078783303499 1000.0
141 M4 2026 3 audio_to_text 0.0010000000474974513 0.004000000189989805 0.007144550327211618 1000.0
142 M4 2026 3 text_to_vision 0.003000000026077032 0.015000000596046448 0.016557173803448677 1000.0
143 M4 2026 3 vision_to_text 0.0010000000474974513 0.005000000353902578 0.008665882982313633 1000.0
144 M4 2026 3 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.008280493319034576 1000.0
145 M4 2026 3 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011866509914398193 1000.0
146 M1 fixed 4 text_to_audio 0.0 0.006000000052154064 0.006950765382498503 1000.0
147 M1 fixed 4 audio_to_text 0.005000000353902578 0.00800000037997961 0.010673508048057556 1000.0
148 M1 fixed 4 text_to_vision 0.0010000000474974513 0.007000000216066837 0.007159821223467588 1000.0
149 M1 fixed 4 vision_to_text 0.009000000543892384 0.009000000543892384 0.01242038607597351 1000.0
150 M1 fixed 4 audio_to_vision 0.0020000000949949026 0.013000000268220901 0.01288546621799469 1000.0
151 M1 fixed 4 vision_to_audio 0.0020000000949949026 0.01100000087171793 0.012257258407771587 1000.0
152 M2 fixed 4 text_to_audio 0.0010000000474974513 0.009000000543892384 0.009888535365462303 1000.0
153 M2 fixed 4 audio_to_text 0.005000000353902578 0.00800000037997961 0.011750306934118271 1000.0
154 M2 fixed 4 text_to_vision 0.0010000000474974513 0.0010000000474974513 0.005150882992893457 1000.0
155 M2 fixed 4 vision_to_text 0.010000000707805157 0.010000000707805157 0.013726433739066124 1000.0
156 M2 fixed 4 audio_to_vision 0.0010000000474974513 0.004000000189989805 0.00889560766518116 1000.0
157 M2 fixed 4 vision_to_audio 0.0020000000949949026 0.01100000087171793 0.011912405490875244 1000.0
158 M3 42 4 text_to_audio 0.02500000037252903 0.036000002175569534 0.038420792669057846 1000.0
159 M3 42 4 audio_to_text 0.00800000037997961 0.023000001907348633 0.02173793874680996 1000.0
160 M3 42 4 text_to_vision 0.0 0.005000000353902578 0.006278287619352341 1000.0
161 M3 42 4 vision_to_text 0.0020000000949949026 0.003000000026077032 0.007351293694227934 1000.0
162 M3 42 4 audio_to_vision 0.004000000189989805 0.01100000087171793 0.010880847461521626 1000.0
163 M3 42 4 vision_to_audio 0.006000000052154064 0.013000000268220901 0.015541859902441502 1000.0
164 M4 42 4 text_to_audio 0.004000000189989805 0.01900000125169754 0.020215420052409172 1000.0
165 M4 42 4 audio_to_text 0.0 0.0 0.004233467858284712 1000.0
166 M4 42 4 text_to_vision 0.0020000000949949026 0.010000000707805157 0.012029506266117096 1000.0
167 M4 42 4 vision_to_text 0.0 0.0 0.002725329017266631 1000.0
168 M4 42 4 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.009229524992406368 1000.0
169 M4 42 4 vision_to_audio 0.003000000026077032 0.015000000596046448 0.017202889546751976 1000.0
170 M3 3407 4 text_to_audio 0.013000000268220901 0.032999999821186066 0.02899651788175106 1000.0
171 M3 3407 4 audio_to_text 0.007000000216066837 0.013000000268220901 0.01741517148911953 1000.0
172 M3 3407 4 text_to_vision 0.003000000026077032 0.006000000052154064 0.009169278666377068 1000.0
173 M3 3407 4 vision_to_text 0.006000000052154064 0.006000000052154064 0.011985262855887413 1000.0
174 M3 3407 4 audio_to_vision 0.0 0.004000000189989805 0.008569329045712948 1000.0
175 M3 3407 4 vision_to_audio 0.007000000216066837 0.010000000707805157 0.01606379821896553 1000.0
176 M4 3407 4 text_to_audio 0.003000000026077032 0.015000000596046448 0.015794366598129272 1000.0
177 M4 3407 4 audio_to_text 0.0 0.0 0.004071671981364489 1000.0
178 M4 3407 4 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011711233295500278 1000.0
179 M4 3407 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.00810206588357687 1000.0
180 M4 3407 4 audio_to_vision 0.0020000000949949026 0.010000000707805157 0.013361790217459202 1000.0
181 M4 3407 4 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012590405531227589 1000.0
182 M3 2026 4 text_to_audio 0.024000000208616257 0.04000000283122063 0.039450231939554214 1000.0
183 M3 2026 4 audio_to_text 0.006000000052154064 0.015000000596046448 0.01841675490140915 1000.0
184 M3 2026 4 text_to_vision 0.01100000087171793 0.01600000075995922 0.01930876635015011 1000.0
185 M3 2026 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.007208418566733599 1000.0
186 M3 2026 4 audio_to_vision 0.00800000037997961 0.017000000923871994 0.01701623573899269 1000.0
187 M3 2026 4 vision_to_audio 0.006000000052154064 0.014000000432133675 0.015153588727116585 1000.0
188 M4 2026 4 text_to_audio 0.0020000000949949026 0.010000000707805157 0.011758673004806042 1000.0
189 M4 2026 4 audio_to_text 0.0010000000474974513 0.005000000353902578 0.008434664458036423 1000.0
190 M4 2026 4 text_to_vision 0.0 0.0 0.0034921157639473677 1000.0
191 M4 2026 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.007859558798372746 1000.0
192 M4 2026 4 audio_to_vision 0.003000000026077032 0.014000000432133675 0.01731833629310131 1000.0
193 M4 2026 4 vision_to_audio 0.003000000026077032 0.015000000596046448 0.016523126512765884 1000.0
194 M1 fixed 5 text_to_audio 0.0 0.007000000216066837 0.008882568217813969 1000.0
195 M1 fixed 5 audio_to_text 0.0010000000474974513 0.003000000026077032 0.008665341883897781 1000.0
196 M1 fixed 5 text_to_vision 0.0 0.005000000353902578 0.00567932054400444 1000.0
197 M1 fixed 5 vision_to_text 0.0020000000949949026 0.009000000543892384 0.00881400890648365 1000.0
198 M1 fixed 5 audio_to_vision 0.0020000000949949026 0.012000000104308128 0.012908315286040306 1000.0
199 M1 fixed 5 vision_to_audio 0.003000000026077032 0.012000000104308128 0.0124351242557168 1000.0
200 M2 fixed 5 text_to_audio 0.0 0.009000000543892384 0.009715890511870384 1000.0
201 M2 fixed 5 audio_to_text 0.0020000000949949026 0.009000000543892384 0.010659169405698776 1000.0
202 M2 fixed 5 text_to_vision 0.0 0.009000000543892384 0.008167226798832417 1000.0
203 M2 fixed 5 vision_to_text 0.006000000052154064 0.009000000543892384 0.013293917290866375 1000.0
204 M2 fixed 5 audio_to_vision 0.0020000000949949026 0.012000000104308128 0.012093348428606987 1000.0
205 M2 fixed 5 vision_to_audio 0.0020000000949949026 0.007000000216066837 0.010753415524959564 1000.0
206 M3 42 5 text_to_audio 0.01600000075995922 0.06400000303983688 0.04854178801178932 1000.0
207 M3 42 5 audio_to_text 0.014000000432133675 0.03700000047683716 0.034194398671388626 1000.0
208 M3 42 5 text_to_vision 0.006000000052154064 0.006000000052154064 0.011698092333972454 1000.0
209 M3 42 5 vision_to_text 0.005000000353902578 0.012000000104308128 0.012927225790917873 1000.0
210 M3 42 5 audio_to_vision 0.013000000268220901 0.026000000536441803 0.02460731379687786 1000.0
211 M3 42 5 vision_to_audio 0.004000000189989805 0.007000000216066837 0.012679936364293098 1000.0
212 M4 42 5 text_to_audio 0.04000000283122063 0.052000001072883606 0.05324704200029373 1000.0
213 M4 42 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.008529611863195896 1000.0
214 M4 42 5 text_to_vision 0.0010000000474974513 0.005000000353902578 0.00876129325479269 1000.0
215 M4 42 5 vision_to_text 0.0020000000949949026 0.010000000707805157 0.011993298307061195 1000.0
216 M4 42 5 audio_to_vision 0.003000000026077032 0.015000000596046448 0.017660561949014664 1000.0
217 M4 42 5 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.014293942600488663 1000.0
218 M3 3407 5 text_to_audio 0.03100000135600567 0.07200000435113907 0.061651285737752914 1000.0
219 M3 3407 5 audio_to_text 0.018000001087784767 0.04000000283122063 0.042421478778123856 1000.0
220 M3 3407 5 text_to_vision 0.009000000543892384 0.021000001579523087 0.020427849143743515 1000.0
221 M3 3407 5 vision_to_text 0.004000000189989805 0.013000000268220901 0.012980340048670769 1000.0
222 M3 3407 5 audio_to_vision 0.007000000216066837 0.018000001087784767 0.021291999146342278 1000.0
223 M3 3407 5 vision_to_audio 0.004000000189989805 0.01600000075995922 0.016669930890202522 1000.0
224 M4 3407 5 text_to_audio 0.0020000000949949026 0.010000000707805157 0.0127327349036932 1000.0
225 M4 3407 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.00789211317896843 1000.0
226 M4 3407 5 text_to_vision 0.0020000000949949026 0.005000000353902578 0.00888831913471222 1000.0
227 M4 3407 5 vision_to_text 0.0010000000474974513 0.005000000353902578 0.008448818698525429 1000.0
228 M4 3407 5 audio_to_vision 0.003000000026077032 0.013000000268220901 0.016633378341794014 1000.0
229 M4 3407 5 vision_to_audio 0.003000000026077032 0.015000000596046448 0.01952446810901165 1000.0
230 M3 2026 5 text_to_audio 0.03100000135600567 0.08500000089406967 0.06357216835021973 1000.0
231 M3 2026 5 audio_to_text 0.005000000353902578 0.032999999821186066 0.030199285596609116 1000.0
232 M3 2026 5 text_to_vision 0.007000000216066837 0.021000001579523087 0.015703614801168442 1000.0
233 M3 2026 5 vision_to_text 0.005000000353902578 0.009000000543892384 0.01244888361543417 1000.0
234 M3 2026 5 audio_to_vision 0.009000000543892384 0.023000001907348633 0.02339860051870346 1000.0
235 M3 2026 5 vision_to_audio 0.004000000189989805 0.015000000596046448 0.01597834937274456 1000.0
236 M4 2026 5 text_to_audio 0.052000001072883606 0.06300000101327896 0.06472568958997726 1000.0
237 M4 2026 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.007973534055054188 1000.0
238 M4 2026 5 text_to_vision 0.0010000000474974513 0.005000000353902578 0.007530428934842348 1000.0
239 M4 2026 5 vision_to_text 0.003000000026077032 0.015000000596046448 0.016602179035544395 1000.0
240 M4 2026 5 audio_to_vision 0.003000000026077032 0.015000000596046448 0.017103707417845726 1000.0
241 M4 2026 5 vision_to_audio 0.003000000026077032 0.014000000432133675 0.01859205961227417 1000.0
@@ -0,0 +1,25 @@
method,direction,n,r_at_1_mean,r_at_1_std,r_at_5_mean,r_at_5_std,mrr_mean,mrr_std
M1,text_to_audio,5,0.0012000000569969416,0.001643167750561775,0.007000000309199094,0.0015811389419980182,0.008951337542384863,0.0015105744211016399
M1,audio_to_text,5,0.0028000001097097994,0.0014832398031884779,0.006000000238418579,0.0021213204717563106,0.009594007954001426,0.0008820339883026645
M1,text_to_vision,5,0.0020000000949949026,0.0020000001240987344,0.007000000402331352,0.0031622779576235397,0.009382870234549045,0.0038973032823256706
M1,vision_to_text,5,0.005000000214204192,0.0038729835451934104,0.007000000402331352,0.004301162959612636,0.010874488297849894,0.0033759486781861315
M1,audio_to_vision,5,0.0010000000474974513,0.0010000000474974513,0.008600000198930502,0.004037325898109636,0.010010761767625808,0.0028143308073864106
M1,vision_to_audio,5,0.002000000048428774,0.0009999999892897908,0.008000000286847353,0.0036742348045336777,0.009903422184288501,0.0023425375089251867
M2,text_to_audio,5,0.0014000000432133675,0.0011401754384133685,0.009200000390410423,0.0022803508359855447,0.010239988192915916,0.0015598764796833854
M2,audio_to_text,5,0.004000000143423677,0.0015811388683705022,0.009200000483542681,0.0030331502218969453,0.011443204618990422,0.0018186032524210385
M2,text_to_vision,5,0.003600000194273889,0.003209361472308511,0.008800000487826764,0.004764451893141854,0.011315117496997118,0.004998444639556679
M2,vision_to_text,5,0.00680000027641654,0.001923538612521058,0.009200000576674939,0.0013038406947566386,0.013902883604168892,0.0007593274807152007
M2,audio_to_vision,5,0.0014000000664964318,0.0005477225835205916,0.006800000183284282,0.003346640098121672,0.009431831166148185,0.0017045066129491178
M2,vision_to_audio,5,0.002000000118277967,0.0018708288289165209,0.00640000025741756,0.0032093615938253205,0.00967191495001316,0.002293345042131113
M3,text_to_audio,15,0.01966666733690848,0.009707631019479971,0.043000001646578315,0.020511321663901527,0.03803744154671828,0.014634778360297347
M3,audio_to_text,15,0.009066667081788183,0.009353889531597692,0.020933334335374334,0.013084706401165278,0.023117945653696854,0.011741696228606028
M3,text_to_vision,15,0.0062666668944681685,0.00413118239876218,0.013800000709791978,0.007466114240091005,0.01513038482517004,0.005506734651604392
M3,vision_to_text,15,0.00506666685299327,0.003634490987677242,0.011266667147477467,0.005573748401698712,0.01275980711604158,0.003981040361844423
M3,audio_to_vision,15,0.006400000179807345,0.0033974781354081164,0.0142666673908631,0.007215723836767552,0.016017937225600083,0.005224730877211012
M3,vision_to_audio,15,0.00426666682275633,0.0024630605290560273,0.00913333367401113,0.005111145810859167,0.012493464909493924,0.0045251231363594895
M4,text_to_audio,15,0.007066666963510215,0.01601546933258453,0.01226666709408164,0.019414526078011158,0.014511969638988376,0.018944261404425476
M4,audio_to_text,15,0.00046666668883214396,0.0005163978040219006,0.002266666820893685,0.00252039318928005,0.005223872559145093,0.0023624842688679767
M4,text_to_vision,15,0.0016000000604738792,0.0011832159931418852,0.00760000046963493,0.005852960507593743,0.0100666221541663,0.005020802122090484
M4,vision_to_text,15,0.0014666667130465308,0.0010600988416483645,0.007333333759258191,0.005300494367735815,0.009886658145114779,0.004472077112598579
M4,audio_to_vision,15,0.002133333380334079,0.0009154754048125681,0.010533333849161863,0.004405624197684311,0.012994256491462389,0.00403471312250859
M4,vision_to_audio,15,0.00206666673378398,0.0007037315398609183,0.010266667200873296,0.003712270612174926,0.012846407542626063,0.003972973272038369
1 method direction n r_at_1_mean r_at_1_std r_at_5_mean r_at_5_std mrr_mean mrr_std
2 M1 text_to_audio 5 0.0012000000569969416 0.001643167750561775 0.007000000309199094 0.0015811389419980182 0.008951337542384863 0.0015105744211016399
3 M1 audio_to_text 5 0.0028000001097097994 0.0014832398031884779 0.006000000238418579 0.0021213204717563106 0.009594007954001426 0.0008820339883026645
4 M1 text_to_vision 5 0.0020000000949949026 0.0020000001240987344 0.007000000402331352 0.0031622779576235397 0.009382870234549045 0.0038973032823256706
5 M1 vision_to_text 5 0.005000000214204192 0.0038729835451934104 0.007000000402331352 0.004301162959612636 0.010874488297849894 0.0033759486781861315
6 M1 audio_to_vision 5 0.0010000000474974513 0.0010000000474974513 0.008600000198930502 0.004037325898109636 0.010010761767625808 0.0028143308073864106
7 M1 vision_to_audio 5 0.002000000048428774 0.0009999999892897908 0.008000000286847353 0.0036742348045336777 0.009903422184288501 0.0023425375089251867
8 M2 text_to_audio 5 0.0014000000432133675 0.0011401754384133685 0.009200000390410423 0.0022803508359855447 0.010239988192915916 0.0015598764796833854
9 M2 audio_to_text 5 0.004000000143423677 0.0015811388683705022 0.009200000483542681 0.0030331502218969453 0.011443204618990422 0.0018186032524210385
10 M2 text_to_vision 5 0.003600000194273889 0.003209361472308511 0.008800000487826764 0.004764451893141854 0.011315117496997118 0.004998444639556679
11 M2 vision_to_text 5 0.00680000027641654 0.001923538612521058 0.009200000576674939 0.0013038406947566386 0.013902883604168892 0.0007593274807152007
12 M2 audio_to_vision 5 0.0014000000664964318 0.0005477225835205916 0.006800000183284282 0.003346640098121672 0.009431831166148185 0.0017045066129491178
13 M2 vision_to_audio 5 0.002000000118277967 0.0018708288289165209 0.00640000025741756 0.0032093615938253205 0.00967191495001316 0.002293345042131113
14 M3 text_to_audio 15 0.01966666733690848 0.009707631019479971 0.043000001646578315 0.020511321663901527 0.03803744154671828 0.014634778360297347
15 M3 audio_to_text 15 0.009066667081788183 0.009353889531597692 0.020933334335374334 0.013084706401165278 0.023117945653696854 0.011741696228606028
16 M3 text_to_vision 15 0.0062666668944681685 0.00413118239876218 0.013800000709791978 0.007466114240091005 0.01513038482517004 0.005506734651604392
17 M3 vision_to_text 15 0.00506666685299327 0.003634490987677242 0.011266667147477467 0.005573748401698712 0.01275980711604158 0.003981040361844423
18 M3 audio_to_vision 15 0.006400000179807345 0.0033974781354081164 0.0142666673908631 0.007215723836767552 0.016017937225600083 0.005224730877211012
19 M3 vision_to_audio 15 0.00426666682275633 0.0024630605290560273 0.00913333367401113 0.005111145810859167 0.012493464909493924 0.0045251231363594895
20 M4 text_to_audio 15 0.007066666963510215 0.01601546933258453 0.01226666709408164 0.019414526078011158 0.014511969638988376 0.018944261404425476
21 M4 audio_to_text 15 0.00046666668883214396 0.0005163978040219006 0.002266666820893685 0.00252039318928005 0.005223872559145093 0.0023624842688679767
22 M4 text_to_vision 15 0.0016000000604738792 0.0011832159931418852 0.00760000046963493 0.005852960507593743 0.0100666221541663 0.005020802122090484
23 M4 vision_to_text 15 0.0014666667130465308 0.0010600988416483645 0.007333333759258191 0.005300494367735815 0.009886658145114779 0.004472077112598579
24 M4 audio_to_vision 15 0.002133333380334079 0.0009154754048125681 0.010533333849161863 0.004405624197684311 0.012994256491462389 0.00403471312250859
25 M4 vision_to_audio 15 0.00206666673378398 0.0007037315398609183 0.010266667200873296 0.003712270612174926 0.012846407542626063 0.003972973272038369
@@ -0,0 +1,39 @@
{
"created_utc": "2026-09-23T09:07:49.626404+00:00",
"sample_count": 100,
"group_count": 37,
"folds": 5,
"seeds": [
42,
3407,
2026
],
"device": "cuda",
"gpu_name": "NVIDIA GeForce RTX 5070 Ti",
"python": "3.14.7",
"torch": "2.14.0+cu130",
"parameters": {
"grid_size": 50,
"hidden_size": 128,
"heads": 4,
"dropout": 0.1,
"batch_size": 8,
"epochs_max": 50,
"early_stopping_patience": 8,
"learning_rate": 0.0001,
"mask_ratio": 0.2,
"retrieval_probe_epochs": 20,
"reconstruction_probe_epochs": 25,
"mvr_epsilon": 0.02
},
"objective": "masked reconstruction + cross-modal contrastive + temporal monotonicity; emotion labels unused",
"split_rule": "GroupKFold by group_id/video_id",
"example_sample_id": "-tPCytz4rww/12",
"elapsed_seconds": 516.2944533824921,
"interpretation_limits": [
"Grid-index retrieval is a representation-consistency probe, not independent temporal ground truth.",
"Masked reconstruction uses a decoder trained on the training fold and reports standardized-feature errors.",
"The emotion probe is a small-sample downstream utility check, not a claim of generalization to MOSEI.",
"No human event timestamps are available, so human IoU/MATE is not reported."
]
}
@@ -0,0 +1,742 @@
[
{
"fold": 1,
"train_sample_ids": [
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-qDkUB0GgYY/6",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-UUCSKoHeMA/0"
],
"train_video_ids": [
"-3nNcZdcdvU",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-3g5yACwYnA",
"-9y-fZ3swSY",
"-I_e4mIh0yE",
"-UUCSKoHeMA",
"-egA8-b7-3M",
"-qDkUB0GgYY"
]
},
{
"fold": 2,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mqbVkbCndg/0",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-HeZS2-Prhc/2"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-mJ2ud6oKI8",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-571d8cVauQ",
"-HeZS2-Prhc",
"-THoVjtIkeU",
"-UuX1xuaiiE",
"-lzEya4AM_4",
"-mqbVkbCndg",
"-wny0OAz3g8"
]
},
{
"fold": 3,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-aqamKhZ1Ec/0",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-AUZQgSxyPQ/2",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-HeZS2-Prhc",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-AUZQgSxyPQ",
"-HwX2H8Z4hY",
"-RfYyzHpjk4",
"-aqamKhZ1Ec",
"-hnBHBN8p5A",
"-s9qJ7ATP7w",
"-vxjVxOeScU",
"-wMB_hJL-3o"
]
},
{
"fold": 4,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-tANM6ETl_M/3",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-aNfi7CP8vM/7",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-571d8cVauQ",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-s9qJ7ATP7w",
"-tANM6ETl_M",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-6rXp3zJ3kc",
"-UacrmKiTn4",
"-aNfi7CP8vM",
"-iRBcNs9oI8",
"-ri04Z7vwnc",
"-t217m2on-s",
"-tPCytz4rww",
"-uywlfIYOS8"
]
},
{
"fold": 5,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8"
],
"validation_sample_ids": [
"-3nNcZdcdvU/5",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-a55Q6RWvTA/3",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-tANM6ETl_M/3",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"train_video_ids": [
"-3g5yACwYnA",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8"
],
"validation_video_ids": [
"-3nNcZdcdvU",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-a55Q6RWvTA",
"-dxfTGcXJoc",
"-mJ2ud6oKI8",
"-tANM6ETl_M",
"-yRb-Jum7EQ"
]
}
]
@@ -0,0 +1,31 @@
method,seed,fold,best_epoch,best_validation_objective,checkpoint
M3,42,1,50,4.937662982940674,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M3.pt
M4,42,1,4,6.076866149902344,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M4.pt
M3,3407,1,50,4.9825927734375,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M3.pt
M4,3407,1,4,5.871845149993897,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M4.pt
M3,2026,1,50,5.102480220794678,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M3.pt
M4,2026,1,3,6.099013900756836,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M4.pt
M3,42,2,1,6.125876426696777,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M3.pt
M4,42,2,1,5.942682838439941,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M4.pt
M3,3407,2,50,5.144535160064697,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M3.pt
M4,3407,2,3,6.067738342285156,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M4.pt
M3,2026,2,50,5.17415599822998,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M3.pt
M4,2026,2,1,5.96431074142456,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M4.pt
M3,42,3,50,4.976429462432861,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M3.pt
M4,42,3,3,6.195817565917968,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M4.pt
M3,3407,3,50,4.659491157531738,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M3.pt
M4,3407,3,9,5.910655784606933,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M4.pt
M3,2026,3,50,4.999933815002441,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M3.pt
M4,2026,3,3,6.024094963073731,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M4.pt
M3,42,4,50,5.152298069000244,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M3.pt
M4,42,4,1,5.998569297790527,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M4.pt
M3,3407,4,50,5.098000621795654,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M3.pt
M4,3407,4,2,6.03455867767334,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M4.pt
M3,2026,4,50,5.260256576538086,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M3.pt
M4,2026,4,2,6.081212425231934,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M4.pt
M3,42,5,50,5.298740291595459,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M3.pt
M4,42,5,4,6.0218939781188965,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M4.pt
M3,3407,5,50,4.953551292419434,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M3.pt
M4,3407,5,8,6.058719825744629,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M4.pt
M3,2026,5,50,5.133688831329346,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M3.pt
M4,2026,5,1,5.882685661315918,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M4.pt
1 method seed fold best_epoch best_validation_objective checkpoint
2 M3 42 1 50 4.937662982940674 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M3.pt
3 M4 42 1 4 6.076866149902344 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M4.pt
4 M3 3407 1 50 4.9825927734375 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M3.pt
5 M4 3407 1 4 5.871845149993897 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M4.pt
6 M3 2026 1 50 5.102480220794678 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M3.pt
7 M4 2026 1 3 6.099013900756836 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M4.pt
8 M3 42 2 1 6.125876426696777 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M3.pt
9 M4 42 2 1 5.942682838439941 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M4.pt
10 M3 3407 2 50 5.144535160064697 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M3.pt
11 M4 3407 2 3 6.067738342285156 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M4.pt
12 M3 2026 2 50 5.17415599822998 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M3.pt
13 M4 2026 2 1 5.96431074142456 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M4.pt
14 M3 42 3 50 4.976429462432861 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M3.pt
15 M4 42 3 3 6.195817565917968 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M4.pt
16 M3 3407 3 50 4.659491157531738 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M3.pt
17 M4 3407 3 9 5.910655784606933 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M4.pt
18 M3 2026 3 50 4.999933815002441 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M3.pt
19 M4 2026 3 3 6.024094963073731 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M4.pt
20 M3 42 4 50 5.152298069000244 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M3.pt
21 M4 42 4 1 5.998569297790527 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M4.pt
22 M3 3407 4 50 5.098000621795654 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M3.pt
23 M4 3407 4 2 6.03455867767334 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M4.pt
24 M3 2026 4 50 5.260256576538086 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M3.pt
25 M4 2026 4 2 6.081212425231934 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M4.pt
26 M3 42 5 50 5.298740291595459 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M3.pt
27 M4 42 5 4 6.0218939781188965 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M4.pt
28 M3 3407 5 50 4.953551292419434 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M3.pt
29 M4 3407 5 8 6.058719825744629 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M4.pt
30 M3 2026 5 50 5.133688831329346 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M3.pt
31 M4 2026 5 1 5.882685661315918 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M4.pt
Binary file not shown.

After

Width:  |  Height:  |  Size: 516 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 282 KiB

@@ -0,0 +1,241 @@
method,seed,fold,direction,r_at_1,r_at_5,mrr,queries
M1,fixed,1,text_to_audio,0.0010000000474974513,0.00800000037997961,0.010086040943861008,1000.0
M1,fixed,1,audio_to_text,0.003000000026077032,0.00800000037997961,0.009918988682329655,1000.0
M1,fixed,1,text_to_vision,0.003000000026077032,0.009000000543892384,0.012364907190203667,1000.0
M1,fixed,1,vision_to_text,0.0,0.0,0.006473104935139418,1000.0
M1,fixed,1,audio_to_vision,0.0,0.00800000037997961,0.009185504168272018,1000.0
M1,fixed,1,vision_to_audio,0.0010000000474974513,0.006000000052154064,0.008802860975265503,1000.0
M2,fixed,1,text_to_audio,0.003000000026077032,0.013000000268220901,0.012891415506601334,1000.0
M2,fixed,1,audio_to_text,0.004000000189989805,0.01100000087171793,0.012235527858138084,1000.0
M2,fixed,1,text_to_vision,0.00800000037997961,0.014000000432133675,0.017583860084414482,1000.0
M2,fixed,1,vision_to_text,0.005000000353902578,0.00800000037997961,0.013148980215191841,1000.0
M2,fixed,1,audio_to_vision,0.0010000000474974513,0.00800000037997961,0.009523994289338589,1000.0
M2,fixed,1,vision_to_audio,0.0010000000474974513,0.006000000052154064,0.009459399618208408,1000.0
M3,42,1,text_to_audio,0.021000001579523087,0.03700000047683716,0.03344433382153511,1000.0
M3,42,1,audio_to_text,0.00800000037997961,0.014000000432133675,0.01682252623140812,1000.0
M3,42,1,text_to_vision,0.006000000052154064,0.027000000700354576,0.018166253343224525,1000.0
M3,42,1,vision_to_text,0.010000000707805157,0.02200000174343586,0.018923873081803322,1000.0
M3,42,1,audio_to_vision,0.003000000026077032,0.003000000026077032,0.010045807808637619,1000.0
M3,42,1,vision_to_audio,0.009000000543892384,0.013000000268220901,0.020716793835163116,1000.0
M4,42,1,text_to_audio,0.0,0.0,0.002745515899732709,1000.0
M4,42,1,audio_to_text,0.0010000000474974513,0.005000000353902578,0.006256598513573408,1000.0
M4,42,1,text_to_vision,0.0010000000474974513,0.005000000353902578,0.007646685931831598,1000.0
M4,42,1,vision_to_text,0.0020000000949949026,0.010000000707805157,0.011485339142382145,1000.0
M4,42,1,audio_to_vision,0.0020000000949949026,0.01100000087171793,0.011041815392673016,1000.0
M4,42,1,vision_to_audio,0.0010000000474974513,0.003000000026077032,0.005618138238787651,1000.0
M3,3407,1,text_to_audio,0.009000000543892384,0.017000000923871994,0.024300672113895416,1000.0
M3,3407,1,audio_to_text,0.005000000353902578,0.00800000037997961,0.016388019546866417,1000.0
M3,3407,1,text_to_vision,0.007000000216066837,0.01100000087171793,0.014694208279252052,1000.0
M3,3407,1,vision_to_text,0.013000000268220901,0.018000001087784767,0.019234079867601395,1000.0
M3,3407,1,audio_to_vision,0.007000000216066837,0.010000000707805157,0.01576545275747776,1000.0
M3,3407,1,vision_to_audio,0.005000000353902578,0.005000000353902578,0.010637721046805382,1000.0
M4,3407,1,text_to_audio,0.0010000000474974513,0.005000000353902578,0.006559826899319887,1000.0
M4,3407,1,audio_to_text,0.0010000000474974513,0.005000000353902578,0.006270123180001974,1000.0
M4,3407,1,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011605769395828247,1000.0
M4,3407,1,vision_to_text,0.003000000026077032,0.015000000596046448,0.016500672325491905,1000.0
M4,3407,1,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.00809124018996954,1000.0
M4,3407,1,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012351044453680515,1000.0
M3,2026,1,text_to_audio,0.017000000923871994,0.02500000037252903,0.026665309444069862,1000.0
M3,2026,1,audio_to_text,0.0010000000474974513,0.01100000087171793,0.011807681992650032,1000.0
M3,2026,1,text_to_vision,0.006000000052154064,0.012000000104308128,0.01321154274046421,1000.0
M3,2026,1,vision_to_text,0.0,0.006000000052154064,0.007229108829051256,1000.0
M3,2026,1,audio_to_vision,0.004000000189989805,0.010000000707805157,0.012219780124723911,1000.0
M3,2026,1,vision_to_audio,0.0,0.0,0.00736302649602294,1000.0
M4,2026,1,text_to_audio,0.0,0.0,0.00321578630246222,1000.0
M4,2026,1,audio_to_text,0.0,0.0,0.0029425874818116426,1000.0
M4,2026,1,text_to_vision,0.0010000000474974513,0.005000000353902578,0.00850450899451971,1000.0
M4,2026,1,vision_to_text,0.0020000000949949026,0.010000000707805157,0.01312504056841135,1000.0
M4,2026,1,audio_to_vision,0.0020000000949949026,0.010000000707805157,0.01287905778735876,1000.0
M4,2026,1,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012796711176633835,1000.0
M1,fixed,2,text_to_audio,0.004000000189989805,0.009000000543892384,0.010721729137003422,1000.0
M1,fixed,2,audio_to_text,0.0020000000949949026,0.005000000353902578,0.008698307909071445,1000.0
M1,fixed,2,text_to_vision,0.0010000000474974513,0.003000000026077032,0.007055815774947405,1000.0
M1,fixed,2,vision_to_text,0.006000000052154064,0.006000000052154064,0.011396339163184166,1000.0
M1,fixed,2,audio_to_vision,0.0010000000474974513,0.003000000026077032,0.006511436775326729,1000.0
M1,fixed,2,vision_to_audio,0.003000000026077032,0.00800000037997961,0.008912489749491215,1000.0
M2,fixed,2,text_to_audio,0.0020000000949949026,0.007000000216066837,0.008748839609324932,1000.0
M2,fixed,2,audio_to_text,0.003000000026077032,0.005000000353902578,0.008852123282849789,1000.0
M2,fixed,2,text_to_vision,0.004000000189989805,0.010000000707805157,0.01082402840256691,1000.0
M2,fixed,2,vision_to_text,0.007000000216066837,0.01100000087171793,0.014397574588656425,1000.0
M2,fixed,2,audio_to_vision,0.0010000000474974513,0.004000000189989805,0.007378880865871906,1000.0
M2,fixed,2,vision_to_audio,0.005000000353902578,0.006000000052154064,0.010349743068218231,1000.0
M3,42,2,text_to_audio,0.0010000000474974513,0.009000000543892384,0.008156482130289078,1000.0
M3,42,2,audio_to_text,0.0,0.0010000000474974513,0.004495462402701378,1000.0
M3,42,2,text_to_vision,0.004000000189989805,0.01100000087171793,0.012809447944164276,1000.0
M3,42,2,vision_to_text,0.00800000037997961,0.009000000543892384,0.013898583129048347,1000.0
M3,42,2,audio_to_vision,0.003000000026077032,0.00800000037997961,0.009653640910983086,1000.0
M3,42,2,vision_to_audio,0.0,0.0010000000474974513,0.0033326284028589725,1000.0
M4,42,2,text_to_audio,0.0,0.0,0.002928863512352109,1000.0
M4,42,2,audio_to_text,0.0,0.0,0.0023521804250776768,1000.0
M4,42,2,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011110157705843449,1000.0
M4,42,2,vision_to_text,0.0010000000474974513,0.005000000353902578,0.006895906291902065,1000.0
M4,42,2,audio_to_vision,0.003000000026077032,0.015000000596046448,0.015711786225438118,1000.0
M4,42,2,vision_to_audio,0.0020000000949949026,0.012000000104308128,0.012692151591181755,1000.0
M3,3407,2,text_to_audio,0.018000001087784767,0.05300000309944153,0.041355911642313004,1000.0
M3,3407,2,audio_to_text,0.007000000216066837,0.03100000135600567,0.023398615419864655,1000.0
M3,3407,2,text_to_vision,0.0010000000474974513,0.005000000353902578,0.009130735881626606,1000.0
M3,3407,2,vision_to_text,0.0,0.006000000052154064,0.008097999729216099,1000.0
M3,3407,2,audio_to_vision,0.006000000052154064,0.010000000707805157,0.013130152598023415,1000.0
M3,3407,2,vision_to_audio,0.0020000000949949026,0.009000000543892384,0.008952965028584003,1000.0
M4,3407,2,text_to_audio,0.0,0.0,0.003208547830581665,1000.0
M4,3407,2,audio_to_text,0.0,0.0,0.0023785873781889677,1000.0
M4,3407,2,text_to_vision,0.0,0.0,0.0033639008179306984,1000.0
M4,3407,2,vision_to_text,0.0020000000949949026,0.010000000707805157,0.01099696196615696,1000.0
M4,3407,2,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.006608974188566208,1000.0
M4,3407,2,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011376568116247654,1000.0
M3,2026,2,text_to_audio,0.024000000208616257,0.04800000041723251,0.04059242084622383,1000.0
M3,2026,2,audio_to_text,0.0010000000474974513,0.013000000268220901,0.014835278503596783,1000.0
M3,2026,2,text_to_vision,0.014000000432133675,0.02200000174343586,0.02588539943099022,1000.0
M3,2026,2,vision_to_text,0.007000000216066837,0.018000001087784767,0.016461974009871483,1000.0
M3,2026,2,audio_to_vision,0.012000000104308128,0.012000000104308128,0.018243461847305298,1000.0
M3,2026,2,vision_to_audio,0.006000000052154064,0.015000000596046448,0.015246634371578693,1000.0
M4,2026,2,text_to_audio,0.0010000000474974513,0.005000000353902578,0.006106700282543898,1000.0
M4,2026,2,audio_to_text,0.0,0.0,0.0024518424179404974,1000.0
M4,2026,2,text_to_vision,0.0,0.0,0.003404181217774749,1000.0
M4,2026,2,vision_to_text,0.0,0.0,0.004238351248204708,1000.0
M4,2026,2,audio_to_vision,0.003000000026077032,0.015000000596046448,0.016335317865014076,1000.0
M4,2026,2,vision_to_audio,0.0010000000474974513,0.005000000353902578,0.007721395697444677,1000.0
M1,fixed,3,text_to_audio,0.0010000000474974513,0.005000000353902578,0.008115584030747414,1000.0
M1,fixed,3,audio_to_text,0.003000000026077032,0.006000000052154064,0.010013893246650696,1000.0
M1,fixed,3,text_to_vision,0.005000000353902578,0.01100000087171793,0.014654486440122128,1000.0
M1,fixed,3,vision_to_text,0.00800000037997961,0.01100000087171793,0.015268602408468723,1000.0
M1,fixed,3,audio_to_vision,0.0,0.007000000216066837,0.0085630863904953,1000.0
M1,fixed,3,vision_to_audio,0.0010000000474974513,0.003000000026077032,0.007109377533197403,1000.0
M2,fixed,3,text_to_audio,0.0010000000474974513,0.00800000037997961,0.009955259971320629,1000.0
M2,fixed,3,audio_to_text,0.006000000052154064,0.013000000268220901,0.013718895614147186,1000.0
M2,fixed,3,text_to_vision,0.005000000353902578,0.010000000707805157,0.014849589206278324,1000.0
M2,fixed,3,vision_to_text,0.006000000052154064,0.00800000037997961,0.014947512187063694,1000.0
M2,fixed,3,audio_to_vision,0.0020000000949949026,0.006000000052154064,0.009267324581742287,1000.0
M2,fixed,3,vision_to_audio,0.0,0.0020000000949949026,0.005884611047804356,1000.0
M3,42,3,text_to_audio,0.014000000432133675,0.0430000014603138,0.03258649632334709,1000.0
M3,42,3,audio_to_text,0.006000000052154064,0.010000000707805157,0.01843179017305374,1000.0
M3,42,3,text_to_vision,0.013000000268220901,0.024000000208616257,0.023365069180727005,1000.0
M3,42,3,vision_to_text,0.006000000052154064,0.015000000596046448,0.015362463891506195,1000.0
M3,42,3,audio_to_vision,0.007000000216066837,0.01900000125169754,0.016374722123146057,1000.0
M3,42,3,vision_to_audio,0.003000000026077032,0.006000000052154064,0.008777403272688389,1000.0
M4,42,3,text_to_audio,0.0,0.0,0.002880551852285862,1000.0
M4,42,3,audio_to_text,0.0,0.0,0.003347013145685196,1000.0
M4,42,3,text_to_vision,0.004000000189989805,0.020000001415610313,0.021214386448264122,1000.0
M4,42,3,vision_to_text,0.003000000026077032,0.015000000596046448,0.016206368803977966,1000.0
M4,42,3,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.00828514900058508,1000.0
M4,42,3,vision_to_audio,0.0010000000474974513,0.005000000353902578,0.007783014792948961,1000.0
M3,3407,3,text_to_audio,0.039000000804662704,0.055000003427267075,0.05350850895047188,1000.0
M3,3407,3,audio_to_text,0.03800000250339508,0.04400000348687172,0.04879831150174141,1000.0
M3,3407,3,text_to_vision,0.0020000000949949026,0.009000000543892384,0.011307715438306332,1000.0
M3,3407,3,vision_to_text,0.003000000026077032,0.013000000268220901,0.011990648694336414,1000.0
M3,3407,3,audio_to_vision,0.007000000216066837,0.02500000037252903,0.022219380363821983,1000.0
M3,3407,3,vision_to_audio,0.003000000026077032,0.006000000052154064,0.009232758544385433,1000.0
M4,3407,3,text_to_audio,0.0,0.0,0.0037817470729351044,1000.0
M4,3407,3,audio_to_text,0.0,0.0,0.004079542122781277,1000.0
M4,3407,3,text_to_vision,0.003000000026077032,0.014000000432133675,0.015179671347141266,1000.0
M4,3407,3,vision_to_text,0.0,0.0,0.004454099107533693,1000.0
M4,3407,3,audio_to_vision,0.003000000026077032,0.015000000596046448,0.016372714191675186,1000.0
M4,3407,3,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011763687245547771,1000.0
M3,2026,3,text_to_audio,0.012000000104308128,0.02800000086426735,0.029318703338503838,1000.0
M3,2026,3,audio_to_text,0.012000000104308128,0.021000001579523087,0.027406470850110054,1000.0
M3,2026,3,text_to_vision,0.005000000353902578,0.01100000087171793,0.01579951122403145,1000.0
M3,2026,3,vision_to_text,0.006000000052154064,0.014000000432133675,0.015296950936317444,1000.0
M3,2026,3,audio_to_vision,0.006000000052154064,0.018000001087784767,0.016852334141731262,1000.0
M3,2026,3,vision_to_audio,0.005000000353902578,0.007000000216066837,0.01105457916855812,1000.0
M4,2026,3,text_to_audio,0.0010000000474974513,0.005000000353902578,0.007778078783303499,1000.0
M4,2026,3,audio_to_text,0.0010000000474974513,0.004000000189989805,0.007144550327211618,1000.0
M4,2026,3,text_to_vision,0.003000000026077032,0.015000000596046448,0.016557173803448677,1000.0
M4,2026,3,vision_to_text,0.0010000000474974513,0.005000000353902578,0.008665882982313633,1000.0
M4,2026,3,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.008280493319034576,1000.0
M4,2026,3,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.011866509914398193,1000.0
M1,fixed,4,text_to_audio,0.0,0.006000000052154064,0.006950765382498503,1000.0
M1,fixed,4,audio_to_text,0.005000000353902578,0.00800000037997961,0.010673508048057556,1000.0
M1,fixed,4,text_to_vision,0.0010000000474974513,0.007000000216066837,0.007159821223467588,1000.0
M1,fixed,4,vision_to_text,0.009000000543892384,0.009000000543892384,0.01242038607597351,1000.0
M1,fixed,4,audio_to_vision,0.0020000000949949026,0.013000000268220901,0.01288546621799469,1000.0
M1,fixed,4,vision_to_audio,0.0020000000949949026,0.01100000087171793,0.012257258407771587,1000.0
M2,fixed,4,text_to_audio,0.0010000000474974513,0.009000000543892384,0.009888535365462303,1000.0
M2,fixed,4,audio_to_text,0.005000000353902578,0.00800000037997961,0.011750306934118271,1000.0
M2,fixed,4,text_to_vision,0.0010000000474974513,0.0010000000474974513,0.005150882992893457,1000.0
M2,fixed,4,vision_to_text,0.010000000707805157,0.010000000707805157,0.013726433739066124,1000.0
M2,fixed,4,audio_to_vision,0.0010000000474974513,0.004000000189989805,0.00889560766518116,1000.0
M2,fixed,4,vision_to_audio,0.0020000000949949026,0.01100000087171793,0.011912405490875244,1000.0
M3,42,4,text_to_audio,0.02500000037252903,0.036000002175569534,0.038420792669057846,1000.0
M3,42,4,audio_to_text,0.00800000037997961,0.023000001907348633,0.02173793874680996,1000.0
M3,42,4,text_to_vision,0.0,0.005000000353902578,0.006278287619352341,1000.0
M3,42,4,vision_to_text,0.0020000000949949026,0.003000000026077032,0.007351293694227934,1000.0
M3,42,4,audio_to_vision,0.004000000189989805,0.01100000087171793,0.010880847461521626,1000.0
M3,42,4,vision_to_audio,0.006000000052154064,0.013000000268220901,0.015541859902441502,1000.0
M4,42,4,text_to_audio,0.004000000189989805,0.01900000125169754,0.020215420052409172,1000.0
M4,42,4,audio_to_text,0.0,0.0,0.004233467858284712,1000.0
M4,42,4,text_to_vision,0.0020000000949949026,0.010000000707805157,0.012029506266117096,1000.0
M4,42,4,vision_to_text,0.0,0.0,0.002725329017266631,1000.0
M4,42,4,audio_to_vision,0.0010000000474974513,0.005000000353902578,0.009229524992406368,1000.0
M4,42,4,vision_to_audio,0.003000000026077032,0.015000000596046448,0.017202889546751976,1000.0
M3,3407,4,text_to_audio,0.013000000268220901,0.032999999821186066,0.02899651788175106,1000.0
M3,3407,4,audio_to_text,0.007000000216066837,0.013000000268220901,0.01741517148911953,1000.0
M3,3407,4,text_to_vision,0.003000000026077032,0.006000000052154064,0.009169278666377068,1000.0
M3,3407,4,vision_to_text,0.006000000052154064,0.006000000052154064,0.011985262855887413,1000.0
M3,3407,4,audio_to_vision,0.0,0.004000000189989805,0.008569329045712948,1000.0
M3,3407,4,vision_to_audio,0.007000000216066837,0.010000000707805157,0.01606379821896553,1000.0
M4,3407,4,text_to_audio,0.003000000026077032,0.015000000596046448,0.015794366598129272,1000.0
M4,3407,4,audio_to_text,0.0,0.0,0.004071671981364489,1000.0
M4,3407,4,text_to_vision,0.0020000000949949026,0.010000000707805157,0.011711233295500278,1000.0
M4,3407,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.00810206588357687,1000.0
M4,3407,4,audio_to_vision,0.0020000000949949026,0.010000000707805157,0.013361790217459202,1000.0
M4,3407,4,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.012590405531227589,1000.0
M3,2026,4,text_to_audio,0.024000000208616257,0.04000000283122063,0.039450231939554214,1000.0
M3,2026,4,audio_to_text,0.006000000052154064,0.015000000596046448,0.01841675490140915,1000.0
M3,2026,4,text_to_vision,0.01100000087171793,0.01600000075995922,0.01930876635015011,1000.0
M3,2026,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.007208418566733599,1000.0
M3,2026,4,audio_to_vision,0.00800000037997961,0.017000000923871994,0.01701623573899269,1000.0
M3,2026,4,vision_to_audio,0.006000000052154064,0.014000000432133675,0.015153588727116585,1000.0
M4,2026,4,text_to_audio,0.0020000000949949026,0.010000000707805157,0.011758673004806042,1000.0
M4,2026,4,audio_to_text,0.0010000000474974513,0.005000000353902578,0.008434664458036423,1000.0
M4,2026,4,text_to_vision,0.0,0.0,0.0034921157639473677,1000.0
M4,2026,4,vision_to_text,0.0010000000474974513,0.005000000353902578,0.007859558798372746,1000.0
M4,2026,4,audio_to_vision,0.003000000026077032,0.014000000432133675,0.01731833629310131,1000.0
M4,2026,4,vision_to_audio,0.003000000026077032,0.015000000596046448,0.016523126512765884,1000.0
M1,fixed,5,text_to_audio,0.0,0.007000000216066837,0.008882568217813969,1000.0
M1,fixed,5,audio_to_text,0.0010000000474974513,0.003000000026077032,0.008665341883897781,1000.0
M1,fixed,5,text_to_vision,0.0,0.005000000353902578,0.00567932054400444,1000.0
M1,fixed,5,vision_to_text,0.0020000000949949026,0.009000000543892384,0.00881400890648365,1000.0
M1,fixed,5,audio_to_vision,0.0020000000949949026,0.012000000104308128,0.012908315286040306,1000.0
M1,fixed,5,vision_to_audio,0.003000000026077032,0.012000000104308128,0.0124351242557168,1000.0
M2,fixed,5,text_to_audio,0.0,0.009000000543892384,0.009715890511870384,1000.0
M2,fixed,5,audio_to_text,0.0020000000949949026,0.009000000543892384,0.010659169405698776,1000.0
M2,fixed,5,text_to_vision,0.0,0.009000000543892384,0.008167226798832417,1000.0
M2,fixed,5,vision_to_text,0.006000000052154064,0.009000000543892384,0.013293917290866375,1000.0
M2,fixed,5,audio_to_vision,0.0020000000949949026,0.012000000104308128,0.012093348428606987,1000.0
M2,fixed,5,vision_to_audio,0.0020000000949949026,0.007000000216066837,0.010753415524959564,1000.0
M3,42,5,text_to_audio,0.01600000075995922,0.06400000303983688,0.04854178801178932,1000.0
M3,42,5,audio_to_text,0.014000000432133675,0.03700000047683716,0.034194398671388626,1000.0
M3,42,5,text_to_vision,0.006000000052154064,0.006000000052154064,0.011698092333972454,1000.0
M3,42,5,vision_to_text,0.005000000353902578,0.012000000104308128,0.012927225790917873,1000.0
M3,42,5,audio_to_vision,0.013000000268220901,0.026000000536441803,0.02460731379687786,1000.0
M3,42,5,vision_to_audio,0.004000000189989805,0.007000000216066837,0.012679936364293098,1000.0
M4,42,5,text_to_audio,0.04000000283122063,0.052000001072883606,0.05324704200029373,1000.0
M4,42,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.008529611863195896,1000.0
M4,42,5,text_to_vision,0.0010000000474974513,0.005000000353902578,0.00876129325479269,1000.0
M4,42,5,vision_to_text,0.0020000000949949026,0.010000000707805157,0.011993298307061195,1000.0
M4,42,5,audio_to_vision,0.003000000026077032,0.015000000596046448,0.017660561949014664,1000.0
M4,42,5,vision_to_audio,0.0020000000949949026,0.010000000707805157,0.014293942600488663,1000.0
M3,3407,5,text_to_audio,0.03100000135600567,0.07200000435113907,0.061651285737752914,1000.0
M3,3407,5,audio_to_text,0.018000001087784767,0.04000000283122063,0.042421478778123856,1000.0
M3,3407,5,text_to_vision,0.009000000543892384,0.021000001579523087,0.020427849143743515,1000.0
M3,3407,5,vision_to_text,0.004000000189989805,0.013000000268220901,0.012980340048670769,1000.0
M3,3407,5,audio_to_vision,0.007000000216066837,0.018000001087784767,0.021291999146342278,1000.0
M3,3407,5,vision_to_audio,0.004000000189989805,0.01600000075995922,0.016669930890202522,1000.0
M4,3407,5,text_to_audio,0.0020000000949949026,0.010000000707805157,0.0127327349036932,1000.0
M4,3407,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.00789211317896843,1000.0
M4,3407,5,text_to_vision,0.0020000000949949026,0.005000000353902578,0.00888831913471222,1000.0
M4,3407,5,vision_to_text,0.0010000000474974513,0.005000000353902578,0.008448818698525429,1000.0
M4,3407,5,audio_to_vision,0.003000000026077032,0.013000000268220901,0.016633378341794014,1000.0
M4,3407,5,vision_to_audio,0.003000000026077032,0.015000000596046448,0.01952446810901165,1000.0
M3,2026,5,text_to_audio,0.03100000135600567,0.08500000089406967,0.06357216835021973,1000.0
M3,2026,5,audio_to_text,0.005000000353902578,0.032999999821186066,0.030199285596609116,1000.0
M3,2026,5,text_to_vision,0.007000000216066837,0.021000001579523087,0.015703614801168442,1000.0
M3,2026,5,vision_to_text,0.005000000353902578,0.009000000543892384,0.01244888361543417,1000.0
M3,2026,5,audio_to_vision,0.009000000543892384,0.023000001907348633,0.02339860051870346,1000.0
M3,2026,5,vision_to_audio,0.004000000189989805,0.015000000596046448,0.01597834937274456,1000.0
M4,2026,5,text_to_audio,0.052000001072883606,0.06300000101327896,0.06472568958997726,1000.0
M4,2026,5,audio_to_text,0.0010000000474974513,0.005000000353902578,0.007973534055054188,1000.0
M4,2026,5,text_to_vision,0.0010000000474974513,0.005000000353902578,0.007530428934842348,1000.0
M4,2026,5,vision_to_text,0.003000000026077032,0.015000000596046448,0.016602179035544395,1000.0
M4,2026,5,audio_to_vision,0.003000000026077032,0.015000000596046448,0.017103707417845726,1000.0
M4,2026,5,vision_to_audio,0.003000000026077032,0.014000000432133675,0.01859205961227417,1000.0
1 method seed fold direction r_at_1 r_at_5 mrr queries
2 M1 fixed 1 text_to_audio 0.0010000000474974513 0.00800000037997961 0.010086040943861008 1000.0
3 M1 fixed 1 audio_to_text 0.003000000026077032 0.00800000037997961 0.009918988682329655 1000.0
4 M1 fixed 1 text_to_vision 0.003000000026077032 0.009000000543892384 0.012364907190203667 1000.0
5 M1 fixed 1 vision_to_text 0.0 0.0 0.006473104935139418 1000.0
6 M1 fixed 1 audio_to_vision 0.0 0.00800000037997961 0.009185504168272018 1000.0
7 M1 fixed 1 vision_to_audio 0.0010000000474974513 0.006000000052154064 0.008802860975265503 1000.0
8 M2 fixed 1 text_to_audio 0.003000000026077032 0.013000000268220901 0.012891415506601334 1000.0
9 M2 fixed 1 audio_to_text 0.004000000189989805 0.01100000087171793 0.012235527858138084 1000.0
10 M2 fixed 1 text_to_vision 0.00800000037997961 0.014000000432133675 0.017583860084414482 1000.0
11 M2 fixed 1 vision_to_text 0.005000000353902578 0.00800000037997961 0.013148980215191841 1000.0
12 M2 fixed 1 audio_to_vision 0.0010000000474974513 0.00800000037997961 0.009523994289338589 1000.0
13 M2 fixed 1 vision_to_audio 0.0010000000474974513 0.006000000052154064 0.009459399618208408 1000.0
14 M3 42 1 text_to_audio 0.021000001579523087 0.03700000047683716 0.03344433382153511 1000.0
15 M3 42 1 audio_to_text 0.00800000037997961 0.014000000432133675 0.01682252623140812 1000.0
16 M3 42 1 text_to_vision 0.006000000052154064 0.027000000700354576 0.018166253343224525 1000.0
17 M3 42 1 vision_to_text 0.010000000707805157 0.02200000174343586 0.018923873081803322 1000.0
18 M3 42 1 audio_to_vision 0.003000000026077032 0.003000000026077032 0.010045807808637619 1000.0
19 M3 42 1 vision_to_audio 0.009000000543892384 0.013000000268220901 0.020716793835163116 1000.0
20 M4 42 1 text_to_audio 0.0 0.0 0.002745515899732709 1000.0
21 M4 42 1 audio_to_text 0.0010000000474974513 0.005000000353902578 0.006256598513573408 1000.0
22 M4 42 1 text_to_vision 0.0010000000474974513 0.005000000353902578 0.007646685931831598 1000.0
23 M4 42 1 vision_to_text 0.0020000000949949026 0.010000000707805157 0.011485339142382145 1000.0
24 M4 42 1 audio_to_vision 0.0020000000949949026 0.01100000087171793 0.011041815392673016 1000.0
25 M4 42 1 vision_to_audio 0.0010000000474974513 0.003000000026077032 0.005618138238787651 1000.0
26 M3 3407 1 text_to_audio 0.009000000543892384 0.017000000923871994 0.024300672113895416 1000.0
27 M3 3407 1 audio_to_text 0.005000000353902578 0.00800000037997961 0.016388019546866417 1000.0
28 M3 3407 1 text_to_vision 0.007000000216066837 0.01100000087171793 0.014694208279252052 1000.0
29 M3 3407 1 vision_to_text 0.013000000268220901 0.018000001087784767 0.019234079867601395 1000.0
30 M3 3407 1 audio_to_vision 0.007000000216066837 0.010000000707805157 0.01576545275747776 1000.0
31 M3 3407 1 vision_to_audio 0.005000000353902578 0.005000000353902578 0.010637721046805382 1000.0
32 M4 3407 1 text_to_audio 0.0010000000474974513 0.005000000353902578 0.006559826899319887 1000.0
33 M4 3407 1 audio_to_text 0.0010000000474974513 0.005000000353902578 0.006270123180001974 1000.0
34 M4 3407 1 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011605769395828247 1000.0
35 M4 3407 1 vision_to_text 0.003000000026077032 0.015000000596046448 0.016500672325491905 1000.0
36 M4 3407 1 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.00809124018996954 1000.0
37 M4 3407 1 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012351044453680515 1000.0
38 M3 2026 1 text_to_audio 0.017000000923871994 0.02500000037252903 0.026665309444069862 1000.0
39 M3 2026 1 audio_to_text 0.0010000000474974513 0.01100000087171793 0.011807681992650032 1000.0
40 M3 2026 1 text_to_vision 0.006000000052154064 0.012000000104308128 0.01321154274046421 1000.0
41 M3 2026 1 vision_to_text 0.0 0.006000000052154064 0.007229108829051256 1000.0
42 M3 2026 1 audio_to_vision 0.004000000189989805 0.010000000707805157 0.012219780124723911 1000.0
43 M3 2026 1 vision_to_audio 0.0 0.0 0.00736302649602294 1000.0
44 M4 2026 1 text_to_audio 0.0 0.0 0.00321578630246222 1000.0
45 M4 2026 1 audio_to_text 0.0 0.0 0.0029425874818116426 1000.0
46 M4 2026 1 text_to_vision 0.0010000000474974513 0.005000000353902578 0.00850450899451971 1000.0
47 M4 2026 1 vision_to_text 0.0020000000949949026 0.010000000707805157 0.01312504056841135 1000.0
48 M4 2026 1 audio_to_vision 0.0020000000949949026 0.010000000707805157 0.01287905778735876 1000.0
49 M4 2026 1 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012796711176633835 1000.0
50 M1 fixed 2 text_to_audio 0.004000000189989805 0.009000000543892384 0.010721729137003422 1000.0
51 M1 fixed 2 audio_to_text 0.0020000000949949026 0.005000000353902578 0.008698307909071445 1000.0
52 M1 fixed 2 text_to_vision 0.0010000000474974513 0.003000000026077032 0.007055815774947405 1000.0
53 M1 fixed 2 vision_to_text 0.006000000052154064 0.006000000052154064 0.011396339163184166 1000.0
54 M1 fixed 2 audio_to_vision 0.0010000000474974513 0.003000000026077032 0.006511436775326729 1000.0
55 M1 fixed 2 vision_to_audio 0.003000000026077032 0.00800000037997961 0.008912489749491215 1000.0
56 M2 fixed 2 text_to_audio 0.0020000000949949026 0.007000000216066837 0.008748839609324932 1000.0
57 M2 fixed 2 audio_to_text 0.003000000026077032 0.005000000353902578 0.008852123282849789 1000.0
58 M2 fixed 2 text_to_vision 0.004000000189989805 0.010000000707805157 0.01082402840256691 1000.0
59 M2 fixed 2 vision_to_text 0.007000000216066837 0.01100000087171793 0.014397574588656425 1000.0
60 M2 fixed 2 audio_to_vision 0.0010000000474974513 0.004000000189989805 0.007378880865871906 1000.0
61 M2 fixed 2 vision_to_audio 0.005000000353902578 0.006000000052154064 0.010349743068218231 1000.0
62 M3 42 2 text_to_audio 0.0010000000474974513 0.009000000543892384 0.008156482130289078 1000.0
63 M3 42 2 audio_to_text 0.0 0.0010000000474974513 0.004495462402701378 1000.0
64 M3 42 2 text_to_vision 0.004000000189989805 0.01100000087171793 0.012809447944164276 1000.0
65 M3 42 2 vision_to_text 0.00800000037997961 0.009000000543892384 0.013898583129048347 1000.0
66 M3 42 2 audio_to_vision 0.003000000026077032 0.00800000037997961 0.009653640910983086 1000.0
67 M3 42 2 vision_to_audio 0.0 0.0010000000474974513 0.0033326284028589725 1000.0
68 M4 42 2 text_to_audio 0.0 0.0 0.002928863512352109 1000.0
69 M4 42 2 audio_to_text 0.0 0.0 0.0023521804250776768 1000.0
70 M4 42 2 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011110157705843449 1000.0
71 M4 42 2 vision_to_text 0.0010000000474974513 0.005000000353902578 0.006895906291902065 1000.0
72 M4 42 2 audio_to_vision 0.003000000026077032 0.015000000596046448 0.015711786225438118 1000.0
73 M4 42 2 vision_to_audio 0.0020000000949949026 0.012000000104308128 0.012692151591181755 1000.0
74 M3 3407 2 text_to_audio 0.018000001087784767 0.05300000309944153 0.041355911642313004 1000.0
75 M3 3407 2 audio_to_text 0.007000000216066837 0.03100000135600567 0.023398615419864655 1000.0
76 M3 3407 2 text_to_vision 0.0010000000474974513 0.005000000353902578 0.009130735881626606 1000.0
77 M3 3407 2 vision_to_text 0.0 0.006000000052154064 0.008097999729216099 1000.0
78 M3 3407 2 audio_to_vision 0.006000000052154064 0.010000000707805157 0.013130152598023415 1000.0
79 M3 3407 2 vision_to_audio 0.0020000000949949026 0.009000000543892384 0.008952965028584003 1000.0
80 M4 3407 2 text_to_audio 0.0 0.0 0.003208547830581665 1000.0
81 M4 3407 2 audio_to_text 0.0 0.0 0.0023785873781889677 1000.0
82 M4 3407 2 text_to_vision 0.0 0.0 0.0033639008179306984 1000.0
83 M4 3407 2 vision_to_text 0.0020000000949949026 0.010000000707805157 0.01099696196615696 1000.0
84 M4 3407 2 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.006608974188566208 1000.0
85 M4 3407 2 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011376568116247654 1000.0
86 M3 2026 2 text_to_audio 0.024000000208616257 0.04800000041723251 0.04059242084622383 1000.0
87 M3 2026 2 audio_to_text 0.0010000000474974513 0.013000000268220901 0.014835278503596783 1000.0
88 M3 2026 2 text_to_vision 0.014000000432133675 0.02200000174343586 0.02588539943099022 1000.0
89 M3 2026 2 vision_to_text 0.007000000216066837 0.018000001087784767 0.016461974009871483 1000.0
90 M3 2026 2 audio_to_vision 0.012000000104308128 0.012000000104308128 0.018243461847305298 1000.0
91 M3 2026 2 vision_to_audio 0.006000000052154064 0.015000000596046448 0.015246634371578693 1000.0
92 M4 2026 2 text_to_audio 0.0010000000474974513 0.005000000353902578 0.006106700282543898 1000.0
93 M4 2026 2 audio_to_text 0.0 0.0 0.0024518424179404974 1000.0
94 M4 2026 2 text_to_vision 0.0 0.0 0.003404181217774749 1000.0
95 M4 2026 2 vision_to_text 0.0 0.0 0.004238351248204708 1000.0
96 M4 2026 2 audio_to_vision 0.003000000026077032 0.015000000596046448 0.016335317865014076 1000.0
97 M4 2026 2 vision_to_audio 0.0010000000474974513 0.005000000353902578 0.007721395697444677 1000.0
98 M1 fixed 3 text_to_audio 0.0010000000474974513 0.005000000353902578 0.008115584030747414 1000.0
99 M1 fixed 3 audio_to_text 0.003000000026077032 0.006000000052154064 0.010013893246650696 1000.0
100 M1 fixed 3 text_to_vision 0.005000000353902578 0.01100000087171793 0.014654486440122128 1000.0
101 M1 fixed 3 vision_to_text 0.00800000037997961 0.01100000087171793 0.015268602408468723 1000.0
102 M1 fixed 3 audio_to_vision 0.0 0.007000000216066837 0.0085630863904953 1000.0
103 M1 fixed 3 vision_to_audio 0.0010000000474974513 0.003000000026077032 0.007109377533197403 1000.0
104 M2 fixed 3 text_to_audio 0.0010000000474974513 0.00800000037997961 0.009955259971320629 1000.0
105 M2 fixed 3 audio_to_text 0.006000000052154064 0.013000000268220901 0.013718895614147186 1000.0
106 M2 fixed 3 text_to_vision 0.005000000353902578 0.010000000707805157 0.014849589206278324 1000.0
107 M2 fixed 3 vision_to_text 0.006000000052154064 0.00800000037997961 0.014947512187063694 1000.0
108 M2 fixed 3 audio_to_vision 0.0020000000949949026 0.006000000052154064 0.009267324581742287 1000.0
109 M2 fixed 3 vision_to_audio 0.0 0.0020000000949949026 0.005884611047804356 1000.0
110 M3 42 3 text_to_audio 0.014000000432133675 0.0430000014603138 0.03258649632334709 1000.0
111 M3 42 3 audio_to_text 0.006000000052154064 0.010000000707805157 0.01843179017305374 1000.0
112 M3 42 3 text_to_vision 0.013000000268220901 0.024000000208616257 0.023365069180727005 1000.0
113 M3 42 3 vision_to_text 0.006000000052154064 0.015000000596046448 0.015362463891506195 1000.0
114 M3 42 3 audio_to_vision 0.007000000216066837 0.01900000125169754 0.016374722123146057 1000.0
115 M3 42 3 vision_to_audio 0.003000000026077032 0.006000000052154064 0.008777403272688389 1000.0
116 M4 42 3 text_to_audio 0.0 0.0 0.002880551852285862 1000.0
117 M4 42 3 audio_to_text 0.0 0.0 0.003347013145685196 1000.0
118 M4 42 3 text_to_vision 0.004000000189989805 0.020000001415610313 0.021214386448264122 1000.0
119 M4 42 3 vision_to_text 0.003000000026077032 0.015000000596046448 0.016206368803977966 1000.0
120 M4 42 3 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.00828514900058508 1000.0
121 M4 42 3 vision_to_audio 0.0010000000474974513 0.005000000353902578 0.007783014792948961 1000.0
122 M3 3407 3 text_to_audio 0.039000000804662704 0.055000003427267075 0.05350850895047188 1000.0
123 M3 3407 3 audio_to_text 0.03800000250339508 0.04400000348687172 0.04879831150174141 1000.0
124 M3 3407 3 text_to_vision 0.0020000000949949026 0.009000000543892384 0.011307715438306332 1000.0
125 M3 3407 3 vision_to_text 0.003000000026077032 0.013000000268220901 0.011990648694336414 1000.0
126 M3 3407 3 audio_to_vision 0.007000000216066837 0.02500000037252903 0.022219380363821983 1000.0
127 M3 3407 3 vision_to_audio 0.003000000026077032 0.006000000052154064 0.009232758544385433 1000.0
128 M4 3407 3 text_to_audio 0.0 0.0 0.0037817470729351044 1000.0
129 M4 3407 3 audio_to_text 0.0 0.0 0.004079542122781277 1000.0
130 M4 3407 3 text_to_vision 0.003000000026077032 0.014000000432133675 0.015179671347141266 1000.0
131 M4 3407 3 vision_to_text 0.0 0.0 0.004454099107533693 1000.0
132 M4 3407 3 audio_to_vision 0.003000000026077032 0.015000000596046448 0.016372714191675186 1000.0
133 M4 3407 3 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011763687245547771 1000.0
134 M3 2026 3 text_to_audio 0.012000000104308128 0.02800000086426735 0.029318703338503838 1000.0
135 M3 2026 3 audio_to_text 0.012000000104308128 0.021000001579523087 0.027406470850110054 1000.0
136 M3 2026 3 text_to_vision 0.005000000353902578 0.01100000087171793 0.01579951122403145 1000.0
137 M3 2026 3 vision_to_text 0.006000000052154064 0.014000000432133675 0.015296950936317444 1000.0
138 M3 2026 3 audio_to_vision 0.006000000052154064 0.018000001087784767 0.016852334141731262 1000.0
139 M3 2026 3 vision_to_audio 0.005000000353902578 0.007000000216066837 0.01105457916855812 1000.0
140 M4 2026 3 text_to_audio 0.0010000000474974513 0.005000000353902578 0.007778078783303499 1000.0
141 M4 2026 3 audio_to_text 0.0010000000474974513 0.004000000189989805 0.007144550327211618 1000.0
142 M4 2026 3 text_to_vision 0.003000000026077032 0.015000000596046448 0.016557173803448677 1000.0
143 M4 2026 3 vision_to_text 0.0010000000474974513 0.005000000353902578 0.008665882982313633 1000.0
144 M4 2026 3 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.008280493319034576 1000.0
145 M4 2026 3 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.011866509914398193 1000.0
146 M1 fixed 4 text_to_audio 0.0 0.006000000052154064 0.006950765382498503 1000.0
147 M1 fixed 4 audio_to_text 0.005000000353902578 0.00800000037997961 0.010673508048057556 1000.0
148 M1 fixed 4 text_to_vision 0.0010000000474974513 0.007000000216066837 0.007159821223467588 1000.0
149 M1 fixed 4 vision_to_text 0.009000000543892384 0.009000000543892384 0.01242038607597351 1000.0
150 M1 fixed 4 audio_to_vision 0.0020000000949949026 0.013000000268220901 0.01288546621799469 1000.0
151 M1 fixed 4 vision_to_audio 0.0020000000949949026 0.01100000087171793 0.012257258407771587 1000.0
152 M2 fixed 4 text_to_audio 0.0010000000474974513 0.009000000543892384 0.009888535365462303 1000.0
153 M2 fixed 4 audio_to_text 0.005000000353902578 0.00800000037997961 0.011750306934118271 1000.0
154 M2 fixed 4 text_to_vision 0.0010000000474974513 0.0010000000474974513 0.005150882992893457 1000.0
155 M2 fixed 4 vision_to_text 0.010000000707805157 0.010000000707805157 0.013726433739066124 1000.0
156 M2 fixed 4 audio_to_vision 0.0010000000474974513 0.004000000189989805 0.00889560766518116 1000.0
157 M2 fixed 4 vision_to_audio 0.0020000000949949026 0.01100000087171793 0.011912405490875244 1000.0
158 M3 42 4 text_to_audio 0.02500000037252903 0.036000002175569534 0.038420792669057846 1000.0
159 M3 42 4 audio_to_text 0.00800000037997961 0.023000001907348633 0.02173793874680996 1000.0
160 M3 42 4 text_to_vision 0.0 0.005000000353902578 0.006278287619352341 1000.0
161 M3 42 4 vision_to_text 0.0020000000949949026 0.003000000026077032 0.007351293694227934 1000.0
162 M3 42 4 audio_to_vision 0.004000000189989805 0.01100000087171793 0.010880847461521626 1000.0
163 M3 42 4 vision_to_audio 0.006000000052154064 0.013000000268220901 0.015541859902441502 1000.0
164 M4 42 4 text_to_audio 0.004000000189989805 0.01900000125169754 0.020215420052409172 1000.0
165 M4 42 4 audio_to_text 0.0 0.0 0.004233467858284712 1000.0
166 M4 42 4 text_to_vision 0.0020000000949949026 0.010000000707805157 0.012029506266117096 1000.0
167 M4 42 4 vision_to_text 0.0 0.0 0.002725329017266631 1000.0
168 M4 42 4 audio_to_vision 0.0010000000474974513 0.005000000353902578 0.009229524992406368 1000.0
169 M4 42 4 vision_to_audio 0.003000000026077032 0.015000000596046448 0.017202889546751976 1000.0
170 M3 3407 4 text_to_audio 0.013000000268220901 0.032999999821186066 0.02899651788175106 1000.0
171 M3 3407 4 audio_to_text 0.007000000216066837 0.013000000268220901 0.01741517148911953 1000.0
172 M3 3407 4 text_to_vision 0.003000000026077032 0.006000000052154064 0.009169278666377068 1000.0
173 M3 3407 4 vision_to_text 0.006000000052154064 0.006000000052154064 0.011985262855887413 1000.0
174 M3 3407 4 audio_to_vision 0.0 0.004000000189989805 0.008569329045712948 1000.0
175 M3 3407 4 vision_to_audio 0.007000000216066837 0.010000000707805157 0.01606379821896553 1000.0
176 M4 3407 4 text_to_audio 0.003000000026077032 0.015000000596046448 0.015794366598129272 1000.0
177 M4 3407 4 audio_to_text 0.0 0.0 0.004071671981364489 1000.0
178 M4 3407 4 text_to_vision 0.0020000000949949026 0.010000000707805157 0.011711233295500278 1000.0
179 M4 3407 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.00810206588357687 1000.0
180 M4 3407 4 audio_to_vision 0.0020000000949949026 0.010000000707805157 0.013361790217459202 1000.0
181 M4 3407 4 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.012590405531227589 1000.0
182 M3 2026 4 text_to_audio 0.024000000208616257 0.04000000283122063 0.039450231939554214 1000.0
183 M3 2026 4 audio_to_text 0.006000000052154064 0.015000000596046448 0.01841675490140915 1000.0
184 M3 2026 4 text_to_vision 0.01100000087171793 0.01600000075995922 0.01930876635015011 1000.0
185 M3 2026 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.007208418566733599 1000.0
186 M3 2026 4 audio_to_vision 0.00800000037997961 0.017000000923871994 0.01701623573899269 1000.0
187 M3 2026 4 vision_to_audio 0.006000000052154064 0.014000000432133675 0.015153588727116585 1000.0
188 M4 2026 4 text_to_audio 0.0020000000949949026 0.010000000707805157 0.011758673004806042 1000.0
189 M4 2026 4 audio_to_text 0.0010000000474974513 0.005000000353902578 0.008434664458036423 1000.0
190 M4 2026 4 text_to_vision 0.0 0.0 0.0034921157639473677 1000.0
191 M4 2026 4 vision_to_text 0.0010000000474974513 0.005000000353902578 0.007859558798372746 1000.0
192 M4 2026 4 audio_to_vision 0.003000000026077032 0.014000000432133675 0.01731833629310131 1000.0
193 M4 2026 4 vision_to_audio 0.003000000026077032 0.015000000596046448 0.016523126512765884 1000.0
194 M1 fixed 5 text_to_audio 0.0 0.007000000216066837 0.008882568217813969 1000.0
195 M1 fixed 5 audio_to_text 0.0010000000474974513 0.003000000026077032 0.008665341883897781 1000.0
196 M1 fixed 5 text_to_vision 0.0 0.005000000353902578 0.00567932054400444 1000.0
197 M1 fixed 5 vision_to_text 0.0020000000949949026 0.009000000543892384 0.00881400890648365 1000.0
198 M1 fixed 5 audio_to_vision 0.0020000000949949026 0.012000000104308128 0.012908315286040306 1000.0
199 M1 fixed 5 vision_to_audio 0.003000000026077032 0.012000000104308128 0.0124351242557168 1000.0
200 M2 fixed 5 text_to_audio 0.0 0.009000000543892384 0.009715890511870384 1000.0
201 M2 fixed 5 audio_to_text 0.0020000000949949026 0.009000000543892384 0.010659169405698776 1000.0
202 M2 fixed 5 text_to_vision 0.0 0.009000000543892384 0.008167226798832417 1000.0
203 M2 fixed 5 vision_to_text 0.006000000052154064 0.009000000543892384 0.013293917290866375 1000.0
204 M2 fixed 5 audio_to_vision 0.0020000000949949026 0.012000000104308128 0.012093348428606987 1000.0
205 M2 fixed 5 vision_to_audio 0.0020000000949949026 0.007000000216066837 0.010753415524959564 1000.0
206 M3 42 5 text_to_audio 0.01600000075995922 0.06400000303983688 0.04854178801178932 1000.0
207 M3 42 5 audio_to_text 0.014000000432133675 0.03700000047683716 0.034194398671388626 1000.0
208 M3 42 5 text_to_vision 0.006000000052154064 0.006000000052154064 0.011698092333972454 1000.0
209 M3 42 5 vision_to_text 0.005000000353902578 0.012000000104308128 0.012927225790917873 1000.0
210 M3 42 5 audio_to_vision 0.013000000268220901 0.026000000536441803 0.02460731379687786 1000.0
211 M3 42 5 vision_to_audio 0.004000000189989805 0.007000000216066837 0.012679936364293098 1000.0
212 M4 42 5 text_to_audio 0.04000000283122063 0.052000001072883606 0.05324704200029373 1000.0
213 M4 42 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.008529611863195896 1000.0
214 M4 42 5 text_to_vision 0.0010000000474974513 0.005000000353902578 0.00876129325479269 1000.0
215 M4 42 5 vision_to_text 0.0020000000949949026 0.010000000707805157 0.011993298307061195 1000.0
216 M4 42 5 audio_to_vision 0.003000000026077032 0.015000000596046448 0.017660561949014664 1000.0
217 M4 42 5 vision_to_audio 0.0020000000949949026 0.010000000707805157 0.014293942600488663 1000.0
218 M3 3407 5 text_to_audio 0.03100000135600567 0.07200000435113907 0.061651285737752914 1000.0
219 M3 3407 5 audio_to_text 0.018000001087784767 0.04000000283122063 0.042421478778123856 1000.0
220 M3 3407 5 text_to_vision 0.009000000543892384 0.021000001579523087 0.020427849143743515 1000.0
221 M3 3407 5 vision_to_text 0.004000000189989805 0.013000000268220901 0.012980340048670769 1000.0
222 M3 3407 5 audio_to_vision 0.007000000216066837 0.018000001087784767 0.021291999146342278 1000.0
223 M3 3407 5 vision_to_audio 0.004000000189989805 0.01600000075995922 0.016669930890202522 1000.0
224 M4 3407 5 text_to_audio 0.0020000000949949026 0.010000000707805157 0.0127327349036932 1000.0
225 M4 3407 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.00789211317896843 1000.0
226 M4 3407 5 text_to_vision 0.0020000000949949026 0.005000000353902578 0.00888831913471222 1000.0
227 M4 3407 5 vision_to_text 0.0010000000474974513 0.005000000353902578 0.008448818698525429 1000.0
228 M4 3407 5 audio_to_vision 0.003000000026077032 0.013000000268220901 0.016633378341794014 1000.0
229 M4 3407 5 vision_to_audio 0.003000000026077032 0.015000000596046448 0.01952446810901165 1000.0
230 M3 2026 5 text_to_audio 0.03100000135600567 0.08500000089406967 0.06357216835021973 1000.0
231 M3 2026 5 audio_to_text 0.005000000353902578 0.032999999821186066 0.030199285596609116 1000.0
232 M3 2026 5 text_to_vision 0.007000000216066837 0.021000001579523087 0.015703614801168442 1000.0
233 M3 2026 5 vision_to_text 0.005000000353902578 0.009000000543892384 0.01244888361543417 1000.0
234 M3 2026 5 audio_to_vision 0.009000000543892384 0.023000001907348633 0.02339860051870346 1000.0
235 M3 2026 5 vision_to_audio 0.004000000189989805 0.015000000596046448 0.01597834937274456 1000.0
236 M4 2026 5 text_to_audio 0.052000001072883606 0.06300000101327896 0.06472568958997726 1000.0
237 M4 2026 5 audio_to_text 0.0010000000474974513 0.005000000353902578 0.007973534055054188 1000.0
238 M4 2026 5 text_to_vision 0.0010000000474974513 0.005000000353902578 0.007530428934842348 1000.0
239 M4 2026 5 vision_to_text 0.003000000026077032 0.015000000596046448 0.016602179035544395 1000.0
240 M4 2026 5 audio_to_vision 0.003000000026077032 0.015000000596046448 0.017103707417845726 1000.0
241 M4 2026 5 vision_to_audio 0.003000000026077032 0.014000000432133675 0.01859205961227417 1000.0
@@ -0,0 +1,25 @@
method,direction,n,r_at_1_mean,r_at_1_std,r_at_5_mean,r_at_5_std,mrr_mean,mrr_std
M1,text_to_audio,5,0.0012000000569969416,0.001643167750561775,0.007000000309199094,0.0015811389419980182,0.008951337542384863,0.0015105744211016399
M1,audio_to_text,5,0.0028000001097097994,0.0014832398031884779,0.006000000238418579,0.0021213204717563106,0.009594007954001426,0.0008820339883026645
M1,text_to_vision,5,0.0020000000949949026,0.0020000001240987344,0.007000000402331352,0.0031622779576235397,0.009382870234549045,0.0038973032823256706
M1,vision_to_text,5,0.005000000214204192,0.0038729835451934104,0.007000000402331352,0.004301162959612636,0.010874488297849894,0.0033759486781861315
M1,audio_to_vision,5,0.0010000000474974513,0.0010000000474974513,0.008600000198930502,0.004037325898109636,0.010010761767625808,0.0028143308073864106
M1,vision_to_audio,5,0.002000000048428774,0.0009999999892897908,0.008000000286847353,0.0036742348045336777,0.009903422184288501,0.0023425375089251867
M2,text_to_audio,5,0.0014000000432133675,0.0011401754384133685,0.009200000390410423,0.0022803508359855447,0.010239988192915916,0.0015598764796833854
M2,audio_to_text,5,0.004000000143423677,0.0015811388683705022,0.009200000483542681,0.0030331502218969453,0.011443204618990422,0.0018186032524210385
M2,text_to_vision,5,0.003600000194273889,0.003209361472308511,0.008800000487826764,0.004764451893141854,0.011315117496997118,0.004998444639556679
M2,vision_to_text,5,0.00680000027641654,0.001923538612521058,0.009200000576674939,0.0013038406947566386,0.013902883604168892,0.0007593274807152007
M2,audio_to_vision,5,0.0014000000664964318,0.0005477225835205916,0.006800000183284282,0.003346640098121672,0.009431831166148185,0.0017045066129491178
M2,vision_to_audio,5,0.002000000118277967,0.0018708288289165209,0.00640000025741756,0.0032093615938253205,0.00967191495001316,0.002293345042131113
M3,text_to_audio,15,0.01966666733690848,0.009707631019479971,0.043000001646578315,0.020511321663901527,0.03803744154671828,0.014634778360297347
M3,audio_to_text,15,0.009066667081788183,0.009353889531597692,0.020933334335374334,0.013084706401165278,0.023117945653696854,0.011741696228606028
M3,text_to_vision,15,0.0062666668944681685,0.00413118239876218,0.013800000709791978,0.007466114240091005,0.01513038482517004,0.005506734651604392
M3,vision_to_text,15,0.00506666685299327,0.003634490987677242,0.011266667147477467,0.005573748401698712,0.01275980711604158,0.003981040361844423
M3,audio_to_vision,15,0.006400000179807345,0.0033974781354081164,0.0142666673908631,0.007215723836767552,0.016017937225600083,0.005224730877211012
M3,vision_to_audio,15,0.00426666682275633,0.0024630605290560273,0.00913333367401113,0.005111145810859167,0.012493464909493924,0.0045251231363594895
M4,text_to_audio,15,0.007066666963510215,0.01601546933258453,0.01226666709408164,0.019414526078011158,0.014511969638988376,0.018944261404425476
M4,audio_to_text,15,0.00046666668883214396,0.0005163978040219006,0.002266666820893685,0.00252039318928005,0.005223872559145093,0.0023624842688679767
M4,text_to_vision,15,0.0016000000604738792,0.0011832159931418852,0.00760000046963493,0.005852960507593743,0.0100666221541663,0.005020802122090484
M4,vision_to_text,15,0.0014666667130465308,0.0010600988416483645,0.007333333759258191,0.005300494367735815,0.009886658145114779,0.004472077112598579
M4,audio_to_vision,15,0.002133333380334079,0.0009154754048125681,0.010533333849161863,0.004405624197684311,0.012994256491462389,0.00403471312250859
M4,vision_to_audio,15,0.00206666673378398,0.0007037315398609183,0.010266667200873296,0.003712270612174926,0.012846407542626063,0.003972973272038369
1 method direction n r_at_1_mean r_at_1_std r_at_5_mean r_at_5_std mrr_mean mrr_std
2 M1 text_to_audio 5 0.0012000000569969416 0.001643167750561775 0.007000000309199094 0.0015811389419980182 0.008951337542384863 0.0015105744211016399
3 M1 audio_to_text 5 0.0028000001097097994 0.0014832398031884779 0.006000000238418579 0.0021213204717563106 0.009594007954001426 0.0008820339883026645
4 M1 text_to_vision 5 0.0020000000949949026 0.0020000001240987344 0.007000000402331352 0.0031622779576235397 0.009382870234549045 0.0038973032823256706
5 M1 vision_to_text 5 0.005000000214204192 0.0038729835451934104 0.007000000402331352 0.004301162959612636 0.010874488297849894 0.0033759486781861315
6 M1 audio_to_vision 5 0.0010000000474974513 0.0010000000474974513 0.008600000198930502 0.004037325898109636 0.010010761767625808 0.0028143308073864106
7 M1 vision_to_audio 5 0.002000000048428774 0.0009999999892897908 0.008000000286847353 0.0036742348045336777 0.009903422184288501 0.0023425375089251867
8 M2 text_to_audio 5 0.0014000000432133675 0.0011401754384133685 0.009200000390410423 0.0022803508359855447 0.010239988192915916 0.0015598764796833854
9 M2 audio_to_text 5 0.004000000143423677 0.0015811388683705022 0.009200000483542681 0.0030331502218969453 0.011443204618990422 0.0018186032524210385
10 M2 text_to_vision 5 0.003600000194273889 0.003209361472308511 0.008800000487826764 0.004764451893141854 0.011315117496997118 0.004998444639556679
11 M2 vision_to_text 5 0.00680000027641654 0.001923538612521058 0.009200000576674939 0.0013038406947566386 0.013902883604168892 0.0007593274807152007
12 M2 audio_to_vision 5 0.0014000000664964318 0.0005477225835205916 0.006800000183284282 0.003346640098121672 0.009431831166148185 0.0017045066129491178
13 M2 vision_to_audio 5 0.002000000118277967 0.0018708288289165209 0.00640000025741756 0.0032093615938253205 0.00967191495001316 0.002293345042131113
14 M3 text_to_audio 15 0.01966666733690848 0.009707631019479971 0.043000001646578315 0.020511321663901527 0.03803744154671828 0.014634778360297347
15 M3 audio_to_text 15 0.009066667081788183 0.009353889531597692 0.020933334335374334 0.013084706401165278 0.023117945653696854 0.011741696228606028
16 M3 text_to_vision 15 0.0062666668944681685 0.00413118239876218 0.013800000709791978 0.007466114240091005 0.01513038482517004 0.005506734651604392
17 M3 vision_to_text 15 0.00506666685299327 0.003634490987677242 0.011266667147477467 0.005573748401698712 0.01275980711604158 0.003981040361844423
18 M3 audio_to_vision 15 0.006400000179807345 0.0033974781354081164 0.0142666673908631 0.007215723836767552 0.016017937225600083 0.005224730877211012
19 M3 vision_to_audio 15 0.00426666682275633 0.0024630605290560273 0.00913333367401113 0.005111145810859167 0.012493464909493924 0.0045251231363594895
20 M4 text_to_audio 15 0.007066666963510215 0.01601546933258453 0.01226666709408164 0.019414526078011158 0.014511969638988376 0.018944261404425476
21 M4 audio_to_text 15 0.00046666668883214396 0.0005163978040219006 0.002266666820893685 0.00252039318928005 0.005223872559145093 0.0023624842688679767
22 M4 text_to_vision 15 0.0016000000604738792 0.0011832159931418852 0.00760000046963493 0.005852960507593743 0.0100666221541663 0.005020802122090484
23 M4 vision_to_text 15 0.0014666667130465308 0.0010600988416483645 0.007333333759258191 0.005300494367735815 0.009886658145114779 0.004472077112598579
24 M4 audio_to_vision 15 0.002133333380334079 0.0009154754048125681 0.010533333849161863 0.004405624197684311 0.012994256491462389 0.00403471312250859
25 M4 vision_to_audio 15 0.00206666673378398 0.0007037315398609183 0.010266667200873296 0.003712270612174926 0.012846407542626063 0.003972973272038369
@@ -0,0 +1,39 @@
{
"created_utc": "2026-09-23T09:07:49.626404+00:00",
"sample_count": 100,
"group_count": 37,
"folds": 5,
"seeds": [
42,
3407,
2026
],
"device": "cuda",
"gpu_name": "NVIDIA GeForce RTX 5070 Ti",
"python": "3.14.7",
"torch": "2.14.0+cu130",
"parameters": {
"grid_size": 50,
"hidden_size": 128,
"heads": 4,
"dropout": 0.1,
"batch_size": 8,
"epochs_max": 50,
"early_stopping_patience": 8,
"learning_rate": 0.0001,
"mask_ratio": 0.2,
"retrieval_probe_epochs": 20,
"reconstruction_probe_epochs": 25,
"mvr_epsilon": 0.02
},
"objective": "masked reconstruction + cross-modal contrastive + temporal monotonicity; emotion labels unused",
"split_rule": "GroupKFold by group_id/video_id",
"example_sample_id": "-tPCytz4rww/12",
"elapsed_seconds": 516.2944533824921,
"interpretation_limits": [
"Grid-index retrieval is a representation-consistency probe, not independent temporal ground truth.",
"Masked reconstruction uses a decoder trained on the training fold and reports standardized-feature errors.",
"The emotion probe is a small-sample downstream utility check, not a claim of generalization to MOSEI.",
"No human event timestamps are available, so human IoU/MATE is not reported."
]
}
@@ -0,0 +1,742 @@
[
{
"fold": 1,
"train_sample_ids": [
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-qDkUB0GgYY/6",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-UUCSKoHeMA/0"
],
"train_video_ids": [
"-3nNcZdcdvU",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-3g5yACwYnA",
"-9y-fZ3swSY",
"-I_e4mIh0yE",
"-UUCSKoHeMA",
"-egA8-b7-3M",
"-qDkUB0GgYY"
]
},
{
"fold": 2,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mqbVkbCndg/0",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-HeZS2-Prhc/2"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-mJ2ud6oKI8",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-571d8cVauQ",
"-HeZS2-Prhc",
"-THoVjtIkeU",
"-UuX1xuaiiE",
"-lzEya4AM_4",
"-mqbVkbCndg",
"-wny0OAz3g8"
]
},
{
"fold": 3,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aNfi7CP8vM/7",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tANM6ETl_M/3",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-aqamKhZ1Ec/0",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-AUZQgSxyPQ/2",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-HeZS2-Prhc",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aNfi7CP8vM",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-t217m2on-s",
"-tANM6ETl_M",
"-tPCytz4rww",
"-uywlfIYOS8",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-AUZQgSxyPQ",
"-HwX2H8Z4hY",
"-RfYyzHpjk4",
"-aqamKhZ1Ec",
"-hnBHBN8p5A",
"-s9qJ7ATP7w",
"-vxjVxOeScU",
"-wMB_hJL-3o"
]
},
{
"fold": 4,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-3nNcZdcdvU/5",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-a55Q6RWvTA/3",
"-aqamKhZ1Ec/0",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-mqbVkbCndg/0",
"-tANM6ETl_M/3",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"validation_sample_ids": [
"-aNfi7CP8vM/7",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5"
],
"train_video_ids": [
"-3g5yACwYnA",
"-3nNcZdcdvU",
"-571d8cVauQ",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UuX1xuaiiE",
"-a55Q6RWvTA",
"-aqamKhZ1Ec",
"-dxfTGcXJoc",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-lzEya4AM_4",
"-mJ2ud6oKI8",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-s9qJ7ATP7w",
"-tANM6ETl_M",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8",
"-yRb-Jum7EQ"
],
"validation_video_ids": [
"-6rXp3zJ3kc",
"-UacrmKiTn4",
"-aNfi7CP8vM",
"-iRBcNs9oI8",
"-ri04Z7vwnc",
"-t217m2on-s",
"-tPCytz4rww",
"-uywlfIYOS8"
]
},
{
"fold": 5,
"train_sample_ids": [
"-3g5yACwYnA/13",
"-3g5yACwYnA/3",
"-3g5yACwYnA/2",
"-3g5yACwYnA/9",
"-HwX2H8Z4hY/2",
"-HwX2H8Z4hY/5",
"-HwX2H8Z4hY/6",
"-HwX2H8Z4hY/9",
"-THoVjtIkeU/12",
"-THoVjtIkeU/2",
"-THoVjtIkeU/6",
"-UuX1xuaiiE/1",
"-UuX1xuaiiE/0",
"-UuX1xuaiiE/3",
"-UuX1xuaiiE/6",
"-aNfi7CP8vM/7",
"-aqamKhZ1Ec/0",
"-egA8-b7-3M/26",
"-egA8-b7-3M/17",
"-egA8-b7-3M/18",
"-egA8-b7-3M/16",
"-egA8-b7-3M/13",
"-egA8-b7-3M/1",
"-egA8-b7-3M/6",
"-egA8-b7-3M/9",
"-egA8-b7-3M/20",
"-iRBcNs9oI8/3",
"-iRBcNs9oI8/7",
"-iRBcNs9oI8/6",
"-iRBcNs9oI8/9",
"-iRBcNs9oI8/8",
"-lzEya4AM_4/5",
"-lzEya4AM_4/6",
"-mqbVkbCndg/0",
"-t217m2on-s/2",
"-t217m2on-s/7",
"-tPCytz4rww/11",
"-tPCytz4rww/10",
"-tPCytz4rww/12",
"-tPCytz4rww/16",
"-tPCytz4rww/18",
"-vxjVxOeScU/4",
"-wMB_hJL-3o/7",
"-wny0OAz3g8/1",
"-wny0OAz3g8/0",
"-wny0OAz3g8/3",
"-wny0OAz3g8/2",
"-wny0OAz3g8/5",
"-wny0OAz3g8/7",
"-wny0OAz3g8/9",
"-571d8cVauQ/0",
"-571d8cVauQ/5",
"-I_e4mIh0yE/1",
"-I_e4mIh0yE/3",
"-UacrmKiTn4/10",
"-UacrmKiTn4/4",
"-hnBHBN8p5A/7",
"-hnBHBN8p5A/6",
"-qDkUB0GgYY/6",
"-uywlfIYOS8/4",
"-6rXp3zJ3kc/8",
"-9y-fZ3swSY/0",
"-9y-fZ3swSY/4",
"-9y-fZ3swSY/8",
"-AUZQgSxyPQ/2",
"-HeZS2-Prhc/2",
"-RfYyzHpjk4/11",
"-RfYyzHpjk4/8",
"-RfYyzHpjk4/2",
"-UUCSKoHeMA/0",
"-ri04Z7vwnc/0",
"-ri04Z7vwnc/2",
"-ri04Z7vwnc/5",
"-s9qJ7ATP7w/1",
"-s9qJ7ATP7w/0",
"-s9qJ7ATP7w/5",
"-s9qJ7ATP7w/4",
"-s9qJ7ATP7w/7",
"-s9qJ7ATP7w/6",
"-s9qJ7ATP7w/8"
],
"validation_sample_ids": [
"-3nNcZdcdvU/5",
"-NFrJFQijFE/1",
"-NFrJFQijFE/2",
"-a55Q6RWvTA/3",
"-dxfTGcXJoc/1",
"-dxfTGcXJoc/0",
"-dxfTGcXJoc/2",
"-dxfTGcXJoc/6",
"-mJ2ud6oKI8/1",
"-mJ2ud6oKI8/2",
"-mJ2ud6oKI8/6",
"-mJ2ud6oKI8/9",
"-mJ2ud6oKI8/8",
"-tANM6ETl_M/3",
"-MeTTeMJBNc/0",
"-MeTTeMJBNc/13",
"-MeTTeMJBNc/7",
"-yRb-Jum7EQ/1",
"-yRb-Jum7EQ/5",
"-yRb-Jum7EQ/6"
],
"train_video_ids": [
"-3g5yACwYnA",
"-571d8cVauQ",
"-6rXp3zJ3kc",
"-9y-fZ3swSY",
"-AUZQgSxyPQ",
"-HeZS2-Prhc",
"-HwX2H8Z4hY",
"-I_e4mIh0yE",
"-RfYyzHpjk4",
"-THoVjtIkeU",
"-UUCSKoHeMA",
"-UacrmKiTn4",
"-UuX1xuaiiE",
"-aNfi7CP8vM",
"-aqamKhZ1Ec",
"-egA8-b7-3M",
"-hnBHBN8p5A",
"-iRBcNs9oI8",
"-lzEya4AM_4",
"-mqbVkbCndg",
"-qDkUB0GgYY",
"-ri04Z7vwnc",
"-s9qJ7ATP7w",
"-t217m2on-s",
"-tPCytz4rww",
"-uywlfIYOS8",
"-vxjVxOeScU",
"-wMB_hJL-3o",
"-wny0OAz3g8"
],
"validation_video_ids": [
"-3nNcZdcdvU",
"-MeTTeMJBNc",
"-NFrJFQijFE",
"-a55Q6RWvTA",
"-dxfTGcXJoc",
"-mJ2ud6oKI8",
"-tANM6ETl_M",
"-yRb-Jum7EQ"
]
}
]
@@ -0,0 +1,634 @@
{
"alignment": [
{
"method": "M1",
"modality": "text",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.0008249471336603165,
"normalized_entropy_std": 0.005233484169432736,
"width80_source_positions_mean": 1.004799998998642,
"width80_source_positions_std": 0.0312849358458409,
"expected_time_start_s_mean": 0.373650000790672,
"expected_time_start_s_std": 0.28760723340035366,
"expected_time_end_s_mean": 7.3229000100512325,
"expected_time_end_s_std": 4.567867965796829,
"trajectory_span_fraction_mean": 0.8612861529795919,
"trajectory_span_fraction_std": 0.07875035859337948
},
{
"method": "M1",
"modality": "audio",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.35824004098773005,
"normalized_entropy_std": 0.05432258900057586,
"width80_source_positions_mean": 12.056200053691864,
"width80_source_positions_std": 7.307551875172266,
"expected_time_start_s_mean": 0.3177500062027157,
"expected_time_start_s_std": 0.2794568898376313,
"expected_time_end_s_mean": 7.441361117547106,
"expected_time_end_s_std": 4.533387053269832,
"trajectory_span_fraction_mean": 0.8937758523714728,
"trajectory_span_fraction_std": 0.06338502509117576
},
{
"method": "M1",
"modality": "vision",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.02091189981205389,
"normalized_entropy_std": 0.03423249778642653,
"width80_source_positions_mean": 1.144999988079071,
"width80_source_positions_std": 0.27983220085257804,
"expected_time_start_s_mean": 0.30618889875375566,
"expected_time_start_s_std": 0.2956530091396046,
"expected_time_end_s_mean": 7.43663269007148,
"expected_time_end_s_std": 4.535796288658362,
"trajectory_span_fraction_mean": 0.8956127409730107,
"trajectory_span_fraction_std": 0.0656268624353691
},
{
"method": "M2",
"modality": "text",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.003971234038472176,
"normalized_entropy_std": 0.010095580232451877,
"width80_source_positions_mean": 1.0215999960899353,
"width80_source_positions_std": 0.060615033764357146,
"expected_time_start_s_mean": 0.3753062509072731,
"expected_time_start_s_std": 0.2858492171110105,
"expected_time_end_s_mean": 7.3229000100512325,
"expected_time_end_s_std": 4.567867965796829,
"trajectory_span_fraction_mean": 0.8611038638930768,
"trajectory_span_fraction_std": 0.07866203063075565
},
{
"method": "M2",
"modality": "audio",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.39199826538562776,
"normalized_entropy_std": 0.04856653297051989,
"width80_source_positions_mean": 12.777399990558624,
"width80_source_positions_std": 7.437441491467199,
"expected_time_start_s_mean": 0.08135000219190694,
"expected_time_start_s_std": 0.046356409936463026,
"expected_time_end_s_mean": 7.716148159798975,
"expected_time_end_s_std": 4.594836781525238,
"trajectory_span_fraction_mean": 0.968455815287307,
"trajectory_span_fraction_std": 0.009448427161791137
},
{
"method": "M2",
"modality": "vision",
"n": 100,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.015714890849776567,
"normalized_entropy_std": 0.04120196183036711,
"width80_source_positions_mean": 1.108399999141693,
"width80_source_positions_std": 0.3072087474461631,
"expected_time_start_s_mean": 0.024000000567226686,
"expected_time_start_s_std": 0.0514830577428781,
"expected_time_end_s_mean": 7.7077327980137875,
"expected_time_end_s_std": 4.617512750249008,
"trajectory_span_fraction_mean": 0.9744537002407014,
"trajectory_span_fraction_std": 0.016307502074103464
},
{
"method": "M3",
"modality": "text",
"n": 300,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.0004981429129838944,
"normalized_entropy_std": 0.004964740878146852,
"width80_source_positions_mean": 1.0029999995231629,
"width80_source_positions_std": 0.02989949245257891,
"expected_time_start_s_mean": 0.37365000079067195,
"expected_time_start_s_std": 0.28664372237282515,
"expected_time_end_s_mean": 7.320650003693404,
"expected_time_end_s_std": 4.545633324884766,
"trajectory_span_fraction_mean": 0.8611844960658346,
"trajectory_span_fraction_std": 0.07838045740132776
},
{
"method": "M3",
"modality": "audio",
"n": 300,
"mvr_mean": 0.01857142824679613,
"mvr_std": 0.028241165293696132,
"normalized_entropy_mean": 0.992265993754069,
"normalized_entropy_std": 0.00445452263162519,
"width80_source_positions_mean": 607.8159977213542,
"width80_source_positions_std": 365.5015286317014,
"expected_time_start_s_mean": 3.883805515007902,
"expected_time_start_s_std": 2.328319755482711,
"expected_time_end_s_mean": 3.8684255829138965,
"expected_time_end_s_std": 2.2993181536785716,
"trajectory_span_fraction_mean": -0.001562372346719104,
"trajectory_span_fraction_std": 0.016018112979249148
},
{
"method": "M3",
"modality": "vision",
"n": 300,
"mvr_mean": 0.06843537279715141,
"mvr_std": 0.06722724739800441,
"normalized_entropy_mean": 0.9827406370639801,
"normalized_entropy_std": 0.021948411615917433,
"width80_source_positions_mean": 31.175333293279014,
"width80_source_positions_std": 18.019629268682312,
"expected_time_start_s_mean": 3.8615792202155004,
"expected_time_start_s_std": 2.266734235588232,
"expected_time_end_s_mean": 3.834497800808173,
"expected_time_end_s_std": 2.259828297195092,
"trajectory_span_fraction_mean": -0.004611149926980335,
"trajectory_span_fraction_std": 0.0433981312652511
},
{
"method": "M4",
"modality": "text",
"n": 300,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.9999777108430863,
"normalized_entropy_std": 4.937972466967483e-06,
"width80_source_positions_mean": 15.891133365631104,
"width80_source_positions_std": 8.431963588888667,
"expected_time_start_s_mean": 3.83314954836363,
"expected_time_start_s_std": 2.345445678904365,
"expected_time_end_s_mean": 3.8348252384682473,
"expected_time_end_s_std": 2.3463801144674252,
"trajectory_span_fraction_mean": 0.00020046850045522113,
"trajectory_span_fraction_std": 0.0014442569299196843
},
{
"method": "M4",
"modality": "audio",
"n": 300,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.9999900078773498,
"normalized_entropy_std": 3.4153207662384735e-06,
"width80_source_positions_mean": 618.1359326680501,
"width80_source_positions_std": 369.4452964156476,
"expected_time_start_s_mean": 3.867822667326802,
"expected_time_start_s_std": 2.309874647070878,
"expected_time_end_s_mean": 3.868319556290074,
"expected_time_end_s_std": 2.3104860442974764,
"trajectory_span_fraction_mean": 4.607309897740703e-05,
"trajectory_span_fraction_std": 0.0006428369960135046
},
{
"method": "M4",
"modality": "vision",
"n": 300,
"mvr_mean": 0.0,
"mvr_std": 0.0,
"normalized_entropy_mean": 0.9999960112571716,
"normalized_entropy_std": 4.416330070963206e-06,
"width80_source_positions_mean": 32.02326666514079,
"width80_source_positions_std": 18.480166711480305,
"expected_time_start_s_mean": 3.8529649544991638,
"expected_time_start_s_std": 2.3080516925693866,
"expected_time_end_s_mean": 3.8542076376005214,
"expected_time_end_s_std": 2.3106234287813843,
"trajectory_span_fraction_mean": 0.00010548800230026034,
"trajectory_span_fraction_std": 0.001267114074095688
}
],
"retrieval": [
{
"method": "M1",
"direction": "text_to_audio",
"n": 5,
"r_at_1_mean": 0.0012000000569969416,
"r_at_1_std": 0.001643167750561775,
"r_at_5_mean": 0.007000000309199094,
"r_at_5_std": 0.0015811389419980182,
"mrr_mean": 0.008951337542384863,
"mrr_std": 0.0015105744211016399
},
{
"method": "M1",
"direction": "audio_to_text",
"n": 5,
"r_at_1_mean": 0.0028000001097097994,
"r_at_1_std": 0.0014832398031884779,
"r_at_5_mean": 0.006000000238418579,
"r_at_5_std": 0.0021213204717563106,
"mrr_mean": 0.009594007954001426,
"mrr_std": 0.0008820339883026645
},
{
"method": "M1",
"direction": "text_to_vision",
"n": 5,
"r_at_1_mean": 0.0020000000949949026,
"r_at_1_std": 0.0020000001240987344,
"r_at_5_mean": 0.007000000402331352,
"r_at_5_std": 0.0031622779576235397,
"mrr_mean": 0.009382870234549045,
"mrr_std": 0.0038973032823256706
},
{
"method": "M1",
"direction": "vision_to_text",
"n": 5,
"r_at_1_mean": 0.005000000214204192,
"r_at_1_std": 0.0038729835451934104,
"r_at_5_mean": 0.007000000402331352,
"r_at_5_std": 0.004301162959612636,
"mrr_mean": 0.010874488297849894,
"mrr_std": 0.0033759486781861315
},
{
"method": "M1",
"direction": "audio_to_vision",
"n": 5,
"r_at_1_mean": 0.0010000000474974513,
"r_at_1_std": 0.0010000000474974513,
"r_at_5_mean": 0.008600000198930502,
"r_at_5_std": 0.004037325898109636,
"mrr_mean": 0.010010761767625808,
"mrr_std": 0.0028143308073864106
},
{
"method": "M1",
"direction": "vision_to_audio",
"n": 5,
"r_at_1_mean": 0.002000000048428774,
"r_at_1_std": 0.0009999999892897908,
"r_at_5_mean": 0.008000000286847353,
"r_at_5_std": 0.0036742348045336777,
"mrr_mean": 0.009903422184288501,
"mrr_std": 0.0023425375089251867
},
{
"method": "M2",
"direction": "text_to_audio",
"n": 5,
"r_at_1_mean": 0.0014000000432133675,
"r_at_1_std": 0.0011401754384133685,
"r_at_5_mean": 0.009200000390410423,
"r_at_5_std": 0.0022803508359855447,
"mrr_mean": 0.010239988192915916,
"mrr_std": 0.0015598764796833854
},
{
"method": "M2",
"direction": "audio_to_text",
"n": 5,
"r_at_1_mean": 0.004000000143423677,
"r_at_1_std": 0.0015811388683705022,
"r_at_5_mean": 0.009200000483542681,
"r_at_5_std": 0.0030331502218969453,
"mrr_mean": 0.011443204618990422,
"mrr_std": 0.0018186032524210385
},
{
"method": "M2",
"direction": "text_to_vision",
"n": 5,
"r_at_1_mean": 0.003600000194273889,
"r_at_1_std": 0.003209361472308511,
"r_at_5_mean": 0.008800000487826764,
"r_at_5_std": 0.004764451893141854,
"mrr_mean": 0.011315117496997118,
"mrr_std": 0.004998444639556679
},
{
"method": "M2",
"direction": "vision_to_text",
"n": 5,
"r_at_1_mean": 0.00680000027641654,
"r_at_1_std": 0.001923538612521058,
"r_at_5_mean": 0.009200000576674939,
"r_at_5_std": 0.0013038406947566386,
"mrr_mean": 0.013902883604168892,
"mrr_std": 0.0007593274807152007
},
{
"method": "M2",
"direction": "audio_to_vision",
"n": 5,
"r_at_1_mean": 0.0014000000664964318,
"r_at_1_std": 0.0005477225835205916,
"r_at_5_mean": 0.006800000183284282,
"r_at_5_std": 0.003346640098121672,
"mrr_mean": 0.009431831166148185,
"mrr_std": 0.0017045066129491178
},
{
"method": "M2",
"direction": "vision_to_audio",
"n": 5,
"r_at_1_mean": 0.002000000118277967,
"r_at_1_std": 0.0018708288289165209,
"r_at_5_mean": 0.00640000025741756,
"r_at_5_std": 0.0032093615938253205,
"mrr_mean": 0.00967191495001316,
"mrr_std": 0.002293345042131113
},
{
"method": "M3",
"direction": "text_to_audio",
"n": 15,
"r_at_1_mean": 0.01966666733690848,
"r_at_1_std": 0.009707631019479971,
"r_at_5_mean": 0.043000001646578315,
"r_at_5_std": 0.020511321663901527,
"mrr_mean": 0.03803744154671828,
"mrr_std": 0.014634778360297347
},
{
"method": "M3",
"direction": "audio_to_text",
"n": 15,
"r_at_1_mean": 0.009066667081788183,
"r_at_1_std": 0.009353889531597692,
"r_at_5_mean": 0.020933334335374334,
"r_at_5_std": 0.013084706401165278,
"mrr_mean": 0.023117945653696854,
"mrr_std": 0.011741696228606028
},
{
"method": "M3",
"direction": "text_to_vision",
"n": 15,
"r_at_1_mean": 0.0062666668944681685,
"r_at_1_std": 0.00413118239876218,
"r_at_5_mean": 0.013800000709791978,
"r_at_5_std": 0.007466114240091005,
"mrr_mean": 0.01513038482517004,
"mrr_std": 0.005506734651604392
},
{
"method": "M3",
"direction": "vision_to_text",
"n": 15,
"r_at_1_mean": 0.00506666685299327,
"r_at_1_std": 0.003634490987677242,
"r_at_5_mean": 0.011266667147477467,
"r_at_5_std": 0.005573748401698712,
"mrr_mean": 0.01275980711604158,
"mrr_std": 0.003981040361844423
},
{
"method": "M3",
"direction": "audio_to_vision",
"n": 15,
"r_at_1_mean": 0.006400000179807345,
"r_at_1_std": 0.0033974781354081164,
"r_at_5_mean": 0.0142666673908631,
"r_at_5_std": 0.007215723836767552,
"mrr_mean": 0.016017937225600083,
"mrr_std": 0.005224730877211012
},
{
"method": "M3",
"direction": "vision_to_audio",
"n": 15,
"r_at_1_mean": 0.00426666682275633,
"r_at_1_std": 0.0024630605290560273,
"r_at_5_mean": 0.00913333367401113,
"r_at_5_std": 0.005111145810859167,
"mrr_mean": 0.012493464909493924,
"mrr_std": 0.0045251231363594895
},
{
"method": "M4",
"direction": "text_to_audio",
"n": 15,
"r_at_1_mean": 0.007066666963510215,
"r_at_1_std": 0.01601546933258453,
"r_at_5_mean": 0.01226666709408164,
"r_at_5_std": 0.019414526078011158,
"mrr_mean": 0.014511969638988376,
"mrr_std": 0.018944261404425476
},
{
"method": "M4",
"direction": "audio_to_text",
"n": 15,
"r_at_1_mean": 0.00046666668883214396,
"r_at_1_std": 0.0005163978040219006,
"r_at_5_mean": 0.002266666820893685,
"r_at_5_std": 0.00252039318928005,
"mrr_mean": 0.005223872559145093,
"mrr_std": 0.0023624842688679767
},
{
"method": "M4",
"direction": "text_to_vision",
"n": 15,
"r_at_1_mean": 0.0016000000604738792,
"r_at_1_std": 0.0011832159931418852,
"r_at_5_mean": 0.00760000046963493,
"r_at_5_std": 0.005852960507593743,
"mrr_mean": 0.0100666221541663,
"mrr_std": 0.005020802122090484
},
{
"method": "M4",
"direction": "vision_to_text",
"n": 15,
"r_at_1_mean": 0.0014666667130465308,
"r_at_1_std": 0.0010600988416483645,
"r_at_5_mean": 0.007333333759258191,
"r_at_5_std": 0.005300494367735815,
"mrr_mean": 0.009886658145114779,
"mrr_std": 0.004472077112598579
},
{
"method": "M4",
"direction": "audio_to_vision",
"n": 15,
"r_at_1_mean": 0.002133333380334079,
"r_at_1_std": 0.0009154754048125681,
"r_at_5_mean": 0.010533333849161863,
"r_at_5_std": 0.004405624197684311,
"mrr_mean": 0.012994256491462389,
"mrr_std": 0.00403471312250859
},
{
"method": "M4",
"direction": "vision_to_audio",
"n": 15,
"r_at_1_mean": 0.00206666673378398,
"r_at_1_std": 0.0007037315398609183,
"r_at_5_mean": 0.010266667200873296,
"r_at_5_std": 0.003712270612174926,
"mrr_mean": 0.012846407542626063,
"mrr_std": 0.003972973272038369
}
],
"reconstruction": [
{
"method": "M1",
"target_modality": "text",
"n": 5,
"mae_standardized_mean": 0.8308010935783386,
"mae_standardized_std": 0.028634584948895196,
"smooth_l1_standardized_mean": 0.4546235024929047,
"smooth_l1_standardized_std": 0.025151457325080215
},
{
"method": "M1",
"target_modality": "audio",
"n": 5,
"mae_standardized_mean": 0.6908806562423706,
"mae_standardized_std": 0.06720606353302543,
"smooth_l1_standardized_mean": 0.34603241086006165,
"smooth_l1_standardized_std": 0.06541831705153776
},
{
"method": "M1",
"target_modality": "vision",
"n": 5,
"mae_standardized_mean": 0.9209612965583801,
"mae_standardized_std": 0.04009937912435952,
"smooth_l1_standardized_mean": 0.5347834944725036,
"smooth_l1_standardized_std": 0.036821434614539225
},
{
"method": "M2",
"target_modality": "text",
"n": 5,
"mae_standardized_mean": 0.8190697312355042,
"mae_standardized_std": 0.03563334190353864,
"smooth_l1_standardized_mean": 0.44534851908683776,
"smooth_l1_standardized_std": 0.030773561031959877
},
{
"method": "M2",
"target_modality": "audio",
"n": 5,
"mae_standardized_mean": 0.6621044635772705,
"mae_standardized_std": 0.05689547728439605,
"smooth_l1_standardized_mean": 0.32293344140052793,
"smooth_l1_standardized_std": 0.05414120404134669
},
{
"method": "M2",
"target_modality": "vision",
"n": 5,
"mae_standardized_mean": 0.9195392370223999,
"mae_standardized_std": 0.03744298345037278,
"smooth_l1_standardized_mean": 0.5340398073196411,
"smooth_l1_standardized_std": 0.0346960350528028
},
{
"method": "M3",
"target_modality": "text",
"n": 15,
"mae_standardized_mean": 0.8311106324195862,
"mae_standardized_std": 0.025205971351551344,
"smooth_l1_standardized_mean": 0.45519693891207375,
"smooth_l1_standardized_std": 0.021717713583393992
},
{
"method": "M3",
"target_modality": "audio",
"n": 15,
"mae_standardized_mean": 0.38449583053588865,
"mae_standardized_std": 0.050754623523549675,
"smooth_l1_standardized_mean": 0.14107545365889868,
"smooth_l1_standardized_std": 0.05381054674326515
},
{
"method": "M3",
"target_modality": "vision",
"n": 15,
"mae_standardized_mean": 0.868217929204305,
"mae_standardized_std": 0.03802189028391118,
"smooth_l1_standardized_mean": 0.48890710473060606,
"smooth_l1_standardized_std": 0.03450529817617854
},
{
"method": "M4",
"target_modality": "text",
"n": 15,
"mae_standardized_mean": 0.4924671709537506,
"mae_standardized_std": 0.011169848357006705,
"smooth_l1_standardized_mean": 0.18583014905452727,
"smooth_l1_standardized_std": 0.008262501869688806
},
{
"method": "M4",
"target_modality": "audio",
"n": 15,
"mae_standardized_mean": 0.37839911778767904,
"mae_standardized_std": 0.05503994105535867,
"smooth_l1_standardized_mean": 0.13528098662694296,
"smooth_l1_standardized_std": 0.05838370221384358
},
{
"method": "M4",
"target_modality": "vision",
"n": 15,
"mae_standardized_mean": 0.8630666335423788,
"mae_standardized_std": 0.03867536107911695,
"smooth_l1_standardized_mean": 0.48466227054595945,
"smooth_l1_standardized_std": 0.03512809212998346
}
],
"emotion": [
{
"method": "M1",
"n": 5,
"accuracy_mean": 0.5900000000000001,
"accuracy_std": 0.1673320053068151,
"macro_f1_mean": 0.43700352733686065,
"macro_f1_std": 0.22785227549938045,
"mae_mean": 0.5521364179253578,
"mae_std": 0.12144893660249381,
"pearson_mean": 0.5305163430956757,
"pearson_std": 0.09576372130151391
},
{
"method": "M2",
"n": 5,
"accuracy_mean": 0.55,
"accuracy_std": 0.1767766952966369,
"macro_f1_mean": 0.3434301254301254,
"macro_f1_std": 0.14084701638388486,
"mae_mean": 0.5691992919147014,
"mae_std": 0.11997495602498047,
"pearson_mean": 0.5082439027512137,
"pearson_std": 0.09789171743073
},
{
"method": "M3",
"n": 15,
"accuracy_mean": 0.5766666666666667,
"accuracy_std": 0.18503538915054937,
"macro_f1_mean": 0.43789820467598245,
"macro_f1_std": 0.21990438278507196,
"mae_mean": 0.5534388141334057,
"mae_std": 0.11230839313313754,
"pearson_mean": 0.5457241196963684,
"pearson_std": 0.09265676731571848
},
{
"method": "M4",
"n": 15,
"accuracy_mean": 0.56,
"accuracy_std": 0.15142183273415838,
"macro_f1_mean": 0.41446068467573843,
"macro_f1_std": 0.13578816810911878,
"mae_mean": 0.6231901267170906,
"mae_std": 0.13326862372482587,
"pearson_mean": 0.5067829543507287,
"pearson_std": 0.1021700904550588
}
]
}
@@ -0,0 +1,31 @@
method,seed,fold,best_epoch,best_validation_objective,checkpoint
M3,42,1,50,4.937662982940674,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M3.pt
M4,42,1,4,6.076866149902344,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M4.pt
M3,3407,1,50,4.9825927734375,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M3.pt
M4,3407,1,4,5.871845149993897,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M4.pt
M3,2026,1,50,5.102480220794678,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M3.pt
M4,2026,1,3,6.099013900756836,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M4.pt
M3,42,2,1,6.125876426696777,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M3.pt
M4,42,2,1,5.942682838439941,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M4.pt
M3,3407,2,50,5.144535160064697,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M3.pt
M4,3407,2,3,6.067738342285156,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M4.pt
M3,2026,2,50,5.17415599822998,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M3.pt
M4,2026,2,1,5.96431074142456,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M4.pt
M3,42,3,50,4.976429462432861,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M3.pt
M4,42,3,3,6.195817565917968,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M4.pt
M3,3407,3,50,4.659491157531738,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M3.pt
M4,3407,3,9,5.910655784606933,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M4.pt
M3,2026,3,50,4.999933815002441,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M3.pt
M4,2026,3,3,6.024094963073731,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M4.pt
M3,42,4,50,5.152298069000244,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M3.pt
M4,42,4,1,5.998569297790527,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M4.pt
M3,3407,4,50,5.098000621795654,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M3.pt
M4,3407,4,2,6.03455867767334,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M4.pt
M3,2026,4,50,5.260256576538086,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M3.pt
M4,2026,4,2,6.081212425231934,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M4.pt
M3,42,5,50,5.298740291595459,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M3.pt
M4,42,5,4,6.0218939781188965,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M4.pt
M3,3407,5,50,4.953551292419434,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M3.pt
M4,3407,5,8,6.058719825744629,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M4.pt
M3,2026,5,50,5.133688831329346,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M3.pt
M4,2026,5,1,5.882685661315918,/home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M4.pt
1 method seed fold best_epoch best_validation_objective checkpoint
2 M3 42 1 50 4.937662982940674 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M3.pt
3 M4 42 1 4 6.076866149902344 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_42/M4.pt
4 M3 3407 1 50 4.9825927734375 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M3.pt
5 M4 3407 1 4 5.871845149993897 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_3407/M4.pt
6 M3 2026 1 50 5.102480220794678 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M3.pt
7 M4 2026 1 3 6.099013900756836 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_01/seed_2026/M4.pt
8 M3 42 2 1 6.125876426696777 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M3.pt
9 M4 42 2 1 5.942682838439941 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_42/M4.pt
10 M3 3407 2 50 5.144535160064697 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M3.pt
11 M4 3407 2 3 6.067738342285156 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_3407/M4.pt
12 M3 2026 2 50 5.17415599822998 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M3.pt
13 M4 2026 2 1 5.96431074142456 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_02/seed_2026/M4.pt
14 M3 42 3 50 4.976429462432861 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M3.pt
15 M4 42 3 3 6.195817565917968 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_42/M4.pt
16 M3 3407 3 50 4.659491157531738 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M3.pt
17 M4 3407 3 9 5.910655784606933 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_3407/M4.pt
18 M3 2026 3 50 4.999933815002441 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M3.pt
19 M4 2026 3 3 6.024094963073731 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_03/seed_2026/M4.pt
20 M3 42 4 50 5.152298069000244 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M3.pt
21 M4 42 4 1 5.998569297790527 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_42/M4.pt
22 M3 3407 4 50 5.098000621795654 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M3.pt
23 M4 3407 4 2 6.03455867767334 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_3407/M4.pt
24 M3 2026 4 50 5.260256576538086 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M3.pt
25 M4 2026 4 2 6.081212425231934 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_04/seed_2026/M4.pt
26 M3 42 5 50 5.298740291595459 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M3.pt
27 M4 42 5 4 6.0218939781188965 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_42/M4.pt
28 M3 3407 5 50 4.953551292419434 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M3.pt
29 M4 3407 5 8 6.058719825744629 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_3407/M4.pt
30 M3 2026 5 50 5.133688831329346 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M3.pt
31 M4 2026 5 1 5.882685661315918 /home/gloamxun/modeling_zhaocui/Q1/outputs/method_comparison/fold_05/seed_2026/M4.pt
Binary file not shown.

After

Width:  |  Height:  |  Size: 516 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 282 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 281 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 306 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 314 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 242 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 280 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 281 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 238 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 237 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 245 KiB

@@ -0,0 +1,386 @@
2026 年中国研究生数学建模竞赛 E 题
复杂场景下多模态情感预测的数学建模与算法设计
一、背景
情感分析是具身智能、自然人机交互等技术领域的关键支撑技术之一。真实交
互中人类情绪通过文本语义、语音韵律、面部表情与肢体动作等多通道信号协同表
达,多模态信息联合建模是突破单模态识别局限、提升复杂场景下情绪理解准确性
与稳定性的核心路径。然而,多模态情感预测在落地中面临三重挑战:文本、语音、
视觉模态在采样频率、特征维度与时间尺度上差异显著,特征提取与时序对齐的质
量直接决定模型性能上限;实际采集常因画面遮挡、环境噪声、转写失败等导致模
态局部缺失,常规固定权重融合模型性能急剧下降;多数模型仅能输出预测结果,
决策依据难以追溯与复核,制约了模型的可信应用。本题以 CMU-MOSEI 英文多
模态情感数据集为基础,围绕多模态特征提取与时序对齐、模态缺失条件下的鲁棒
预测、可解释性情感预测三个层级开展系统性建模研究。
二、数据说明
本题提供如下数据与材料:
1.附件 1:原始多模态样本
包含从 CMU-MOSEI 数据集中筛选的 100 条英文原始视频,具体描述如下:
 附件包含 37 个子文件夹,其命名为原始视频标识 video_id;每个子
文件夹包含一个或多个视频片段,命名 clip_id.mp4。video_id 和 clip_id 共
同确定一条视频样本。所有视频时长介于 2.648 秒和 34.567 秒之间。
 附件同时提供视频标注对应表(label-100.xlsx),每一行对应一个视
频样本。其中,video_id 列和 clip_id 列对应子文件夹名称和文件名;text
列为该视频片段对应的英文转写文本;label 列为连续情感强度标注,取值:
[-3,0)(负向)、0(中性)、(0,3](正向),数值越接近-3 表示负向程度
越强,越接近 3 表示正向程度越强;0 仅属于中性,不同时归入负向或正
向。;annotation 列为情感极性的英文文字标签,取值:Negative(负向情
感)、Neutral(中性情感)、Positive(正向情感)。
注:本附件所有内容由赛题统一提供。参赛队不得自行替换、增补、删除或修
改样本及其标签;如需在模型处理中对极短、无效或异常片段采取掩码、截断或其
他处理,应保留原始样本,并在报告中说明处理规则。
2.附件 2:CMU-MOSEI 标准化多模态特征文件(.pkl 格式)
本附件提供:对 CMU-MOSEI 公开数据中有效的部分视频片段进行处理后、可
直接用于建模的特征文件,包含对齐和非对齐各 4850 条样本,以及样本标签对应
的 excel 表格。样本按训练/验证/测试划分组织,包括文本、语音、视觉三模态各
自的特征矩阵、标签映射、维度定义与存储格式说明(数据形式和解释见后面补充
说明)。参赛队可以根据情况选择附件 2 中的数据开展问题 2 和问题 3 建模。
3.附件 3:模态局部缺失专项测试集
包含若干无标签样本对应的处理后的带有随机模态缺失的特征文件,用于模态
局部缺失专项测试。其中,模态缺失指样本局部信息不可用,即一个或多个模态中
存在特征值全部为零的随机连续序列区间。
4.附件 4:可解释性专项测试集
包含若干无标签真实场景视频及对应的三模态特征文件,三模态信息完整。特
征字段、张量维度和时序组织方式与附件 2 保持一致;配套原始视频用于将模型给
出的关键证据时间位置回看至文本、语音或视觉片段。
注:以上特征文件分别提供与附件 2 aligned_50.pkl 和 unaligned_50.pkl 对应的
两套版本。参赛队应在训练、验证和专项测试中保持同一特征版本和同一输入接口。
三、需要解决的问题
本题依次要求完成多模态特征提取与时序对齐、模态缺失条件下的鲁棒情感预
测、可解释性情感预测三项建模任务。
问题 1:多模态情感特征提取与时序对齐的数学建模
基于附件 1 提供的原始视频,建立文本、语音、视觉三模态情感特征提取与时
序对齐的数学模型,模型须明确原始样本的处理流程、各模态情感特征的定义与提
取方法、跨模态时序对齐的规则与实现逻辑。参赛队可使用开源工具或预训练模型
完成特征提取基础处理。
参赛队在问题 1 中自行生成的特征文件需达到如下要求:一是原始样本覆盖完
整性,包括样本编号、模态文件与输出特征是否一一对应;二是时序组织的可核验
性,包括序列位置、有效长度、填充规则及其与原始素材的对应记录;三是方法的
合理性与可复现性,包括特征提取工具、关键参数、处理日志和复现实验说明。
问题 2:模态信息缺失条件下的情感预测建模与验证
针对局部时段模态信息缺失的复杂场景,构建鲁棒性多模态情感预测模型。模
态局部缺失指文本、语音或视觉中的部分连续时间段不可用,并不表示整个模态完
全不存在。模型需在局部模态信息不完整的条件下稳定输出情感极性与情感强度预
测,并分析缺失模态类型、缺失位置、缺失时长对预测性能的影响规律。
使用所建立的鲁棒性情感预测模型,对附件 3 中的测试样本完成推理预测,按
第四条结果与提交说明提交预测结果文件。
问题 3:可解释性情感预测建模与验证
针对三模态信息完整但决策依据难以追溯的场景,构建可解释性多模态情感预
测模型。可解释性指模型在给出情感预测结果的同时,能够以可量化、可复核的方
式说明本次判断主要依据了哪些输入信息,包括不同模态在当前样本中的作用差异、
对预测起主要作用的模态,以及各模态中与判断密切相关的局部片段。
模型需同时输出情感极性、情感强度预测结果,以及主要参考模态、模态作用
程度、关键证据定位等可解释性分析结果,关键证据需可对应至原始文本片段、语
音时段或视觉关键帧。
使用所建立的可解释性情感预测模型,对附件 4 中的测试样本完成推理预测,
按第四条要求格式提交预测结果与解释文件。
说明:问题 2 与问题 3 的模型均在附件 2 提供的训练集上学习模型参数,在验
证集上选择模型结构、超参数与决策阈值;附件 3、附件 4 为无标签专项测试集,
仅用于相应专项模型的最终推理预测。分类任务(情感极性)采用准确率
(Accuracy)、F1 值评价,回归任务(情感强度)采用平均绝对误差(MAE)、
皮尔逊相关系数评价。
四、结果与提交说明
所有提交内容需严格对应本题各问题的任务要求,在参赛论文中完整、规范呈
现。
(一)论文正文要求
参赛论文需按照竞赛统一格式规范撰写,所有问题的方法原理、建模过程、实
验结果与分析结论均须系统纳入正文,不得仅存放于附件。
1.论文组织要求
论文应依次回应问题 1、问题 2 和问题 3,并在每个问题中说明数据输入、方
法或模型、求解过程、评价指标和主要结果。参赛队可自主安排理论推导、算法细
节、消融实验和创新讨论,但不得省略题目规定的核心输出、结果文件说明和必要
的实验依据。
2.问题 1 相关内容
(1)特征提取与时序对齐整体方案:说明文本、语音、视觉三类模态的特征定义、
提取方法与所用工具,阐述时序对齐的核心规则与跨模态时间对应逻辑,完整呈现
从原始素材到标准化时序特征的处理流程。
(2)特征文件规范与全量结果汇总:说明自生成特征文件的存储格式、组织结构
与读取方式;以表格形式汇总附件 1 全部 100 条样本的特征提取结果,包含样本编
号、模态类型、原始有效时长、特征维度、对齐粒度等关键信息,确保全量结果可
追溯、可核验。全量原始特征文件随附件提交。
(3)典型样本验证:选取至少 1 个典型样本,展示其文本片段、对应语音时段、
视频帧段与三类特征的对应关系,呈现时序对齐效果。
(4)方案可复现性说明:标注所用工具与模型的版本、核心参数及完整运行流程,
保障结果可复现。
3.问题 2 相关内容
(1)鲁棒性模型的建模原理、网络结构、目标函数、训练方案与关键参数;
(2)缺失模态类型、缺失率对预测性能影响的规律分析与消融实验结果;
(3)附件 3 测试集的全量预测结果汇总与结果展示;
(4)验证集上的基础性能评价、可视化分析与错误归因结论。
4.问题 3 相关内容
(1)可解释性模型的建模原理、网络结构、目标函数、训练方案与关键参数;
(2)典型样本解释卡(含预测结果、三模态作用程度、主要参考模态及关键证据
定位);
(3)主要参考模态内局部片段重要性分布可视化、三模态作用差异对比分析;
(4)附件 4 测试集的全量预测与解释结果汇总;
(5)验证集上的基础性能评价、可视化分析与错误归因结论。
(二)附件提交要求
1.可复现核心材料:问题 1 自主生成的 100 条样本多模态时序特征文件,问题
2 和问题 3 两类专项模型的核心代码、说明文档、模型参数文件、配置文件与运行
环境说明。代码须附带详细运行说明、数据集处理规则与参数配置,确保实验结果
可完整复现。
2.专项测试结果文件:附件 3 测试集预测结果 CSV 文件、附件 4 测试集预测与
解释结果 CSV 文件,文件命名清晰规范。
总附件大小须符合竞赛统一要求(≤50MB),所有提交材料中严禁出现参赛
单位、队员姓名、队伍编号等身份信息,违者按竞赛违规处理。
五、补充说明
1.数据与工具使用规范
本题所有建模任务均以赛题提供的 CMU-MOSEI 系列数据为唯一数据来源,参
赛队不得引入其他任何公开或私有情感数据集参与模型训练、微调、参数优化、阈
值选择或结果统计。
参赛队可在合理论证前提下,对附件 2 的训练集、验证集数据开展清洗、重采
样、标准化等基础预处理,也可基于该数据开展消融实验与规律分析。
本题所有问题的建模工作均允许使用开源特征提取工具、预训练模型与开源算
法框架,但须在论文中统一标注所涉工具与模型的名称、版本号及核心参数设置,
明确其使用场景。
2.关于附件 2 的数据解释和样例说明
CMU-MOSEI 数 据 集 提 供 两 个 已 处 理 特 征 文 件 : aligned_50.pkl 和
unaligned_50.pkl,选手可自行选择使用,但须在训练、验证和专项测试中保持同一
版本。二者均以 Python Pickle 字典格式保存。
文件结构:最外层按数据用途划分为`train`、`valid`、`test`三个顶层键。进入
任一顶层键后,文件按字段集中保存该划分下的全部样本,即`data['train']['audio']`
表示训练集所有样本的语音特征。各字段中同一索引位置的取值共同描述同一条样
本,读取方式为`data['train'][字段名][j]`(`j`为样本索引,Python 从 0 开始计数),
而非`data['train'][j][字段名]`。
特征形状与时序组织:两个文件的矩阵形状如下表所示,第一维`N`为当前划
分的样本数,后两维分别为序列位置数和特征维度。
表 1 附件 2 两类特征文件的矩阵形状
文件 文本特征 语音特征 视觉特征 时序含义
三模态均按最多 50 个相互对应
aligned_50.pkl (N, 50, 768) (N, 50, 74) (N, 50, 35)
的序列位置组织
文本保持 50 个位置;音、保留
unaligned_50.pkl (N, 50, 768) (N, 500, 74) (N, 500, 35) 独立时 序,有效长 度见
audio_lengths、vision_lengths。
文件名中的 50 表示统一采用的最大序列位置数。所有样本存储形状相同,有
效位置数可能小于最大长度,其余位置按文件规则填充。语音特征维度 74 由原始
提取流程确定,本赛题将其视为统一输入维度,不对其每一维的具体物理含义作额
外假定。
字段含义:主要字段如表 2 所示。
表 2 附件 2 主要字段含义
字段 含义 使用说明
字符串,格式为
id 唯一标识一条建模样本
video_id$_$clip_id
raw_text 原始英文转写文本 用于阅读核查,非数值特征
已经预计算的 BERT 类文本表示,可直
text 连续文本特征,50×768
接输入模型
词元编号、注意力掩码和分段编号,需
BERT 三路整数输入,
text_bert 经 BERT 编码后使用;与 text 二选一,不
3×50
构成第四种模态
audio / vision 语音或视觉时序特征 形状见表 1
字段 含义 使用说明
annotations /
文字极性、三分类标
classification_labels 情感标注定义与问题 2 一致
签、连续情感强度
/ regression_labels
读取示例:
附件 2 保存的是处理后的数值特征和标签,不保存原始视频帧、音频波形或逐
词时间戳。为帮助参赛者理解这些字段在原始视频中的来源及其时间关系,图 3 用
一个独立的多模态样本示意同一时间轴上的视觉、音频、文本和标签信息。
图 1 多模态视频片段的原始信号、文本时间戳与情感标签示例
参考文献
[1] Zhang S, Yang Y, Chen C, et al. Deep learning-based multimodal emotion recognition
from audio, visual, and text modalities: A systematic review of recent advancements and
future prospects[J]. Expert Systems with Applications, 2024, 237: 121692.
[2] Qiu X, Fang Y, Zhou Q, et al. Beyond Missing Modalities: Hypergraph Conditioned
Diffusion for Uncertainty-Aware Multimodal Emotion Recognition[C]//Proceedings of the
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2026.
[3] Yang Z, Li M. Factorize, Reconstruct, Enhance: A Unified Framework for Multimodal
Sentiment Analysis[C]//Proceedings of the IEEE/CVF Conference on Computer Vision
and Pattern Recognition (CVPR). 2026.
[4] Zhuang Y, Liu M, Bai W, et al. CMAD: Correlation-Aware and Modalities-Aware
Distillation for Multimodal Sentiment Analysis with Missing Modalities[C]//Proceedings
of the IEEE/CVF International Conference on Computer Vision (ICCV). 2025.
[5] Zhu A, Hu M, Wang X, et al. Proxy-Driven Robust Multimodal Sentiment Analysis
with Incomplete Data[C]//Proceedings of the 63rd Annual Meeting of the Association for
Computational Linguistics (ACL). 2025.
[6] Mai S, Han S. Learning Invariant Modality Representation for Robust Multimodal
Learning from a Causal Inference Perspective[C]//Proceedings of the 64th Annual Meeting
of the Association for Computational Linguistics (ACL). 2026.
[7] Fang Y, Huang W, Wan G, et al. EMOE: Modality-Specific Enhanced Dynamic
Emotion Experts[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and
Pattern Recognition (CVPR). 2025.
[8] Wan J, Gong C, Fu G. Locate and Explain: Joint Multimodal Emotion Cause Extraction
and Summarization in Conversation[C]//Proceedings of the 64th Annual Meeting of the
Association for Computational Linguistics (ACL). 2026.
[9] Mai S, Han S. CaReFlow: Cyclic Adaptive Rectified Flow for Multimodal
Fusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern
Recognition (CVPR). 2026.
[10] 王楠,王淇,欧阳丹彤。基于知识蒸馏与动态调整机制的多模态情感分析模型
[J]. 计算机学报,2025, 48 (8): 1923-1942.

Some files were not shown because too many files have changed in this diff Show More