Files

14 KiB

1moderolescaleshapendimdtypeverified_pairsminimum_direct_errormedian_direct_errormaximum_cosineminimum_scaled_errorpairs_under_5pct_direct_errorpairs_under_20pct_direct_error
2tensormodel.language_model.layers.*.input_layernorm.weightwhole(2048,)1BF16740.139121081919607260.36180188968632330.99708574921504440.07628898159153481010
3tensormodel.language_model.layers.*.linear_attn.A_logwhole(16,)1F32740.279031954865745560.52269475509125040.96275913880532680.270360575245033400
4tensormodel.language_model.layers.*.linear_attn.conv1d.weightwhole(6144, 1, 4)3BF16741.0989158027131691.61667026790055910.028079990778775030.999577680656060500
5tensormodel.language_model.layers.*.linear_attn.dt_biaswhole(16,)1BF16740.58851146170581971.00445284219408080.80921287139569940.587515556191942200
6tensormodel.language_model.layers.*.linear_attn.in_proj_a.weightwhole(16, 2048)2BF16741.13499521895345111.21681846424637370.297591718737855670.954693232896645600
7tensormodel.language_model.layers.*.linear_attn.in_proj_b.weightwhole(16, 2048)2BF16741.08589013789407461.22259628116522910.32117699910791060.947019184200635600
8tensormodel.language_model.layers.*.linear_attn.in_proj_qkv.weightwhole(6144, 2048)2BF16741.27233498170011531.46352964205279030.00101353533584107140.999998924051265900
9tensormodel.language_model.layers.*.linear_attn.in_proj_z.weightwhole(2048, 2048)2BF16741.29108305421636321.45378584035599450.0135318130331016010.999908440826477500
10tensormodel.language_model.layers.*.linear_attn.norm.weightwhole(128,)1F32740.09962848461018260.144539211790483460.99620832159442170.08699988497707883074
11tensormodel.language_model.layers.*.linear_attn.out_proj.weightwhole(2048, 2048)2BF16741.33580196136744681.40335561079590.00165859438638679170.999998624531384900
12tensormodel.language_model.layers.*.mlp.down_proj.weightwhole(2048, 6144)2BF16741.2893246167894221.34219675411502060.00066281697661898730.999999780336803700
13tensormodel.language_model.layers.*.mlp.gate_proj.weightwhole(6144, 2048)2BF16741.1562017814391961.67201960788715720.0099224189799254940.999950771589075200
14tensormodel.language_model.layers.*.mlp.up_proj.weightwhole(6144, 2048)2BF16731.22106134050904421.32936689663597420.00087376016405094510.999999618271515100
15tensormodel.language_model.layers.*.post_attention_layernorm.weightwhole(2048,)1BF16730.083892050163051050.335508480218248640.99820243975747740.05993237240606879015
16tensormodel.language_model.layers.*.self_attn.k_norm.weightwhole(256,)1BF16150.42203548694097630.57307671727198820.91093269551260660.4125549954201702700
17tensormodel.language_model.layers.*.self_attn.k_proj.weightwhole(512, 2048)2BF16151.32165007967171341.3909286776086350.00299253374180675740.999991587596552300
18tensormodel.language_model.layers.*.self_attn.o_proj.weightwhole(2048, 2048)2BF16151.18179600075844581.35821790331685470.0006617897590010420.999999781017133500
19tensormodel.language_model.layers.*.self_attn.q_norm.weightwhole(256,)1BF16150.394040647924993550.5439362007952080.92305095427081480.3846777038245628400
20tensormodel.language_model.layers.*.self_attn.q_proj.weightwhole(4096, 2048)2BF16151.28226249002734651.4523255074752760.0311039706888872580.999516154450434500
21tensormodel.language_model.layers.*.self_attn.v_proj.weightwhole(512, 2048)2BF16151.13815356635104831.2798557298834910.0042749339659219050.999990862428045900
22tensormodel.visual.layers.*.attn.proj.biaswhole(1024,)1BF16730.66809796842798161.03268438043278570.76478431410321590.64428639043648400
23tensormodel.visual.layers.*.attn.proj.weightwhole(1024, 1024)2BF16731.22631099923204181.4940304952785360.00228937795859618250.999997379370847500
24tensormodel.visual.layers.*.attn.qkv.biaswhole(3072,)1BF16731.15082350972250061.28953586068968870.078800748878007160.996890386138950200
25tensormodel.visual.layers.*.attn.qkv.weightwhole(3072, 1024)2BF16731.19194095807913581.35283526122666740.00124202254344408720.999997251271210100
26tensormodel.visual.layers.*.mlp.linear_fc1.biaswhole(4096,)1BF16730.34232551204141030.45066084579667910.9428322268754030.3332677481622975400
27tensormodel.visual.layers.*.mlp.linear_fc1.weightwhole(4096, 1024)2BF16731.23184201085776861.56416455897597540.00249982209904134330.99999687543985500
28tensormodel.visual.layers.*.mlp.linear_fc2.biaswhole(1024,)1BF16730.68566257420299311.0089781071266440.79221473964057180.610242416008770500
29tensormodel.visual.layers.*.mlp.linear_fc2.weightwhole(1024, 4096)2BF16731.25838004203856221.36604553222984020.00191110096578416380.999996218510318200
30tensormodel.visual.layers.*.norm1.biaswhole(1024,)1BF16730.251159439476569930.6908838548091810.96912594744868830.2465662141934337700
31tensormodel.visual.layers.*.norm1.weightwhole(1024,)1BF16730.048237311661066430.145045063064817240.99922803584221510.03928526933736397158
32tensormodel.visual.layers.*.norm2.biaswhole(1024,)1BF16730.462457161110339130.96059892019127570.92556646050940130.378585165029084300
33tensormodel.visual.layers.*.norm2.weightwhole(1024,)1BF16730.081096034914990540.255849698123852940.9980975721771660.06165416785625297029
34blockmodel.language_model.embed_tokens.weight2d-128(128, 128)2BF162741.09714316559198341.1949961656344020.383713842853735050.923452049000065400
35blockmodel.language_model.embed_tokens.weight2d-64(64, 64)2BF162740.78671925209976381.22511361727129750.64175200960300890.766912223250157600
36blockmodel.language_model.layers.*.linear_attn.conv1d.weight3d-axis0-slice(1, 1, 4)3BF162740.0300154707143710440.141073632591451560.99997646939729760.00686007665521580512264
37blockmodel.language_model.layers.*.linear_attn.conv1d.weight3d-volume(4, 1, 4)3BF162740.204348666388623530.65113268823758050.99395307258210220.1098056897642306400
38blockmodel.language_model.layers.*.linear_attn.in_proj_qkv.weight2d-128(128, 128)2BF162741.08591943819046041.16727681872903920.0225064912896195680.999746696843670600
39blockmodel.language_model.layers.*.linear_attn.in_proj_qkv.weight2d-64(64, 64)2BF162741.0904316051852191.87708569974590730.0495804324660010160.998770134073143200
40blockmodel.language_model.layers.*.linear_attn.in_proj_z.weight2d-128(128, 128)2BF162741.11049489128624961.87738648485714750.0234758509826428380.999724404233807300
41blockmodel.language_model.layers.*.linear_attn.in_proj_z.weight2d-64(64, 64)2BF162741.09916612446335731.85144860501249920.0434855288740225650.998841835987462900
42blockmodel.language_model.layers.*.linear_attn.out_proj.weight2d-128(128, 128)2BF162741.19518817599831941.65949421349807840.0204685456169650080.999790497374488100
43blockmodel.language_model.layers.*.linear_attn.out_proj.weight2d-64(64, 64)2BF162741.20723927846511961.55826775357150240.044540567003889260.999007576493277600
44blockmodel.language_model.layers.*.mlp.down_proj.weight2d-128(128, 128)2BF162741.25826173225798831.30900999386021420.0235712308853066840.999722159939726200
45blockmodel.language_model.layers.*.mlp.down_proj.weight2d-64(64, 64)2BF162741.2458524337732061.29406291244666380.0558581584661288560.998438714259805200
46blockmodel.language_model.layers.*.mlp.gate_proj.weight2d-128(128, 128)2BF162741.14614923722416061.19662550746018010.0219463621699511230.999759149589292500
47blockmodel.language_model.layers.*.mlp.gate_proj.weight2d-64(64, 64)2BF162741.12570857315927421.17607533993126710.059644058019983050.998219708452457300
48blockmodel.language_model.layers.*.mlp.up_proj.weight2d-128(128, 128)2BF162741.19388965374658221.57437360467051040.0247467194028587430.999693753045799700
49blockmodel.language_model.layers.*.mlp.up_proj.weight2d-64(64, 64)2BF162741.19828690324878641.28167743755540390.055394643180264740.998464537931583900
50blockmodel.language_model.layers.*.self_attn.k_proj.weight2d-128(128, 128)2BF162741.14328715431965771.1957610623400630.0297706416636919940.999556756214939600
51blockmodel.language_model.layers.*.self_attn.k_proj.weight2d-64(64, 64)2BF162741.0898438005598862.07693224728822570.0477141667360517840.998861030520604600
52blockmodel.language_model.layers.*.self_attn.o_proj.weight2d-128(128, 128)2BF162741.07598653693893141.95526679220727950.37173124139663410.928340392404487200
53blockmodel.language_model.layers.*.self_attn.o_proj.weight2d-64(64, 64)2BF162741.0901856360687111.51838312369606140.40061430233579770.916246790315798100
54blockmodel.language_model.layers.*.self_attn.q_proj.weight2d-128(128, 128)2BF162740.6810621202927951.61107916550113070.7665897298400590.642137201930977800
55blockmodel.language_model.layers.*.self_attn.q_proj.weight2d-64(64, 64)2BF162741.07319679529521821.1325864371183160.144462005238312850.989510348123012600
56blockmodel.language_model.layers.*.self_attn.v_proj.weight2d-128(128, 128)2BF162741.07037722374150032.30645129183926570.0255832253599119560.999672695726048100
57blockmodel.language_model.layers.*.self_attn.v_proj.weight2d-64(64, 64)2BF162741.06159395390515862.2421448038148840.059796412830882670.998210593518501400
58blockmodel.visual.layers.*.attn.proj.weight2d-128(128, 128)2BF162741.09648982120688231.249176675930920.0303776660737441230.999538492207234700
59blockmodel.visual.layers.*.attn.proj.weight2d-64(64, 64)2BF162741.09293197765574361.20367093662283330.041908888256001630.999121436605754100
60blockmodel.visual.layers.*.attn.qkv.weight2d-128(128, 128)2BF162741.05256748646845731.14726291661163970.34972906283020710.936850886006895500
61blockmodel.visual.layers.*.attn.qkv.weight2d-64(64, 64)2BF162741.02763029406764632.1539829409967720.050393064081329990.99872946241336800
62blockmodel.visual.layers.*.mlp.linear_fc1.weight2d-128(128, 128)2BF162741.17840376712194231.26100173576509040.054462052044824530.99851584107968300
63blockmodel.visual.layers.*.mlp.linear_fc1.weight2d-64(64, 64)2BF162741.19309327644690491.56352164502282220.057428885876377690.998349599622796500
64blockmodel.visual.layers.*.mlp.linear_fc2.weight2d-128(128, 128)2BF162741.19769862001561631.58642887357831760.0281903877598622470.999602572044484400
65blockmodel.visual.layers.*.mlp.linear_fc2.weight2d-64(64, 64)2BF162741.09735762409373621.28210713291342020.0527527944279967340.998607601953859300
66blockmodel.visual.merger.linear_fc1.weight2d-128(128, 128)2BF162740.56904823770691831.38817332011640150.83944678751469130.543441892874725600
67blockmodel.visual.merger.linear_fc1.weight2d-64(64, 64)2BF162740.57041601778384911.3837152887818340.83827413555938430.545249001514140200
68blockmodel.visual.merger.linear_fc2.weight2d-128(128, 128)2BF162741.3616028809229121.40765210053970160.0227168373032965370.999741939353819300
69blockmodel.visual.merger.linear_fc2.weight2d-64(64, 64)2BF162741.31723352522289551.38852383018451730.0549167673767164060.998490935692904400
70blockmodel.visual.pos_embed.weight2d-128(128, 128)2BF162730.18949074790984470.83688140652083690.98200687566663210.1888451644693597301
71blockmodel.visual.pos_embed.weight2d-64(64, 64)2BF162730.094944303100750260.57610562307428750.99605025668255860.0887912504845429507
72blockmtp.fc.weight2d-128(128, 128)2BF162730.397951581875416741.30091984188304030.92198776956759620.387218998459230100
73blockmtp.fc.weight2d-64(64, 64)2BF162730.28394094426924661.21878888243739850.95981325316373730.280639482345666500
74blockmtp.layers.*.mlp.down_proj.weight2d-128(128, 128)2BF162731.3100740095351211.42150540260602960.0275546716852172640.999620297947335500
75blockmtp.layers.*.mlp.down_proj.weight2d-64(64, 64)2BF162731.2498075789421721.4700990010925120.0444347626047828450.999012288148777700
76blockmtp.layers.*.mlp.gate_proj.weight2d-128(128, 128)2BF162731.30148022902135051.3616819965635590.033739963989720230.999430645332617600
77blockmtp.layers.*.mlp.gate_proj.weight2d-64(64, 64)2BF162731.25646811408446731.46259573755341470.05421569467959960.998529247668994600
78blockmtp.layers.*.mlp.up_proj.weight2d-128(128, 128)2BF162731.33148151361286751.40836788846501570.0297583005333780340.999557123704976200
79blockmtp.layers.*.mlp.up_proj.weight2d-64(64, 64)2BF162731.29390279958913611.35584024184669080.062464930860104710.998047159413142300
80blockmtp.layers.*.self_attn.k_proj.weight2d-128(128, 128)2BF162731.22110713469066751.65684604784646060.027739416699410690.999615188340581600
81blockmtp.layers.*.self_attn.k_proj.weight2d-64(64, 64)2BF162731.1007499720374442.3212676168968190.063913983133134510.997955411208365300
82blockmtp.layers.*.self_attn.o_proj.weight2d-128(128, 128)2BF162730.88210454274679731.2061705866120790.57673168012398480.816933638150226800
83blockmtp.layers.*.self_attn.o_proj.weight2d-64(64, 64)2BF162730.94255228499647151.26966011130993860.52756645803834220.84951376230928600
84blockmtp.layers.*.self_attn.q_proj.weight2d-128(128, 128)2BF162731.02377778191180372.83715505496135560.48986646552387830.871797479897219600
85blockmtp.layers.*.self_attn.q_proj.weight2d-64(64, 64)2BF162731.0082804962878952.5243954562058660.54304023633068290.839706676004136500
86blockmtp.layers.*.self_attn.v_proj.weight2d-128(128, 128)2BF162731.35346773091810361.4023894379519310.0248380606915191250.999691487780647400
87blockmtp.layers.*.self_attn.v_proj.weight2d-64(64, 64)2BF162731.2698342740853241.33817272153333720.057985526492290150.998317423827417700