Download research_code/pretraining_mamba2_outer/native_initialization_seed1.json from Nicholas0228/tail-drop-5b-dense: direct link, hf CLI and curl.
- Browser
- Download file 17.4 kB
-
https://huggingface.co/Nicholas0228/tail-drop-5b-dense/resolve/main/research_code/pretraining_mamba2_outer/native_initialization_seed1.json
- Command line
-
hf download hf://Nicholas0228/tail-drop-5b-dense/research_code/pretraining_mamba2_outer/native_initialization_seed1.json
-
curl -L -o native_initialization_seed1.json https://huggingface.co/Nicholas0228/tail-drop-5b-dense/resolve/main/research_code/pretraining_mamba2_outer/native_initialization_seed1.json
17.4 kB
| { | |
| "schema": 1, | |
| "description": "Native upstream seed-1 scratch initialization, frozen before any scientific update; only host-dependent last-bit transcendental rounding is normalized.", | |
| "mamba_revision": "d7b1ceb3c367ec9022925e812f507bcf706937c6", | |
| "expected_model_sha256": "45edc93a6f0d060f10820c0ba626141d9739c7a0d16110e291c8175d50821207", | |
| "source_artifact": "art-pip6e", | |
| "source_artifact_sha256": "3e828c7855e009ff782a4cc14e9cd086b71ef74302f78ee07d2e2adfeb15dc58", | |
| "original_audit_artifact": "art-ci8n5", | |
| "cpu_reproduction_matches_original_audit": true, | |
| "maximum_allowed_ulp": 1, | |
| "config": { | |
| "seed": 1, | |
| "layers": 28, | |
| "dense_ends": 4, | |
| "hidden": 1024, | |
| "mamba_d_state": 128, | |
| "mamba_d_conv": 4, | |
| "mamba_expand": 2, | |
| "mamba_headdim": 64 | |
| }, | |
| "vectors": { | |
| "layers.0.attn.dt_bias": { | |
| "values": [ | |
| -3.7554869651794434, | |
| -2.89456844329834, | |
| -4.961036205291748, | |
| -2.7677934169769287, | |
| -6.515177249908447, | |
| -5.1998515129089355, | |
| -3.911940813064575, | |
| -4.382295608520508, | |
| -6.211663246154785, | |
| -4.00882625579834, | |
| -4.266192436218262, | |
| -5.378808498382568, | |
| -3.7056217193603516, | |
| -2.303309917449951, | |
| -4.29649019241333, | |
| -5.096563339233398, | |
| -3.6716136932373047, | |
| -4.715573787689209, | |
| -2.9057633876800537, | |
| -3.0622637271881104, | |
| -5.064349174499512, | |
| -6.219362735748291, | |
| -3.725118637084961, | |
| -3.838928461074829, | |
| -3.3714468479156494, | |
| -3.410841703414917, | |
| -5.164243221282959, | |
| -6.393595218658447, | |
| -6.7128448486328125, | |
| -4.139652729034424, | |
| -2.678271770477295, | |
| -5.7577619552612305 | |
| ], | |
| "sha256": "1801b6a8b5c57d9837404ba9322af80b14e500fa6d180e009698db655fdbefde" | |
| }, | |
| "layers.0.attn.A_log": { | |
| "values": [ | |
| 1.730581521987915, | |
| 2.0241286754608154, | |
| 1.5153424739837646, | |
| 2.6725432872772217, | |
| 2.7714650630950928, | |
| 2.517014741897583, | |
| 0.28924328088760376, | |
| 2.0145440101623535, | |
| 2.1449573040008545, | |
| 2.2256762981414795, | |
| 2.371535539627075, | |
| 2.271897792816162, | |
| 2.3200573921203613, | |
| 2.6965723037719727, | |
| 2.6953208446502686, | |
| 0.6480419039726257, | |
| 2.7003047466278076, | |
| 2.2425732612609863, | |
| 2.723991870880127, | |
| 1.809671401977539, | |
| 2.524787664413452, | |
| 2.6669509410858154, | |
| 1.794349193572998, | |
| 2.3228650093078613, | |
| 0.5621227622032166, | |
| 2.1177496910095215, | |
| 2.360903263092041, | |
| 2.4662232398986816, | |
| 1.4346588850021362, | |
| 2.5053441524505615, | |
| 2.6499335765838623, | |
| 2.743143081665039 | |
| ], | |
| "sha256": "b124b79c8deafd97c58c10dd9a3fd598fc7facaee1eed6710b4258fbe123414f" | |
| }, | |
| "layers.1.attn.dt_bias": { | |
| "values": [ | |
| -5.221463680267334, | |
| -4.339187145233154, | |
| -6.700262069702148, | |
| -3.0825932025909424, | |
| -2.8782331943511963, | |
| -6.509949684143066, | |
| -5.656719207763672, | |
| -5.35589599609375, | |
| -3.4449169635772705, | |
| -4.370367527008057, | |
| -6.304828643798828, | |
| -5.196429252624512, | |
| -6.284307479858398, | |
| -5.914492607116699, | |
| -5.1533708572387695, | |
| -5.737362861633301, | |
| -5.937115669250488, | |
| -2.414966106414795, | |
| -3.1989028453826904, | |
| -2.6013505458831787, | |
| -3.636624336242676, | |
| -4.960628509521484, | |
| -3.3915233612060547, | |
| -3.1007297039031982, | |
| -3.3484277725219727, | |
| -4.0297770500183105, | |
| -3.778918981552124, | |
| -3.6957545280456543, | |
| -6.079647541046143, | |
| -3.487917900085449, | |
| -2.5595157146453857, | |
| -4.887794494628906 | |
| ], | |
| "sha256": "7ef0e8e54843a7e7a914522fed3d8e7d4c73bc8c1491a26d878992ac08fedd39" | |
| }, | |
| "layers.1.attn.A_log": { | |
| "values": [ | |
| 2.7357425689697266, | |
| 2.6726226806640625, | |
| 2.255178928375244, | |
| 2.1083714962005615, | |
| 2.361412525177002, | |
| 1.1302341222763062, | |
| 1.1614139080047607, | |
| 2.5925230979919434, | |
| 2.2649929523468018, | |
| 1.3362315893173218, | |
| 1.8114635944366455, | |
| 1.099045753479004, | |
| 2.345737934112549, | |
| 2.239605188369751, | |
| 1.8964868783950806, | |
| 2.573024272918701, | |
| 2.0770387649536133, | |
| 0.971108078956604, | |
| 2.523940324783325, | |
| 2.759037733078003, | |
| 1.6339459419250488, | |
| 1.4179774522781372, | |
| 2.4383842945098877, | |
| 2.161860942840576, | |
| 2.090899705886841, | |
| 0.11402231454849243, | |
| 1.1591370105743408, | |
| 1.240274429321289, | |
| 0.8050137758255005, | |
| 2.171171188354492, | |
| 1.9447535276412964, | |
| 2.592294454574585 | |
| ], | |
| "sha256": "6182023a04600d400ad96f261cd7ca8bab4e33400c8e05e0c585c38e43d0291d" | |
| }, | |
| "layers.2.attn.dt_bias": { | |
| "values": [ | |
| -5.566741466522217, | |
| -6.232064723968506, | |
| -6.336475849151611, | |
| -5.357068061828613, | |
| -5.011998176574707, | |
| -2.8345494270324707, | |
| -5.813052177429199, | |
| -6.505034923553467, | |
| -5.366116523742676, | |
| -6.872137546539307, | |
| -5.947303295135498, | |
| -4.408282279968262, | |
| -6.898733139038086, | |
| -4.002575874328613, | |
| -2.5516865253448486, | |
| -5.7927141189575195, | |
| -4.391349792480469, | |
| -3.4711239337921143, | |
| -3.211191415786743, | |
| -4.4015326499938965, | |
| -5.708187103271484, | |
| -5.7143425941467285, | |
| -6.16478157043457, | |
| -4.655795574188232, | |
| -2.724674701690674, | |
| -6.641811847686768, | |
| -3.9453041553497314, | |
| -6.235372066497803, | |
| -4.929118633270264, | |
| -6.420401096343994, | |
| -6.283128261566162, | |
| -2.8920445442199707 | |
| ], | |
| "sha256": "a48e3e67c4da594901346919aec184b191c718a4595970dbbf215bb1a806d0aa" | |
| }, | |
| "layers.2.attn.A_log": { | |
| "values": [ | |
| 2.676557779312134, | |
| 2.2557921409606934, | |
| 1.8576364517211914, | |
| 1.846186637878418, | |
| 2.4556260108947754, | |
| 2.23270320892334, | |
| 0.9019314646720886, | |
| 2.4762966632843018, | |
| 2.255948543548584, | |
| 2.500607967376709, | |
| 2.297694683074951, | |
| 1.1867787837982178, | |
| 2.189927339553833, | |
| 1.8986485004425049, | |
| 2.361891031265259, | |
| 2.4243123531341553, | |
| 2.483764171600342, | |
| 1.205756425857544, | |
| 1.912316083908081, | |
| 2.6718246936798096, | |
| 1.9304218292236328, | |
| 2.472172260284424, | |
| 0.8554327487945557, | |
| 2.0944204330444336, | |
| 2.76684832572937, | |
| 1.5109045505523682, | |
| 2.641317129135132, | |
| 2.19612717628479, | |
| 1.46620774269104, | |
| 1.1964620351791382, | |
| 0.27422037720680237, | |
| 2.1852316856384277 | |
| ], | |
| "sha256": "ecad43ed886d4ae247359aab831ee83df735379b05b1079d66ffb02d9e830690" | |
| }, | |
| "layers.3.attn.dt_bias": { | |
| "values": [ | |
| -4.804631233215332, | |
| -3.899235963821411, | |
| -2.488706111907959, | |
| -5.67948579788208, | |
| -6.092596054077148, | |
| -3.723757743835449, | |
| -6.866397380828857, | |
| -5.145601272583008, | |
| -4.019090175628662, | |
| -4.736316680908203, | |
| -5.6807451248168945, | |
| -6.823186874389648, | |
| -3.815753936767578, | |
| -2.9954864978790283, | |
| -5.326148509979248, | |
| -6.220427989959717, | |
| -3.9218523502349854, | |
| -2.9459688663482666, | |
| -6.578087329864502, | |
| -5.847710132598877, | |
| -6.645875453948975, | |
| -3.6275055408477783, | |
| -4.552668571472168, | |
| -3.443135976791382, | |
| -5.74568510055542, | |
| -3.2314822673797607, | |
| -3.6912028789520264, | |
| -4.7605671882629395, | |
| -5.25062894821167, | |
| -4.338469982147217, | |
| -2.7038638591766357, | |
| -3.796093225479126 | |
| ], | |
| "sha256": "4578e6afe45549dba32cbd08ad7fab3eb7e8ca31c474b21b2b08670d6c7d5943" | |
| }, | |
| "layers.3.attn.A_log": { | |
| "values": [ | |
| 1.3121625185012817, | |
| 2.744684934616089, | |
| 1.024592399597168, | |
| 2.330152750015259, | |
| 2.7133681774139404, | |
| 1.9468700885772705, | |
| 1.142308235168457, | |
| 2.3645870685577393, | |
| 1.619040608406067, | |
| 1.5045323371887207, | |
| 1.3756338357925415, | |
| 2.5798349380493164, | |
| 2.558901786804199, | |
| 2.6624372005462646, | |
| 2.52777099609375, | |
| 1.9410635232925415, | |
| 1.687807321548462, | |
| 1.5303276777267456, | |
| 2.766721725463867, | |
| 2.4653491973876953, | |
| 2.2931737899780273, | |
| 2.636115550994873, | |
| 2.5432093143463135, | |
| 1.9790256023406982, | |
| 2.6622986793518066, | |
| 1.9605681896209717, | |
| 1.0559009313583374, | |
| 0.40600109100341797, | |
| 2.5760157108306885, | |
| 2.4247400760650635, | |
| 1.5102198123931885, | |
| 2.5201799869537354 | |
| ], | |
| "sha256": "13efd71259059e19e0ab0628ef9c130ea4771b3db92139f47844d45d500ff60b" | |
| }, | |
| "layers.24.attn.dt_bias": { | |
| "values": [ | |
| -6.811822891235352, | |
| -2.4707183837890625, | |
| -4.037230014801025, | |
| -3.126941204071045, | |
| -3.7725186347961426, | |
| -2.506024122238159, | |
| -6.060723304748535, | |
| -6.299108505249023, | |
| -4.949989318847656, | |
| -6.820126056671143, | |
| -5.9252166748046875, | |
| -4.965448379516602, | |
| -2.26172137260437, | |
| -4.704195976257324, | |
| -4.545362949371338, | |
| -6.195320129394531, | |
| -3.49263334274292, | |
| -5.598983287811279, | |
| -6.862378120422363, | |
| -3.933257579803467, | |
| -5.867007732391357, | |
| -3.922450304031372, | |
| -4.719234943389893, | |
| -6.2294182777404785, | |
| -3.8939015865325928, | |
| -6.788502216339111, | |
| -4.9644775390625, | |
| -5.412384510040283, | |
| -3.053751230239868, | |
| -4.095732688903809, | |
| -5.8144707679748535, | |
| -3.916658639907837 | |
| ], | |
| "sha256": "220ac92ea0d483cd1acc2ce67373fdde17b73fa6f587581c993db23772bd8cdf" | |
| }, | |
| "layers.24.attn.A_log": { | |
| "values": [ | |
| 1.935854196548462, | |
| 2.402085542678833, | |
| 2.569786787033081, | |
| 2.6451468467712402, | |
| 2.6867244243621826, | |
| 2.465416193008423, | |
| 2.0113072395324707, | |
| 2.726113796234131, | |
| 2.750795364379883, | |
| 2.479898691177368, | |
| 2.6629059314727783, | |
| 2.579301118850708, | |
| 2.193319320678711, | |
| 1.1620546579360962, | |
| 1.777323842048645, | |
| 1.4181650876998901, | |
| 0.8534590005874634, | |
| 2.501542329788208, | |
| 1.9093964099884033, | |
| 0.8884909152984619, | |
| 0.7735486030578613, | |
| 0.4444018602371216, | |
| 2.735567092895508, | |
| 2.32594895362854, | |
| 2.5412023067474365, | |
| 2.1243083477020264, | |
| 1.4774413108825684, | |
| 2.058936357498169, | |
| 1.7353363037109375, | |
| 2.028571367263794, | |
| 1.5982521772384644, | |
| 2.324718952178955 | |
| ], | |
| "sha256": "ab3da344d36e57b3e8d372a071ed8dd2ec948813b8763f343a06028cc55d0a9a" | |
| }, | |
| "layers.25.attn.dt_bias": { | |
| "values": [ | |
| -3.0411536693573, | |
| -2.777629852294922, | |
| -6.884170055389404, | |
| -3.169262647628784, | |
| -3.6260154247283936, | |
| -6.88244104385376, | |
| -4.655452251434326, | |
| -3.2058005332946777, | |
| -4.123000621795654, | |
| -5.914124965667725, | |
| -2.758500814437866, | |
| -5.921397686004639, | |
| -5.475053787231445, | |
| -5.100183963775635, | |
| -4.400990962982178, | |
| -2.7898807525634766, | |
| -4.158336162567139, | |
| -6.282515048980713, | |
| -3.2546143531799316, | |
| -4.511543273925781, | |
| -4.193739891052246, | |
| -2.90832257270813, | |
| -4.28714656829834, | |
| -4.5907301902771, | |
| -6.904921531677246, | |
| -5.753597736358643, | |
| -4.951962471008301, | |
| -6.057634353637695, | |
| -3.810419797897339, | |
| -4.752627849578857, | |
| -6.392106056213379, | |
| -5.934003829956055 | |
| ], | |
| "sha256": "80faaf649020ec2a0d9e91ef904d4be902d433075723a1304fa4dec36fccf636" | |
| }, | |
| "layers.25.attn.A_log": { | |
| "values": [ | |
| 2.7150585651397705, | |
| 1.8667364120483398, | |
| 1.9881675243377686, | |
| 1.7351881265640259, | |
| 0.6903443336486816, | |
| 2.5331451892852783, | |
| 2.1228251457214355, | |
| 1.603857159614563, | |
| 1.8874999284744263, | |
| 2.243292808532715, | |
| 2.347374677658081, | |
| 1.791951298713684, | |
| 2.5444576740264893, | |
| 2.7163050174713135, | |
| 2.117619514465332, | |
| 2.33750581741333, | |
| 1.2096391916275024, | |
| 0.36090177297592163, | |
| 0.33749765157699585, | |
| 0.966330885887146, | |
| 1.5235062837600708, | |
| 2.471710205078125, | |
| 0.6161686778068542, | |
| 2.501368761062622, | |
| 1.686753273010254, | |
| 2.4261200428009033, | |
| 1.4209768772125244, | |
| 1.8571573495864868, | |
| 2.3713276386260986, | |
| 1.246005654335022, | |
| 2.7106332778930664, | |
| 2.088240146636963 | |
| ], | |
| "sha256": "924e13791ade5585f6edc145d3e673eb615f0a77c778c3ca2250c46eacd28ef2" | |
| }, | |
| "layers.26.attn.dt_bias": { | |
| "values": [ | |
| -5.508322715759277, | |
| -4.659005641937256, | |
| -4.707586765289307, | |
| -5.092001438140869, | |
| -5.235030651092529, | |
| -6.055210113525391, | |
| -6.8153910636901855, | |
| -6.120583534240723, | |
| -2.4891443252563477, | |
| -2.258056879043579, | |
| -2.4085402488708496, | |
| -2.7803595066070557, | |
| -5.2153191566467285, | |
| -3.990525245666504, | |
| -4.1779561042785645, | |
| -5.093641757965088, | |
| -3.245126962661743, | |
| -4.875733375549316, | |
| -6.14001989364624, | |
| -6.191963195800781, | |
| -3.395951747894287, | |
| -2.801229476928711, | |
| -3.6102380752563477, | |
| -5.156657695770264, | |
| -6.411379337310791, | |
| -4.295237064361572, | |
| -6.105694770812988, | |
| -4.919986724853516, | |
| -6.16376256942749, | |
| -4.770781517028809, | |
| -2.798285961151123, | |
| -4.960533142089844 | |
| ], | |
| "sha256": "a3f8785e943f20db94ded4962eeed8f81b09dbfbebe0e40d693aa967b2ea2956" | |
| }, | |
| "layers.26.attn.A_log": { | |
| "values": [ | |
| 2.609185218811035, | |
| 1.8760918378829956, | |
| 2.5963172912597656, | |
| 1.858278751373291, | |
| 2.6693291664123535, | |
| 1.3640241622924805, | |
| 2.33142352104187, | |
| 1.9357331991195679, | |
| 2.649726152420044, | |
| 2.485280990600586, | |
| 1.6772164106369019, | |
| 2.5923445224761963, | |
| 2.323786497116089, | |
| 2.5850675106048584, | |
| 1.8785045146942139, | |
| 1.8661174774169922, | |
| 2.6580183506011963, | |
| 2.5347843170166016, | |
| 0.6509512066841125, | |
| 2.3700318336486816, | |
| 2.6578171253204346, | |
| 2.130983591079712, | |
| 2.5079891681671143, | |
| 1.7549562454223633, | |
| 2.5272843837738037, | |
| 2.757739543914795, | |
| 2.2930736541748047, | |
| 2.408629894256592, | |
| 2.7679288387298584, | |
| 2.06329345703125, | |
| 0.5408400297164917, | |
| 2.6919751167297363 | |
| ], | |
| "sha256": "e0a7c6e11bd100df903ae03b45dcccc15da5d8fb41982c69953f18529bc4c9aa" | |
| }, | |
| "layers.27.attn.dt_bias": { | |
| "values": [ | |
| -6.83493185043335, | |
| -4.799566268920898, | |
| -5.654449462890625, | |
| -5.148915767669678, | |
| -4.8133931159973145, | |
| -4.306173801422119, | |
| -6.193814754486084, | |
| -5.2967023849487305, | |
| -3.6741726398468018, | |
| -4.374444007873535, | |
| -2.7135963439941406, | |
| -3.629587173461914, | |
| -3.334251880645752, | |
| -2.405266761779785, | |
| -5.942895412445068, | |
| -3.6998093128204346, | |
| -5.175762176513672, | |
| -5.032937526702881, | |
| -2.5548362731933594, | |
| -4.477755069732666, | |
| -5.7006940841674805, | |
| -4.7214131355285645, | |
| -3.8110363483428955, | |
| -6.837927341461182, | |
| -2.3956873416900635, | |
| -3.6135594844818115, | |
| -4.882908344268799, | |
| -5.650901794433594, | |
| -3.1220805644989014, | |
| -3.3826849460601807, | |
| -6.2744460105896, | |
| -6.248389720916748 | |
| ], | |
| "sha256": "f3e477f5993196a87ebbfda05a708c7a6a587a8b603dc8eca9b45f33e5c6afcf" | |
| }, | |
| "layers.27.attn.A_log": { | |
| "values": [ | |
| 2.0527842044830322, | |
| 2.7384886741638184, | |
| 2.6923394203186035, | |
| 0.5760841965675354, | |
| 2.0624935626983643, | |
| 2.6220438480377197, | |
| 0.6734188199043274, | |
| 1.1576848030090332, | |
| 0.9968636631965637, | |
| 2.0503878593444824, | |
| 1.3686782121658325, | |
| 0.9141130447387695, | |
| 2.3804163932800293, | |
| 0.31117433309555054, | |
| 1.4641815423965454, | |
| 1.5327024459838867, | |
| 2.6774961948394775, | |
| 2.6854703426361084, | |
| 1.9678905010223389, | |
| 2.5495262145996094, | |
| 1.0027847290039062, | |
| 0.7813264727592468, | |
| 1.8085479736328125, | |
| 2.628911018371582, | |
| 1.8065228462219238, | |
| 2.486243724822998, | |
| 2.568321704864502, | |
| 1.5529464483261108, | |
| 1.1472079753875732, | |
| 2.681427001953125, | |
| 1.8278661966323853, | |
| 2.0854036808013916 | |
| ], | |
| "sha256": "49a0bc69a6a0ed9da6c0b915a544a04689ee670914559097ba6744a3dde36413" | |
| } | |
| } | |
| } | |