{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.2341526520051747, "eval_steps": 500, "global_step": 2500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0129366106080207, "grad_norm": 4.402749538421631, "learning_rate": 3.8809831824062096e-07, "loss": 1.5887, "step": 10 }, { "epoch": 0.0258732212160414, "grad_norm": 2.115481376647949, "learning_rate": 7.761966364812419e-07, "loss": 1.6039, "step": 20 }, { "epoch": 0.03880983182406209, "grad_norm": 4.390326023101807, "learning_rate": 1.1642949547218628e-06, "loss": 1.2713, "step": 30 }, { "epoch": 0.0517464424320828, "grad_norm": 2.4916300773620605, "learning_rate": 1.5523932729624839e-06, "loss": 1.4527, "step": 40 }, { "epoch": 0.0646830530401035, "grad_norm": 1.259100317955017, "learning_rate": 1.940491591203105e-06, "loss": 1.3702, "step": 50 }, { "epoch": 0.07761966364812418, "grad_norm": 4.703212738037109, "learning_rate": 2.3285899094437256e-06, "loss": 1.6409, "step": 60 }, { "epoch": 0.09055627425614489, "grad_norm": 6.5729146003723145, "learning_rate": 2.7166882276843466e-06, "loss": 1.6568, "step": 70 }, { "epoch": 0.1034928848641656, "grad_norm": 4.498919486999512, "learning_rate": 3.1047865459249677e-06, "loss": 1.4528, "step": 80 }, { "epoch": 0.11642949547218628, "grad_norm": 1.7919808626174927, "learning_rate": 3.492884864165589e-06, "loss": 1.3359, "step": 90 }, { "epoch": 0.129366106080207, "grad_norm": 1.8573850393295288, "learning_rate": 3.88098318240621e-06, "loss": 1.0611, "step": 100 }, { "epoch": 0.1423027166882277, "grad_norm": 1.8852155208587646, "learning_rate": 4.2690815006468305e-06, "loss": 0.8434, "step": 110 }, { "epoch": 0.15523932729624837, "grad_norm": 1.1542340517044067, "learning_rate": 4.657179818887451e-06, "loss": 0.839, "step": 120 }, { "epoch": 0.16817593790426907, "grad_norm": 2.5517771244049072, "learning_rate": 5.045278137128073e-06, "loss": 0.7732, "step": 130 }, { "epoch": 0.18111254851228978, "grad_norm": 2.299227476119995, "learning_rate": 5.433376455368693e-06, "loss": 0.7517, "step": 140 }, { "epoch": 0.19404915912031048, "grad_norm": 1.0255972146987915, "learning_rate": 5.821474773609315e-06, "loss": 0.74, "step": 150 }, { "epoch": 0.2069857697283312, "grad_norm": 6.213005542755127, "learning_rate": 6.2095730918499354e-06, "loss": 0.6561, "step": 160 }, { "epoch": 0.21992238033635186, "grad_norm": 1.065950632095337, "learning_rate": 6.597671410090556e-06, "loss": 0.6735, "step": 170 }, { "epoch": 0.23285899094437257, "grad_norm": 1.7846940755844116, "learning_rate": 6.985769728331178e-06, "loss": 0.5362, "step": 180 }, { "epoch": 0.24579560155239327, "grad_norm": 1.3279622793197632, "learning_rate": 7.373868046571798e-06, "loss": 0.6241, "step": 190 }, { "epoch": 0.258732212160414, "grad_norm": 2.3642659187316895, "learning_rate": 7.76196636481242e-06, "loss": 0.5634, "step": 200 }, { "epoch": 0.2716688227684347, "grad_norm": 1.5506231784820557, "learning_rate": 8.15006468305304e-06, "loss": 0.4842, "step": 210 }, { "epoch": 0.2846054333764554, "grad_norm": 1.9322525262832642, "learning_rate": 8.538163001293661e-06, "loss": 0.5793, "step": 220 }, { "epoch": 0.2975420439844761, "grad_norm": 2.907057523727417, "learning_rate": 8.926261319534283e-06, "loss": 0.6017, "step": 230 }, { "epoch": 0.31047865459249674, "grad_norm": 1.1939823627471924, "learning_rate": 9.314359637774902e-06, "loss": 0.5463, "step": 240 }, { "epoch": 0.32341526520051744, "grad_norm": 1.6144232749938965, "learning_rate": 9.702457956015523e-06, "loss": 0.5068, "step": 250 }, { "epoch": 0.33635187580853815, "grad_norm": 1.599927306175232, "learning_rate": 1.0090556274256145e-05, "loss": 0.4741, "step": 260 }, { "epoch": 0.34928848641655885, "grad_norm": 1.0453033447265625, "learning_rate": 1.0478654592496766e-05, "loss": 0.444, "step": 270 }, { "epoch": 0.36222509702457956, "grad_norm": 1.6979962587356567, "learning_rate": 1.0866752910737387e-05, "loss": 0.5465, "step": 280 }, { "epoch": 0.37516170763260026, "grad_norm": 1.794843316078186, "learning_rate": 1.1254851228978009e-05, "loss": 0.4947, "step": 290 }, { "epoch": 0.38809831824062097, "grad_norm": 1.308774709701538, "learning_rate": 1.164294954721863e-05, "loss": 0.5153, "step": 300 }, { "epoch": 0.40103492884864167, "grad_norm": 1.2728034257888794, "learning_rate": 1.203104786545925e-05, "loss": 0.5684, "step": 310 }, { "epoch": 0.4139715394566624, "grad_norm": 1.312259316444397, "learning_rate": 1.2419146183699871e-05, "loss": 0.4749, "step": 320 }, { "epoch": 0.4269081500646831, "grad_norm": 1.1889984607696533, "learning_rate": 1.2807244501940493e-05, "loss": 0.4884, "step": 330 }, { "epoch": 0.4398447606727037, "grad_norm": 1.4888752698898315, "learning_rate": 1.3195342820181112e-05, "loss": 0.4353, "step": 340 }, { "epoch": 0.45278137128072443, "grad_norm": 1.481899380683899, "learning_rate": 1.3583441138421733e-05, "loss": 0.4649, "step": 350 }, { "epoch": 0.46571798188874514, "grad_norm": 1.5792471170425415, "learning_rate": 1.3971539456662355e-05, "loss": 0.4314, "step": 360 }, { "epoch": 0.47865459249676584, "grad_norm": 6.622572422027588, "learning_rate": 1.4359637774902976e-05, "loss": 0.6105, "step": 370 }, { "epoch": 0.49159120310478654, "grad_norm": 2.20646333694458, "learning_rate": 1.4747736093143596e-05, "loss": 0.5326, "step": 380 }, { "epoch": 0.5045278137128072, "grad_norm": 1.741729974746704, "learning_rate": 1.5135834411384215e-05, "loss": 0.5013, "step": 390 }, { "epoch": 0.517464424320828, "grad_norm": 1.5974177122116089, "learning_rate": 1.552393272962484e-05, "loss": 0.4441, "step": 400 }, { "epoch": 0.5304010349288486, "grad_norm": 1.5022289752960205, "learning_rate": 1.591203104786546e-05, "loss": 0.477, "step": 410 }, { "epoch": 0.5433376455368694, "grad_norm": 2.8963606357574463, "learning_rate": 1.630012936610608e-05, "loss": 0.4674, "step": 420 }, { "epoch": 0.55627425614489, "grad_norm": 1.0391566753387451, "learning_rate": 1.66882276843467e-05, "loss": 0.4624, "step": 430 }, { "epoch": 0.5692108667529108, "grad_norm": 2.012328624725342, "learning_rate": 1.7076326002587322e-05, "loss": 0.4798, "step": 440 }, { "epoch": 0.5821474773609314, "grad_norm": 1.6585012674331665, "learning_rate": 1.7464424320827943e-05, "loss": 0.3837, "step": 450 }, { "epoch": 0.5950840879689522, "grad_norm": 2.2663848400115967, "learning_rate": 1.7852522639068567e-05, "loss": 0.4888, "step": 460 }, { "epoch": 0.6080206985769728, "grad_norm": 1.234695553779602, "learning_rate": 1.8240620957309184e-05, "loss": 0.3898, "step": 470 }, { "epoch": 0.6209573091849935, "grad_norm": 1.9593554735183716, "learning_rate": 1.8628719275549805e-05, "loss": 0.4126, "step": 480 }, { "epoch": 0.6338939197930142, "grad_norm": 1.539140224456787, "learning_rate": 1.901681759379043e-05, "loss": 0.3865, "step": 490 }, { "epoch": 0.6468305304010349, "grad_norm": 1.70127534866333, "learning_rate": 1.9404915912031046e-05, "loss": 0.3849, "step": 500 }, { "epoch": 0.6468305304010349, "eval_loss": 0.49836617708206177, "eval_runtime": 53.9941, "eval_samples_per_second": 3.704, "eval_steps_per_second": 1.852, "step": 500 }, { "epoch": 0.6597671410090556, "grad_norm": 1.2516582012176514, "learning_rate": 1.979301423027167e-05, "loss": 0.4551, "step": 510 }, { "epoch": 0.6727037516170763, "grad_norm": 2.031634569168091, "learning_rate": 2.018111254851229e-05, "loss": 0.4455, "step": 520 }, { "epoch": 0.685640362225097, "grad_norm": 1.4083653688430786, "learning_rate": 2.056921086675291e-05, "loss": 0.3805, "step": 530 }, { "epoch": 0.6985769728331177, "grad_norm": 1.8752052783966064, "learning_rate": 2.0957309184993532e-05, "loss": 0.4295, "step": 540 }, { "epoch": 0.7115135834411385, "grad_norm": 2.337958574295044, "learning_rate": 2.1345407503234156e-05, "loss": 0.3906, "step": 550 }, { "epoch": 0.7244501940491591, "grad_norm": 1.7030471563339233, "learning_rate": 2.1733505821474773e-05, "loss": 0.4151, "step": 560 }, { "epoch": 0.7373868046571799, "grad_norm": 2.75461745262146, "learning_rate": 2.2121604139715397e-05, "loss": 0.4353, "step": 570 }, { "epoch": 0.7503234152652005, "grad_norm": 1.3487416505813599, "learning_rate": 2.2509702457956018e-05, "loss": 0.3256, "step": 580 }, { "epoch": 0.7632600258732212, "grad_norm": 2.2425625324249268, "learning_rate": 2.2897800776196635e-05, "loss": 0.3754, "step": 590 }, { "epoch": 0.7761966364812419, "grad_norm": 2.3429338932037354, "learning_rate": 2.328589909443726e-05, "loss": 0.4642, "step": 600 }, { "epoch": 0.7891332470892626, "grad_norm": 1.7922463417053223, "learning_rate": 2.3673997412677876e-05, "loss": 0.4097, "step": 610 }, { "epoch": 0.8020698576972833, "grad_norm": 1.4449383020401, "learning_rate": 2.40620957309185e-05, "loss": 0.4068, "step": 620 }, { "epoch": 0.815006468305304, "grad_norm": 2.2355751991271973, "learning_rate": 2.445019404915912e-05, "loss": 0.4196, "step": 630 }, { "epoch": 0.8279430789133247, "grad_norm": 1.9109554290771484, "learning_rate": 2.4838292367399742e-05, "loss": 0.4531, "step": 640 }, { "epoch": 0.8408796895213454, "grad_norm": 2.0991272926330566, "learning_rate": 2.5226390685640362e-05, "loss": 0.4161, "step": 650 }, { "epoch": 0.8538163001293662, "grad_norm": 3.535731554031372, "learning_rate": 2.5614489003880986e-05, "loss": 0.3607, "step": 660 }, { "epoch": 0.8667529107373868, "grad_norm": 1.3505184650421143, "learning_rate": 2.6002587322121604e-05, "loss": 0.401, "step": 670 }, { "epoch": 0.8796895213454075, "grad_norm": 1.0390079021453857, "learning_rate": 2.6390685640362224e-05, "loss": 0.411, "step": 680 }, { "epoch": 0.8926261319534282, "grad_norm": 4.166248798370361, "learning_rate": 2.677878395860285e-05, "loss": 0.456, "step": 690 }, { "epoch": 0.9055627425614489, "grad_norm": 1.9140669107437134, "learning_rate": 2.7166882276843466e-05, "loss": 0.4109, "step": 700 }, { "epoch": 0.9184993531694696, "grad_norm": 1.4948009252548218, "learning_rate": 2.755498059508409e-05, "loss": 0.4113, "step": 710 }, { "epoch": 0.9314359637774903, "grad_norm": 1.2608532905578613, "learning_rate": 2.794307891332471e-05, "loss": 0.4074, "step": 720 }, { "epoch": 0.944372574385511, "grad_norm": 1.6317986249923706, "learning_rate": 2.833117723156533e-05, "loss": 0.4994, "step": 730 }, { "epoch": 0.9573091849935317, "grad_norm": 2.5210256576538086, "learning_rate": 2.871927554980595e-05, "loss": 0.3803, "step": 740 }, { "epoch": 0.9702457956015524, "grad_norm": 1.3882018327713013, "learning_rate": 2.9107373868046576e-05, "loss": 0.3962, "step": 750 }, { "epoch": 0.9831824062095731, "grad_norm": 2.999624013900757, "learning_rate": 2.9495472186287193e-05, "loss": 0.3772, "step": 760 }, { "epoch": 0.9961190168175937, "grad_norm": 1.3309364318847656, "learning_rate": 2.9883570504527814e-05, "loss": 0.3445, "step": 770 }, { "epoch": 1.0090556274256144, "grad_norm": 1.9435688257217407, "learning_rate": 2.99996206183792e-05, "loss": 0.3399, "step": 780 }, { "epoch": 1.0219922380336353, "grad_norm": 1.2193248271942139, "learning_rate": 2.999776246888373e-05, "loss": 0.3159, "step": 790 }, { "epoch": 1.034928848641656, "grad_norm": 1.2065479755401611, "learning_rate": 2.999435606075697e-05, "loss": 0.3086, "step": 800 }, { "epoch": 1.0478654592496766, "grad_norm": 2.735750198364258, "learning_rate": 2.9989401745651436e-05, "loss": 0.3718, "step": 810 }, { "epoch": 1.0608020698576972, "grad_norm": 2.377659320831299, "learning_rate": 2.998290003501415e-05, "loss": 0.2935, "step": 820 }, { "epoch": 1.073738680465718, "grad_norm": 1.961264967918396, "learning_rate": 2.997485160003388e-05, "loss": 0.3441, "step": 830 }, { "epoch": 1.0866752910737387, "grad_norm": 1.8460434675216675, "learning_rate": 2.9965257271571833e-05, "loss": 0.3485, "step": 840 }, { "epoch": 1.0996119016817594, "grad_norm": 1.7050330638885498, "learning_rate": 2.995411804007586e-05, "loss": 0.3739, "step": 850 }, { "epoch": 1.11254851228978, "grad_norm": 2.2543892860412598, "learning_rate": 2.994143505547826e-05, "loss": 0.307, "step": 860 }, { "epoch": 1.1254851228978007, "grad_norm": 3.6111979484558105, "learning_rate": 2.9927209627077032e-05, "loss": 0.2837, "step": 870 }, { "epoch": 1.1384217335058215, "grad_norm": 1.4579052925109863, "learning_rate": 2.9911443223400712e-05, "loss": 0.2957, "step": 880 }, { "epoch": 1.1513583441138422, "grad_norm": 2.5956053733825684, "learning_rate": 2.9894137472056805e-05, "loss": 0.2659, "step": 890 }, { "epoch": 1.1642949547218628, "grad_norm": 2.876161813735962, "learning_rate": 2.9875294159563723e-05, "loss": 0.3203, "step": 900 }, { "epoch": 1.1772315653298835, "grad_norm": 2.269209384918213, "learning_rate": 2.9854915231166383e-05, "loss": 0.3527, "step": 910 }, { "epoch": 1.1901681759379044, "grad_norm": 1.9031386375427246, "learning_rate": 2.983300279063539e-05, "loss": 0.3527, "step": 920 }, { "epoch": 1.203104786545925, "grad_norm": 1.621424913406372, "learning_rate": 2.9809559100049852e-05, "loss": 0.2666, "step": 930 }, { "epoch": 1.2160413971539457, "grad_norm": 4.12821626663208, "learning_rate": 2.9784586579563867e-05, "loss": 0.2957, "step": 940 }, { "epoch": 1.2289780077619663, "grad_norm": 2.230564832687378, "learning_rate": 2.9758087807156683e-05, "loss": 0.3305, "step": 950 }, { "epoch": 1.2419146183699872, "grad_norm": 2.618791103363037, "learning_rate": 2.9730065518366562e-05, "loss": 0.3508, "step": 960 }, { "epoch": 1.2548512289780078, "grad_norm": 2.047571897506714, "learning_rate": 2.9700522606008392e-05, "loss": 0.2711, "step": 970 }, { "epoch": 1.2677878395860285, "grad_norm": 2.2815372943878174, "learning_rate": 2.9669462119875037e-05, "loss": 0.3719, "step": 980 }, { "epoch": 1.2807244501940491, "grad_norm": 2.5117783546447754, "learning_rate": 2.963688726642252e-05, "loss": 0.2837, "step": 990 }, { "epoch": 1.2936610608020698, "grad_norm": 1.2821422815322876, "learning_rate": 2.9602801408439003e-05, "loss": 0.3029, "step": 1000 }, { "epoch": 1.2936610608020698, "eval_loss": 0.46682319045066833, "eval_runtime": 53.9865, "eval_samples_per_second": 3.705, "eval_steps_per_second": 1.852, "step": 1000 }, { "epoch": 1.3065976714100906, "grad_norm": 1.465987205505371, "learning_rate": 2.956720806469762e-05, "loss": 0.3424, "step": 1010 }, { "epoch": 1.3195342820181113, "grad_norm": 2.9073352813720703, "learning_rate": 2.9530110909593264e-05, "loss": 0.3734, "step": 1020 }, { "epoch": 1.332470892626132, "grad_norm": 1.9318575859069824, "learning_rate": 2.949151377276323e-05, "loss": 0.3153, "step": 1030 }, { "epoch": 1.3454075032341526, "grad_norm": 2.4026870727539062, "learning_rate": 2.94514206386919e-05, "loss": 0.2888, "step": 1040 }, { "epoch": 1.3583441138421732, "grad_norm": 1.4837349653244019, "learning_rate": 2.9409835646299393e-05, "loss": 0.221, "step": 1050 }, { "epoch": 1.371280724450194, "grad_norm": 1.4667383432388306, "learning_rate": 2.9366763088514306e-05, "loss": 0.2935, "step": 1060 }, { "epoch": 1.3842173350582148, "grad_norm": 1.7328715324401855, "learning_rate": 2.932220741183055e-05, "loss": 0.3634, "step": 1070 }, { "epoch": 1.3971539456662354, "grad_norm": 1.5297213792800903, "learning_rate": 2.9276173215848297e-05, "loss": 0.282, "step": 1080 }, { "epoch": 1.4100905562742563, "grad_norm": 1.0233724117279053, "learning_rate": 2.9228665252799187e-05, "loss": 0.3941, "step": 1090 }, { "epoch": 1.4230271668822767, "grad_norm": 1.3520108461380005, "learning_rate": 2.917968842705572e-05, "loss": 0.3211, "step": 1100 }, { "epoch": 1.4359637774902976, "grad_norm": 2.3806064128875732, "learning_rate": 2.9129247794624977e-05, "loss": 0.2752, "step": 1110 }, { "epoch": 1.4489003880983182, "grad_norm": 1.543283224105835, "learning_rate": 2.907734856262666e-05, "loss": 0.3198, "step": 1120 }, { "epoch": 1.4618369987063389, "grad_norm": 4.661839962005615, "learning_rate": 2.9023996088755573e-05, "loss": 0.3588, "step": 1130 }, { "epoch": 1.4747736093143597, "grad_norm": 0.8752655386924744, "learning_rate": 2.8969195880728497e-05, "loss": 0.3085, "step": 1140 }, { "epoch": 1.4877102199223804, "grad_norm": 1.2036000490188599, "learning_rate": 2.891295359571565e-05, "loss": 0.2737, "step": 1150 }, { "epoch": 1.500646830530401, "grad_norm": 1.5076147317886353, "learning_rate": 2.8855275039756653e-05, "loss": 0.2985, "step": 1160 }, { "epoch": 1.5135834411384217, "grad_norm": 1.462025761604309, "learning_rate": 2.879616616716118e-05, "loss": 0.3275, "step": 1170 }, { "epoch": 1.5265200517464423, "grad_norm": 3.9805407524108887, "learning_rate": 2.873563307989427e-05, "loss": 0.267, "step": 1180 }, { "epoch": 1.5394566623544632, "grad_norm": 1.7805012464523315, "learning_rate": 2.8673682026946385e-05, "loss": 0.2511, "step": 1190 }, { "epoch": 1.5523932729624839, "grad_norm": 2.0648598670959473, "learning_rate": 2.8610319403688362e-05, "loss": 0.3148, "step": 1200 }, { "epoch": 1.5653298835705045, "grad_norm": 1.9050356149673462, "learning_rate": 2.854555175121114e-05, "loss": 0.3273, "step": 1210 }, { "epoch": 1.5782664941785254, "grad_norm": 2.525097370147705, "learning_rate": 2.847938575565055e-05, "loss": 0.2969, "step": 1220 }, { "epoch": 1.5912031047865458, "grad_norm": 2.5983824729919434, "learning_rate": 2.8411828247497072e-05, "loss": 0.2368, "step": 1230 }, { "epoch": 1.6041397153945667, "grad_norm": 1.126298189163208, "learning_rate": 2.83428862008907e-05, "loss": 0.2437, "step": 1240 }, { "epoch": 1.6170763260025873, "grad_norm": 2.4499635696411133, "learning_rate": 2.827256673290099e-05, "loss": 0.269, "step": 1250 }, { "epoch": 1.630012936610608, "grad_norm": 1.302224040031433, "learning_rate": 2.8200877102792354e-05, "loss": 0.2963, "step": 1260 }, { "epoch": 1.6429495472186288, "grad_norm": 3.0382964611053467, "learning_rate": 2.8127824711274648e-05, "loss": 0.2381, "step": 1270 }, { "epoch": 1.6558861578266493, "grad_norm": 1.313506007194519, "learning_rate": 2.8053417099739195e-05, "loss": 0.2552, "step": 1280 }, { "epoch": 1.6688227684346701, "grad_norm": 1.7527904510498047, "learning_rate": 2.7977661949480256e-05, "loss": 0.2691, "step": 1290 }, { "epoch": 1.6817593790426908, "grad_norm": 2.4050045013427734, "learning_rate": 2.7900567080902077e-05, "loss": 0.2874, "step": 1300 }, { "epoch": 1.6946959896507114, "grad_norm": 1.7919877767562866, "learning_rate": 2.7822140452711557e-05, "loss": 0.2413, "step": 1310 }, { "epoch": 1.7076326002587323, "grad_norm": 2.414724349975586, "learning_rate": 2.774239016109666e-05, "loss": 0.291, "step": 1320 }, { "epoch": 1.720569210866753, "grad_norm": 1.6505646705627441, "learning_rate": 2.766132443889063e-05, "loss": 0.3146, "step": 1330 }, { "epoch": 1.7335058214747736, "grad_norm": 2.0308473110198975, "learning_rate": 2.757895165472208e-05, "loss": 0.3366, "step": 1340 }, { "epoch": 1.7464424320827943, "grad_norm": 1.5665441751480103, "learning_rate": 2.7495280312151075e-05, "loss": 0.3427, "step": 1350 }, { "epoch": 1.759379042690815, "grad_norm": 1.963440179824829, "learning_rate": 2.74103190487913e-05, "loss": 0.2648, "step": 1360 }, { "epoch": 1.7723156532988358, "grad_norm": 2.129335403442383, "learning_rate": 2.732407663541837e-05, "loss": 0.3138, "step": 1370 }, { "epoch": 1.7852522639068564, "grad_norm": 2.803375005722046, "learning_rate": 2.7236561975064396e-05, "loss": 0.3078, "step": 1380 }, { "epoch": 1.798188874514877, "grad_norm": 2.7263729572296143, "learning_rate": 2.7147784102098906e-05, "loss": 0.269, "step": 1390 }, { "epoch": 1.811125485122898, "grad_norm": 2.3668291568756104, "learning_rate": 2.7057752181296207e-05, "loss": 0.2619, "step": 1400 }, { "epoch": 1.8240620957309184, "grad_norm": 2.556081533432007, "learning_rate": 2.6966475506889274e-05, "loss": 0.3413, "step": 1410 }, { "epoch": 1.8369987063389392, "grad_norm": 2.335906744003296, "learning_rate": 2.687396350161028e-05, "loss": 0.3316, "step": 1420 }, { "epoch": 1.84993531694696, "grad_norm": 1.4382126331329346, "learning_rate": 2.6780225715717873e-05, "loss": 0.2713, "step": 1430 }, { "epoch": 1.8628719275549805, "grad_norm": 1.600113868713379, "learning_rate": 2.6685271826011267e-05, "loss": 0.373, "step": 1440 }, { "epoch": 1.8758085381630014, "grad_norm": 5.542492866516113, "learning_rate": 2.658911163483128e-05, "loss": 0.3024, "step": 1450 }, { "epoch": 1.8887451487710218, "grad_norm": 2.238302707672119, "learning_rate": 2.649175506904843e-05, "loss": 0.2843, "step": 1460 }, { "epoch": 1.9016817593790427, "grad_norm": 2.0862436294555664, "learning_rate": 2.6393212179038136e-05, "loss": 0.3868, "step": 1470 }, { "epoch": 1.9146183699870634, "grad_norm": 1.8203462362289429, "learning_rate": 2.6293493137643202e-05, "loss": 0.2422, "step": 1480 }, { "epoch": 1.927554980595084, "grad_norm": 1.6436854600906372, "learning_rate": 2.619260823912364e-05, "loss": 0.2936, "step": 1490 }, { "epoch": 1.9404915912031049, "grad_norm": 1.905551791191101, "learning_rate": 2.609056789809399e-05, "loss": 0.336, "step": 1500 }, { "epoch": 1.9404915912031049, "eval_loss": 0.4133800268173218, "eval_runtime": 54.0053, "eval_samples_per_second": 3.703, "eval_steps_per_second": 1.852, "step": 1500 }, { "epoch": 1.9534282018111255, "grad_norm": 1.8538085222244263, "learning_rate": 2.598738264844816e-05, "loss": 0.3038, "step": 1510 }, { "epoch": 1.9663648124191462, "grad_norm": 2.1020994186401367, "learning_rate": 2.5883063142271986e-05, "loss": 0.3022, "step": 1520 }, { "epoch": 1.9793014230271668, "grad_norm": 2.7591235637664795, "learning_rate": 2.5777620148743637e-05, "loss": 0.2968, "step": 1530 }, { "epoch": 1.9922380336351875, "grad_norm": 1.4124302864074707, "learning_rate": 2.5671064553021808e-05, "loss": 0.2635, "step": 1540 }, { "epoch": 2.0051746442432083, "grad_norm": 1.7690317630767822, "learning_rate": 2.556340735512209e-05, "loss": 0.3312, "step": 1550 }, { "epoch": 2.0181112548512288, "grad_norm": 2.0017142295837402, "learning_rate": 2.5454659668781362e-05, "loss": 0.1666, "step": 1560 }, { "epoch": 2.0310478654592496, "grad_norm": 3.2062392234802246, "learning_rate": 2.53448327203105e-05, "loss": 0.115, "step": 1570 }, { "epoch": 2.0439844760672705, "grad_norm": 3.323155641555786, "learning_rate": 2.523393784743547e-05, "loss": 0.1774, "step": 1580 }, { "epoch": 2.056921086675291, "grad_norm": 2.3531577587127686, "learning_rate": 2.5121986498126892e-05, "loss": 0.1482, "step": 1590 }, { "epoch": 2.069857697283312, "grad_norm": 2.0525689125061035, "learning_rate": 2.5008990229418233e-05, "loss": 0.1395, "step": 1600 }, { "epoch": 2.0827943078913327, "grad_norm": 2.6297266483306885, "learning_rate": 2.489496070621276e-05, "loss": 0.1005, "step": 1610 }, { "epoch": 2.095730918499353, "grad_norm": 2.2044808864593506, "learning_rate": 2.4779909700079322e-05, "loss": 0.1475, "step": 1620 }, { "epoch": 2.108667529107374, "grad_norm": 2.0108914375305176, "learning_rate": 2.466384908803715e-05, "loss": 0.1694, "step": 1630 }, { "epoch": 2.1216041397153944, "grad_norm": 1.6785058975219727, "learning_rate": 2.454679085132976e-05, "loss": 0.1611, "step": 1640 }, { "epoch": 2.1345407503234153, "grad_norm": 2.2977287769317627, "learning_rate": 2.44287470741881e-05, "loss": 0.1325, "step": 1650 }, { "epoch": 2.147477360931436, "grad_norm": 1.995399832725525, "learning_rate": 2.430972994258306e-05, "loss": 0.1171, "step": 1660 }, { "epoch": 2.1604139715394566, "grad_norm": 3.472907066345215, "learning_rate": 2.4189751742967475e-05, "loss": 0.1902, "step": 1670 }, { "epoch": 2.1733505821474774, "grad_norm": 2.1005077362060547, "learning_rate": 2.4068824861007773e-05, "loss": 0.1704, "step": 1680 }, { "epoch": 2.186287192755498, "grad_norm": 2.8117411136627197, "learning_rate": 2.3946961780305375e-05, "loss": 0.1313, "step": 1690 }, { "epoch": 2.1992238033635187, "grad_norm": 2.1987087726593018, "learning_rate": 2.3824175081107958e-05, "loss": 0.1592, "step": 1700 }, { "epoch": 2.2121604139715396, "grad_norm": 1.832919716835022, "learning_rate": 2.370047743901078e-05, "loss": 0.121, "step": 1710 }, { "epoch": 2.22509702457956, "grad_norm": 1.8450496196746826, "learning_rate": 2.3575881623648147e-05, "loss": 0.1185, "step": 1720 }, { "epoch": 2.238033635187581, "grad_norm": 3.0566859245300293, "learning_rate": 2.3450400497375146e-05, "loss": 0.1112, "step": 1730 }, { "epoch": 2.2509702457956013, "grad_norm": 1.326285481452942, "learning_rate": 2.332404701393987e-05, "loss": 0.1673, "step": 1740 }, { "epoch": 2.263906856403622, "grad_norm": 3.9111506938934326, "learning_rate": 2.3196834217146126e-05, "loss": 0.1585, "step": 1750 }, { "epoch": 2.276843467011643, "grad_norm": 1.3035225868225098, "learning_rate": 2.3068775239506928e-05, "loss": 0.1046, "step": 1760 }, { "epoch": 2.2897800776196635, "grad_norm": 1.153611183166504, "learning_rate": 2.293988330088875e-05, "loss": 0.1362, "step": 1770 }, { "epoch": 2.3027166882276844, "grad_norm": 2.1432673931121826, "learning_rate": 2.281017170714685e-05, "loss": 0.1112, "step": 1780 }, { "epoch": 2.315653298835705, "grad_norm": 3.1188621520996094, "learning_rate": 2.2679653848751642e-05, "loss": 0.1071, "step": 1790 }, { "epoch": 2.3285899094437257, "grad_norm": 5.124244213104248, "learning_rate": 2.2548343199406362e-05, "loss": 0.1404, "step": 1800 }, { "epoch": 2.3415265200517466, "grad_norm": 1.3150992393493652, "learning_rate": 2.2416253314656154e-05, "loss": 0.1169, "step": 1810 }, { "epoch": 2.354463130659767, "grad_norm": 2.1793763637542725, "learning_rate": 2.2283397830488702e-05, "loss": 0.1248, "step": 1820 }, { "epoch": 2.367399741267788, "grad_norm": 3.1768717765808105, "learning_rate": 2.214979046192652e-05, "loss": 0.2202, "step": 1830 }, { "epoch": 2.3803363518758087, "grad_norm": 1.7203766107559204, "learning_rate": 2.201544500161115e-05, "loss": 0.1172, "step": 1840 }, { "epoch": 2.393272962483829, "grad_norm": 2.7805585861206055, "learning_rate": 2.188037531837929e-05, "loss": 0.1222, "step": 1850 }, { "epoch": 2.40620957309185, "grad_norm": 1.8473186492919922, "learning_rate": 2.1744595355831075e-05, "loss": 0.1519, "step": 1860 }, { "epoch": 2.4191461836998704, "grad_norm": 0.9544809460639954, "learning_rate": 2.160811913089067e-05, "loss": 0.1494, "step": 1870 }, { "epoch": 2.4320827943078913, "grad_norm": 0.7487724423408508, "learning_rate": 2.1470960732359217e-05, "loss": 0.1028, "step": 1880 }, { "epoch": 2.445019404915912, "grad_norm": 3.580900192260742, "learning_rate": 2.1333134319460462e-05, "loss": 0.1208, "step": 1890 }, { "epoch": 2.4579560155239326, "grad_norm": 2.701359987258911, "learning_rate": 2.1194654120379035e-05, "loss": 0.1526, "step": 1900 }, { "epoch": 2.4708926261319535, "grad_norm": 3.648247480392456, "learning_rate": 2.105553443079163e-05, "loss": 0.1302, "step": 1910 }, { "epoch": 2.4838292367399744, "grad_norm": 1.9671046733856201, "learning_rate": 2.0915789612391252e-05, "loss": 0.1192, "step": 1920 }, { "epoch": 2.496765847347995, "grad_norm": 2.657763719558716, "learning_rate": 2.077543409140458e-05, "loss": 0.0996, "step": 1930 }, { "epoch": 2.5097024579560157, "grad_norm": 2.28479266166687, "learning_rate": 2.0634482357102756e-05, "loss": 0.1008, "step": 1940 }, { "epoch": 2.522639068564036, "grad_norm": 3.152531862258911, "learning_rate": 2.049294896030559e-05, "loss": 0.1027, "step": 1950 }, { "epoch": 2.535575679172057, "grad_norm": 2.3287220001220703, "learning_rate": 2.035084851187944e-05, "loss": 0.173, "step": 1960 }, { "epoch": 2.548512289780078, "grad_norm": 2.7931973934173584, "learning_rate": 2.0208195681228925e-05, "loss": 0.1159, "step": 1970 }, { "epoch": 2.5614489003880982, "grad_norm": 2.251265048980713, "learning_rate": 2.0065005194782503e-05, "loss": 0.1111, "step": 1980 }, { "epoch": 2.574385510996119, "grad_norm": 1.9223365783691406, "learning_rate": 1.992129183447229e-05, "loss": 0.111, "step": 1990 }, { "epoch": 2.5873221216041395, "grad_norm": 1.6581416130065918, "learning_rate": 1.977707043620804e-05, "loss": 0.1782, "step": 2000 }, { "epoch": 2.5873221216041395, "eval_loss": 0.4399765133857727, "eval_runtime": 54.0058, "eval_samples_per_second": 3.703, "eval_steps_per_second": 1.852, "step": 2000 }, { "epoch": 2.6002587322121604, "grad_norm": 2.0683767795562744, "learning_rate": 1.9632355888345613e-05, "loss": 0.1118, "step": 2010 }, { "epoch": 2.6131953428201813, "grad_norm": 2.32218861579895, "learning_rate": 1.948716313015e-05, "loss": 0.1636, "step": 2020 }, { "epoch": 2.6261319534282017, "grad_norm": 1.1012799739837646, "learning_rate": 1.93415071502531e-05, "loss": 0.1289, "step": 2030 }, { "epoch": 2.6390685640362226, "grad_norm": 0.9127896428108215, "learning_rate": 1.919540298510643e-05, "loss": 0.1059, "step": 2040 }, { "epoch": 2.652005174644243, "grad_norm": 3.4647481441497803, "learning_rate": 1.904886571742884e-05, "loss": 0.1094, "step": 2050 }, { "epoch": 2.664941785252264, "grad_norm": 1.8314989805221558, "learning_rate": 1.89019104746495e-05, "loss": 0.1278, "step": 2060 }, { "epoch": 2.6778783958602848, "grad_norm": 1.384616732597351, "learning_rate": 1.8754552427346256e-05, "loss": 0.1479, "step": 2070 }, { "epoch": 2.690815006468305, "grad_norm": 2.36779522895813, "learning_rate": 1.8606806787679525e-05, "loss": 0.1186, "step": 2080 }, { "epoch": 2.703751617076326, "grad_norm": 1.426782488822937, "learning_rate": 1.8458688807821922e-05, "loss": 0.1042, "step": 2090 }, { "epoch": 2.7166882276843465, "grad_norm": 0.9588170647621155, "learning_rate": 1.8310213778383704e-05, "loss": 0.0943, "step": 2100 }, { "epoch": 2.7296248382923674, "grad_norm": 1.778329610824585, "learning_rate": 1.8161397026834316e-05, "loss": 0.1382, "step": 2110 }, { "epoch": 2.742561448900388, "grad_norm": 3.2746925354003906, "learning_rate": 1.801225391592006e-05, "loss": 0.1758, "step": 2120 }, { "epoch": 2.7554980595084086, "grad_norm": 3.0417640209198, "learning_rate": 1.786279984207819e-05, "loss": 0.1341, "step": 2130 }, { "epoch": 2.7684346701164295, "grad_norm": 1.301560878753662, "learning_rate": 1.771305023384748e-05, "loss": 0.1444, "step": 2140 }, { "epoch": 2.78137128072445, "grad_norm": 1.202136754989624, "learning_rate": 1.7563020550275512e-05, "loss": 0.1478, "step": 2150 }, { "epoch": 2.794307891332471, "grad_norm": 1.7543171644210815, "learning_rate": 1.741272627932277e-05, "loss": 0.135, "step": 2160 }, { "epoch": 2.8072445019404917, "grad_norm": 2.2829203605651855, "learning_rate": 1.726218293626379e-05, "loss": 0.1439, "step": 2170 }, { "epoch": 2.8201811125485126, "grad_norm": 1.9965158700942993, "learning_rate": 1.7111406062085496e-05, "loss": 0.1509, "step": 2180 }, { "epoch": 2.833117723156533, "grad_norm": 1.459793210029602, "learning_rate": 1.6960411221882833e-05, "loss": 0.1227, "step": 2190 }, { "epoch": 2.8460543337645534, "grad_norm": 2.8442864418029785, "learning_rate": 1.6809214003251953e-05, "loss": 0.1291, "step": 2200 }, { "epoch": 2.8589909443725743, "grad_norm": 2.614272356033325, "learning_rate": 1.6657830014681087e-05, "loss": 0.1288, "step": 2210 }, { "epoch": 2.871927554980595, "grad_norm": 1.6810243129730225, "learning_rate": 1.6506274883939186e-05, "loss": 0.1809, "step": 2220 }, { "epoch": 2.884864165588616, "grad_norm": 2.1350340843200684, "learning_rate": 1.6354564256462697e-05, "loss": 0.1295, "step": 2230 }, { "epoch": 2.8978007761966365, "grad_norm": 1.7097618579864502, "learning_rate": 1.6202713793740377e-05, "loss": 0.1165, "step": 2240 }, { "epoch": 2.9107373868046573, "grad_norm": 1.1933066844940186, "learning_rate": 1.605073917169656e-05, "loss": 0.0981, "step": 2250 }, { "epoch": 2.9236739974126777, "grad_norm": 2.428194046020508, "learning_rate": 1.5898656079072867e-05, "loss": 0.1546, "step": 2260 }, { "epoch": 2.9366106080206986, "grad_norm": 2.83290696144104, "learning_rate": 1.574648021580863e-05, "loss": 0.1217, "step": 2270 }, { "epoch": 2.9495472186287195, "grad_norm": 1.9912760257720947, "learning_rate": 1.5594227291420138e-05, "loss": 0.1378, "step": 2280 }, { "epoch": 2.96248382923674, "grad_norm": 0.6878167986869812, "learning_rate": 1.5441913023378893e-05, "loss": 0.1213, "step": 2290 }, { "epoch": 2.975420439844761, "grad_norm": 1.880091905593872, "learning_rate": 1.5289553135489078e-05, "loss": 0.144, "step": 2300 }, { "epoch": 2.988357050452781, "grad_norm": 2.091439962387085, "learning_rate": 1.5137163356264331e-05, "loss": 0.155, "step": 2310 }, { "epoch": 3.001293661060802, "grad_norm": 1.475691556930542, "learning_rate": 1.498475941730404e-05, "loss": 0.0894, "step": 2320 }, { "epoch": 3.014230271668823, "grad_norm": 0.95653235912323, "learning_rate": 1.4832357051669358e-05, "loss": 0.0481, "step": 2330 }, { "epoch": 3.0271668822768434, "grad_norm": 2.0062646865844727, "learning_rate": 1.4679971992259008e-05, "loss": 0.0493, "step": 2340 }, { "epoch": 3.0401034928848643, "grad_norm": 1.8597626686096191, "learning_rate": 1.4527619970185141e-05, "loss": 0.0422, "step": 2350 }, { "epoch": 3.0530401034928847, "grad_norm": 1.7422715425491333, "learning_rate": 1.437531671314939e-05, "loss": 0.0481, "step": 2360 }, { "epoch": 3.0659767141009056, "grad_norm": 2.963845729827881, "learning_rate": 1.4223077943819228e-05, "loss": 0.041, "step": 2370 }, { "epoch": 3.0789133247089264, "grad_norm": 3.0557303428649902, "learning_rate": 1.4070919378204913e-05, "loss": 0.051, "step": 2380 }, { "epoch": 3.091849935316947, "grad_norm": 1.0821400880813599, "learning_rate": 1.3918856724037048e-05, "loss": 0.0513, "step": 2390 }, { "epoch": 3.1047865459249677, "grad_norm": 1.5353697538375854, "learning_rate": 1.3766905679145038e-05, "loss": 0.0456, "step": 2400 }, { "epoch": 3.117723156532988, "grad_norm": 2.533022165298462, "learning_rate": 1.3615081929836583e-05, "loss": 0.0572, "step": 2410 }, { "epoch": 3.130659767141009, "grad_norm": 1.9726536273956299, "learning_rate": 1.3463401149278308e-05, "loss": 0.0595, "step": 2420 }, { "epoch": 3.14359637774903, "grad_norm": 1.1939815282821655, "learning_rate": 1.3311878995877814e-05, "loss": 0.0403, "step": 2430 }, { "epoch": 3.1565329883570503, "grad_norm": 1.2100446224212646, "learning_rate": 1.3160531111667193e-05, "loss": 0.0579, "step": 2440 }, { "epoch": 3.169469598965071, "grad_norm": 1.7639714479446411, "learning_rate": 1.3009373120688257e-05, "loss": 0.0341, "step": 2450 }, { "epoch": 3.1824062095730916, "grad_norm": 1.6528544425964355, "learning_rate": 1.2858420627379652e-05, "loss": 0.0438, "step": 2460 }, { "epoch": 3.1953428201811125, "grad_norm": 2.3253066539764404, "learning_rate": 1.2707689214965964e-05, "loss": 0.0413, "step": 2470 }, { "epoch": 3.2082794307891334, "grad_norm": 0.2878986597061157, "learning_rate": 1.2557194443848987e-05, "loss": 0.0452, "step": 2480 }, { "epoch": 3.221216041397154, "grad_norm": 1.8581609725952148, "learning_rate": 1.240695185000144e-05, "loss": 0.0431, "step": 2490 }, { "epoch": 3.2341526520051747, "grad_norm": 1.1797322034835815, "learning_rate": 1.2256976943363091e-05, "loss": 0.0407, "step": 2500 }, { "epoch": 3.2341526520051747, "eval_loss": 0.5326685905456543, "eval_runtime": 54.0355, "eval_samples_per_second": 3.701, "eval_steps_per_second": 1.851, "step": 2500 } ], "logging_steps": 10, "max_steps": 3865, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 1.4544010942957486e+18, "train_batch_size": 8, "trial_name": null, "trial_params": null }