Files
summ_gpt2_inst_cnnxsumsam/checkpoint-10904/trainer_state.json

1797 lines
45 KiB
JSON
Raw Normal View History

{
"best_global_step": 10904,
"best_metric": 2.4245221614837646,
"best_model_checkpoint": "/common/home/users/m/mq.nguyen.2023/scratchDirectory/finetune/train_nbtpj_summ_gpt2_inst_cnnxsumsam/checkpoint-10904",
"epoch": 2.0,
"eval_steps": 500,
"global_step": 10904,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.009172208209126348,
"grad_norm": 7.051512241363525,
"learning_rate": 9.800000000000001e-06,
"loss": 6.4476,
"num_tokens": 3070741.0,
"step": 50
},
{
"epoch": 0.018344416418252695,
"grad_norm": 6.664697170257568,
"learning_rate": 1.98e-05,
"loss": 5.4332,
"num_tokens": 6150092.0,
"step": 100
},
{
"epoch": 0.02751662462737904,
"grad_norm": 5.514298439025879,
"learning_rate": 1.9999839379410334e-05,
"loss": 4.9984,
"num_tokens": 9221796.0,
"step": 150
},
{
"epoch": 0.03668883283650539,
"grad_norm": 4.882246494293213,
"learning_rate": 1.9999344344268732e-05,
"loss": 4.7556,
"num_tokens": 12294380.0,
"step": 200
},
{
"epoch": 0.04586104104563173,
"grad_norm": 4.8477373123168945,
"learning_rate": 1.9998514844235922e-05,
"loss": 4.6238,
"num_tokens": 15366995.0,
"step": 250
},
{
"epoch": 0.05503324925475808,
"grad_norm": 4.187108516693115,
"learning_rate": 1.9997350907057584e-05,
"loss": 4.5485,
"num_tokens": 18447066.0,
"step": 300
},
{
"epoch": 0.06420545746388442,
"grad_norm": 5.372704982757568,
"learning_rate": 1.9995852571665885e-05,
"loss": 4.4819,
"num_tokens": 21511375.0,
"step": 350
},
{
"epoch": 0.07337766567301078,
"grad_norm": 5.349086761474609,
"learning_rate": 1.9994019888178163e-05,
"loss": 4.3881,
"num_tokens": 24591962.0,
"step": 400
},
{
"epoch": 0.08254987388213712,
"grad_norm": 4.023826599121094,
"learning_rate": 1.9991852917895272e-05,
"loss": 4.3618,
"num_tokens": 27676055.0,
"step": 450
},
{
"epoch": 0.09172208209126347,
"grad_norm": 3.742589235305786,
"learning_rate": 1.9989351733299514e-05,
"loss": 4.3175,
"num_tokens": 30745102.0,
"step": 500
},
{
"epoch": 0.10089429030038982,
"grad_norm": 3.394568681716919,
"learning_rate": 1.9986516418052213e-05,
"loss": 4.2946,
"num_tokens": 33797572.0,
"step": 550
},
{
"epoch": 0.11006649850951616,
"grad_norm": 3.7485358715057373,
"learning_rate": 1.998334706699093e-05,
"loss": 4.2073,
"num_tokens": 36862581.0,
"step": 600
},
{
"epoch": 0.1192387067186425,
"grad_norm": 3.286533832550049,
"learning_rate": 1.997984378612628e-05,
"loss": 4.1903,
"num_tokens": 39940075.0,
"step": 650
},
{
"epoch": 0.12841091492776885,
"grad_norm": 3.757556676864624,
"learning_rate": 1.997600669263839e-05,
"loss": 4.1797,
"num_tokens": 43022248.0,
"step": 700
},
{
"epoch": 0.13758312313689522,
"grad_norm": 3.648007392883301,
"learning_rate": 1.9971835914872985e-05,
"loss": 4.1506,
"num_tokens": 46088933.0,
"step": 750
},
{
"epoch": 0.14675533134602156,
"grad_norm": 3.151127815246582,
"learning_rate": 1.996733159233708e-05,
"loss": 4.0749,
"num_tokens": 49170683.0,
"step": 800
},
{
"epoch": 0.1559275395551479,
"grad_norm": 3.370467185974121,
"learning_rate": 1.9962493875694335e-05,
"loss": 4.0687,
"num_tokens": 52242629.0,
"step": 850
},
{
"epoch": 0.16509974776427425,
"grad_norm": 3.516242027282715,
"learning_rate": 1.995732292675999e-05,
"loss": 3.9537,
"num_tokens": 55316954.0,
"step": 900
},
{
"epoch": 0.1742719559734006,
"grad_norm": 3.9187300205230713,
"learning_rate": 1.9951818918495474e-05,
"loss": 3.9209,
"num_tokens": 58384220.0,
"step": 950
},
{
"epoch": 0.18344416418252693,
"grad_norm": 3.647482395172119,
"learning_rate": 1.9945982035002613e-05,
"loss": 3.8353,
"num_tokens": 61463865.0,
"step": 1000
},
{
"epoch": 0.1926163723916533,
"grad_norm": 4.165699005126953,
"learning_rate": 1.9939812471517463e-05,
"loss": 3.8031,
"num_tokens": 64528502.0,
"step": 1050
},
{
"epoch": 0.20178858060077964,
"grad_norm": 3.750986099243164,
"learning_rate": 1.99333104344038e-05,
"loss": 3.723,
"num_tokens": 67601444.0,
"step": 1100
},
{
"epoch": 0.21096078880990599,
"grad_norm": 4.958056926727295,
"learning_rate": 1.9926476141146186e-05,
"loss": 3.6597,
"num_tokens": 70661859.0,
"step": 1150
},
{
"epoch": 0.22013299701903233,
"grad_norm": 3.76973295211792,
"learning_rate": 1.9919309820342735e-05,
"loss": 3.5977,
"num_tokens": 73730542.0,
"step": 1200
},
{
"epoch": 0.22930520522815867,
"grad_norm": 4.332479953765869,
"learning_rate": 1.991181171169743e-05,
"loss": 3.5533,
"num_tokens": 76796322.0,
"step": 1250
},
{
"epoch": 0.238477413437285,
"grad_norm": 3.7759413719177246,
"learning_rate": 1.9903982066012126e-05,
"loss": 3.5245,
"num_tokens": 79858262.0,
"step": 1300
},
{
"epoch": 0.24764962164641138,
"grad_norm": 3.8706881999969482,
"learning_rate": 1.9895821145178148e-05,
"loss": 3.4821,
"num_tokens": 82926316.0,
"step": 1350
},
{
"epoch": 0.2568218298555377,
"grad_norm": 3.8657612800598145,
"learning_rate": 1.9887329222167548e-05,
"loss": 3.3993,
"num_tokens": 85997475.0,
"step": 1400
},
{
"epoch": 0.2659940380646641,
"grad_norm": 3.64652156829834,
"learning_rate": 1.9878506581023957e-05,
"loss": 3.4047,
"num_tokens": 89061497.0,
"step": 1450
},
{
"epoch": 0.27516624627379044,
"grad_norm": 3.8434200286865234,
"learning_rate": 1.98693535168531e-05,
"loss": 3.3561,
"num_tokens": 92130920.0,
"step": 1500
},
{
"epoch": 0.2843384544829168,
"grad_norm": 3.7041072845458984,
"learning_rate": 1.9859870335812904e-05,
"loss": 3.2907,
"num_tokens": 95201616.0,
"step": 1550
},
{
"epoch": 0.2935106626920431,
"grad_norm": 3.672961711883545,
"learning_rate": 1.9850057355103283e-05,
"loss": 3.2824,
"num_tokens": 98262928.0,
"step": 1600
},
{
"epoch": 0.30268287090116947,
"grad_norm": 4.109488487243652,
"learning_rate": 1.9839914902955514e-05,
"loss": 3.2584,
"num_tokens": 101331029.0,
"step": 1650
},
{
"epoch": 0.3118550791102958,
"grad_norm": 3.8466403484344482,
"learning_rate": 1.982944331862125e-05,
"loss": 3.2185,
"num_tokens": 104400096.0,
"step": 1700
},
{
"epoch": 0.32102728731942215,
"grad_norm": 3.791473150253296,
"learning_rate": 1.9818642952361188e-05,
"loss": 3.2011,
"num_tokens": 107470267.0,
"step": 1750
},
{
"epoch": 0.3301994955285485,
"grad_norm": 4.251014232635498,
"learning_rate": 1.9807514165433342e-05,
"loss": 3.1644,
"num_tokens": 110539470.0,
"step": 1800
},
{
"epoch": 0.33937170373767483,
"grad_norm": 3.785540819168091,
"learning_rate": 1.979605733008098e-05,
"loss": 3.129,
"num_tokens": 113605849.0,
"step": 1850
},
{
"epoch": 0.3485439119468012,
"grad_norm": 3.8080990314483643,
"learning_rate": 1.978427282952014e-05,
"loss": 3.1142,
"num_tokens": 116669302.0,
"step": 1900
},
{
"epoch": 0.3577161201559275,
"grad_norm": 4.565866470336914,
"learning_rate": 1.977216105792684e-05,
"loss": 3.0942,
"num_tokens": 119736059.0,
"step": 1950
},
{
"epoch": 0.36688832836505386,
"grad_norm": 3.6725170612335205,
"learning_rate": 1.9759722420423875e-05,
"loss": 3.0913,
"num_tokens": 122795575.0,
"step": 2000
},
{
"epoch": 0.37606053657418026,
"grad_norm": 4.128929138183594,
"learning_rate": 1.974695733306728e-05,
"loss": 3.0232,
"num_tokens": 125860925.0,
"step": 2050
},
{
"epoch": 0.3852327447833066,
"grad_norm": 4.008982181549072,
"learning_rate": 1.9733866222832404e-05,
"loss": 3.0281,
"num_tokens": 128939549.0,
"step": 2100
},
{
"epoch": 0.39440495299243294,
"grad_norm": 3.8258960247039795,
"learning_rate": 1.972044952759963e-05,
"loss": 3.005,
"num_tokens": 132002665.0,
"step": 2150
},
{
"epoch": 0.4035771612015593,
"grad_norm": 3.722099781036377,
"learning_rate": 1.970670769613974e-05,
"loss": 2.9717,
"num_tokens": 135066326.0,
"step": 2200
},
{
"epoch": 0.41274936941068563,
"grad_norm": 3.6987600326538086,
"learning_rate": 1.9692641188098886e-05,
"loss": 2.9571,
"num_tokens": 138128630.0,
"step": 2250
},
{
"epoch": 0.42192157761981197,
"grad_norm": 6.789268493652344,
"learning_rate": 1.9678250473983223e-05,
"loss": 2.9399,
"num_tokens": 141192291.0,
"step": 2300
},
{
"epoch": 0.4310937858289383,
"grad_norm": 4.319190979003906,
"learning_rate": 1.966353603514317e-05,
"loss": 2.9201,
"num_tokens": 144272511.0,
"step": 2350
},
{
"epoch": 0.44026599403806466,
"grad_norm": 4.209902763366699,
"learning_rate": 1.9648498363757325e-05,
"loss": 2.924,
"num_tokens": 147343206.0,
"step": 2400
},
{
"epoch": 0.449438202247191,
"grad_norm": 3.9288434982299805,
"learning_rate": 1.9633137962815973e-05,
"loss": 2.8886,
"num_tokens": 150412293.0,
"step": 2450
},
{
"epoch": 0.45861041045631734,
"grad_norm": 4.768758296966553,
"learning_rate": 1.9617455346104283e-05,
"loss": 2.8774,
"num_tokens": 153481791.0,
"step": 2500
},
{
"epoch": 0.4677826186654437,
"grad_norm": 4.089212894439697,
"learning_rate": 1.9601451038185117e-05,
"loss": 2.8393,
"num_tokens": 156551627.0,
"step": 2550
},
{
"epoch": 0.47695482687457,
"grad_norm": 3.6109139919281006,
"learning_rate": 1.958512557438149e-05,
"loss": 2.8697,
"num_tokens": 159628649.0,
"step": 2600
},
{
"epoch": 0.4861270350836964,
"grad_norm": 4.69704008102417,
"learning_rate": 1.9568479500758648e-05,
"loss": 2.8156,
"num_tokens": 162692475.0,
"step": 2650
},
{
"epoch": 0.49529924329282277,
"grad_norm": 3.726196527481079,
"learning_rate": 1.9551513374105823e-05,
"loss": 2.8183,
"num_tokens": 165764216.0,
"step": 2700
},
{
"epoch": 0.504471451501949,
"grad_norm": 3.8877694606781006,
"learning_rate": 1.9534227761917585e-05,
"loss": 2.8098,
"num_tokens": 168832733.0,
"step": 2750
},
{
"epoch": 0.5136436597110754,
"grad_norm": 4.2488112449646,
"learning_rate": 1.9516623242374888e-05,
"loss": 2.8193,
"num_tokens": 171887021.0,
"step": 2800
},
{
"epoch": 0.5228158679202017,
"grad_norm": 3.4136576652526855,
"learning_rate": 1.949870040432571e-05,
"loss": 2.7806,
"num_tokens": 174960619.0,
"step": 2850
},
{
"epoch": 0.5319880761293282,
"grad_norm": 4.282463550567627,
"learning_rate": 1.948045984726537e-05,
"loss": 2.779,
"num_tokens": 178028517.0,
"step": 2900
},
{
"epoch": 0.5411602843384545,
"grad_norm": 4.703866958618164,
"learning_rate": 1.9461902181316455e-05,
"loss": 2.761,
"num_tokens": 181092191.0,
"step": 2950
},
{
"epoch": 0.5503324925475809,
"grad_norm": 3.5504965782165527,
"learning_rate": 1.9443028027208443e-05,
"loss": 2.7438,
"num_tokens": 184160542.0,
"step": 3000
},
{
"epoch": 0.5595047007567072,
"grad_norm": 4.609411239624023,
"learning_rate": 1.9423838016256913e-05,
"loss": 2.7322,
"num_tokens": 187232548.0,
"step": 3050
},
{
"epoch": 0.5686769089658336,
"grad_norm": 3.5375072956085205,
"learning_rate": 1.9404332790342446e-05,
"loss": 2.7147,
"num_tokens": 190297910.0,
"step": 3100
},
{
"epoch": 0.5778491171749599,
"grad_norm": 3.5921809673309326,
"learning_rate": 1.9384513001889144e-05,
"loss": 2.7105,
"num_tokens": 193364684.0,
"step": 3150
},
{
"epoch": 0.5870213253840862,
"grad_norm": 3.7573354244232178,
"learning_rate": 1.936437931384281e-05,
"loss": 2.7081,
"num_tokens": 196431776.0,
"step": 3200
},
{
"epoch": 0.5961935335932126,
"grad_norm": 4.44867467880249,
"learning_rate": 1.934393239964878e-05,
"loss": 2.6803,
"num_tokens": 199500117.0,
"step": 3250
},
{
"epoch": 0.6053657418023389,
"grad_norm": 3.4919068813323975,
"learning_rate": 1.9323172943229386e-05,
"loss": 2.6674,
"num_tokens": 202569017.0,
"step": 3300
},
{
"epoch": 0.6145379500114653,
"grad_norm": 3.851284980773926,
"learning_rate": 1.930210163896109e-05,
"loss": 2.6882,
"num_tokens": 205635983.0,
"step": 3350
},
{
"epoch": 0.6237101582205916,
"grad_norm": 5.8265910148620605,
"learning_rate": 1.9280719191651252e-05,
"loss": 2.648,
"num_tokens": 208697081.0,
"step": 3400
},
{
"epoch": 0.632882366429718,
"grad_norm": 3.4978857040405273,
"learning_rate": 1.925902631651456e-05,
"loss": 2.6435,
"num_tokens": 211770956.0,
"step": 3450
},
{
"epoch": 0.6420545746388443,
"grad_norm": 3.321568250656128,
"learning_rate": 1.9237023739149094e-05,
"loss": 2.6556,
"num_tokens": 214830999.0,
"step": 3500
},
{
"epoch": 0.6512267828479706,
"grad_norm": 3.848578691482544,
"learning_rate": 1.9214712195512074e-05,
"loss": 2.6334,
"num_tokens": 217899025.0,
"step": 3550
},
{
"epoch": 0.660398991057097,
"grad_norm": 3.511373996734619,
"learning_rate": 1.919209243189524e-05,
"loss": 2.6138,
"num_tokens": 220972192.0,
"step": 3600
},
{
"epoch": 0.6695711992662233,
"grad_norm": 3.4985859394073486,
"learning_rate": 1.9169165204899866e-05,
"loss": 2.6251,
"num_tokens": 224041762.0,
"step": 3650
},
{
"epoch": 0.6787434074753497,
"grad_norm": 5.905818462371826,
"learning_rate": 1.9145931281411485e-05,
"loss": 2.6089,
"num_tokens": 227110870.0,
"step": 3700
},
{
"epoch": 0.687915615684476,
"grad_norm": 8.281390190124512,
"learning_rate": 1.9122391438574226e-05,
"loss": 2.6084,
"num_tokens": 230193112.0,
"step": 3750
},
{
"epoch": 0.6970878238936024,
"grad_norm": 3.555816888809204,
"learning_rate": 1.9098546463764802e-05,
"loss": 2.6074,
"num_tokens": 233261587.0,
"step": 3800
},
{
"epoch": 0.7062600321027287,
"grad_norm": 4.379006862640381,
"learning_rate": 1.9074397154566203e-05,
"loss": 2.5825,
"num_tokens": 236318086.0,
"step": 3850
},
{
"epoch": 0.715432240311855,
"grad_norm": 4.370838642120361,
"learning_rate": 1.9049944318740998e-05,
"loss": 2.6007,
"num_tokens": 239390375.0,
"step": 3900
},
{
"epoch": 0.7246044485209814,
"grad_norm": 3.347520112991333,
"learning_rate": 1.902518877420432e-05,
"loss": 2.5861,
"num_tokens": 242454603.0,
"step": 3950
},
{
"epoch": 0.7337766567301077,
"grad_norm": 3.8170406818389893,
"learning_rate": 1.900013134899651e-05,
"loss": 2.5956,
"num_tokens": 245520349.0,
"step": 4000
},
{
"epoch": 0.7429488649392341,
"grad_norm": 4.084133625030518,
"learning_rate": 1.8974772881255423e-05,
"loss": 2.5744,
"num_tokens": 248591742.0,
"step": 4050
},
{
"epoch": 0.7521210731483605,
"grad_norm": 4.162419319152832,
"learning_rate": 1.8949114219188386e-05,
"loss": 2.5557,
"num_tokens": 251668641.0,
"step": 4100
},
{
"epoch": 0.7612932813574869,
"grad_norm": 4.139531135559082,
"learning_rate": 1.8923156221043834e-05,
"loss": 2.5529,
"num_tokens": 254744782.0,
"step": 4150
},
{
"epoch": 0.7704654895666132,
"grad_norm": 3.5236408710479736,
"learning_rate": 1.8896899755082594e-05,
"loss": 2.5342,
"num_tokens": 257825042.0,
"step": 4200
},
{
"epoch": 0.7796376977757395,
"grad_norm": 7.408937931060791,
"learning_rate": 1.8870345699548856e-05,
"loss": 2.5367,
"num_tokens": 260890737.0,
"step": 4250
},
{
"epoch": 0.7888099059848659,
"grad_norm": 4.408638000488281,
"learning_rate": 1.8843494942640782e-05,
"loss": 2.5397,
"num_tokens": 263956000.0,
"step": 4300
},
{
"epoch": 0.7979821141939922,
"grad_norm": 3.64819073677063,
"learning_rate": 1.8816348382480807e-05,
"loss": 2.5326,
"num_tokens": 267024229.0,
"step": 4350
},
{
"epoch": 0.8071543224031186,
"grad_norm": 3.3967065811157227,
"learning_rate": 1.8788906927085596e-05,
"loss": 2.5291,
"num_tokens": 270093879.0,
"step": 4400
},
{
"epoch": 0.8163265306122449,
"grad_norm": 3.771216869354248,
"learning_rate": 1.876117149433567e-05,
"loss": 2.5167,
"num_tokens": 273153385.0,
"step": 4450
},
{
"epoch": 0.8254987388213713,
"grad_norm": 3.552034854888916,
"learning_rate": 1.87331430119447e-05,
"loss": 2.516,
"num_tokens": 276228897.0,
"step": 4500
},
{
"epoch": 0.8346709470304976,
"grad_norm": 4.362802982330322,
"learning_rate": 1.870482241742849e-05,
"loss": 2.5083,
"num_tokens": 279299108.0,
"step": 4550
},
{
"epoch": 0.8438431552396239,
"grad_norm": 3.8707432746887207,
"learning_rate": 1.867621065807361e-05,
"loss": 2.4962,
"num_tokens": 282367606.0,
"step": 4600
},
{
"epoch": 0.8530153634487503,
"grad_norm": 3.7560534477233887,
"learning_rate": 1.8647308690905695e-05,
"loss": 2.5022,
"num_tokens": 285431292.0,
"step": 4650
},
{
"epoch": 0.8621875716578766,
"grad_norm": 4.073310375213623,
"learning_rate": 1.8618117482657473e-05,
"loss": 2.5052,
"num_tokens": 288509576.0,
"step": 4700
},
{
"epoch": 0.871359779867003,
"grad_norm": 4.2512288093566895,
"learning_rate": 1.858863800973638e-05,
"loss": 2.4922,
"num_tokens": 291577926.0,
"step": 4750
},
{
"epoch": 0.8805319880761293,
"grad_norm": 3.6681947708129883,
"learning_rate": 1.8558871258191943e-05,
"loss": 2.4687,
"num_tokens": 294652257.0,
"step": 4800
},
{
"epoch": 0.8897041962852557,
"grad_norm": 3.7940540313720703,
"learning_rate": 1.8528818223682777e-05,
"loss": 2.4622,
"num_tokens": 297724675.0,
"step": 4850
},
{
"epoch": 0.898876404494382,
"grad_norm": 4.411775588989258,
"learning_rate": 1.849847991144328e-05,
"loss": 2.472,
"num_tokens": 300795065.0,
"step": 4900
},
{
"epoch": 0.9080486127035083,
"grad_norm": 3.331857442855835,
"learning_rate": 1.8467857336250015e-05,
"loss": 2.4734,
"num_tokens": 303858894.0,
"step": 4950
},
{
"epoch": 0.9172208209126347,
"grad_norm": 5.33769416809082,
"learning_rate": 1.8436951522387766e-05,
"loss": 2.4534,
"num_tokens": 306912692.0,
"step": 5000
},
{
"epoch": 0.926393029121761,
"grad_norm": 3.5148868560791016,
"learning_rate": 1.8405763503615285e-05,
"loss": 2.4871,
"num_tokens": 310002551.0,
"step": 5050
},
{
"epoch": 0.9355652373308874,
"grad_norm": 3.6607162952423096,
"learning_rate": 1.8374294323130697e-05,
"loss": 2.4801,
"num_tokens": 313070094.0,
"step": 5100
},
{
"epoch": 0.9447374455400137,
"grad_norm": 6.60593318939209,
"learning_rate": 1.8342545033536622e-05,
"loss": 2.431,
"num_tokens": 316134272.0,
"step": 5150
},
{
"epoch": 0.95390965374914,
"grad_norm": 3.5577642917633057,
"learning_rate": 1.8310516696804953e-05,
"loss": 2.425,
"num_tokens": 319208656.0,
"step": 5200
},
{
"epoch": 0.9630818619582665,
"grad_norm": 3.443085193634033,
"learning_rate": 1.8278210384241353e-05,
"loss": 2.419,
"num_tokens": 322277785.0,
"step": 5250
},
{
"epoch": 0.9722540701673928,
"grad_norm": 4.775812149047852,
"learning_rate": 1.8245627176449395e-05,
"loss": 2.4398,
"num_tokens": 325343977.0,
"step": 5300
},
{
"epoch": 0.9814262783765192,
"grad_norm": 3.54435133934021,
"learning_rate": 1.8212768163294453e-05,
"loss": 2.4288,
"num_tokens": 328414604.0,
"step": 5350
},
{
"epoch": 0.9905984865856455,
"grad_norm": 3.6476333141326904,
"learning_rate": 1.8179634443867207e-05,
"loss": 2.4228,
"num_tokens": 331477653.0,
"step": 5400
},
{
"epoch": 0.9997706947947719,
"grad_norm": 4.548381805419922,
"learning_rate": 1.814622712644691e-05,
"loss": 2.4154,
"num_tokens": 334553911.0,
"step": 5450
},
{
"epoch": 1.0,
"eval_loss": 2.662151575088501,
"eval_num_tokens": 334630136.0,
"eval_runtime": 5.3343,
"eval_samples_per_second": 90.922,
"eval_steps_per_second": 5.811,
"step": 5452
},
{
"epoch": 1.0088053198807614,
"grad_norm": 3.1079516410827637,
"learning_rate": 1.8112547328464308e-05,
"loss": 2.4206,
"num_tokens": 337566175.0,
"step": 5500
},
{
"epoch": 1.0179775280898877,
"grad_norm": 3.530143976211548,
"learning_rate": 1.8078596176464257e-05,
"loss": 2.426,
"num_tokens": 340636573.0,
"step": 5550
},
{
"epoch": 1.027149736299014,
"grad_norm": 3.131134033203125,
"learning_rate": 1.8044374806068057e-05,
"loss": 2.3843,
"num_tokens": 343696926.0,
"step": 5600
},
{
"epoch": 1.0363219445081404,
"grad_norm": 3.275998592376709,
"learning_rate": 1.8009884361935445e-05,
"loss": 2.4091,
"num_tokens": 346772520.0,
"step": 5650
},
{
"epoch": 1.0454941527172668,
"grad_norm": 3.3414154052734375,
"learning_rate": 1.7975125997726328e-05,
"loss": 2.3759,
"num_tokens": 349844605.0,
"step": 5700
},
{
"epoch": 1.054666360926393,
"grad_norm": 4.579761981964111,
"learning_rate": 1.7940100876062184e-05,
"loss": 2.3762,
"num_tokens": 352917933.0,
"step": 5750
},
{
"epoch": 1.0638385691355194,
"grad_norm": 3.126434326171875,
"learning_rate": 1.7904810168487175e-05,
"loss": 2.371,
"num_tokens": 355987563.0,
"step": 5800
},
{
"epoch": 1.0730107773446458,
"grad_norm": 3.765179395675659,
"learning_rate": 1.7869255055428964e-05,
"loss": 2.369,
"num_tokens": 359052419.0,
"step": 5850
},
{
"epoch": 1.0821829855537721,
"grad_norm": 3.1499340534210205,
"learning_rate": 1.783343672615923e-05,
"loss": 2.3784,
"num_tokens": 362111794.0,
"step": 5900
},
{
"epoch": 1.0913551937628985,
"grad_norm": 3.7670187950134277,
"learning_rate": 1.7797356378753886e-05,
"loss": 2.362,
"num_tokens": 365184289.0,
"step": 5950
},
{
"epoch": 1.1005274019720248,
"grad_norm": 4.070444583892822,
"learning_rate": 1.7761015220053006e-05,
"loss": 2.3939,
"num_tokens": 368258725.0,
"step": 6000
},
{
"epoch": 1.1096996101811512,
"grad_norm": 3.125969886779785,
"learning_rate": 1.772441446562046e-05,
"loss": 2.3923,
"num_tokens": 371333945.0,
"step": 6050
},
{
"epoch": 1.1188718183902775,
"grad_norm": 3.1043412685394287,
"learning_rate": 1.7687555339703245e-05,
"loss": 2.3639,
"num_tokens": 374400787.0,
"step": 6100
},
{
"epoch": 1.1280440265994038,
"grad_norm": 3.3613622188568115,
"learning_rate": 1.7650439075190552e-05,
"loss": 2.3593,
"num_tokens": 377470930.0,
"step": 6150
},
{
"epoch": 1.1372162348085302,
"grad_norm": 3.4898386001586914,
"learning_rate": 1.7613066913572525e-05,
"loss": 2.398,
"num_tokens": 380540191.0,
"step": 6200
},
{
"epoch": 1.1463884430176565,
"grad_norm": 2.7355456352233887,
"learning_rate": 1.7575440104898712e-05,
"loss": 2.37,
"num_tokens": 383603448.0,
"step": 6250
},
{
"epoch": 1.1555606512267829,
"grad_norm": 3.0368947982788086,
"learning_rate": 1.7537559907736282e-05,
"loss": 2.347,
"num_tokens": 386679512.0,
"step": 6300
},
{
"epoch": 1.1647328594359092,
"grad_norm": 3.0240418910980225,
"learning_rate": 1.7499427589127913e-05,
"loss": 2.3726,
"num_tokens": 389735095.0,
"step": 6350
},
{
"epoch": 1.1739050676450355,
"grad_norm": 3.9326765537261963,
"learning_rate": 1.7461044424549408e-05,
"loss": 2.3671,
"num_tokens": 392794314.0,
"step": 6400
},
{
"epoch": 1.183077275854162,
"grad_norm": 2.9769821166992188,
"learning_rate": 1.7422411697867045e-05,
"loss": 2.3217,
"num_tokens": 395861368.0,
"step": 6450
},
{
"epoch": 1.1922494840632882,
"grad_norm": 3.362730026245117,
"learning_rate": 1.738353070129462e-05,
"loss": 2.3555,
"num_tokens": 398934759.0,
"step": 6500
},
{
"epoch": 1.2014216922724146,
"grad_norm": 3.829068899154663,
"learning_rate": 1.7344402735350235e-05,
"loss": 2.3512,
"num_tokens": 401999347.0,
"step": 6550
},
{
"epoch": 1.210593900481541,
"grad_norm": 2.987025737762451,
"learning_rate": 1.7305029108812777e-05,
"loss": 2.3352,
"num_tokens": 405069670.0,
"step": 6600
},
{
"epoch": 1.2197661086906673,
"grad_norm": 3.466670274734497,
"learning_rate": 1.7265411138678178e-05,
"loss": 2.3183,
"num_tokens": 408145682.0,
"step": 6650
},
{
"epoch": 1.2289383168997936,
"grad_norm": 3.421823263168335,
"learning_rate": 1.722555015011533e-05,
"loss": 2.342,
"num_tokens": 411222733.0,
"step": 6700
},
{
"epoch": 1.23811052510892,
"grad_norm": 4.740370750427246,
"learning_rate": 1.7185447476421764e-05,
"loss": 2.33,
"num_tokens": 414293206.0,
"step": 6750
},
{
"epoch": 1.2472827333180463,
"grad_norm": 3.6490085124969482,
"learning_rate": 1.7145104458979074e-05,
"loss": 2.3191,
"num_tokens": 417348351.0,
"step": 6800
},
{
"epoch": 1.2564549415271726,
"grad_norm": 2.6619176864624023,
"learning_rate": 1.7104522447208028e-05,
"loss": 2.3287,
"num_tokens": 420420095.0,
"step": 6850
},
{
"epoch": 1.265627149736299,
"grad_norm": 3.2309610843658447,
"learning_rate": 1.706370279852344e-05,
"loss": 2.3252,
"num_tokens": 423491946.0,
"step": 6900
},
{
"epoch": 1.2747993579454253,
"grad_norm": 3.0654044151306152,
"learning_rate": 1.702264687828877e-05,
"loss": 2.3279,
"num_tokens": 426558795.0,
"step": 6950
},
{
"epoch": 1.2839715661545517,
"grad_norm": 3.152662992477417,
"learning_rate": 1.6981356059770442e-05,
"loss": 2.326,
"num_tokens": 429622896.0,
"step": 7000
},
{
"epoch": 1.293143774363678,
"grad_norm": 2.96634578704834,
"learning_rate": 1.6939831724091927e-05,
"loss": 2.3234,
"num_tokens": 432687860.0,
"step": 7050
},
{
"epoch": 1.3023159825728043,
"grad_norm": 5.338396072387695,
"learning_rate": 1.6898075260187527e-05,
"loss": 2.3119,
"num_tokens": 435752672.0,
"step": 7100
},
{
"epoch": 1.3114881907819307,
"grad_norm": 3.5085184574127197,
"learning_rate": 1.6856088064755936e-05,
"loss": 2.3226,
"num_tokens": 438820637.0,
"step": 7150
},
{
"epoch": 1.320660398991057,
"grad_norm": 2.9175689220428467,
"learning_rate": 1.6813871542213514e-05,
"loss": 2.2943,
"num_tokens": 441888429.0,
"step": 7200
},
{
"epoch": 1.3298326072001834,
"grad_norm": 4.301224708557129,
"learning_rate": 1.6771427104647303e-05,
"loss": 2.3095,
"num_tokens": 444961887.0,
"step": 7250
},
{
"epoch": 1.3390048154093097,
"grad_norm": 3.6543874740600586,
"learning_rate": 1.6728756171767807e-05,
"loss": 2.3122,
"num_tokens": 448044553.0,
"step": 7300
},
{
"epoch": 1.3481770236184363,
"grad_norm": 2.869036912918091,
"learning_rate": 1.66858601708615e-05,
"loss": 2.2963,
"num_tokens": 451118849.0,
"step": 7350
},
{
"epoch": 1.3573492318275626,
"grad_norm": 2.6811025142669678,
"learning_rate": 1.6642740536743086e-05,
"loss": 2.2987,
"num_tokens": 454182648.0,
"step": 7400
},
{
"epoch": 1.366521440036689,
"grad_norm": 3.101973295211792,
"learning_rate": 1.65993987117075e-05,
"loss": 2.2987,
"num_tokens": 457255403.0,
"step": 7450
},
{
"epoch": 1.3756936482458153,
"grad_norm": 5.224907875061035,
"learning_rate": 1.6555836145481673e-05,
"loss": 2.2874,
"num_tokens": 460321588.0,
"step": 7500
},
{
"epoch": 1.3848658564549416,
"grad_norm": 3.0296120643615723,
"learning_rate": 1.6512054295176042e-05,
"loss": 2.2832,
"num_tokens": 463395678.0,
"step": 7550
},
{
"epoch": 1.394038064664068,
"grad_norm": 3.4018895626068115,
"learning_rate": 1.64680546252358e-05,
"loss": 2.2759,
"num_tokens": 466458283.0,
"step": 7600
},
{
"epoch": 1.4032102728731943,
"grad_norm": 4.157562255859375,
"learning_rate": 1.642383860739193e-05,
"loss": 2.3066,
"num_tokens": 469533607.0,
"step": 7650
},
{
"epoch": 1.4123824810823207,
"grad_norm": 3.008226156234741,
"learning_rate": 1.637940772061196e-05,
"loss": 2.2832,
"num_tokens": 472604436.0,
"step": 7700
},
{
"epoch": 1.421554689291447,
"grad_norm": 3.0457003116607666,
"learning_rate": 1.6334763451050496e-05,
"loss": 2.2693,
"num_tokens": 475671029.0,
"step": 7750
},
{
"epoch": 1.4307268975005734,
"grad_norm": 2.6443099975585938,
"learning_rate": 1.628990729199953e-05,
"loss": 2.2919,
"num_tokens": 478727676.0,
"step": 7800
},
{
"epoch": 1.4398991057096997,
"grad_norm": 4.169058322906494,
"learning_rate": 1.6244840743838463e-05,
"loss": 2.271,
"num_tokens": 481796442.0,
"step": 7850
},
{
"epoch": 1.449071313918826,
"grad_norm": 4.345733642578125,
"learning_rate": 1.619956531398395e-05,
"loss": 2.2957,
"num_tokens": 484867456.0,
"step": 7900
},
{
"epoch": 1.4582435221279524,
"grad_norm": 4.380826950073242,
"learning_rate": 1.615408251683945e-05,
"loss": 2.2822,
"num_tokens": 487941363.0,
"step": 7950
},
{
"epoch": 1.4674157303370787,
"grad_norm": 3.4112560749053955,
"learning_rate": 1.6108393873744593e-05,
"loss": 2.287,
"num_tokens": 491013621.0,
"step": 8000
},
{
"epoch": 1.476587938546205,
"grad_norm": 4.045591354370117,
"learning_rate": 1.606250091292428e-05,
"loss": 2.2905,
"num_tokens": 494090566.0,
"step": 8050
},
{
"epoch": 1.4857601467553314,
"grad_norm": 3.067234754562378,
"learning_rate": 1.6016405169437582e-05,
"loss": 2.2676,
"num_tokens": 497162524.0,
"step": 8100
},
{
"epoch": 1.4949323549644578,
"grad_norm": 3.4898037910461426,
"learning_rate": 1.5970108185126362e-05,
"loss": 2.2657,
"num_tokens": 500225621.0,
"step": 8150
},
{
"epoch": 1.504104563173584,
"grad_norm": 3.147956609725952,
"learning_rate": 1.592361150856374e-05,
"loss": 2.2511,
"num_tokens": 503293194.0,
"step": 8200
},
{
"epoch": 1.5132767713827104,
"grad_norm": 2.8965494632720947,
"learning_rate": 1.5876916695002277e-05,
"loss": 2.2704,
"num_tokens": 506366343.0,
"step": 8250
},
{
"epoch": 1.5224489795918368,
"grad_norm": 4.127537250518799,
"learning_rate": 1.5830025306321945e-05,
"loss": 2.2745,
"num_tokens": 509436830.0,
"step": 8300
},
{
"epoch": 1.5316211878009631,
"grad_norm": 3.495630979537964,
"learning_rate": 1.578293891097791e-05,
"loss": 2.2761,
"num_tokens": 512509535.0,
"step": 8350
},
{
"epoch": 1.5407933960100895,
"grad_norm": 2.908043622970581,
"learning_rate": 1.573565908394804e-05,
"loss": 2.2706,
"num_tokens": 515566764.0,
"step": 8400
},
{
"epoch": 1.5499656042192158,
"grad_norm": 3.0703983306884766,
"learning_rate": 1.568818740668024e-05,
"loss": 2.2665,
"num_tokens": 518634607.0,
"step": 8450
},
{
"epoch": 1.5591378124283422,
"grad_norm": 3.2771713733673096,
"learning_rate": 1.564052546703957e-05,
"loss": 2.2468,
"num_tokens": 521690844.0,
"step": 8500
},
{
"epoch": 1.5683100206374685,
"grad_norm": 3.3331332206726074,
"learning_rate": 1.5592674859255093e-05,
"loss": 2.2691,
"num_tokens": 524759416.0,
"step": 8550
},
{
"epoch": 1.5774822288465948,
"grad_norm": 3.091536283493042,
"learning_rate": 1.554463718386659e-05,
"loss": 2.2526,
"num_tokens": 527834639.0,
"step": 8600
},
{
"epoch": 1.5866544370557212,
"grad_norm": 2.8725991249084473,
"learning_rate": 1.5496414047670985e-05,
"loss": 2.2763,
"num_tokens": 530900064.0,
"step": 8650
},
{
"epoch": 1.5958266452648475,
"grad_norm": 5.627836227416992,
"learning_rate": 1.5448007063668645e-05,
"loss": 2.262,
"num_tokens": 533979089.0,
"step": 8700
},
{
"epoch": 1.6049988534739739,
"grad_norm": 2.99426007270813,
"learning_rate": 1.5399417851009394e-05,
"loss": 2.243,
"num_tokens": 537049251.0,
"step": 8750
},
{
"epoch": 1.6141710616831002,
"grad_norm": 2.950291395187378,
"learning_rate": 1.5350648034938364e-05,
"loss": 2.2536,
"num_tokens": 540115638.0,
"step": 8800
},
{
"epoch": 1.6233432698922265,
"grad_norm": 4.661383152008057,
"learning_rate": 1.5301699246741626e-05,
"loss": 2.2568,
"num_tokens": 543188518.0,
"step": 8850
},
{
"epoch": 1.632515478101353,
"grad_norm": 3.3682801723480225,
"learning_rate": 1.5252573123691654e-05,
"loss": 2.269,
"num_tokens": 546259101.0,
"step": 8900
},
{
"epoch": 1.6416876863104792,
"grad_norm": 3.7050347328186035,
"learning_rate": 1.520327130899252e-05,
"loss": 2.2242,
"num_tokens": 549332246.0,
"step": 8950
},
{
"epoch": 1.6508598945196056,
"grad_norm": 2.5407910346984863,
"learning_rate": 1.5153795451724953e-05,
"loss": 2.2399,
"num_tokens": 552401113.0,
"step": 9000
},
{
"epoch": 1.660032102728732,
"grad_norm": 2.9334683418273926,
"learning_rate": 1.5104147206791182e-05,
"loss": 2.23,
"num_tokens": 555478577.0,
"step": 9050
},
{
"epoch": 1.6692043109378583,
"grad_norm": 3.646679401397705,
"learning_rate": 1.5054328234859572e-05,
"loss": 2.2431,
"num_tokens": 558544082.0,
"step": 9100
},
{
"epoch": 1.6783765191469846,
"grad_norm": 3.0259110927581787,
"learning_rate": 1.5004340202309089e-05,
"loss": 2.2374,
"num_tokens": 561616790.0,
"step": 9150
},
{
"epoch": 1.687548727356111,
"grad_norm": 5.128026485443115,
"learning_rate": 1.495418478117354e-05,
"loss": 2.2372,
"num_tokens": 564687010.0,
"step": 9200
},
{
"epoch": 1.6967209355652373,
"grad_norm": 2.761215925216675,
"learning_rate": 1.4903863649085668e-05,
"loss": 2.2316,
"num_tokens": 567756890.0,
"step": 9250
},
{
"epoch": 1.7058931437743636,
"grad_norm": 3.4462618827819824,
"learning_rate": 1.4853378489221033e-05,
"loss": 2.2451,
"num_tokens": 570823485.0,
"step": 9300
},
{
"epoch": 1.71506535198349,
"grad_norm": 3.8544795513153076,
"learning_rate": 1.4802730990241702e-05,
"loss": 2.2442,
"num_tokens": 573890500.0,
"step": 9350
},
{
"epoch": 1.7242375601926163,
"grad_norm": 3.1527886390686035,
"learning_rate": 1.475192284623977e-05,
"loss": 2.2257,
"num_tokens": 576956808.0,
"step": 9400
},
{
"epoch": 1.7334097684017427,
"grad_norm": 3.3589236736297607,
"learning_rate": 1.4700955756680704e-05,
"loss": 2.2256,
"num_tokens": 580028975.0,
"step": 9450
},
{
"epoch": 1.742581976610869,
"grad_norm": 2.782637119293213,
"learning_rate": 1.464983142634648e-05,
"loss": 2.2284,
"num_tokens": 583101191.0,
"step": 9500
},
{
"epoch": 1.7517541848199953,
"grad_norm": 2.791273832321167,
"learning_rate": 1.4598551565278578e-05,
"loss": 2.2466,
"num_tokens": 586177194.0,
"step": 9550
},
{
"epoch": 1.7609263930291217,
"grad_norm": 3.3659918308258057,
"learning_rate": 1.4547117888720776e-05,
"loss": 2.2151,
"num_tokens": 589248537.0,
"step": 9600
},
{
"epoch": 1.770098601238248,
"grad_norm": 3.0505056381225586,
"learning_rate": 1.4495532117061772e-05,
"loss": 2.2645,
"num_tokens": 592318226.0,
"step": 9650
},
{
"epoch": 1.7792708094473744,
"grad_norm": 3.632218360900879,
"learning_rate": 1.4443795975777652e-05,
"loss": 2.2267,
"num_tokens": 595386393.0,
"step": 9700
},
{
"epoch": 1.7884430176565007,
"grad_norm": 3.192918300628662,
"learning_rate": 1.4391911195374164e-05,
"loss": 2.204,
"num_tokens": 598456081.0,
"step": 9750
},
{
"epoch": 1.797615225865627,
"grad_norm": 2.828756809234619,
"learning_rate": 1.4339879511328837e-05,
"loss": 2.2332,
"num_tokens": 601526298.0,
"step": 9800
},
{
"epoch": 1.8067874340747534,
"grad_norm": 2.9059629440307617,
"learning_rate": 1.4287702664032939e-05,
"loss": 2.2152,
"num_tokens": 604587712.0,
"step": 9850
},
{
"epoch": 1.8159596422838797,
"grad_norm": 3.254810094833374,
"learning_rate": 1.4235382398733247e-05,
"loss": 2.21,
"num_tokens": 607658810.0,
"step": 9900
},
{
"epoch": 1.825131850493006,
"grad_norm": 2.805896043777466,
"learning_rate": 1.4182920465473692e-05,
"loss": 2.216,
"num_tokens": 610716018.0,
"step": 9950
},
{
"epoch": 1.8343040587021324,
"grad_norm": 3.190450429916382,
"learning_rate": 1.4130318619036808e-05,
"loss": 2.2301,
"num_tokens": 613780917.0,
"step": 10000
},
{
"epoch": 1.8434762669112588,
"grad_norm": 2.745699644088745,
"learning_rate": 1.407757861888504e-05,
"loss": 2.2432,
"num_tokens": 616850698.0,
"step": 10050
},
{
"epoch": 1.852648475120385,
"grad_norm": 3.0942554473876953,
"learning_rate": 1.4024702229101885e-05,
"loss": 2.2301,
"num_tokens": 619916687.0,
"step": 10100
},
{
"epoch": 1.8618206833295114,
"grad_norm": 2.85457444190979,
"learning_rate": 1.3971691218332912e-05,
"loss": 2.205,
"num_tokens": 622985743.0,
"step": 10150
},
{
"epoch": 1.8709928915386378,
"grad_norm": 3.447561025619507,
"learning_rate": 1.3918547359726569e-05,
"loss": 2.2299,
"num_tokens": 626061264.0,
"step": 10200
},
{
"epoch": 1.8801650997477641,
"grad_norm": 3.3711204528808594,
"learning_rate": 1.3865272430874892e-05,
"loss": 2.2207,
"num_tokens": 629127449.0,
"step": 10250
},
{
"epoch": 1.8893373079568905,
"grad_norm": 3.6021645069122314,
"learning_rate": 1.3811868213754043e-05,
"loss": 2.2439,
"num_tokens": 632195091.0,
"step": 10300
},
{
"epoch": 1.8985095161660168,
"grad_norm": 2.6557374000549316,
"learning_rate": 1.3758336494664717e-05,
"loss": 2.2255,
"num_tokens": 635272904.0,
"step": 10350
},
{
"epoch": 1.9076817243751432,
"grad_norm": 2.8971426486968994,
"learning_rate": 1.3704679064172367e-05,
"loss": 2.1996,
"num_tokens": 638348466.0,
"step": 10400
},
{
"epoch": 1.9168539325842695,
"grad_norm": 2.906564950942993,
"learning_rate": 1.3650897717047336e-05,
"loss": 2.2005,
"num_tokens": 641418105.0,
"step": 10450
},
{
"epoch": 1.9260261407933958,
"grad_norm": 3.114825487136841,
"learning_rate": 1.359699425220481e-05,
"loss": 2.2054,
"num_tokens": 644495331.0,
"step": 10500
},
{
"epoch": 1.9351983490025224,
"grad_norm": 2.6857142448425293,
"learning_rate": 1.3542970472644656e-05,
"loss": 2.2049,
"num_tokens": 647574515.0,
"step": 10550
},
{
"epoch": 1.9443705572116488,
"grad_norm": 3.0475053787231445,
"learning_rate": 1.348882818539111e-05,
"loss": 2.2099,
"num_tokens": 650651609.0,
"step": 10600
},
{
"epoch": 1.953542765420775,
"grad_norm": 2.635979413986206,
"learning_rate": 1.3434569201432326e-05,
"loss": 2.1955,
"num_tokens": 653727158.0,
"step": 10650
},
{
"epoch": 1.9627149736299014,
"grad_norm": 2.658064365386963,
"learning_rate": 1.3380195335659822e-05,
"loss": 2.215,
"num_tokens": 656801882.0,
"step": 10700
},
{
"epoch": 1.9718871818390278,
"grad_norm": 2.7926363945007324,
"learning_rate": 1.3325708406807753e-05,
"loss": 2.1944,
"num_tokens": 659872536.0,
"step": 10750
},
{
"epoch": 1.9810593900481541,
"grad_norm": 2.6653809547424316,
"learning_rate": 1.3271110237392083e-05,
"loss": 2.2009,
"num_tokens": 662940974.0,
"step": 10800
},
{
"epoch": 1.9902315982572805,
"grad_norm": 2.3410346508026123,
"learning_rate": 1.3216402653649631e-05,
"loss": 2.1746,
"num_tokens": 665999433.0,
"step": 10850
},
{
"epoch": 1.9994038064664068,
"grad_norm": 2.8723654747009277,
"learning_rate": 1.3161587485476978e-05,
"loss": 2.225,
"num_tokens": 669062836.0,
"step": 10900
},
{
"epoch": 2.0,
"eval_loss": 2.4245221614837646,
"eval_num_tokens": 669261246.0,
"eval_runtime": 5.3377,
"eval_samples_per_second": 90.863,
"eval_steps_per_second": 5.808,
"step": 10904
}
],
"logging_steps": 50,
"max_steps": 27260,
"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": 3.41548107238656e+17,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}