4440 lines
112 KiB
JSON
4440 lines
112 KiB
JSON
{
|
|
"best_global_step": 21808,
|
|
"best_metric": 2.3043127059936523,
|
|
"best_model_checkpoint": "/common/home/users/m/mq.nguyen.2023/scratchDirectory/finetune/train_nbtpj_summ_gpt2_inst_cnnxsumsam/checkpoint-21808",
|
|
"epoch": 5.0,
|
|
"eval_steps": 500,
|
|
"global_step": 27260,
|
|
"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
|
|
},
|
|
{
|
|
"epoch": 2.0084384315523964,
|
|
"grad_norm": 2.9814906120300293,
|
|
"learning_rate": 1.3106666566369266e-05,
|
|
"loss": 2.1629,
|
|
"num_tokens": 672079827.0,
|
|
"step": 10950
|
|
},
|
|
{
|
|
"epoch": 2.0176106397615228,
|
|
"grad_norm": 2.8018527030944824,
|
|
"learning_rate": 1.3051641733358869e-05,
|
|
"loss": 2.179,
|
|
"num_tokens": 675156433.0,
|
|
"step": 11000
|
|
},
|
|
{
|
|
"epoch": 2.026782847970649,
|
|
"grad_norm": 2.977572202682495,
|
|
"learning_rate": 1.2996514826953931e-05,
|
|
"loss": 2.1796,
|
|
"num_tokens": 678221060.0,
|
|
"step": 11050
|
|
},
|
|
{
|
|
"epoch": 2.0359550561797755,
|
|
"grad_norm": 3.5244157314300537,
|
|
"learning_rate": 1.2941287691076833e-05,
|
|
"loss": 2.1746,
|
|
"num_tokens": 681279496.0,
|
|
"step": 11100
|
|
},
|
|
{
|
|
"epoch": 2.045127264388902,
|
|
"grad_norm": 4.0321431159973145,
|
|
"learning_rate": 1.288596217300249e-05,
|
|
"loss": 2.1692,
|
|
"num_tokens": 684357409.0,
|
|
"step": 11150
|
|
},
|
|
{
|
|
"epoch": 2.054299472598028,
|
|
"grad_norm": 3.5743963718414307,
|
|
"learning_rate": 1.2830540123296568e-05,
|
|
"loss": 2.1903,
|
|
"num_tokens": 687419819.0,
|
|
"step": 11200
|
|
},
|
|
{
|
|
"epoch": 2.0634716808071545,
|
|
"grad_norm": 2.714646339416504,
|
|
"learning_rate": 1.27750233957536e-05,
|
|
"loss": 2.1826,
|
|
"num_tokens": 690478740.0,
|
|
"step": 11250
|
|
},
|
|
{
|
|
"epoch": 2.072643889016281,
|
|
"grad_norm": 3.1788077354431152,
|
|
"learning_rate": 1.2719413847334963e-05,
|
|
"loss": 2.1772,
|
|
"num_tokens": 693539914.0,
|
|
"step": 11300
|
|
},
|
|
{
|
|
"epoch": 2.081816097225407,
|
|
"grad_norm": 2.851320266723633,
|
|
"learning_rate": 1.2663713338106765e-05,
|
|
"loss": 2.1902,
|
|
"num_tokens": 696609574.0,
|
|
"step": 11350
|
|
},
|
|
{
|
|
"epoch": 2.0909883054345335,
|
|
"grad_norm": 3.043736219406128,
|
|
"learning_rate": 1.2607923731177642e-05,
|
|
"loss": 2.164,
|
|
"num_tokens": 699671518.0,
|
|
"step": 11400
|
|
},
|
|
{
|
|
"epoch": 2.10016051364366,
|
|
"grad_norm": 2.4879658222198486,
|
|
"learning_rate": 1.2552046892636427e-05,
|
|
"loss": 2.1851,
|
|
"num_tokens": 702735604.0,
|
|
"step": 11450
|
|
},
|
|
{
|
|
"epoch": 2.109332721852786,
|
|
"grad_norm": 3.014916181564331,
|
|
"learning_rate": 1.2496084691489731e-05,
|
|
"loss": 2.167,
|
|
"num_tokens": 705808446.0,
|
|
"step": 11500
|
|
},
|
|
{
|
|
"epoch": 2.1185049300619125,
|
|
"grad_norm": 5.272231101989746,
|
|
"learning_rate": 1.2440038999599438e-05,
|
|
"loss": 2.1686,
|
|
"num_tokens": 708874044.0,
|
|
"step": 11550
|
|
},
|
|
{
|
|
"epoch": 2.127677138271039,
|
|
"grad_norm": 2.8708434104919434,
|
|
"learning_rate": 1.2383911691620083e-05,
|
|
"loss": 2.1651,
|
|
"num_tokens": 711930253.0,
|
|
"step": 11600
|
|
},
|
|
{
|
|
"epoch": 2.1368493464801652,
|
|
"grad_norm": 2.5448362827301025,
|
|
"learning_rate": 1.2327704644936156e-05,
|
|
"loss": 2.1723,
|
|
"num_tokens": 714992067.0,
|
|
"step": 11650
|
|
},
|
|
{
|
|
"epoch": 2.1460215546892916,
|
|
"grad_norm": 2.435708999633789,
|
|
"learning_rate": 1.22714197395993e-05,
|
|
"loss": 2.1682,
|
|
"num_tokens": 718068361.0,
|
|
"step": 11700
|
|
},
|
|
{
|
|
"epoch": 2.155193762898418,
|
|
"grad_norm": 2.605455160140991,
|
|
"learning_rate": 1.2215058858265423e-05,
|
|
"loss": 2.178,
|
|
"num_tokens": 721130428.0,
|
|
"step": 11750
|
|
},
|
|
{
|
|
"epoch": 2.1643659711075443,
|
|
"grad_norm": 3.2399065494537354,
|
|
"learning_rate": 1.2158623886131733e-05,
|
|
"loss": 2.1682,
|
|
"num_tokens": 724200490.0,
|
|
"step": 11800
|
|
},
|
|
{
|
|
"epoch": 2.1735381793166706,
|
|
"grad_norm": 2.644134998321533,
|
|
"learning_rate": 1.2102116710873681e-05,
|
|
"loss": 2.1561,
|
|
"num_tokens": 727279410.0,
|
|
"step": 11850
|
|
},
|
|
{
|
|
"epoch": 2.182710387525797,
|
|
"grad_norm": 2.811669111251831,
|
|
"learning_rate": 1.2045539222581812e-05,
|
|
"loss": 2.168,
|
|
"num_tokens": 730340441.0,
|
|
"step": 11900
|
|
},
|
|
{
|
|
"epoch": 2.1918825957349233,
|
|
"grad_norm": 3.6238574981689453,
|
|
"learning_rate": 1.1988893313698556e-05,
|
|
"loss": 2.1496,
|
|
"num_tokens": 733410874.0,
|
|
"step": 11950
|
|
},
|
|
{
|
|
"epoch": 2.2010548039440496,
|
|
"grad_norm": 2.5450801849365234,
|
|
"learning_rate": 1.1932180878954913e-05,
|
|
"loss": 2.1789,
|
|
"num_tokens": 736478072.0,
|
|
"step": 12000
|
|
},
|
|
{
|
|
"epoch": 2.210227012153176,
|
|
"grad_norm": 3.001086711883545,
|
|
"learning_rate": 1.1875403815307092e-05,
|
|
"loss": 2.1639,
|
|
"num_tokens": 739547375.0,
|
|
"step": 12050
|
|
},
|
|
{
|
|
"epoch": 2.2193992203623023,
|
|
"grad_norm": 2.8402042388916016,
|
|
"learning_rate": 1.1818564021873051e-05,
|
|
"loss": 2.1697,
|
|
"num_tokens": 742621016.0,
|
|
"step": 12100
|
|
},
|
|
{
|
|
"epoch": 2.2285714285714286,
|
|
"grad_norm": 3.5109944343566895,
|
|
"learning_rate": 1.1761663399868972e-05,
|
|
"loss": 2.1792,
|
|
"num_tokens": 745682838.0,
|
|
"step": 12150
|
|
},
|
|
{
|
|
"epoch": 2.237743636780555,
|
|
"grad_norm": 3.4116134643554688,
|
|
"learning_rate": 1.1704703852545685e-05,
|
|
"loss": 2.1695,
|
|
"num_tokens": 748753387.0,
|
|
"step": 12200
|
|
},
|
|
{
|
|
"epoch": 2.2469158449896813,
|
|
"grad_norm": 3.222461700439453,
|
|
"learning_rate": 1.1647687285124985e-05,
|
|
"loss": 2.1513,
|
|
"num_tokens": 751827240.0,
|
|
"step": 12250
|
|
},
|
|
{
|
|
"epoch": 2.2560880531988077,
|
|
"grad_norm": 2.8149147033691406,
|
|
"learning_rate": 1.1590615604735913e-05,
|
|
"loss": 2.1506,
|
|
"num_tokens": 754897622.0,
|
|
"step": 12300
|
|
},
|
|
{
|
|
"epoch": 2.265260261407934,
|
|
"grad_norm": 3.038508176803589,
|
|
"learning_rate": 1.1533490720350972e-05,
|
|
"loss": 2.1518,
|
|
"num_tokens": 757971210.0,
|
|
"step": 12350
|
|
},
|
|
{
|
|
"epoch": 2.2744324696170604,
|
|
"grad_norm": 2.754978656768799,
|
|
"learning_rate": 1.1476314542722267e-05,
|
|
"loss": 2.1507,
|
|
"num_tokens": 761038532.0,
|
|
"step": 12400
|
|
},
|
|
{
|
|
"epoch": 2.2836046778261867,
|
|
"grad_norm": 2.7081894874572754,
|
|
"learning_rate": 1.1419088984317589e-05,
|
|
"loss": 2.1854,
|
|
"num_tokens": 764119294.0,
|
|
"step": 12450
|
|
},
|
|
{
|
|
"epoch": 2.292776886035313,
|
|
"grad_norm": 3.2845664024353027,
|
|
"learning_rate": 1.1361815959256456e-05,
|
|
"loss": 2.1522,
|
|
"num_tokens": 767188411.0,
|
|
"step": 12500
|
|
},
|
|
{
|
|
"epoch": 2.3019490942444394,
|
|
"grad_norm": 3.1656436920166016,
|
|
"learning_rate": 1.1304497383246083e-05,
|
|
"loss": 2.162,
|
|
"num_tokens": 770249106.0,
|
|
"step": 12550
|
|
},
|
|
{
|
|
"epoch": 2.3111213024535657,
|
|
"grad_norm": 3.843442440032959,
|
|
"learning_rate": 1.1247135173517303e-05,
|
|
"loss": 2.1501,
|
|
"num_tokens": 773312866.0,
|
|
"step": 12600
|
|
},
|
|
{
|
|
"epoch": 2.320293510662692,
|
|
"grad_norm": 2.4421207904815674,
|
|
"learning_rate": 1.1189731248760438e-05,
|
|
"loss": 2.1549,
|
|
"num_tokens": 776389614.0,
|
|
"step": 12650
|
|
},
|
|
{
|
|
"epoch": 2.3294657188718184,
|
|
"grad_norm": 2.5875680446624756,
|
|
"learning_rate": 1.1132287529061127e-05,
|
|
"loss": 2.1563,
|
|
"num_tokens": 779462718.0,
|
|
"step": 12700
|
|
},
|
|
{
|
|
"epoch": 2.3386379270809448,
|
|
"grad_norm": 3.222823143005371,
|
|
"learning_rate": 1.1074805935836096e-05,
|
|
"loss": 2.1493,
|
|
"num_tokens": 782526270.0,
|
|
"step": 12750
|
|
},
|
|
{
|
|
"epoch": 2.347810135290071,
|
|
"grad_norm": 2.859107494354248,
|
|
"learning_rate": 1.1017288391768885e-05,
|
|
"loss": 2.1694,
|
|
"num_tokens": 785606459.0,
|
|
"step": 12800
|
|
},
|
|
{
|
|
"epoch": 2.3569823434991974,
|
|
"grad_norm": 3.297732353210449,
|
|
"learning_rate": 1.095973682074555e-05,
|
|
"loss": 2.1502,
|
|
"num_tokens": 788671501.0,
|
|
"step": 12850
|
|
},
|
|
{
|
|
"epoch": 2.366154551708324,
|
|
"grad_norm": 2.5810225009918213,
|
|
"learning_rate": 1.09021531477903e-05,
|
|
"loss": 2.168,
|
|
"num_tokens": 791747077.0,
|
|
"step": 12900
|
|
},
|
|
{
|
|
"epoch": 2.37532675991745,
|
|
"grad_norm": 2.870887517929077,
|
|
"learning_rate": 1.0844539299001112e-05,
|
|
"loss": 2.1531,
|
|
"num_tokens": 794816268.0,
|
|
"step": 12950
|
|
},
|
|
{
|
|
"epoch": 2.3844989681265765,
|
|
"grad_norm": 3.0016744136810303,
|
|
"learning_rate": 1.0786897201485303e-05,
|
|
"loss": 2.1369,
|
|
"num_tokens": 797885120.0,
|
|
"step": 13000
|
|
},
|
|
{
|
|
"epoch": 2.393671176335703,
|
|
"grad_norm": 3.464268684387207,
|
|
"learning_rate": 1.0729228783295072e-05,
|
|
"loss": 2.146,
|
|
"num_tokens": 800951727.0,
|
|
"step": 13050
|
|
},
|
|
{
|
|
"epoch": 2.402843384544829,
|
|
"grad_norm": 2.76597261428833,
|
|
"learning_rate": 1.0671535973363019e-05,
|
|
"loss": 2.1607,
|
|
"num_tokens": 804020260.0,
|
|
"step": 13100
|
|
},
|
|
{
|
|
"epoch": 2.4120155927539555,
|
|
"grad_norm": 2.9096412658691406,
|
|
"learning_rate": 1.06138207014376e-05,
|
|
"loss": 2.141,
|
|
"num_tokens": 807091595.0,
|
|
"step": 13150
|
|
},
|
|
{
|
|
"epoch": 2.421187800963082,
|
|
"grad_norm": 3.2226877212524414,
|
|
"learning_rate": 1.0556084898018615e-05,
|
|
"loss": 2.1548,
|
|
"num_tokens": 810167283.0,
|
|
"step": 13200
|
|
},
|
|
{
|
|
"epoch": 2.430360009172208,
|
|
"grad_norm": 2.603027820587158,
|
|
"learning_rate": 1.0498330494292596e-05,
|
|
"loss": 2.149,
|
|
"num_tokens": 813226356.0,
|
|
"step": 13250
|
|
},
|
|
{
|
|
"epoch": 2.4395322173813345,
|
|
"grad_norm": 3.0920653343200684,
|
|
"learning_rate": 1.0440559422068241e-05,
|
|
"loss": 2.1495,
|
|
"num_tokens": 816295716.0,
|
|
"step": 13300
|
|
},
|
|
{
|
|
"epoch": 2.448704425590461,
|
|
"grad_norm": 2.6723990440368652,
|
|
"learning_rate": 1.0382773613711787e-05,
|
|
"loss": 2.1556,
|
|
"num_tokens": 819371405.0,
|
|
"step": 13350
|
|
},
|
|
{
|
|
"epoch": 2.457876633799587,
|
|
"grad_norm": 2.6446688175201416,
|
|
"learning_rate": 1.032497500208237e-05,
|
|
"loss": 2.1454,
|
|
"num_tokens": 822442259.0,
|
|
"step": 13400
|
|
},
|
|
{
|
|
"epoch": 2.4670488420087135,
|
|
"grad_norm": 2.5830724239349365,
|
|
"learning_rate": 1.0267165520467385e-05,
|
|
"loss": 2.1227,
|
|
"num_tokens": 825509067.0,
|
|
"step": 13450
|
|
},
|
|
{
|
|
"epoch": 2.47622105021784,
|
|
"grad_norm": 2.625577688217163,
|
|
"learning_rate": 1.020934710251781e-05,
|
|
"loss": 2.1478,
|
|
"num_tokens": 828588777.0,
|
|
"step": 13500
|
|
},
|
|
{
|
|
"epoch": 2.4853932584269662,
|
|
"grad_norm": 2.7063612937927246,
|
|
"learning_rate": 1.0151521682183534e-05,
|
|
"loss": 2.1322,
|
|
"num_tokens": 831656407.0,
|
|
"step": 13550
|
|
},
|
|
{
|
|
"epoch": 2.4945654666360926,
|
|
"grad_norm": 2.5114853382110596,
|
|
"learning_rate": 1.0093691193648666e-05,
|
|
"loss": 2.1542,
|
|
"num_tokens": 834726053.0,
|
|
"step": 13600
|
|
},
|
|
{
|
|
"epoch": 2.503737674845219,
|
|
"grad_norm": 2.416917324066162,
|
|
"learning_rate": 1.0035857571266832e-05,
|
|
"loss": 2.1343,
|
|
"num_tokens": 837801784.0,
|
|
"step": 13650
|
|
},
|
|
{
|
|
"epoch": 2.5129098830543453,
|
|
"grad_norm": 2.9593300819396973,
|
|
"learning_rate": 9.978022749496495e-06,
|
|
"loss": 2.151,
|
|
"num_tokens": 840873527.0,
|
|
"step": 13700
|
|
},
|
|
{
|
|
"epoch": 2.5220820912634716,
|
|
"grad_norm": 2.8286054134368896,
|
|
"learning_rate": 9.920188662836223e-06,
|
|
"loss": 2.1556,
|
|
"num_tokens": 843955836.0,
|
|
"step": 13750
|
|
},
|
|
{
|
|
"epoch": 2.531254299472598,
|
|
"grad_norm": 2.9257073402404785,
|
|
"learning_rate": 9.862357245760006e-06,
|
|
"loss": 2.1223,
|
|
"num_tokens": 847025588.0,
|
|
"step": 13800
|
|
},
|
|
{
|
|
"epoch": 2.5404265076817243,
|
|
"grad_norm": 3.1148195266723633,
|
|
"learning_rate": 9.804530432652533e-06,
|
|
"loss": 2.1481,
|
|
"num_tokens": 850085106.0,
|
|
"step": 13850
|
|
},
|
|
{
|
|
"epoch": 2.5495987158908506,
|
|
"grad_norm": 3.747142791748047,
|
|
"learning_rate": 9.7467101577445e-06,
|
|
"loss": 2.1562,
|
|
"num_tokens": 853174410.0,
|
|
"step": 13900
|
|
},
|
|
{
|
|
"epoch": 2.558770924099977,
|
|
"grad_norm": 2.368061065673828,
|
|
"learning_rate": 9.688898355047905e-06,
|
|
"loss": 2.1477,
|
|
"num_tokens": 856251300.0,
|
|
"step": 13950
|
|
},
|
|
{
|
|
"epoch": 2.5679431323091033,
|
|
"grad_norm": 2.6596877574920654,
|
|
"learning_rate": 9.631096958291361e-06,
|
|
"loss": 2.1328,
|
|
"num_tokens": 859334390.0,
|
|
"step": 14000
|
|
},
|
|
{
|
|
"epoch": 2.5771153405182297,
|
|
"grad_norm": 2.4528298377990723,
|
|
"learning_rate": 9.573307900855428e-06,
|
|
"loss": 2.1233,
|
|
"num_tokens": 862423963.0,
|
|
"step": 14050
|
|
},
|
|
{
|
|
"epoch": 2.586287548727356,
|
|
"grad_norm": 2.7410616874694824,
|
|
"learning_rate": 9.515533115707915e-06,
|
|
"loss": 2.1542,
|
|
"num_tokens": 865499353.0,
|
|
"step": 14100
|
|
},
|
|
{
|
|
"epoch": 2.5954597569364823,
|
|
"grad_norm": 2.893670082092285,
|
|
"learning_rate": 9.457774535339251e-06,
|
|
"loss": 2.1284,
|
|
"num_tokens": 868572178.0,
|
|
"step": 14150
|
|
},
|
|
{
|
|
"epoch": 2.6046319651456087,
|
|
"grad_norm": 2.4961345195770264,
|
|
"learning_rate": 9.40003409169784e-06,
|
|
"loss": 2.1387,
|
|
"num_tokens": 871636138.0,
|
|
"step": 14200
|
|
},
|
|
{
|
|
"epoch": 2.613804173354735,
|
|
"grad_norm": 2.655097723007202,
|
|
"learning_rate": 9.342313716125427e-06,
|
|
"loss": 2.1272,
|
|
"num_tokens": 874711460.0,
|
|
"step": 14250
|
|
},
|
|
{
|
|
"epoch": 2.6229763815638614,
|
|
"grad_norm": 2.478729724884033,
|
|
"learning_rate": 9.284615339292515e-06,
|
|
"loss": 2.1134,
|
|
"num_tokens": 877775972.0,
|
|
"step": 14300
|
|
},
|
|
{
|
|
"epoch": 2.6321485897729877,
|
|
"grad_norm": 2.3555896282196045,
|
|
"learning_rate": 9.22694089113377e-06,
|
|
"loss": 2.1121,
|
|
"num_tokens": 880850816.0,
|
|
"step": 14350
|
|
},
|
|
{
|
|
"epoch": 2.641320797982114,
|
|
"grad_norm": 3.220005512237549,
|
|
"learning_rate": 9.16929230078348e-06,
|
|
"loss": 2.1338,
|
|
"num_tokens": 883915833.0,
|
|
"step": 14400
|
|
},
|
|
{
|
|
"epoch": 2.6504930061912404,
|
|
"grad_norm": 2.8448081016540527,
|
|
"learning_rate": 9.111671496511027e-06,
|
|
"loss": 2.1302,
|
|
"num_tokens": 886986053.0,
|
|
"step": 14450
|
|
},
|
|
{
|
|
"epoch": 2.6596652144003667,
|
|
"grad_norm": 2.555471181869507,
|
|
"learning_rate": 9.05408040565638e-06,
|
|
"loss": 2.1248,
|
|
"num_tokens": 890052402.0,
|
|
"step": 14500
|
|
},
|
|
{
|
|
"epoch": 2.668837422609493,
|
|
"grad_norm": 2.9007201194763184,
|
|
"learning_rate": 8.996520954565632e-06,
|
|
"loss": 2.1339,
|
|
"num_tokens": 893109082.0,
|
|
"step": 14550
|
|
},
|
|
{
|
|
"epoch": 2.6780096308186194,
|
|
"grad_norm": 4.271463394165039,
|
|
"learning_rate": 8.938995068526573e-06,
|
|
"loss": 2.1229,
|
|
"num_tokens": 896176023.0,
|
|
"step": 14600
|
|
},
|
|
{
|
|
"epoch": 2.6871818390277458,
|
|
"grad_norm": 2.6032025814056396,
|
|
"learning_rate": 8.881504671704282e-06,
|
|
"loss": 2.1437,
|
|
"num_tokens": 899239231.0,
|
|
"step": 14650
|
|
},
|
|
{
|
|
"epoch": 2.6963540472368726,
|
|
"grad_norm": 2.927783489227295,
|
|
"learning_rate": 8.824051687076774e-06,
|
|
"loss": 2.1421,
|
|
"num_tokens": 902305139.0,
|
|
"step": 14700
|
|
},
|
|
{
|
|
"epoch": 2.7055262554459985,
|
|
"grad_norm": 2.9317362308502197,
|
|
"learning_rate": 8.76663803637067e-06,
|
|
"loss": 2.1258,
|
|
"num_tokens": 905384894.0,
|
|
"step": 14750
|
|
},
|
|
{
|
|
"epoch": 2.7146984636551252,
|
|
"grad_norm": 2.6969382762908936,
|
|
"learning_rate": 8.709265639996932e-06,
|
|
"loss": 2.132,
|
|
"num_tokens": 908453576.0,
|
|
"step": 14800
|
|
},
|
|
{
|
|
"epoch": 2.723870671864251,
|
|
"grad_norm": 2.784421443939209,
|
|
"learning_rate": 8.651936416986608e-06,
|
|
"loss": 2.1265,
|
|
"num_tokens": 911520406.0,
|
|
"step": 14850
|
|
},
|
|
{
|
|
"epoch": 2.733042880073378,
|
|
"grad_norm": 2.754910469055176,
|
|
"learning_rate": 8.59465228492666e-06,
|
|
"loss": 2.1287,
|
|
"num_tokens": 914588413.0,
|
|
"step": 14900
|
|
},
|
|
{
|
|
"epoch": 2.742215088282504,
|
|
"grad_norm": 2.538515329360962,
|
|
"learning_rate": 8.53741515989582e-06,
|
|
"loss": 2.1435,
|
|
"num_tokens": 917659649.0,
|
|
"step": 14950
|
|
},
|
|
{
|
|
"epoch": 2.7513872964916306,
|
|
"grad_norm": 3.178069591522217,
|
|
"learning_rate": 8.480226956400491e-06,
|
|
"loss": 2.1234,
|
|
"num_tokens": 920724902.0,
|
|
"step": 15000
|
|
},
|
|
{
|
|
"epoch": 2.7605595047007565,
|
|
"grad_norm": 3.3320059776306152,
|
|
"learning_rate": 8.423089587310718e-06,
|
|
"loss": 2.1176,
|
|
"num_tokens": 923803197.0,
|
|
"step": 15050
|
|
},
|
|
{
|
|
"epoch": 2.7697317129098833,
|
|
"grad_norm": 2.3997714519500732,
|
|
"learning_rate": 8.366004963796202e-06,
|
|
"loss": 2.1208,
|
|
"num_tokens": 926880331.0,
|
|
"step": 15100
|
|
},
|
|
{
|
|
"epoch": 2.778903921119009,
|
|
"grad_norm": 2.872494697570801,
|
|
"learning_rate": 8.308974995262375e-06,
|
|
"loss": 2.1405,
|
|
"num_tokens": 929947114.0,
|
|
"step": 15150
|
|
},
|
|
{
|
|
"epoch": 2.788076129328136,
|
|
"grad_norm": 2.511892557144165,
|
|
"learning_rate": 8.25200158928653e-06,
|
|
"loss": 2.1139,
|
|
"num_tokens": 933010582.0,
|
|
"step": 15200
|
|
},
|
|
{
|
|
"epoch": 2.797248337537262,
|
|
"grad_norm": 2.8765101432800293,
|
|
"learning_rate": 8.195086651554013e-06,
|
|
"loss": 2.1321,
|
|
"num_tokens": 936084714.0,
|
|
"step": 15250
|
|
},
|
|
{
|
|
"epoch": 2.8064205457463887,
|
|
"grad_norm": 2.4140682220458984,
|
|
"learning_rate": 8.138232085794494e-06,
|
|
"loss": 2.1322,
|
|
"num_tokens": 939135927.0,
|
|
"step": 15300
|
|
},
|
|
{
|
|
"epoch": 2.8155927539555146,
|
|
"grad_norm": 3.0238280296325684,
|
|
"learning_rate": 8.081439793718271e-06,
|
|
"loss": 2.1255,
|
|
"num_tokens": 942203988.0,
|
|
"step": 15350
|
|
},
|
|
{
|
|
"epoch": 2.8247649621646413,
|
|
"grad_norm": 3.9417409896850586,
|
|
"learning_rate": 8.024711674952678e-06,
|
|
"loss": 2.1234,
|
|
"num_tokens": 945271149.0,
|
|
"step": 15400
|
|
},
|
|
{
|
|
"epoch": 2.8339371703737672,
|
|
"grad_norm": 2.6532320976257324,
|
|
"learning_rate": 7.968049626978527e-06,
|
|
"loss": 2.1159,
|
|
"num_tokens": 948330185.0,
|
|
"step": 15450
|
|
},
|
|
{
|
|
"epoch": 2.843109378582894,
|
|
"grad_norm": 2.8720295429229736,
|
|
"learning_rate": 7.911455545066658e-06,
|
|
"loss": 2.1192,
|
|
"num_tokens": 951405046.0,
|
|
"step": 15500
|
|
},
|
|
{
|
|
"epoch": 2.85228158679202,
|
|
"grad_norm": 2.9304111003875732,
|
|
"learning_rate": 7.854931322214525e-06,
|
|
"loss": 2.1237,
|
|
"num_tokens": 954478327.0,
|
|
"step": 15550
|
|
},
|
|
{
|
|
"epoch": 2.8614537950011467,
|
|
"grad_norm": 2.6521966457366943,
|
|
"learning_rate": 7.798478849082902e-06,
|
|
"loss": 2.122,
|
|
"num_tokens": 957538224.0,
|
|
"step": 15600
|
|
},
|
|
{
|
|
"epoch": 2.8706260032102726,
|
|
"grad_norm": 2.3696980476379395,
|
|
"learning_rate": 7.742100013932618e-06,
|
|
"loss": 2.1414,
|
|
"num_tokens": 960610742.0,
|
|
"step": 15650
|
|
},
|
|
{
|
|
"epoch": 2.8797982114193994,
|
|
"grad_norm": 2.6325161457061768,
|
|
"learning_rate": 7.685796702561416e-06,
|
|
"loss": 2.1172,
|
|
"num_tokens": 963681234.0,
|
|
"step": 15700
|
|
},
|
|
{
|
|
"epoch": 2.8889704196285253,
|
|
"grad_norm": 4.786980152130127,
|
|
"learning_rate": 7.629570798240861e-06,
|
|
"loss": 2.1083,
|
|
"num_tokens": 966763405.0,
|
|
"step": 15750
|
|
},
|
|
{
|
|
"epoch": 2.898142627837652,
|
|
"grad_norm": 2.6920909881591797,
|
|
"learning_rate": 7.573424181653363e-06,
|
|
"loss": 2.1262,
|
|
"num_tokens": 969828117.0,
|
|
"step": 15800
|
|
},
|
|
{
|
|
"epoch": 2.9073148360467784,
|
|
"grad_norm": 2.511882781982422,
|
|
"learning_rate": 7.51735873082925e-06,
|
|
"loss": 2.1066,
|
|
"num_tokens": 972905740.0,
|
|
"step": 15850
|
|
},
|
|
{
|
|
"epoch": 2.9164870442559048,
|
|
"grad_norm": 2.9187281131744385,
|
|
"learning_rate": 7.461376321083975e-06,
|
|
"loss": 2.1116,
|
|
"num_tokens": 975971572.0,
|
|
"step": 15900
|
|
},
|
|
{
|
|
"epoch": 2.925659252465031,
|
|
"grad_norm": 3.1962969303131104,
|
|
"learning_rate": 7.40547882495537e-06,
|
|
"loss": 2.1431,
|
|
"num_tokens": 979041359.0,
|
|
"step": 15950
|
|
},
|
|
{
|
|
"epoch": 2.9348314606741575,
|
|
"grad_norm": 2.8684613704681396,
|
|
"learning_rate": 7.34966811214102e-06,
|
|
"loss": 2.1283,
|
|
"num_tokens": 982097803.0,
|
|
"step": 16000
|
|
},
|
|
{
|
|
"epoch": 2.944003668883284,
|
|
"grad_norm": 3.554492950439453,
|
|
"learning_rate": 7.293946049435718e-06,
|
|
"loss": 2.1225,
|
|
"num_tokens": 985160591.0,
|
|
"step": 16050
|
|
},
|
|
{
|
|
"epoch": 2.95317587709241,
|
|
"grad_norm": 2.750422239303589,
|
|
"learning_rate": 7.238314500669033e-06,
|
|
"loss": 2.1216,
|
|
"num_tokens": 988236506.0,
|
|
"step": 16100
|
|
},
|
|
{
|
|
"epoch": 2.9623480853015365,
|
|
"grad_norm": 3.027853012084961,
|
|
"learning_rate": 7.182775326642961e-06,
|
|
"loss": 2.1338,
|
|
"num_tokens": 991298224.0,
|
|
"step": 16150
|
|
},
|
|
{
|
|
"epoch": 2.971520293510663,
|
|
"grad_norm": 2.8112590312957764,
|
|
"learning_rate": 7.127330385069682e-06,
|
|
"loss": 2.1197,
|
|
"num_tokens": 994361927.0,
|
|
"step": 16200
|
|
},
|
|
{
|
|
"epoch": 2.980692501719789,
|
|
"grad_norm": 2.5668013095855713,
|
|
"learning_rate": 7.071981530509424e-06,
|
|
"loss": 2.0874,
|
|
"num_tokens": 997437861.0,
|
|
"step": 16250
|
|
},
|
|
{
|
|
"epoch": 2.9898647099289155,
|
|
"grad_norm": 2.909646987915039,
|
|
"learning_rate": 7.01673061430844e-06,
|
|
"loss": 2.1306,
|
|
"num_tokens": 1000499652.0,
|
|
"step": 16300
|
|
},
|
|
{
|
|
"epoch": 2.999036918138042,
|
|
"grad_norm": 2.9382026195526123,
|
|
"learning_rate": 6.961579484537065e-06,
|
|
"loss": 2.1142,
|
|
"num_tokens": 1003569038.0,
|
|
"step": 16350
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 2.342665433883667,
|
|
"eval_num_tokens": 1003891312.0,
|
|
"eval_runtime": 5.3817,
|
|
"eval_samples_per_second": 90.12,
|
|
"eval_steps_per_second": 5.76,
|
|
"step": 16356
|
|
},
|
|
{
|
|
"epoch": 3.0080715432240313,
|
|
"grad_norm": 2.643911838531494,
|
|
"learning_rate": 6.906529985927908e-06,
|
|
"loss": 2.0771,
|
|
"num_tokens": 1006580635.0,
|
|
"step": 16400
|
|
},
|
|
{
|
|
"epoch": 3.0172437514331576,
|
|
"grad_norm": 2.9734792709350586,
|
|
"learning_rate": 6.8515839598141584e-06,
|
|
"loss": 2.1001,
|
|
"num_tokens": 1009650049.0,
|
|
"step": 16450
|
|
},
|
|
{
|
|
"epoch": 3.026415959642284,
|
|
"grad_norm": 2.1990256309509277,
|
|
"learning_rate": 6.796743244067978e-06,
|
|
"loss": 2.0876,
|
|
"num_tokens": 1012719707.0,
|
|
"step": 16500
|
|
},
|
|
{
|
|
"epoch": 3.0355881678514103,
|
|
"grad_norm": 3.03206467628479,
|
|
"learning_rate": 6.742009673039043e-06,
|
|
"loss": 2.1055,
|
|
"num_tokens": 1015792221.0,
|
|
"step": 16550
|
|
},
|
|
{
|
|
"epoch": 3.0447603760605366,
|
|
"grad_norm": 2.538322687149048,
|
|
"learning_rate": 6.687385077493174e-06,
|
|
"loss": 2.1058,
|
|
"num_tokens": 1018852853.0,
|
|
"step": 16600
|
|
},
|
|
{
|
|
"epoch": 3.053932584269663,
|
|
"grad_norm": 2.883906841278076,
|
|
"learning_rate": 6.63287128455111e-06,
|
|
"loss": 2.1025,
|
|
"num_tokens": 1021929777.0,
|
|
"step": 16650
|
|
},
|
|
{
|
|
"epoch": 3.0631047924787893,
|
|
"grad_norm": 2.341115951538086,
|
|
"learning_rate": 6.578470117627383e-06,
|
|
"loss": 2.0897,
|
|
"num_tokens": 1024996761.0,
|
|
"step": 16700
|
|
},
|
|
{
|
|
"epoch": 3.0722770006879156,
|
|
"grad_norm": 2.923607110977173,
|
|
"learning_rate": 6.5241833963693416e-06,
|
|
"loss": 2.1083,
|
|
"num_tokens": 1028072129.0,
|
|
"step": 16750
|
|
},
|
|
{
|
|
"epoch": 3.081449208897042,
|
|
"grad_norm": 2.836818218231201,
|
|
"learning_rate": 6.4700129365962725e-06,
|
|
"loss": 2.1047,
|
|
"num_tokens": 1031140847.0,
|
|
"step": 16800
|
|
},
|
|
{
|
|
"epoch": 3.0906214171061683,
|
|
"grad_norm": 2.681126594543457,
|
|
"learning_rate": 6.415960550238667e-06,
|
|
"loss": 2.1164,
|
|
"num_tokens": 1034208969.0,
|
|
"step": 16850
|
|
},
|
|
{
|
|
"epoch": 3.0997936253152947,
|
|
"grad_norm": 2.6722464561462402,
|
|
"learning_rate": 6.362028045277617e-06,
|
|
"loss": 2.0995,
|
|
"num_tokens": 1037278379.0,
|
|
"step": 16900
|
|
},
|
|
{
|
|
"epoch": 3.108965833524421,
|
|
"grad_norm": 2.5728116035461426,
|
|
"learning_rate": 6.308217225684348e-06,
|
|
"loss": 2.0881,
|
|
"num_tokens": 1040343794.0,
|
|
"step": 16950
|
|
},
|
|
{
|
|
"epoch": 3.1181380417335474,
|
|
"grad_norm": 2.4531943798065186,
|
|
"learning_rate": 6.2545298913598595e-06,
|
|
"loss": 2.1073,
|
|
"num_tokens": 1043403537.0,
|
|
"step": 17000
|
|
},
|
|
{
|
|
"epoch": 3.1273102499426737,
|
|
"grad_norm": 2.8585636615753174,
|
|
"learning_rate": 6.2009678380747354e-06,
|
|
"loss": 2.078,
|
|
"num_tokens": 1046475339.0,
|
|
"step": 17050
|
|
},
|
|
{
|
|
"epoch": 3.1364824581518,
|
|
"grad_norm": 2.4157869815826416,
|
|
"learning_rate": 6.147532857409075e-06,
|
|
"loss": 2.1005,
|
|
"num_tokens": 1049537238.0,
|
|
"step": 17100
|
|
},
|
|
{
|
|
"epoch": 3.1456546663609264,
|
|
"grad_norm": 2.7901792526245117,
|
|
"learning_rate": 6.094226736692561e-06,
|
|
"loss": 2.0975,
|
|
"num_tokens": 1052609023.0,
|
|
"step": 17150
|
|
},
|
|
{
|
|
"epoch": 3.1548268745700527,
|
|
"grad_norm": 2.8096275329589844,
|
|
"learning_rate": 6.041051258944692e-06,
|
|
"loss": 2.0914,
|
|
"num_tokens": 1055666197.0,
|
|
"step": 17200
|
|
},
|
|
{
|
|
"epoch": 3.163999082779179,
|
|
"grad_norm": 2.6654086112976074,
|
|
"learning_rate": 5.988008202815121e-06,
|
|
"loss": 2.0932,
|
|
"num_tokens": 1058734997.0,
|
|
"step": 17250
|
|
},
|
|
{
|
|
"epoch": 3.1731712909883054,
|
|
"grad_norm": 2.52492618560791,
|
|
"learning_rate": 5.935099342524175e-06,
|
|
"loss": 2.0951,
|
|
"num_tokens": 1061803902.0,
|
|
"step": 17300
|
|
},
|
|
{
|
|
"epoch": 3.1823434991974318,
|
|
"grad_norm": 2.548250913619995,
|
|
"learning_rate": 5.882326447803508e-06,
|
|
"loss": 2.1118,
|
|
"num_tokens": 1064870870.0,
|
|
"step": 17350
|
|
},
|
|
{
|
|
"epoch": 3.191515707406558,
|
|
"grad_norm": 2.477153778076172,
|
|
"learning_rate": 5.829691283836902e-06,
|
|
"loss": 2.0961,
|
|
"num_tokens": 1067932406.0,
|
|
"step": 17400
|
|
},
|
|
{
|
|
"epoch": 3.2006879156156844,
|
|
"grad_norm": 2.7208263874053955,
|
|
"learning_rate": 5.777195611201234e-06,
|
|
"loss": 2.0881,
|
|
"num_tokens": 1070985556.0,
|
|
"step": 17450
|
|
},
|
|
{
|
|
"epoch": 3.209860123824811,
|
|
"grad_norm": 2.7050225734710693,
|
|
"learning_rate": 5.72484118580757e-06,
|
|
"loss": 2.1023,
|
|
"num_tokens": 1074068510.0,
|
|
"step": 17500
|
|
},
|
|
{
|
|
"epoch": 3.219032332033937,
|
|
"grad_norm": 2.5750534534454346,
|
|
"learning_rate": 5.672629758842449e-06,
|
|
"loss": 2.082,
|
|
"num_tokens": 1077138470.0,
|
|
"step": 17550
|
|
},
|
|
{
|
|
"epoch": 3.2282045402430635,
|
|
"grad_norm": 2.9645302295684814,
|
|
"learning_rate": 5.6205630767093e-06,
|
|
"loss": 2.0903,
|
|
"num_tokens": 1080214390.0,
|
|
"step": 17600
|
|
},
|
|
{
|
|
"epoch": 3.23737674845219,
|
|
"grad_norm": 4.345239639282227,
|
|
"learning_rate": 5.568642880970026e-06,
|
|
"loss": 2.1013,
|
|
"num_tokens": 1083282043.0,
|
|
"step": 17650
|
|
},
|
|
{
|
|
"epoch": 3.246548956661316,
|
|
"grad_norm": 3.1888034343719482,
|
|
"learning_rate": 5.516870908286759e-06,
|
|
"loss": 2.1227,
|
|
"num_tokens": 1086349075.0,
|
|
"step": 17700
|
|
},
|
|
{
|
|
"epoch": 3.2557211648704425,
|
|
"grad_norm": 2.669917106628418,
|
|
"learning_rate": 5.465248890363752e-06,
|
|
"loss": 2.1028,
|
|
"num_tokens": 1089424002.0,
|
|
"step": 17750
|
|
},
|
|
{
|
|
"epoch": 3.264893373079569,
|
|
"grad_norm": 3.3849666118621826,
|
|
"learning_rate": 5.413778553889484e-06,
|
|
"loss": 2.097,
|
|
"num_tokens": 1092481733.0,
|
|
"step": 17800
|
|
},
|
|
{
|
|
"epoch": 3.274065581288695,
|
|
"grad_norm": 2.4466021060943604,
|
|
"learning_rate": 5.362461620478876e-06,
|
|
"loss": 2.0896,
|
|
"num_tokens": 1095559998.0,
|
|
"step": 17850
|
|
},
|
|
{
|
|
"epoch": 3.2832377894978215,
|
|
"grad_norm": 2.5363235473632812,
|
|
"learning_rate": 5.311299806615734e-06,
|
|
"loss": 2.08,
|
|
"num_tokens": 1098626669.0,
|
|
"step": 17900
|
|
},
|
|
{
|
|
"epoch": 3.292409997706948,
|
|
"grad_norm": 2.906054973602295,
|
|
"learning_rate": 5.260294823595301e-06,
|
|
"loss": 2.1055,
|
|
"num_tokens": 1101687719.0,
|
|
"step": 17950
|
|
},
|
|
{
|
|
"epoch": 3.301582205916074,
|
|
"grad_norm": 2.772920846939087,
|
|
"learning_rate": 5.20944837746705e-06,
|
|
"loss": 2.0877,
|
|
"num_tokens": 1104758931.0,
|
|
"step": 18000
|
|
},
|
|
{
|
|
"epoch": 3.3107544141252006,
|
|
"grad_norm": 2.481189489364624,
|
|
"learning_rate": 5.158762168977603e-06,
|
|
"loss": 2.0931,
|
|
"num_tokens": 1107829262.0,
|
|
"step": 18050
|
|
},
|
|
{
|
|
"epoch": 3.319926622334327,
|
|
"grad_norm": 2.398188352584839,
|
|
"learning_rate": 5.10823789351383e-06,
|
|
"loss": 2.0938,
|
|
"num_tokens": 1110906941.0,
|
|
"step": 18100
|
|
},
|
|
{
|
|
"epoch": 3.3290988305434532,
|
|
"grad_norm": 2.560267448425293,
|
|
"learning_rate": 5.057877241046173e-06,
|
|
"loss": 2.1287,
|
|
"num_tokens": 1113970900.0,
|
|
"step": 18150
|
|
},
|
|
{
|
|
"epoch": 3.3382710387525796,
|
|
"grad_norm": 2.5596182346343994,
|
|
"learning_rate": 5.007681896072085e-06,
|
|
"loss": 2.1011,
|
|
"num_tokens": 1117040694.0,
|
|
"step": 18200
|
|
},
|
|
{
|
|
"epoch": 3.347443246961706,
|
|
"grad_norm": 2.5481984615325928,
|
|
"learning_rate": 4.957653537559714e-06,
|
|
"loss": 2.1034,
|
|
"num_tokens": 1120115063.0,
|
|
"step": 18250
|
|
},
|
|
{
|
|
"epoch": 3.3566154551708323,
|
|
"grad_norm": 2.254716157913208,
|
|
"learning_rate": 4.907793838891725e-06,
|
|
"loss": 2.0861,
|
|
"num_tokens": 1123183440.0,
|
|
"step": 18300
|
|
},
|
|
{
|
|
"epoch": 3.3657876633799586,
|
|
"grad_norm": 2.2420766353607178,
|
|
"learning_rate": 4.858104467809323e-06,
|
|
"loss": 2.0859,
|
|
"num_tokens": 1126252549.0,
|
|
"step": 18350
|
|
},
|
|
{
|
|
"epoch": 3.374959871589085,
|
|
"grad_norm": 2.909041404724121,
|
|
"learning_rate": 4.808587086356507e-06,
|
|
"loss": 2.0851,
|
|
"num_tokens": 1129323810.0,
|
|
"step": 18400
|
|
},
|
|
{
|
|
"epoch": 3.3841320797982113,
|
|
"grad_norm": 3.050262689590454,
|
|
"learning_rate": 4.759243350824421e-06,
|
|
"loss": 2.0907,
|
|
"num_tokens": 1132392455.0,
|
|
"step": 18450
|
|
},
|
|
{
|
|
"epoch": 3.3933042880073376,
|
|
"grad_norm": 3.359714984893799,
|
|
"learning_rate": 4.710074911695999e-06,
|
|
"loss": 2.1075,
|
|
"num_tokens": 1135464524.0,
|
|
"step": 18500
|
|
},
|
|
{
|
|
"epoch": 3.402476496216464,
|
|
"grad_norm": 2.33805251121521,
|
|
"learning_rate": 4.66108341359073e-06,
|
|
"loss": 2.0909,
|
|
"num_tokens": 1138522853.0,
|
|
"step": 18550
|
|
},
|
|
{
|
|
"epoch": 3.4116487044255903,
|
|
"grad_norm": 3.151082754135132,
|
|
"learning_rate": 4.612270495209669e-06,
|
|
"loss": 2.0941,
|
|
"num_tokens": 1141590183.0,
|
|
"step": 18600
|
|
},
|
|
{
|
|
"epoch": 3.4208209126347167,
|
|
"grad_norm": 2.5991766452789307,
|
|
"learning_rate": 4.563637789280614e-06,
|
|
"loss": 2.0912,
|
|
"num_tokens": 1144660646.0,
|
|
"step": 18650
|
|
},
|
|
{
|
|
"epoch": 3.429993120843843,
|
|
"grad_norm": 2.696897268295288,
|
|
"learning_rate": 4.515186922503481e-06,
|
|
"loss": 2.099,
|
|
"num_tokens": 1147731684.0,
|
|
"step": 18700
|
|
},
|
|
{
|
|
"epoch": 3.4391653290529693,
|
|
"grad_norm": 3.167006492614746,
|
|
"learning_rate": 4.466919515495924e-06,
|
|
"loss": 2.0855,
|
|
"num_tokens": 1150798166.0,
|
|
"step": 18750
|
|
},
|
|
{
|
|
"epoch": 3.4483375372620957,
|
|
"grad_norm": 2.953187942504883,
|
|
"learning_rate": 4.418837182739094e-06,
|
|
"loss": 2.0956,
|
|
"num_tokens": 1153876907.0,
|
|
"step": 18800
|
|
},
|
|
{
|
|
"epoch": 3.457509745471222,
|
|
"grad_norm": 2.6491310596466064,
|
|
"learning_rate": 4.3709415325236614e-06,
|
|
"loss": 2.1077,
|
|
"num_tokens": 1156946926.0,
|
|
"step": 18850
|
|
},
|
|
{
|
|
"epoch": 3.4666819536803484,
|
|
"grad_norm": 2.720600128173828,
|
|
"learning_rate": 4.323234166896014e-06,
|
|
"loss": 2.0993,
|
|
"num_tokens": 1160022802.0,
|
|
"step": 18900
|
|
},
|
|
{
|
|
"epoch": 3.4758541618894747,
|
|
"grad_norm": 2.6700637340545654,
|
|
"learning_rate": 4.275716681604655e-06,
|
|
"loss": 2.09,
|
|
"num_tokens": 1163086712.0,
|
|
"step": 18950
|
|
},
|
|
{
|
|
"epoch": 3.485026370098601,
|
|
"grad_norm": 2.5062873363494873,
|
|
"learning_rate": 4.228390666046857e-06,
|
|
"loss": 2.0725,
|
|
"num_tokens": 1166162929.0,
|
|
"step": 19000
|
|
},
|
|
{
|
|
"epoch": 3.4941985783077274,
|
|
"grad_norm": 2.452350378036499,
|
|
"learning_rate": 4.181257703215467e-06,
|
|
"loss": 2.0928,
|
|
"num_tokens": 1169227751.0,
|
|
"step": 19050
|
|
},
|
|
{
|
|
"epoch": 3.5033707865168537,
|
|
"grad_norm": 2.891571521759033,
|
|
"learning_rate": 4.134319369645986e-06,
|
|
"loss": 2.0793,
|
|
"num_tokens": 1172298354.0,
|
|
"step": 19100
|
|
},
|
|
{
|
|
"epoch": 3.5125429947259805,
|
|
"grad_norm": 2.5535755157470703,
|
|
"learning_rate": 4.087577235363812e-06,
|
|
"loss": 2.0915,
|
|
"num_tokens": 1175366979.0,
|
|
"step": 19150
|
|
},
|
|
{
|
|
"epoch": 3.5217152029351064,
|
|
"grad_norm": 3.2136306762695312,
|
|
"learning_rate": 4.041032863831742e-06,
|
|
"loss": 2.1079,
|
|
"num_tokens": 1178431092.0,
|
|
"step": 19200
|
|
},
|
|
{
|
|
"epoch": 3.530887411144233,
|
|
"grad_norm": 2.785170316696167,
|
|
"learning_rate": 3.994687811897668e-06,
|
|
"loss": 2.1137,
|
|
"num_tokens": 1181490876.0,
|
|
"step": 19250
|
|
},
|
|
{
|
|
"epoch": 3.540059619353359,
|
|
"grad_norm": 3.704871416091919,
|
|
"learning_rate": 3.948543629742504e-06,
|
|
"loss": 2.0724,
|
|
"num_tokens": 1184556615.0,
|
|
"step": 19300
|
|
},
|
|
{
|
|
"epoch": 3.549231827562486,
|
|
"grad_norm": 2.566716432571411,
|
|
"learning_rate": 3.9026018608283365e-06,
|
|
"loss": 2.0915,
|
|
"num_tokens": 1187624909.0,
|
|
"step": 19350
|
|
},
|
|
{
|
|
"epoch": 3.558404035771612,
|
|
"grad_norm": 2.6187422275543213,
|
|
"learning_rate": 3.85686404184679e-06,
|
|
"loss": 2.0903,
|
|
"num_tokens": 1190686253.0,
|
|
"step": 19400
|
|
},
|
|
{
|
|
"epoch": 3.5675762439807386,
|
|
"grad_norm": 3.740200996398926,
|
|
"learning_rate": 3.81133170266764e-06,
|
|
"loss": 2.1008,
|
|
"num_tokens": 1193763731.0,
|
|
"step": 19450
|
|
},
|
|
{
|
|
"epoch": 3.5767484521898645,
|
|
"grad_norm": 2.2109391689300537,
|
|
"learning_rate": 3.7660063662876245e-06,
|
|
"loss": 2.0784,
|
|
"num_tokens": 1196837953.0,
|
|
"step": 19500
|
|
},
|
|
{
|
|
"epoch": 3.5859206603989913,
|
|
"grad_norm": 2.734866142272949,
|
|
"learning_rate": 3.720889548779518e-06,
|
|
"loss": 2.0897,
|
|
"num_tokens": 1199906139.0,
|
|
"step": 19550
|
|
},
|
|
{
|
|
"epoch": 3.595092868608117,
|
|
"grad_norm": 2.3316569328308105,
|
|
"learning_rate": 3.6759827592414165e-06,
|
|
"loss": 2.091,
|
|
"num_tokens": 1202967461.0,
|
|
"step": 19600
|
|
},
|
|
{
|
|
"epoch": 3.604265076817244,
|
|
"grad_norm": 2.5883240699768066,
|
|
"learning_rate": 3.6312874997462444e-06,
|
|
"loss": 2.082,
|
|
"num_tokens": 1206018711.0,
|
|
"step": 19650
|
|
},
|
|
{
|
|
"epoch": 3.61343728502637,
|
|
"grad_norm": 2.4006330966949463,
|
|
"learning_rate": 3.5868052652915386e-06,
|
|
"loss": 2.0942,
|
|
"num_tokens": 1209091227.0,
|
|
"step": 19700
|
|
},
|
|
{
|
|
"epoch": 3.6226094932354966,
|
|
"grad_norm": 2.8134772777557373,
|
|
"learning_rate": 3.542537543749417e-06,
|
|
"loss": 2.086,
|
|
"num_tokens": 1212157315.0,
|
|
"step": 19750
|
|
},
|
|
{
|
|
"epoch": 3.6317817014446225,
|
|
"grad_norm": 3.291865348815918,
|
|
"learning_rate": 3.4984858158168324e-06,
|
|
"loss": 2.0811,
|
|
"num_tokens": 1215232699.0,
|
|
"step": 19800
|
|
},
|
|
{
|
|
"epoch": 3.6409539096537493,
|
|
"grad_norm": 2.5171632766723633,
|
|
"learning_rate": 3.4546515549660376e-06,
|
|
"loss": 2.0882,
|
|
"num_tokens": 1218294413.0,
|
|
"step": 19850
|
|
},
|
|
{
|
|
"epoch": 3.650126117862875,
|
|
"grad_norm": 2.6282389163970947,
|
|
"learning_rate": 3.4110362273952834e-06,
|
|
"loss": 2.0766,
|
|
"num_tokens": 1221363586.0,
|
|
"step": 19900
|
|
},
|
|
{
|
|
"epoch": 3.659298326072002,
|
|
"grad_norm": 3.0376970767974854,
|
|
"learning_rate": 3.367641291979812e-06,
|
|
"loss": 2.0809,
|
|
"num_tokens": 1224437413.0,
|
|
"step": 19950
|
|
},
|
|
{
|
|
"epoch": 3.6684705342811283,
|
|
"grad_norm": 3.1428987979888916,
|
|
"learning_rate": 3.3244682002230165e-06,
|
|
"loss": 2.0775,
|
|
"num_tokens": 1227505865.0,
|
|
"step": 20000
|
|
},
|
|
{
|
|
"epoch": 3.6776427424902547,
|
|
"grad_norm": 2.4935078620910645,
|
|
"learning_rate": 3.281518396207929e-06,
|
|
"loss": 2.0862,
|
|
"num_tokens": 1230574609.0,
|
|
"step": 20050
|
|
},
|
|
{
|
|
"epoch": 3.686814950699381,
|
|
"grad_norm": 2.975177526473999,
|
|
"learning_rate": 3.238793316548894e-06,
|
|
"loss": 2.0759,
|
|
"num_tokens": 1233650321.0,
|
|
"step": 20100
|
|
},
|
|
{
|
|
"epoch": 3.6959871589085074,
|
|
"grad_norm": 2.8056142330169678,
|
|
"learning_rate": 3.196294390343522e-06,
|
|
"loss": 2.0691,
|
|
"num_tokens": 1236726919.0,
|
|
"step": 20150
|
|
},
|
|
{
|
|
"epoch": 3.7051593671176337,
|
|
"grad_norm": 2.7456674575805664,
|
|
"learning_rate": 3.1540230391248938e-06,
|
|
"loss": 2.0852,
|
|
"num_tokens": 1239804275.0,
|
|
"step": 20200
|
|
},
|
|
{
|
|
"epoch": 3.71433157532676,
|
|
"grad_norm": 2.4337079524993896,
|
|
"learning_rate": 3.111980676813996e-06,
|
|
"loss": 2.0787,
|
|
"num_tokens": 1242867601.0,
|
|
"step": 20250
|
|
},
|
|
{
|
|
"epoch": 3.7235037835358864,
|
|
"grad_norm": 3.432335615158081,
|
|
"learning_rate": 3.070168709672452e-06,
|
|
"loss": 2.109,
|
|
"num_tokens": 1245937215.0,
|
|
"step": 20300
|
|
},
|
|
{
|
|
"epoch": 3.7326759917450127,
|
|
"grad_norm": 3.535491704940796,
|
|
"learning_rate": 3.0285885362554546e-06,
|
|
"loss": 2.076,
|
|
"num_tokens": 1249009885.0,
|
|
"step": 20350
|
|
},
|
|
{
|
|
"epoch": 3.741848199954139,
|
|
"grad_norm": 2.634734869003296,
|
|
"learning_rate": 2.9872415473650174e-06,
|
|
"loss": 2.0918,
|
|
"num_tokens": 1252083496.0,
|
|
"step": 20400
|
|
},
|
|
{
|
|
"epoch": 3.7510204081632654,
|
|
"grad_norm": 2.6964564323425293,
|
|
"learning_rate": 2.9461291260034263e-06,
|
|
"loss": 2.0968,
|
|
"num_tokens": 1255148836.0,
|
|
"step": 20450
|
|
},
|
|
{
|
|
"epoch": 3.7601926163723918,
|
|
"grad_norm": 3.015977382659912,
|
|
"learning_rate": 2.9052526473270015e-06,
|
|
"loss": 2.0866,
|
|
"num_tokens": 1258221193.0,
|
|
"step": 20500
|
|
},
|
|
{
|
|
"epoch": 3.769364824581518,
|
|
"grad_norm": 2.6833744049072266,
|
|
"learning_rate": 2.8646134786000903e-06,
|
|
"loss": 2.0845,
|
|
"num_tokens": 1261291526.0,
|
|
"step": 20550
|
|
},
|
|
{
|
|
"epoch": 3.7785370327906445,
|
|
"grad_norm": 2.608642101287842,
|
|
"learning_rate": 2.8242129791493245e-06,
|
|
"loss": 2.0594,
|
|
"num_tokens": 1264356594.0,
|
|
"step": 20600
|
|
},
|
|
{
|
|
"epoch": 3.787709240999771,
|
|
"grad_norm": 2.5900356769561768,
|
|
"learning_rate": 2.7840525003181805e-06,
|
|
"loss": 2.0807,
|
|
"num_tokens": 1267430188.0,
|
|
"step": 20650
|
|
},
|
|
{
|
|
"epoch": 3.796881449208897,
|
|
"grad_norm": 2.877882242202759,
|
|
"learning_rate": 2.744133385421743e-06,
|
|
"loss": 2.0751,
|
|
"num_tokens": 1270498926.0,
|
|
"step": 20700
|
|
},
|
|
{
|
|
"epoch": 3.8060536574180235,
|
|
"grad_norm": 3.001420259475708,
|
|
"learning_rate": 2.7044569697018043e-06,
|
|
"loss": 2.0857,
|
|
"num_tokens": 1273570408.0,
|
|
"step": 20750
|
|
},
|
|
{
|
|
"epoch": 3.81522586562715,
|
|
"grad_norm": 2.6310830116271973,
|
|
"learning_rate": 2.6650245802821883e-06,
|
|
"loss": 2.0902,
|
|
"num_tokens": 1276642637.0,
|
|
"step": 20800
|
|
},
|
|
{
|
|
"epoch": 3.824398073836276,
|
|
"grad_norm": 2.6124022006988525,
|
|
"learning_rate": 2.6258375361243505e-06,
|
|
"loss": 2.0777,
|
|
"num_tokens": 1279715797.0,
|
|
"step": 20850
|
|
},
|
|
{
|
|
"epoch": 3.8335702820454025,
|
|
"grad_norm": 2.7061257362365723,
|
|
"learning_rate": 2.5868971479832796e-06,
|
|
"loss": 2.1073,
|
|
"num_tokens": 1282783417.0,
|
|
"step": 20900
|
|
},
|
|
{
|
|
"epoch": 3.842742490254529,
|
|
"grad_norm": 2.277985095977783,
|
|
"learning_rate": 2.5482047183636448e-06,
|
|
"loss": 2.0848,
|
|
"num_tokens": 1285850165.0,
|
|
"step": 20950
|
|
},
|
|
{
|
|
"epoch": 3.851914698463655,
|
|
"grad_norm": 2.4407973289489746,
|
|
"learning_rate": 2.509761541476231e-06,
|
|
"loss": 2.0905,
|
|
"num_tokens": 1288930804.0,
|
|
"step": 21000
|
|
},
|
|
{
|
|
"epoch": 3.8610869066727815,
|
|
"grad_norm": 2.4228036403656006,
|
|
"learning_rate": 2.4715689031946377e-06,
|
|
"loss": 2.0792,
|
|
"num_tokens": 1292002479.0,
|
|
"step": 21050
|
|
},
|
|
{
|
|
"epoch": 3.870259114881908,
|
|
"grad_norm": 2.5259807109832764,
|
|
"learning_rate": 2.433628081012289e-06,
|
|
"loss": 2.0891,
|
|
"num_tokens": 1295068549.0,
|
|
"step": 21100
|
|
},
|
|
{
|
|
"epoch": 3.8794313230910342,
|
|
"grad_norm": 2.8560571670532227,
|
|
"learning_rate": 2.395940343999691e-06,
|
|
"loss": 2.0894,
|
|
"num_tokens": 1298144870.0,
|
|
"step": 21150
|
|
},
|
|
{
|
|
"epoch": 3.8886035313001606,
|
|
"grad_norm": 2.9525187015533447,
|
|
"learning_rate": 2.3585069527619775e-06,
|
|
"loss": 2.0933,
|
|
"num_tokens": 1301221403.0,
|
|
"step": 21200
|
|
},
|
|
{
|
|
"epoch": 3.897775739509287,
|
|
"grad_norm": 3.031400680541992,
|
|
"learning_rate": 2.32132915939676e-06,
|
|
"loss": 2.0627,
|
|
"num_tokens": 1304299358.0,
|
|
"step": 21250
|
|
},
|
|
{
|
|
"epoch": 3.9069479477184132,
|
|
"grad_norm": 3.7090675830841064,
|
|
"learning_rate": 2.284408207452229e-06,
|
|
"loss": 2.0806,
|
|
"num_tokens": 1307356084.0,
|
|
"step": 21300
|
|
},
|
|
{
|
|
"epoch": 3.9161201559275396,
|
|
"grad_norm": 2.952364921569824,
|
|
"learning_rate": 2.247745331885576e-06,
|
|
"loss": 2.0811,
|
|
"num_tokens": 1310430987.0,
|
|
"step": 21350
|
|
},
|
|
{
|
|
"epoch": 3.925292364136666,
|
|
"grad_norm": 2.5101613998413086,
|
|
"learning_rate": 2.2113417590216746e-06,
|
|
"loss": 2.1017,
|
|
"num_tokens": 1313508657.0,
|
|
"step": 21400
|
|
},
|
|
{
|
|
"epoch": 3.9344645723457923,
|
|
"grad_norm": 2.5011069774627686,
|
|
"learning_rate": 2.175198706512063e-06,
|
|
"loss": 2.0741,
|
|
"num_tokens": 1316571388.0,
|
|
"step": 21450
|
|
},
|
|
{
|
|
"epoch": 3.9436367805549186,
|
|
"grad_norm": 2.638942003250122,
|
|
"learning_rate": 2.1393173832942203e-06,
|
|
"loss": 2.0839,
|
|
"num_tokens": 1319658216.0,
|
|
"step": 21500
|
|
},
|
|
{
|
|
"epoch": 3.952808988764045,
|
|
"grad_norm": 2.4227914810180664,
|
|
"learning_rate": 2.103698989551122e-06,
|
|
"loss": 2.0774,
|
|
"num_tokens": 1322733413.0,
|
|
"step": 21550
|
|
},
|
|
{
|
|
"epoch": 3.9619811969731713,
|
|
"grad_norm": 2.9436216354370117,
|
|
"learning_rate": 2.0683447166711023e-06,
|
|
"loss": 2.0878,
|
|
"num_tokens": 1325805455.0,
|
|
"step": 21600
|
|
},
|
|
{
|
|
"epoch": 3.9711534051822976,
|
|
"grad_norm": 2.806403160095215,
|
|
"learning_rate": 2.033255747207997e-06,
|
|
"loss": 2.0781,
|
|
"num_tokens": 1328875872.0,
|
|
"step": 21650
|
|
},
|
|
{
|
|
"epoch": 3.980325613391424,
|
|
"grad_norm": 2.7450454235076904,
|
|
"learning_rate": 1.998433254841596e-06,
|
|
"loss": 2.0654,
|
|
"num_tokens": 1331941040.0,
|
|
"step": 21700
|
|
},
|
|
{
|
|
"epoch": 3.9894978216005503,
|
|
"grad_norm": 3.459977626800537,
|
|
"learning_rate": 1.9638784043383807e-06,
|
|
"loss": 2.0883,
|
|
"num_tokens": 1335001783.0,
|
|
"step": 21750
|
|
},
|
|
{
|
|
"epoch": 3.9986700298096767,
|
|
"grad_norm": 2.5629732608795166,
|
|
"learning_rate": 1.9295923515125625e-06,
|
|
"loss": 2.0792,
|
|
"num_tokens": 1338081325.0,
|
|
"step": 21800
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"eval_loss": 2.3043127059936523,
|
|
"eval_num_tokens": 1338522099.0,
|
|
"eval_runtime": 5.3826,
|
|
"eval_samples_per_second": 90.105,
|
|
"eval_steps_per_second": 5.759,
|
|
"step": 21808
|
|
},
|
|
{
|
|
"epoch": 4.007704654895666,
|
|
"grad_norm": 2.3756043910980225,
|
|
"learning_rate": 1.8955762431874315e-06,
|
|
"loss": 2.0661,
|
|
"num_tokens": 1341101813.0,
|
|
"step": 21850
|
|
},
|
|
{
|
|
"epoch": 4.016876863104793,
|
|
"grad_norm": 2.3425350189208984,
|
|
"learning_rate": 1.8618312171569774e-06,
|
|
"loss": 2.069,
|
|
"num_tokens": 1344174255.0,
|
|
"step": 21900
|
|
},
|
|
{
|
|
"epoch": 4.026049071313919,
|
|
"grad_norm": 2.5926780700683594,
|
|
"learning_rate": 1.8283584021478585e-06,
|
|
"loss": 2.0834,
|
|
"num_tokens": 1347245490.0,
|
|
"step": 21950
|
|
},
|
|
{
|
|
"epoch": 4.0352212795230455,
|
|
"grad_norm": 2.451519012451172,
|
|
"learning_rate": 1.7951589177816197e-06,
|
|
"loss": 2.0688,
|
|
"num_tokens": 1350320208.0,
|
|
"step": 22000
|
|
},
|
|
{
|
|
"epoch": 4.044393487732171,
|
|
"grad_norm": 2.5455305576324463,
|
|
"learning_rate": 1.762233874537268e-06,
|
|
"loss": 2.0834,
|
|
"num_tokens": 1353388964.0,
|
|
"step": 22050
|
|
},
|
|
{
|
|
"epoch": 4.053565695941298,
|
|
"grad_norm": 2.239051103591919,
|
|
"learning_rate": 1.7295843737141105e-06,
|
|
"loss": 2.065,
|
|
"num_tokens": 1356456883.0,
|
|
"step": 22100
|
|
},
|
|
{
|
|
"epoch": 4.062737904150424,
|
|
"grad_norm": 2.482544183731079,
|
|
"learning_rate": 1.6972115073949225e-06,
|
|
"loss": 2.0493,
|
|
"num_tokens": 1359525123.0,
|
|
"step": 22150
|
|
},
|
|
{
|
|
"epoch": 4.071910112359551,
|
|
"grad_norm": 2.696545362472534,
|
|
"learning_rate": 1.6651163584094244e-06,
|
|
"loss": 2.096,
|
|
"num_tokens": 1362598592.0,
|
|
"step": 22200
|
|
},
|
|
{
|
|
"epoch": 4.081082320568677,
|
|
"grad_norm": 2.9893839359283447,
|
|
"learning_rate": 1.6333000002980515e-06,
|
|
"loss": 2.0766,
|
|
"num_tokens": 1365667640.0,
|
|
"step": 22250
|
|
},
|
|
{
|
|
"epoch": 4.090254528777804,
|
|
"grad_norm": 2.5702953338623047,
|
|
"learning_rate": 1.6017634972760566e-06,
|
|
"loss": 2.0687,
|
|
"num_tokens": 1368733353.0,
|
|
"step": 22300
|
|
},
|
|
{
|
|
"epoch": 4.0994267369869295,
|
|
"grad_norm": 3.0168826580047607,
|
|
"learning_rate": 1.5705079041979087e-06,
|
|
"loss": 2.0936,
|
|
"num_tokens": 1371803756.0,
|
|
"step": 22350
|
|
},
|
|
{
|
|
"epoch": 4.108598945196056,
|
|
"grad_norm": 2.9977757930755615,
|
|
"learning_rate": 1.5395342665220048e-06,
|
|
"loss": 2.0762,
|
|
"num_tokens": 1374885435.0,
|
|
"step": 22400
|
|
},
|
|
{
|
|
"epoch": 4.117771153405182,
|
|
"grad_norm": 3.476008892059326,
|
|
"learning_rate": 1.5088436202757117e-06,
|
|
"loss": 2.0378,
|
|
"num_tokens": 1377961692.0,
|
|
"step": 22450
|
|
},
|
|
{
|
|
"epoch": 4.126943361614309,
|
|
"grad_norm": 2.5779595375061035,
|
|
"learning_rate": 1.4784369920206988e-06,
|
|
"loss": 2.0774,
|
|
"num_tokens": 1381035718.0,
|
|
"step": 22500
|
|
},
|
|
{
|
|
"epoch": 4.136115569823435,
|
|
"grad_norm": 2.893066644668579,
|
|
"learning_rate": 1.4483153988186115e-06,
|
|
"loss": 2.1011,
|
|
"num_tokens": 1384108091.0,
|
|
"step": 22550
|
|
},
|
|
{
|
|
"epoch": 4.145287778032562,
|
|
"grad_norm": 2.950747489929199,
|
|
"learning_rate": 1.4184798481970508e-06,
|
|
"loss": 2.0854,
|
|
"num_tokens": 1387173009.0,
|
|
"step": 22600
|
|
},
|
|
{
|
|
"epoch": 4.1544599862416876,
|
|
"grad_norm": 2.361630916595459,
|
|
"learning_rate": 1.3889313381158664e-06,
|
|
"loss": 2.045,
|
|
"num_tokens": 1390236725.0,
|
|
"step": 22650
|
|
},
|
|
{
|
|
"epoch": 4.163632194450814,
|
|
"grad_norm": 2.812408924102783,
|
|
"learning_rate": 1.3596708569337846e-06,
|
|
"loss": 2.0747,
|
|
"num_tokens": 1393309356.0,
|
|
"step": 22700
|
|
},
|
|
{
|
|
"epoch": 4.17280440265994,
|
|
"grad_norm": 3.1163132190704346,
|
|
"learning_rate": 1.3306993833753358e-06,
|
|
"loss": 2.0606,
|
|
"num_tokens": 1396381958.0,
|
|
"step": 22750
|
|
},
|
|
{
|
|
"epoch": 4.181976610869067,
|
|
"grad_norm": 2.686840534210205,
|
|
"learning_rate": 1.302017886498138e-06,
|
|
"loss": 2.1085,
|
|
"num_tokens": 1399456162.0,
|
|
"step": 22800
|
|
},
|
|
{
|
|
"epoch": 4.191148819078193,
|
|
"grad_norm": 3.056447982788086,
|
|
"learning_rate": 1.2736273256604592e-06,
|
|
"loss": 2.0799,
|
|
"num_tokens": 1402525285.0,
|
|
"step": 22850
|
|
},
|
|
{
|
|
"epoch": 4.20032102728732,
|
|
"grad_norm": 2.601339101791382,
|
|
"learning_rate": 1.2455286504891528e-06,
|
|
"loss": 2.0744,
|
|
"num_tokens": 1405597752.0,
|
|
"step": 22900
|
|
},
|
|
{
|
|
"epoch": 4.209493235496446,
|
|
"grad_norm": 2.8501698970794678,
|
|
"learning_rate": 1.2177228008478693e-06,
|
|
"loss": 2.059,
|
|
"num_tokens": 1408656836.0,
|
|
"step": 22950
|
|
},
|
|
{
|
|
"epoch": 4.218665443705572,
|
|
"grad_norm": 2.3979616165161133,
|
|
"learning_rate": 1.190210706805639e-06,
|
|
"loss": 2.0643,
|
|
"num_tokens": 1411731246.0,
|
|
"step": 23000
|
|
},
|
|
{
|
|
"epoch": 4.227837651914698,
|
|
"grad_norm": 2.7723591327667236,
|
|
"learning_rate": 1.162993288605755e-06,
|
|
"loss": 2.0828,
|
|
"num_tokens": 1414807082.0,
|
|
"step": 23050
|
|
},
|
|
{
|
|
"epoch": 4.237009860123825,
|
|
"grad_norm": 3.6521003246307373,
|
|
"learning_rate": 1.136071456634984e-06,
|
|
"loss": 2.0899,
|
|
"num_tokens": 1417875267.0,
|
|
"step": 23100
|
|
},
|
|
{
|
|
"epoch": 4.246182068332951,
|
|
"grad_norm": 4.563591003417969,
|
|
"learning_rate": 1.1094461113931332e-06,
|
|
"loss": 2.0788,
|
|
"num_tokens": 1420943916.0,
|
|
"step": 23150
|
|
},
|
|
{
|
|
"epoch": 4.255354276542078,
|
|
"grad_norm": 2.5819358825683594,
|
|
"learning_rate": 1.083118143462909e-06,
|
|
"loss": 2.0563,
|
|
"num_tokens": 1424010773.0,
|
|
"step": 23200
|
|
},
|
|
{
|
|
"epoch": 4.264526484751204,
|
|
"grad_norm": 2.5750629901885986,
|
|
"learning_rate": 1.0570884334801456e-06,
|
|
"loss": 2.0649,
|
|
"num_tokens": 1427075160.0,
|
|
"step": 23250
|
|
},
|
|
{
|
|
"epoch": 4.2736986929603304,
|
|
"grad_norm": 2.886565923690796,
|
|
"learning_rate": 1.0313578521043432e-06,
|
|
"loss": 2.0837,
|
|
"num_tokens": 1430150670.0,
|
|
"step": 23300
|
|
},
|
|
{
|
|
"epoch": 4.282870901169456,
|
|
"grad_norm": 3.7653214931488037,
|
|
"learning_rate": 1.0059272599895364e-06,
|
|
"loss": 2.0467,
|
|
"num_tokens": 1433226218.0,
|
|
"step": 23350
|
|
},
|
|
{
|
|
"epoch": 4.292043109378583,
|
|
"grad_norm": 3.291280508041382,
|
|
"learning_rate": 9.807975077555233e-07,
|
|
"loss": 2.0897,
|
|
"num_tokens": 1436290750.0,
|
|
"step": 23400
|
|
},
|
|
{
|
|
"epoch": 4.301215317587709,
|
|
"grad_norm": 2.654538869857788,
|
|
"learning_rate": 9.55969435959393e-07,
|
|
"loss": 2.0647,
|
|
"num_tokens": 1439372071.0,
|
|
"step": 23450
|
|
},
|
|
{
|
|
"epoch": 4.310387525796836,
|
|
"grad_norm": 4.152379989624023,
|
|
"learning_rate": 9.314438750674393e-07,
|
|
"loss": 2.059,
|
|
"num_tokens": 1442435628.0,
|
|
"step": 23500
|
|
},
|
|
{
|
|
"epoch": 4.319559734005962,
|
|
"grad_norm": 3.305551528930664,
|
|
"learning_rate": 9.072216454273464e-07,
|
|
"loss": 2.0645,
|
|
"num_tokens": 1445511713.0,
|
|
"step": 23550
|
|
},
|
|
{
|
|
"epoch": 4.3287319422150885,
|
|
"grad_norm": 2.8625693321228027,
|
|
"learning_rate": 8.833035572407778e-07,
|
|
"loss": 2.0651,
|
|
"num_tokens": 1448586063.0,
|
|
"step": 23600
|
|
},
|
|
{
|
|
"epoch": 4.337904150424214,
|
|
"grad_norm": 2.5643680095672607,
|
|
"learning_rate": 8.596904105362669e-07,
|
|
"loss": 2.0809,
|
|
"num_tokens": 1451664598.0,
|
|
"step": 23650
|
|
},
|
|
{
|
|
"epoch": 4.347076358633341,
|
|
"grad_norm": 2.9321987628936768,
|
|
"learning_rate": 8.363829951424495e-07,
|
|
"loss": 2.0732,
|
|
"num_tokens": 1454737624.0,
|
|
"step": 23700
|
|
},
|
|
{
|
|
"epoch": 4.356248566842467,
|
|
"grad_norm": 2.9986279010772705,
|
|
"learning_rate": 8.133820906616596e-07,
|
|
"loss": 2.0497,
|
|
"num_tokens": 1457812848.0,
|
|
"step": 23750
|
|
},
|
|
{
|
|
"epoch": 4.365420775051594,
|
|
"grad_norm": 2.493105173110962,
|
|
"learning_rate": 7.906884664438386e-07,
|
|
"loss": 2.1094,
|
|
"num_tokens": 1460882685.0,
|
|
"step": 23800
|
|
},
|
|
{
|
|
"epoch": 4.37459298326072,
|
|
"grad_norm": 2.2998056411743164,
|
|
"learning_rate": 7.683028815608118e-07,
|
|
"loss": 2.0745,
|
|
"num_tokens": 1463958664.0,
|
|
"step": 23850
|
|
},
|
|
{
|
|
"epoch": 4.383765191469847,
|
|
"grad_norm": 2.9117794036865234,
|
|
"learning_rate": 7.462260847808966e-07,
|
|
"loss": 2.0958,
|
|
"num_tokens": 1467029251.0,
|
|
"step": 23900
|
|
},
|
|
{
|
|
"epoch": 4.3929373996789725,
|
|
"grad_norm": 2.936800241470337,
|
|
"learning_rate": 7.24458814543848e-07,
|
|
"loss": 2.0653,
|
|
"num_tokens": 1470095122.0,
|
|
"step": 23950
|
|
},
|
|
{
|
|
"epoch": 4.402109607888099,
|
|
"grad_norm": 2.805723190307617,
|
|
"learning_rate": 7.030017989361748e-07,
|
|
"loss": 2.0705,
|
|
"num_tokens": 1473162799.0,
|
|
"step": 24000
|
|
},
|
|
{
|
|
"epoch": 4.411281816097225,
|
|
"grad_norm": 2.4307007789611816,
|
|
"learning_rate": 6.818557556667693e-07,
|
|
"loss": 2.0691,
|
|
"num_tokens": 1476226733.0,
|
|
"step": 24050
|
|
},
|
|
{
|
|
"epoch": 4.420454024306352,
|
|
"grad_norm": 2.804394006729126,
|
|
"learning_rate": 6.610213920429165e-07,
|
|
"loss": 2.0818,
|
|
"num_tokens": 1479296397.0,
|
|
"step": 24100
|
|
},
|
|
{
|
|
"epoch": 4.429626232515478,
|
|
"grad_norm": 3.200984477996826,
|
|
"learning_rate": 6.404994049466206e-07,
|
|
"loss": 2.0833,
|
|
"num_tokens": 1482374453.0,
|
|
"step": 24150
|
|
},
|
|
{
|
|
"epoch": 4.438798440724605,
|
|
"grad_norm": 2.858473300933838,
|
|
"learning_rate": 6.202904808113075e-07,
|
|
"loss": 2.0574,
|
|
"num_tokens": 1485446701.0,
|
|
"step": 24200
|
|
},
|
|
{
|
|
"epoch": 4.4479706489337305,
|
|
"grad_norm": 2.65733003616333,
|
|
"learning_rate": 6.003952955988613e-07,
|
|
"loss": 2.0773,
|
|
"num_tokens": 1488516075.0,
|
|
"step": 24250
|
|
},
|
|
{
|
|
"epoch": 4.457142857142857,
|
|
"grad_norm": 2.6960391998291016,
|
|
"learning_rate": 5.808145147770017e-07,
|
|
"loss": 2.0768,
|
|
"num_tokens": 1491579965.0,
|
|
"step": 24300
|
|
},
|
|
{
|
|
"epoch": 4.466315065351983,
|
|
"grad_norm": 2.5901076793670654,
|
|
"learning_rate": 5.615487932970531e-07,
|
|
"loss": 2.058,
|
|
"num_tokens": 1494634094.0,
|
|
"step": 24350
|
|
},
|
|
{
|
|
"epoch": 4.47548727356111,
|
|
"grad_norm": 2.5276620388031006,
|
|
"learning_rate": 5.425987755720053e-07,
|
|
"loss": 2.0739,
|
|
"num_tokens": 1497712306.0,
|
|
"step": 24400
|
|
},
|
|
{
|
|
"epoch": 4.484659481770236,
|
|
"grad_norm": 2.5469558238983154,
|
|
"learning_rate": 5.239650954549802e-07,
|
|
"loss": 2.0957,
|
|
"num_tokens": 1500776934.0,
|
|
"step": 24450
|
|
},
|
|
{
|
|
"epoch": 4.493831689979363,
|
|
"grad_norm": 2.691767454147339,
|
|
"learning_rate": 5.056483762180209e-07,
|
|
"loss": 2.0868,
|
|
"num_tokens": 1503853639.0,
|
|
"step": 24500
|
|
},
|
|
{
|
|
"epoch": 4.503003898188489,
|
|
"grad_norm": 2.740900754928589,
|
|
"learning_rate": 4.876492305312497e-07,
|
|
"loss": 2.0723,
|
|
"num_tokens": 1506905888.0,
|
|
"step": 24550
|
|
},
|
|
{
|
|
"epoch": 4.512176106397615,
|
|
"grad_norm": 3.29970383644104,
|
|
"learning_rate": 4.6996826044237165e-07,
|
|
"loss": 2.0774,
|
|
"num_tokens": 1509971293.0,
|
|
"step": 24600
|
|
},
|
|
{
|
|
"epoch": 4.521348314606741,
|
|
"grad_norm": 2.7559125423431396,
|
|
"learning_rate": 4.5260605735653387e-07,
|
|
"loss": 2.07,
|
|
"num_tokens": 1513026402.0,
|
|
"step": 24650
|
|
},
|
|
{
|
|
"epoch": 4.530520522815868,
|
|
"grad_norm": 2.5944247245788574,
|
|
"learning_rate": 4.3556320201655233e-07,
|
|
"loss": 2.0448,
|
|
"num_tokens": 1516094159.0,
|
|
"step": 24700
|
|
},
|
|
{
|
|
"epoch": 4.539692731024994,
|
|
"grad_norm": 2.798725128173828,
|
|
"learning_rate": 4.1884026448347546e-07,
|
|
"loss": 2.0718,
|
|
"num_tokens": 1519158840.0,
|
|
"step": 24750
|
|
},
|
|
{
|
|
"epoch": 4.548864939234121,
|
|
"grad_norm": 2.508417844772339,
|
|
"learning_rate": 4.0243780411752807e-07,
|
|
"loss": 2.0742,
|
|
"num_tokens": 1522222683.0,
|
|
"step": 24800
|
|
},
|
|
{
|
|
"epoch": 4.558037147443247,
|
|
"grad_norm": 2.406092882156372,
|
|
"learning_rate": 3.863563695593975e-07,
|
|
"loss": 2.0822,
|
|
"num_tokens": 1525293067.0,
|
|
"step": 24850
|
|
},
|
|
{
|
|
"epoch": 4.567209355652373,
|
|
"grad_norm": 2.7515735626220703,
|
|
"learning_rate": 3.7059649871187153e-07,
|
|
"loss": 2.0682,
|
|
"num_tokens": 1528361274.0,
|
|
"step": 24900
|
|
},
|
|
{
|
|
"epoch": 4.576381563861499,
|
|
"grad_norm": 2.784795045852661,
|
|
"learning_rate": 3.551587187218675e-07,
|
|
"loss": 2.0675,
|
|
"num_tokens": 1531414162.0,
|
|
"step": 24950
|
|
},
|
|
{
|
|
"epoch": 4.585553772070626,
|
|
"grad_norm": 2.497493267059326,
|
|
"learning_rate": 3.400435459627782e-07,
|
|
"loss": 2.0657,
|
|
"num_tokens": 1534477418.0,
|
|
"step": 25000
|
|
},
|
|
{
|
|
"epoch": 4.594725980279752,
|
|
"grad_norm": 2.718258857727051,
|
|
"learning_rate": 3.252514860172151e-07,
|
|
"loss": 2.063,
|
|
"num_tokens": 1537539682.0,
|
|
"step": 25050
|
|
},
|
|
{
|
|
"epoch": 4.603898188488879,
|
|
"grad_norm": 2.7915687561035156,
|
|
"learning_rate": 3.107830336600914e-07,
|
|
"loss": 2.0806,
|
|
"num_tokens": 1540605734.0,
|
|
"step": 25100
|
|
},
|
|
{
|
|
"epoch": 4.613070396698005,
|
|
"grad_norm": 2.699873685836792,
|
|
"learning_rate": 2.9663867284207006e-07,
|
|
"loss": 2.0684,
|
|
"num_tokens": 1543676637.0,
|
|
"step": 25150
|
|
},
|
|
{
|
|
"epoch": 4.6222426049071315,
|
|
"grad_norm": 2.657258987426758,
|
|
"learning_rate": 2.8281887667338303e-07,
|
|
"loss": 2.0669,
|
|
"num_tokens": 1546747161.0,
|
|
"step": 25200
|
|
},
|
|
{
|
|
"epoch": 4.631414813116257,
|
|
"grad_norm": 2.951749563217163,
|
|
"learning_rate": 2.693241074079989e-07,
|
|
"loss": 2.0777,
|
|
"num_tokens": 1549813714.0,
|
|
"step": 25250
|
|
},
|
|
{
|
|
"epoch": 4.640587021325384,
|
|
"grad_norm": 2.455998420715332,
|
|
"learning_rate": 2.5615481642816687e-07,
|
|
"loss": 2.0541,
|
|
"num_tokens": 1552886589.0,
|
|
"step": 25300
|
|
},
|
|
{
|
|
"epoch": 4.64975922953451,
|
|
"grad_norm": 2.936232805252075,
|
|
"learning_rate": 2.433114442293161e-07,
|
|
"loss": 2.0693,
|
|
"num_tokens": 1555962138.0,
|
|
"step": 25350
|
|
},
|
|
{
|
|
"epoch": 4.658931437743637,
|
|
"grad_norm": 2.830545425415039,
|
|
"learning_rate": 2.307944204053214e-07,
|
|
"loss": 2.0754,
|
|
"num_tokens": 1559040179.0,
|
|
"step": 25400
|
|
},
|
|
{
|
|
"epoch": 4.668103645952764,
|
|
"grad_norm": 2.661842107772827,
|
|
"learning_rate": 2.1860416363413405e-07,
|
|
"loss": 2.0586,
|
|
"num_tokens": 1562118797.0,
|
|
"step": 25450
|
|
},
|
|
{
|
|
"epoch": 4.6772758541618895,
|
|
"grad_norm": 2.921616554260254,
|
|
"learning_rate": 2.0674108166377937e-07,
|
|
"loss": 2.0595,
|
|
"num_tokens": 1565180802.0,
|
|
"step": 25500
|
|
},
|
|
{
|
|
"epoch": 4.686448062371015,
|
|
"grad_norm": 2.7570571899414062,
|
|
"learning_rate": 1.9520557129871775e-07,
|
|
"loss": 2.0723,
|
|
"num_tokens": 1568244479.0,
|
|
"step": 25550
|
|
},
|
|
{
|
|
"epoch": 4.695620270580142,
|
|
"grad_norm": 2.5084118843078613,
|
|
"learning_rate": 1.839980183865664e-07,
|
|
"loss": 2.0763,
|
|
"num_tokens": 1571310946.0,
|
|
"step": 25600
|
|
},
|
|
{
|
|
"epoch": 4.704792478789269,
|
|
"grad_norm": 2.5618412494659424,
|
|
"learning_rate": 1.7311879780520292e-07,
|
|
"loss": 2.0477,
|
|
"num_tokens": 1574377673.0,
|
|
"step": 25650
|
|
},
|
|
{
|
|
"epoch": 4.713964686998395,
|
|
"grad_norm": 3.21343731880188,
|
|
"learning_rate": 1.6256827345021764e-07,
|
|
"loss": 2.0823,
|
|
"num_tokens": 1577444823.0,
|
|
"step": 25700
|
|
},
|
|
{
|
|
"epoch": 4.723136895207521,
|
|
"grad_norm": 2.692683458328247,
|
|
"learning_rate": 1.523467982227489e-07,
|
|
"loss": 2.0593,
|
|
"num_tokens": 1580523382.0,
|
|
"step": 25750
|
|
},
|
|
{
|
|
"epoch": 4.732309103416648,
|
|
"grad_norm": 2.242737054824829,
|
|
"learning_rate": 1.4245471401767242e-07,
|
|
"loss": 2.0705,
|
|
"num_tokens": 1583594375.0,
|
|
"step": 25800
|
|
},
|
|
{
|
|
"epoch": 4.741481311625774,
|
|
"grad_norm": 2.356287956237793,
|
|
"learning_rate": 1.328923517121672e-07,
|
|
"loss": 2.0907,
|
|
"num_tokens": 1586662093.0,
|
|
"step": 25850
|
|
},
|
|
{
|
|
"epoch": 4.7506535198349,
|
|
"grad_norm": 2.768916368484497,
|
|
"learning_rate": 1.236600311546543e-07,
|
|
"loss": 2.0816,
|
|
"num_tokens": 1589738417.0,
|
|
"step": 25900
|
|
},
|
|
{
|
|
"epoch": 4.759825728044026,
|
|
"grad_norm": 2.7122697830200195,
|
|
"learning_rate": 1.1475806115408772e-07,
|
|
"loss": 2.057,
|
|
"num_tokens": 1592816998.0,
|
|
"step": 25950
|
|
},
|
|
{
|
|
"epoch": 4.768997936253153,
|
|
"grad_norm": 2.5500218868255615,
|
|
"learning_rate": 1.0618673946963365e-07,
|
|
"loss": 2.0846,
|
|
"num_tokens": 1595892204.0,
|
|
"step": 26000
|
|
},
|
|
{
|
|
"epoch": 4.77817014446228,
|
|
"grad_norm": 2.6489052772521973,
|
|
"learning_rate": 9.794635280070741e-08,
|
|
"loss": 2.1002,
|
|
"num_tokens": 1598955484.0,
|
|
"step": 26050
|
|
},
|
|
{
|
|
"epoch": 4.787342352671406,
|
|
"grad_norm": 2.427790880203247,
|
|
"learning_rate": 9.003717677738332e-08,
|
|
"loss": 2.0623,
|
|
"num_tokens": 1602016989.0,
|
|
"step": 26100
|
|
},
|
|
{
|
|
"epoch": 4.7965145608805315,
|
|
"grad_norm": 2.5858287811279297,
|
|
"learning_rate": 8.245947595117765e-08,
|
|
"loss": 2.0443,
|
|
"num_tokens": 1605082091.0,
|
|
"step": 26150
|
|
},
|
|
{
|
|
"epoch": 4.805686769089658,
|
|
"grad_norm": 2.5199549198150635,
|
|
"learning_rate": 7.521350378619674e-08,
|
|
"loss": 2.0557,
|
|
"num_tokens": 1608155305.0,
|
|
"step": 26200
|
|
},
|
|
{
|
|
"epoch": 4.814858977298785,
|
|
"grad_norm": 2.4867303371429443,
|
|
"learning_rate": 6.829950265066054e-08,
|
|
"loss": 2.0798,
|
|
"num_tokens": 1611218030.0,
|
|
"step": 26250
|
|
},
|
|
{
|
|
"epoch": 4.824031185507911,
|
|
"grad_norm": 2.278090000152588,
|
|
"learning_rate": 6.171770380879571e-08,
|
|
"loss": 2.0508,
|
|
"num_tokens": 1614283215.0,
|
|
"step": 26300
|
|
},
|
|
{
|
|
"epoch": 4.833203393717037,
|
|
"grad_norm": 2.397080183029175,
|
|
"learning_rate": 5.546832741310071e-08,
|
|
"loss": 2.0656,
|
|
"num_tokens": 1617357636.0,
|
|
"step": 26350
|
|
},
|
|
{
|
|
"epoch": 4.842375601926164,
|
|
"grad_norm": 2.516777992248535,
|
|
"learning_rate": 4.955158249697945e-08,
|
|
"loss": 2.0871,
|
|
"num_tokens": 1620412351.0,
|
|
"step": 26400
|
|
},
|
|
{
|
|
"epoch": 4.8515478101352905,
|
|
"grad_norm": 2.862271308898926,
|
|
"learning_rate": 4.3967666967753565e-08,
|
|
"loss": 2.0826,
|
|
"num_tokens": 1623478926.0,
|
|
"step": 26450
|
|
},
|
|
{
|
|
"epoch": 4.860720018344416,
|
|
"grad_norm": 2.5883853435516357,
|
|
"learning_rate": 3.871676760003884e-08,
|
|
"loss": 2.093,
|
|
"num_tokens": 1626541158.0,
|
|
"step": 26500
|
|
},
|
|
{
|
|
"epoch": 4.869892226553543,
|
|
"grad_norm": 2.6614222526550293,
|
|
"learning_rate": 3.379906002949795e-08,
|
|
"loss": 2.0671,
|
|
"num_tokens": 1629599793.0,
|
|
"step": 26550
|
|
},
|
|
{
|
|
"epoch": 4.879064434762669,
|
|
"grad_norm": 2.4612748622894287,
|
|
"learning_rate": 2.9214708746970745e-08,
|
|
"loss": 2.0579,
|
|
"num_tokens": 1632664837.0,
|
|
"step": 26600
|
|
},
|
|
{
|
|
"epoch": 4.888236642971796,
|
|
"grad_norm": 2.594634532928467,
|
|
"learning_rate": 2.4963867092964167e-08,
|
|
"loss": 2.0643,
|
|
"num_tokens": 1635721838.0,
|
|
"step": 26650
|
|
},
|
|
{
|
|
"epoch": 4.897408851180922,
|
|
"grad_norm": 2.5785598754882812,
|
|
"learning_rate": 2.1046677252531954e-08,
|
|
"loss": 2.0654,
|
|
"num_tokens": 1638785520.0,
|
|
"step": 26700
|
|
},
|
|
{
|
|
"epoch": 4.9065810593900485,
|
|
"grad_norm": 2.447727680206299,
|
|
"learning_rate": 1.7463270250510645e-08,
|
|
"loss": 2.0731,
|
|
"num_tokens": 1641853644.0,
|
|
"step": 26750
|
|
},
|
|
{
|
|
"epoch": 4.915753267599174,
|
|
"grad_norm": 2.757690668106079,
|
|
"learning_rate": 1.4213765947143077e-08,
|
|
"loss": 2.0709,
|
|
"num_tokens": 1644924784.0,
|
|
"step": 26800
|
|
},
|
|
{
|
|
"epoch": 4.924925475808301,
|
|
"grad_norm": 2.783698797225952,
|
|
"learning_rate": 1.1298273034066054e-08,
|
|
"loss": 2.0639,
|
|
"num_tokens": 1647996055.0,
|
|
"step": 26850
|
|
},
|
|
{
|
|
"epoch": 4.934097684017427,
|
|
"grad_norm": 2.42083477973938,
|
|
"learning_rate": 8.716889030675468e-09,
|
|
"loss": 2.0706,
|
|
"num_tokens": 1651076792.0,
|
|
"step": 26900
|
|
},
|
|
{
|
|
"epoch": 4.943269892226554,
|
|
"grad_norm": 2.450413227081299,
|
|
"learning_rate": 6.469700280863356e-09,
|
|
"loss": 2.0566,
|
|
"num_tokens": 1654159645.0,
|
|
"step": 26950
|
|
},
|
|
{
|
|
"epoch": 4.95244210043568,
|
|
"grad_norm": 2.1979095935821533,
|
|
"learning_rate": 4.556781950133537e-09,
|
|
"loss": 2.0627,
|
|
"num_tokens": 1657232226.0,
|
|
"step": 27000
|
|
},
|
|
{
|
|
"epoch": 4.961614308644807,
|
|
"grad_norm": 2.521437168121338,
|
|
"learning_rate": 2.9781980230825235e-09,
|
|
"loss": 2.0674,
|
|
"num_tokens": 1660311253.0,
|
|
"step": 27050
|
|
},
|
|
{
|
|
"epoch": 4.9707865168539325,
|
|
"grad_norm": 2.599595546722412,
|
|
"learning_rate": 1.7340013012634437e-09,
|
|
"loss": 2.0672,
|
|
"num_tokens": 1663386873.0,
|
|
"step": 27100
|
|
},
|
|
{
|
|
"epoch": 4.979958725063059,
|
|
"grad_norm": 2.4748663902282715,
|
|
"learning_rate": 8.24233401415242e-10,
|
|
"loss": 2.0704,
|
|
"num_tokens": 1666451138.0,
|
|
"step": 27150
|
|
},
|
|
{
|
|
"epoch": 4.989130933272185,
|
|
"grad_norm": 2.5703213214874268,
|
|
"learning_rate": 2.489247540771178e-10,
|
|
"loss": 2.0664,
|
|
"num_tokens": 1669520986.0,
|
|
"step": 27200
|
|
},
|
|
{
|
|
"epoch": 4.998303141481312,
|
|
"grad_norm": 3.184465169906616,
|
|
"learning_rate": 8.094602563790688e-12,
|
|
"loss": 2.0632,
|
|
"num_tokens": 1672582468.0,
|
|
"step": 27250
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"eval_loss": 2.3054072856903076,
|
|
"eval_num_tokens": 1673152022.0,
|
|
"eval_runtime": 5.4129,
|
|
"eval_samples_per_second": 89.601,
|
|
"eval_steps_per_second": 5.727,
|
|
"step": 27260
|
|
}
|
|
],
|
|
"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": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 8.53869710299392e+17,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|