bgem3-vi-e1 / trainer_state.json
nntoan209's picture
Upload folder using huggingface_hub
8b955d0 verified
raw
history blame contribute delete
No virus
68.5 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 8546,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 9.972599255002624,
"learning_rate": 4.934477379095164e-07,
"loss": 0.4175,
"step": 20
},
{
"epoch": 0.0,
"grad_norm": 9.535056632903434,
"learning_rate": 8.023400936037442e-07,
"loss": 0.4234,
"step": 40
},
{
"epoch": 0.01,
"grad_norm": 8.69996199325503,
"learning_rate": 1.1112324492979721e-06,
"loss": 0.4021,
"step": 60
},
{
"epoch": 0.01,
"grad_norm": 7.56775274786591,
"learning_rate": 1.4201248049921997e-06,
"loss": 0.393,
"step": 80
},
{
"epoch": 0.01,
"grad_norm": 9.287484589226363,
"learning_rate": 1.7290171606864275e-06,
"loss": 0.4086,
"step": 100
},
{
"epoch": 0.01,
"grad_norm": 8.052654553706383,
"learning_rate": 2.037909516380655e-06,
"loss": 0.3761,
"step": 120
},
{
"epoch": 0.02,
"grad_norm": 12.124856535072642,
"learning_rate": 2.346801872074883e-06,
"loss": 0.3431,
"step": 140
},
{
"epoch": 0.02,
"grad_norm": 7.602396496836849,
"learning_rate": 2.655694227769111e-06,
"loss": 0.3907,
"step": 160
},
{
"epoch": 0.02,
"grad_norm": 7.226437129402364,
"learning_rate": 2.964586583463339e-06,
"loss": 0.3281,
"step": 180
},
{
"epoch": 0.02,
"grad_norm": 23.40605051442028,
"learning_rate": 3.2734789391575668e-06,
"loss": 0.3253,
"step": 200
},
{
"epoch": 0.03,
"grad_norm": 7.708682270572112,
"learning_rate": 3.5823712948517946e-06,
"loss": 0.336,
"step": 220
},
{
"epoch": 0.03,
"grad_norm": 7.7105187144857865,
"learning_rate": 3.891263650546022e-06,
"loss": 0.3283,
"step": 240
},
{
"epoch": 0.03,
"grad_norm": 6.988204554760109,
"learning_rate": 4.200156006240251e-06,
"loss": 0.3512,
"step": 260
},
{
"epoch": 0.03,
"grad_norm": 8.836133501086515,
"learning_rate": 4.509048361934478e-06,
"loss": 0.3266,
"step": 280
},
{
"epoch": 0.04,
"grad_norm": 7.405216965880714,
"learning_rate": 4.817940717628705e-06,
"loss": 0.3259,
"step": 300
},
{
"epoch": 0.04,
"grad_norm": 8.288061025740957,
"learning_rate": 5.126833073322934e-06,
"loss": 0.2918,
"step": 320
},
{
"epoch": 0.04,
"grad_norm": 9.310442601222784,
"learning_rate": 5.435725429017161e-06,
"loss": 0.3229,
"step": 340
},
{
"epoch": 0.04,
"grad_norm": 6.570232957864239,
"learning_rate": 5.744617784711389e-06,
"loss": 0.3063,
"step": 360
},
{
"epoch": 0.04,
"grad_norm": 8.212480249482766,
"learning_rate": 6.053510140405617e-06,
"loss": 0.2867,
"step": 380
},
{
"epoch": 0.05,
"grad_norm": 7.688921541109132,
"learning_rate": 6.362402496099844e-06,
"loss": 0.2987,
"step": 400
},
{
"epoch": 0.05,
"grad_norm": 6.800952048131668,
"learning_rate": 6.671294851794072e-06,
"loss": 0.3047,
"step": 420
},
{
"epoch": 0.05,
"grad_norm": 6.193094209969253,
"learning_rate": 6.9801872074883005e-06,
"loss": 0.316,
"step": 440
},
{
"epoch": 0.05,
"grad_norm": 7.878186262498564,
"learning_rate": 7.289079563182527e-06,
"loss": 0.331,
"step": 460
},
{
"epoch": 0.06,
"grad_norm": 9.807655645103456,
"learning_rate": 7.597971918876755e-06,
"loss": 0.2761,
"step": 480
},
{
"epoch": 0.06,
"grad_norm": 10.286905976546715,
"learning_rate": 7.906864274570983e-06,
"loss": 0.2978,
"step": 500
},
{
"epoch": 0.06,
"grad_norm": 6.140392036969283,
"learning_rate": 8.215756630265212e-06,
"loss": 0.3164,
"step": 520
},
{
"epoch": 0.06,
"grad_norm": 6.714849999088133,
"learning_rate": 8.52464898595944e-06,
"loss": 0.2659,
"step": 540
},
{
"epoch": 0.07,
"grad_norm": 6.369031618010756,
"learning_rate": 8.833541341653666e-06,
"loss": 0.2994,
"step": 560
},
{
"epoch": 0.07,
"grad_norm": 7.515486432421921,
"learning_rate": 9.142433697347895e-06,
"loss": 0.295,
"step": 580
},
{
"epoch": 0.07,
"grad_norm": 6.180107864376804,
"learning_rate": 9.451326053042123e-06,
"loss": 0.3205,
"step": 600
},
{
"epoch": 0.07,
"grad_norm": 8.619489153176001,
"learning_rate": 9.760218408736351e-06,
"loss": 0.2835,
"step": 620
},
{
"epoch": 0.07,
"grad_norm": 5.689814505344143,
"learning_rate": 1.0069110764430576e-05,
"loss": 0.2983,
"step": 640
},
{
"epoch": 0.08,
"grad_norm": 7.447069695249161,
"learning_rate": 1.0378003120124806e-05,
"loss": 0.2736,
"step": 660
},
{
"epoch": 0.08,
"grad_norm": 6.997342558926417,
"learning_rate": 1.0686895475819034e-05,
"loss": 0.2697,
"step": 680
},
{
"epoch": 0.08,
"grad_norm": 9.33003253698042,
"learning_rate": 1.099578783151326e-05,
"loss": 0.2897,
"step": 700
},
{
"epoch": 0.08,
"grad_norm": 7.585581715153782,
"learning_rate": 1.1304680187207489e-05,
"loss": 0.2864,
"step": 720
},
{
"epoch": 0.09,
"grad_norm": 8.595084235914209,
"learning_rate": 1.1613572542901717e-05,
"loss": 0.273,
"step": 740
},
{
"epoch": 0.09,
"grad_norm": 6.3292154008480415,
"learning_rate": 1.1922464898595945e-05,
"loss": 0.2817,
"step": 760
},
{
"epoch": 0.09,
"grad_norm": 14.605530389011415,
"learning_rate": 1.2231357254290172e-05,
"loss": 0.2553,
"step": 780
},
{
"epoch": 0.09,
"grad_norm": 6.541062863744024,
"learning_rate": 1.25402496099844e-05,
"loss": 0.2841,
"step": 800
},
{
"epoch": 0.1,
"grad_norm": 5.8850560260656515,
"learning_rate": 1.2849141965678628e-05,
"loss": 0.2807,
"step": 820
},
{
"epoch": 0.1,
"grad_norm": 6.8846024470910585,
"learning_rate": 1.3158034321372857e-05,
"loss": 0.2676,
"step": 840
},
{
"epoch": 0.1,
"grad_norm": 8.570829697532297,
"learning_rate": 1.3466926677067085e-05,
"loss": 0.2927,
"step": 860
},
{
"epoch": 0.1,
"grad_norm": 5.270646324408607,
"learning_rate": 1.3775819032761311e-05,
"loss": 0.2602,
"step": 880
},
{
"epoch": 0.11,
"grad_norm": 9.01642423951394,
"learning_rate": 1.4084711388455538e-05,
"loss": 0.3052,
"step": 900
},
{
"epoch": 0.11,
"grad_norm": 7.50913242007014,
"learning_rate": 1.4393603744149766e-05,
"loss": 0.2528,
"step": 920
},
{
"epoch": 0.11,
"grad_norm": 9.734458861747513,
"learning_rate": 1.4702496099843994e-05,
"loss": 0.2873,
"step": 940
},
{
"epoch": 0.11,
"grad_norm": 7.558759284363097,
"learning_rate": 1.5011388455538223e-05,
"loss": 0.2422,
"step": 960
},
{
"epoch": 0.11,
"grad_norm": 6.004819523801505,
"learning_rate": 1.532028081123245e-05,
"loss": 0.286,
"step": 980
},
{
"epoch": 0.12,
"grad_norm": 6.551216816169416,
"learning_rate": 1.562917316692668e-05,
"loss": 0.2656,
"step": 1000
},
{
"epoch": 0.12,
"grad_norm": 8.811195222833438,
"learning_rate": 1.5938065522620907e-05,
"loss": 0.2875,
"step": 1020
},
{
"epoch": 0.12,
"grad_norm": 7.722575980788583,
"learning_rate": 1.6246957878315132e-05,
"loss": 0.3019,
"step": 1040
},
{
"epoch": 0.12,
"grad_norm": 7.541514386348537,
"learning_rate": 1.655585023400936e-05,
"loss": 0.2558,
"step": 1060
},
{
"epoch": 0.13,
"grad_norm": 5.48873197937876,
"learning_rate": 1.686474258970359e-05,
"loss": 0.2688,
"step": 1080
},
{
"epoch": 0.13,
"grad_norm": 6.341266315002328,
"learning_rate": 1.7173634945397817e-05,
"loss": 0.2737,
"step": 1100
},
{
"epoch": 0.13,
"grad_norm": 5.9824355844807,
"learning_rate": 1.748252730109205e-05,
"loss": 0.2894,
"step": 1120
},
{
"epoch": 0.13,
"grad_norm": 7.806450783046352,
"learning_rate": 1.7791419656786273e-05,
"loss": 0.2642,
"step": 1140
},
{
"epoch": 0.14,
"grad_norm": 6.224028987297279,
"learning_rate": 1.81003120124805e-05,
"loss": 0.2963,
"step": 1160
},
{
"epoch": 0.14,
"grad_norm": 6.9210132897774,
"learning_rate": 1.840920436817473e-05,
"loss": 0.267,
"step": 1180
},
{
"epoch": 0.14,
"grad_norm": 7.506216143453127,
"learning_rate": 1.8718096723868958e-05,
"loss": 0.2855,
"step": 1200
},
{
"epoch": 0.14,
"grad_norm": 7.678726275837568,
"learning_rate": 1.9026989079563183e-05,
"loss": 0.257,
"step": 1220
},
{
"epoch": 0.15,
"grad_norm": 11.103121729216026,
"learning_rate": 1.933588143525741e-05,
"loss": 0.2615,
"step": 1240
},
{
"epoch": 0.15,
"grad_norm": 7.821432637451867,
"learning_rate": 1.964477379095164e-05,
"loss": 0.2802,
"step": 1260
},
{
"epoch": 0.15,
"grad_norm": 6.305517886474544,
"learning_rate": 1.9953666146645868e-05,
"loss": 0.2799,
"step": 1280
},
{
"epoch": 0.15,
"grad_norm": 6.67203594790867,
"learning_rate": 1.9999973049973136e-05,
"loss": 0.2708,
"step": 1300
},
{
"epoch": 0.15,
"grad_norm": 5.618638234623427,
"learning_rate": 1.999987988957295e-05,
"loss": 0.2964,
"step": 1320
},
{
"epoch": 0.16,
"grad_norm": 7.048756147457391,
"learning_rate": 1.9999720186702903e-05,
"loss": 0.2883,
"step": 1340
},
{
"epoch": 0.16,
"grad_norm": 7.697343680221464,
"learning_rate": 1.999949394242581e-05,
"loss": 0.3157,
"step": 1360
},
{
"epoch": 0.16,
"grad_norm": 5.891830361487415,
"learning_rate": 1.9999201158247326e-05,
"loss": 0.2823,
"step": 1380
},
{
"epoch": 0.16,
"grad_norm": 6.585748092805747,
"learning_rate": 1.9998841836115927e-05,
"loss": 0.2284,
"step": 1400
},
{
"epoch": 0.17,
"grad_norm": 5.437814286894138,
"learning_rate": 1.99984159784229e-05,
"loss": 0.2702,
"step": 1420
},
{
"epoch": 0.17,
"grad_norm": 6.817107156388952,
"learning_rate": 1.9997923588002327e-05,
"loss": 0.2801,
"step": 1440
},
{
"epoch": 0.17,
"grad_norm": 18.42457670763344,
"learning_rate": 1.9997364668131065e-05,
"loss": 0.3078,
"step": 1460
},
{
"epoch": 0.17,
"grad_norm": 9.998003569634188,
"learning_rate": 1.9996739222528713e-05,
"loss": 0.2839,
"step": 1480
},
{
"epoch": 0.18,
"grad_norm": 6.684087435193273,
"learning_rate": 1.999604725535761e-05,
"loss": 0.3049,
"step": 1500
},
{
"epoch": 0.18,
"grad_norm": 6.376706960696693,
"learning_rate": 1.9995288771222792e-05,
"loss": 0.2556,
"step": 1520
},
{
"epoch": 0.18,
"grad_norm": 8.261169271029942,
"learning_rate": 1.999446377517197e-05,
"loss": 0.2766,
"step": 1540
},
{
"epoch": 0.18,
"grad_norm": 6.805041788997332,
"learning_rate": 1.999357227269548e-05,
"loss": 0.2727,
"step": 1560
},
{
"epoch": 0.18,
"grad_norm": 6.8738954523752795,
"learning_rate": 1.999261426972627e-05,
"loss": 0.2616,
"step": 1580
},
{
"epoch": 0.19,
"grad_norm": 5.946552322782328,
"learning_rate": 1.9991589772639836e-05,
"loss": 0.2782,
"step": 1600
},
{
"epoch": 0.19,
"grad_norm": 8.221627199562976,
"learning_rate": 1.9990498788254206e-05,
"loss": 0.2675,
"step": 1620
},
{
"epoch": 0.19,
"grad_norm": 8.064560737732597,
"learning_rate": 1.9989341323829865e-05,
"loss": 0.2947,
"step": 1640
},
{
"epoch": 0.19,
"grad_norm": 15.364919443710688,
"learning_rate": 1.9988117387069733e-05,
"loss": 0.2752,
"step": 1660
},
{
"epoch": 0.2,
"grad_norm": 5.108588948577897,
"learning_rate": 1.9986826986119102e-05,
"loss": 0.2588,
"step": 1680
},
{
"epoch": 0.2,
"grad_norm": 7.676958998385404,
"learning_rate": 1.998547012956558e-05,
"loss": 0.2639,
"step": 1700
},
{
"epoch": 0.2,
"grad_norm": 6.998387014654503,
"learning_rate": 1.9984046826439038e-05,
"loss": 0.2687,
"step": 1720
},
{
"epoch": 0.2,
"grad_norm": 5.572858958673421,
"learning_rate": 1.9982557086211542e-05,
"loss": 0.2903,
"step": 1740
},
{
"epoch": 0.21,
"grad_norm": 8.209097528142939,
"learning_rate": 1.9981000918797306e-05,
"loss": 0.2608,
"step": 1760
},
{
"epoch": 0.21,
"grad_norm": 7.106558508035025,
"learning_rate": 1.9979378334552614e-05,
"loss": 0.2847,
"step": 1780
},
{
"epoch": 0.21,
"grad_norm": 4.4357068374392234,
"learning_rate": 1.997768934427575e-05,
"loss": 0.2348,
"step": 1800
},
{
"epoch": 0.21,
"grad_norm": 6.012720927979585,
"learning_rate": 1.9975933959206926e-05,
"loss": 0.2597,
"step": 1820
},
{
"epoch": 0.22,
"grad_norm": 7.101130876339373,
"learning_rate": 1.997411219102823e-05,
"loss": 0.2804,
"step": 1840
},
{
"epoch": 0.22,
"grad_norm": 6.753118571675995,
"learning_rate": 1.99722240518635e-05,
"loss": 0.2635,
"step": 1860
},
{
"epoch": 0.22,
"grad_norm": 5.548898247312993,
"learning_rate": 1.9970269554278296e-05,
"loss": 0.2312,
"step": 1880
},
{
"epoch": 0.22,
"grad_norm": 6.659248070869921,
"learning_rate": 1.9968248711279784e-05,
"loss": 0.2457,
"step": 1900
},
{
"epoch": 0.22,
"grad_norm": 7.084690577306244,
"learning_rate": 1.996616153631666e-05,
"loss": 0.2654,
"step": 1920
},
{
"epoch": 0.23,
"grad_norm": 5.444008275594034,
"learning_rate": 1.9964008043279056e-05,
"loss": 0.24,
"step": 1940
},
{
"epoch": 0.23,
"grad_norm": 5.36653730738069,
"learning_rate": 1.996178824649845e-05,
"loss": 0.2841,
"step": 1960
},
{
"epoch": 0.23,
"grad_norm": 5.808350425195198,
"learning_rate": 1.9959502160747575e-05,
"loss": 0.2706,
"step": 1980
},
{
"epoch": 0.23,
"grad_norm": 5.156993650505354,
"learning_rate": 1.9957149801240313e-05,
"loss": 0.2696,
"step": 2000
},
{
"epoch": 0.24,
"grad_norm": 5.642951633427938,
"learning_rate": 1.99547311836316e-05,
"loss": 0.2644,
"step": 2020
},
{
"epoch": 0.24,
"grad_norm": 3.4649157253216916,
"learning_rate": 1.995224632401732e-05,
"loss": 0.2567,
"step": 2040
},
{
"epoch": 0.24,
"grad_norm": 10.299176851731511,
"learning_rate": 1.9949695238934187e-05,
"loss": 0.264,
"step": 2060
},
{
"epoch": 0.24,
"grad_norm": 9.341352786786667,
"learning_rate": 1.994707794535966e-05,
"loss": 0.288,
"step": 2080
},
{
"epoch": 0.25,
"grad_norm": 5.416770268941113,
"learning_rate": 1.9944394460711806e-05,
"loss": 0.2342,
"step": 2100
},
{
"epoch": 0.25,
"grad_norm": 6.406230051893485,
"learning_rate": 1.9941644802849197e-05,
"loss": 0.2684,
"step": 2120
},
{
"epoch": 0.25,
"grad_norm": 5.161284052745025,
"learning_rate": 1.993882899007078e-05,
"loss": 0.2733,
"step": 2140
},
{
"epoch": 0.25,
"grad_norm": 5.193421993616942,
"learning_rate": 1.9935947041115774e-05,
"loss": 0.2623,
"step": 2160
},
{
"epoch": 0.26,
"grad_norm": 13.024715697248084,
"learning_rate": 1.9932998975163525e-05,
"loss": 0.2702,
"step": 2180
},
{
"epoch": 0.26,
"grad_norm": 6.8241910797414125,
"learning_rate": 1.9929984811833385e-05,
"loss": 0.2566,
"step": 2200
},
{
"epoch": 0.26,
"grad_norm": 6.730073692519145,
"learning_rate": 1.9926904571184594e-05,
"loss": 0.242,
"step": 2220
},
{
"epoch": 0.26,
"grad_norm": 9.451222932996513,
"learning_rate": 1.992375827371613e-05,
"loss": 0.278,
"step": 2240
},
{
"epoch": 0.26,
"grad_norm": 8.019858508300759,
"learning_rate": 1.9920545940366576e-05,
"loss": 0.2341,
"step": 2260
},
{
"epoch": 0.27,
"grad_norm": 5.705986412954751,
"learning_rate": 1.9917267592513983e-05,
"loss": 0.2598,
"step": 2280
},
{
"epoch": 0.27,
"grad_norm": 6.536565196669616,
"learning_rate": 1.991392325197574e-05,
"loss": 0.3296,
"step": 2300
},
{
"epoch": 0.27,
"grad_norm": 6.445428798666668,
"learning_rate": 1.99105129410084e-05,
"loss": 0.26,
"step": 2320
},
{
"epoch": 0.27,
"grad_norm": 7.151819181763656,
"learning_rate": 1.9907036682307563e-05,
"loss": 0.2571,
"step": 2340
},
{
"epoch": 0.28,
"grad_norm": 5.051575919032304,
"learning_rate": 1.99034944990077e-05,
"loss": 0.2623,
"step": 2360
},
{
"epoch": 0.28,
"grad_norm": 4.396965905437974,
"learning_rate": 1.9899886414682012e-05,
"loss": 0.222,
"step": 2380
},
{
"epoch": 0.28,
"grad_norm": 5.550275936655984,
"learning_rate": 1.989621245334228e-05,
"loss": 0.2472,
"step": 2400
},
{
"epoch": 0.28,
"grad_norm": 7.484173415620986,
"learning_rate": 1.9892472639438684e-05,
"loss": 0.2572,
"step": 2420
},
{
"epoch": 0.29,
"grad_norm": 6.330013944541192,
"learning_rate": 1.9888666997859657e-05,
"loss": 0.256,
"step": 2440
},
{
"epoch": 0.29,
"grad_norm": 6.839909560291623,
"learning_rate": 1.9884795553931713e-05,
"loss": 0.2505,
"step": 2460
},
{
"epoch": 0.29,
"grad_norm": 3.7368937022928987,
"learning_rate": 1.988085833341929e-05,
"loss": 0.2232,
"step": 2480
},
{
"epoch": 0.29,
"grad_norm": 7.700733239077536,
"learning_rate": 1.9876855362524555e-05,
"loss": 0.2513,
"step": 2500
},
{
"epoch": 0.29,
"grad_norm": 9.003459689305725,
"learning_rate": 1.9872786667887246e-05,
"loss": 0.2701,
"step": 2520
},
{
"epoch": 0.3,
"grad_norm": 6.173066538462164,
"learning_rate": 1.9868652276584498e-05,
"loss": 0.2641,
"step": 2540
},
{
"epoch": 0.3,
"grad_norm": 7.517799617986063,
"learning_rate": 1.9864452216130656e-05,
"loss": 0.2602,
"step": 2560
},
{
"epoch": 0.3,
"grad_norm": 5.206716769795917,
"learning_rate": 1.986018651447709e-05,
"loss": 0.2702,
"step": 2580
},
{
"epoch": 0.3,
"grad_norm": 5.352469307581187,
"learning_rate": 1.9855855200012012e-05,
"loss": 0.2551,
"step": 2600
},
{
"epoch": 0.31,
"grad_norm": 4.307397892605768,
"learning_rate": 1.985145830156028e-05,
"loss": 0.2641,
"step": 2620
},
{
"epoch": 0.31,
"grad_norm": 4.264075140901128,
"learning_rate": 1.9846995848383227e-05,
"loss": 0.2531,
"step": 2640
},
{
"epoch": 0.31,
"grad_norm": 7.436472589390343,
"learning_rate": 1.9842467870178446e-05,
"loss": 0.2338,
"step": 2660
},
{
"epoch": 0.31,
"grad_norm": 7.224638151723576,
"learning_rate": 1.9837874397079597e-05,
"loss": 0.2704,
"step": 2680
},
{
"epoch": 0.32,
"grad_norm": 5.457029152871823,
"learning_rate": 1.9833215459656204e-05,
"loss": 0.2268,
"step": 2700
},
{
"epoch": 0.32,
"grad_norm": 8.049908882715352,
"learning_rate": 1.9828491088913466e-05,
"loss": 0.2761,
"step": 2720
},
{
"epoch": 0.32,
"grad_norm": 5.589531357284561,
"learning_rate": 1.9823701316292027e-05,
"loss": 0.2226,
"step": 2740
},
{
"epoch": 0.32,
"grad_norm": 5.958210692666308,
"learning_rate": 1.9818846173667795e-05,
"loss": 0.2351,
"step": 2760
},
{
"epoch": 0.33,
"grad_norm": 4.946607085809116,
"learning_rate": 1.9813925693351706e-05,
"loss": 0.2428,
"step": 2780
},
{
"epoch": 0.33,
"grad_norm": 5.0976059249790815,
"learning_rate": 1.9808939908089514e-05,
"loss": 0.2694,
"step": 2800
},
{
"epoch": 0.33,
"grad_norm": 3.8642265929763013,
"learning_rate": 1.9803888851061584e-05,
"loss": 0.2507,
"step": 2820
},
{
"epoch": 0.33,
"grad_norm": 5.4071017790539,
"learning_rate": 1.979877255588267e-05,
"loss": 0.2443,
"step": 2840
},
{
"epoch": 0.33,
"grad_norm": 5.291228849466517,
"learning_rate": 1.979359105660167e-05,
"loss": 0.2449,
"step": 2860
},
{
"epoch": 0.34,
"grad_norm": 18.084778243085054,
"learning_rate": 1.9788344387701422e-05,
"loss": 0.2587,
"step": 2880
},
{
"epoch": 0.34,
"grad_norm": 7.733243732371791,
"learning_rate": 1.9783032584098475e-05,
"loss": 0.2534,
"step": 2900
},
{
"epoch": 0.34,
"grad_norm": 5.9419057461797875,
"learning_rate": 1.9777655681142828e-05,
"loss": 0.2887,
"step": 2920
},
{
"epoch": 0.34,
"grad_norm": 6.194967060632705,
"learning_rate": 1.9772213714617746e-05,
"loss": 0.271,
"step": 2940
},
{
"epoch": 0.35,
"grad_norm": 5.416199101539707,
"learning_rate": 1.9766706720739462e-05,
"loss": 0.2479,
"step": 2960
},
{
"epoch": 0.35,
"grad_norm": 7.381640790370163,
"learning_rate": 1.9761134736156984e-05,
"loss": 0.253,
"step": 2980
},
{
"epoch": 0.35,
"grad_norm": 5.289607190437299,
"learning_rate": 1.975549779795183e-05,
"loss": 0.2398,
"step": 3000
},
{
"epoch": 0.35,
"grad_norm": 6.393511952492927,
"learning_rate": 1.9749795943637778e-05,
"loss": 0.2527,
"step": 3020
},
{
"epoch": 0.36,
"grad_norm": 5.786789008808832,
"learning_rate": 1.9744029211160623e-05,
"loss": 0.264,
"step": 3040
},
{
"epoch": 0.36,
"grad_norm": 6.325619802149641,
"learning_rate": 1.9738197638897923e-05,
"loss": 0.2604,
"step": 3060
},
{
"epoch": 0.36,
"grad_norm": 5.138427821373369,
"learning_rate": 1.973230126565876e-05,
"loss": 0.2501,
"step": 3080
},
{
"epoch": 0.36,
"grad_norm": 5.754983152469265,
"learning_rate": 1.972634013068344e-05,
"loss": 0.2563,
"step": 3100
},
{
"epoch": 0.37,
"grad_norm": 8.479269441048931,
"learning_rate": 1.9720314273643277e-05,
"loss": 0.2428,
"step": 3120
},
{
"epoch": 0.37,
"grad_norm": 5.213630952429399,
"learning_rate": 1.9714223734640303e-05,
"loss": 0.2161,
"step": 3140
},
{
"epoch": 0.37,
"grad_norm": 4.644334537071628,
"learning_rate": 1.970806855420701e-05,
"loss": 0.2529,
"step": 3160
},
{
"epoch": 0.37,
"grad_norm": 6.1305106149806665,
"learning_rate": 1.9701848773306074e-05,
"loss": 0.236,
"step": 3180
},
{
"epoch": 0.37,
"grad_norm": 5.353262482329062,
"learning_rate": 1.9695564433330096e-05,
"loss": 0.2549,
"step": 3200
},
{
"epoch": 0.38,
"grad_norm": 5.060708753381077,
"learning_rate": 1.9689215576101306e-05,
"loss": 0.215,
"step": 3220
},
{
"epoch": 0.38,
"grad_norm": 6.674948634260756,
"learning_rate": 1.9682802243871302e-05,
"loss": 0.2383,
"step": 3240
},
{
"epoch": 0.38,
"grad_norm": 7.099655314930459,
"learning_rate": 1.9676324479320758e-05,
"loss": 0.2343,
"step": 3260
},
{
"epoch": 0.38,
"grad_norm": 6.495667113827656,
"learning_rate": 1.9669782325559156e-05,
"loss": 0.271,
"step": 3280
},
{
"epoch": 0.39,
"grad_norm": 8.167355047783982,
"learning_rate": 1.9663175826124467e-05,
"loss": 0.275,
"step": 3300
},
{
"epoch": 0.39,
"grad_norm": 5.097460933758238,
"learning_rate": 1.9656505024982906e-05,
"loss": 0.2525,
"step": 3320
},
{
"epoch": 0.39,
"grad_norm": 3.5743234754551723,
"learning_rate": 1.9649769966528606e-05,
"loss": 0.2607,
"step": 3340
},
{
"epoch": 0.39,
"grad_norm": 6.011608021133435,
"learning_rate": 1.9642970695583317e-05,
"loss": 0.2493,
"step": 3360
},
{
"epoch": 0.4,
"grad_norm": 4.773080465946433,
"learning_rate": 1.9636107257396144e-05,
"loss": 0.231,
"step": 3380
},
{
"epoch": 0.4,
"grad_norm": 5.449977370748874,
"learning_rate": 1.9629179697643214e-05,
"loss": 0.2466,
"step": 3400
},
{
"epoch": 0.4,
"grad_norm": 5.36397841910189,
"learning_rate": 1.9622188062427393e-05,
"loss": 0.2531,
"step": 3420
},
{
"epoch": 0.4,
"grad_norm": 5.451392655529927,
"learning_rate": 1.9615132398277943e-05,
"loss": 0.2354,
"step": 3440
},
{
"epoch": 0.4,
"grad_norm": 7.143537532553735,
"learning_rate": 1.960801275215027e-05,
"loss": 0.2632,
"step": 3460
},
{
"epoch": 0.41,
"grad_norm": 3.4010860952344717,
"learning_rate": 1.960082917142556e-05,
"loss": 0.2362,
"step": 3480
},
{
"epoch": 0.41,
"grad_norm": 5.840079557321102,
"learning_rate": 1.959358170391049e-05,
"loss": 0.2402,
"step": 3500
},
{
"epoch": 0.41,
"grad_norm": 5.938052824284585,
"learning_rate": 1.9586270397836898e-05,
"loss": 0.2407,
"step": 3520
},
{
"epoch": 0.41,
"grad_norm": 5.132153426638341,
"learning_rate": 1.9578895301861475e-05,
"loss": 0.2401,
"step": 3540
},
{
"epoch": 0.42,
"grad_norm": 5.87746718526918,
"learning_rate": 1.9571456465065425e-05,
"loss": 0.2749,
"step": 3560
},
{
"epoch": 0.42,
"grad_norm": 11.830889239002573,
"learning_rate": 1.956395393695415e-05,
"loss": 0.2302,
"step": 3580
},
{
"epoch": 0.42,
"grad_norm": 4.661847737277734,
"learning_rate": 1.955638776745692e-05,
"loss": 0.2359,
"step": 3600
},
{
"epoch": 0.42,
"grad_norm": 4.245057766614552,
"learning_rate": 1.9548758006926537e-05,
"loss": 0.2534,
"step": 3620
},
{
"epoch": 0.43,
"grad_norm": 6.400102149625017,
"learning_rate": 1.9541064706138998e-05,
"loss": 0.2513,
"step": 3640
},
{
"epoch": 0.43,
"grad_norm": 7.7926185012927105,
"learning_rate": 1.953330791629316e-05,
"loss": 0.2508,
"step": 3660
},
{
"epoch": 0.43,
"grad_norm": 5.502373096713324,
"learning_rate": 1.9525487689010402e-05,
"loss": 0.2316,
"step": 3680
},
{
"epoch": 0.43,
"grad_norm": 4.600917699743126,
"learning_rate": 1.951760407633428e-05,
"loss": 0.2207,
"step": 3700
},
{
"epoch": 0.44,
"grad_norm": 5.748567153866237,
"learning_rate": 1.950965713073017e-05,
"loss": 0.2383,
"step": 3720
},
{
"epoch": 0.44,
"grad_norm": 7.113580486910231,
"learning_rate": 1.950164690508494e-05,
"loss": 0.2362,
"step": 3740
},
{
"epoch": 0.44,
"grad_norm": 4.385449600398098,
"learning_rate": 1.9493573452706576e-05,
"loss": 0.2646,
"step": 3760
},
{
"epoch": 0.44,
"grad_norm": 6.270456330213243,
"learning_rate": 1.948543682732385e-05,
"loss": 0.2558,
"step": 3780
},
{
"epoch": 0.44,
"grad_norm": 6.787086330896131,
"learning_rate": 1.9477237083085934e-05,
"loss": 0.2446,
"step": 3800
},
{
"epoch": 0.45,
"grad_norm": 4.424978723852011,
"learning_rate": 1.946897427456207e-05,
"loss": 0.2658,
"step": 3820
},
{
"epoch": 0.45,
"grad_norm": 5.168068951870145,
"learning_rate": 1.946064845674118e-05,
"loss": 0.2553,
"step": 3840
},
{
"epoch": 0.45,
"grad_norm": 5.442116191965272,
"learning_rate": 1.9452259685031528e-05,
"loss": 0.27,
"step": 3860
},
{
"epoch": 0.45,
"grad_norm": 6.214077368189247,
"learning_rate": 1.944380801526032e-05,
"loss": 0.2472,
"step": 3880
},
{
"epoch": 0.46,
"grad_norm": 5.0773790016952125,
"learning_rate": 1.943529350367335e-05,
"loss": 0.2416,
"step": 3900
},
{
"epoch": 0.46,
"grad_norm": 5.691248067465491,
"learning_rate": 1.9426716206934634e-05,
"loss": 0.2666,
"step": 3920
},
{
"epoch": 0.46,
"grad_norm": 8.072722349391876,
"learning_rate": 1.9418076182126016e-05,
"loss": 0.2642,
"step": 3940
},
{
"epoch": 0.46,
"grad_norm": 6.147111800674382,
"learning_rate": 1.9409373486746796e-05,
"loss": 0.2736,
"step": 3960
},
{
"epoch": 0.47,
"grad_norm": 6.10278933178705,
"learning_rate": 1.9400608178713344e-05,
"loss": 0.2241,
"step": 3980
},
{
"epoch": 0.47,
"grad_norm": 11.62252289006471,
"learning_rate": 1.939178031635872e-05,
"loss": 0.2305,
"step": 4000
},
{
"epoch": 0.47,
"grad_norm": 5.344565702945431,
"learning_rate": 1.9382889958432278e-05,
"loss": 0.2534,
"step": 4020
},
{
"epoch": 0.47,
"grad_norm": 5.186472227935148,
"learning_rate": 1.9373937164099286e-05,
"loss": 0.2196,
"step": 4040
},
{
"epoch": 0.48,
"grad_norm": 5.878287511563509,
"learning_rate": 1.9364921992940528e-05,
"loss": 0.2304,
"step": 4060
},
{
"epoch": 0.48,
"grad_norm": 5.994204901496365,
"learning_rate": 1.935584450495188e-05,
"loss": 0.2237,
"step": 4080
},
{
"epoch": 0.48,
"grad_norm": 6.796206246284836,
"learning_rate": 1.9346704760543978e-05,
"loss": 0.2437,
"step": 4100
},
{
"epoch": 0.48,
"grad_norm": 7.852752747158656,
"learning_rate": 1.9337502820541742e-05,
"loss": 0.2438,
"step": 4120
},
{
"epoch": 0.48,
"grad_norm": 4.969180490330714,
"learning_rate": 1.9328238746184015e-05,
"loss": 0.2216,
"step": 4140
},
{
"epoch": 0.49,
"grad_norm": 6.896232416719303,
"learning_rate": 1.9318912599123136e-05,
"loss": 0.2268,
"step": 4160
},
{
"epoch": 0.49,
"grad_norm": 7.925370929112004,
"learning_rate": 1.930952444142455e-05,
"loss": 0.2182,
"step": 4180
},
{
"epoch": 0.49,
"grad_norm": 5.198662571209722,
"learning_rate": 1.930007433556637e-05,
"loss": 0.2506,
"step": 4200
},
{
"epoch": 0.49,
"grad_norm": 5.826437370857378,
"learning_rate": 1.9290562344438983e-05,
"loss": 0.2437,
"step": 4220
},
{
"epoch": 0.5,
"grad_norm": 6.687619217139761,
"learning_rate": 1.9280988531344616e-05,
"loss": 0.224,
"step": 4240
},
{
"epoch": 0.5,
"grad_norm": 5.17828230403087,
"learning_rate": 1.9271352959996922e-05,
"loss": 0.2304,
"step": 4260
},
{
"epoch": 0.5,
"grad_norm": 5.633269271636816,
"learning_rate": 1.9261655694520554e-05,
"loss": 0.2501,
"step": 4280
},
{
"epoch": 0.5,
"grad_norm": 8.027110417077195,
"learning_rate": 1.925189679945075e-05,
"loss": 0.24,
"step": 4300
},
{
"epoch": 0.51,
"grad_norm": 3.9526632288043206,
"learning_rate": 1.9242076339732868e-05,
"loss": 0.2464,
"step": 4320
},
{
"epoch": 0.51,
"grad_norm": 24.011155396941934,
"learning_rate": 1.9232194380722e-05,
"loss": 0.242,
"step": 4340
},
{
"epoch": 0.51,
"grad_norm": 6.901238880338521,
"learning_rate": 1.922225098818251e-05,
"loss": 0.2233,
"step": 4360
},
{
"epoch": 0.51,
"grad_norm": 4.6063725407318925,
"learning_rate": 1.9212246228287598e-05,
"loss": 0.2436,
"step": 4380
},
{
"epoch": 0.51,
"grad_norm": 6.595648758844639,
"learning_rate": 1.9202180167618868e-05,
"loss": 0.2274,
"step": 4400
},
{
"epoch": 0.52,
"grad_norm": 4.8087240491206,
"learning_rate": 1.9192052873165873e-05,
"loss": 0.2259,
"step": 4420
},
{
"epoch": 0.52,
"grad_norm": 5.967897364247756,
"learning_rate": 1.9181864412325683e-05,
"loss": 0.2087,
"step": 4440
},
{
"epoch": 0.52,
"grad_norm": 5.531641886993624,
"learning_rate": 1.9171614852902424e-05,
"loss": 0.2467,
"step": 4460
},
{
"epoch": 0.52,
"grad_norm": 4.809099854052731,
"learning_rate": 1.9161304263106844e-05,
"loss": 0.253,
"step": 4480
},
{
"epoch": 0.53,
"grad_norm": 5.503146225359231,
"learning_rate": 1.9150932711555832e-05,
"loss": 0.2634,
"step": 4500
},
{
"epoch": 0.53,
"grad_norm": 6.074743433428479,
"learning_rate": 1.914050026727199e-05,
"loss": 0.241,
"step": 4520
},
{
"epoch": 0.53,
"grad_norm": 9.683286476569032,
"learning_rate": 1.9130006999683156e-05,
"loss": 0.2378,
"step": 4540
},
{
"epoch": 0.53,
"grad_norm": 4.549479050209584,
"learning_rate": 1.9119452978621942e-05,
"loss": 0.234,
"step": 4560
},
{
"epoch": 0.54,
"grad_norm": 5.11133278951115,
"learning_rate": 1.9108838274325284e-05,
"loss": 0.1995,
"step": 4580
},
{
"epoch": 0.54,
"grad_norm": 3.9316257072882044,
"learning_rate": 1.9098162957433953e-05,
"loss": 0.2238,
"step": 4600
},
{
"epoch": 0.54,
"grad_norm": 5.645831888158784,
"learning_rate": 1.90874270989921e-05,
"loss": 0.2437,
"step": 4620
},
{
"epoch": 0.54,
"grad_norm": 6.604181075055343,
"learning_rate": 1.9076630770446795e-05,
"loss": 0.2181,
"step": 4640
},
{
"epoch": 0.55,
"grad_norm": 5.8750461008413986,
"learning_rate": 1.9065774043647508e-05,
"loss": 0.2271,
"step": 4660
},
{
"epoch": 0.55,
"grad_norm": 4.745625740492316,
"learning_rate": 1.9054856990845676e-05,
"loss": 0.2568,
"step": 4680
},
{
"epoch": 0.55,
"grad_norm": 6.812269574389316,
"learning_rate": 1.904387968469421e-05,
"loss": 0.2371,
"step": 4700
},
{
"epoch": 0.55,
"grad_norm": 5.799364897057997,
"learning_rate": 1.903284219824699e-05,
"loss": 0.2314,
"step": 4720
},
{
"epoch": 0.55,
"grad_norm": 6.48625343415292,
"learning_rate": 1.9021744604958413e-05,
"loss": 0.219,
"step": 4740
},
{
"epoch": 0.56,
"grad_norm": 6.278707264795213,
"learning_rate": 1.9010586978682876e-05,
"loss": 0.2276,
"step": 4760
},
{
"epoch": 0.56,
"grad_norm": 5.371555555431128,
"learning_rate": 1.899936939367429e-05,
"loss": 0.2363,
"step": 4780
},
{
"epoch": 0.56,
"grad_norm": 5.40795130778857,
"learning_rate": 1.898809192458561e-05,
"loss": 0.2366,
"step": 4800
},
{
"epoch": 0.56,
"grad_norm": 7.243141086845891,
"learning_rate": 1.8976754646468297e-05,
"loss": 0.2236,
"step": 4820
},
{
"epoch": 0.57,
"grad_norm": 5.568031389685097,
"learning_rate": 1.8965357634771863e-05,
"loss": 0.2233,
"step": 4840
},
{
"epoch": 0.57,
"grad_norm": 4.403544142992904,
"learning_rate": 1.895390096534332e-05,
"loss": 0.231,
"step": 4860
},
{
"epoch": 0.57,
"grad_norm": 5.810384057592795,
"learning_rate": 1.8942384714426724e-05,
"loss": 0.227,
"step": 4880
},
{
"epoch": 0.57,
"grad_norm": 7.307934240436817,
"learning_rate": 1.8930808958662634e-05,
"loss": 0.2374,
"step": 4900
},
{
"epoch": 0.58,
"grad_norm": 6.000264161970516,
"learning_rate": 1.891917377508762e-05,
"loss": 0.2357,
"step": 4920
},
{
"epoch": 0.58,
"grad_norm": 6.840584636933369,
"learning_rate": 1.8907479241133743e-05,
"loss": 0.2209,
"step": 4940
},
{
"epoch": 0.58,
"grad_norm": 5.060412317899446,
"learning_rate": 1.8895725434628033e-05,
"loss": 0.2165,
"step": 4960
},
{
"epoch": 0.58,
"grad_norm": 5.819876963045717,
"learning_rate": 1.888391243379199e-05,
"loss": 0.2222,
"step": 4980
},
{
"epoch": 0.59,
"grad_norm": 5.534379838696889,
"learning_rate": 1.8872040317241033e-05,
"loss": 0.2319,
"step": 5000
},
{
"epoch": 0.59,
"grad_norm": 5.311173217247157,
"learning_rate": 1.8860109163984017e-05,
"loss": 0.2184,
"step": 5020
},
{
"epoch": 0.59,
"grad_norm": 4.196681646351116,
"learning_rate": 1.884811905342267e-05,
"loss": 0.2136,
"step": 5040
},
{
"epoch": 0.59,
"grad_norm": 4.147633735105589,
"learning_rate": 1.8836070065351094e-05,
"loss": 0.2228,
"step": 5060
},
{
"epoch": 0.59,
"grad_norm": 5.977670444689344,
"learning_rate": 1.882396227995521e-05,
"loss": 0.2196,
"step": 5080
},
{
"epoch": 0.6,
"grad_norm": 5.9509298928281344,
"learning_rate": 1.881179577781224e-05,
"loss": 0.2128,
"step": 5100
},
{
"epoch": 0.6,
"grad_norm": 4.308487061660745,
"learning_rate": 1.8799570639890162e-05,
"loss": 0.2034,
"step": 5120
},
{
"epoch": 0.6,
"grad_norm": 5.579788771975886,
"learning_rate": 1.8787286947547173e-05,
"loss": 0.2163,
"step": 5140
},
{
"epoch": 0.6,
"grad_norm": 5.331086082666531,
"learning_rate": 1.8774944782531157e-05,
"loss": 0.2387,
"step": 5160
},
{
"epoch": 0.61,
"grad_norm": 4.859846184262665,
"learning_rate": 1.8762544226979123e-05,
"loss": 0.228,
"step": 5180
},
{
"epoch": 0.61,
"grad_norm": 8.956508243561895,
"learning_rate": 1.8750085363416685e-05,
"loss": 0.205,
"step": 5200
},
{
"epoch": 0.61,
"grad_norm": 6.011375134855825,
"learning_rate": 1.8737568274757472e-05,
"loss": 0.2554,
"step": 5220
},
{
"epoch": 0.61,
"grad_norm": 4.884307583607841,
"learning_rate": 1.8724993044302627e-05,
"loss": 0.2426,
"step": 5240
},
{
"epoch": 0.62,
"grad_norm": 5.502332192912132,
"learning_rate": 1.87123597557402e-05,
"loss": 0.2113,
"step": 5260
},
{
"epoch": 0.62,
"grad_norm": 4.10529243533445,
"learning_rate": 1.8699668493144644e-05,
"loss": 0.212,
"step": 5280
},
{
"epoch": 0.62,
"grad_norm": 5.764936204410069,
"learning_rate": 1.8686919340976204e-05,
"loss": 0.2411,
"step": 5300
},
{
"epoch": 0.62,
"grad_norm": 5.522034635540896,
"learning_rate": 1.86741123840804e-05,
"loss": 0.2216,
"step": 5320
},
{
"epoch": 0.62,
"grad_norm": 7.991532135733726,
"learning_rate": 1.8661247707687427e-05,
"loss": 0.2194,
"step": 5340
},
{
"epoch": 0.63,
"grad_norm": 9.518656030242417,
"learning_rate": 1.864832539741161e-05,
"loss": 0.2587,
"step": 5360
},
{
"epoch": 0.63,
"grad_norm": 3.577063390505517,
"learning_rate": 1.863534553925083e-05,
"loss": 0.2102,
"step": 5380
},
{
"epoch": 0.63,
"grad_norm": 6.534884231114006,
"learning_rate": 1.862230821958594e-05,
"loss": 0.2557,
"step": 5400
},
{
"epoch": 0.63,
"grad_norm": 11.231289372192048,
"learning_rate": 1.8609213525180207e-05,
"loss": 0.2181,
"step": 5420
},
{
"epoch": 0.64,
"grad_norm": 6.42773971626943,
"learning_rate": 1.8596061543178722e-05,
"loss": 0.2048,
"step": 5440
},
{
"epoch": 0.64,
"grad_norm": 5.645681381526835,
"learning_rate": 1.858285236110782e-05,
"loss": 0.2239,
"step": 5460
},
{
"epoch": 0.64,
"grad_norm": 5.521382987053305,
"learning_rate": 1.8569586066874516e-05,
"loss": 0.2536,
"step": 5480
},
{
"epoch": 0.64,
"grad_norm": 6.458490316472477,
"learning_rate": 1.8556262748765894e-05,
"loss": 0.2274,
"step": 5500
},
{
"epoch": 0.65,
"grad_norm": 5.396911317032735,
"learning_rate": 1.854288249544853e-05,
"loss": 0.2208,
"step": 5520
},
{
"epoch": 0.65,
"grad_norm": 4.837238093735751,
"learning_rate": 1.852944539596792e-05,
"loss": 0.2088,
"step": 5540
},
{
"epoch": 0.65,
"grad_norm": 4.897482454931544,
"learning_rate": 1.851595153974785e-05,
"loss": 0.2387,
"step": 5560
},
{
"epoch": 0.65,
"grad_norm": 4.932207578796922,
"learning_rate": 1.850240101658983e-05,
"loss": 0.204,
"step": 5580
},
{
"epoch": 0.66,
"grad_norm": 5.536196980329085,
"learning_rate": 1.8488793916672495e-05,
"loss": 0.212,
"step": 5600
},
{
"epoch": 0.66,
"grad_norm": 7.074855563033988,
"learning_rate": 1.8475130330550987e-05,
"loss": 0.2258,
"step": 5620
},
{
"epoch": 0.66,
"grad_norm": 4.714164678662987,
"learning_rate": 1.8461410349156373e-05,
"loss": 0.2142,
"step": 5640
},
{
"epoch": 0.66,
"grad_norm": 5.118777335330323,
"learning_rate": 1.8447634063795017e-05,
"loss": 0.2258,
"step": 5660
},
{
"epoch": 0.66,
"grad_norm": 5.367581593178051,
"learning_rate": 1.8433801566148005e-05,
"loss": 0.1945,
"step": 5680
},
{
"epoch": 0.67,
"grad_norm": 1.9698062666888698,
"learning_rate": 1.8419912948270494e-05,
"loss": 0.201,
"step": 5700
},
{
"epoch": 0.67,
"grad_norm": 4.6618833323265,
"learning_rate": 1.8405968302591135e-05,
"loss": 0.2409,
"step": 5720
},
{
"epoch": 0.67,
"grad_norm": 5.546962683615939,
"learning_rate": 1.839196772191144e-05,
"loss": 0.1949,
"step": 5740
},
{
"epoch": 0.67,
"grad_norm": 5.457193425471358,
"learning_rate": 1.8377911299405163e-05,
"loss": 0.2171,
"step": 5760
},
{
"epoch": 0.68,
"grad_norm": 15.65983957649386,
"learning_rate": 1.8363799128617697e-05,
"loss": 0.2076,
"step": 5780
},
{
"epoch": 0.68,
"grad_norm": 5.987492877103834,
"learning_rate": 1.8349631303465426e-05,
"loss": 0.2199,
"step": 5800
},
{
"epoch": 0.68,
"grad_norm": 6.583898334965102,
"learning_rate": 1.8335407918235116e-05,
"loss": 0.1861,
"step": 5820
},
{
"epoch": 0.68,
"grad_norm": 5.51874555542228,
"learning_rate": 1.832112906758329e-05,
"loss": 0.2392,
"step": 5840
},
{
"epoch": 0.69,
"grad_norm": 3.579186427839362,
"learning_rate": 1.83067948465356e-05,
"loss": 0.2204,
"step": 5860
},
{
"epoch": 0.69,
"grad_norm": 3.519966665890388,
"learning_rate": 1.8292405350486165e-05,
"loss": 0.2151,
"step": 5880
},
{
"epoch": 0.69,
"grad_norm": 4.244150062547031,
"learning_rate": 1.8277960675196983e-05,
"loss": 0.2297,
"step": 5900
},
{
"epoch": 0.69,
"grad_norm": 5.344839063171692,
"learning_rate": 1.826346091679725e-05,
"loss": 0.2245,
"step": 5920
},
{
"epoch": 0.7,
"grad_norm": 5.0436570623407775,
"learning_rate": 1.8248906171782755e-05,
"loss": 0.2378,
"step": 5940
},
{
"epoch": 0.7,
"grad_norm": 7.185397163940519,
"learning_rate": 1.8234296537015207e-05,
"loss": 0.2294,
"step": 5960
},
{
"epoch": 0.7,
"grad_norm": 4.591681189724662,
"learning_rate": 1.8219632109721624e-05,
"loss": 0.2256,
"step": 5980
},
{
"epoch": 0.7,
"grad_norm": 5.248378321548811,
"learning_rate": 1.8204912987493658e-05,
"loss": 0.2351,
"step": 6000
},
{
"epoch": 0.7,
"grad_norm": 3.251787867657218,
"learning_rate": 1.819013926828695e-05,
"loss": 0.1976,
"step": 6020
},
{
"epoch": 0.71,
"grad_norm": 3.355709045119318,
"learning_rate": 1.817531105042049e-05,
"loss": 0.1958,
"step": 6040
},
{
"epoch": 0.71,
"grad_norm": 2.8975140933466705,
"learning_rate": 1.8160428432575958e-05,
"loss": 0.1763,
"step": 6060
},
{
"epoch": 0.71,
"grad_norm": 7.07650583370445,
"learning_rate": 1.814549151379706e-05,
"loss": 0.1839,
"step": 6080
},
{
"epoch": 0.71,
"grad_norm": 2.3292924654795906,
"learning_rate": 1.8130500393488883e-05,
"loss": 0.1737,
"step": 6100
},
{
"epoch": 0.72,
"grad_norm": 3.6110322128545116,
"learning_rate": 1.811545517141722e-05,
"loss": 0.2161,
"step": 6120
},
{
"epoch": 0.72,
"grad_norm": 3.2080012723487368,
"learning_rate": 1.8100355947707896e-05,
"loss": 0.1741,
"step": 6140
},
{
"epoch": 0.72,
"grad_norm": 3.4674409337915906,
"learning_rate": 1.8085202822846145e-05,
"loss": 0.1983,
"step": 6160
},
{
"epoch": 0.72,
"grad_norm": 3.24481638760689,
"learning_rate": 1.8069995897675893e-05,
"loss": 0.1974,
"step": 6180
},
{
"epoch": 0.73,
"grad_norm": 2.1527007372890807,
"learning_rate": 1.8054735273399105e-05,
"loss": 0.1787,
"step": 6200
},
{
"epoch": 0.73,
"grad_norm": 3.8433408441880994,
"learning_rate": 1.8039421051575134e-05,
"loss": 0.2045,
"step": 6220
},
{
"epoch": 0.73,
"grad_norm": 3.1482360623951675,
"learning_rate": 1.802405333412e-05,
"loss": 0.1823,
"step": 6240
},
{
"epoch": 0.73,
"grad_norm": 3.638738970462984,
"learning_rate": 1.800863222330575e-05,
"loss": 0.163,
"step": 6260
},
{
"epoch": 0.73,
"grad_norm": 2.6939386819489473,
"learning_rate": 1.7993157821759756e-05,
"loss": 0.1506,
"step": 6280
},
{
"epoch": 0.74,
"grad_norm": 2.460255555625638,
"learning_rate": 1.7977630232464046e-05,
"loss": 0.2013,
"step": 6300
},
{
"epoch": 0.74,
"grad_norm": 3.419539678056426,
"learning_rate": 1.7962049558754607e-05,
"loss": 0.1746,
"step": 6320
},
{
"epoch": 0.74,
"grad_norm": 4.6827563888655295,
"learning_rate": 1.794641590432071e-05,
"loss": 0.1907,
"step": 6340
},
{
"epoch": 0.74,
"grad_norm": 3.156495792902755,
"learning_rate": 1.7930729373204197e-05,
"loss": 0.1932,
"step": 6360
},
{
"epoch": 0.75,
"grad_norm": 1.8329842047169478,
"learning_rate": 1.7914990069798825e-05,
"loss": 0.1839,
"step": 6380
},
{
"epoch": 0.75,
"grad_norm": 3.1302938253922066,
"learning_rate": 1.7899198098849532e-05,
"loss": 0.1958,
"step": 6400
},
{
"epoch": 0.75,
"grad_norm": 3.389701612050505,
"learning_rate": 1.788335356545176e-05,
"loss": 0.1915,
"step": 6420
},
{
"epoch": 0.75,
"grad_norm": 2.4493770274440805,
"learning_rate": 1.786745657505077e-05,
"loss": 0.1864,
"step": 6440
},
{
"epoch": 0.76,
"grad_norm": 3.7081248335498316,
"learning_rate": 1.7851507233440907e-05,
"loss": 0.1737,
"step": 6460
},
{
"epoch": 0.76,
"grad_norm": 2.9880342269224034,
"learning_rate": 1.7835505646764915e-05,
"loss": 0.1737,
"step": 6480
},
{
"epoch": 0.76,
"grad_norm": 3.243289769328672,
"learning_rate": 1.781945192151324e-05,
"loss": 0.1719,
"step": 6500
},
{
"epoch": 0.76,
"grad_norm": 3.4374399353328338,
"learning_rate": 1.7803346164523294e-05,
"loss": 0.1758,
"step": 6520
},
{
"epoch": 0.77,
"grad_norm": 2.8036052347770863,
"learning_rate": 1.778718848297877e-05,
"loss": 0.1782,
"step": 6540
},
{
"epoch": 0.77,
"grad_norm": 2.2879959069195395,
"learning_rate": 1.777097898440892e-05,
"loss": 0.191,
"step": 6560
},
{
"epoch": 0.77,
"grad_norm": 3.040554749166379,
"learning_rate": 1.7754717776687835e-05,
"loss": 0.1831,
"step": 6580
},
{
"epoch": 0.77,
"grad_norm": 3.071599714764158,
"learning_rate": 1.773840496803372e-05,
"loss": 0.185,
"step": 6600
},
{
"epoch": 0.77,
"grad_norm": 3.501269791372487,
"learning_rate": 1.7722040667008204e-05,
"loss": 0.1719,
"step": 6620
},
{
"epoch": 0.78,
"grad_norm": 4.16368161404422,
"learning_rate": 1.7705624982515573e-05,
"loss": 0.184,
"step": 6640
},
{
"epoch": 0.78,
"grad_norm": 2.7540351566960117,
"learning_rate": 1.7689158023802095e-05,
"loss": 0.1946,
"step": 6660
},
{
"epoch": 0.78,
"grad_norm": 3.2355776407579153,
"learning_rate": 1.7672639900455252e-05,
"loss": 0.1712,
"step": 6680
},
{
"epoch": 0.78,
"grad_norm": 2.6997084000974327,
"learning_rate": 1.765607072240303e-05,
"loss": 0.1784,
"step": 6700
},
{
"epoch": 0.79,
"grad_norm": 2.841998053177577,
"learning_rate": 1.763945059991318e-05,
"loss": 0.1742,
"step": 6720
},
{
"epoch": 0.79,
"grad_norm": 2.7453638357487713,
"learning_rate": 1.7622779643592492e-05,
"loss": 0.1944,
"step": 6740
},
{
"epoch": 0.79,
"grad_norm": 3.3176797639024938,
"learning_rate": 1.760605796438606e-05,
"loss": 0.1899,
"step": 6760
},
{
"epoch": 0.79,
"grad_norm": 2.366476424276501,
"learning_rate": 1.758928567357652e-05,
"loss": 0.1759,
"step": 6780
},
{
"epoch": 0.8,
"grad_norm": 3.7220494355163054,
"learning_rate": 1.7572462882783353e-05,
"loss": 0.1837,
"step": 6800
},
{
"epoch": 0.8,
"grad_norm": 3.038915792420121,
"learning_rate": 1.7555589703962098e-05,
"loss": 0.1646,
"step": 6820
},
{
"epoch": 0.8,
"grad_norm": 3.4617122704323244,
"learning_rate": 1.7538666249403633e-05,
"loss": 0.1758,
"step": 6840
},
{
"epoch": 0.8,
"grad_norm": 2.362854772501653,
"learning_rate": 1.752169263173342e-05,
"loss": 0.1634,
"step": 6860
},
{
"epoch": 0.81,
"grad_norm": 3.00174582863279,
"learning_rate": 1.7504668963910752e-05,
"loss": 0.1701,
"step": 6880
},
{
"epoch": 0.81,
"grad_norm": 3.2309255588354,
"learning_rate": 1.748759535922802e-05,
"loss": 0.1995,
"step": 6900
},
{
"epoch": 0.81,
"grad_norm": 2.662821998288258,
"learning_rate": 1.7470471931309922e-05,
"loss": 0.1672,
"step": 6920
},
{
"epoch": 0.81,
"grad_norm": 2.8772004041302623,
"learning_rate": 1.7453298794112747e-05,
"loss": 0.1898,
"step": 6940
},
{
"epoch": 0.81,
"grad_norm": 2.9237695763280365,
"learning_rate": 1.7436076061923592e-05,
"loss": 0.1762,
"step": 6960
},
{
"epoch": 0.82,
"grad_norm": 2.5504855909062036,
"learning_rate": 1.741880384935961e-05,
"loss": 0.1992,
"step": 6980
},
{
"epoch": 0.82,
"grad_norm": 7.919165394281912,
"learning_rate": 1.7401482271367242e-05,
"loss": 0.1692,
"step": 7000
},
{
"epoch": 0.82,
"grad_norm": 3.05371171825022,
"learning_rate": 1.7384111443221457e-05,
"loss": 0.1908,
"step": 7020
},
{
"epoch": 0.82,
"grad_norm": 2.621096842688656,
"learning_rate": 1.7366691480524992e-05,
"loss": 0.1685,
"step": 7040
},
{
"epoch": 0.83,
"grad_norm": 3.1230713042352862,
"learning_rate": 1.734922249920756e-05,
"loss": 0.1792,
"step": 7060
},
{
"epoch": 0.83,
"grad_norm": 3.104105869867001,
"learning_rate": 1.73317046155251e-05,
"loss": 0.1729,
"step": 7080
},
{
"epoch": 0.83,
"grad_norm": 3.1591947811656436,
"learning_rate": 1.7314137946059e-05,
"loss": 0.1718,
"step": 7100
},
{
"epoch": 0.83,
"grad_norm": 3.519623783770741,
"learning_rate": 1.7296522607715304e-05,
"loss": 0.1764,
"step": 7120
},
{
"epoch": 0.84,
"grad_norm": 2.915870412492841,
"learning_rate": 1.7278858717723962e-05,
"loss": 0.1558,
"step": 7140
},
{
"epoch": 0.84,
"grad_norm": 2.3279234523707295,
"learning_rate": 1.726114639363803e-05,
"loss": 0.1913,
"step": 7160
},
{
"epoch": 0.84,
"grad_norm": 2.6716434311486,
"learning_rate": 1.724338575333288e-05,
"loss": 0.1745,
"step": 7180
},
{
"epoch": 0.84,
"grad_norm": 2.878117736740971,
"learning_rate": 1.7225576915005445e-05,
"loss": 0.1727,
"step": 7200
},
{
"epoch": 0.84,
"grad_norm": 2.7046376665043277,
"learning_rate": 1.720771999717341e-05,
"loss": 0.1772,
"step": 7220
},
{
"epoch": 0.85,
"grad_norm": 3.194537005719014,
"learning_rate": 1.7189815118674426e-05,
"loss": 0.1796,
"step": 7240
},
{
"epoch": 0.85,
"grad_norm": 3.463727321763896,
"learning_rate": 1.7171862398665327e-05,
"loss": 0.1919,
"step": 7260
},
{
"epoch": 0.85,
"grad_norm": 3.13725471905633,
"learning_rate": 1.7153861956621326e-05,
"loss": 0.1811,
"step": 7280
},
{
"epoch": 0.85,
"grad_norm": 4.544353611874395,
"learning_rate": 1.7135813912335225e-05,
"loss": 0.1692,
"step": 7300
},
{
"epoch": 0.86,
"grad_norm": 2.1203551297270153,
"learning_rate": 1.7117718385916627e-05,
"loss": 0.1794,
"step": 7320
},
{
"epoch": 0.86,
"grad_norm": 2.8114219294750917,
"learning_rate": 1.709957549779112e-05,
"loss": 0.1568,
"step": 7340
},
{
"epoch": 0.86,
"grad_norm": 2.772982105960882,
"learning_rate": 1.7081385368699487e-05,
"loss": 0.1731,
"step": 7360
},
{
"epoch": 0.86,
"grad_norm": 2.169539808531856,
"learning_rate": 1.7063148119696903e-05,
"loss": 0.196,
"step": 7380
},
{
"epoch": 0.87,
"grad_norm": 3.269632171568159,
"learning_rate": 1.704486387215212e-05,
"loss": 0.1733,
"step": 7400
},
{
"epoch": 0.87,
"grad_norm": 2.304673017036061,
"learning_rate": 1.7026532747746664e-05,
"loss": 0.1752,
"step": 7420
},
{
"epoch": 0.87,
"grad_norm": 3.871844945364525,
"learning_rate": 1.700815486847403e-05,
"loss": 0.1614,
"step": 7440
},
{
"epoch": 0.87,
"grad_norm": 3.4589296462620522,
"learning_rate": 1.698973035663887e-05,
"loss": 0.1795,
"step": 7460
},
{
"epoch": 0.88,
"grad_norm": 2.449963810711825,
"learning_rate": 1.6971259334856162e-05,
"loss": 0.1607,
"step": 7480
},
{
"epoch": 0.88,
"grad_norm": 3.769043316663817,
"learning_rate": 1.695274192605042e-05,
"loss": 0.1649,
"step": 7500
},
{
"epoch": 0.88,
"grad_norm": 3.1777138173025103,
"learning_rate": 1.6934178253454868e-05,
"loss": 0.1579,
"step": 7520
},
{
"epoch": 0.88,
"grad_norm": 2.55585511382277,
"learning_rate": 1.691556844061059e-05,
"loss": 0.1483,
"step": 7540
},
{
"epoch": 0.88,
"grad_norm": 2.9610046540146118,
"learning_rate": 1.6896912611365762e-05,
"loss": 0.1671,
"step": 7560
},
{
"epoch": 0.89,
"grad_norm": 2.916410998083297,
"learning_rate": 1.687821088987478e-05,
"loss": 0.1715,
"step": 7580
},
{
"epoch": 0.89,
"grad_norm": 2.5312840023818173,
"learning_rate": 1.6859463400597465e-05,
"loss": 0.1715,
"step": 7600
},
{
"epoch": 0.89,
"grad_norm": 3.5116131080293282,
"learning_rate": 1.684067026829821e-05,
"loss": 0.1579,
"step": 7620
},
{
"epoch": 0.89,
"grad_norm": 2.5798326789951567,
"learning_rate": 1.6821831618045166e-05,
"loss": 0.1713,
"step": 7640
},
{
"epoch": 0.9,
"grad_norm": 2.7929950419435565,
"learning_rate": 1.6802947575209415e-05,
"loss": 0.173,
"step": 7660
},
{
"epoch": 0.9,
"grad_norm": 3.2236373979916646,
"learning_rate": 1.6784018265464113e-05,
"loss": 0.1808,
"step": 7680
},
{
"epoch": 0.9,
"grad_norm": 3.1749239394321243,
"learning_rate": 1.6765043814783676e-05,
"loss": 0.1759,
"step": 7700
},
{
"epoch": 0.9,
"grad_norm": 2.032005169681671,
"learning_rate": 1.674602434944293e-05,
"loss": 0.1814,
"step": 7720
},
{
"epoch": 0.91,
"grad_norm": 3.2976168841859366,
"learning_rate": 1.6726959996016273e-05,
"loss": 0.1487,
"step": 7740
},
{
"epoch": 0.91,
"grad_norm": 2.3455040026914995,
"learning_rate": 1.6707850881376823e-05,
"loss": 0.1686,
"step": 7760
},
{
"epoch": 0.91,
"grad_norm": 3.271968976468327,
"learning_rate": 1.668869713269561e-05,
"loss": 0.2027,
"step": 7780
},
{
"epoch": 0.91,
"grad_norm": 3.220964133175747,
"learning_rate": 1.666949887744067e-05,
"loss": 0.1694,
"step": 7800
},
{
"epoch": 0.92,
"grad_norm": 3.5721944777889676,
"learning_rate": 1.6650256243376257e-05,
"loss": 0.1908,
"step": 7820
},
{
"epoch": 0.92,
"grad_norm": 2.3813766405767627,
"learning_rate": 1.6630969358561948e-05,
"loss": 0.1824,
"step": 7840
},
{
"epoch": 0.92,
"grad_norm": 3.129649556197066,
"learning_rate": 1.661163835135182e-05,
"loss": 0.1787,
"step": 7860
},
{
"epoch": 0.92,
"grad_norm": 3.415549227335631,
"learning_rate": 1.659226335039358e-05,
"loss": 0.1825,
"step": 7880
},
{
"epoch": 0.92,
"grad_norm": 2.3764217538181778,
"learning_rate": 1.6572844484627707e-05,
"loss": 0.1462,
"step": 7900
},
{
"epoch": 0.93,
"grad_norm": 2.9671229672357287,
"learning_rate": 1.6553381883286612e-05,
"loss": 0.1914,
"step": 7920
},
{
"epoch": 0.93,
"grad_norm": 3.060295245568844,
"learning_rate": 1.6533875675893755e-05,
"loss": 0.181,
"step": 7940
},
{
"epoch": 0.93,
"grad_norm": 3.7842972511980086,
"learning_rate": 1.6514325992262798e-05,
"loss": 0.1507,
"step": 7960
},
{
"epoch": 0.93,
"grad_norm": 2.8230206703837326,
"learning_rate": 1.6494732962496735e-05,
"loss": 0.1736,
"step": 7980
},
{
"epoch": 0.94,
"grad_norm": 3.453554704952226,
"learning_rate": 1.6475096716987044e-05,
"loss": 0.1765,
"step": 8000
},
{
"epoch": 0.94,
"grad_norm": 2.9060450655966794,
"learning_rate": 1.6455417386412776e-05,
"loss": 0.1755,
"step": 8020
},
{
"epoch": 0.94,
"grad_norm": 2.54357724739597,
"learning_rate": 1.6435695101739728e-05,
"loss": 0.178,
"step": 8040
},
{
"epoch": 0.94,
"grad_norm": 2.2841639263364417,
"learning_rate": 1.641592999421956e-05,
"loss": 0.1671,
"step": 8060
},
{
"epoch": 0.95,
"grad_norm": 2.922479250844846,
"learning_rate": 1.639612219538891e-05,
"loss": 0.1722,
"step": 8080
},
{
"epoch": 0.95,
"grad_norm": 3.364819041086794,
"learning_rate": 1.6376271837068535e-05,
"loss": 0.1743,
"step": 8100
},
{
"epoch": 0.95,
"grad_norm": 3.2343077721726976,
"learning_rate": 1.6356379051362408e-05,
"loss": 0.1754,
"step": 8120
},
{
"epoch": 0.95,
"grad_norm": 3.988702555642947,
"learning_rate": 1.6336443970656877e-05,
"loss": 0.1719,
"step": 8140
},
{
"epoch": 0.95,
"grad_norm": 3.586344112040257,
"learning_rate": 1.6316466727619746e-05,
"loss": 0.1727,
"step": 8160
},
{
"epoch": 0.96,
"grad_norm": 3.49168839611008,
"learning_rate": 1.629644745519942e-05,
"loss": 0.1812,
"step": 8180
},
{
"epoch": 0.96,
"grad_norm": 3.103292218300282,
"learning_rate": 1.6276386286624007e-05,
"loss": 0.1703,
"step": 8200
},
{
"epoch": 0.96,
"grad_norm": 3.5623729750088957,
"learning_rate": 1.6256283355400428e-05,
"loss": 0.1873,
"step": 8220
},
{
"epoch": 0.96,
"grad_norm": 2.295795628214393,
"learning_rate": 1.6236138795313535e-05,
"loss": 0.1807,
"step": 8240
},
{
"epoch": 0.97,
"grad_norm": 3.365176421212747,
"learning_rate": 1.6215952740425232e-05,
"loss": 0.1433,
"step": 8260
},
{
"epoch": 0.97,
"grad_norm": 3.3396715621732143,
"learning_rate": 1.6195725325073552e-05,
"loss": 0.1686,
"step": 8280
},
{
"epoch": 0.97,
"grad_norm": 4.290597554096145,
"learning_rate": 1.6175456683871797e-05,
"loss": 0.1687,
"step": 8300
},
{
"epoch": 0.97,
"grad_norm": 2.5200221330185415,
"learning_rate": 1.6155146951707615e-05,
"loss": 0.1562,
"step": 8320
},
{
"epoch": 0.98,
"grad_norm": 4.318015952324211,
"learning_rate": 1.6134796263742125e-05,
"loss": 0.1731,
"step": 8340
},
{
"epoch": 0.98,
"grad_norm": 2.503160957906815,
"learning_rate": 1.6114404755409e-05,
"loss": 0.1592,
"step": 8360
},
{
"epoch": 0.98,
"grad_norm": 2.9078381208780253,
"learning_rate": 1.6093972562413572e-05,
"loss": 0.1664,
"step": 8380
},
{
"epoch": 0.98,
"grad_norm": 2.480511330066373,
"learning_rate": 1.607349982073193e-05,
"loss": 0.181,
"step": 8400
},
{
"epoch": 0.99,
"grad_norm": 2.548494912067414,
"learning_rate": 1.6052986666610015e-05,
"loss": 0.1612,
"step": 8420
},
{
"epoch": 0.99,
"grad_norm": 2.9336394244218176,
"learning_rate": 1.6032433236562715e-05,
"loss": 0.1597,
"step": 8440
},
{
"epoch": 0.99,
"grad_norm": 3.7724275678856776,
"learning_rate": 1.6011839667372948e-05,
"loss": 0.1782,
"step": 8460
},
{
"epoch": 0.99,
"grad_norm": 4.105531004225903,
"learning_rate": 1.599120609609076e-05,
"loss": 0.1746,
"step": 8480
},
{
"epoch": 0.99,
"grad_norm": 2.6233575314573505,
"learning_rate": 1.5970532660032413e-05,
"loss": 0.1654,
"step": 8500
},
{
"epoch": 1.0,
"grad_norm": 2.71752665389956,
"learning_rate": 1.5949819496779467e-05,
"loss": 0.1866,
"step": 8520
},
{
"epoch": 1.0,
"grad_norm": 3.052677830070527,
"learning_rate": 1.592906674417787e-05,
"loss": 0.1705,
"step": 8540
}
],
"logging_steps": 20,
"max_steps": 25638,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}