| { |
| "best_global_step": 1000, |
| "best_metric": 0.9389243125915527, |
| "best_model_checkpoint": "/project2/robinjia_875/chuang95/hyper_search/TLS/llama/Llama-rank:4-layerR:(0, 27)-includeGolden:False-maskNeutral:False-modules:all-iteration:0-lr:0.0001/checkpoint-1000", |
| "epoch": 1.0, |
| "eval_steps": 200, |
| "global_step": 1391, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0007189072609633358, |
| "grad_norm": 13.526103019714355, |
| "learning_rate": 0.0001, |
| "loss": 0.9279, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0014378145219266715, |
| "grad_norm": 9.46270751953125, |
| "learning_rate": 9.992810927390367e-05, |
| "loss": 0.8179, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.002156721782890007, |
| "grad_norm": 6.017407417297363, |
| "learning_rate": 9.985621854780733e-05, |
| "loss": 0.8241, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.002875629043853343, |
| "grad_norm": 5.485882759094238, |
| "learning_rate": 9.9784327821711e-05, |
| "loss": 0.7669, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0035945363048166786, |
| "grad_norm": 5.696507930755615, |
| "learning_rate": 9.971243709561468e-05, |
| "loss": 0.593, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.004313443565780014, |
| "grad_norm": 7.812379837036133, |
| "learning_rate": 9.964054636951834e-05, |
| "loss": 0.7126, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.0050323508267433505, |
| "grad_norm": 6.563722133636475, |
| "learning_rate": 9.9568655643422e-05, |
| "loss": 0.6582, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.005751258087706686, |
| "grad_norm": 5.59401273727417, |
| "learning_rate": 9.949676491732567e-05, |
| "loss": 0.6228, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.006470165348670022, |
| "grad_norm": 4.154993057250977, |
| "learning_rate": 9.942487419122934e-05, |
| "loss": 0.5712, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.007189072609633357, |
| "grad_norm": 3.4424068927764893, |
| "learning_rate": 9.9352983465133e-05, |
| "loss": 0.5496, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.007907979870596693, |
| "grad_norm": 2.921372413635254, |
| "learning_rate": 9.928109273903668e-05, |
| "loss": 0.5147, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.008626887131560028, |
| "grad_norm": 3.346285104751587, |
| "learning_rate": 9.920920201294034e-05, |
| "loss": 0.5317, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.009345794392523364, |
| "grad_norm": 2.7753374576568604, |
| "learning_rate": 9.913731128684401e-05, |
| "loss": 0.5164, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.010064701653486701, |
| "grad_norm": 2.2643685340881348, |
| "learning_rate": 9.906542056074767e-05, |
| "loss": 0.4489, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.010783608914450037, |
| "grad_norm": 2.5159664154052734, |
| "learning_rate": 9.899352983465134e-05, |
| "loss": 0.4372, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.011502516175413372, |
| "grad_norm": 2.186605453491211, |
| "learning_rate": 9.8921639108555e-05, |
| "loss": 0.3904, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.012221423436376708, |
| "grad_norm": 2.0177955627441406, |
| "learning_rate": 9.884974838245867e-05, |
| "loss": 0.3503, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.012940330697340043, |
| "grad_norm": 3.426478862762451, |
| "learning_rate": 9.877785765636233e-05, |
| "loss": 0.4509, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.013659237958303379, |
| "grad_norm": 2.4085726737976074, |
| "learning_rate": 9.8705966930266e-05, |
| "loss": 0.3563, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.014378145219266714, |
| "grad_norm": 1.7231711149215698, |
| "learning_rate": 9.863407620416966e-05, |
| "loss": 0.3661, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.01509705248023005, |
| "grad_norm": 2.3881592750549316, |
| "learning_rate": 9.856218547807333e-05, |
| "loss": 0.3481, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.015815959741193385, |
| "grad_norm": 1.1663906574249268, |
| "learning_rate": 9.849029475197699e-05, |
| "loss": 0.2612, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.01653486700215672, |
| "grad_norm": 1.8138469457626343, |
| "learning_rate": 9.841840402588066e-05, |
| "loss": 0.3572, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.017253774263120056, |
| "grad_norm": 2.513244152069092, |
| "learning_rate": 9.834651329978433e-05, |
| "loss": 0.3389, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.017972681524083392, |
| "grad_norm": 4.380908966064453, |
| "learning_rate": 9.8274622573688e-05, |
| "loss": 0.3061, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.018691588785046728, |
| "grad_norm": 1.3064526319503784, |
| "learning_rate": 9.820273184759166e-05, |
| "loss": 0.2627, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.019410496046010063, |
| "grad_norm": 1.2855287790298462, |
| "learning_rate": 9.813084112149533e-05, |
| "loss": 0.2732, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.020129403306973402, |
| "grad_norm": 1.6424367427825928, |
| "learning_rate": 9.805895039539899e-05, |
| "loss": 0.2681, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.020848310567936738, |
| "grad_norm": 1.0976167917251587, |
| "learning_rate": 9.798705966930266e-05, |
| "loss": 0.1806, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.021567217828900073, |
| "grad_norm": 1.1664091348648071, |
| "learning_rate": 9.791516894320632e-05, |
| "loss": 0.2007, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.02228612508986341, |
| "grad_norm": 1.5214357376098633, |
| "learning_rate": 9.784327821711e-05, |
| "loss": 0.277, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.023005032350826744, |
| "grad_norm": 1.171317458152771, |
| "learning_rate": 9.777138749101366e-05, |
| "loss": 0.1947, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.02372393961179008, |
| "grad_norm": 1.6442270278930664, |
| "learning_rate": 9.769949676491733e-05, |
| "loss": 0.2636, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.024442846872753415, |
| "grad_norm": 3.0155868530273438, |
| "learning_rate": 9.7627606038821e-05, |
| "loss": 0.4081, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.02516175413371675, |
| "grad_norm": 1.3478777408599854, |
| "learning_rate": 9.755571531272466e-05, |
| "loss": 0.2287, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.025880661394680086, |
| "grad_norm": 1.3787730932235718, |
| "learning_rate": 9.748382458662832e-05, |
| "loss": 0.1657, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.026599568655643422, |
| "grad_norm": 2.977001190185547, |
| "learning_rate": 9.7411933860532e-05, |
| "loss": 0.2048, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.027318475916606758, |
| "grad_norm": 1.2369728088378906, |
| "learning_rate": 9.734004313443567e-05, |
| "loss": 0.1736, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.028037383177570093, |
| "grad_norm": 3.8605053424835205, |
| "learning_rate": 9.726815240833933e-05, |
| "loss": 0.3035, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.02875629043853343, |
| "grad_norm": 1.1683523654937744, |
| "learning_rate": 9.7196261682243e-05, |
| "loss": 0.2056, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.029475197699496764, |
| "grad_norm": 1.7133461236953735, |
| "learning_rate": 9.712437095614666e-05, |
| "loss": 0.2429, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.0301941049604601, |
| "grad_norm": 1.103882908821106, |
| "learning_rate": 9.705248023005032e-05, |
| "loss": 0.145, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.030913012221423435, |
| "grad_norm": 2.404679775238037, |
| "learning_rate": 9.698058950395399e-05, |
| "loss": 0.1862, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.03163191948238677, |
| "grad_norm": 1.614344835281372, |
| "learning_rate": 9.690869877785767e-05, |
| "loss": 0.2224, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.03235082674335011, |
| "grad_norm": 4.089974880218506, |
| "learning_rate": 9.683680805176133e-05, |
| "loss": 0.3103, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.03306973400431344, |
| "grad_norm": 4.1694416999816895, |
| "learning_rate": 9.6764917325665e-05, |
| "loss": 0.3185, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.03378864126527678, |
| "grad_norm": 4.718451976776123, |
| "learning_rate": 9.669302659956866e-05, |
| "loss": 0.3463, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.03450754852624011, |
| "grad_norm": 1.8312580585479736, |
| "learning_rate": 9.662113587347233e-05, |
| "loss": 0.2105, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.03522645578720345, |
| "grad_norm": 1.377999186515808, |
| "learning_rate": 9.654924514737599e-05, |
| "loss": 0.2113, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.035945363048166784, |
| "grad_norm": 1.6277997493743896, |
| "learning_rate": 9.647735442127967e-05, |
| "loss": 0.236, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.03666427030913012, |
| "grad_norm": 4.6866631507873535, |
| "learning_rate": 9.640546369518333e-05, |
| "loss": 0.1771, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.037383177570093455, |
| "grad_norm": 5.198335647583008, |
| "learning_rate": 9.6333572969087e-05, |
| "loss": 0.2932, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.038102084831056794, |
| "grad_norm": 4.452822685241699, |
| "learning_rate": 9.626168224299066e-05, |
| "loss": 0.2381, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.038820992092020126, |
| "grad_norm": 1.527959942817688, |
| "learning_rate": 9.618979151689433e-05, |
| "loss": 0.1416, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.039539899352983465, |
| "grad_norm": 1.2298065423965454, |
| "learning_rate": 9.611790079079799e-05, |
| "loss": 0.1902, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.040258806613946804, |
| "grad_norm": 2.105191946029663, |
| "learning_rate": 9.604601006470166e-05, |
| "loss": 0.3549, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.040977713874910136, |
| "grad_norm": 2.3031511306762695, |
| "learning_rate": 9.597411933860532e-05, |
| "loss": 0.2307, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.041696621135873475, |
| "grad_norm": 1.0506740808486938, |
| "learning_rate": 9.590222861250899e-05, |
| "loss": 0.1595, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.04241552839683681, |
| "grad_norm": 1.4700311422348022, |
| "learning_rate": 9.583033788641265e-05, |
| "loss": 0.2773, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.043134435657800146, |
| "grad_norm": 1.4296544790267944, |
| "learning_rate": 9.575844716031632e-05, |
| "loss": 0.198, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.04385334291876348, |
| "grad_norm": 1.3201313018798828, |
| "learning_rate": 9.568655643421998e-05, |
| "loss": 0.1818, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.04457225017972682, |
| "grad_norm": 0.8356322646141052, |
| "learning_rate": 9.561466570812365e-05, |
| "loss": 0.1723, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.04529115744069015, |
| "grad_norm": 2.923663854598999, |
| "learning_rate": 9.554277498202732e-05, |
| "loss": 0.341, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.04601006470165349, |
| "grad_norm": 0.9205687046051025, |
| "learning_rate": 9.547088425593099e-05, |
| "loss": 0.1479, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.04672897196261682, |
| "grad_norm": 2.386075258255005, |
| "learning_rate": 9.539899352983465e-05, |
| "loss": 0.1776, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.04744787922358016, |
| "grad_norm": 1.443748950958252, |
| "learning_rate": 9.532710280373832e-05, |
| "loss": 0.1654, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.04816678648454349, |
| "grad_norm": 0.9766165614128113, |
| "learning_rate": 9.525521207764198e-05, |
| "loss": 0.1655, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.04888569374550683, |
| "grad_norm": 1.0120283365249634, |
| "learning_rate": 9.518332135154565e-05, |
| "loss": 0.1878, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.04960460100647016, |
| "grad_norm": 2.3093202114105225, |
| "learning_rate": 9.511143062544933e-05, |
| "loss": 0.1783, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.0503235082674335, |
| "grad_norm": 1.5764620304107666, |
| "learning_rate": 9.503953989935299e-05, |
| "loss": 0.239, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.051042415528396834, |
| "grad_norm": 1.97128427028656, |
| "learning_rate": 9.496764917325666e-05, |
| "loss": 0.2963, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.05176132278936017, |
| "grad_norm": 0.9656399488449097, |
| "learning_rate": 9.489575844716032e-05, |
| "loss": 0.0934, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.052480230050323505, |
| "grad_norm": 1.0600707530975342, |
| "learning_rate": 9.482386772106398e-05, |
| "loss": 0.1884, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.053199137311286844, |
| "grad_norm": 1.6577003002166748, |
| "learning_rate": 9.475197699496765e-05, |
| "loss": 0.3079, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.05391804457225018, |
| "grad_norm": 1.2079652547836304, |
| "learning_rate": 9.468008626887131e-05, |
| "loss": 0.1873, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.054636951833213515, |
| "grad_norm": 0.8267244100570679, |
| "learning_rate": 9.460819554277499e-05, |
| "loss": 0.1605, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.055355859094176854, |
| "grad_norm": 2.803229570388794, |
| "learning_rate": 9.453630481667866e-05, |
| "loss": 0.2788, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.056074766355140186, |
| "grad_norm": 1.5047197341918945, |
| "learning_rate": 9.446441409058232e-05, |
| "loss": 0.2684, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.056793673616103525, |
| "grad_norm": 0.8882524371147156, |
| "learning_rate": 9.439252336448599e-05, |
| "loss": 0.1438, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.05751258087706686, |
| "grad_norm": 0.8985353708267212, |
| "learning_rate": 9.432063263838965e-05, |
| "loss": 0.1427, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.058231488138030196, |
| "grad_norm": 2.100515604019165, |
| "learning_rate": 9.424874191229332e-05, |
| "loss": 0.2139, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.05895039539899353, |
| "grad_norm": 1.758117437362671, |
| "learning_rate": 9.4176851186197e-05, |
| "loss": 0.1958, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.05966930265995687, |
| "grad_norm": 2.295666456222534, |
| "learning_rate": 9.410496046010066e-05, |
| "loss": 0.3502, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.0603882099209202, |
| "grad_norm": 1.415231704711914, |
| "learning_rate": 9.403306973400432e-05, |
| "loss": 0.2319, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.06110711718188354, |
| "grad_norm": 0.877133846282959, |
| "learning_rate": 9.396117900790799e-05, |
| "loss": 0.1646, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.06182602444284687, |
| "grad_norm": 0.8490685224533081, |
| "learning_rate": 9.388928828181165e-05, |
| "loss": 0.1625, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.0625449317038102, |
| "grad_norm": 1.605905532836914, |
| "learning_rate": 9.381739755571532e-05, |
| "loss": 0.2371, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.06326383896477354, |
| "grad_norm": 1.5608909130096436, |
| "learning_rate": 9.374550682961898e-05, |
| "loss": 0.1792, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.06398274622573688, |
| "grad_norm": 1.5868877172470093, |
| "learning_rate": 9.367361610352266e-05, |
| "loss": 0.2214, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.06470165348670022, |
| "grad_norm": 2.048278331756592, |
| "learning_rate": 9.360172537742632e-05, |
| "loss": 0.1878, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.06542056074766354, |
| "grad_norm": 2.5156164169311523, |
| "learning_rate": 9.352983465132999e-05, |
| "loss": 0.204, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.06613946800862688, |
| "grad_norm": 1.3117547035217285, |
| "learning_rate": 9.345794392523365e-05, |
| "loss": 0.1995, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.06685837526959022, |
| "grad_norm": 1.7099689245224, |
| "learning_rate": 9.338605319913732e-05, |
| "loss": 0.2499, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.06757728253055356, |
| "grad_norm": 0.970403254032135, |
| "learning_rate": 9.331416247304098e-05, |
| "loss": 0.1899, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.0682961897915169, |
| "grad_norm": 5.142948627471924, |
| "learning_rate": 9.324227174694465e-05, |
| "loss": 0.436, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.06901509705248023, |
| "grad_norm": 1.407037615776062, |
| "learning_rate": 9.317038102084831e-05, |
| "loss": 0.1908, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.06973400431344356, |
| "grad_norm": 0.7759674191474915, |
| "learning_rate": 9.309849029475198e-05, |
| "loss": 0.1788, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.0704529115744069, |
| "grad_norm": 1.683898687362671, |
| "learning_rate": 9.302659956865564e-05, |
| "loss": 0.142, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.07117181883537024, |
| "grad_norm": 0.765933632850647, |
| "learning_rate": 9.295470884255931e-05, |
| "loss": 0.1443, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.07189072609633357, |
| "grad_norm": 2.1621007919311523, |
| "learning_rate": 9.288281811646297e-05, |
| "loss": 0.1286, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.07260963335729691, |
| "grad_norm": 1.0329022407531738, |
| "learning_rate": 9.281092739036664e-05, |
| "loss": 0.1943, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.07332854061826025, |
| "grad_norm": 2.71134352684021, |
| "learning_rate": 9.273903666427031e-05, |
| "loss": 0.1554, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.07404744787922359, |
| "grad_norm": 0.7791102528572083, |
| "learning_rate": 9.266714593817398e-05, |
| "loss": 0.1184, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.07476635514018691, |
| "grad_norm": 2.5555965900421143, |
| "learning_rate": 9.259525521207764e-05, |
| "loss": 0.2266, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.07548526240115025, |
| "grad_norm": 0.8810281753540039, |
| "learning_rate": 9.252336448598131e-05, |
| "loss": 0.1211, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.07620416966211359, |
| "grad_norm": 1.2278245687484741, |
| "learning_rate": 9.245147375988497e-05, |
| "loss": 0.2086, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.07692307692307693, |
| "grad_norm": 0.9289319515228271, |
| "learning_rate": 9.237958303378864e-05, |
| "loss": 0.1416, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.07764198418404025, |
| "grad_norm": 2.2514514923095703, |
| "learning_rate": 9.230769230769232e-05, |
| "loss": 0.2697, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.07836089144500359, |
| "grad_norm": 0.9639162421226501, |
| "learning_rate": 9.223580158159598e-05, |
| "loss": 0.1499, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.07907979870596693, |
| "grad_norm": 1.9107799530029297, |
| "learning_rate": 9.216391085549965e-05, |
| "loss": 0.3683, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.07979870596693027, |
| "grad_norm": 2.26277232170105, |
| "learning_rate": 9.209202012940331e-05, |
| "loss": 0.1314, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.08051761322789361, |
| "grad_norm": 1.6424294710159302, |
| "learning_rate": 9.202012940330698e-05, |
| "loss": 0.1381, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.08123652048885693, |
| "grad_norm": 1.5212277173995972, |
| "learning_rate": 9.194823867721064e-05, |
| "loss": 0.2319, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.08195542774982027, |
| "grad_norm": 1.9586520195007324, |
| "learning_rate": 9.18763479511143e-05, |
| "loss": 0.2182, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.08267433501078361, |
| "grad_norm": 2.01633620262146, |
| "learning_rate": 9.180445722501798e-05, |
| "loss": 0.2325, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.08339324227174695, |
| "grad_norm": 1.0720397233963013, |
| "learning_rate": 9.173256649892165e-05, |
| "loss": 0.1907, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.08411214953271028, |
| "grad_norm": 0.8959536552429199, |
| "learning_rate": 9.166067577282531e-05, |
| "loss": 0.1561, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.08483105679367361, |
| "grad_norm": 1.5228979587554932, |
| "learning_rate": 9.158878504672898e-05, |
| "loss": 0.1872, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.08554996405463695, |
| "grad_norm": 1.0474015474319458, |
| "learning_rate": 9.151689432063264e-05, |
| "loss": 0.1422, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.08626887131560029, |
| "grad_norm": 1.6714234352111816, |
| "learning_rate": 9.14450035945363e-05, |
| "loss": 0.163, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.08698777857656362, |
| "grad_norm": 1.194766879081726, |
| "learning_rate": 9.137311286843998e-05, |
| "loss": 0.1749, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.08770668583752696, |
| "grad_norm": 1.9764833450317383, |
| "learning_rate": 9.130122214234365e-05, |
| "loss": 0.1317, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.0884255930984903, |
| "grad_norm": 2.3686954975128174, |
| "learning_rate": 9.122933141624731e-05, |
| "loss": 0.1776, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.08914450035945364, |
| "grad_norm": 2.600713014602661, |
| "learning_rate": 9.115744069015098e-05, |
| "loss": 0.1385, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.08986340762041696, |
| "grad_norm": 0.7720825672149658, |
| "learning_rate": 9.108554996405464e-05, |
| "loss": 0.1393, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.0905823148813803, |
| "grad_norm": 1.3123085498809814, |
| "learning_rate": 9.101365923795831e-05, |
| "loss": 0.1949, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.09130122214234364, |
| "grad_norm": 1.2883367538452148, |
| "learning_rate": 9.094176851186197e-05, |
| "loss": 0.1568, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.09202012940330698, |
| "grad_norm": 0.7956113815307617, |
| "learning_rate": 9.086987778576565e-05, |
| "loss": 0.1218, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.0927390366642703, |
| "grad_norm": 0.9514375329017639, |
| "learning_rate": 9.079798705966932e-05, |
| "loss": 0.1173, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.09345794392523364, |
| "grad_norm": 0.5750173926353455, |
| "learning_rate": 9.072609633357298e-05, |
| "loss": 0.1188, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.09417685118619698, |
| "grad_norm": 1.8377829790115356, |
| "learning_rate": 9.065420560747664e-05, |
| "loss": 0.1974, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.09489575844716032, |
| "grad_norm": 1.0993359088897705, |
| "learning_rate": 9.058231488138031e-05, |
| "loss": 0.1465, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.09561466570812366, |
| "grad_norm": 1.0178059339523315, |
| "learning_rate": 9.051042415528397e-05, |
| "loss": 0.1441, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.09633357296908698, |
| "grad_norm": 1.1786631345748901, |
| "learning_rate": 9.043853342918764e-05, |
| "loss": 0.1766, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.09705248023005032, |
| "grad_norm": 1.5105781555175781, |
| "learning_rate": 9.03666427030913e-05, |
| "loss": 0.1045, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.09777138749101366, |
| "grad_norm": 2.000307083129883, |
| "learning_rate": 9.029475197699497e-05, |
| "loss": 0.2065, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.098490294751977, |
| "grad_norm": 1.0840668678283691, |
| "learning_rate": 9.022286125089863e-05, |
| "loss": 0.1789, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.09920920201294033, |
| "grad_norm": 0.6337581872940063, |
| "learning_rate": 9.01509705248023e-05, |
| "loss": 0.1066, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.09992810927390366, |
| "grad_norm": 1.520429253578186, |
| "learning_rate": 9.007907979870596e-05, |
| "loss": 0.1904, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.100647016534867, |
| "grad_norm": 1.298966407775879, |
| "learning_rate": 9.000718907260963e-05, |
| "loss": 0.1729, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.10136592379583034, |
| "grad_norm": 0.7120006084442139, |
| "learning_rate": 8.99352983465133e-05, |
| "loss": 0.1189, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.10208483105679367, |
| "grad_norm": 2.07332444190979, |
| "learning_rate": 8.986340762041697e-05, |
| "loss": 0.2364, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.102803738317757, |
| "grad_norm": 1.0223716497421265, |
| "learning_rate": 8.979151689432063e-05, |
| "loss": 0.2145, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.10352264557872035, |
| "grad_norm": 2.1979446411132812, |
| "learning_rate": 8.97196261682243e-05, |
| "loss": 0.1874, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.10424155283968368, |
| "grad_norm": 0.6637475490570068, |
| "learning_rate": 8.964773544212796e-05, |
| "loss": 0.1253, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.10496046010064701, |
| "grad_norm": 1.6761113405227661, |
| "learning_rate": 8.957584471603163e-05, |
| "loss": 0.2649, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.10567936736161035, |
| "grad_norm": 4.166442394256592, |
| "learning_rate": 8.950395398993531e-05, |
| "loss": 0.2894, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.10639827462257369, |
| "grad_norm": 2.4235761165618896, |
| "learning_rate": 8.943206326383897e-05, |
| "loss": 0.1754, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.10711718188353703, |
| "grad_norm": 1.6823021173477173, |
| "learning_rate": 8.936017253774264e-05, |
| "loss": 0.114, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.10783608914450037, |
| "grad_norm": 0.8887488842010498, |
| "learning_rate": 8.92882818116463e-05, |
| "loss": 0.1974, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.10855499640546369, |
| "grad_norm": 0.7510696649551392, |
| "learning_rate": 8.921639108554997e-05, |
| "loss": 0.1662, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.10927390366642703, |
| "grad_norm": 3.53305983543396, |
| "learning_rate": 8.914450035945363e-05, |
| "loss": 0.1979, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.10999281092739037, |
| "grad_norm": 1.014398217201233, |
| "learning_rate": 8.90726096333573e-05, |
| "loss": 0.1453, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.11071171818835371, |
| "grad_norm": 2.1397502422332764, |
| "learning_rate": 8.900071890726097e-05, |
| "loss": 0.1531, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.11143062544931703, |
| "grad_norm": 0.8437989354133606, |
| "learning_rate": 8.892882818116464e-05, |
| "loss": 0.1404, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.11214953271028037, |
| "grad_norm": 1.4224001169204712, |
| "learning_rate": 8.88569374550683e-05, |
| "loss": 0.2316, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.11286843997124371, |
| "grad_norm": 0.9540480375289917, |
| "learning_rate": 8.878504672897197e-05, |
| "loss": 0.1113, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.11358734723220705, |
| "grad_norm": 1.2439966201782227, |
| "learning_rate": 8.871315600287563e-05, |
| "loss": 0.1491, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.11430625449317038, |
| "grad_norm": 0.7045279741287231, |
| "learning_rate": 8.86412652767793e-05, |
| "loss": 0.0709, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.11502516175413371, |
| "grad_norm": 2.2379260063171387, |
| "learning_rate": 8.856937455068297e-05, |
| "loss": 0.1878, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.11574406901509705, |
| "grad_norm": 1.0175137519836426, |
| "learning_rate": 8.849748382458664e-05, |
| "loss": 0.1495, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.11646297627606039, |
| "grad_norm": 0.7115600109100342, |
| "learning_rate": 8.84255930984903e-05, |
| "loss": 0.0864, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.11718188353702372, |
| "grad_norm": 0.9618721008300781, |
| "learning_rate": 8.835370237239397e-05, |
| "loss": 0.187, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.11790079079798706, |
| "grad_norm": 1.248335361480713, |
| "learning_rate": 8.828181164629763e-05, |
| "loss": 0.1845, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.1186196980589504, |
| "grad_norm": 2.709212064743042, |
| "learning_rate": 8.82099209202013e-05, |
| "loss": 0.287, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.11933860531991373, |
| "grad_norm": 1.1670459508895874, |
| "learning_rate": 8.813803019410498e-05, |
| "loss": 0.1465, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.12005751258087707, |
| "grad_norm": 0.9553388953208923, |
| "learning_rate": 8.806613946800864e-05, |
| "loss": 0.1718, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.1207764198418404, |
| "grad_norm": 0.9523774981498718, |
| "learning_rate": 8.79942487419123e-05, |
| "loss": 0.1573, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.12149532710280374, |
| "grad_norm": 0.9836131930351257, |
| "learning_rate": 8.792235801581597e-05, |
| "loss": 0.1456, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.12221423436376708, |
| "grad_norm": 1.6975351572036743, |
| "learning_rate": 8.785046728971964e-05, |
| "loss": 0.1844, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.12293314162473042, |
| "grad_norm": 1.543567419052124, |
| "learning_rate": 8.77785765636233e-05, |
| "loss": 0.1194, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.12365204888569374, |
| "grad_norm": 1.4463262557983398, |
| "learning_rate": 8.770668583752696e-05, |
| "loss": 0.2811, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.12437095614665708, |
| "grad_norm": 2.0609545707702637, |
| "learning_rate": 8.763479511143063e-05, |
| "loss": 0.1532, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.1250898634076204, |
| "grad_norm": 2.18935489654541, |
| "learning_rate": 8.75629043853343e-05, |
| "loss": 0.1788, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.12580877066858376, |
| "grad_norm": 0.794236421585083, |
| "learning_rate": 8.749101365923796e-05, |
| "loss": 0.1168, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.12652767792954708, |
| "grad_norm": 0.7721659541130066, |
| "learning_rate": 8.741912293314162e-05, |
| "loss": 0.1463, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.12724658519051044, |
| "grad_norm": 1.2313865423202515, |
| "learning_rate": 8.734723220704529e-05, |
| "loss": 0.18, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.12796549245147376, |
| "grad_norm": 0.9938852190971375, |
| "learning_rate": 8.727534148094895e-05, |
| "loss": 0.2109, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.1286843997124371, |
| "grad_norm": 1.9981725215911865, |
| "learning_rate": 8.720345075485263e-05, |
| "loss": 0.1886, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.12940330697340044, |
| "grad_norm": 0.8175665736198425, |
| "learning_rate": 8.71315600287563e-05, |
| "loss": 0.143, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.13012221423436376, |
| "grad_norm": 1.4282681941986084, |
| "learning_rate": 8.705966930265996e-05, |
| "loss": 0.144, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.1308411214953271, |
| "grad_norm": 1.064103126525879, |
| "learning_rate": 8.698777857656363e-05, |
| "loss": 0.2546, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.13156002875629044, |
| "grad_norm": 1.3310757875442505, |
| "learning_rate": 8.691588785046729e-05, |
| "loss": 0.2736, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.13227893601725377, |
| "grad_norm": 0.5591844320297241, |
| "learning_rate": 8.684399712437095e-05, |
| "loss": 0.1234, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.13299784327821712, |
| "grad_norm": 0.9298526644706726, |
| "learning_rate": 8.677210639827462e-05, |
| "loss": 0.1482, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.13371675053918045, |
| "grad_norm": 0.6556968092918396, |
| "learning_rate": 8.67002156721783e-05, |
| "loss": 0.1267, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.13443565780014377, |
| "grad_norm": 0.9258823990821838, |
| "learning_rate": 8.662832494608196e-05, |
| "loss": 0.1653, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.13515456506110712, |
| "grad_norm": 0.8357483744621277, |
| "learning_rate": 8.655643421998563e-05, |
| "loss": 0.1471, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.13587347232207045, |
| "grad_norm": 1.7659372091293335, |
| "learning_rate": 8.648454349388929e-05, |
| "loss": 0.1537, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.1365923795830338, |
| "grad_norm": 1.6333738565444946, |
| "learning_rate": 8.641265276779296e-05, |
| "loss": 0.2463, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.13731128684399713, |
| "grad_norm": 1.2712441682815552, |
| "learning_rate": 8.634076204169662e-05, |
| "loss": 0.2327, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.13803019410496045, |
| "grad_norm": 0.6965837478637695, |
| "learning_rate": 8.62688713156003e-05, |
| "loss": 0.1042, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.1387491013659238, |
| "grad_norm": 1.7786089181900024, |
| "learning_rate": 8.619698058950396e-05, |
| "loss": 0.1823, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.13946800862688713, |
| "grad_norm": 0.6405283212661743, |
| "learning_rate": 8.612508986340763e-05, |
| "loss": 0.1384, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.14018691588785046, |
| "grad_norm": 1.1369792222976685, |
| "learning_rate": 8.60531991373113e-05, |
| "loss": 0.1335, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.1409058231488138, |
| "grad_norm": 1.310003638267517, |
| "learning_rate": 8.598130841121496e-05, |
| "loss": 0.1353, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.14162473040977713, |
| "grad_norm": 0.7499133944511414, |
| "learning_rate": 8.590941768511862e-05, |
| "loss": 0.084, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.14234363767074049, |
| "grad_norm": 0.6289011836051941, |
| "learning_rate": 8.583752695902229e-05, |
| "loss": 0.1196, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.1430625449317038, |
| "grad_norm": 1.2842230796813965, |
| "learning_rate": 8.576563623292597e-05, |
| "loss": 0.2293, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.14378145219266714, |
| "grad_norm": 1.2408102750778198, |
| "learning_rate": 8.569374550682963e-05, |
| "loss": 0.1837, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.14378145219266714, |
| "eval_accuracy": 0.9261622428894043, |
| "eval_loss": 0.15695695579051971, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6554238796234131, |
| "eval_precision": 0.9221140742301941, |
| "eval_recall": 0.9636628031730652, |
| "eval_runtime": 602.7343, |
| "eval_samples_per_second": 4.617, |
| "eval_steps_per_second": 0.577, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.1445003594536305, |
| "grad_norm": 1.3462035655975342, |
| "learning_rate": 8.56218547807333e-05, |
| "loss": 0.1982, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.14521926671459381, |
| "grad_norm": 1.3314146995544434, |
| "learning_rate": 8.554996405463696e-05, |
| "loss": 0.1869, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.14593817397555714, |
| "grad_norm": 1.9233165979385376, |
| "learning_rate": 8.547807332854062e-05, |
| "loss": 0.1611, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.1466570812365205, |
| "grad_norm": 2.1303205490112305, |
| "learning_rate": 8.540618260244429e-05, |
| "loss": 0.2212, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.14737598849748382, |
| "grad_norm": 0.8199381232261658, |
| "learning_rate": 8.533429187634795e-05, |
| "loss": 0.1873, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.14809489575844717, |
| "grad_norm": 2.357494592666626, |
| "learning_rate": 8.526240115025162e-05, |
| "loss": 0.1636, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.1488138030194105, |
| "grad_norm": 0.9244757294654846, |
| "learning_rate": 8.519051042415528e-05, |
| "loss": 0.1181, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.14953271028037382, |
| "grad_norm": 2.8323566913604736, |
| "learning_rate": 8.511861969805895e-05, |
| "loss": 0.1763, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.15025161754133717, |
| "grad_norm": 0.9575947523117065, |
| "learning_rate": 8.504672897196261e-05, |
| "loss": 0.1585, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.1509705248023005, |
| "grad_norm": 2.096196174621582, |
| "learning_rate": 8.497483824586628e-05, |
| "loss": 0.1428, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.15168943206326385, |
| "grad_norm": 0.6452504992485046, |
| "learning_rate": 8.490294751976994e-05, |
| "loss": 0.1167, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.15240833932422718, |
| "grad_norm": 2.3544774055480957, |
| "learning_rate": 8.483105679367362e-05, |
| "loss": 0.1487, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.1531272465851905, |
| "grad_norm": 2.7052340507507324, |
| "learning_rate": 8.475916606757728e-05, |
| "loss": 0.1799, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.15384615384615385, |
| "grad_norm": 2.5094029903411865, |
| "learning_rate": 8.468727534148095e-05, |
| "loss": 0.1525, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.15456506110711718, |
| "grad_norm": 2.472966194152832, |
| "learning_rate": 8.461538461538461e-05, |
| "loss": 0.1581, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.1552839683680805, |
| "grad_norm": 2.283250331878662, |
| "learning_rate": 8.454349388928828e-05, |
| "loss": 0.246, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.15600287562904386, |
| "grad_norm": 0.9398806691169739, |
| "learning_rate": 8.447160316319194e-05, |
| "loss": 0.1909, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.15672178289000718, |
| "grad_norm": 1.2464269399642944, |
| "learning_rate": 8.439971243709562e-05, |
| "loss": 0.1708, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.15744069015097054, |
| "grad_norm": 2.4313812255859375, |
| "learning_rate": 8.432782171099929e-05, |
| "loss": 0.222, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.15815959741193386, |
| "grad_norm": 1.9309693574905396, |
| "learning_rate": 8.425593098490295e-05, |
| "loss": 0.1741, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.1588785046728972, |
| "grad_norm": 0.9127328395843506, |
| "learning_rate": 8.418404025880662e-05, |
| "loss": 0.1135, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.15959741193386054, |
| "grad_norm": 0.780059278011322, |
| "learning_rate": 8.411214953271028e-05, |
| "loss": 0.107, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.16031631919482386, |
| "grad_norm": 1.071359395980835, |
| "learning_rate": 8.404025880661395e-05, |
| "loss": 0.1909, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.16103522645578722, |
| "grad_norm": 0.6312058568000793, |
| "learning_rate": 8.396836808051761e-05, |
| "loss": 0.0999, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.16175413371675054, |
| "grad_norm": 1.3417737483978271, |
| "learning_rate": 8.389647735442129e-05, |
| "loss": 0.237, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.16247304097771387, |
| "grad_norm": 0.99446040391922, |
| "learning_rate": 8.382458662832495e-05, |
| "loss": 0.2125, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.16319194823867722, |
| "grad_norm": 0.6472429037094116, |
| "learning_rate": 8.375269590222862e-05, |
| "loss": 0.0938, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.16391085549964055, |
| "grad_norm": 1.0624991655349731, |
| "learning_rate": 8.368080517613228e-05, |
| "loss": 0.1394, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.16462976276060387, |
| "grad_norm": 0.7551208734512329, |
| "learning_rate": 8.360891445003595e-05, |
| "loss": 0.1258, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.16534867002156722, |
| "grad_norm": 0.7304654717445374, |
| "learning_rate": 8.353702372393961e-05, |
| "loss": 0.1489, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.16606757728253055, |
| "grad_norm": 0.6864830851554871, |
| "learning_rate": 8.346513299784329e-05, |
| "loss": 0.1256, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.1667864845434939, |
| "grad_norm": 1.5767005681991577, |
| "learning_rate": 8.339324227174695e-05, |
| "loss": 0.1786, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.16750539180445723, |
| "grad_norm": 1.1987712383270264, |
| "learning_rate": 8.332135154565062e-05, |
| "loss": 0.1845, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.16822429906542055, |
| "grad_norm": 0.8182716369628906, |
| "learning_rate": 8.324946081955428e-05, |
| "loss": 0.1517, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.1689432063263839, |
| "grad_norm": 0.6728079319000244, |
| "learning_rate": 8.317757009345795e-05, |
| "loss": 0.1448, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.16966211358734723, |
| "grad_norm": 2.4024782180786133, |
| "learning_rate": 8.310567936736161e-05, |
| "loss": 0.1667, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.17038102084831055, |
| "grad_norm": 1.7827162742614746, |
| "learning_rate": 8.303378864126528e-05, |
| "loss": 0.1616, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.1710999281092739, |
| "grad_norm": 1.3755327463150024, |
| "learning_rate": 8.296189791516896e-05, |
| "loss": 0.1243, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.17181883537023723, |
| "grad_norm": 1.0310641527175903, |
| "learning_rate": 8.289000718907262e-05, |
| "loss": 0.1544, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.17253774263120059, |
| "grad_norm": 1.8722699880599976, |
| "learning_rate": 8.281811646297629e-05, |
| "loss": 0.1671, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.1732566498921639, |
| "grad_norm": 2.2485873699188232, |
| "learning_rate": 8.274622573687995e-05, |
| "loss": 0.3335, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.17397555715312724, |
| "grad_norm": 1.9893066883087158, |
| "learning_rate": 8.267433501078362e-05, |
| "loss": 0.1711, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.1746944644140906, |
| "grad_norm": 3.5491175651550293, |
| "learning_rate": 8.260244428468728e-05, |
| "loss": 0.2039, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.17541337167505391, |
| "grad_norm": 3.9351868629455566, |
| "learning_rate": 8.253055355859094e-05, |
| "loss": 0.1916, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.17613227893601727, |
| "grad_norm": 2.2302541732788086, |
| "learning_rate": 8.245866283249461e-05, |
| "loss": 0.1443, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.1768511861969806, |
| "grad_norm": 2.324260950088501, |
| "learning_rate": 8.238677210639827e-05, |
| "loss": 0.1041, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.17757009345794392, |
| "grad_norm": 1.3087395429611206, |
| "learning_rate": 8.231488138030194e-05, |
| "loss": 0.2378, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.17828900071890727, |
| "grad_norm": 1.3952845335006714, |
| "learning_rate": 8.22429906542056e-05, |
| "loss": 0.1211, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.1790079079798706, |
| "grad_norm": 1.1960985660552979, |
| "learning_rate": 8.217109992810927e-05, |
| "loss": 0.1417, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.17972681524083392, |
| "grad_norm": 0.8309913277626038, |
| "learning_rate": 8.209920920201293e-05, |
| "loss": 0.1069, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.18044572250179727, |
| "grad_norm": 3.9855003356933594, |
| "learning_rate": 8.202731847591661e-05, |
| "loss": 0.2895, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.1811646297627606, |
| "grad_norm": 1.8458303213119507, |
| "learning_rate": 8.195542774982028e-05, |
| "loss": 0.1474, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.18188353702372395, |
| "grad_norm": 0.9199703931808472, |
| "learning_rate": 8.188353702372394e-05, |
| "loss": 0.1651, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.18260244428468728, |
| "grad_norm": 0.9246107935905457, |
| "learning_rate": 8.18116462976276e-05, |
| "loss": 0.164, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.1833213515456506, |
| "grad_norm": 0.976622998714447, |
| "learning_rate": 8.173975557153127e-05, |
| "loss": 0.1424, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.18404025880661395, |
| "grad_norm": 1.1131577491760254, |
| "learning_rate": 8.166786484543493e-05, |
| "loss": 0.2277, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.18475916606757728, |
| "grad_norm": 1.3994272947311401, |
| "learning_rate": 8.159597411933861e-05, |
| "loss": 0.1354, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.1854780733285406, |
| "grad_norm": 0.7919973134994507, |
| "learning_rate": 8.152408339324228e-05, |
| "loss": 0.184, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.18619698058950396, |
| "grad_norm": 1.959978461265564, |
| "learning_rate": 8.145219266714594e-05, |
| "loss": 0.2788, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.18691588785046728, |
| "grad_norm": 0.6807904839515686, |
| "learning_rate": 8.13803019410496e-05, |
| "loss": 0.1433, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.18763479511143064, |
| "grad_norm": 1.2130250930786133, |
| "learning_rate": 8.130841121495327e-05, |
| "loss": 0.1794, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.18835370237239396, |
| "grad_norm": 0.6109574437141418, |
| "learning_rate": 8.123652048885694e-05, |
| "loss": 0.1482, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.18907260963335729, |
| "grad_norm": 0.98033607006073, |
| "learning_rate": 8.11646297627606e-05, |
| "loss": 0.1314, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.18979151689432064, |
| "grad_norm": 1.1905217170715332, |
| "learning_rate": 8.109273903666428e-05, |
| "loss": 0.0872, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.19051042415528396, |
| "grad_norm": 1.0579410791397095, |
| "learning_rate": 8.102084831056794e-05, |
| "loss": 0.1578, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.19122933141624732, |
| "grad_norm": 1.1417217254638672, |
| "learning_rate": 8.094895758447161e-05, |
| "loss": 0.1151, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.19194823867721064, |
| "grad_norm": 0.803637683391571, |
| "learning_rate": 8.087706685837527e-05, |
| "loss": 0.1411, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.19266714593817397, |
| "grad_norm": 1.1087104082107544, |
| "learning_rate": 8.080517613227894e-05, |
| "loss": 0.2009, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.19338605319913732, |
| "grad_norm": 0.838955819606781, |
| "learning_rate": 8.07332854061826e-05, |
| "loss": 0.1048, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.19410496046010065, |
| "grad_norm": 1.6386674642562866, |
| "learning_rate": 8.066139468008628e-05, |
| "loss": 0.2484, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.19482386772106397, |
| "grad_norm": 0.9644877910614014, |
| "learning_rate": 8.058950395398995e-05, |
| "loss": 0.1945, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.19554277498202732, |
| "grad_norm": 1.1747089624404907, |
| "learning_rate": 8.051761322789361e-05, |
| "loss": 0.1618, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.19626168224299065, |
| "grad_norm": 0.8988797068595886, |
| "learning_rate": 8.044572250179727e-05, |
| "loss": 0.1622, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.196980589503954, |
| "grad_norm": 0.949789822101593, |
| "learning_rate": 8.037383177570094e-05, |
| "loss": 0.1465, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.19769949676491733, |
| "grad_norm": 2.2505388259887695, |
| "learning_rate": 8.03019410496046e-05, |
| "loss": 0.2112, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.19841840402588065, |
| "grad_norm": 3.6812849044799805, |
| "learning_rate": 8.023005032350828e-05, |
| "loss": 0.2604, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.199137311286844, |
| "grad_norm": 2.220285177230835, |
| "learning_rate": 8.015815959741195e-05, |
| "loss": 0.1805, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.19985621854780733, |
| "grad_norm": 1.4569875001907349, |
| "learning_rate": 8.008626887131561e-05, |
| "loss": 0.1445, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.20057512580877068, |
| "grad_norm": 0.9023877382278442, |
| "learning_rate": 8.001437814521928e-05, |
| "loss": 0.1165, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.201294033069734, |
| "grad_norm": 0.596118152141571, |
| "learning_rate": 7.994248741912294e-05, |
| "loss": 0.145, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.20201294033069733, |
| "grad_norm": 0.9356397986412048, |
| "learning_rate": 7.98705966930266e-05, |
| "loss": 0.1418, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.20273184759166069, |
| "grad_norm": 1.6694836616516113, |
| "learning_rate": 7.979870596693027e-05, |
| "loss": 0.1663, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.203450754852624, |
| "grad_norm": 1.472245454788208, |
| "learning_rate": 7.972681524083394e-05, |
| "loss": 0.0837, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.20416966211358734, |
| "grad_norm": 0.7129232883453369, |
| "learning_rate": 7.96549245147376e-05, |
| "loss": 0.1358, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.2048885693745507, |
| "grad_norm": 1.2116553783416748, |
| "learning_rate": 7.958303378864126e-05, |
| "loss": 0.1293, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.205607476635514, |
| "grad_norm": 1.1829216480255127, |
| "learning_rate": 7.951114306254493e-05, |
| "loss": 0.2561, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.20632638389647737, |
| "grad_norm": 0.5829026699066162, |
| "learning_rate": 7.94392523364486e-05, |
| "loss": 0.1027, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.2070452911574407, |
| "grad_norm": 0.7739739418029785, |
| "learning_rate": 7.936736161035226e-05, |
| "loss": 0.1112, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.20776419841840402, |
| "grad_norm": 1.601771593093872, |
| "learning_rate": 7.929547088425594e-05, |
| "loss": 0.1623, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.20848310567936737, |
| "grad_norm": 1.032884955406189, |
| "learning_rate": 7.92235801581596e-05, |
| "loss": 0.163, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.2092020129403307, |
| "grad_norm": 2.129141092300415, |
| "learning_rate": 7.915168943206327e-05, |
| "loss": 0.2317, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.20992092020129402, |
| "grad_norm": 1.0240572690963745, |
| "learning_rate": 7.907979870596693e-05, |
| "loss": 0.1878, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.21063982746225737, |
| "grad_norm": 0.9452721476554871, |
| "learning_rate": 7.90079079798706e-05, |
| "loss": 0.1759, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.2113587347232207, |
| "grad_norm": 0.9061615467071533, |
| "learning_rate": 7.893601725377426e-05, |
| "loss": 0.195, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.21207764198418405, |
| "grad_norm": 1.401107668876648, |
| "learning_rate": 7.886412652767792e-05, |
| "loss": 0.1762, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.21279654924514738, |
| "grad_norm": 1.7064363956451416, |
| "learning_rate": 7.87922358015816e-05, |
| "loss": 0.1083, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.2135154565061107, |
| "grad_norm": 0.8996630311012268, |
| "learning_rate": 7.872034507548527e-05, |
| "loss": 0.2256, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.21423436376707405, |
| "grad_norm": 0.5467312932014465, |
| "learning_rate": 7.864845434938893e-05, |
| "loss": 0.1586, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.21495327102803738, |
| "grad_norm": 1.2373936176300049, |
| "learning_rate": 7.85765636232926e-05, |
| "loss": 0.1597, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.21567217828900073, |
| "grad_norm": 1.0149507522583008, |
| "learning_rate": 7.850467289719626e-05, |
| "loss": 0.1528, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.21639108554996406, |
| "grad_norm": 1.3829774856567383, |
| "learning_rate": 7.843278217109993e-05, |
| "loss": 0.3025, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.21710999281092738, |
| "grad_norm": 1.1181117296218872, |
| "learning_rate": 7.83608914450036e-05, |
| "loss": 0.1425, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.21782890007189074, |
| "grad_norm": 1.2984338998794556, |
| "learning_rate": 7.828900071890727e-05, |
| "loss": 0.1471, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.21854780733285406, |
| "grad_norm": 2.4325575828552246, |
| "learning_rate": 7.821710999281093e-05, |
| "loss": 0.2435, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.21926671459381739, |
| "grad_norm": 1.392596960067749, |
| "learning_rate": 7.81452192667146e-05, |
| "loss": 0.108, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.21998562185478074, |
| "grad_norm": 0.6541122198104858, |
| "learning_rate": 7.807332854061826e-05, |
| "loss": 0.1291, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.22070452911574406, |
| "grad_norm": 0.9223505854606628, |
| "learning_rate": 7.800143781452193e-05, |
| "loss": 0.1326, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.22142343637670742, |
| "grad_norm": 0.7688751220703125, |
| "learning_rate": 7.792954708842559e-05, |
| "loss": 0.1524, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.22214234363767074, |
| "grad_norm": 0.5062466859817505, |
| "learning_rate": 7.785765636232927e-05, |
| "loss": 0.1204, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.22286125089863407, |
| "grad_norm": 1.205576777458191, |
| "learning_rate": 7.778576563623294e-05, |
| "loss": 0.2146, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.22358015815959742, |
| "grad_norm": 0.8479394316673279, |
| "learning_rate": 7.77138749101366e-05, |
| "loss": 0.126, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.22429906542056074, |
| "grad_norm": 1.1954373121261597, |
| "learning_rate": 7.764198418404027e-05, |
| "loss": 0.2473, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.22501797268152407, |
| "grad_norm": 0.707917332649231, |
| "learning_rate": 7.757009345794393e-05, |
| "loss": 0.1423, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.22573687994248742, |
| "grad_norm": 1.0723934173583984, |
| "learning_rate": 7.74982027318476e-05, |
| "loss": 0.1669, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.22645578720345075, |
| "grad_norm": 1.4507575035095215, |
| "learning_rate": 7.742631200575127e-05, |
| "loss": 0.1896, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.2271746944644141, |
| "grad_norm": 0.7972711324691772, |
| "learning_rate": 7.735442127965494e-05, |
| "loss": 0.1267, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.22789360172537743, |
| "grad_norm": 1.5650343894958496, |
| "learning_rate": 7.72825305535586e-05, |
| "loss": 0.1037, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.22861250898634075, |
| "grad_norm": 1.0155292749404907, |
| "learning_rate": 7.721063982746227e-05, |
| "loss": 0.1708, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.2293314162473041, |
| "grad_norm": 2.5561273097991943, |
| "learning_rate": 7.713874910136593e-05, |
| "loss": 0.1824, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.23005032350826743, |
| "grad_norm": 0.7330106496810913, |
| "learning_rate": 7.70668583752696e-05, |
| "loss": 0.1848, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.23076923076923078, |
| "grad_norm": 0.8648838400840759, |
| "learning_rate": 7.699496764917326e-05, |
| "loss": 0.1973, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.2314881380301941, |
| "grad_norm": 1.2262929677963257, |
| "learning_rate": 7.692307692307693e-05, |
| "loss": 0.1131, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.23220704529115743, |
| "grad_norm": 0.8531995415687561, |
| "learning_rate": 7.685118619698059e-05, |
| "loss": 0.1599, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.23292595255212079, |
| "grad_norm": 2.8929901123046875, |
| "learning_rate": 7.677929547088426e-05, |
| "loss": 0.2661, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.2336448598130841, |
| "grad_norm": 2.1622095108032227, |
| "learning_rate": 7.670740474478792e-05, |
| "loss": 0.2437, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.23436376707404744, |
| "grad_norm": 0.7595067620277405, |
| "learning_rate": 7.663551401869158e-05, |
| "loss": 0.1249, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.2350826743350108, |
| "grad_norm": 0.8510514497756958, |
| "learning_rate": 7.656362329259525e-05, |
| "loss": 0.1756, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.2358015815959741, |
| "grad_norm": 1.0503634214401245, |
| "learning_rate": 7.649173256649893e-05, |
| "loss": 0.1922, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.23652048885693747, |
| "grad_norm": 1.3306595087051392, |
| "learning_rate": 7.641984184040259e-05, |
| "loss": 0.1788, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.2372393961179008, |
| "grad_norm": 1.438962459564209, |
| "learning_rate": 7.634795111430626e-05, |
| "loss": 0.1514, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.23795830337886412, |
| "grad_norm": 0.5268650650978088, |
| "learning_rate": 7.627606038820992e-05, |
| "loss": 0.14, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.23867721063982747, |
| "grad_norm": 1.43551504611969, |
| "learning_rate": 7.620416966211359e-05, |
| "loss": 0.1413, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.2393961179007908, |
| "grad_norm": 0.611075758934021, |
| "learning_rate": 7.613227893601725e-05, |
| "loss": 0.104, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.24011502516175415, |
| "grad_norm": 0.7788759469985962, |
| "learning_rate": 7.606038820992092e-05, |
| "loss": 0.167, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.24083393242271747, |
| "grad_norm": 0.8650516867637634, |
| "learning_rate": 7.59884974838246e-05, |
| "loss": 0.1181, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.2415528396836808, |
| "grad_norm": 0.6475949287414551, |
| "learning_rate": 7.591660675772826e-05, |
| "loss": 0.134, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.24227174694464415, |
| "grad_norm": 1.0974619388580322, |
| "learning_rate": 7.584471603163192e-05, |
| "loss": 0.0979, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.24299065420560748, |
| "grad_norm": 0.7453069686889648, |
| "learning_rate": 7.577282530553559e-05, |
| "loss": 0.1248, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.2437095614665708, |
| "grad_norm": 0.5186426043510437, |
| "learning_rate": 7.570093457943925e-05, |
| "loss": 0.0912, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.24442846872753415, |
| "grad_norm": 0.5677504539489746, |
| "learning_rate": 7.562904385334292e-05, |
| "loss": 0.141, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.24514737598849748, |
| "grad_norm": 0.8022885322570801, |
| "learning_rate": 7.55571531272466e-05, |
| "loss": 0.1272, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.24586628324946083, |
| "grad_norm": 0.862316906452179, |
| "learning_rate": 7.548526240115026e-05, |
| "loss": 0.0977, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.24658519051042416, |
| "grad_norm": 1.0201530456542969, |
| "learning_rate": 7.541337167505392e-05, |
| "loss": 0.1285, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.24730409777138748, |
| "grad_norm": 0.5684915781021118, |
| "learning_rate": 7.534148094895759e-05, |
| "loss": 0.1082, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.24802300503235084, |
| "grad_norm": 0.5687134265899658, |
| "learning_rate": 7.526959022286125e-05, |
| "loss": 0.1131, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.24874191229331416, |
| "grad_norm": 0.9496104717254639, |
| "learning_rate": 7.519769949676492e-05, |
| "loss": 0.1551, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.24946081955427749, |
| "grad_norm": 1.4114607572555542, |
| "learning_rate": 7.512580877066858e-05, |
| "loss": 0.2625, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.2501797268152408, |
| "grad_norm": 1.1259725093841553, |
| "learning_rate": 7.505391804457226e-05, |
| "loss": 0.1388, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.25089863407620416, |
| "grad_norm": 1.1572023630142212, |
| "learning_rate": 7.498202731847593e-05, |
| "loss": 0.1778, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.2516175413371675, |
| "grad_norm": 0.5434879660606384, |
| "learning_rate": 7.491013659237959e-05, |
| "loss": 0.0998, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2523364485981308, |
| "grad_norm": 1.2463257312774658, |
| "learning_rate": 7.483824586628326e-05, |
| "loss": 0.2216, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.25305535585909417, |
| "grad_norm": 0.7039005160331726, |
| "learning_rate": 7.476635514018692e-05, |
| "loss": 0.1593, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.2537742631200575, |
| "grad_norm": 1.324094295501709, |
| "learning_rate": 7.469446441409059e-05, |
| "loss": 0.1102, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.2544931703810209, |
| "grad_norm": 0.6719451546669006, |
| "learning_rate": 7.462257368799426e-05, |
| "loss": 0.1196, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.25521207764198417, |
| "grad_norm": 1.0878773927688599, |
| "learning_rate": 7.455068296189793e-05, |
| "loss": 0.1507, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.2559309849029475, |
| "grad_norm": 0.6032963395118713, |
| "learning_rate": 7.447879223580159e-05, |
| "loss": 0.0934, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.2566498921639109, |
| "grad_norm": 0.8666496276855469, |
| "learning_rate": 7.440690150970526e-05, |
| "loss": 0.1951, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.2573687994248742, |
| "grad_norm": 0.6647478342056274, |
| "learning_rate": 7.433501078360892e-05, |
| "loss": 0.1087, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.2580877066858375, |
| "grad_norm": 0.6436010003089905, |
| "learning_rate": 7.426312005751259e-05, |
| "loss": 0.0992, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.2588066139468009, |
| "grad_norm": 0.7569960951805115, |
| "learning_rate": 7.419122933141625e-05, |
| "loss": 0.1413, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.2595255212077642, |
| "grad_norm": 1.5900118350982666, |
| "learning_rate": 7.411933860531992e-05, |
| "loss": 0.2225, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.26024442846872753, |
| "grad_norm": 1.2691010236740112, |
| "learning_rate": 7.404744787922358e-05, |
| "loss": 0.1759, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.2609633357296909, |
| "grad_norm": 0.5943278670310974, |
| "learning_rate": 7.397555715312725e-05, |
| "loss": 0.0994, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.2616822429906542, |
| "grad_norm": 1.376136064529419, |
| "learning_rate": 7.390366642703091e-05, |
| "loss": 0.2242, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.26240115025161753, |
| "grad_norm": 1.6708033084869385, |
| "learning_rate": 7.383177570093458e-05, |
| "loss": 0.1336, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.2631200575125809, |
| "grad_norm": 1.465207576751709, |
| "learning_rate": 7.375988497483824e-05, |
| "loss": 0.1883, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.26383896477354424, |
| "grad_norm": 0.965944230556488, |
| "learning_rate": 7.368799424874192e-05, |
| "loss": 0.2182, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.26455787203450754, |
| "grad_norm": 0.556839108467102, |
| "learning_rate": 7.361610352264558e-05, |
| "loss": 0.1009, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.2652767792954709, |
| "grad_norm": 0.7746458053588867, |
| "learning_rate": 7.354421279654925e-05, |
| "loss": 0.1431, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.26599568655643424, |
| "grad_norm": 1.3613864183425903, |
| "learning_rate": 7.347232207045291e-05, |
| "loss": 0.2002, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.26671459381739754, |
| "grad_norm": 0.8492037057876587, |
| "learning_rate": 7.340043134435658e-05, |
| "loss": 0.1721, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.2674335010783609, |
| "grad_norm": 2.1364450454711914, |
| "learning_rate": 7.332854061826024e-05, |
| "loss": 0.2187, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.26815240833932424, |
| "grad_norm": 0.8353592753410339, |
| "learning_rate": 7.32566498921639e-05, |
| "loss": 0.1501, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.26887131560028754, |
| "grad_norm": 0.4625248610973358, |
| "learning_rate": 7.318475916606758e-05, |
| "loss": 0.1069, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.2695902228612509, |
| "grad_norm": 0.5995030403137207, |
| "learning_rate": 7.311286843997125e-05, |
| "loss": 0.1006, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.27030913012221425, |
| "grad_norm": 1.5556994676589966, |
| "learning_rate": 7.304097771387491e-05, |
| "loss": 0.1315, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.27102803738317754, |
| "grad_norm": 1.143564224243164, |
| "learning_rate": 7.296908698777858e-05, |
| "loss": 0.1446, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.2717469446441409, |
| "grad_norm": 0.8274374008178711, |
| "learning_rate": 7.289719626168224e-05, |
| "loss": 0.1881, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.27246585190510425, |
| "grad_norm": 0.810071587562561, |
| "learning_rate": 7.282530553558591e-05, |
| "loss": 0.1376, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.2731847591660676, |
| "grad_norm": 1.0889372825622559, |
| "learning_rate": 7.275341480948959e-05, |
| "loss": 0.2517, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.2739036664270309, |
| "grad_norm": 0.6273843050003052, |
| "learning_rate": 7.268152408339325e-05, |
| "loss": 0.1309, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.27462257368799425, |
| "grad_norm": 0.7900193333625793, |
| "learning_rate": 7.260963335729692e-05, |
| "loss": 0.1054, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.2753414809489576, |
| "grad_norm": 0.5866711735725403, |
| "learning_rate": 7.253774263120058e-05, |
| "loss": 0.1496, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.2760603882099209, |
| "grad_norm": 0.7079902291297913, |
| "learning_rate": 7.246585190510424e-05, |
| "loss": 0.1325, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.27677929547088426, |
| "grad_norm": 1.9546867609024048, |
| "learning_rate": 7.239396117900791e-05, |
| "loss": 0.2521, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.2774982027318476, |
| "grad_norm": 1.137825608253479, |
| "learning_rate": 7.232207045291159e-05, |
| "loss": 0.1667, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.2782171099928109, |
| "grad_norm": 1.0918433666229248, |
| "learning_rate": 7.225017972681525e-05, |
| "loss": 0.1966, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.27893601725377426, |
| "grad_norm": 1.2907887697219849, |
| "learning_rate": 7.217828900071892e-05, |
| "loss": 0.1466, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.2796549245147376, |
| "grad_norm": 1.178578495979309, |
| "learning_rate": 7.210639827462258e-05, |
| "loss": 0.1192, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.2803738317757009, |
| "grad_norm": 1.241025686264038, |
| "learning_rate": 7.203450754852625e-05, |
| "loss": 0.1042, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.28109273903666426, |
| "grad_norm": 1.9735612869262695, |
| "learning_rate": 7.196261682242991e-05, |
| "loss": 0.3748, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.2818116462976276, |
| "grad_norm": 0.8372847437858582, |
| "learning_rate": 7.189072609633358e-05, |
| "loss": 0.1542, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.28253055355859097, |
| "grad_norm": 1.099506139755249, |
| "learning_rate": 7.181883537023724e-05, |
| "loss": 0.2261, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.28324946081955427, |
| "grad_norm": 0.685853123664856, |
| "learning_rate": 7.174694464414092e-05, |
| "loss": 0.1284, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.2839683680805176, |
| "grad_norm": 0.6166328191757202, |
| "learning_rate": 7.167505391804458e-05, |
| "loss": 0.1015, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.28468727534148097, |
| "grad_norm": 1.4274548292160034, |
| "learning_rate": 7.160316319194825e-05, |
| "loss": 0.2144, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.28540618260244427, |
| "grad_norm": 1.2263768911361694, |
| "learning_rate": 7.153127246585191e-05, |
| "loss": 0.1473, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.2861250898634076, |
| "grad_norm": 0.6739816069602966, |
| "learning_rate": 7.145938173975558e-05, |
| "loss": 0.144, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.286843997124371, |
| "grad_norm": 0.8352803587913513, |
| "learning_rate": 7.138749101365924e-05, |
| "loss": 0.1581, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.2875629043853343, |
| "grad_norm": 1.2863707542419434, |
| "learning_rate": 7.131560028756291e-05, |
| "loss": 0.196, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.2875629043853343, |
| "eval_accuracy": 0.928896963596344, |
| "eval_loss": 0.14465543627738953, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6563354730606079, |
| "eval_precision": 0.9236111044883728, |
| "eval_recall": 0.9665697813034058, |
| "eval_runtime": 602.5363, |
| "eval_samples_per_second": 4.619, |
| "eval_steps_per_second": 0.578, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.2882818116462976, |
| "grad_norm": 1.1578876972198486, |
| "learning_rate": 7.124370956146657e-05, |
| "loss": 0.2142, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.289000718907261, |
| "grad_norm": 1.1095978021621704, |
| "learning_rate": 7.117181883537024e-05, |
| "loss": 0.1497, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.2897196261682243, |
| "grad_norm": 0.5568669438362122, |
| "learning_rate": 7.10999281092739e-05, |
| "loss": 0.1303, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.29043853342918763, |
| "grad_norm": 0.8726722002029419, |
| "learning_rate": 7.102803738317757e-05, |
| "loss": 0.1287, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.291157440690151, |
| "grad_norm": 0.6691792607307434, |
| "learning_rate": 7.095614665708123e-05, |
| "loss": 0.1221, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.2918763479511143, |
| "grad_norm": 1.4088934659957886, |
| "learning_rate": 7.088425593098491e-05, |
| "loss": 0.1305, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.29259525521207763, |
| "grad_norm": 1.162762999534607, |
| "learning_rate": 7.081236520488857e-05, |
| "loss": 0.25, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.293314162473041, |
| "grad_norm": 1.0753746032714844, |
| "learning_rate": 7.074047447879224e-05, |
| "loss": 0.1345, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.29403306973400434, |
| "grad_norm": 0.9136525392532349, |
| "learning_rate": 7.06685837526959e-05, |
| "loss": 0.1408, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.29475197699496763, |
| "grad_norm": 0.6772589683532715, |
| "learning_rate": 7.059669302659957e-05, |
| "loss": 0.0981, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.295470884255931, |
| "grad_norm": 0.9310998320579529, |
| "learning_rate": 7.052480230050323e-05, |
| "loss": 0.1573, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.29618979151689434, |
| "grad_norm": 0.9734662175178528, |
| "learning_rate": 7.045291157440691e-05, |
| "loss": 0.1146, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.29690869877785764, |
| "grad_norm": 0.8142755031585693, |
| "learning_rate": 7.038102084831058e-05, |
| "loss": 0.1164, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.297627606038821, |
| "grad_norm": 1.020463228225708, |
| "learning_rate": 7.030913012221424e-05, |
| "loss": 0.1172, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.29834651329978434, |
| "grad_norm": 0.6402932405471802, |
| "learning_rate": 7.02372393961179e-05, |
| "loss": 0.1356, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.29906542056074764, |
| "grad_norm": 1.4168362617492676, |
| "learning_rate": 7.016534867002157e-05, |
| "loss": 0.0994, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.299784327821711, |
| "grad_norm": 0.5625162124633789, |
| "learning_rate": 7.009345794392523e-05, |
| "loss": 0.0913, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.30050323508267435, |
| "grad_norm": 2.9484524726867676, |
| "learning_rate": 7.00215672178289e-05, |
| "loss": 0.2973, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.30122214234363764, |
| "grad_norm": 0.9706459641456604, |
| "learning_rate": 6.994967649173258e-05, |
| "loss": 0.1829, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.301941049604601, |
| "grad_norm": 1.4434211254119873, |
| "learning_rate": 6.987778576563624e-05, |
| "loss": 0.1488, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.30265995686556435, |
| "grad_norm": 0.8742904663085938, |
| "learning_rate": 6.98058950395399e-05, |
| "loss": 0.0903, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.3033788641265277, |
| "grad_norm": 1.1030147075653076, |
| "learning_rate": 6.973400431344357e-05, |
| "loss": 0.139, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.304097771387491, |
| "grad_norm": 1.611258625984192, |
| "learning_rate": 6.966211358734724e-05, |
| "loss": 0.131, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.30481667864845435, |
| "grad_norm": 0.9128326177597046, |
| "learning_rate": 6.95902228612509e-05, |
| "loss": 0.1568, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.3055355859094177, |
| "grad_norm": 1.0490553379058838, |
| "learning_rate": 6.951833213515458e-05, |
| "loss": 0.2333, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.306254493170381, |
| "grad_norm": 0.667241096496582, |
| "learning_rate": 6.944644140905824e-05, |
| "loss": 0.1707, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.30697340043134436, |
| "grad_norm": 1.1369763612747192, |
| "learning_rate": 6.937455068296191e-05, |
| "loss": 0.1327, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.3076923076923077, |
| "grad_norm": 0.905947744846344, |
| "learning_rate": 6.930265995686557e-05, |
| "loss": 0.1792, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.308411214953271, |
| "grad_norm": 0.5325162410736084, |
| "learning_rate": 6.923076923076924e-05, |
| "loss": 0.1391, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.30913012221423436, |
| "grad_norm": 0.675025999546051, |
| "learning_rate": 6.91588785046729e-05, |
| "loss": 0.1095, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.3098490294751977, |
| "grad_norm": 1.2129236459732056, |
| "learning_rate": 6.908698777857657e-05, |
| "loss": 0.1636, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.310567936736161, |
| "grad_norm": 1.8984676599502563, |
| "learning_rate": 6.901509705248023e-05, |
| "loss": 0.1706, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.31128684399712436, |
| "grad_norm": 1.3937430381774902, |
| "learning_rate": 6.89432063263839e-05, |
| "loss": 0.1453, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.3120057512580877, |
| "grad_norm": 1.6447863578796387, |
| "learning_rate": 6.887131560028756e-05, |
| "loss": 0.1787, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.31272465851905107, |
| "grad_norm": 1.4637572765350342, |
| "learning_rate": 6.879942487419123e-05, |
| "loss": 0.1373, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.31344356578001437, |
| "grad_norm": 2.5144917964935303, |
| "learning_rate": 6.872753414809489e-05, |
| "loss": 0.3381, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.3141624730409777, |
| "grad_norm": 0.8186586499214172, |
| "learning_rate": 6.865564342199855e-05, |
| "loss": 0.1728, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.31488138030194107, |
| "grad_norm": 2.0116798877716064, |
| "learning_rate": 6.858375269590223e-05, |
| "loss": 0.2191, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.31560028756290437, |
| "grad_norm": 1.6542249917984009, |
| "learning_rate": 6.85118619698059e-05, |
| "loss": 0.1163, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.3163191948238677, |
| "grad_norm": 0.8176520466804504, |
| "learning_rate": 6.843997124370956e-05, |
| "loss": 0.1913, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.3170381020848311, |
| "grad_norm": 1.4219037294387817, |
| "learning_rate": 6.836808051761323e-05, |
| "loss": 0.2051, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.3177570093457944, |
| "grad_norm": 0.6978217363357544, |
| "learning_rate": 6.829618979151689e-05, |
| "loss": 0.1312, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.3184759166067577, |
| "grad_norm": 1.081991195678711, |
| "learning_rate": 6.822429906542056e-05, |
| "loss": 0.154, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.3191948238677211, |
| "grad_norm": 0.7945512533187866, |
| "learning_rate": 6.815240833932422e-05, |
| "loss": 0.1637, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.3199137311286844, |
| "grad_norm": 0.6147925853729248, |
| "learning_rate": 6.80805176132279e-05, |
| "loss": 0.1714, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.32063263838964773, |
| "grad_norm": 0.9328286051750183, |
| "learning_rate": 6.800862688713156e-05, |
| "loss": 0.1017, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.3213515456506111, |
| "grad_norm": 0.8529399037361145, |
| "learning_rate": 6.793673616103523e-05, |
| "loss": 0.1285, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.32207045291157443, |
| "grad_norm": 0.9937113523483276, |
| "learning_rate": 6.78648454349389e-05, |
| "loss": 0.1514, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.32278936017253773, |
| "grad_norm": 0.8380501866340637, |
| "learning_rate": 6.779295470884256e-05, |
| "loss": 0.1065, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.3235082674335011, |
| "grad_norm": 0.8374715447425842, |
| "learning_rate": 6.772106398274622e-05, |
| "loss": 0.153, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.32422717469446444, |
| "grad_norm": 0.6685544848442078, |
| "learning_rate": 6.76491732566499e-05, |
| "loss": 0.0815, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.32494608195542773, |
| "grad_norm": 0.8247376680374146, |
| "learning_rate": 6.757728253055357e-05, |
| "loss": 0.1414, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.3256649892163911, |
| "grad_norm": 0.726820170879364, |
| "learning_rate": 6.750539180445723e-05, |
| "loss": 0.1158, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.32638389647735444, |
| "grad_norm": 2.300973415374756, |
| "learning_rate": 6.74335010783609e-05, |
| "loss": 0.2397, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.32710280373831774, |
| "grad_norm": 0.6339260339736938, |
| "learning_rate": 6.736161035226456e-05, |
| "loss": 0.1416, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.3278217109992811, |
| "grad_norm": 1.2476907968521118, |
| "learning_rate": 6.728971962616822e-05, |
| "loss": 0.254, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.32854061826024444, |
| "grad_norm": 0.6578892469406128, |
| "learning_rate": 6.721782890007189e-05, |
| "loss": 0.1119, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.32925952552120774, |
| "grad_norm": 0.8718114495277405, |
| "learning_rate": 6.714593817397557e-05, |
| "loss": 0.1778, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.3299784327821711, |
| "grad_norm": 1.5130765438079834, |
| "learning_rate": 6.707404744787923e-05, |
| "loss": 0.1268, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.33069734004313445, |
| "grad_norm": 0.9510946273803711, |
| "learning_rate": 6.70021567217829e-05, |
| "loss": 0.1543, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.33141624730409774, |
| "grad_norm": 1.3413060903549194, |
| "learning_rate": 6.693026599568656e-05, |
| "loss": 0.2444, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.3321351545650611, |
| "grad_norm": 0.9424805045127869, |
| "learning_rate": 6.685837526959023e-05, |
| "loss": 0.1868, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.33285406182602445, |
| "grad_norm": 1.6345001459121704, |
| "learning_rate": 6.678648454349389e-05, |
| "loss": 0.2082, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.3335729690869878, |
| "grad_norm": 1.112050175666809, |
| "learning_rate": 6.671459381739757e-05, |
| "loss": 0.1043, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.3342918763479511, |
| "grad_norm": 1.4402129650115967, |
| "learning_rate": 6.664270309130123e-05, |
| "loss": 0.1747, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.33501078360891445, |
| "grad_norm": 0.699443519115448, |
| "learning_rate": 6.65708123652049e-05, |
| "loss": 0.1039, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.3357296908698778, |
| "grad_norm": 0.8008235096931458, |
| "learning_rate": 6.649892163910856e-05, |
| "loss": 0.134, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.3364485981308411, |
| "grad_norm": 0.7060474157333374, |
| "learning_rate": 6.642703091301223e-05, |
| "loss": 0.0884, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.33716750539180446, |
| "grad_norm": 1.0643765926361084, |
| "learning_rate": 6.635514018691589e-05, |
| "loss": 0.1451, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.3378864126527678, |
| "grad_norm": 1.6296634674072266, |
| "learning_rate": 6.628324946081956e-05, |
| "loss": 0.2087, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.3386053199137311, |
| "grad_norm": 0.7258971929550171, |
| "learning_rate": 6.621135873472322e-05, |
| "loss": 0.1716, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.33932422717469446, |
| "grad_norm": 1.0323922634124756, |
| "learning_rate": 6.613946800862689e-05, |
| "loss": 0.164, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.3400431344356578, |
| "grad_norm": 2.1684961318969727, |
| "learning_rate": 6.606757728253055e-05, |
| "loss": 0.1853, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.3407620416966211, |
| "grad_norm": 0.5301715135574341, |
| "learning_rate": 6.599568655643422e-05, |
| "loss": 0.1267, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.34148094895758446, |
| "grad_norm": 0.7696582078933716, |
| "learning_rate": 6.592379583033788e-05, |
| "loss": 0.1232, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.3421998562185478, |
| "grad_norm": 0.8466119766235352, |
| "learning_rate": 6.585190510424155e-05, |
| "loss": 0.185, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.34291876347951117, |
| "grad_norm": 1.5882998704910278, |
| "learning_rate": 6.578001437814522e-05, |
| "loss": 0.0998, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.34363767074047447, |
| "grad_norm": 1.439219355583191, |
| "learning_rate": 6.570812365204889e-05, |
| "loss": 0.1292, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.3443565780014378, |
| "grad_norm": 1.376057744026184, |
| "learning_rate": 6.563623292595255e-05, |
| "loss": 0.1345, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.34507548526240117, |
| "grad_norm": 0.6745831966400146, |
| "learning_rate": 6.556434219985622e-05, |
| "loss": 0.1121, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.34579439252336447, |
| "grad_norm": 0.9137468934059143, |
| "learning_rate": 6.549245147375988e-05, |
| "loss": 0.1001, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.3465132997843278, |
| "grad_norm": 0.8462786674499512, |
| "learning_rate": 6.542056074766355e-05, |
| "loss": 0.1321, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.3472322070452912, |
| "grad_norm": 0.9228039383888245, |
| "learning_rate": 6.534867002156723e-05, |
| "loss": 0.1656, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.34795111430625447, |
| "grad_norm": 3.7959492206573486, |
| "learning_rate": 6.527677929547089e-05, |
| "loss": 0.3162, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.3486700215672178, |
| "grad_norm": 1.453232765197754, |
| "learning_rate": 6.520488856937455e-05, |
| "loss": 0.1764, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.3493889288281812, |
| "grad_norm": 0.813624918460846, |
| "learning_rate": 6.513299784327822e-05, |
| "loss": 0.139, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.3501078360891445, |
| "grad_norm": 2.408250570297241, |
| "learning_rate": 6.506110711718188e-05, |
| "loss": 0.1875, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.35082674335010783, |
| "grad_norm": 0.781344473361969, |
| "learning_rate": 6.498921639108555e-05, |
| "loss": 0.1161, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.3515456506110712, |
| "grad_norm": 0.6290162801742554, |
| "learning_rate": 6.491732566498921e-05, |
| "loss": 0.0917, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.35226455787203453, |
| "grad_norm": 0.8420344591140747, |
| "learning_rate": 6.484543493889289e-05, |
| "loss": 0.1644, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.35298346513299783, |
| "grad_norm": 0.5526477694511414, |
| "learning_rate": 6.477354421279656e-05, |
| "loss": 0.1538, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.3537023723939612, |
| "grad_norm": 0.6922248005867004, |
| "learning_rate": 6.470165348670022e-05, |
| "loss": 0.1361, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.35442127965492454, |
| "grad_norm": 1.2012991905212402, |
| "learning_rate": 6.462976276060389e-05, |
| "loss": 0.1253, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.35514018691588783, |
| "grad_norm": 2.9389891624450684, |
| "learning_rate": 6.455787203450755e-05, |
| "loss": 0.2912, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.3558590941768512, |
| "grad_norm": 2.280164957046509, |
| "learning_rate": 6.448598130841122e-05, |
| "loss": 0.1256, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.35657800143781454, |
| "grad_norm": 2.0135037899017334, |
| "learning_rate": 6.44140905823149e-05, |
| "loss": 0.2304, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.35729690869877784, |
| "grad_norm": 0.6317698359489441, |
| "learning_rate": 6.434219985621856e-05, |
| "loss": 0.1154, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.3580158159597412, |
| "grad_norm": 1.062667727470398, |
| "learning_rate": 6.427030913012222e-05, |
| "loss": 0.109, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.35873472322070454, |
| "grad_norm": 0.6957636475563049, |
| "learning_rate": 6.419841840402589e-05, |
| "loss": 0.0949, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.35945363048166784, |
| "grad_norm": 0.9724053144454956, |
| "learning_rate": 6.412652767792955e-05, |
| "loss": 0.1677, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.3601725377426312, |
| "grad_norm": 0.656159520149231, |
| "learning_rate": 6.405463695183322e-05, |
| "loss": 0.0953, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.36089144500359455, |
| "grad_norm": 0.6304846405982971, |
| "learning_rate": 6.398274622573688e-05, |
| "loss": 0.1344, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.3616103522645579, |
| "grad_norm": 0.5929521918296814, |
| "learning_rate": 6.391085549964056e-05, |
| "loss": 0.1124, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.3623292595255212, |
| "grad_norm": 0.46629002690315247, |
| "learning_rate": 6.383896477354422e-05, |
| "loss": 0.0846, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.36304816678648455, |
| "grad_norm": 2.310432195663452, |
| "learning_rate": 6.376707404744789e-05, |
| "loss": 0.2651, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.3637670740474479, |
| "grad_norm": 2.4397852420806885, |
| "learning_rate": 6.369518332135155e-05, |
| "loss": 0.2162, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.3644859813084112, |
| "grad_norm": 1.6224433183670044, |
| "learning_rate": 6.362329259525522e-05, |
| "loss": 0.1382, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.36520488856937455, |
| "grad_norm": 0.6871057748794556, |
| "learning_rate": 6.355140186915888e-05, |
| "loss": 0.1083, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.3659237958303379, |
| "grad_norm": 0.9529922604560852, |
| "learning_rate": 6.347951114306255e-05, |
| "loss": 0.1173, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.3666427030913012, |
| "grad_norm": 1.0535115003585815, |
| "learning_rate": 6.340762041696621e-05, |
| "loss": 0.1091, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.36736161035226456, |
| "grad_norm": 0.6320528984069824, |
| "learning_rate": 6.333572969086988e-05, |
| "loss": 0.0602, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.3680805176132279, |
| "grad_norm": 1.0191007852554321, |
| "learning_rate": 6.326383896477354e-05, |
| "loss": 0.197, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.3687994248741912, |
| "grad_norm": 0.9638345241546631, |
| "learning_rate": 6.31919482386772e-05, |
| "loss": 0.1155, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.36951833213515456, |
| "grad_norm": 0.7040844559669495, |
| "learning_rate": 6.312005751258087e-05, |
| "loss": 0.1566, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.3702372393961179, |
| "grad_norm": 1.1568024158477783, |
| "learning_rate": 6.304816678648454e-05, |
| "loss": 0.1855, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.3709561466570812, |
| "grad_norm": 1.1298298835754395, |
| "learning_rate": 6.297627606038821e-05, |
| "loss": 0.177, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.37167505391804456, |
| "grad_norm": 1.572418451309204, |
| "learning_rate": 6.290438533429188e-05, |
| "loss": 0.1763, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.3723939611790079, |
| "grad_norm": 0.5524023175239563, |
| "learning_rate": 6.283249460819554e-05, |
| "loss": 0.128, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.37311286843997127, |
| "grad_norm": 0.5786755084991455, |
| "learning_rate": 6.276060388209921e-05, |
| "loss": 0.0964, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.37383177570093457, |
| "grad_norm": 0.9922318458557129, |
| "learning_rate": 6.268871315600287e-05, |
| "loss": 0.1904, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.3745506829618979, |
| "grad_norm": 2.362755060195923, |
| "learning_rate": 6.261682242990654e-05, |
| "loss": 0.2071, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.37526959022286127, |
| "grad_norm": 0.7869640588760376, |
| "learning_rate": 6.254493170381022e-05, |
| "loss": 0.1004, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.37598849748382457, |
| "grad_norm": 1.2965569496154785, |
| "learning_rate": 6.247304097771388e-05, |
| "loss": 0.1087, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.3767074047447879, |
| "grad_norm": 0.697258472442627, |
| "learning_rate": 6.240115025161755e-05, |
| "loss": 0.1186, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.3774263120057513, |
| "grad_norm": 0.5020483732223511, |
| "learning_rate": 6.232925952552121e-05, |
| "loss": 0.0954, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.37814521926671457, |
| "grad_norm": 0.5413784980773926, |
| "learning_rate": 6.225736879942487e-05, |
| "loss": 0.1227, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.3788641265276779, |
| "grad_norm": 1.4515224695205688, |
| "learning_rate": 6.218547807332854e-05, |
| "loss": 0.1679, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.3795830337886413, |
| "grad_norm": 0.780626654624939, |
| "learning_rate": 6.21135873472322e-05, |
| "loss": 0.0964, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.3803019410496046, |
| "grad_norm": 1.2292068004608154, |
| "learning_rate": 6.204169662113588e-05, |
| "loss": 0.2376, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.38102084831056793, |
| "grad_norm": 0.5723634958267212, |
| "learning_rate": 6.196980589503955e-05, |
| "loss": 0.1406, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.3817397555715313, |
| "grad_norm": 1.6000092029571533, |
| "learning_rate": 6.189791516894321e-05, |
| "loss": 0.1962, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.38245866283249463, |
| "grad_norm": 0.7020848393440247, |
| "learning_rate": 6.182602444284688e-05, |
| "loss": 0.1004, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.38317757009345793, |
| "grad_norm": 0.9194239974021912, |
| "learning_rate": 6.175413371675054e-05, |
| "loss": 0.2008, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.3838964773544213, |
| "grad_norm": 1.377670407295227, |
| "learning_rate": 6.16822429906542e-05, |
| "loss": 0.133, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.38461538461538464, |
| "grad_norm": 0.6439110040664673, |
| "learning_rate": 6.161035226455788e-05, |
| "loss": 0.184, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.38533429187634793, |
| "grad_norm": 0.7055622935295105, |
| "learning_rate": 6.153846153846155e-05, |
| "loss": 0.0987, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.3860531991373113, |
| "grad_norm": 0.8457476496696472, |
| "learning_rate": 6.146657081236521e-05, |
| "loss": 0.218, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.38677210639827464, |
| "grad_norm": 1.7412817478179932, |
| "learning_rate": 6.139468008626888e-05, |
| "loss": 0.1071, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.38749101365923794, |
| "grad_norm": 1.6000370979309082, |
| "learning_rate": 6.132278936017254e-05, |
| "loss": 0.1603, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.3882099209202013, |
| "grad_norm": 1.1800726652145386, |
| "learning_rate": 6.125089863407621e-05, |
| "loss": 0.0968, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.38892882818116464, |
| "grad_norm": 1.320655107498169, |
| "learning_rate": 6.117900790797987e-05, |
| "loss": 0.2518, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.38964773544212794, |
| "grad_norm": 0.8056125044822693, |
| "learning_rate": 6.110711718188355e-05, |
| "loss": 0.1505, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.3903666427030913, |
| "grad_norm": 0.6257234215736389, |
| "learning_rate": 6.103522645578722e-05, |
| "loss": 0.1105, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.39108554996405465, |
| "grad_norm": 1.016261100769043, |
| "learning_rate": 6.096333572969087e-05, |
| "loss": 0.1194, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.391804457225018, |
| "grad_norm": 0.4871998727321625, |
| "learning_rate": 6.089144500359454e-05, |
| "loss": 0.1126, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.3925233644859813, |
| "grad_norm": 0.5652322769165039, |
| "learning_rate": 6.08195542774982e-05, |
| "loss": 0.1466, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.39324227174694465, |
| "grad_norm": 0.7007426619529724, |
| "learning_rate": 6.074766355140187e-05, |
| "loss": 0.1083, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.393961179007908, |
| "grad_norm": 0.8851130604743958, |
| "learning_rate": 6.0675772825305545e-05, |
| "loss": 0.1624, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.3946800862688713, |
| "grad_norm": 0.86240553855896, |
| "learning_rate": 6.060388209920921e-05, |
| "loss": 0.1748, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.39539899352983465, |
| "grad_norm": 0.9123683571815491, |
| "learning_rate": 6.0531991373112875e-05, |
| "loss": 0.0903, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.396117900790798, |
| "grad_norm": 0.5733093619346619, |
| "learning_rate": 6.046010064701654e-05, |
| "loss": 0.0969, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.3968368080517613, |
| "grad_norm": 4.72888708114624, |
| "learning_rate": 6.0388209920920204e-05, |
| "loss": 0.4002, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.39755571531272466, |
| "grad_norm": 0.4290800094604492, |
| "learning_rate": 6.031631919482387e-05, |
| "loss": 0.0837, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.398274622573688, |
| "grad_norm": 0.9166390895843506, |
| "learning_rate": 6.0244428468727533e-05, |
| "loss": 0.1113, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.3989935298346513, |
| "grad_norm": 0.44564205408096313, |
| "learning_rate": 6.0172537742631205e-05, |
| "loss": 0.1139, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.39971243709561466, |
| "grad_norm": 1.640791416168213, |
| "learning_rate": 6.010064701653487e-05, |
| "loss": 0.2098, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.400431344356578, |
| "grad_norm": 0.9517141580581665, |
| "learning_rate": 6.0028756290438534e-05, |
| "loss": 0.1351, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.40115025161754136, |
| "grad_norm": 1.3287968635559082, |
| "learning_rate": 5.99568655643422e-05, |
| "loss": 0.0997, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.40186915887850466, |
| "grad_norm": 0.48180484771728516, |
| "learning_rate": 5.9884974838245864e-05, |
| "loss": 0.1008, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.402588066139468, |
| "grad_norm": 0.8257919549942017, |
| "learning_rate": 5.981308411214953e-05, |
| "loss": 0.2259, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.40330697340043137, |
| "grad_norm": 1.2000943422317505, |
| "learning_rate": 5.9741193386053207e-05, |
| "loss": 0.2331, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.40402588066139467, |
| "grad_norm": 0.6794840693473816, |
| "learning_rate": 5.966930265995687e-05, |
| "loss": 0.1602, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.404744787922358, |
| "grad_norm": 1.0928236246109009, |
| "learning_rate": 5.9597411933860536e-05, |
| "loss": 0.1749, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.40546369518332137, |
| "grad_norm": 1.5526889562606812, |
| "learning_rate": 5.95255212077642e-05, |
| "loss": 0.1679, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.40618260244428467, |
| "grad_norm": 1.131243348121643, |
| "learning_rate": 5.9453630481667865e-05, |
| "loss": 0.243, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.406901509705248, |
| "grad_norm": 0.720497727394104, |
| "learning_rate": 5.938173975557153e-05, |
| "loss": 0.1605, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.4076204169662114, |
| "grad_norm": 1.18629789352417, |
| "learning_rate": 5.9309849029475195e-05, |
| "loss": 0.2818, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.40833932422717467, |
| "grad_norm": 0.7498906850814819, |
| "learning_rate": 5.923795830337887e-05, |
| "loss": 0.1231, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.409058231488138, |
| "grad_norm": 1.3997340202331543, |
| "learning_rate": 5.916606757728254e-05, |
| "loss": 0.2796, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.4097771387491014, |
| "grad_norm": 2.068100690841675, |
| "learning_rate": 5.90941768511862e-05, |
| "loss": 0.1586, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.4104960460100647, |
| "grad_norm": 0.7314684391021729, |
| "learning_rate": 5.902228612508987e-05, |
| "loss": 0.1388, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.411214953271028, |
| "grad_norm": 0.6927562952041626, |
| "learning_rate": 5.895039539899353e-05, |
| "loss": 0.1396, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.4119338605319914, |
| "grad_norm": 0.8310578465461731, |
| "learning_rate": 5.8878504672897196e-05, |
| "loss": 0.1419, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.41265276779295473, |
| "grad_norm": 0.7011169195175171, |
| "learning_rate": 5.880661394680087e-05, |
| "loss": 0.1106, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.41337167505391803, |
| "grad_norm": 1.0497771501541138, |
| "learning_rate": 5.873472322070453e-05, |
| "loss": 0.0876, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.4140905823148814, |
| "grad_norm": 0.6327552199363708, |
| "learning_rate": 5.86628324946082e-05, |
| "loss": 0.1308, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.41480948957584474, |
| "grad_norm": 0.594848096370697, |
| "learning_rate": 5.859094176851186e-05, |
| "loss": 0.1071, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.41552839683680803, |
| "grad_norm": 0.6702591180801392, |
| "learning_rate": 5.851905104241553e-05, |
| "loss": 0.111, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.4162473040977714, |
| "grad_norm": 1.3994640111923218, |
| "learning_rate": 5.844716031631919e-05, |
| "loss": 0.1677, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.41696621135873474, |
| "grad_norm": 0.7137076258659363, |
| "learning_rate": 5.8375269590222856e-05, |
| "loss": 0.1021, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.41768511861969804, |
| "grad_norm": 0.7051552534103394, |
| "learning_rate": 5.8303378864126534e-05, |
| "loss": 0.0809, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.4184040258806614, |
| "grad_norm": 0.750128984451294, |
| "learning_rate": 5.82314881380302e-05, |
| "loss": 0.153, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.41912293314162474, |
| "grad_norm": 1.175628900527954, |
| "learning_rate": 5.8159597411933864e-05, |
| "loss": 0.1542, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.41984184040258804, |
| "grad_norm": 1.0719283819198608, |
| "learning_rate": 5.808770668583753e-05, |
| "loss": 0.2254, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.4205607476635514, |
| "grad_norm": 0.8681702613830566, |
| "learning_rate": 5.801581595974119e-05, |
| "loss": 0.1069, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.42127965492451475, |
| "grad_norm": 0.8094863891601562, |
| "learning_rate": 5.794392523364486e-05, |
| "loss": 0.1535, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.4219985621854781, |
| "grad_norm": 0.7679921388626099, |
| "learning_rate": 5.7872034507548536e-05, |
| "loss": 0.1175, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.4227174694464414, |
| "grad_norm": 0.6886312961578369, |
| "learning_rate": 5.78001437814522e-05, |
| "loss": 0.084, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.42343637670740475, |
| "grad_norm": 0.9979172945022583, |
| "learning_rate": 5.7728253055355865e-05, |
| "loss": 0.116, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.4241552839683681, |
| "grad_norm": 2.0505921840667725, |
| "learning_rate": 5.765636232925953e-05, |
| "loss": 0.2421, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.4248741912293314, |
| "grad_norm": 1.1256647109985352, |
| "learning_rate": 5.7584471603163195e-05, |
| "loss": 0.1118, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.42559309849029475, |
| "grad_norm": 0.830562174320221, |
| "learning_rate": 5.751258087706686e-05, |
| "loss": 0.1129, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.4263120057512581, |
| "grad_norm": 0.8950141668319702, |
| "learning_rate": 5.744069015097053e-05, |
| "loss": 0.1468, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.4270309130122214, |
| "grad_norm": 0.4330369234085083, |
| "learning_rate": 5.7368799424874196e-05, |
| "loss": 0.1272, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.42774982027318476, |
| "grad_norm": 0.8154317736625671, |
| "learning_rate": 5.729690869877786e-05, |
| "loss": 0.0936, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.4284687275341481, |
| "grad_norm": 1.090325951576233, |
| "learning_rate": 5.7225017972681525e-05, |
| "loss": 0.1479, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.4291876347951114, |
| "grad_norm": 0.861066997051239, |
| "learning_rate": 5.715312724658519e-05, |
| "loss": 0.0994, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.42990654205607476, |
| "grad_norm": 0.6655673384666443, |
| "learning_rate": 5.7081236520488854e-05, |
| "loss": 0.095, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.4306254493170381, |
| "grad_norm": 1.1570032835006714, |
| "learning_rate": 5.700934579439252e-05, |
| "loss": 0.194, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.43134435657800146, |
| "grad_norm": 0.6067618727684021, |
| "learning_rate": 5.69374550682962e-05, |
| "loss": 0.1134, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.43134435657800146, |
| "eval_accuracy": 0.935278058052063, |
| "eval_loss": 0.1366163194179535, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6572470664978027, |
| "eval_precision": 0.9278779625892639, |
| "eval_recall": 0.9723837375640869, |
| "eval_runtime": 602.7804, |
| "eval_samples_per_second": 4.617, |
| "eval_steps_per_second": 0.577, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.43206326383896476, |
| "grad_norm": 0.4303855001926422, |
| "learning_rate": 5.686556434219986e-05, |
| "loss": 0.111, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.4327821710999281, |
| "grad_norm": 0.7033389806747437, |
| "learning_rate": 5.679367361610353e-05, |
| "loss": 0.1176, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.43350107836089147, |
| "grad_norm": 1.008163571357727, |
| "learning_rate": 5.672178289000719e-05, |
| "loss": 0.1428, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.43421998562185476, |
| "grad_norm": 1.0969762802124023, |
| "learning_rate": 5.6649892163910856e-05, |
| "loss": 0.2394, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.4349388928828181, |
| "grad_norm": 1.8316993713378906, |
| "learning_rate": 5.657800143781452e-05, |
| "loss": 0.1898, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.43565780014378147, |
| "grad_norm": 2.6679141521453857, |
| "learning_rate": 5.65061107117182e-05, |
| "loss": 0.3538, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.43637670740474477, |
| "grad_norm": 0.7877079248428345, |
| "learning_rate": 5.6434219985621864e-05, |
| "loss": 0.1176, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.4370956146657081, |
| "grad_norm": 0.9411882758140564, |
| "learning_rate": 5.636232925952553e-05, |
| "loss": 0.0893, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.4378145219266715, |
| "grad_norm": 1.1060857772827148, |
| "learning_rate": 5.629043853342919e-05, |
| "loss": 0.1644, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.43853342918763477, |
| "grad_norm": 0.8070788383483887, |
| "learning_rate": 5.621854780733286e-05, |
| "loss": 0.0917, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.4392523364485981, |
| "grad_norm": 0.6507598757743835, |
| "learning_rate": 5.614665708123652e-05, |
| "loss": 0.1176, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.4399712437095615, |
| "grad_norm": 0.756272554397583, |
| "learning_rate": 5.607476635514019e-05, |
| "loss": 0.1023, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.44069015097052483, |
| "grad_norm": 0.6685004234313965, |
| "learning_rate": 5.600287562904386e-05, |
| "loss": 0.0911, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.4414090582314881, |
| "grad_norm": 0.6946046352386475, |
| "learning_rate": 5.593098490294752e-05, |
| "loss": 0.1393, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.4421279654924515, |
| "grad_norm": 1.3382245302200317, |
| "learning_rate": 5.585909417685119e-05, |
| "loss": 0.1999, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.44284687275341483, |
| "grad_norm": 0.8030561804771423, |
| "learning_rate": 5.578720345075485e-05, |
| "loss": 0.1523, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.44356578001437813, |
| "grad_norm": 1.3797811269760132, |
| "learning_rate": 5.571531272465852e-05, |
| "loss": 0.1644, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.4442846872753415, |
| "grad_norm": 1.432765007019043, |
| "learning_rate": 5.564342199856218e-05, |
| "loss": 0.1535, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.44500359453630484, |
| "grad_norm": 0.5737884044647217, |
| "learning_rate": 5.557153127246586e-05, |
| "loss": 0.0795, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.44572250179726813, |
| "grad_norm": 0.4333879053592682, |
| "learning_rate": 5.5499640546369525e-05, |
| "loss": 0.0963, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.4464414090582315, |
| "grad_norm": 0.9941200613975525, |
| "learning_rate": 5.542774982027319e-05, |
| "loss": 0.18, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.44716031631919484, |
| "grad_norm": 1.332000732421875, |
| "learning_rate": 5.5355859094176854e-05, |
| "loss": 0.2362, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.44787922358015814, |
| "grad_norm": 2.3299262523651123, |
| "learning_rate": 5.528396836808052e-05, |
| "loss": 0.3902, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.4485981308411215, |
| "grad_norm": 0.7755465507507324, |
| "learning_rate": 5.5212077641984184e-05, |
| "loss": 0.1603, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.44931703810208484, |
| "grad_norm": 0.6063105463981628, |
| "learning_rate": 5.514018691588785e-05, |
| "loss": 0.1278, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.45003594536304814, |
| "grad_norm": 1.2312532663345337, |
| "learning_rate": 5.5068296189791527e-05, |
| "loss": 0.1754, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.4507548526240115, |
| "grad_norm": 2.011659860610962, |
| "learning_rate": 5.499640546369519e-05, |
| "loss": 0.3376, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.45147375988497485, |
| "grad_norm": 0.5915288329124451, |
| "learning_rate": 5.4924514737598856e-05, |
| "loss": 0.1057, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.4521926671459382, |
| "grad_norm": 1.204221248626709, |
| "learning_rate": 5.485262401150252e-05, |
| "loss": 0.072, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.4529115744069015, |
| "grad_norm": 0.8110437989234924, |
| "learning_rate": 5.4780733285406185e-05, |
| "loss": 0.182, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.45363048166786485, |
| "grad_norm": 1.9918159246444702, |
| "learning_rate": 5.470884255930985e-05, |
| "loss": 0.1259, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.4543493889288282, |
| "grad_norm": 1.0023044347763062, |
| "learning_rate": 5.463695183321352e-05, |
| "loss": 0.1762, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.4550682961897915, |
| "grad_norm": 0.952862560749054, |
| "learning_rate": 5.4565061107117186e-05, |
| "loss": 0.1746, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.45578720345075485, |
| "grad_norm": 1.728900671005249, |
| "learning_rate": 5.449317038102085e-05, |
| "loss": 0.1492, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.4565061107117182, |
| "grad_norm": 0.798629105091095, |
| "learning_rate": 5.4421279654924516e-05, |
| "loss": 0.1876, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.4572250179726815, |
| "grad_norm": 0.7590505480766296, |
| "learning_rate": 5.434938892882818e-05, |
| "loss": 0.1492, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.45794392523364486, |
| "grad_norm": 1.2391388416290283, |
| "learning_rate": 5.4277498202731845e-05, |
| "loss": 0.0846, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.4586628324946082, |
| "grad_norm": 0.7551385760307312, |
| "learning_rate": 5.420560747663551e-05, |
| "loss": 0.1517, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.4593817397555715, |
| "grad_norm": 0.7110326290130615, |
| "learning_rate": 5.413371675053919e-05, |
| "loss": 0.0716, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.46010064701653486, |
| "grad_norm": 1.065392017364502, |
| "learning_rate": 5.406182602444285e-05, |
| "loss": 0.1583, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.4608195542774982, |
| "grad_norm": 1.618969202041626, |
| "learning_rate": 5.398993529834652e-05, |
| "loss": 0.1094, |
| "step": 641 |
| }, |
| { |
| "epoch": 0.46153846153846156, |
| "grad_norm": 0.6073803901672363, |
| "learning_rate": 5.391804457225018e-05, |
| "loss": 0.1267, |
| "step": 642 |
| }, |
| { |
| "epoch": 0.46225736879942486, |
| "grad_norm": 1.156285047531128, |
| "learning_rate": 5.384615384615385e-05, |
| "loss": 0.2088, |
| "step": 643 |
| }, |
| { |
| "epoch": 0.4629762760603882, |
| "grad_norm": 1.3841142654418945, |
| "learning_rate": 5.377426312005751e-05, |
| "loss": 0.1774, |
| "step": 644 |
| }, |
| { |
| "epoch": 0.46369518332135157, |
| "grad_norm": 0.8162432909011841, |
| "learning_rate": 5.370237239396119e-05, |
| "loss": 0.1117, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.46441409058231486, |
| "grad_norm": 0.8136155009269714, |
| "learning_rate": 5.3630481667864854e-05, |
| "loss": 0.1961, |
| "step": 646 |
| }, |
| { |
| "epoch": 0.4651329978432782, |
| "grad_norm": 0.9473450779914856, |
| "learning_rate": 5.355859094176852e-05, |
| "loss": 0.1263, |
| "step": 647 |
| }, |
| { |
| "epoch": 0.46585190510424157, |
| "grad_norm": 1.213666319847107, |
| "learning_rate": 5.3486700215672184e-05, |
| "loss": 0.1817, |
| "step": 648 |
| }, |
| { |
| "epoch": 0.46657081236520487, |
| "grad_norm": 0.7157108783721924, |
| "learning_rate": 5.341480948957585e-05, |
| "loss": 0.0914, |
| "step": 649 |
| }, |
| { |
| "epoch": 0.4672897196261682, |
| "grad_norm": 1.8709887266159058, |
| "learning_rate": 5.334291876347951e-05, |
| "loss": 0.125, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.4680086268871316, |
| "grad_norm": 0.8363114595413208, |
| "learning_rate": 5.327102803738318e-05, |
| "loss": 0.1664, |
| "step": 651 |
| }, |
| { |
| "epoch": 0.46872753414809487, |
| "grad_norm": 0.9694750905036926, |
| "learning_rate": 5.319913731128685e-05, |
| "loss": 0.2049, |
| "step": 652 |
| }, |
| { |
| "epoch": 0.4694464414090582, |
| "grad_norm": 0.6881033778190613, |
| "learning_rate": 5.3127246585190514e-05, |
| "loss": 0.0849, |
| "step": 653 |
| }, |
| { |
| "epoch": 0.4701653486700216, |
| "grad_norm": 1.2331490516662598, |
| "learning_rate": 5.305535585909418e-05, |
| "loss": 0.11, |
| "step": 654 |
| }, |
| { |
| "epoch": 0.47088425593098493, |
| "grad_norm": 1.0406458377838135, |
| "learning_rate": 5.298346513299784e-05, |
| "loss": 0.1549, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.4716031631919482, |
| "grad_norm": 0.5542729496955872, |
| "learning_rate": 5.291157440690151e-05, |
| "loss": 0.1178, |
| "step": 656 |
| }, |
| { |
| "epoch": 0.4723220704529116, |
| "grad_norm": 0.9304601550102234, |
| "learning_rate": 5.283968368080517e-05, |
| "loss": 0.1555, |
| "step": 657 |
| }, |
| { |
| "epoch": 0.47304097771387493, |
| "grad_norm": 0.4800335466861725, |
| "learning_rate": 5.276779295470885e-05, |
| "loss": 0.1206, |
| "step": 658 |
| }, |
| { |
| "epoch": 0.47375988497483823, |
| "grad_norm": 0.5330154299736023, |
| "learning_rate": 5.2695902228612516e-05, |
| "loss": 0.1064, |
| "step": 659 |
| }, |
| { |
| "epoch": 0.4744787922358016, |
| "grad_norm": 1.4122358560562134, |
| "learning_rate": 5.262401150251618e-05, |
| "loss": 0.1485, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.47519769949676494, |
| "grad_norm": 0.44964709877967834, |
| "learning_rate": 5.2552120776419845e-05, |
| "loss": 0.1013, |
| "step": 661 |
| }, |
| { |
| "epoch": 0.47591660675772823, |
| "grad_norm": 0.7293593883514404, |
| "learning_rate": 5.248023005032351e-05, |
| "loss": 0.1516, |
| "step": 662 |
| }, |
| { |
| "epoch": 0.4766355140186916, |
| "grad_norm": 1.1407926082611084, |
| "learning_rate": 5.2408339324227174e-05, |
| "loss": 0.1083, |
| "step": 663 |
| }, |
| { |
| "epoch": 0.47735442127965494, |
| "grad_norm": 1.0616534948349, |
| "learning_rate": 5.233644859813084e-05, |
| "loss": 0.2365, |
| "step": 664 |
| }, |
| { |
| "epoch": 0.47807332854061824, |
| "grad_norm": 1.0046387910842896, |
| "learning_rate": 5.226455787203452e-05, |
| "loss": 0.1227, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.4787922358015816, |
| "grad_norm": 0.6443594694137573, |
| "learning_rate": 5.219266714593818e-05, |
| "loss": 0.1766, |
| "step": 666 |
| }, |
| { |
| "epoch": 0.47951114306254494, |
| "grad_norm": 1.1169304847717285, |
| "learning_rate": 5.212077641984185e-05, |
| "loss": 0.1451, |
| "step": 667 |
| }, |
| { |
| "epoch": 0.4802300503235083, |
| "grad_norm": 0.6844356656074524, |
| "learning_rate": 5.204888569374551e-05, |
| "loss": 0.0947, |
| "step": 668 |
| }, |
| { |
| "epoch": 0.4809489575844716, |
| "grad_norm": 0.8872304558753967, |
| "learning_rate": 5.1976994967649176e-05, |
| "loss": 0.113, |
| "step": 669 |
| }, |
| { |
| "epoch": 0.48166786484543495, |
| "grad_norm": 0.614462673664093, |
| "learning_rate": 5.190510424155284e-05, |
| "loss": 0.1705, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.4823867721063983, |
| "grad_norm": 0.5005216598510742, |
| "learning_rate": 5.183321351545651e-05, |
| "loss": 0.0981, |
| "step": 671 |
| }, |
| { |
| "epoch": 0.4831056793673616, |
| "grad_norm": 0.6912380456924438, |
| "learning_rate": 5.176132278936018e-05, |
| "loss": 0.1171, |
| "step": 672 |
| }, |
| { |
| "epoch": 0.48382458662832495, |
| "grad_norm": 0.6769587397575378, |
| "learning_rate": 5.168943206326384e-05, |
| "loss": 0.185, |
| "step": 673 |
| }, |
| { |
| "epoch": 0.4845434938892883, |
| "grad_norm": 1.0327954292297363, |
| "learning_rate": 5.1617541337167506e-05, |
| "loss": 0.125, |
| "step": 674 |
| }, |
| { |
| "epoch": 0.4852624011502516, |
| "grad_norm": 1.8892732858657837, |
| "learning_rate": 5.154565061107117e-05, |
| "loss": 0.2393, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.48598130841121495, |
| "grad_norm": 0.5028102993965149, |
| "learning_rate": 5.1473759884974836e-05, |
| "loss": 0.1455, |
| "step": 676 |
| }, |
| { |
| "epoch": 0.4867002156721783, |
| "grad_norm": 1.4603813886642456, |
| "learning_rate": 5.14018691588785e-05, |
| "loss": 0.1954, |
| "step": 677 |
| }, |
| { |
| "epoch": 0.4874191229331416, |
| "grad_norm": 0.6014376282691956, |
| "learning_rate": 5.132997843278218e-05, |
| "loss": 0.1548, |
| "step": 678 |
| }, |
| { |
| "epoch": 0.48813803019410495, |
| "grad_norm": 0.7407705187797546, |
| "learning_rate": 5.125808770668584e-05, |
| "loss": 0.1471, |
| "step": 679 |
| }, |
| { |
| "epoch": 0.4888569374550683, |
| "grad_norm": 0.48685091733932495, |
| "learning_rate": 5.118619698058951e-05, |
| "loss": 0.1359, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.4895758447160316, |
| "grad_norm": 0.6102830767631531, |
| "learning_rate": 5.111430625449317e-05, |
| "loss": 0.1426, |
| "step": 681 |
| }, |
| { |
| "epoch": 0.49029475197699496, |
| "grad_norm": 0.9747264981269836, |
| "learning_rate": 5.104241552839684e-05, |
| "loss": 0.129, |
| "step": 682 |
| }, |
| { |
| "epoch": 0.4910136592379583, |
| "grad_norm": 0.7752413153648376, |
| "learning_rate": 5.09705248023005e-05, |
| "loss": 0.1013, |
| "step": 683 |
| }, |
| { |
| "epoch": 0.49173256649892166, |
| "grad_norm": 1.4923985004425049, |
| "learning_rate": 5.089863407620418e-05, |
| "loss": 0.314, |
| "step": 684 |
| }, |
| { |
| "epoch": 0.49245147375988496, |
| "grad_norm": 1.1001107692718506, |
| "learning_rate": 5.0826743350107845e-05, |
| "loss": 0.1286, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.4931703810208483, |
| "grad_norm": 1.2768259048461914, |
| "learning_rate": 5.075485262401151e-05, |
| "loss": 0.1833, |
| "step": 686 |
| }, |
| { |
| "epoch": 0.49388928828181167, |
| "grad_norm": 0.5535606145858765, |
| "learning_rate": 5.0682961897915174e-05, |
| "loss": 0.1334, |
| "step": 687 |
| }, |
| { |
| "epoch": 0.49460819554277496, |
| "grad_norm": 0.5561098456382751, |
| "learning_rate": 5.061107117181884e-05, |
| "loss": 0.088, |
| "step": 688 |
| }, |
| { |
| "epoch": 0.4953271028037383, |
| "grad_norm": 0.6551526188850403, |
| "learning_rate": 5.0539180445722504e-05, |
| "loss": 0.1461, |
| "step": 689 |
| }, |
| { |
| "epoch": 0.49604601006470167, |
| "grad_norm": 0.7609792947769165, |
| "learning_rate": 5.046728971962617e-05, |
| "loss": 0.187, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.49676491732566497, |
| "grad_norm": 1.0811909437179565, |
| "learning_rate": 5.039539899352984e-05, |
| "loss": 0.1364, |
| "step": 691 |
| }, |
| { |
| "epoch": 0.4974838245866283, |
| "grad_norm": 0.5842167139053345, |
| "learning_rate": 5.0323508267433505e-05, |
| "loss": 0.1031, |
| "step": 692 |
| }, |
| { |
| "epoch": 0.4982027318475917, |
| "grad_norm": 0.7106841206550598, |
| "learning_rate": 5.025161754133717e-05, |
| "loss": 0.108, |
| "step": 693 |
| }, |
| { |
| "epoch": 0.49892163910855497, |
| "grad_norm": 1.2816544771194458, |
| "learning_rate": 5.0179726815240834e-05, |
| "loss": 0.1421, |
| "step": 694 |
| }, |
| { |
| "epoch": 0.4996405463695183, |
| "grad_norm": 1.206875205039978, |
| "learning_rate": 5.01078360891445e-05, |
| "loss": 0.1859, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.5003594536304816, |
| "grad_norm": 1.0266493558883667, |
| "learning_rate": 5.003594536304816e-05, |
| "loss": 0.1408, |
| "step": 696 |
| }, |
| { |
| "epoch": 0.501078360891445, |
| "grad_norm": 0.8125544190406799, |
| "learning_rate": 4.9964054636951835e-05, |
| "loss": 0.1115, |
| "step": 697 |
| }, |
| { |
| "epoch": 0.5017972681524083, |
| "grad_norm": 0.5617983937263489, |
| "learning_rate": 4.98921639108555e-05, |
| "loss": 0.0848, |
| "step": 698 |
| }, |
| { |
| "epoch": 0.5025161754133717, |
| "grad_norm": 2.0314903259277344, |
| "learning_rate": 4.982027318475917e-05, |
| "loss": 0.1919, |
| "step": 699 |
| }, |
| { |
| "epoch": 0.503235082674335, |
| "grad_norm": 2.6272692680358887, |
| "learning_rate": 4.9748382458662836e-05, |
| "loss": 0.2338, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.5039539899352984, |
| "grad_norm": 0.7299609780311584, |
| "learning_rate": 4.96764917325665e-05, |
| "loss": 0.1537, |
| "step": 701 |
| }, |
| { |
| "epoch": 0.5046728971962616, |
| "grad_norm": 2.1801068782806396, |
| "learning_rate": 4.960460100647017e-05, |
| "loss": 0.2158, |
| "step": 702 |
| }, |
| { |
| "epoch": 0.505391804457225, |
| "grad_norm": 0.8830720782279968, |
| "learning_rate": 4.9532710280373836e-05, |
| "loss": 0.1625, |
| "step": 703 |
| }, |
| { |
| "epoch": 0.5061107117181883, |
| "grad_norm": 2.140620231628418, |
| "learning_rate": 4.94608195542775e-05, |
| "loss": 0.2953, |
| "step": 704 |
| }, |
| { |
| "epoch": 0.5068296189791517, |
| "grad_norm": 0.6284368634223938, |
| "learning_rate": 4.9388928828181166e-05, |
| "loss": 0.1618, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.507548526240115, |
| "grad_norm": 0.6372418999671936, |
| "learning_rate": 4.931703810208483e-05, |
| "loss": 0.1731, |
| "step": 706 |
| }, |
| { |
| "epoch": 0.5082674335010784, |
| "grad_norm": 0.934623658657074, |
| "learning_rate": 4.9245147375988495e-05, |
| "loss": 0.2133, |
| "step": 707 |
| }, |
| { |
| "epoch": 0.5089863407620417, |
| "grad_norm": 0.6417623162269592, |
| "learning_rate": 4.917325664989217e-05, |
| "loss": 0.0725, |
| "step": 708 |
| }, |
| { |
| "epoch": 0.509705248023005, |
| "grad_norm": 0.49728158116340637, |
| "learning_rate": 4.910136592379583e-05, |
| "loss": 0.0925, |
| "step": 709 |
| }, |
| { |
| "epoch": 0.5104241552839683, |
| "grad_norm": 0.4544738829135895, |
| "learning_rate": 4.9029475197699496e-05, |
| "loss": 0.1183, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.5111430625449317, |
| "grad_norm": 0.7206600308418274, |
| "learning_rate": 4.895758447160316e-05, |
| "loss": 0.1379, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.511861969805895, |
| "grad_norm": 0.7607790231704712, |
| "learning_rate": 4.888569374550683e-05, |
| "loss": 0.172, |
| "step": 712 |
| }, |
| { |
| "epoch": 0.5125808770668584, |
| "grad_norm": 0.8933032155036926, |
| "learning_rate": 4.88138030194105e-05, |
| "loss": 0.1136, |
| "step": 713 |
| }, |
| { |
| "epoch": 0.5132997843278218, |
| "grad_norm": 0.5635328888893127, |
| "learning_rate": 4.874191229331416e-05, |
| "loss": 0.1271, |
| "step": 714 |
| }, |
| { |
| "epoch": 0.514018691588785, |
| "grad_norm": 0.6393849849700928, |
| "learning_rate": 4.867002156721783e-05, |
| "loss": 0.1481, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.5147375988497483, |
| "grad_norm": 0.5831479430198669, |
| "learning_rate": 4.85981308411215e-05, |
| "loss": 0.1142, |
| "step": 716 |
| }, |
| { |
| "epoch": 0.5154565061107117, |
| "grad_norm": 0.9907329082489014, |
| "learning_rate": 4.852624011502516e-05, |
| "loss": 0.2316, |
| "step": 717 |
| }, |
| { |
| "epoch": 0.516175413371675, |
| "grad_norm": 0.9734897017478943, |
| "learning_rate": 4.8454349388928834e-05, |
| "loss": 0.1154, |
| "step": 718 |
| }, |
| { |
| "epoch": 0.5168943206326384, |
| "grad_norm": 0.6978167295455933, |
| "learning_rate": 4.83824586628325e-05, |
| "loss": 0.1267, |
| "step": 719 |
| }, |
| { |
| "epoch": 0.5176132278936018, |
| "grad_norm": 1.4382848739624023, |
| "learning_rate": 4.831056793673616e-05, |
| "loss": 0.2097, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.5183321351545651, |
| "grad_norm": 0.9241942763328552, |
| "learning_rate": 4.8238677210639835e-05, |
| "loss": 0.1925, |
| "step": 721 |
| }, |
| { |
| "epoch": 0.5190510424155284, |
| "grad_norm": 0.7044418454170227, |
| "learning_rate": 4.81667864845435e-05, |
| "loss": 0.1156, |
| "step": 722 |
| }, |
| { |
| "epoch": 0.5197699496764917, |
| "grad_norm": 0.7493116855621338, |
| "learning_rate": 4.8094895758447164e-05, |
| "loss": 0.1668, |
| "step": 723 |
| }, |
| { |
| "epoch": 0.5204888569374551, |
| "grad_norm": 0.43593427538871765, |
| "learning_rate": 4.802300503235083e-05, |
| "loss": 0.0931, |
| "step": 724 |
| }, |
| { |
| "epoch": 0.5212077641984184, |
| "grad_norm": 0.9037162661552429, |
| "learning_rate": 4.7951114306254493e-05, |
| "loss": 0.1083, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.5219266714593818, |
| "grad_norm": 0.7315430641174316, |
| "learning_rate": 4.787922358015816e-05, |
| "loss": 0.1355, |
| "step": 726 |
| }, |
| { |
| "epoch": 0.5226455787203451, |
| "grad_norm": 0.8982157111167908, |
| "learning_rate": 4.780733285406182e-05, |
| "loss": 0.1081, |
| "step": 727 |
| }, |
| { |
| "epoch": 0.5233644859813084, |
| "grad_norm": 1.5741866827011108, |
| "learning_rate": 4.7735442127965494e-05, |
| "loss": 0.1362, |
| "step": 728 |
| }, |
| { |
| "epoch": 0.5240833932422717, |
| "grad_norm": 0.8856979012489319, |
| "learning_rate": 4.766355140186916e-05, |
| "loss": 0.0754, |
| "step": 729 |
| }, |
| { |
| "epoch": 0.5248023005032351, |
| "grad_norm": 0.6115644574165344, |
| "learning_rate": 4.7591660675772824e-05, |
| "loss": 0.1055, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.5255212077641984, |
| "grad_norm": 1.9210646152496338, |
| "learning_rate": 4.7519769949676495e-05, |
| "loss": 0.2276, |
| "step": 731 |
| }, |
| { |
| "epoch": 0.5262401150251618, |
| "grad_norm": 0.8324475288391113, |
| "learning_rate": 4.744787922358016e-05, |
| "loss": 0.1293, |
| "step": 732 |
| }, |
| { |
| "epoch": 0.5269590222861251, |
| "grad_norm": 0.8234529495239258, |
| "learning_rate": 4.7375988497483825e-05, |
| "loss": 0.113, |
| "step": 733 |
| }, |
| { |
| "epoch": 0.5276779295470885, |
| "grad_norm": 1.8535352945327759, |
| "learning_rate": 4.7304097771387496e-05, |
| "loss": 0.2818, |
| "step": 734 |
| }, |
| { |
| "epoch": 0.5283968368080517, |
| "grad_norm": 0.7025673389434814, |
| "learning_rate": 4.723220704529116e-05, |
| "loss": 0.0918, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.5291157440690151, |
| "grad_norm": 0.9405072927474976, |
| "learning_rate": 4.7160316319194825e-05, |
| "loss": 0.1032, |
| "step": 736 |
| }, |
| { |
| "epoch": 0.5298346513299784, |
| "grad_norm": 1.0390485525131226, |
| "learning_rate": 4.70884255930985e-05, |
| "loss": 0.1719, |
| "step": 737 |
| }, |
| { |
| "epoch": 0.5305535585909418, |
| "grad_norm": 0.7070658802986145, |
| "learning_rate": 4.701653486700216e-05, |
| "loss": 0.1264, |
| "step": 738 |
| }, |
| { |
| "epoch": 0.5312724658519051, |
| "grad_norm": 0.5530199408531189, |
| "learning_rate": 4.6944644140905826e-05, |
| "loss": 0.1511, |
| "step": 739 |
| }, |
| { |
| "epoch": 0.5319913731128685, |
| "grad_norm": 0.8632076382637024, |
| "learning_rate": 4.687275341480949e-05, |
| "loss": 0.0907, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.5327102803738317, |
| "grad_norm": 1.1795481443405151, |
| "learning_rate": 4.680086268871316e-05, |
| "loss": 0.2211, |
| "step": 741 |
| }, |
| { |
| "epoch": 0.5334291876347951, |
| "grad_norm": 0.6648495197296143, |
| "learning_rate": 4.672897196261683e-05, |
| "loss": 0.1095, |
| "step": 742 |
| }, |
| { |
| "epoch": 0.5341480948957584, |
| "grad_norm": 1.1685370206832886, |
| "learning_rate": 4.665708123652049e-05, |
| "loss": 0.2392, |
| "step": 743 |
| }, |
| { |
| "epoch": 0.5348670021567218, |
| "grad_norm": 1.0606144666671753, |
| "learning_rate": 4.6585190510424156e-05, |
| "loss": 0.1627, |
| "step": 744 |
| }, |
| { |
| "epoch": 0.5355859094176851, |
| "grad_norm": 1.2027032375335693, |
| "learning_rate": 4.651329978432782e-05, |
| "loss": 0.1773, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.5363048166786485, |
| "grad_norm": 0.8189231157302856, |
| "learning_rate": 4.6441409058231486e-05, |
| "loss": 0.1511, |
| "step": 746 |
| }, |
| { |
| "epoch": 0.5370237239396118, |
| "grad_norm": 1.100556492805481, |
| "learning_rate": 4.636951833213516e-05, |
| "loss": 0.1602, |
| "step": 747 |
| }, |
| { |
| "epoch": 0.5377426312005751, |
| "grad_norm": 0.7999088168144226, |
| "learning_rate": 4.629762760603882e-05, |
| "loss": 0.0875, |
| "step": 748 |
| }, |
| { |
| "epoch": 0.5384615384615384, |
| "grad_norm": 1.1077264547348022, |
| "learning_rate": 4.622573687994249e-05, |
| "loss": 0.1283, |
| "step": 749 |
| }, |
| { |
| "epoch": 0.5391804457225018, |
| "grad_norm": 0.6350802779197693, |
| "learning_rate": 4.615384615384616e-05, |
| "loss": 0.0871, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.5398993529834651, |
| "grad_norm": 0.7382431626319885, |
| "learning_rate": 4.608195542774982e-05, |
| "loss": 0.1256, |
| "step": 751 |
| }, |
| { |
| "epoch": 0.5406182602444285, |
| "grad_norm": 0.4942343235015869, |
| "learning_rate": 4.601006470165349e-05, |
| "loss": 0.0831, |
| "step": 752 |
| }, |
| { |
| "epoch": 0.5413371675053918, |
| "grad_norm": 0.7007788419723511, |
| "learning_rate": 4.593817397555715e-05, |
| "loss": 0.1414, |
| "step": 753 |
| }, |
| { |
| "epoch": 0.5420560747663551, |
| "grad_norm": 0.5757505893707275, |
| "learning_rate": 4.5866283249460824e-05, |
| "loss": 0.1172, |
| "step": 754 |
| }, |
| { |
| "epoch": 0.5427749820273184, |
| "grad_norm": 0.8589261770248413, |
| "learning_rate": 4.579439252336449e-05, |
| "loss": 0.1996, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.5434938892882818, |
| "grad_norm": 1.4060238599777222, |
| "learning_rate": 4.572250179726815e-05, |
| "loss": 0.241, |
| "step": 756 |
| }, |
| { |
| "epoch": 0.5442127965492451, |
| "grad_norm": 1.7768281698226929, |
| "learning_rate": 4.5650611071171825e-05, |
| "loss": 0.2292, |
| "step": 757 |
| }, |
| { |
| "epoch": 0.5449317038102085, |
| "grad_norm": 1.020683765411377, |
| "learning_rate": 4.557872034507549e-05, |
| "loss": 0.2003, |
| "step": 758 |
| }, |
| { |
| "epoch": 0.5456506110711719, |
| "grad_norm": 1.3989146947860718, |
| "learning_rate": 4.5506829618979154e-05, |
| "loss": 0.1097, |
| "step": 759 |
| }, |
| { |
| "epoch": 0.5463695183321352, |
| "grad_norm": 1.1977224349975586, |
| "learning_rate": 4.5434938892882825e-05, |
| "loss": 0.2283, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.5470884255930984, |
| "grad_norm": 0.6626883745193481, |
| "learning_rate": 4.536304816678649e-05, |
| "loss": 0.1272, |
| "step": 761 |
| }, |
| { |
| "epoch": 0.5478073328540618, |
| "grad_norm": 1.4868335723876953, |
| "learning_rate": 4.5291157440690155e-05, |
| "loss": 0.1721, |
| "step": 762 |
| }, |
| { |
| "epoch": 0.5485262401150252, |
| "grad_norm": 1.3106938600540161, |
| "learning_rate": 4.521926671459382e-05, |
| "loss": 0.1749, |
| "step": 763 |
| }, |
| { |
| "epoch": 0.5492451473759885, |
| "grad_norm": 0.5769041180610657, |
| "learning_rate": 4.5147375988497484e-05, |
| "loss": 0.1039, |
| "step": 764 |
| }, |
| { |
| "epoch": 0.5499640546369519, |
| "grad_norm": 1.991715669631958, |
| "learning_rate": 4.507548526240115e-05, |
| "loss": 0.2503, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.5506829618979152, |
| "grad_norm": 0.6081489324569702, |
| "learning_rate": 4.5003594536304814e-05, |
| "loss": 0.1425, |
| "step": 766 |
| }, |
| { |
| "epoch": 0.5514018691588785, |
| "grad_norm": 0.6078304648399353, |
| "learning_rate": 4.4931703810208485e-05, |
| "loss": 0.1013, |
| "step": 767 |
| }, |
| { |
| "epoch": 0.5521207764198418, |
| "grad_norm": 0.7449146509170532, |
| "learning_rate": 4.485981308411215e-05, |
| "loss": 0.1894, |
| "step": 768 |
| }, |
| { |
| "epoch": 0.5528396836808052, |
| "grad_norm": 0.5999807119369507, |
| "learning_rate": 4.4787922358015814e-05, |
| "loss": 0.1545, |
| "step": 769 |
| }, |
| { |
| "epoch": 0.5535585909417685, |
| "grad_norm": 1.127963662147522, |
| "learning_rate": 4.4716031631919486e-05, |
| "loss": 0.1205, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.5542774982027319, |
| "grad_norm": 0.5055525898933411, |
| "learning_rate": 4.464414090582315e-05, |
| "loss": 0.093, |
| "step": 771 |
| }, |
| { |
| "epoch": 0.5549964054636952, |
| "grad_norm": 0.7007226347923279, |
| "learning_rate": 4.4572250179726815e-05, |
| "loss": 0.1181, |
| "step": 772 |
| }, |
| { |
| "epoch": 0.5557153127246586, |
| "grad_norm": 2.161555290222168, |
| "learning_rate": 4.450035945363049e-05, |
| "loss": 0.2331, |
| "step": 773 |
| }, |
| { |
| "epoch": 0.5564342199856218, |
| "grad_norm": 0.7132800817489624, |
| "learning_rate": 4.442846872753415e-05, |
| "loss": 0.1566, |
| "step": 774 |
| }, |
| { |
| "epoch": 0.5571531272465852, |
| "grad_norm": 1.0908145904541016, |
| "learning_rate": 4.4356578001437816e-05, |
| "loss": 0.1167, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.5578720345075485, |
| "grad_norm": 1.0569723844528198, |
| "learning_rate": 4.428468727534149e-05, |
| "loss": 0.1746, |
| "step": 776 |
| }, |
| { |
| "epoch": 0.5585909417685119, |
| "grad_norm": 1.0999054908752441, |
| "learning_rate": 4.421279654924515e-05, |
| "loss": 0.1751, |
| "step": 777 |
| }, |
| { |
| "epoch": 0.5593098490294752, |
| "grad_norm": 0.9633118510246277, |
| "learning_rate": 4.414090582314882e-05, |
| "loss": 0.1028, |
| "step": 778 |
| }, |
| { |
| "epoch": 0.5600287562904386, |
| "grad_norm": 0.8112704157829285, |
| "learning_rate": 4.406901509705249e-05, |
| "loss": 0.2061, |
| "step": 779 |
| }, |
| { |
| "epoch": 0.5607476635514018, |
| "grad_norm": 0.7662085294723511, |
| "learning_rate": 4.399712437095615e-05, |
| "loss": 0.1139, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.5614665708123652, |
| "grad_norm": 1.3713099956512451, |
| "learning_rate": 4.392523364485982e-05, |
| "loss": 0.18, |
| "step": 781 |
| }, |
| { |
| "epoch": 0.5621854780733285, |
| "grad_norm": 1.0912457704544067, |
| "learning_rate": 4.385334291876348e-05, |
| "loss": 0.1553, |
| "step": 782 |
| }, |
| { |
| "epoch": 0.5629043853342919, |
| "grad_norm": 1.713066577911377, |
| "learning_rate": 4.378145219266715e-05, |
| "loss": 0.1708, |
| "step": 783 |
| }, |
| { |
| "epoch": 0.5636232925952552, |
| "grad_norm": 0.5156126618385315, |
| "learning_rate": 4.370956146657081e-05, |
| "loss": 0.0714, |
| "step": 784 |
| }, |
| { |
| "epoch": 0.5643421998562186, |
| "grad_norm": 0.8495787978172302, |
| "learning_rate": 4.3637670740474476e-05, |
| "loss": 0.1817, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.5650611071171819, |
| "grad_norm": 1.591055154800415, |
| "learning_rate": 4.356578001437815e-05, |
| "loss": 0.3295, |
| "step": 786 |
| }, |
| { |
| "epoch": 0.5657800143781452, |
| "grad_norm": 0.6924194693565369, |
| "learning_rate": 4.349388928828181e-05, |
| "loss": 0.1515, |
| "step": 787 |
| }, |
| { |
| "epoch": 0.5664989216391085, |
| "grad_norm": 0.6013800501823425, |
| "learning_rate": 4.342199856218548e-05, |
| "loss": 0.1374, |
| "step": 788 |
| }, |
| { |
| "epoch": 0.5672178289000719, |
| "grad_norm": 0.6470313668251038, |
| "learning_rate": 4.335010783608915e-05, |
| "loss": 0.1711, |
| "step": 789 |
| }, |
| { |
| "epoch": 0.5679367361610352, |
| "grad_norm": 0.7522788643836975, |
| "learning_rate": 4.3278217109992813e-05, |
| "loss": 0.1125, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.5686556434219986, |
| "grad_norm": 0.5333603024482727, |
| "learning_rate": 4.320632638389648e-05, |
| "loss": 0.1327, |
| "step": 791 |
| }, |
| { |
| "epoch": 0.5693745506829619, |
| "grad_norm": 0.8220458030700684, |
| "learning_rate": 4.313443565780015e-05, |
| "loss": 0.1766, |
| "step": 792 |
| }, |
| { |
| "epoch": 0.5700934579439252, |
| "grad_norm": 1.6966160535812378, |
| "learning_rate": 4.3062544931703814e-05, |
| "loss": 0.1266, |
| "step": 793 |
| }, |
| { |
| "epoch": 0.5708123652048885, |
| "grad_norm": 0.5166921615600586, |
| "learning_rate": 4.299065420560748e-05, |
| "loss": 0.1027, |
| "step": 794 |
| }, |
| { |
| "epoch": 0.5715312724658519, |
| "grad_norm": 1.0604074001312256, |
| "learning_rate": 4.2918763479511144e-05, |
| "loss": 0.1557, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.5722501797268152, |
| "grad_norm": 0.6916298270225525, |
| "learning_rate": 4.2846872753414815e-05, |
| "loss": 0.131, |
| "step": 796 |
| }, |
| { |
| "epoch": 0.5729690869877786, |
| "grad_norm": 1.0458672046661377, |
| "learning_rate": 4.277498202731848e-05, |
| "loss": 0.1987, |
| "step": 797 |
| }, |
| { |
| "epoch": 0.573687994248742, |
| "grad_norm": 0.7854750752449036, |
| "learning_rate": 4.2703091301222145e-05, |
| "loss": 0.1251, |
| "step": 798 |
| }, |
| { |
| "epoch": 0.5744069015097053, |
| "grad_norm": 0.6716087460517883, |
| "learning_rate": 4.263120057512581e-05, |
| "loss": 0.1163, |
| "step": 799 |
| }, |
| { |
| "epoch": 0.5751258087706685, |
| "grad_norm": 2.01017165184021, |
| "learning_rate": 4.2559309849029474e-05, |
| "loss": 0.2086, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.5751258087706685, |
| "eval_accuracy": 0.935278058052063, |
| "eval_loss": 0.13323242962360382, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6627165079116821, |
| "eval_precision": 0.9243466258049011, |
| "eval_recall": 0.9767441749572754, |
| "eval_runtime": 602.4644, |
| "eval_samples_per_second": 4.619, |
| "eval_steps_per_second": 0.578, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.5758447160316319, |
| "grad_norm": 0.6006670594215393, |
| "learning_rate": 4.248741912293314e-05, |
| "loss": 0.1118, |
| "step": 801 |
| }, |
| { |
| "epoch": 0.5765636232925953, |
| "grad_norm": 0.5900166034698486, |
| "learning_rate": 4.241552839683681e-05, |
| "loss": 0.1266, |
| "step": 802 |
| }, |
| { |
| "epoch": 0.5772825305535586, |
| "grad_norm": 0.776298463344574, |
| "learning_rate": 4.2343637670740475e-05, |
| "loss": 0.1913, |
| "step": 803 |
| }, |
| { |
| "epoch": 0.578001437814522, |
| "grad_norm": 1.318871259689331, |
| "learning_rate": 4.227174694464414e-05, |
| "loss": 0.1747, |
| "step": 804 |
| }, |
| { |
| "epoch": 0.5787203450754853, |
| "grad_norm": 1.4930387735366821, |
| "learning_rate": 4.219985621854781e-05, |
| "loss": 0.2106, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.5794392523364486, |
| "grad_norm": 0.6400780081748962, |
| "learning_rate": 4.2127965492451476e-05, |
| "loss": 0.1052, |
| "step": 806 |
| }, |
| { |
| "epoch": 0.5801581595974119, |
| "grad_norm": 1.0664925575256348, |
| "learning_rate": 4.205607476635514e-05, |
| "loss": 0.1764, |
| "step": 807 |
| }, |
| { |
| "epoch": 0.5808770668583753, |
| "grad_norm": 1.0541273355484009, |
| "learning_rate": 4.1984184040258805e-05, |
| "loss": 0.1616, |
| "step": 808 |
| }, |
| { |
| "epoch": 0.5815959741193386, |
| "grad_norm": 1.064652681350708, |
| "learning_rate": 4.1912293314162476e-05, |
| "loss": 0.1574, |
| "step": 809 |
| }, |
| { |
| "epoch": 0.582314881380302, |
| "grad_norm": 1.6458402872085571, |
| "learning_rate": 4.184040258806614e-05, |
| "loss": 0.1116, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.5830337886412653, |
| "grad_norm": 1.2804372310638428, |
| "learning_rate": 4.1768511861969806e-05, |
| "loss": 0.2654, |
| "step": 811 |
| }, |
| { |
| "epoch": 0.5837526959022286, |
| "grad_norm": 0.8314998149871826, |
| "learning_rate": 4.169662113587348e-05, |
| "loss": 0.1015, |
| "step": 812 |
| }, |
| { |
| "epoch": 0.5844716031631919, |
| "grad_norm": 0.8012036681175232, |
| "learning_rate": 4.162473040977714e-05, |
| "loss": 0.1615, |
| "step": 813 |
| }, |
| { |
| "epoch": 0.5851905104241553, |
| "grad_norm": 0.7409771084785461, |
| "learning_rate": 4.155283968368081e-05, |
| "loss": 0.1842, |
| "step": 814 |
| }, |
| { |
| "epoch": 0.5859094176851186, |
| "grad_norm": 1.1309490203857422, |
| "learning_rate": 4.148094895758448e-05, |
| "loss": 0.1238, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.586628324946082, |
| "grad_norm": 1.952232837677002, |
| "learning_rate": 4.140905823148814e-05, |
| "loss": 0.2051, |
| "step": 816 |
| }, |
| { |
| "epoch": 0.5873472322070453, |
| "grad_norm": 0.9179194569587708, |
| "learning_rate": 4.133716750539181e-05, |
| "loss": 0.1188, |
| "step": 817 |
| }, |
| { |
| "epoch": 0.5880661394680087, |
| "grad_norm": 0.5914908051490784, |
| "learning_rate": 4.126527677929547e-05, |
| "loss": 0.1075, |
| "step": 818 |
| }, |
| { |
| "epoch": 0.5887850467289719, |
| "grad_norm": 0.6003243327140808, |
| "learning_rate": 4.119338605319914e-05, |
| "loss": 0.1328, |
| "step": 819 |
| }, |
| { |
| "epoch": 0.5895039539899353, |
| "grad_norm": 0.5188689827919006, |
| "learning_rate": 4.11214953271028e-05, |
| "loss": 0.1119, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.5902228612508986, |
| "grad_norm": 1.6032989025115967, |
| "learning_rate": 4.1049604601006466e-05, |
| "loss": 0.2053, |
| "step": 821 |
| }, |
| { |
| "epoch": 0.590941768511862, |
| "grad_norm": 0.7089391350746155, |
| "learning_rate": 4.097771387491014e-05, |
| "loss": 0.155, |
| "step": 822 |
| }, |
| { |
| "epoch": 0.5916606757728253, |
| "grad_norm": 1.0963640213012695, |
| "learning_rate": 4.09058231488138e-05, |
| "loss": 0.162, |
| "step": 823 |
| }, |
| { |
| "epoch": 0.5923795830337887, |
| "grad_norm": 0.6405614018440247, |
| "learning_rate": 4.083393242271747e-05, |
| "loss": 0.1501, |
| "step": 824 |
| }, |
| { |
| "epoch": 0.5930984902947519, |
| "grad_norm": 0.801033079624176, |
| "learning_rate": 4.076204169662114e-05, |
| "loss": 0.1013, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.5938173975557153, |
| "grad_norm": 0.4776207506656647, |
| "learning_rate": 4.06901509705248e-05, |
| "loss": 0.0896, |
| "step": 826 |
| }, |
| { |
| "epoch": 0.5945363048166786, |
| "grad_norm": 0.9710808992385864, |
| "learning_rate": 4.061826024442847e-05, |
| "loss": 0.1447, |
| "step": 827 |
| }, |
| { |
| "epoch": 0.595255212077642, |
| "grad_norm": 0.7523439526557922, |
| "learning_rate": 4.054636951833214e-05, |
| "loss": 0.1488, |
| "step": 828 |
| }, |
| { |
| "epoch": 0.5959741193386053, |
| "grad_norm": 1.388016700744629, |
| "learning_rate": 4.0474478792235804e-05, |
| "loss": 0.2255, |
| "step": 829 |
| }, |
| { |
| "epoch": 0.5966930265995687, |
| "grad_norm": 0.5801938772201538, |
| "learning_rate": 4.040258806613947e-05, |
| "loss": 0.1055, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.597411933860532, |
| "grad_norm": 1.0491878986358643, |
| "learning_rate": 4.033069734004314e-05, |
| "loss": 0.2178, |
| "step": 831 |
| }, |
| { |
| "epoch": 0.5981308411214953, |
| "grad_norm": 0.7289500832557678, |
| "learning_rate": 4.0258806613946805e-05, |
| "loss": 0.1306, |
| "step": 832 |
| }, |
| { |
| "epoch": 0.5988497483824586, |
| "grad_norm": 3.3592841625213623, |
| "learning_rate": 4.018691588785047e-05, |
| "loss": 0.3017, |
| "step": 833 |
| }, |
| { |
| "epoch": 0.599568655643422, |
| "grad_norm": 0.5456299781799316, |
| "learning_rate": 4.011502516175414e-05, |
| "loss": 0.0784, |
| "step": 834 |
| }, |
| { |
| "epoch": 0.6002875629043853, |
| "grad_norm": 0.7469555139541626, |
| "learning_rate": 4.0043134435657806e-05, |
| "loss": 0.1392, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.6010064701653487, |
| "grad_norm": 1.229032278060913, |
| "learning_rate": 3.997124370956147e-05, |
| "loss": 0.1407, |
| "step": 836 |
| }, |
| { |
| "epoch": 0.601725377426312, |
| "grad_norm": 0.5570354461669922, |
| "learning_rate": 3.9899352983465135e-05, |
| "loss": 0.1017, |
| "step": 837 |
| }, |
| { |
| "epoch": 0.6024442846872753, |
| "grad_norm": 0.7418543696403503, |
| "learning_rate": 3.98274622573688e-05, |
| "loss": 0.0562, |
| "step": 838 |
| }, |
| { |
| "epoch": 0.6031631919482386, |
| "grad_norm": 0.7655563354492188, |
| "learning_rate": 3.9755571531272465e-05, |
| "loss": 0.1667, |
| "step": 839 |
| }, |
| { |
| "epoch": 0.603882099209202, |
| "grad_norm": 0.8355135321617126, |
| "learning_rate": 3.968368080517613e-05, |
| "loss": 0.0933, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.6046010064701653, |
| "grad_norm": 0.4018462300300598, |
| "learning_rate": 3.96117900790798e-05, |
| "loss": 0.0958, |
| "step": 841 |
| }, |
| { |
| "epoch": 0.6053199137311287, |
| "grad_norm": 0.47250309586524963, |
| "learning_rate": 3.9539899352983465e-05, |
| "loss": 0.0965, |
| "step": 842 |
| }, |
| { |
| "epoch": 0.606038820992092, |
| "grad_norm": 0.4766501188278198, |
| "learning_rate": 3.946800862688713e-05, |
| "loss": 0.0902, |
| "step": 843 |
| }, |
| { |
| "epoch": 0.6067577282530554, |
| "grad_norm": 0.43635791540145874, |
| "learning_rate": 3.93961179007908e-05, |
| "loss": 0.0988, |
| "step": 844 |
| }, |
| { |
| "epoch": 0.6074766355140186, |
| "grad_norm": 0.846659243106842, |
| "learning_rate": 3.9324227174694466e-05, |
| "loss": 0.1371, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.608195542774982, |
| "grad_norm": 1.3217209577560425, |
| "learning_rate": 3.925233644859813e-05, |
| "loss": 0.1699, |
| "step": 846 |
| }, |
| { |
| "epoch": 0.6089144500359454, |
| "grad_norm": 0.6459369659423828, |
| "learning_rate": 3.91804457225018e-05, |
| "loss": 0.0599, |
| "step": 847 |
| }, |
| { |
| "epoch": 0.6096333572969087, |
| "grad_norm": 1.2141060829162598, |
| "learning_rate": 3.910855499640547e-05, |
| "loss": 0.1451, |
| "step": 848 |
| }, |
| { |
| "epoch": 0.6103522645578721, |
| "grad_norm": 0.6874876618385315, |
| "learning_rate": 3.903666427030913e-05, |
| "loss": 0.1281, |
| "step": 849 |
| }, |
| { |
| "epoch": 0.6110711718188354, |
| "grad_norm": 0.5104864835739136, |
| "learning_rate": 3.8964773544212796e-05, |
| "loss": 0.0772, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.6117900790797987, |
| "grad_norm": 0.5834945440292358, |
| "learning_rate": 3.889288281811647e-05, |
| "loss": 0.1016, |
| "step": 851 |
| }, |
| { |
| "epoch": 0.612508986340762, |
| "grad_norm": 3.122201442718506, |
| "learning_rate": 3.882099209202013e-05, |
| "loss": 0.369, |
| "step": 852 |
| }, |
| { |
| "epoch": 0.6132278936017254, |
| "grad_norm": 1.7977139949798584, |
| "learning_rate": 3.87491013659238e-05, |
| "loss": 0.1376, |
| "step": 853 |
| }, |
| { |
| "epoch": 0.6139468008626887, |
| "grad_norm": 0.6076405644416809, |
| "learning_rate": 3.867721063982747e-05, |
| "loss": 0.0833, |
| "step": 854 |
| }, |
| { |
| "epoch": 0.6146657081236521, |
| "grad_norm": 0.9768205881118774, |
| "learning_rate": 3.8605319913731133e-05, |
| "loss": 0.153, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.6153846153846154, |
| "grad_norm": 2.236875057220459, |
| "learning_rate": 3.85334291876348e-05, |
| "loss": 0.3848, |
| "step": 856 |
| }, |
| { |
| "epoch": 0.6161035226455788, |
| "grad_norm": 1.3891701698303223, |
| "learning_rate": 3.846153846153846e-05, |
| "loss": 0.1796, |
| "step": 857 |
| }, |
| { |
| "epoch": 0.616822429906542, |
| "grad_norm": 2.8574094772338867, |
| "learning_rate": 3.838964773544213e-05, |
| "loss": 0.3733, |
| "step": 858 |
| }, |
| { |
| "epoch": 0.6175413371675054, |
| "grad_norm": 1.3593533039093018, |
| "learning_rate": 3.831775700934579e-05, |
| "loss": 0.272, |
| "step": 859 |
| }, |
| { |
| "epoch": 0.6182602444284687, |
| "grad_norm": 0.6665904521942139, |
| "learning_rate": 3.8245866283249464e-05, |
| "loss": 0.1101, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.6189791516894321, |
| "grad_norm": 0.6988992094993591, |
| "learning_rate": 3.817397555715313e-05, |
| "loss": 0.1463, |
| "step": 861 |
| }, |
| { |
| "epoch": 0.6196980589503954, |
| "grad_norm": 0.9991676211357117, |
| "learning_rate": 3.810208483105679e-05, |
| "loss": 0.2135, |
| "step": 862 |
| }, |
| { |
| "epoch": 0.6204169662113588, |
| "grad_norm": 0.8370771408081055, |
| "learning_rate": 3.803019410496046e-05, |
| "loss": 0.0998, |
| "step": 863 |
| }, |
| { |
| "epoch": 0.621135873472322, |
| "grad_norm": 1.0214239358901978, |
| "learning_rate": 3.795830337886413e-05, |
| "loss": 0.1648, |
| "step": 864 |
| }, |
| { |
| "epoch": 0.6218547807332854, |
| "grad_norm": 1.1405718326568604, |
| "learning_rate": 3.7886412652767794e-05, |
| "loss": 0.1216, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.6225736879942487, |
| "grad_norm": 1.0623610019683838, |
| "learning_rate": 3.781452192667146e-05, |
| "loss": 0.101, |
| "step": 866 |
| }, |
| { |
| "epoch": 0.6232925952552121, |
| "grad_norm": 0.6313748955726624, |
| "learning_rate": 3.774263120057513e-05, |
| "loss": 0.0872, |
| "step": 867 |
| }, |
| { |
| "epoch": 0.6240115025161754, |
| "grad_norm": 0.7465776801109314, |
| "learning_rate": 3.7670740474478795e-05, |
| "loss": 0.0971, |
| "step": 868 |
| }, |
| { |
| "epoch": 0.6247304097771388, |
| "grad_norm": 0.862188994884491, |
| "learning_rate": 3.759884974838246e-05, |
| "loss": 0.1605, |
| "step": 869 |
| }, |
| { |
| "epoch": 0.6254493170381021, |
| "grad_norm": 0.6960712671279907, |
| "learning_rate": 3.752695902228613e-05, |
| "loss": 0.1134, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.6261682242990654, |
| "grad_norm": 1.1138195991516113, |
| "learning_rate": 3.7455068296189796e-05, |
| "loss": 0.1147, |
| "step": 871 |
| }, |
| { |
| "epoch": 0.6268871315600287, |
| "grad_norm": 0.8084585666656494, |
| "learning_rate": 3.738317757009346e-05, |
| "loss": 0.1016, |
| "step": 872 |
| }, |
| { |
| "epoch": 0.6276060388209921, |
| "grad_norm": 1.7191016674041748, |
| "learning_rate": 3.731128684399713e-05, |
| "loss": 0.2537, |
| "step": 873 |
| }, |
| { |
| "epoch": 0.6283249460819554, |
| "grad_norm": 2.128603219985962, |
| "learning_rate": 3.7239396117900796e-05, |
| "loss": 0.2308, |
| "step": 874 |
| }, |
| { |
| "epoch": 0.6290438533429188, |
| "grad_norm": 0.5702608823776245, |
| "learning_rate": 3.716750539180446e-05, |
| "loss": 0.1835, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.6297627606038821, |
| "grad_norm": 0.5635088086128235, |
| "learning_rate": 3.7095614665708126e-05, |
| "loss": 0.0834, |
| "step": 876 |
| }, |
| { |
| "epoch": 0.6304816678648454, |
| "grad_norm": 0.5227687954902649, |
| "learning_rate": 3.702372393961179e-05, |
| "loss": 0.0899, |
| "step": 877 |
| }, |
| { |
| "epoch": 0.6312005751258087, |
| "grad_norm": 0.7848190069198608, |
| "learning_rate": 3.6951833213515455e-05, |
| "loss": 0.1183, |
| "step": 878 |
| }, |
| { |
| "epoch": 0.6319194823867721, |
| "grad_norm": 0.9584906101226807, |
| "learning_rate": 3.687994248741912e-05, |
| "loss": 0.1022, |
| "step": 879 |
| }, |
| { |
| "epoch": 0.6326383896477354, |
| "grad_norm": 1.682355284690857, |
| "learning_rate": 3.680805176132279e-05, |
| "loss": 0.233, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.6333572969086988, |
| "grad_norm": 1.6860936880111694, |
| "learning_rate": 3.6736161035226456e-05, |
| "loss": 0.1344, |
| "step": 881 |
| }, |
| { |
| "epoch": 0.6340762041696621, |
| "grad_norm": 1.0452550649642944, |
| "learning_rate": 3.666427030913012e-05, |
| "loss": 0.0767, |
| "step": 882 |
| }, |
| { |
| "epoch": 0.6347951114306255, |
| "grad_norm": 0.9689773917198181, |
| "learning_rate": 3.659237958303379e-05, |
| "loss": 0.1741, |
| "step": 883 |
| }, |
| { |
| "epoch": 0.6355140186915887, |
| "grad_norm": 2.4641709327697754, |
| "learning_rate": 3.652048885693746e-05, |
| "loss": 0.2494, |
| "step": 884 |
| }, |
| { |
| "epoch": 0.6362329259525521, |
| "grad_norm": 0.7344926595687866, |
| "learning_rate": 3.644859813084112e-05, |
| "loss": 0.1251, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.6369518332135155, |
| "grad_norm": 0.9594895243644714, |
| "learning_rate": 3.637670740474479e-05, |
| "loss": 0.1264, |
| "step": 886 |
| }, |
| { |
| "epoch": 0.6376707404744788, |
| "grad_norm": 1.5171008110046387, |
| "learning_rate": 3.630481667864846e-05, |
| "loss": 0.208, |
| "step": 887 |
| }, |
| { |
| "epoch": 0.6383896477354422, |
| "grad_norm": 0.5646158456802368, |
| "learning_rate": 3.623292595255212e-05, |
| "loss": 0.1046, |
| "step": 888 |
| }, |
| { |
| "epoch": 0.6391085549964055, |
| "grad_norm": 1.4685351848602295, |
| "learning_rate": 3.6161035226455794e-05, |
| "loss": 0.2528, |
| "step": 889 |
| }, |
| { |
| "epoch": 0.6398274622573688, |
| "grad_norm": 0.9118368029594421, |
| "learning_rate": 3.608914450035946e-05, |
| "loss": 0.1503, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.6405463695183321, |
| "grad_norm": 1.2091461420059204, |
| "learning_rate": 3.601725377426312e-05, |
| "loss": 0.139, |
| "step": 891 |
| }, |
| { |
| "epoch": 0.6412652767792955, |
| "grad_norm": 0.7984756231307983, |
| "learning_rate": 3.594536304816679e-05, |
| "loss": 0.1523, |
| "step": 892 |
| }, |
| { |
| "epoch": 0.6419841840402588, |
| "grad_norm": 1.1727027893066406, |
| "learning_rate": 3.587347232207046e-05, |
| "loss": 0.2125, |
| "step": 893 |
| }, |
| { |
| "epoch": 0.6427030913012222, |
| "grad_norm": 0.41643765568733215, |
| "learning_rate": 3.5801581595974124e-05, |
| "loss": 0.1034, |
| "step": 894 |
| }, |
| { |
| "epoch": 0.6434219985621855, |
| "grad_norm": 1.5034055709838867, |
| "learning_rate": 3.572969086987779e-05, |
| "loss": 0.1743, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.6441409058231489, |
| "grad_norm": 0.6446381211280823, |
| "learning_rate": 3.5657800143781454e-05, |
| "loss": 0.1417, |
| "step": 896 |
| }, |
| { |
| "epoch": 0.6448598130841121, |
| "grad_norm": 0.8232998847961426, |
| "learning_rate": 3.558590941768512e-05, |
| "loss": 0.1406, |
| "step": 897 |
| }, |
| { |
| "epoch": 0.6455787203450755, |
| "grad_norm": 2.0873069763183594, |
| "learning_rate": 3.551401869158878e-05, |
| "loss": 0.3039, |
| "step": 898 |
| }, |
| { |
| "epoch": 0.6462976276060388, |
| "grad_norm": 1.3885002136230469, |
| "learning_rate": 3.5442127965492454e-05, |
| "loss": 0.2175, |
| "step": 899 |
| }, |
| { |
| "epoch": 0.6470165348670022, |
| "grad_norm": 1.215692162513733, |
| "learning_rate": 3.537023723939612e-05, |
| "loss": 0.0805, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.6477354421279655, |
| "grad_norm": 0.7876285910606384, |
| "learning_rate": 3.5298346513299784e-05, |
| "loss": 0.1466, |
| "step": 901 |
| }, |
| { |
| "epoch": 0.6484543493889289, |
| "grad_norm": 1.7122427225112915, |
| "learning_rate": 3.5226455787203455e-05, |
| "loss": 0.1379, |
| "step": 902 |
| }, |
| { |
| "epoch": 0.6491732566498921, |
| "grad_norm": 1.2600849866867065, |
| "learning_rate": 3.515456506110712e-05, |
| "loss": 0.2033, |
| "step": 903 |
| }, |
| { |
| "epoch": 0.6498921639108555, |
| "grad_norm": 0.6112054586410522, |
| "learning_rate": 3.5082674335010785e-05, |
| "loss": 0.1273, |
| "step": 904 |
| }, |
| { |
| "epoch": 0.6506110711718188, |
| "grad_norm": 1.0513567924499512, |
| "learning_rate": 3.501078360891445e-05, |
| "loss": 0.258, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.6513299784327822, |
| "grad_norm": 0.7198784947395325, |
| "learning_rate": 3.493889288281812e-05, |
| "loss": 0.0991, |
| "step": 906 |
| }, |
| { |
| "epoch": 0.6520488856937455, |
| "grad_norm": 0.6129292249679565, |
| "learning_rate": 3.4867002156721785e-05, |
| "loss": 0.1271, |
| "step": 907 |
| }, |
| { |
| "epoch": 0.6527677929547089, |
| "grad_norm": 1.0110605955123901, |
| "learning_rate": 3.479511143062545e-05, |
| "loss": 0.1972, |
| "step": 908 |
| }, |
| { |
| "epoch": 0.6534867002156722, |
| "grad_norm": 1.962942123413086, |
| "learning_rate": 3.472322070452912e-05, |
| "loss": 0.2162, |
| "step": 909 |
| }, |
| { |
| "epoch": 0.6542056074766355, |
| "grad_norm": 0.7024399042129517, |
| "learning_rate": 3.4651329978432786e-05, |
| "loss": 0.0828, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.6549245147375988, |
| "grad_norm": 1.23806631565094, |
| "learning_rate": 3.457943925233645e-05, |
| "loss": 0.1188, |
| "step": 911 |
| }, |
| { |
| "epoch": 0.6556434219985622, |
| "grad_norm": 0.7516190409660339, |
| "learning_rate": 3.4507548526240116e-05, |
| "loss": 0.1542, |
| "step": 912 |
| }, |
| { |
| "epoch": 0.6563623292595255, |
| "grad_norm": 0.636863648891449, |
| "learning_rate": 3.443565780014378e-05, |
| "loss": 0.1436, |
| "step": 913 |
| }, |
| { |
| "epoch": 0.6570812365204889, |
| "grad_norm": 0.4343777298927307, |
| "learning_rate": 3.4363767074047445e-05, |
| "loss": 0.1294, |
| "step": 914 |
| }, |
| { |
| "epoch": 0.6578001437814522, |
| "grad_norm": 0.5764321088790894, |
| "learning_rate": 3.4291876347951116e-05, |
| "loss": 0.1116, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.6585190510424155, |
| "grad_norm": 1.4002054929733276, |
| "learning_rate": 3.421998562185478e-05, |
| "loss": 0.1333, |
| "step": 916 |
| }, |
| { |
| "epoch": 0.6592379583033788, |
| "grad_norm": 0.5198495984077454, |
| "learning_rate": 3.4148094895758446e-05, |
| "loss": 0.1748, |
| "step": 917 |
| }, |
| { |
| "epoch": 0.6599568655643422, |
| "grad_norm": 1.6034399271011353, |
| "learning_rate": 3.407620416966211e-05, |
| "loss": 0.2485, |
| "step": 918 |
| }, |
| { |
| "epoch": 0.6606757728253055, |
| "grad_norm": 0.6112004518508911, |
| "learning_rate": 3.400431344356578e-05, |
| "loss": 0.1656, |
| "step": 919 |
| }, |
| { |
| "epoch": 0.6613946800862689, |
| "grad_norm": 0.9225429892539978, |
| "learning_rate": 3.393242271746945e-05, |
| "loss": 0.1233, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.6621135873472322, |
| "grad_norm": 0.5716907978057861, |
| "learning_rate": 3.386053199137311e-05, |
| "loss": 0.1472, |
| "step": 921 |
| }, |
| { |
| "epoch": 0.6628324946081955, |
| "grad_norm": 0.7885197997093201, |
| "learning_rate": 3.378864126527678e-05, |
| "loss": 0.0917, |
| "step": 922 |
| }, |
| { |
| "epoch": 0.6635514018691588, |
| "grad_norm": 0.558391809463501, |
| "learning_rate": 3.371675053918045e-05, |
| "loss": 0.1058, |
| "step": 923 |
| }, |
| { |
| "epoch": 0.6642703091301222, |
| "grad_norm": 0.5962051749229431, |
| "learning_rate": 3.364485981308411e-05, |
| "loss": 0.0976, |
| "step": 924 |
| }, |
| { |
| "epoch": 0.6649892163910855, |
| "grad_norm": 0.6547393202781677, |
| "learning_rate": 3.3572969086987784e-05, |
| "loss": 0.1143, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.6657081236520489, |
| "grad_norm": 1.5053023099899292, |
| "learning_rate": 3.350107836089145e-05, |
| "loss": 0.1963, |
| "step": 926 |
| }, |
| { |
| "epoch": 0.6664270309130123, |
| "grad_norm": 1.1532018184661865, |
| "learning_rate": 3.342918763479511e-05, |
| "loss": 0.2101, |
| "step": 927 |
| }, |
| { |
| "epoch": 0.6671459381739756, |
| "grad_norm": 0.6440216302871704, |
| "learning_rate": 3.3357296908698785e-05, |
| "loss": 0.1366, |
| "step": 928 |
| }, |
| { |
| "epoch": 0.6678648454349388, |
| "grad_norm": 0.9856230020523071, |
| "learning_rate": 3.328540618260245e-05, |
| "loss": 0.163, |
| "step": 929 |
| }, |
| { |
| "epoch": 0.6685837526959022, |
| "grad_norm": 1.1584689617156982, |
| "learning_rate": 3.3213515456506114e-05, |
| "loss": 0.1567, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.6693026599568656, |
| "grad_norm": 0.7799404859542847, |
| "learning_rate": 3.314162473040978e-05, |
| "loss": 0.1292, |
| "step": 931 |
| }, |
| { |
| "epoch": 0.6700215672178289, |
| "grad_norm": 1.7383888959884644, |
| "learning_rate": 3.306973400431344e-05, |
| "loss": 0.1633, |
| "step": 932 |
| }, |
| { |
| "epoch": 0.6707404744787923, |
| "grad_norm": 0.778204083442688, |
| "learning_rate": 3.299784327821711e-05, |
| "loss": 0.0838, |
| "step": 933 |
| }, |
| { |
| "epoch": 0.6714593817397556, |
| "grad_norm": 1.3401838541030884, |
| "learning_rate": 3.292595255212077e-05, |
| "loss": 0.1235, |
| "step": 934 |
| }, |
| { |
| "epoch": 0.6721782890007189, |
| "grad_norm": 0.675614595413208, |
| "learning_rate": 3.2854061826024444e-05, |
| "loss": 0.1102, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.6728971962616822, |
| "grad_norm": 0.795048177242279, |
| "learning_rate": 3.278217109992811e-05, |
| "loss": 0.1559, |
| "step": 936 |
| }, |
| { |
| "epoch": 0.6736161035226456, |
| "grad_norm": 2.1541974544525146, |
| "learning_rate": 3.2710280373831774e-05, |
| "loss": 0.3036, |
| "step": 937 |
| }, |
| { |
| "epoch": 0.6743350107836089, |
| "grad_norm": 1.3489084243774414, |
| "learning_rate": 3.2638389647735445e-05, |
| "loss": 0.2732, |
| "step": 938 |
| }, |
| { |
| "epoch": 0.6750539180445723, |
| "grad_norm": 0.5745984315872192, |
| "learning_rate": 3.256649892163911e-05, |
| "loss": 0.0881, |
| "step": 939 |
| }, |
| { |
| "epoch": 0.6757728253055356, |
| "grad_norm": 1.369889736175537, |
| "learning_rate": 3.2494608195542774e-05, |
| "loss": 0.2149, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.676491732566499, |
| "grad_norm": 0.6655487418174744, |
| "learning_rate": 3.2422717469446446e-05, |
| "loss": 0.1524, |
| "step": 941 |
| }, |
| { |
| "epoch": 0.6772106398274622, |
| "grad_norm": 1.8854354619979858, |
| "learning_rate": 3.235082674335011e-05, |
| "loss": 0.2042, |
| "step": 942 |
| }, |
| { |
| "epoch": 0.6779295470884256, |
| "grad_norm": 0.8670031428337097, |
| "learning_rate": 3.2278936017253775e-05, |
| "loss": 0.1211, |
| "step": 943 |
| }, |
| { |
| "epoch": 0.6786484543493889, |
| "grad_norm": 1.337445616722107, |
| "learning_rate": 3.220704529115745e-05, |
| "loss": 0.1147, |
| "step": 944 |
| }, |
| { |
| "epoch": 0.6793673616103523, |
| "grad_norm": 0.5259565114974976, |
| "learning_rate": 3.213515456506111e-05, |
| "loss": 0.1028, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.6800862688713156, |
| "grad_norm": 1.4589345455169678, |
| "learning_rate": 3.2063263838964776e-05, |
| "loss": 0.2753, |
| "step": 946 |
| }, |
| { |
| "epoch": 0.680805176132279, |
| "grad_norm": 2.461632251739502, |
| "learning_rate": 3.199137311286844e-05, |
| "loss": 0.2834, |
| "step": 947 |
| }, |
| { |
| "epoch": 0.6815240833932422, |
| "grad_norm": 0.6888427734375, |
| "learning_rate": 3.191948238677211e-05, |
| "loss": 0.1118, |
| "step": 948 |
| }, |
| { |
| "epoch": 0.6822429906542056, |
| "grad_norm": 1.5547517538070679, |
| "learning_rate": 3.184759166067578e-05, |
| "loss": 0.1287, |
| "step": 949 |
| }, |
| { |
| "epoch": 0.6829618979151689, |
| "grad_norm": 0.882874071598053, |
| "learning_rate": 3.177570093457944e-05, |
| "loss": 0.2116, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.6836808051761323, |
| "grad_norm": 0.79315185546875, |
| "learning_rate": 3.1703810208483106e-05, |
| "loss": 0.1034, |
| "step": 951 |
| }, |
| { |
| "epoch": 0.6843997124370956, |
| "grad_norm": 1.387408971786499, |
| "learning_rate": 3.163191948238677e-05, |
| "loss": 0.25, |
| "step": 952 |
| }, |
| { |
| "epoch": 0.685118619698059, |
| "grad_norm": 0.7738803625106812, |
| "learning_rate": 3.1560028756290436e-05, |
| "loss": 0.1813, |
| "step": 953 |
| }, |
| { |
| "epoch": 0.6858375269590223, |
| "grad_norm": 0.6423011422157288, |
| "learning_rate": 3.148813803019411e-05, |
| "loss": 0.1236, |
| "step": 954 |
| }, |
| { |
| "epoch": 0.6865564342199856, |
| "grad_norm": 1.7567709684371948, |
| "learning_rate": 3.141624730409777e-05, |
| "loss": 0.1894, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.6872753414809489, |
| "grad_norm": 0.5689891576766968, |
| "learning_rate": 3.1344356578001437e-05, |
| "loss": 0.1337, |
| "step": 956 |
| }, |
| { |
| "epoch": 0.6879942487419123, |
| "grad_norm": 1.0814508199691772, |
| "learning_rate": 3.127246585190511e-05, |
| "loss": 0.0838, |
| "step": 957 |
| }, |
| { |
| "epoch": 0.6887131560028756, |
| "grad_norm": 0.8408949375152588, |
| "learning_rate": 3.120057512580877e-05, |
| "loss": 0.1288, |
| "step": 958 |
| }, |
| { |
| "epoch": 0.689432063263839, |
| "grad_norm": 0.9981017708778381, |
| "learning_rate": 3.112868439971244e-05, |
| "loss": 0.1968, |
| "step": 959 |
| }, |
| { |
| "epoch": 0.6901509705248023, |
| "grad_norm": 2.650418758392334, |
| "learning_rate": 3.10567936736161e-05, |
| "loss": 0.3508, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.6908698777857656, |
| "grad_norm": 2.785691022872925, |
| "learning_rate": 3.0984902947519774e-05, |
| "loss": 0.2305, |
| "step": 961 |
| }, |
| { |
| "epoch": 0.6915887850467289, |
| "grad_norm": 0.9513468146324158, |
| "learning_rate": 3.091301222142344e-05, |
| "loss": 0.1277, |
| "step": 962 |
| }, |
| { |
| "epoch": 0.6923076923076923, |
| "grad_norm": 0.7549009919166565, |
| "learning_rate": 3.08411214953271e-05, |
| "loss": 0.1175, |
| "step": 963 |
| }, |
| { |
| "epoch": 0.6930265995686556, |
| "grad_norm": 0.7141247987747192, |
| "learning_rate": 3.0769230769230774e-05, |
| "loss": 0.2082, |
| "step": 964 |
| }, |
| { |
| "epoch": 0.693745506829619, |
| "grad_norm": 0.7268842458724976, |
| "learning_rate": 3.069734004313444e-05, |
| "loss": 0.1328, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.6944644140905823, |
| "grad_norm": 0.630649983882904, |
| "learning_rate": 3.0625449317038104e-05, |
| "loss": 0.0766, |
| "step": 966 |
| }, |
| { |
| "epoch": 0.6951833213515457, |
| "grad_norm": 0.7553198933601379, |
| "learning_rate": 3.0553558590941775e-05, |
| "loss": 0.1198, |
| "step": 967 |
| }, |
| { |
| "epoch": 0.6959022286125089, |
| "grad_norm": 0.6632238626480103, |
| "learning_rate": 3.0481667864845436e-05, |
| "loss": 0.128, |
| "step": 968 |
| }, |
| { |
| "epoch": 0.6966211358734723, |
| "grad_norm": 0.5249783992767334, |
| "learning_rate": 3.04097771387491e-05, |
| "loss": 0.1097, |
| "step": 969 |
| }, |
| { |
| "epoch": 0.6973400431344356, |
| "grad_norm": 0.8321582078933716, |
| "learning_rate": 3.0337886412652773e-05, |
| "loss": 0.157, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.698058950395399, |
| "grad_norm": 1.0057244300842285, |
| "learning_rate": 3.0265995686556437e-05, |
| "loss": 0.0965, |
| "step": 971 |
| }, |
| { |
| "epoch": 0.6987778576563624, |
| "grad_norm": 1.271035075187683, |
| "learning_rate": 3.0194104960460102e-05, |
| "loss": 0.262, |
| "step": 972 |
| }, |
| { |
| "epoch": 0.6994967649173257, |
| "grad_norm": 0.660385012626648, |
| "learning_rate": 3.0122214234363767e-05, |
| "loss": 0.1281, |
| "step": 973 |
| }, |
| { |
| "epoch": 0.700215672178289, |
| "grad_norm": 0.7803755402565002, |
| "learning_rate": 3.0050323508267435e-05, |
| "loss": 0.1149, |
| "step": 974 |
| }, |
| { |
| "epoch": 0.7009345794392523, |
| "grad_norm": 1.5343230962753296, |
| "learning_rate": 2.99784327821711e-05, |
| "loss": 0.1426, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.7016534867002157, |
| "grad_norm": 0.6262452602386475, |
| "learning_rate": 2.9906542056074764e-05, |
| "loss": 0.1286, |
| "step": 976 |
| }, |
| { |
| "epoch": 0.702372393961179, |
| "grad_norm": 0.887011706829071, |
| "learning_rate": 2.9834651329978436e-05, |
| "loss": 0.101, |
| "step": 977 |
| }, |
| { |
| "epoch": 0.7030913012221424, |
| "grad_norm": 0.6919823288917542, |
| "learning_rate": 2.97627606038821e-05, |
| "loss": 0.0955, |
| "step": 978 |
| }, |
| { |
| "epoch": 0.7038102084831057, |
| "grad_norm": 0.9775611758232117, |
| "learning_rate": 2.9690869877785765e-05, |
| "loss": 0.1761, |
| "step": 979 |
| }, |
| { |
| "epoch": 0.7045291157440691, |
| "grad_norm": 0.6349767446517944, |
| "learning_rate": 2.9618979151689436e-05, |
| "loss": 0.1569, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.7052480230050323, |
| "grad_norm": 0.8385133147239685, |
| "learning_rate": 2.95470884255931e-05, |
| "loss": 0.1312, |
| "step": 981 |
| }, |
| { |
| "epoch": 0.7059669302659957, |
| "grad_norm": 0.568722665309906, |
| "learning_rate": 2.9475197699496766e-05, |
| "loss": 0.1172, |
| "step": 982 |
| }, |
| { |
| "epoch": 0.706685837526959, |
| "grad_norm": 1.0487462282180786, |
| "learning_rate": 2.9403306973400434e-05, |
| "loss": 0.1222, |
| "step": 983 |
| }, |
| { |
| "epoch": 0.7074047447879224, |
| "grad_norm": 0.7681689262390137, |
| "learning_rate": 2.93314162473041e-05, |
| "loss": 0.1354, |
| "step": 984 |
| }, |
| { |
| "epoch": 0.7081236520488857, |
| "grad_norm": 0.43179476261138916, |
| "learning_rate": 2.9259525521207763e-05, |
| "loss": 0.0842, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.7088425593098491, |
| "grad_norm": 0.8151119351387024, |
| "learning_rate": 2.9187634795111428e-05, |
| "loss": 0.1245, |
| "step": 986 |
| }, |
| { |
| "epoch": 0.7095614665708123, |
| "grad_norm": 1.0838408470153809, |
| "learning_rate": 2.91157440690151e-05, |
| "loss": 0.1414, |
| "step": 987 |
| }, |
| { |
| "epoch": 0.7102803738317757, |
| "grad_norm": 0.9984173774719238, |
| "learning_rate": 2.9043853342918764e-05, |
| "loss": 0.1812, |
| "step": 988 |
| }, |
| { |
| "epoch": 0.710999281092739, |
| "grad_norm": 0.4085555672645569, |
| "learning_rate": 2.897196261682243e-05, |
| "loss": 0.0691, |
| "step": 989 |
| }, |
| { |
| "epoch": 0.7117181883537024, |
| "grad_norm": 0.5325964093208313, |
| "learning_rate": 2.89000718907261e-05, |
| "loss": 0.1054, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.7124370956146657, |
| "grad_norm": 0.9540407657623291, |
| "learning_rate": 2.8828181164629765e-05, |
| "loss": 0.1817, |
| "step": 991 |
| }, |
| { |
| "epoch": 0.7131560028756291, |
| "grad_norm": 0.44249624013900757, |
| "learning_rate": 2.875629043853343e-05, |
| "loss": 0.0989, |
| "step": 992 |
| }, |
| { |
| "epoch": 0.7138749101365924, |
| "grad_norm": 0.48831212520599365, |
| "learning_rate": 2.8684399712437098e-05, |
| "loss": 0.0721, |
| "step": 993 |
| }, |
| { |
| "epoch": 0.7145938173975557, |
| "grad_norm": 2.0384461879730225, |
| "learning_rate": 2.8612508986340762e-05, |
| "loss": 0.2277, |
| "step": 994 |
| }, |
| { |
| "epoch": 0.715312724658519, |
| "grad_norm": 0.7207070589065552, |
| "learning_rate": 2.8540618260244427e-05, |
| "loss": 0.1763, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.7160316319194824, |
| "grad_norm": 0.7955511212348938, |
| "learning_rate": 2.84687275341481e-05, |
| "loss": 0.185, |
| "step": 996 |
| }, |
| { |
| "epoch": 0.7167505391804457, |
| "grad_norm": 0.7830111980438232, |
| "learning_rate": 2.8396836808051763e-05, |
| "loss": 0.1586, |
| "step": 997 |
| }, |
| { |
| "epoch": 0.7174694464414091, |
| "grad_norm": 0.9649351835250854, |
| "learning_rate": 2.8324946081955428e-05, |
| "loss": 0.1004, |
| "step": 998 |
| }, |
| { |
| "epoch": 0.7181883537023724, |
| "grad_norm": 0.6637257933616638, |
| "learning_rate": 2.82530553558591e-05, |
| "loss": 0.1441, |
| "step": 999 |
| }, |
| { |
| "epoch": 0.7189072609633357, |
| "grad_norm": 1.3482410907745361, |
| "learning_rate": 2.8181164629762764e-05, |
| "loss": 0.2113, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7189072609633357, |
| "eval_accuracy": 0.9389243125915527, |
| "eval_loss": 0.13201281428337097, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6608933210372925, |
| "eval_precision": 0.9282758831977844, |
| "eval_recall": 0.9781976938247681, |
| "eval_runtime": 601.6475, |
| "eval_samples_per_second": 4.626, |
| "eval_steps_per_second": 0.578, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.719626168224299, |
| "grad_norm": 0.4483204782009125, |
| "learning_rate": 2.810927390366643e-05, |
| "loss": 0.1101, |
| "step": 1001 |
| }, |
| { |
| "epoch": 0.7203450754852624, |
| "grad_norm": 1.527923583984375, |
| "learning_rate": 2.8037383177570094e-05, |
| "loss": 0.2485, |
| "step": 1002 |
| }, |
| { |
| "epoch": 0.7210639827462257, |
| "grad_norm": 0.6009194254875183, |
| "learning_rate": 2.796549245147376e-05, |
| "loss": 0.1266, |
| "step": 1003 |
| }, |
| { |
| "epoch": 0.7217828900071891, |
| "grad_norm": 1.3639483451843262, |
| "learning_rate": 2.7893601725377426e-05, |
| "loss": 0.1608, |
| "step": 1004 |
| }, |
| { |
| "epoch": 0.7225017972681524, |
| "grad_norm": 0.6033057570457458, |
| "learning_rate": 2.782171099928109e-05, |
| "loss": 0.1194, |
| "step": 1005 |
| }, |
| { |
| "epoch": 0.7232207045291158, |
| "grad_norm": 0.4444285035133362, |
| "learning_rate": 2.7749820273184762e-05, |
| "loss": 0.1163, |
| "step": 1006 |
| }, |
| { |
| "epoch": 0.723939611790079, |
| "grad_norm": 1.5719153881072998, |
| "learning_rate": 2.7677929547088427e-05, |
| "loss": 0.1096, |
| "step": 1007 |
| }, |
| { |
| "epoch": 0.7246585190510424, |
| "grad_norm": 0.6401962041854858, |
| "learning_rate": 2.7606038820992092e-05, |
| "loss": 0.1509, |
| "step": 1008 |
| }, |
| { |
| "epoch": 0.7253774263120057, |
| "grad_norm": 0.5552104115486145, |
| "learning_rate": 2.7534148094895763e-05, |
| "loss": 0.1511, |
| "step": 1009 |
| }, |
| { |
| "epoch": 0.7260963335729691, |
| "grad_norm": 1.3960051536560059, |
| "learning_rate": 2.7462257368799428e-05, |
| "loss": 0.1642, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.7268152408339325, |
| "grad_norm": 0.5197566151618958, |
| "learning_rate": 2.7390366642703093e-05, |
| "loss": 0.1208, |
| "step": 1011 |
| }, |
| { |
| "epoch": 0.7275341480948958, |
| "grad_norm": 0.8309020400047302, |
| "learning_rate": 2.731847591660676e-05, |
| "loss": 0.1361, |
| "step": 1012 |
| }, |
| { |
| "epoch": 0.728253055355859, |
| "grad_norm": 0.6530104279518127, |
| "learning_rate": 2.7246585190510425e-05, |
| "loss": 0.08, |
| "step": 1013 |
| }, |
| { |
| "epoch": 0.7289719626168224, |
| "grad_norm": 0.6603716611862183, |
| "learning_rate": 2.717469446441409e-05, |
| "loss": 0.09, |
| "step": 1014 |
| }, |
| { |
| "epoch": 0.7296908698777858, |
| "grad_norm": 1.6257810592651367, |
| "learning_rate": 2.7102803738317755e-05, |
| "loss": 0.1952, |
| "step": 1015 |
| }, |
| { |
| "epoch": 0.7304097771387491, |
| "grad_norm": 0.47092491388320923, |
| "learning_rate": 2.7030913012221426e-05, |
| "loss": 0.1213, |
| "step": 1016 |
| }, |
| { |
| "epoch": 0.7311286843997125, |
| "grad_norm": 1.7538068294525146, |
| "learning_rate": 2.695902228612509e-05, |
| "loss": 0.1758, |
| "step": 1017 |
| }, |
| { |
| "epoch": 0.7318475916606758, |
| "grad_norm": 1.6580126285552979, |
| "learning_rate": 2.6887131560028756e-05, |
| "loss": 0.2174, |
| "step": 1018 |
| }, |
| { |
| "epoch": 0.732566498921639, |
| "grad_norm": 1.0007431507110596, |
| "learning_rate": 2.6815240833932427e-05, |
| "loss": 0.1448, |
| "step": 1019 |
| }, |
| { |
| "epoch": 0.7332854061826024, |
| "grad_norm": 0.8419533967971802, |
| "learning_rate": 2.6743350107836092e-05, |
| "loss": 0.1748, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.7340043134435658, |
| "grad_norm": 0.9676351547241211, |
| "learning_rate": 2.6671459381739757e-05, |
| "loss": 0.1689, |
| "step": 1021 |
| }, |
| { |
| "epoch": 0.7347232207045291, |
| "grad_norm": 0.8018962740898132, |
| "learning_rate": 2.6599568655643425e-05, |
| "loss": 0.1504, |
| "step": 1022 |
| }, |
| { |
| "epoch": 0.7354421279654925, |
| "grad_norm": 1.185584306716919, |
| "learning_rate": 2.652767792954709e-05, |
| "loss": 0.1657, |
| "step": 1023 |
| }, |
| { |
| "epoch": 0.7361610352264558, |
| "grad_norm": 0.5642529726028442, |
| "learning_rate": 2.6455787203450754e-05, |
| "loss": 0.1274, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.7368799424874192, |
| "grad_norm": 1.1524418592453003, |
| "learning_rate": 2.6383896477354425e-05, |
| "loss": 0.0948, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.7375988497483824, |
| "grad_norm": 2.0588278770446777, |
| "learning_rate": 2.631200575125809e-05, |
| "loss": 0.1813, |
| "step": 1026 |
| }, |
| { |
| "epoch": 0.7383177570093458, |
| "grad_norm": 1.9449617862701416, |
| "learning_rate": 2.6240115025161755e-05, |
| "loss": 0.1064, |
| "step": 1027 |
| }, |
| { |
| "epoch": 0.7390366642703091, |
| "grad_norm": 0.6734256744384766, |
| "learning_rate": 2.616822429906542e-05, |
| "loss": 0.1727, |
| "step": 1028 |
| }, |
| { |
| "epoch": 0.7397555715312725, |
| "grad_norm": 0.9582931995391846, |
| "learning_rate": 2.609633357296909e-05, |
| "loss": 0.0729, |
| "step": 1029 |
| }, |
| { |
| "epoch": 0.7404744787922358, |
| "grad_norm": 0.8061708807945251, |
| "learning_rate": 2.6024442846872756e-05, |
| "loss": 0.142, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.7411933860531992, |
| "grad_norm": 1.691681146621704, |
| "learning_rate": 2.595255212077642e-05, |
| "loss": 0.2275, |
| "step": 1031 |
| }, |
| { |
| "epoch": 0.7419122933141624, |
| "grad_norm": 1.4140758514404297, |
| "learning_rate": 2.588066139468009e-05, |
| "loss": 0.1692, |
| "step": 1032 |
| }, |
| { |
| "epoch": 0.7426312005751258, |
| "grad_norm": 0.8371263146400452, |
| "learning_rate": 2.5808770668583753e-05, |
| "loss": 0.1104, |
| "step": 1033 |
| }, |
| { |
| "epoch": 0.7433501078360891, |
| "grad_norm": 0.6498920917510986, |
| "learning_rate": 2.5736879942487418e-05, |
| "loss": 0.1211, |
| "step": 1034 |
| }, |
| { |
| "epoch": 0.7440690150970525, |
| "grad_norm": 0.648303210735321, |
| "learning_rate": 2.566498921639109e-05, |
| "loss": 0.1807, |
| "step": 1035 |
| }, |
| { |
| "epoch": 0.7447879223580158, |
| "grad_norm": 0.7186898589134216, |
| "learning_rate": 2.5593098490294754e-05, |
| "loss": 0.1449, |
| "step": 1036 |
| }, |
| { |
| "epoch": 0.7455068296189792, |
| "grad_norm": 0.4701125919818878, |
| "learning_rate": 2.552120776419842e-05, |
| "loss": 0.1178, |
| "step": 1037 |
| }, |
| { |
| "epoch": 0.7462257368799425, |
| "grad_norm": 0.8611357808113098, |
| "learning_rate": 2.544931703810209e-05, |
| "loss": 0.1335, |
| "step": 1038 |
| }, |
| { |
| "epoch": 0.7469446441409058, |
| "grad_norm": 0.9387286305427551, |
| "learning_rate": 2.5377426312005755e-05, |
| "loss": 0.1353, |
| "step": 1039 |
| }, |
| { |
| "epoch": 0.7476635514018691, |
| "grad_norm": 0.8511605262756348, |
| "learning_rate": 2.530553558590942e-05, |
| "loss": 0.192, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.7483824586628325, |
| "grad_norm": 0.665184736251831, |
| "learning_rate": 2.5233644859813084e-05, |
| "loss": 0.13, |
| "step": 1041 |
| }, |
| { |
| "epoch": 0.7491013659237958, |
| "grad_norm": 0.527923583984375, |
| "learning_rate": 2.5161754133716752e-05, |
| "loss": 0.1051, |
| "step": 1042 |
| }, |
| { |
| "epoch": 0.7498202731847592, |
| "grad_norm": 0.8178277015686035, |
| "learning_rate": 2.5089863407620417e-05, |
| "loss": 0.1818, |
| "step": 1043 |
| }, |
| { |
| "epoch": 0.7505391804457225, |
| "grad_norm": 1.2879143953323364, |
| "learning_rate": 2.501797268152408e-05, |
| "loss": 0.2244, |
| "step": 1044 |
| }, |
| { |
| "epoch": 0.7512580877066858, |
| "grad_norm": 0.7047324776649475, |
| "learning_rate": 2.494608195542775e-05, |
| "loss": 0.1224, |
| "step": 1045 |
| }, |
| { |
| "epoch": 0.7519769949676491, |
| "grad_norm": 0.7401057481765747, |
| "learning_rate": 2.4874191229331418e-05, |
| "loss": 0.1698, |
| "step": 1046 |
| }, |
| { |
| "epoch": 0.7526959022286125, |
| "grad_norm": 1.147862434387207, |
| "learning_rate": 2.4802300503235086e-05, |
| "loss": 0.1731, |
| "step": 1047 |
| }, |
| { |
| "epoch": 0.7534148094895758, |
| "grad_norm": 0.5870615839958191, |
| "learning_rate": 2.473040977713875e-05, |
| "loss": 0.1, |
| "step": 1048 |
| }, |
| { |
| "epoch": 0.7541337167505392, |
| "grad_norm": 0.6847332715988159, |
| "learning_rate": 2.4658519051042415e-05, |
| "loss": 0.178, |
| "step": 1049 |
| }, |
| { |
| "epoch": 0.7548526240115025, |
| "grad_norm": 0.8262477517127991, |
| "learning_rate": 2.4586628324946083e-05, |
| "loss": 0.1343, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.7555715312724659, |
| "grad_norm": 1.9897246360778809, |
| "learning_rate": 2.4514737598849748e-05, |
| "loss": 0.1833, |
| "step": 1051 |
| }, |
| { |
| "epoch": 0.7562904385334291, |
| "grad_norm": 0.6993079781532288, |
| "learning_rate": 2.4442846872753416e-05, |
| "loss": 0.1208, |
| "step": 1052 |
| }, |
| { |
| "epoch": 0.7570093457943925, |
| "grad_norm": 0.7409127950668335, |
| "learning_rate": 2.437095614665708e-05, |
| "loss": 0.1452, |
| "step": 1053 |
| }, |
| { |
| "epoch": 0.7577282530553558, |
| "grad_norm": 0.46567317843437195, |
| "learning_rate": 2.429906542056075e-05, |
| "loss": 0.1235, |
| "step": 1054 |
| }, |
| { |
| "epoch": 0.7584471603163192, |
| "grad_norm": 0.5836708545684814, |
| "learning_rate": 2.4227174694464417e-05, |
| "loss": 0.0988, |
| "step": 1055 |
| }, |
| { |
| "epoch": 0.7591660675772826, |
| "grad_norm": 0.6743710041046143, |
| "learning_rate": 2.415528396836808e-05, |
| "loss": 0.1847, |
| "step": 1056 |
| }, |
| { |
| "epoch": 0.7598849748382459, |
| "grad_norm": 0.5269002914428711, |
| "learning_rate": 2.408339324227175e-05, |
| "loss": 0.125, |
| "step": 1057 |
| }, |
| { |
| "epoch": 0.7606038820992091, |
| "grad_norm": 1.1138023138046265, |
| "learning_rate": 2.4011502516175414e-05, |
| "loss": 0.1264, |
| "step": 1058 |
| }, |
| { |
| "epoch": 0.7613227893601725, |
| "grad_norm": 1.1032012701034546, |
| "learning_rate": 2.393961179007908e-05, |
| "loss": 0.156, |
| "step": 1059 |
| }, |
| { |
| "epoch": 0.7620416966211359, |
| "grad_norm": 0.7601384520530701, |
| "learning_rate": 2.3867721063982747e-05, |
| "loss": 0.0847, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.7627606038820992, |
| "grad_norm": 0.8220066428184509, |
| "learning_rate": 2.3795830337886412e-05, |
| "loss": 0.1394, |
| "step": 1061 |
| }, |
| { |
| "epoch": 0.7634795111430626, |
| "grad_norm": 1.8573328256607056, |
| "learning_rate": 2.372393961179008e-05, |
| "loss": 0.2568, |
| "step": 1062 |
| }, |
| { |
| "epoch": 0.7641984184040259, |
| "grad_norm": 1.170921802520752, |
| "learning_rate": 2.3652048885693748e-05, |
| "loss": 0.2878, |
| "step": 1063 |
| }, |
| { |
| "epoch": 0.7649173256649893, |
| "grad_norm": 0.8639873266220093, |
| "learning_rate": 2.3580158159597413e-05, |
| "loss": 0.1114, |
| "step": 1064 |
| }, |
| { |
| "epoch": 0.7656362329259525, |
| "grad_norm": 1.3616185188293457, |
| "learning_rate": 2.350826743350108e-05, |
| "loss": 0.084, |
| "step": 1065 |
| }, |
| { |
| "epoch": 0.7663551401869159, |
| "grad_norm": 0.630680501461029, |
| "learning_rate": 2.3436376707404745e-05, |
| "loss": 0.1511, |
| "step": 1066 |
| }, |
| { |
| "epoch": 0.7670740474478792, |
| "grad_norm": 1.677143931388855, |
| "learning_rate": 2.3364485981308414e-05, |
| "loss": 0.145, |
| "step": 1067 |
| }, |
| { |
| "epoch": 0.7677929547088426, |
| "grad_norm": 0.9498730301856995, |
| "learning_rate": 2.3292595255212078e-05, |
| "loss": 0.1269, |
| "step": 1068 |
| }, |
| { |
| "epoch": 0.7685118619698059, |
| "grad_norm": 1.716393232345581, |
| "learning_rate": 2.3220704529115743e-05, |
| "loss": 0.1519, |
| "step": 1069 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 0.6883776187896729, |
| "learning_rate": 2.314881380301941e-05, |
| "loss": 0.1698, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.7699496764917325, |
| "grad_norm": 0.7949609160423279, |
| "learning_rate": 2.307692307692308e-05, |
| "loss": 0.1373, |
| "step": 1071 |
| }, |
| { |
| "epoch": 0.7706685837526959, |
| "grad_norm": 0.6971243619918823, |
| "learning_rate": 2.3005032350826744e-05, |
| "loss": 0.113, |
| "step": 1072 |
| }, |
| { |
| "epoch": 0.7713874910136592, |
| "grad_norm": 0.5017929673194885, |
| "learning_rate": 2.2933141624730412e-05, |
| "loss": 0.1094, |
| "step": 1073 |
| }, |
| { |
| "epoch": 0.7721063982746226, |
| "grad_norm": 0.5453867316246033, |
| "learning_rate": 2.2861250898634077e-05, |
| "loss": 0.0969, |
| "step": 1074 |
| }, |
| { |
| "epoch": 0.7728253055355859, |
| "grad_norm": 0.6106514930725098, |
| "learning_rate": 2.2789360172537745e-05, |
| "loss": 0.1489, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.7735442127965493, |
| "grad_norm": 0.4236951768398285, |
| "learning_rate": 2.2717469446441413e-05, |
| "loss": 0.0965, |
| "step": 1076 |
| }, |
| { |
| "epoch": 0.7742631200575126, |
| "grad_norm": 0.602499783039093, |
| "learning_rate": 2.2645578720345077e-05, |
| "loss": 0.1458, |
| "step": 1077 |
| }, |
| { |
| "epoch": 0.7749820273184759, |
| "grad_norm": 0.5535038709640503, |
| "learning_rate": 2.2573687994248742e-05, |
| "loss": 0.0922, |
| "step": 1078 |
| }, |
| { |
| "epoch": 0.7757009345794392, |
| "grad_norm": 0.6116883754730225, |
| "learning_rate": 2.2501797268152407e-05, |
| "loss": 0.1125, |
| "step": 1079 |
| }, |
| { |
| "epoch": 0.7764198418404026, |
| "grad_norm": 0.3836880326271057, |
| "learning_rate": 2.2429906542056075e-05, |
| "loss": 0.0718, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.7771387491013659, |
| "grad_norm": 0.5877097249031067, |
| "learning_rate": 2.2358015815959743e-05, |
| "loss": 0.1318, |
| "step": 1081 |
| }, |
| { |
| "epoch": 0.7778576563623293, |
| "grad_norm": 0.673808217048645, |
| "learning_rate": 2.2286125089863408e-05, |
| "loss": 0.1352, |
| "step": 1082 |
| }, |
| { |
| "epoch": 0.7785765636232926, |
| "grad_norm": 0.8240437507629395, |
| "learning_rate": 2.2214234363767076e-05, |
| "loss": 0.1074, |
| "step": 1083 |
| }, |
| { |
| "epoch": 0.7792954708842559, |
| "grad_norm": 0.7427266836166382, |
| "learning_rate": 2.2142343637670744e-05, |
| "loss": 0.0881, |
| "step": 1084 |
| }, |
| { |
| "epoch": 0.7800143781452192, |
| "grad_norm": 1.7619154453277588, |
| "learning_rate": 2.207045291157441e-05, |
| "loss": 0.2362, |
| "step": 1085 |
| }, |
| { |
| "epoch": 0.7807332854061826, |
| "grad_norm": 0.6677937507629395, |
| "learning_rate": 2.1998562185478077e-05, |
| "loss": 0.168, |
| "step": 1086 |
| }, |
| { |
| "epoch": 0.7814521926671459, |
| "grad_norm": 1.6103730201721191, |
| "learning_rate": 2.192667145938174e-05, |
| "loss": 0.1294, |
| "step": 1087 |
| }, |
| { |
| "epoch": 0.7821710999281093, |
| "grad_norm": 0.6742684245109558, |
| "learning_rate": 2.1854780733285406e-05, |
| "loss": 0.1029, |
| "step": 1088 |
| }, |
| { |
| "epoch": 0.7828900071890726, |
| "grad_norm": 1.0827564001083374, |
| "learning_rate": 2.1782890007189074e-05, |
| "loss": 0.2341, |
| "step": 1089 |
| }, |
| { |
| "epoch": 0.783608914450036, |
| "grad_norm": 0.5942773222923279, |
| "learning_rate": 2.171099928109274e-05, |
| "loss": 0.1349, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.7843278217109992, |
| "grad_norm": 0.7376931309700012, |
| "learning_rate": 2.1639108554996407e-05, |
| "loss": 0.12, |
| "step": 1091 |
| }, |
| { |
| "epoch": 0.7850467289719626, |
| "grad_norm": 1.202217698097229, |
| "learning_rate": 2.1567217828900075e-05, |
| "loss": 0.1605, |
| "step": 1092 |
| }, |
| { |
| "epoch": 0.785765636232926, |
| "grad_norm": 0.7461795210838318, |
| "learning_rate": 2.149532710280374e-05, |
| "loss": 0.0918, |
| "step": 1093 |
| }, |
| { |
| "epoch": 0.7864845434938893, |
| "grad_norm": 0.7658885717391968, |
| "learning_rate": 2.1423436376707408e-05, |
| "loss": 0.1687, |
| "step": 1094 |
| }, |
| { |
| "epoch": 0.7872034507548527, |
| "grad_norm": 2.174950122833252, |
| "learning_rate": 2.1351545650611072e-05, |
| "loss": 0.2991, |
| "step": 1095 |
| }, |
| { |
| "epoch": 0.787922358015816, |
| "grad_norm": 1.2805111408233643, |
| "learning_rate": 2.1279654924514737e-05, |
| "loss": 0.1442, |
| "step": 1096 |
| }, |
| { |
| "epoch": 0.7886412652767792, |
| "grad_norm": 0.8354223966598511, |
| "learning_rate": 2.1207764198418405e-05, |
| "loss": 0.1316, |
| "step": 1097 |
| }, |
| { |
| "epoch": 0.7893601725377426, |
| "grad_norm": 0.5744108557701111, |
| "learning_rate": 2.113587347232207e-05, |
| "loss": 0.1445, |
| "step": 1098 |
| }, |
| { |
| "epoch": 0.790079079798706, |
| "grad_norm": 0.8944901823997498, |
| "learning_rate": 2.1063982746225738e-05, |
| "loss": 0.0961, |
| "step": 1099 |
| }, |
| { |
| "epoch": 0.7907979870596693, |
| "grad_norm": 0.8875666260719299, |
| "learning_rate": 2.0992092020129402e-05, |
| "loss": 0.1717, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.7915168943206327, |
| "grad_norm": 1.418696403503418, |
| "learning_rate": 2.092020129403307e-05, |
| "loss": 0.1085, |
| "step": 1101 |
| }, |
| { |
| "epoch": 0.792235801581596, |
| "grad_norm": 2.632425308227539, |
| "learning_rate": 2.084831056793674e-05, |
| "loss": 0.339, |
| "step": 1102 |
| }, |
| { |
| "epoch": 0.7929547088425594, |
| "grad_norm": 1.0444191694259644, |
| "learning_rate": 2.0776419841840403e-05, |
| "loss": 0.0763, |
| "step": 1103 |
| }, |
| { |
| "epoch": 0.7936736161035226, |
| "grad_norm": 1.2826745510101318, |
| "learning_rate": 2.070452911574407e-05, |
| "loss": 0.1055, |
| "step": 1104 |
| }, |
| { |
| "epoch": 0.794392523364486, |
| "grad_norm": 1.4323136806488037, |
| "learning_rate": 2.0632638389647736e-05, |
| "loss": 0.1795, |
| "step": 1105 |
| }, |
| { |
| "epoch": 0.7951114306254493, |
| "grad_norm": 1.06536066532135, |
| "learning_rate": 2.05607476635514e-05, |
| "loss": 0.1018, |
| "step": 1106 |
| }, |
| { |
| "epoch": 0.7958303378864127, |
| "grad_norm": 1.1474952697753906, |
| "learning_rate": 2.048885693745507e-05, |
| "loss": 0.1786, |
| "step": 1107 |
| }, |
| { |
| "epoch": 0.796549245147376, |
| "grad_norm": 1.3087904453277588, |
| "learning_rate": 2.0416966211358734e-05, |
| "loss": 0.1902, |
| "step": 1108 |
| }, |
| { |
| "epoch": 0.7972681524083394, |
| "grad_norm": 0.8624778985977173, |
| "learning_rate": 2.03450754852624e-05, |
| "loss": 0.1596, |
| "step": 1109 |
| }, |
| { |
| "epoch": 0.7979870596693026, |
| "grad_norm": 1.1763020753860474, |
| "learning_rate": 2.027318475916607e-05, |
| "loss": 0.2194, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.798705966930266, |
| "grad_norm": 0.6417412161827087, |
| "learning_rate": 2.0201294033069734e-05, |
| "loss": 0.1462, |
| "step": 1111 |
| }, |
| { |
| "epoch": 0.7994248741912293, |
| "grad_norm": 0.74468594789505, |
| "learning_rate": 2.0129403306973402e-05, |
| "loss": 0.1148, |
| "step": 1112 |
| }, |
| { |
| "epoch": 0.8001437814521927, |
| "grad_norm": 0.714839518070221, |
| "learning_rate": 2.005751258087707e-05, |
| "loss": 0.1258, |
| "step": 1113 |
| }, |
| { |
| "epoch": 0.800862688713156, |
| "grad_norm": 1.1390016078948975, |
| "learning_rate": 1.9985621854780735e-05, |
| "loss": 0.168, |
| "step": 1114 |
| }, |
| { |
| "epoch": 0.8015815959741194, |
| "grad_norm": 0.6036468744277954, |
| "learning_rate": 1.99137311286844e-05, |
| "loss": 0.1242, |
| "step": 1115 |
| }, |
| { |
| "epoch": 0.8023005032350827, |
| "grad_norm": 1.0092874765396118, |
| "learning_rate": 1.9841840402588065e-05, |
| "loss": 0.227, |
| "step": 1116 |
| }, |
| { |
| "epoch": 0.803019410496046, |
| "grad_norm": 1.2119369506835938, |
| "learning_rate": 1.9769949676491733e-05, |
| "loss": 0.2376, |
| "step": 1117 |
| }, |
| { |
| "epoch": 0.8037383177570093, |
| "grad_norm": 0.5614322423934937, |
| "learning_rate": 1.96980589503954e-05, |
| "loss": 0.1366, |
| "step": 1118 |
| }, |
| { |
| "epoch": 0.8044572250179727, |
| "grad_norm": 0.5976343154907227, |
| "learning_rate": 1.9626168224299065e-05, |
| "loss": 0.1717, |
| "step": 1119 |
| }, |
| { |
| "epoch": 0.805176132278936, |
| "grad_norm": 1.6751735210418701, |
| "learning_rate": 1.9554277498202734e-05, |
| "loss": 0.2478, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.8058950395398994, |
| "grad_norm": 0.5886757373809814, |
| "learning_rate": 1.9482386772106398e-05, |
| "loss": 0.0935, |
| "step": 1121 |
| }, |
| { |
| "epoch": 0.8066139468008627, |
| "grad_norm": 1.0640392303466797, |
| "learning_rate": 1.9410496046010066e-05, |
| "loss": 0.0883, |
| "step": 1122 |
| }, |
| { |
| "epoch": 0.807332854061826, |
| "grad_norm": 0.8485746383666992, |
| "learning_rate": 1.9338605319913734e-05, |
| "loss": 0.1059, |
| "step": 1123 |
| }, |
| { |
| "epoch": 0.8080517613227893, |
| "grad_norm": 0.655572772026062, |
| "learning_rate": 1.92667145938174e-05, |
| "loss": 0.1152, |
| "step": 1124 |
| }, |
| { |
| "epoch": 0.8087706685837527, |
| "grad_norm": 0.512906551361084, |
| "learning_rate": 1.9194823867721064e-05, |
| "loss": 0.1329, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.809489575844716, |
| "grad_norm": 0.6943735480308533, |
| "learning_rate": 1.9122933141624732e-05, |
| "loss": 0.1133, |
| "step": 1126 |
| }, |
| { |
| "epoch": 0.8102084831056794, |
| "grad_norm": 1.063551664352417, |
| "learning_rate": 1.9051042415528397e-05, |
| "loss": 0.2183, |
| "step": 1127 |
| }, |
| { |
| "epoch": 0.8109273903666427, |
| "grad_norm": 2.0170514583587646, |
| "learning_rate": 1.8979151689432065e-05, |
| "loss": 0.2107, |
| "step": 1128 |
| }, |
| { |
| "epoch": 0.811646297627606, |
| "grad_norm": 1.3455495834350586, |
| "learning_rate": 1.890726096333573e-05, |
| "loss": 0.136, |
| "step": 1129 |
| }, |
| { |
| "epoch": 0.8123652048885693, |
| "grad_norm": 1.0634162425994873, |
| "learning_rate": 1.8835370237239397e-05, |
| "loss": 0.1372, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.8130841121495327, |
| "grad_norm": 0.661889910697937, |
| "learning_rate": 1.8763479511143065e-05, |
| "loss": 0.121, |
| "step": 1131 |
| }, |
| { |
| "epoch": 0.813803019410496, |
| "grad_norm": 1.825216293334961, |
| "learning_rate": 1.869158878504673e-05, |
| "loss": 0.22, |
| "step": 1132 |
| }, |
| { |
| "epoch": 0.8145219266714594, |
| "grad_norm": 0.44640716910362244, |
| "learning_rate": 1.8619698058950398e-05, |
| "loss": 0.1404, |
| "step": 1133 |
| }, |
| { |
| "epoch": 0.8152408339324227, |
| "grad_norm": 1.0852537155151367, |
| "learning_rate": 1.8547807332854063e-05, |
| "loss": 0.2598, |
| "step": 1134 |
| }, |
| { |
| "epoch": 0.8159597411933861, |
| "grad_norm": 1.2837765216827393, |
| "learning_rate": 1.8475916606757728e-05, |
| "loss": 0.1382, |
| "step": 1135 |
| }, |
| { |
| "epoch": 0.8166786484543493, |
| "grad_norm": 1.2876685857772827, |
| "learning_rate": 1.8404025880661396e-05, |
| "loss": 0.1085, |
| "step": 1136 |
| }, |
| { |
| "epoch": 0.8173975557153127, |
| "grad_norm": 0.4106619358062744, |
| "learning_rate": 1.833213515456506e-05, |
| "loss": 0.0965, |
| "step": 1137 |
| }, |
| { |
| "epoch": 0.818116462976276, |
| "grad_norm": 1.0273689031600952, |
| "learning_rate": 1.826024442846873e-05, |
| "loss": 0.0866, |
| "step": 1138 |
| }, |
| { |
| "epoch": 0.8188353702372394, |
| "grad_norm": 1.3030614852905273, |
| "learning_rate": 1.8188353702372397e-05, |
| "loss": 0.1186, |
| "step": 1139 |
| }, |
| { |
| "epoch": 0.8195542774982028, |
| "grad_norm": 1.150269865989685, |
| "learning_rate": 1.811646297627606e-05, |
| "loss": 0.1895, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.8202731847591661, |
| "grad_norm": 0.514682948589325, |
| "learning_rate": 1.804457225017973e-05, |
| "loss": 0.1294, |
| "step": 1141 |
| }, |
| { |
| "epoch": 0.8209920920201293, |
| "grad_norm": 0.44898927211761475, |
| "learning_rate": 1.7972681524083394e-05, |
| "loss": 0.0981, |
| "step": 1142 |
| }, |
| { |
| "epoch": 0.8217109992810927, |
| "grad_norm": 0.6963998675346375, |
| "learning_rate": 1.7900790797987062e-05, |
| "loss": 0.1678, |
| "step": 1143 |
| }, |
| { |
| "epoch": 0.822429906542056, |
| "grad_norm": 0.48149439692497253, |
| "learning_rate": 1.7828900071890727e-05, |
| "loss": 0.0917, |
| "step": 1144 |
| }, |
| { |
| "epoch": 0.8231488138030194, |
| "grad_norm": 0.6283948421478271, |
| "learning_rate": 1.775700934579439e-05, |
| "loss": 0.0996, |
| "step": 1145 |
| }, |
| { |
| "epoch": 0.8238677210639828, |
| "grad_norm": 0.5601471066474915, |
| "learning_rate": 1.768511861969806e-05, |
| "loss": 0.118, |
| "step": 1146 |
| }, |
| { |
| "epoch": 0.8245866283249461, |
| "grad_norm": 0.6269745826721191, |
| "learning_rate": 1.7613227893601728e-05, |
| "loss": 0.1411, |
| "step": 1147 |
| }, |
| { |
| "epoch": 0.8253055355859095, |
| "grad_norm": 2.561779022216797, |
| "learning_rate": 1.7541337167505392e-05, |
| "loss": 0.2081, |
| "step": 1148 |
| }, |
| { |
| "epoch": 0.8260244428468727, |
| "grad_norm": 0.8736401796340942, |
| "learning_rate": 1.746944644140906e-05, |
| "loss": 0.133, |
| "step": 1149 |
| }, |
| { |
| "epoch": 0.8267433501078361, |
| "grad_norm": 0.7084770202636719, |
| "learning_rate": 1.7397555715312725e-05, |
| "loss": 0.0829, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.8274622573687994, |
| "grad_norm": 1.2156176567077637, |
| "learning_rate": 1.7325664989216393e-05, |
| "loss": 0.1417, |
| "step": 1151 |
| }, |
| { |
| "epoch": 0.8281811646297628, |
| "grad_norm": 0.6964887380599976, |
| "learning_rate": 1.7253774263120058e-05, |
| "loss": 0.1165, |
| "step": 1152 |
| }, |
| { |
| "epoch": 0.8289000718907261, |
| "grad_norm": 0.4572407603263855, |
| "learning_rate": 1.7181883537023723e-05, |
| "loss": 0.1019, |
| "step": 1153 |
| }, |
| { |
| "epoch": 0.8296189791516895, |
| "grad_norm": 1.0549578666687012, |
| "learning_rate": 1.710999281092739e-05, |
| "loss": 0.2232, |
| "step": 1154 |
| }, |
| { |
| "epoch": 0.8303378864126527, |
| "grad_norm": 1.239549994468689, |
| "learning_rate": 1.7038102084831055e-05, |
| "loss": 0.2371, |
| "step": 1155 |
| }, |
| { |
| "epoch": 0.8310567936736161, |
| "grad_norm": 0.6848434209823608, |
| "learning_rate": 1.6966211358734723e-05, |
| "loss": 0.1669, |
| "step": 1156 |
| }, |
| { |
| "epoch": 0.8317757009345794, |
| "grad_norm": 1.0799546241760254, |
| "learning_rate": 1.689432063263839e-05, |
| "loss": 0.1382, |
| "step": 1157 |
| }, |
| { |
| "epoch": 0.8324946081955428, |
| "grad_norm": 0.6110497713088989, |
| "learning_rate": 1.6822429906542056e-05, |
| "loss": 0.1197, |
| "step": 1158 |
| }, |
| { |
| "epoch": 0.8332135154565061, |
| "grad_norm": 1.0383644104003906, |
| "learning_rate": 1.6750539180445724e-05, |
| "loss": 0.1653, |
| "step": 1159 |
| }, |
| { |
| "epoch": 0.8339324227174695, |
| "grad_norm": 1.049656629562378, |
| "learning_rate": 1.6678648454349392e-05, |
| "loss": 0.256, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.8346513299784328, |
| "grad_norm": 0.7413502335548401, |
| "learning_rate": 1.6606757728253057e-05, |
| "loss": 0.0893, |
| "step": 1161 |
| }, |
| { |
| "epoch": 0.8353702372393961, |
| "grad_norm": 1.5397248268127441, |
| "learning_rate": 1.653486700215672e-05, |
| "loss": 0.1297, |
| "step": 1162 |
| }, |
| { |
| "epoch": 0.8360891445003594, |
| "grad_norm": 0.9013486504554749, |
| "learning_rate": 1.6462976276060386e-05, |
| "loss": 0.168, |
| "step": 1163 |
| }, |
| { |
| "epoch": 0.8368080517613228, |
| "grad_norm": 1.724400520324707, |
| "learning_rate": 1.6391085549964054e-05, |
| "loss": 0.1218, |
| "step": 1164 |
| }, |
| { |
| "epoch": 0.8375269590222861, |
| "grad_norm": 1.069604754447937, |
| "learning_rate": 1.6319194823867722e-05, |
| "loss": 0.118, |
| "step": 1165 |
| }, |
| { |
| "epoch": 0.8382458662832495, |
| "grad_norm": 1.2329519987106323, |
| "learning_rate": 1.6247304097771387e-05, |
| "loss": 0.2542, |
| "step": 1166 |
| }, |
| { |
| "epoch": 0.8389647735442128, |
| "grad_norm": 0.5549449920654297, |
| "learning_rate": 1.6175413371675055e-05, |
| "loss": 0.0964, |
| "step": 1167 |
| }, |
| { |
| "epoch": 0.8396836808051761, |
| "grad_norm": 0.7650552988052368, |
| "learning_rate": 1.6103522645578723e-05, |
| "loss": 0.1622, |
| "step": 1168 |
| }, |
| { |
| "epoch": 0.8404025880661394, |
| "grad_norm": 0.6795350313186646, |
| "learning_rate": 1.6031631919482388e-05, |
| "loss": 0.088, |
| "step": 1169 |
| }, |
| { |
| "epoch": 0.8411214953271028, |
| "grad_norm": 1.0138665437698364, |
| "learning_rate": 1.5959741193386056e-05, |
| "loss": 0.1404, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.8418404025880661, |
| "grad_norm": 0.5485491156578064, |
| "learning_rate": 1.588785046728972e-05, |
| "loss": 0.0926, |
| "step": 1171 |
| }, |
| { |
| "epoch": 0.8425593098490295, |
| "grad_norm": 1.2163599729537964, |
| "learning_rate": 1.5815959741193385e-05, |
| "loss": 0.1965, |
| "step": 1172 |
| }, |
| { |
| "epoch": 0.8432782171099928, |
| "grad_norm": 2.7217113971710205, |
| "learning_rate": 1.5744069015097054e-05, |
| "loss": 0.2459, |
| "step": 1173 |
| }, |
| { |
| "epoch": 0.8439971243709562, |
| "grad_norm": 1.0199272632598877, |
| "learning_rate": 1.5672178289000718e-05, |
| "loss": 0.1179, |
| "step": 1174 |
| }, |
| { |
| "epoch": 0.8447160316319194, |
| "grad_norm": 0.5198338031768799, |
| "learning_rate": 1.5600287562904386e-05, |
| "loss": 0.155, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.8454349388928828, |
| "grad_norm": 0.724979817867279, |
| "learning_rate": 1.552839683680805e-05, |
| "loss": 0.1169, |
| "step": 1176 |
| }, |
| { |
| "epoch": 0.8461538461538461, |
| "grad_norm": 1.433951497077942, |
| "learning_rate": 1.545650611071172e-05, |
| "loss": 0.1763, |
| "step": 1177 |
| }, |
| { |
| "epoch": 0.8468727534148095, |
| "grad_norm": 1.2934740781784058, |
| "learning_rate": 1.5384615384615387e-05, |
| "loss": 0.1572, |
| "step": 1178 |
| }, |
| { |
| "epoch": 0.8475916606757729, |
| "grad_norm": 0.5693601965904236, |
| "learning_rate": 1.5312724658519052e-05, |
| "loss": 0.1161, |
| "step": 1179 |
| }, |
| { |
| "epoch": 0.8483105679367362, |
| "grad_norm": 0.961031973361969, |
| "learning_rate": 1.5240833932422718e-05, |
| "loss": 0.1729, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.8490294751976994, |
| "grad_norm": 0.4837125837802887, |
| "learning_rate": 1.5168943206326386e-05, |
| "loss": 0.1648, |
| "step": 1181 |
| }, |
| { |
| "epoch": 0.8497483824586628, |
| "grad_norm": 0.5368496179580688, |
| "learning_rate": 1.5097052480230051e-05, |
| "loss": 0.12, |
| "step": 1182 |
| }, |
| { |
| "epoch": 0.8504672897196262, |
| "grad_norm": 0.6104084849357605, |
| "learning_rate": 1.5025161754133717e-05, |
| "loss": 0.1461, |
| "step": 1183 |
| }, |
| { |
| "epoch": 0.8511861969805895, |
| "grad_norm": 1.1371060609817505, |
| "learning_rate": 1.4953271028037382e-05, |
| "loss": 0.1465, |
| "step": 1184 |
| }, |
| { |
| "epoch": 0.8519051042415529, |
| "grad_norm": 0.750773012638092, |
| "learning_rate": 1.488138030194105e-05, |
| "loss": 0.1386, |
| "step": 1185 |
| }, |
| { |
| "epoch": 0.8526240115025162, |
| "grad_norm": 0.49192139506340027, |
| "learning_rate": 1.4809489575844718e-05, |
| "loss": 0.1267, |
| "step": 1186 |
| }, |
| { |
| "epoch": 0.8533429187634796, |
| "grad_norm": 1.0582698583602905, |
| "learning_rate": 1.4737598849748383e-05, |
| "loss": 0.1, |
| "step": 1187 |
| }, |
| { |
| "epoch": 0.8540618260244428, |
| "grad_norm": 0.6468744277954102, |
| "learning_rate": 1.466570812365205e-05, |
| "loss": 0.1478, |
| "step": 1188 |
| }, |
| { |
| "epoch": 0.8547807332854062, |
| "grad_norm": 0.7181344032287598, |
| "learning_rate": 1.4593817397555714e-05, |
| "loss": 0.1331, |
| "step": 1189 |
| }, |
| { |
| "epoch": 0.8554996405463695, |
| "grad_norm": 0.6007802486419678, |
| "learning_rate": 1.4521926671459382e-05, |
| "loss": 0.1472, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.8562185478073329, |
| "grad_norm": 1.2862764596939087, |
| "learning_rate": 1.445003594536305e-05, |
| "loss": 0.1171, |
| "step": 1191 |
| }, |
| { |
| "epoch": 0.8569374550682962, |
| "grad_norm": 0.8010552525520325, |
| "learning_rate": 1.4378145219266715e-05, |
| "loss": 0.1114, |
| "step": 1192 |
| }, |
| { |
| "epoch": 0.8576563623292596, |
| "grad_norm": 0.5855614542961121, |
| "learning_rate": 1.4306254493170381e-05, |
| "loss": 0.1434, |
| "step": 1193 |
| }, |
| { |
| "epoch": 0.8583752695902228, |
| "grad_norm": 0.834697425365448, |
| "learning_rate": 1.423436376707405e-05, |
| "loss": 0.1178, |
| "step": 1194 |
| }, |
| { |
| "epoch": 0.8590941768511862, |
| "grad_norm": 1.7494937181472778, |
| "learning_rate": 1.4162473040977714e-05, |
| "loss": 0.1778, |
| "step": 1195 |
| }, |
| { |
| "epoch": 0.8598130841121495, |
| "grad_norm": 1.145895004272461, |
| "learning_rate": 1.4090582314881382e-05, |
| "loss": 0.1398, |
| "step": 1196 |
| }, |
| { |
| "epoch": 0.8605319913731129, |
| "grad_norm": 1.2310457229614258, |
| "learning_rate": 1.4018691588785047e-05, |
| "loss": 0.0936, |
| "step": 1197 |
| }, |
| { |
| "epoch": 0.8612508986340762, |
| "grad_norm": 0.574589729309082, |
| "learning_rate": 1.3946800862688713e-05, |
| "loss": 0.116, |
| "step": 1198 |
| }, |
| { |
| "epoch": 0.8619698058950396, |
| "grad_norm": 0.8383525013923645, |
| "learning_rate": 1.3874910136592381e-05, |
| "loss": 0.1146, |
| "step": 1199 |
| }, |
| { |
| "epoch": 0.8626887131560029, |
| "grad_norm": 0.4317038357257843, |
| "learning_rate": 1.3803019410496046e-05, |
| "loss": 0.1114, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.8626887131560029, |
| "eval_accuracy": 0.935278058052063, |
| "eval_loss": 0.12901614606380463, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6572470664978027, |
| "eval_precision": 0.9278779625892639, |
| "eval_recall": 0.9723837375640869, |
| "eval_runtime": 605.0393, |
| "eval_samples_per_second": 4.6, |
| "eval_steps_per_second": 0.575, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.8634076204169662, |
| "grad_norm": 0.5123474597930908, |
| "learning_rate": 1.3731128684399714e-05, |
| "loss": 0.0877, |
| "step": 1201 |
| }, |
| { |
| "epoch": 0.8641265276779295, |
| "grad_norm": 1.885433316230774, |
| "learning_rate": 1.365923795830338e-05, |
| "loss": 0.1107, |
| "step": 1202 |
| }, |
| { |
| "epoch": 0.8648454349388929, |
| "grad_norm": 0.5070977210998535, |
| "learning_rate": 1.3587347232207045e-05, |
| "loss": 0.0845, |
| "step": 1203 |
| }, |
| { |
| "epoch": 0.8655643421998562, |
| "grad_norm": 0.6891940236091614, |
| "learning_rate": 1.3515456506110713e-05, |
| "loss": 0.1428, |
| "step": 1204 |
| }, |
| { |
| "epoch": 0.8662832494608196, |
| "grad_norm": 0.7068796753883362, |
| "learning_rate": 1.3443565780014378e-05, |
| "loss": 0.1481, |
| "step": 1205 |
| }, |
| { |
| "epoch": 0.8670021567217829, |
| "grad_norm": 0.8890682458877563, |
| "learning_rate": 1.3371675053918046e-05, |
| "loss": 0.141, |
| "step": 1206 |
| }, |
| { |
| "epoch": 0.8677210639827462, |
| "grad_norm": 0.9557405114173889, |
| "learning_rate": 1.3299784327821712e-05, |
| "loss": 0.106, |
| "step": 1207 |
| }, |
| { |
| "epoch": 0.8684399712437095, |
| "grad_norm": 0.4333617389202118, |
| "learning_rate": 1.3227893601725377e-05, |
| "loss": 0.081, |
| "step": 1208 |
| }, |
| { |
| "epoch": 0.8691588785046729, |
| "grad_norm": 0.7648006081581116, |
| "learning_rate": 1.3156002875629045e-05, |
| "loss": 0.1676, |
| "step": 1209 |
| }, |
| { |
| "epoch": 0.8698777857656362, |
| "grad_norm": 0.5260509252548218, |
| "learning_rate": 1.308411214953271e-05, |
| "loss": 0.0696, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.8705966930265996, |
| "grad_norm": 0.8238019347190857, |
| "learning_rate": 1.3012221423436378e-05, |
| "loss": 0.0857, |
| "step": 1211 |
| }, |
| { |
| "epoch": 0.8713156002875629, |
| "grad_norm": 0.3782735764980316, |
| "learning_rate": 1.2940330697340044e-05, |
| "loss": 0.0784, |
| "step": 1212 |
| }, |
| { |
| "epoch": 0.8720345075485263, |
| "grad_norm": 0.959868848323822, |
| "learning_rate": 1.2868439971243709e-05, |
| "loss": 0.151, |
| "step": 1213 |
| }, |
| { |
| "epoch": 0.8727534148094895, |
| "grad_norm": 0.5242748856544495, |
| "learning_rate": 1.2796549245147377e-05, |
| "loss": 0.1021, |
| "step": 1214 |
| }, |
| { |
| "epoch": 0.8734723220704529, |
| "grad_norm": 0.6725254654884338, |
| "learning_rate": 1.2724658519051045e-05, |
| "loss": 0.1287, |
| "step": 1215 |
| }, |
| { |
| "epoch": 0.8741912293314162, |
| "grad_norm": 0.6339101791381836, |
| "learning_rate": 1.265276779295471e-05, |
| "loss": 0.1488, |
| "step": 1216 |
| }, |
| { |
| "epoch": 0.8749101365923796, |
| "grad_norm": 0.8921059370040894, |
| "learning_rate": 1.2580877066858376e-05, |
| "loss": 0.1936, |
| "step": 1217 |
| }, |
| { |
| "epoch": 0.875629043853343, |
| "grad_norm": 3.053365468978882, |
| "learning_rate": 1.250898634076204e-05, |
| "loss": 0.2308, |
| "step": 1218 |
| }, |
| { |
| "epoch": 0.8763479511143063, |
| "grad_norm": 0.5890373587608337, |
| "learning_rate": 1.2437095614665709e-05, |
| "loss": 0.1533, |
| "step": 1219 |
| }, |
| { |
| "epoch": 0.8770668583752695, |
| "grad_norm": 1.7572747468948364, |
| "learning_rate": 1.2365204888569375e-05, |
| "loss": 0.1628, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.8777857656362329, |
| "grad_norm": 2.0376241207122803, |
| "learning_rate": 1.2293314162473042e-05, |
| "loss": 0.2253, |
| "step": 1221 |
| }, |
| { |
| "epoch": 0.8785046728971962, |
| "grad_norm": 1.0158257484436035, |
| "learning_rate": 1.2221423436376708e-05, |
| "loss": 0.1882, |
| "step": 1222 |
| }, |
| { |
| "epoch": 0.8792235801581596, |
| "grad_norm": 0.6919081211090088, |
| "learning_rate": 1.2149532710280374e-05, |
| "loss": 0.1708, |
| "step": 1223 |
| }, |
| { |
| "epoch": 0.879942487419123, |
| "grad_norm": 0.5944404602050781, |
| "learning_rate": 1.207764198418404e-05, |
| "loss": 0.093, |
| "step": 1224 |
| }, |
| { |
| "epoch": 0.8806613946800863, |
| "grad_norm": 0.4297444522380829, |
| "learning_rate": 1.2005751258087707e-05, |
| "loss": 0.0746, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.8813803019410497, |
| "grad_norm": 0.863753616809845, |
| "learning_rate": 1.1933860531991374e-05, |
| "loss": 0.1563, |
| "step": 1226 |
| }, |
| { |
| "epoch": 0.8820992092020129, |
| "grad_norm": 0.8307890295982361, |
| "learning_rate": 1.186196980589504e-05, |
| "loss": 0.094, |
| "step": 1227 |
| }, |
| { |
| "epoch": 0.8828181164629763, |
| "grad_norm": 0.5512217283248901, |
| "learning_rate": 1.1790079079798706e-05, |
| "loss": 0.1091, |
| "step": 1228 |
| }, |
| { |
| "epoch": 0.8835370237239396, |
| "grad_norm": 0.7317647337913513, |
| "learning_rate": 1.1718188353702373e-05, |
| "loss": 0.1563, |
| "step": 1229 |
| }, |
| { |
| "epoch": 0.884255930984903, |
| "grad_norm": 0.8947478532791138, |
| "learning_rate": 1.1646297627606039e-05, |
| "loss": 0.0928, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.8849748382458663, |
| "grad_norm": 1.199896216392517, |
| "learning_rate": 1.1574406901509705e-05, |
| "loss": 0.2151, |
| "step": 1231 |
| }, |
| { |
| "epoch": 0.8856937455068297, |
| "grad_norm": 0.877432644367218, |
| "learning_rate": 1.1502516175413372e-05, |
| "loss": 0.1336, |
| "step": 1232 |
| }, |
| { |
| "epoch": 0.8864126527677929, |
| "grad_norm": 0.7458563446998596, |
| "learning_rate": 1.1430625449317038e-05, |
| "loss": 0.1623, |
| "step": 1233 |
| }, |
| { |
| "epoch": 0.8871315600287563, |
| "grad_norm": 1.2673143148422241, |
| "learning_rate": 1.1358734723220706e-05, |
| "loss": 0.1721, |
| "step": 1234 |
| }, |
| { |
| "epoch": 0.8878504672897196, |
| "grad_norm": 0.8390029072761536, |
| "learning_rate": 1.1286843997124371e-05, |
| "loss": 0.1064, |
| "step": 1235 |
| }, |
| { |
| "epoch": 0.888569374550683, |
| "grad_norm": 1.2902019023895264, |
| "learning_rate": 1.1214953271028037e-05, |
| "loss": 0.0825, |
| "step": 1236 |
| }, |
| { |
| "epoch": 0.8892882818116463, |
| "grad_norm": 1.12763249874115, |
| "learning_rate": 1.1143062544931704e-05, |
| "loss": 0.1878, |
| "step": 1237 |
| }, |
| { |
| "epoch": 0.8900071890726097, |
| "grad_norm": 0.6045496463775635, |
| "learning_rate": 1.1071171818835372e-05, |
| "loss": 0.1234, |
| "step": 1238 |
| }, |
| { |
| "epoch": 0.8907260963335729, |
| "grad_norm": 0.8368903994560242, |
| "learning_rate": 1.0999281092739038e-05, |
| "loss": 0.1762, |
| "step": 1239 |
| }, |
| { |
| "epoch": 0.8914450035945363, |
| "grad_norm": 0.4690748453140259, |
| "learning_rate": 1.0927390366642703e-05, |
| "loss": 0.0718, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.8921639108554996, |
| "grad_norm": 0.5496010184288025, |
| "learning_rate": 1.085549964054637e-05, |
| "loss": 0.1243, |
| "step": 1241 |
| }, |
| { |
| "epoch": 0.892882818116463, |
| "grad_norm": 0.5322665572166443, |
| "learning_rate": 1.0783608914450037e-05, |
| "loss": 0.0866, |
| "step": 1242 |
| }, |
| { |
| "epoch": 0.8936017253774263, |
| "grad_norm": 0.7182154059410095, |
| "learning_rate": 1.0711718188353704e-05, |
| "loss": 0.0742, |
| "step": 1243 |
| }, |
| { |
| "epoch": 0.8943206326383897, |
| "grad_norm": 0.5049268007278442, |
| "learning_rate": 1.0639827462257368e-05, |
| "loss": 0.1587, |
| "step": 1244 |
| }, |
| { |
| "epoch": 0.895039539899353, |
| "grad_norm": 2.293687105178833, |
| "learning_rate": 1.0567936736161035e-05, |
| "loss": 0.2252, |
| "step": 1245 |
| }, |
| { |
| "epoch": 0.8957584471603163, |
| "grad_norm": 1.0248525142669678, |
| "learning_rate": 1.0496046010064701e-05, |
| "loss": 0.1158, |
| "step": 1246 |
| }, |
| { |
| "epoch": 0.8964773544212796, |
| "grad_norm": 0.7257125973701477, |
| "learning_rate": 1.042415528396837e-05, |
| "loss": 0.1845, |
| "step": 1247 |
| }, |
| { |
| "epoch": 0.897196261682243, |
| "grad_norm": 0.9349539279937744, |
| "learning_rate": 1.0352264557872036e-05, |
| "loss": 0.0823, |
| "step": 1248 |
| }, |
| { |
| "epoch": 0.8979151689432063, |
| "grad_norm": 0.7310618758201599, |
| "learning_rate": 1.02803738317757e-05, |
| "loss": 0.1256, |
| "step": 1249 |
| }, |
| { |
| "epoch": 0.8986340762041697, |
| "grad_norm": 1.7009799480438232, |
| "learning_rate": 1.0208483105679367e-05, |
| "loss": 0.1628, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.899352983465133, |
| "grad_norm": 0.4560825824737549, |
| "learning_rate": 1.0136592379583035e-05, |
| "loss": 0.139, |
| "step": 1251 |
| }, |
| { |
| "epoch": 0.9000718907260963, |
| "grad_norm": 0.7323896288871765, |
| "learning_rate": 1.0064701653486701e-05, |
| "loss": 0.1153, |
| "step": 1252 |
| }, |
| { |
| "epoch": 0.9007907979870596, |
| "grad_norm": 0.8497075438499451, |
| "learning_rate": 9.992810927390368e-06, |
| "loss": 0.1548, |
| "step": 1253 |
| }, |
| { |
| "epoch": 0.901509705248023, |
| "grad_norm": 0.4729348421096802, |
| "learning_rate": 9.920920201294032e-06, |
| "loss": 0.093, |
| "step": 1254 |
| }, |
| { |
| "epoch": 0.9022286125089863, |
| "grad_norm": 0.49565210938453674, |
| "learning_rate": 9.8490294751977e-06, |
| "loss": 0.1337, |
| "step": 1255 |
| }, |
| { |
| "epoch": 0.9029475197699497, |
| "grad_norm": 0.48817089200019836, |
| "learning_rate": 9.777138749101367e-06, |
| "loss": 0.0649, |
| "step": 1256 |
| }, |
| { |
| "epoch": 0.903666427030913, |
| "grad_norm": 0.573188066482544, |
| "learning_rate": 9.705248023005033e-06, |
| "loss": 0.1006, |
| "step": 1257 |
| }, |
| { |
| "epoch": 0.9043853342918764, |
| "grad_norm": 0.45638272166252136, |
| "learning_rate": 9.6333572969087e-06, |
| "loss": 0.0946, |
| "step": 1258 |
| }, |
| { |
| "epoch": 0.9051042415528396, |
| "grad_norm": 0.7156769037246704, |
| "learning_rate": 9.561466570812366e-06, |
| "loss": 0.1406, |
| "step": 1259 |
| }, |
| { |
| "epoch": 0.905823148813803, |
| "grad_norm": 0.4424365162849426, |
| "learning_rate": 9.489575844716032e-06, |
| "loss": 0.0854, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.9065420560747663, |
| "grad_norm": 1.0452828407287598, |
| "learning_rate": 9.417685118619699e-06, |
| "loss": 0.2, |
| "step": 1261 |
| }, |
| { |
| "epoch": 0.9072609633357297, |
| "grad_norm": 1.5137022733688354, |
| "learning_rate": 9.345794392523365e-06, |
| "loss": 0.1965, |
| "step": 1262 |
| }, |
| { |
| "epoch": 0.907979870596693, |
| "grad_norm": 0.807569146156311, |
| "learning_rate": 9.273903666427031e-06, |
| "loss": 0.1726, |
| "step": 1263 |
| }, |
| { |
| "epoch": 0.9086987778576564, |
| "grad_norm": 1.1127057075500488, |
| "learning_rate": 9.202012940330698e-06, |
| "loss": 0.1116, |
| "step": 1264 |
| }, |
| { |
| "epoch": 0.9094176851186196, |
| "grad_norm": 1.5959947109222412, |
| "learning_rate": 9.130122214234364e-06, |
| "loss": 0.2429, |
| "step": 1265 |
| }, |
| { |
| "epoch": 0.910136592379583, |
| "grad_norm": 1.2956665754318237, |
| "learning_rate": 9.05823148813803e-06, |
| "loss": 0.1809, |
| "step": 1266 |
| }, |
| { |
| "epoch": 0.9108554996405464, |
| "grad_norm": 0.9982889294624329, |
| "learning_rate": 8.986340762041697e-06, |
| "loss": 0.1431, |
| "step": 1267 |
| }, |
| { |
| "epoch": 0.9115744069015097, |
| "grad_norm": 0.6282898187637329, |
| "learning_rate": 8.914450035945363e-06, |
| "loss": 0.0983, |
| "step": 1268 |
| }, |
| { |
| "epoch": 0.9122933141624731, |
| "grad_norm": 1.4935771226882935, |
| "learning_rate": 8.84255930984903e-06, |
| "loss": 0.1082, |
| "step": 1269 |
| }, |
| { |
| "epoch": 0.9130122214234364, |
| "grad_norm": 0.9130723476409912, |
| "learning_rate": 8.770668583752696e-06, |
| "loss": 0.1087, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.9137311286843998, |
| "grad_norm": 0.8891334533691406, |
| "learning_rate": 8.698777857656363e-06, |
| "loss": 0.1042, |
| "step": 1271 |
| }, |
| { |
| "epoch": 0.914450035945363, |
| "grad_norm": 1.134940266609192, |
| "learning_rate": 8.626887131560029e-06, |
| "loss": 0.2398, |
| "step": 1272 |
| }, |
| { |
| "epoch": 0.9151689432063264, |
| "grad_norm": 0.5197407603263855, |
| "learning_rate": 8.554996405463695e-06, |
| "loss": 0.0807, |
| "step": 1273 |
| }, |
| { |
| "epoch": 0.9158878504672897, |
| "grad_norm": 0.8537404537200928, |
| "learning_rate": 8.483105679367362e-06, |
| "loss": 0.0884, |
| "step": 1274 |
| }, |
| { |
| "epoch": 0.9166067577282531, |
| "grad_norm": 1.1788685321807861, |
| "learning_rate": 8.411214953271028e-06, |
| "loss": 0.1095, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.9173256649892164, |
| "grad_norm": 0.445051908493042, |
| "learning_rate": 8.339324227174696e-06, |
| "loss": 0.1009, |
| "step": 1276 |
| }, |
| { |
| "epoch": 0.9180445722501798, |
| "grad_norm": 0.41120973229408264, |
| "learning_rate": 8.26743350107836e-06, |
| "loss": 0.0836, |
| "step": 1277 |
| }, |
| { |
| "epoch": 0.918763479511143, |
| "grad_norm": 0.5132132768630981, |
| "learning_rate": 8.195542774982027e-06, |
| "loss": 0.0852, |
| "step": 1278 |
| }, |
| { |
| "epoch": 0.9194823867721064, |
| "grad_norm": 0.5090476870536804, |
| "learning_rate": 8.123652048885694e-06, |
| "loss": 0.0848, |
| "step": 1279 |
| }, |
| { |
| "epoch": 0.9202012940330697, |
| "grad_norm": 0.7417925596237183, |
| "learning_rate": 8.051761322789362e-06, |
| "loss": 0.1451, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.9209202012940331, |
| "grad_norm": 0.5560879111289978, |
| "learning_rate": 7.979870596693028e-06, |
| "loss": 0.1138, |
| "step": 1281 |
| }, |
| { |
| "epoch": 0.9216391085549964, |
| "grad_norm": 0.6955628395080566, |
| "learning_rate": 7.907979870596693e-06, |
| "loss": 0.0699, |
| "step": 1282 |
| }, |
| { |
| "epoch": 0.9223580158159598, |
| "grad_norm": 1.3208876848220825, |
| "learning_rate": 7.836089144500359e-06, |
| "loss": 0.1963, |
| "step": 1283 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 1.715364933013916, |
| "learning_rate": 7.764198418404026e-06, |
| "loss": 0.1423, |
| "step": 1284 |
| }, |
| { |
| "epoch": 0.9237958303378864, |
| "grad_norm": 0.8714684247970581, |
| "learning_rate": 7.692307692307694e-06, |
| "loss": 0.078, |
| "step": 1285 |
| }, |
| { |
| "epoch": 0.9245147375988497, |
| "grad_norm": 2.256927490234375, |
| "learning_rate": 7.620416966211359e-06, |
| "loss": 0.2242, |
| "step": 1286 |
| }, |
| { |
| "epoch": 0.9252336448598131, |
| "grad_norm": 0.7324031591415405, |
| "learning_rate": 7.5485262401150255e-06, |
| "loss": 0.0673, |
| "step": 1287 |
| }, |
| { |
| "epoch": 0.9259525521207764, |
| "grad_norm": 0.8331296443939209, |
| "learning_rate": 7.476635514018691e-06, |
| "loss": 0.0842, |
| "step": 1288 |
| }, |
| { |
| "epoch": 0.9266714593817398, |
| "grad_norm": 1.0670703649520874, |
| "learning_rate": 7.404744787922359e-06, |
| "loss": 0.1727, |
| "step": 1289 |
| }, |
| { |
| "epoch": 0.9273903666427031, |
| "grad_norm": 0.6906760931015015, |
| "learning_rate": 7.332854061826025e-06, |
| "loss": 0.1341, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.9281092739036664, |
| "grad_norm": 0.5483196973800659, |
| "learning_rate": 7.260963335729691e-06, |
| "loss": 0.0928, |
| "step": 1291 |
| }, |
| { |
| "epoch": 0.9288281811646297, |
| "grad_norm": 0.8699132800102234, |
| "learning_rate": 7.1890726096333574e-06, |
| "loss": 0.1499, |
| "step": 1292 |
| }, |
| { |
| "epoch": 0.9295470884255931, |
| "grad_norm": 0.6977189183235168, |
| "learning_rate": 7.117181883537025e-06, |
| "loss": 0.1802, |
| "step": 1293 |
| }, |
| { |
| "epoch": 0.9302659956865564, |
| "grad_norm": 1.273950457572937, |
| "learning_rate": 7.045291157440691e-06, |
| "loss": 0.1358, |
| "step": 1294 |
| }, |
| { |
| "epoch": 0.9309849029475198, |
| "grad_norm": 0.6622292995452881, |
| "learning_rate": 6.9734004313443566e-06, |
| "loss": 0.1106, |
| "step": 1295 |
| }, |
| { |
| "epoch": 0.9317038102084831, |
| "grad_norm": 1.048762559890747, |
| "learning_rate": 6.901509705248023e-06, |
| "loss": 0.1778, |
| "step": 1296 |
| }, |
| { |
| "epoch": 0.9324227174694465, |
| "grad_norm": 0.7745311260223389, |
| "learning_rate": 6.82961897915169e-06, |
| "loss": 0.131, |
| "step": 1297 |
| }, |
| { |
| "epoch": 0.9331416247304097, |
| "grad_norm": 1.3940768241882324, |
| "learning_rate": 6.7577282530553566e-06, |
| "loss": 0.1341, |
| "step": 1298 |
| }, |
| { |
| "epoch": 0.9338605319913731, |
| "grad_norm": 0.6330265998840332, |
| "learning_rate": 6.685837526959023e-06, |
| "loss": 0.1633, |
| "step": 1299 |
| }, |
| { |
| "epoch": 0.9345794392523364, |
| "grad_norm": 2.3914084434509277, |
| "learning_rate": 6.6139468008626885e-06, |
| "loss": 0.2455, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.9352983465132998, |
| "grad_norm": 0.9312193989753723, |
| "learning_rate": 6.542056074766355e-06, |
| "loss": 0.119, |
| "step": 1301 |
| }, |
| { |
| "epoch": 0.9360172537742631, |
| "grad_norm": 0.8084568381309509, |
| "learning_rate": 6.470165348670022e-06, |
| "loss": 0.1669, |
| "step": 1302 |
| }, |
| { |
| "epoch": 0.9367361610352265, |
| "grad_norm": 0.511916995048523, |
| "learning_rate": 6.3982746225736885e-06, |
| "loss": 0.0892, |
| "step": 1303 |
| }, |
| { |
| "epoch": 0.9374550682961897, |
| "grad_norm": 0.6594161987304688, |
| "learning_rate": 6.326383896477355e-06, |
| "loss": 0.0972, |
| "step": 1304 |
| }, |
| { |
| "epoch": 0.9381739755571531, |
| "grad_norm": 0.4225222170352936, |
| "learning_rate": 6.25449317038102e-06, |
| "loss": 0.076, |
| "step": 1305 |
| }, |
| { |
| "epoch": 0.9388928828181164, |
| "grad_norm": 0.5570008754730225, |
| "learning_rate": 6.182602444284688e-06, |
| "loss": 0.1262, |
| "step": 1306 |
| }, |
| { |
| "epoch": 0.9396117900790798, |
| "grad_norm": 0.6197625994682312, |
| "learning_rate": 6.110711718188354e-06, |
| "loss": 0.1194, |
| "step": 1307 |
| }, |
| { |
| "epoch": 0.9403306973400432, |
| "grad_norm": 0.682424783706665, |
| "learning_rate": 6.03882099209202e-06, |
| "loss": 0.0887, |
| "step": 1308 |
| }, |
| { |
| "epoch": 0.9410496046010065, |
| "grad_norm": 1.267173171043396, |
| "learning_rate": 5.966930265995687e-06, |
| "loss": 0.1657, |
| "step": 1309 |
| }, |
| { |
| "epoch": 0.9417685118619699, |
| "grad_norm": 1.0641415119171143, |
| "learning_rate": 5.895039539899353e-06, |
| "loss": 0.1663, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.9424874191229331, |
| "grad_norm": 0.4511805474758148, |
| "learning_rate": 5.8231488138030196e-06, |
| "loss": 0.0862, |
| "step": 1311 |
| }, |
| { |
| "epoch": 0.9432063263838965, |
| "grad_norm": 0.8811210989952087, |
| "learning_rate": 5.751258087706686e-06, |
| "loss": 0.1511, |
| "step": 1312 |
| }, |
| { |
| "epoch": 0.9439252336448598, |
| "grad_norm": 1.1976115703582764, |
| "learning_rate": 5.679367361610353e-06, |
| "loss": 0.1451, |
| "step": 1313 |
| }, |
| { |
| "epoch": 0.9446441409058232, |
| "grad_norm": 0.6144181489944458, |
| "learning_rate": 5.607476635514019e-06, |
| "loss": 0.0862, |
| "step": 1314 |
| }, |
| { |
| "epoch": 0.9453630481667865, |
| "grad_norm": 0.5378531217575073, |
| "learning_rate": 5.535585909417686e-06, |
| "loss": 0.1186, |
| "step": 1315 |
| }, |
| { |
| "epoch": 0.9460819554277499, |
| "grad_norm": 0.9116848707199097, |
| "learning_rate": 5.4636951833213515e-06, |
| "loss": 0.0991, |
| "step": 1316 |
| }, |
| { |
| "epoch": 0.9468008626887131, |
| "grad_norm": 1.1218457221984863, |
| "learning_rate": 5.391804457225019e-06, |
| "loss": 0.1193, |
| "step": 1317 |
| }, |
| { |
| "epoch": 0.9475197699496765, |
| "grad_norm": 0.677040159702301, |
| "learning_rate": 5.319913731128684e-06, |
| "loss": 0.1077, |
| "step": 1318 |
| }, |
| { |
| "epoch": 0.9482386772106398, |
| "grad_norm": 1.1725988388061523, |
| "learning_rate": 5.248023005032351e-06, |
| "loss": 0.1745, |
| "step": 1319 |
| }, |
| { |
| "epoch": 0.9489575844716032, |
| "grad_norm": 0.5412241220474243, |
| "learning_rate": 5.176132278936018e-06, |
| "loss": 0.1307, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.9496764917325665, |
| "grad_norm": 0.6392889618873596, |
| "learning_rate": 5.104241552839683e-06, |
| "loss": 0.1727, |
| "step": 1321 |
| }, |
| { |
| "epoch": 0.9503953989935299, |
| "grad_norm": 1.362302541732788, |
| "learning_rate": 5.032350826743351e-06, |
| "loss": 0.2428, |
| "step": 1322 |
| }, |
| { |
| "epoch": 0.9511143062544932, |
| "grad_norm": 0.45226263999938965, |
| "learning_rate": 4.960460100647016e-06, |
| "loss": 0.1168, |
| "step": 1323 |
| }, |
| { |
| "epoch": 0.9518332135154565, |
| "grad_norm": 0.6780915856361389, |
| "learning_rate": 4.888569374550683e-06, |
| "loss": 0.1473, |
| "step": 1324 |
| }, |
| { |
| "epoch": 0.9525521207764198, |
| "grad_norm": 2.142181158065796, |
| "learning_rate": 4.81667864845435e-06, |
| "loss": 0.238, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.9532710280373832, |
| "grad_norm": 0.5581359267234802, |
| "learning_rate": 4.744787922358016e-06, |
| "loss": 0.0794, |
| "step": 1326 |
| }, |
| { |
| "epoch": 0.9539899352983465, |
| "grad_norm": 1.5413146018981934, |
| "learning_rate": 4.6728971962616825e-06, |
| "loss": 0.1664, |
| "step": 1327 |
| }, |
| { |
| "epoch": 0.9547088425593099, |
| "grad_norm": 0.5772722363471985, |
| "learning_rate": 4.601006470165349e-06, |
| "loss": 0.1159, |
| "step": 1328 |
| }, |
| { |
| "epoch": 0.9554277498202732, |
| "grad_norm": 0.7360172867774963, |
| "learning_rate": 4.529115744069015e-06, |
| "loss": 0.1726, |
| "step": 1329 |
| }, |
| { |
| "epoch": 0.9561466570812365, |
| "grad_norm": 0.698972225189209, |
| "learning_rate": 4.457225017972682e-06, |
| "loss": 0.1219, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.9568655643421998, |
| "grad_norm": 0.7607867121696472, |
| "learning_rate": 4.385334291876348e-06, |
| "loss": 0.1139, |
| "step": 1331 |
| }, |
| { |
| "epoch": 0.9575844716031632, |
| "grad_norm": 0.6001602411270142, |
| "learning_rate": 4.3134435657800145e-06, |
| "loss": 0.148, |
| "step": 1332 |
| }, |
| { |
| "epoch": 0.9583033788641265, |
| "grad_norm": 0.8180660009384155, |
| "learning_rate": 4.241552839683681e-06, |
| "loss": 0.1843, |
| "step": 1333 |
| }, |
| { |
| "epoch": 0.9590222861250899, |
| "grad_norm": 1.0921334028244019, |
| "learning_rate": 4.169662113587348e-06, |
| "loss": 0.1703, |
| "step": 1334 |
| }, |
| { |
| "epoch": 0.9597411933860532, |
| "grad_norm": 1.6129685640335083, |
| "learning_rate": 4.097771387491014e-06, |
| "loss": 0.2274, |
| "step": 1335 |
| }, |
| { |
| "epoch": 0.9604601006470166, |
| "grad_norm": 1.4993046522140503, |
| "learning_rate": 4.025880661394681e-06, |
| "loss": 0.2663, |
| "step": 1336 |
| }, |
| { |
| "epoch": 0.9611790079079798, |
| "grad_norm": 1.416886568069458, |
| "learning_rate": 3.953989935298346e-06, |
| "loss": 0.2684, |
| "step": 1337 |
| }, |
| { |
| "epoch": 0.9618979151689432, |
| "grad_norm": 0.6574516296386719, |
| "learning_rate": 3.882099209202013e-06, |
| "loss": 0.1619, |
| "step": 1338 |
| }, |
| { |
| "epoch": 0.9626168224299065, |
| "grad_norm": 0.5548767447471619, |
| "learning_rate": 3.8102084831056796e-06, |
| "loss": 0.1215, |
| "step": 1339 |
| }, |
| { |
| "epoch": 0.9633357296908699, |
| "grad_norm": 0.5321860313415527, |
| "learning_rate": 3.7383177570093455e-06, |
| "loss": 0.1086, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.9640546369518332, |
| "grad_norm": 0.5427797436714172, |
| "learning_rate": 3.6664270309130123e-06, |
| "loss": 0.1673, |
| "step": 1341 |
| }, |
| { |
| "epoch": 0.9647735442127966, |
| "grad_norm": 0.64905846118927, |
| "learning_rate": 3.5945363048166787e-06, |
| "loss": 0.1377, |
| "step": 1342 |
| }, |
| { |
| "epoch": 0.9654924514737598, |
| "grad_norm": 1.4600318670272827, |
| "learning_rate": 3.5226455787203455e-06, |
| "loss": 0.0961, |
| "step": 1343 |
| }, |
| { |
| "epoch": 0.9662113587347232, |
| "grad_norm": 1.3111642599105835, |
| "learning_rate": 3.4507548526240115e-06, |
| "loss": 0.0925, |
| "step": 1344 |
| }, |
| { |
| "epoch": 0.9669302659956865, |
| "grad_norm": 0.9490992426872253, |
| "learning_rate": 3.3788641265276783e-06, |
| "loss": 0.1665, |
| "step": 1345 |
| }, |
| { |
| "epoch": 0.9676491732566499, |
| "grad_norm": 0.5945020914077759, |
| "learning_rate": 3.3069734004313442e-06, |
| "loss": 0.0843, |
| "step": 1346 |
| }, |
| { |
| "epoch": 0.9683680805176133, |
| "grad_norm": 0.7232512831687927, |
| "learning_rate": 3.235082674335011e-06, |
| "loss": 0.0955, |
| "step": 1347 |
| }, |
| { |
| "epoch": 0.9690869877785766, |
| "grad_norm": 0.6511152386665344, |
| "learning_rate": 3.1631919482386774e-06, |
| "loss": 0.1476, |
| "step": 1348 |
| }, |
| { |
| "epoch": 0.9698058950395398, |
| "grad_norm": 0.441866934299469, |
| "learning_rate": 3.091301222142344e-06, |
| "loss": 0.1086, |
| "step": 1349 |
| }, |
| { |
| "epoch": 0.9705248023005032, |
| "grad_norm": 0.701013445854187, |
| "learning_rate": 3.01941049604601e-06, |
| "loss": 0.1202, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.9712437095614666, |
| "grad_norm": 0.6116084456443787, |
| "learning_rate": 2.9475197699496766e-06, |
| "loss": 0.1058, |
| "step": 1351 |
| }, |
| { |
| "epoch": 0.9719626168224299, |
| "grad_norm": 0.8121954798698425, |
| "learning_rate": 2.875629043853343e-06, |
| "loss": 0.0945, |
| "step": 1352 |
| }, |
| { |
| "epoch": 0.9726815240833933, |
| "grad_norm": 0.8431273102760315, |
| "learning_rate": 2.8037383177570094e-06, |
| "loss": 0.129, |
| "step": 1353 |
| }, |
| { |
| "epoch": 0.9734004313443566, |
| "grad_norm": 0.6913970708847046, |
| "learning_rate": 2.7318475916606757e-06, |
| "loss": 0.1382, |
| "step": 1354 |
| }, |
| { |
| "epoch": 0.97411933860532, |
| "grad_norm": 1.012267827987671, |
| "learning_rate": 2.659956865564342e-06, |
| "loss": 0.1593, |
| "step": 1355 |
| }, |
| { |
| "epoch": 0.9748382458662832, |
| "grad_norm": 1.3991543054580688, |
| "learning_rate": 2.588066139468009e-06, |
| "loss": 0.2431, |
| "step": 1356 |
| }, |
| { |
| "epoch": 0.9755571531272466, |
| "grad_norm": 0.6293736696243286, |
| "learning_rate": 2.5161754133716753e-06, |
| "loss": 0.1001, |
| "step": 1357 |
| }, |
| { |
| "epoch": 0.9762760603882099, |
| "grad_norm": 0.8893879055976868, |
| "learning_rate": 2.4442846872753417e-06, |
| "loss": 0.1163, |
| "step": 1358 |
| }, |
| { |
| "epoch": 0.9769949676491733, |
| "grad_norm": 0.9843087196350098, |
| "learning_rate": 2.372393961179008e-06, |
| "loss": 0.2068, |
| "step": 1359 |
| }, |
| { |
| "epoch": 0.9777138749101366, |
| "grad_norm": 0.6816713809967041, |
| "learning_rate": 2.3005032350826745e-06, |
| "loss": 0.0862, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.9784327821711, |
| "grad_norm": 1.829386830329895, |
| "learning_rate": 2.228612508986341e-06, |
| "loss": 0.2509, |
| "step": 1361 |
| }, |
| { |
| "epoch": 0.9791516894320632, |
| "grad_norm": 0.9720048308372498, |
| "learning_rate": 2.1567217828900072e-06, |
| "loss": 0.0891, |
| "step": 1362 |
| }, |
| { |
| "epoch": 0.9798705966930266, |
| "grad_norm": 1.201979637145996, |
| "learning_rate": 2.084831056793674e-06, |
| "loss": 0.2173, |
| "step": 1363 |
| }, |
| { |
| "epoch": 0.9805895039539899, |
| "grad_norm": 0.5717848539352417, |
| "learning_rate": 2.0129403306973404e-06, |
| "loss": 0.1372, |
| "step": 1364 |
| }, |
| { |
| "epoch": 0.9813084112149533, |
| "grad_norm": 0.6442174911499023, |
| "learning_rate": 1.9410496046010064e-06, |
| "loss": 0.1769, |
| "step": 1365 |
| }, |
| { |
| "epoch": 0.9820273184759166, |
| "grad_norm": 0.5268913507461548, |
| "learning_rate": 1.8691588785046728e-06, |
| "loss": 0.1043, |
| "step": 1366 |
| }, |
| { |
| "epoch": 0.98274622573688, |
| "grad_norm": 0.441861093044281, |
| "learning_rate": 1.7972681524083394e-06, |
| "loss": 0.0953, |
| "step": 1367 |
| }, |
| { |
| "epoch": 0.9834651329978433, |
| "grad_norm": 0.9365966320037842, |
| "learning_rate": 1.7253774263120057e-06, |
| "loss": 0.1051, |
| "step": 1368 |
| }, |
| { |
| "epoch": 0.9841840402588066, |
| "grad_norm": 0.5820815563201904, |
| "learning_rate": 1.6534867002156721e-06, |
| "loss": 0.1582, |
| "step": 1369 |
| }, |
| { |
| "epoch": 0.9849029475197699, |
| "grad_norm": 0.5702938437461853, |
| "learning_rate": 1.5815959741193387e-06, |
| "loss": 0.1033, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.9856218547807333, |
| "grad_norm": 0.8433749675750732, |
| "learning_rate": 1.509705248023005e-06, |
| "loss": 0.1821, |
| "step": 1371 |
| }, |
| { |
| "epoch": 0.9863407620416966, |
| "grad_norm": 0.6916269063949585, |
| "learning_rate": 1.4378145219266715e-06, |
| "loss": 0.1452, |
| "step": 1372 |
| }, |
| { |
| "epoch": 0.98705966930266, |
| "grad_norm": 0.5802507996559143, |
| "learning_rate": 1.3659237958303379e-06, |
| "loss": 0.1296, |
| "step": 1373 |
| }, |
| { |
| "epoch": 0.9877785765636233, |
| "grad_norm": 0.48257482051849365, |
| "learning_rate": 1.2940330697340045e-06, |
| "loss": 0.0982, |
| "step": 1374 |
| }, |
| { |
| "epoch": 0.9884974838245866, |
| "grad_norm": 1.1281664371490479, |
| "learning_rate": 1.2221423436376708e-06, |
| "loss": 0.2641, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.9892163910855499, |
| "grad_norm": 1.0424281358718872, |
| "learning_rate": 1.1502516175413372e-06, |
| "loss": 0.1087, |
| "step": 1376 |
| }, |
| { |
| "epoch": 0.9899352983465133, |
| "grad_norm": 2.2509899139404297, |
| "learning_rate": 1.0783608914450036e-06, |
| "loss": 0.2221, |
| "step": 1377 |
| }, |
| { |
| "epoch": 0.9906542056074766, |
| "grad_norm": 0.4823886454105377, |
| "learning_rate": 1.0064701653486702e-06, |
| "loss": 0.1485, |
| "step": 1378 |
| }, |
| { |
| "epoch": 0.99137311286844, |
| "grad_norm": 1.1936447620391846, |
| "learning_rate": 9.345794392523364e-07, |
| "loss": 0.1093, |
| "step": 1379 |
| }, |
| { |
| "epoch": 0.9920920201294033, |
| "grad_norm": 0.7681226134300232, |
| "learning_rate": 8.626887131560029e-07, |
| "loss": 0.1347, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.9928109273903667, |
| "grad_norm": 1.534011960029602, |
| "learning_rate": 7.907979870596694e-07, |
| "loss": 0.1779, |
| "step": 1381 |
| }, |
| { |
| "epoch": 0.9935298346513299, |
| "grad_norm": 0.5000225305557251, |
| "learning_rate": 7.189072609633357e-07, |
| "loss": 0.0757, |
| "step": 1382 |
| }, |
| { |
| "epoch": 0.9942487419122933, |
| "grad_norm": 0.7270702719688416, |
| "learning_rate": 6.470165348670022e-07, |
| "loss": 0.1658, |
| "step": 1383 |
| }, |
| { |
| "epoch": 0.9949676491732566, |
| "grad_norm": 0.4887916147708893, |
| "learning_rate": 5.751258087706686e-07, |
| "loss": 0.1055, |
| "step": 1384 |
| }, |
| { |
| "epoch": 0.99568655643422, |
| "grad_norm": 0.5311588048934937, |
| "learning_rate": 5.032350826743351e-07, |
| "loss": 0.1321, |
| "step": 1385 |
| }, |
| { |
| "epoch": 0.9964054636951833, |
| "grad_norm": 0.655939519405365, |
| "learning_rate": 4.3134435657800143e-07, |
| "loss": 0.0895, |
| "step": 1386 |
| }, |
| { |
| "epoch": 0.9971243709561467, |
| "grad_norm": 0.5082830786705017, |
| "learning_rate": 3.5945363048166787e-07, |
| "loss": 0.1336, |
| "step": 1387 |
| }, |
| { |
| "epoch": 0.9978432782171099, |
| "grad_norm": 0.4612111449241638, |
| "learning_rate": 2.875629043853343e-07, |
| "loss": 0.104, |
| "step": 1388 |
| }, |
| { |
| "epoch": 0.9985621854780733, |
| "grad_norm": 0.8493375778198242, |
| "learning_rate": 2.1567217828900072e-07, |
| "loss": 0.0975, |
| "step": 1389 |
| }, |
| { |
| "epoch": 0.9992810927390366, |
| "grad_norm": 1.3611900806427002, |
| "learning_rate": 1.4378145219266715e-07, |
| "loss": 0.203, |
| "step": 1390 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.9593672156333923, |
| "learning_rate": 7.189072609633358e-08, |
| "loss": 0.1676, |
| "step": 1391 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 1391, |
| "total_flos": 1.0669986532055274e+18, |
| "train_loss": 0.16190836855527055, |
| "train_runtime": 21087.0624, |
| "train_samples_per_second": 1.055, |
| "train_steps_per_second": 0.066 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.936189591884613, |
| "eval_loss": 0.12847697734832764, |
| "eval_percentage_1_labels": 0.6271650195121765, |
| "eval_percentage_pos_predictions": 0.6581586003303528, |
| "eval_precision": 0.9279778599739075, |
| "eval_recall": 0.9738371968269348, |
| "eval_runtime": 602.3174, |
| "eval_samples_per_second": 4.62, |
| "eval_steps_per_second": 0.578, |
| "step": 1391 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1391, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 100, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.0669986532055274e+18, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
