| { | |
| "best_metric": 1.542808175086975, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.0052082316100597, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 5.208231610059699e-05, | |
| "grad_norm": 0.7850510478019714, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.4988, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 5.208231610059699e-05, | |
| "eval_loss": 2.3925089836120605, | |
| "eval_runtime": 1165.5151, | |
| "eval_samples_per_second": 27.746, | |
| "eval_steps_per_second": 6.937, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00010416463220119398, | |
| "grad_norm": 1.204688310623169, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.641, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00015624694830179098, | |
| "grad_norm": 1.1601476669311523, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.6859, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00020832926440238797, | |
| "grad_norm": 1.301094651222229, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.6955, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.000260411580502985, | |
| "grad_norm": 1.086885929107666, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.81, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00031249389660358197, | |
| "grad_norm": 1.0889768600463867, | |
| "learning_rate": 3e-06, | |
| "loss": 0.7875, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00036457621270417895, | |
| "grad_norm": 1.2988945245742798, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.8043, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00041665852880477594, | |
| "grad_norm": 1.2984793186187744, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.8566, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0004687408449053729, | |
| "grad_norm": 1.8114720582962036, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.8709, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00052082316100597, | |
| "grad_norm": 1.5189226865768433, | |
| "learning_rate": 5e-06, | |
| "loss": 0.9123, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0005729054771065669, | |
| "grad_norm": 1.3734829425811768, | |
| "learning_rate": 4.99847706754774e-06, | |
| "loss": 0.8988, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0006249877932071639, | |
| "grad_norm": 1.8695842027664185, | |
| "learning_rate": 4.993910125649561e-06, | |
| "loss": 0.9392, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.000677070109307761, | |
| "grad_norm": 1.7758443355560303, | |
| "learning_rate": 4.986304738420684e-06, | |
| "loss": 0.948, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0007291524254083579, | |
| "grad_norm": 1.7251638174057007, | |
| "learning_rate": 4.975670171853926e-06, | |
| "loss": 1.0075, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0007812347415089549, | |
| "grad_norm": 1.9016642570495605, | |
| "learning_rate": 4.962019382530521e-06, | |
| "loss": 1.0206, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0008333170576095519, | |
| "grad_norm": 2.261992931365967, | |
| "learning_rate": 4.9453690018345144e-06, | |
| "loss": 1.0041, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0008853993737101489, | |
| "grad_norm": 1.9689737558364868, | |
| "learning_rate": 4.925739315689991e-06, | |
| "loss": 0.9991, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0009374816898107458, | |
| "grad_norm": 2.2839338779449463, | |
| "learning_rate": 4.903154239845798e-06, | |
| "loss": 1.0233, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0009895640059113429, | |
| "grad_norm": 2.3685903549194336, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "loss": 1.0451, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00104164632201194, | |
| "grad_norm": 2.4562370777130127, | |
| "learning_rate": 4.849231551964771e-06, | |
| "loss": 1.1865, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.001093728638112537, | |
| "grad_norm": 2.2600936889648438, | |
| "learning_rate": 4.817959636416969e-06, | |
| "loss": 1.2595, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0011458109542131338, | |
| "grad_norm": 2.520331859588623, | |
| "learning_rate": 4.783863644106502e-06, | |
| "loss": 1.1546, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0011978932703137308, | |
| "grad_norm": 2.7857682704925537, | |
| "learning_rate": 4.746985115747918e-06, | |
| "loss": 1.1655, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0012499755864143279, | |
| "grad_norm": 2.817095994949341, | |
| "learning_rate": 4.707368982147318e-06, | |
| "loss": 1.3266, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.001302057902514925, | |
| "grad_norm": 2.6866202354431152, | |
| "learning_rate": 4.665063509461098e-06, | |
| "loss": 1.3189, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.001354140218615522, | |
| "grad_norm": 3.191648244857788, | |
| "learning_rate": 4.620120240391065e-06, | |
| "loss": 1.3855, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0014062225347161188, | |
| "grad_norm": 3.362218141555786, | |
| "learning_rate": 4.572593931387604e-06, | |
| "loss": 1.4789, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0014583048508167158, | |
| "grad_norm": 3.0076942443847656, | |
| "learning_rate": 4.522542485937369e-06, | |
| "loss": 1.4305, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0015103871669173128, | |
| "grad_norm": 3.998861312866211, | |
| "learning_rate": 4.470026884016805e-06, | |
| "loss": 1.5949, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0015624694830179099, | |
| "grad_norm": 4.050288200378418, | |
| "learning_rate": 4.415111107797445e-06, | |
| "loss": 1.6032, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0016145517991185067, | |
| "grad_norm": 5.438502788543701, | |
| "learning_rate": 4.357862063693486e-06, | |
| "loss": 1.8929, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0016666341152191037, | |
| "grad_norm": 4.320279121398926, | |
| "learning_rate": 4.2983495008466285e-06, | |
| "loss": 1.7818, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0017187164313197008, | |
| "grad_norm": 5.677204132080078, | |
| "learning_rate": 4.236645926147493e-06, | |
| "loss": 1.9708, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0017707987474202978, | |
| "grad_norm": 4.794656276702881, | |
| "learning_rate": 4.172826515897146e-06, | |
| "loss": 1.9317, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0018228810635208949, | |
| "grad_norm": 4.484780311584473, | |
| "learning_rate": 4.106969024216348e-06, | |
| "loss": 1.8344, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0018749633796214917, | |
| "grad_norm": 5.888218879699707, | |
| "learning_rate": 4.039153688314146e-06, | |
| "loss": 2.0714, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0019270456957220887, | |
| "grad_norm": 6.086559772491455, | |
| "learning_rate": 3.969463130731183e-06, | |
| "loss": 2.2107, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0019791280118226858, | |
| "grad_norm": 5.582540512084961, | |
| "learning_rate": 3.897982258676867e-06, | |
| "loss": 2.2159, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0020312103279232826, | |
| "grad_norm": 6.883957862854004, | |
| "learning_rate": 3.824798160583012e-06, | |
| "loss": 2.5386, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00208329264402388, | |
| "grad_norm": 6.9683098793029785, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 2.7286, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0021353749601244767, | |
| "grad_norm": 7.439619541168213, | |
| "learning_rate": 3.6736789069647273e-06, | |
| "loss": 2.9706, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.002187457276225074, | |
| "grad_norm": 8.695789337158203, | |
| "learning_rate": 3.595927866972694e-06, | |
| "loss": 3.1264, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0022395395923256707, | |
| "grad_norm": 8.851489067077637, | |
| "learning_rate": 3.516841607689501e-06, | |
| "loss": 3.2198, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0022916219084262676, | |
| "grad_norm": 11.42151165008545, | |
| "learning_rate": 3.436516483539781e-06, | |
| "loss": 3.5362, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.002343704224526865, | |
| "grad_norm": 9.101286888122559, | |
| "learning_rate": 3.3550503583141726e-06, | |
| "loss": 3.4341, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0023957865406274616, | |
| "grad_norm": 9.989834785461426, | |
| "learning_rate": 3.272542485937369e-06, | |
| "loss": 3.8915, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.002447868856728059, | |
| "grad_norm": 12.451299667358398, | |
| "learning_rate": 3.189093389542498e-06, | |
| "loss": 4.4809, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0024999511728286557, | |
| "grad_norm": 15.89113998413086, | |
| "learning_rate": 3.1048047389991693e-06, | |
| "loss": 5.9077, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0025520334889292525, | |
| "grad_norm": 22.119922637939453, | |
| "learning_rate": 3.019779227044398e-06, | |
| "loss": 7.5703, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.00260411580502985, | |
| "grad_norm": 23.528160095214844, | |
| "learning_rate": 2.9341204441673267e-06, | |
| "loss": 8.9154, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.00260411580502985, | |
| "eval_loss": 1.7300430536270142, | |
| "eval_runtime": 1166.7865, | |
| "eval_samples_per_second": 27.715, | |
| "eval_steps_per_second": 6.929, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0026561981211304466, | |
| "grad_norm": 1.0742809772491455, | |
| "learning_rate": 2.847932752400164e-06, | |
| "loss": 0.3675, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.002708280437231044, | |
| "grad_norm": 1.0545445680618286, | |
| "learning_rate": 2.761321158169134e-06, | |
| "loss": 0.401, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0027603627533316407, | |
| "grad_norm": 1.108581304550171, | |
| "learning_rate": 2.6743911843603134e-06, | |
| "loss": 0.4626, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0028124450694322375, | |
| "grad_norm": 1.2265070676803589, | |
| "learning_rate": 2.587248741756253e-06, | |
| "loss": 0.5719, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0028645273855328348, | |
| "grad_norm": 1.1157441139221191, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.5564, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0029166097016334316, | |
| "grad_norm": 1.1121352910995483, | |
| "learning_rate": 2.4127512582437486e-06, | |
| "loss": 0.5619, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0029686920177340284, | |
| "grad_norm": 1.0687549114227295, | |
| "learning_rate": 2.325608815639687e-06, | |
| "loss": 0.591, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0030207743338346257, | |
| "grad_norm": 1.1351648569107056, | |
| "learning_rate": 2.238678841830867e-06, | |
| "loss": 0.6076, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0030728566499352225, | |
| "grad_norm": 1.1356217861175537, | |
| "learning_rate": 2.1520672475998374e-06, | |
| "loss": 0.6677, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0031249389660358198, | |
| "grad_norm": 1.1470251083374023, | |
| "learning_rate": 2.0658795558326745e-06, | |
| "loss": 0.664, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0031770212821364166, | |
| "grad_norm": 1.2547721862792969, | |
| "learning_rate": 1.9802207729556023e-06, | |
| "loss": 0.6278, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0032291035982370134, | |
| "grad_norm": 1.313974142074585, | |
| "learning_rate": 1.895195261000831e-06, | |
| "loss": 0.6576, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0032811859143376107, | |
| "grad_norm": 1.2617135047912598, | |
| "learning_rate": 1.8109066104575023e-06, | |
| "loss": 0.6497, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0033332682304382075, | |
| "grad_norm": 1.3057128190994263, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "loss": 0.6896, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0033853505465388047, | |
| "grad_norm": 1.3191314935684204, | |
| "learning_rate": 1.6449496416858285e-06, | |
| "loss": 0.679, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0034374328626394016, | |
| "grad_norm": 1.597451090812683, | |
| "learning_rate": 1.56348351646022e-06, | |
| "loss": 0.7526, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0034895151787399984, | |
| "grad_norm": 1.440009355545044, | |
| "learning_rate": 1.4831583923105e-06, | |
| "loss": 0.7096, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0035415974948405956, | |
| "grad_norm": 1.5627026557922363, | |
| "learning_rate": 1.4040721330273063e-06, | |
| "loss": 0.724, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0035936798109411925, | |
| "grad_norm": 1.503731608390808, | |
| "learning_rate": 1.3263210930352737e-06, | |
| "loss": 0.652, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0036457621270417897, | |
| "grad_norm": 1.6435801982879639, | |
| "learning_rate": 1.2500000000000007e-06, | |
| "loss": 0.8539, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0036978444431423865, | |
| "grad_norm": 1.7257816791534424, | |
| "learning_rate": 1.1752018394169882e-06, | |
| "loss": 0.761, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0037499267592429834, | |
| "grad_norm": 1.9037182331085205, | |
| "learning_rate": 1.1020177413231334e-06, | |
| "loss": 0.7582, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0038020090753435806, | |
| "grad_norm": 1.8669556379318237, | |
| "learning_rate": 1.0305368692688175e-06, | |
| "loss": 0.8375, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0038540913914441774, | |
| "grad_norm": 2.0030980110168457, | |
| "learning_rate": 9.608463116858544e-07, | |
| "loss": 0.9134, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0039061737075447747, | |
| "grad_norm": 1.8826111555099487, | |
| "learning_rate": 8.930309757836517e-07, | |
| "loss": 0.7892, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0039582560236453715, | |
| "grad_norm": 2.3302550315856934, | |
| "learning_rate": 8.271734841028553e-07, | |
| "loss": 0.864, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.004010338339745969, | |
| "grad_norm": 2.2271835803985596, | |
| "learning_rate": 7.633540738525066e-07, | |
| "loss": 0.9407, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.004062420655846565, | |
| "grad_norm": 2.2922518253326416, | |
| "learning_rate": 7.016504991533727e-07, | |
| "loss": 0.9515, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.004114502971947162, | |
| "grad_norm": 2.4218201637268066, | |
| "learning_rate": 6.421379363065142e-07, | |
| "loss": 0.9807, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.00416658528804776, | |
| "grad_norm": 2.886337995529175, | |
| "learning_rate": 5.848888922025553e-07, | |
| "loss": 1.0902, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.004218667604148356, | |
| "grad_norm": 2.7967350482940674, | |
| "learning_rate": 5.299731159831953e-07, | |
| "loss": 1.2078, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.004270749920248953, | |
| "grad_norm": 3.2428112030029297, | |
| "learning_rate": 4.774575140626317e-07, | |
| "loss": 1.1927, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.004322832236349551, | |
| "grad_norm": 3.46563458442688, | |
| "learning_rate": 4.27406068612396e-07, | |
| "loss": 1.3252, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.004374914552450148, | |
| "grad_norm": 3.441267490386963, | |
| "learning_rate": 3.798797596089351e-07, | |
| "loss": 1.3008, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.004426996868550744, | |
| "grad_norm": 4.186681270599365, | |
| "learning_rate": 3.3493649053890325e-07, | |
| "loss": 1.5252, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0044790791846513415, | |
| "grad_norm": 3.94539737701416, | |
| "learning_rate": 2.9263101785268253e-07, | |
| "loss": 1.4843, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.004531161500751939, | |
| "grad_norm": 4.773882865905762, | |
| "learning_rate": 2.53014884252083e-07, | |
| "loss": 1.7177, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.004583243816852535, | |
| "grad_norm": 5.422726631164551, | |
| "learning_rate": 2.1613635589349756e-07, | |
| "loss": 1.7822, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.004635326132953132, | |
| "grad_norm": 5.55891227722168, | |
| "learning_rate": 1.8204036358303173e-07, | |
| "loss": 1.9766, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.00468740844905373, | |
| "grad_norm": 5.845249652862549, | |
| "learning_rate": 1.507684480352292e-07, | |
| "loss": 1.9855, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.004739490765154326, | |
| "grad_norm": 5.987824440002441, | |
| "learning_rate": 1.223587092621162e-07, | |
| "loss": 2.0612, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.004791573081254923, | |
| "grad_norm": 6.048609256744385, | |
| "learning_rate": 9.684576015420277e-08, | |
| "loss": 2.1425, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0048436553973555205, | |
| "grad_norm": 7.93657112121582, | |
| "learning_rate": 7.426068431000883e-08, | |
| "loss": 2.7068, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.004895737713456118, | |
| "grad_norm": 8.366839408874512, | |
| "learning_rate": 5.463099816548578e-08, | |
| "loss": 2.9492, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.004947820029556714, | |
| "grad_norm": 9.729937553405762, | |
| "learning_rate": 3.798061746947995e-08, | |
| "loss": 3.1672, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0049999023456573114, | |
| "grad_norm": 10.509390830993652, | |
| "learning_rate": 2.4329828146074096e-08, | |
| "loss": 3.2586, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.005051984661757909, | |
| "grad_norm": 13.249284744262695, | |
| "learning_rate": 1.3695261579316776e-08, | |
| "loss": 3.8453, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.005104066977858505, | |
| "grad_norm": 18.896095275878906, | |
| "learning_rate": 6.089874350439507e-09, | |
| "loss": 5.4556, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.005156149293959102, | |
| "grad_norm": 24.21651268005371, | |
| "learning_rate": 1.5229324522605949e-09, | |
| "loss": 7.5273, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0052082316100597, | |
| "grad_norm": 27.522985458374023, | |
| "learning_rate": 0.0, | |
| "loss": 8.2395, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0052082316100597, | |
| "eval_loss": 1.542808175086975, | |
| "eval_runtime": 1166.5968, | |
| "eval_samples_per_second": 27.72, | |
| "eval_steps_per_second": 6.93, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.949858011689779e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |