| { | |
| "best_metric": 3.187575340270996, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.006732193348592971, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.732193348592971e-05, | |
| "grad_norm": 1.1772795915603638, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.6814, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 6.732193348592971e-05, | |
| "eval_loss": 4.549241542816162, | |
| "eval_runtime": 1967.2208, | |
| "eval_samples_per_second": 12.717, | |
| "eval_steps_per_second": 1.59, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00013464386697185942, | |
| "grad_norm": 1.767252802848816, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 2.0678, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00020196580045778916, | |
| "grad_norm": 2.2500200271606445, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.5754, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00026928773394371884, | |
| "grad_norm": 2.0921666622161865, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.5746, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0003366096674296486, | |
| "grad_norm": 2.19606614112854, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.5383, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0004039316009155783, | |
| "grad_norm": 2.4099624156951904, | |
| "learning_rate": 3e-06, | |
| "loss": 2.7149, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.000471253534401508, | |
| "grad_norm": 2.390442132949829, | |
| "learning_rate": 3.5e-06, | |
| "loss": 2.6377, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0005385754678874377, | |
| "grad_norm": 2.94857120513916, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.7157, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0006058974013733675, | |
| "grad_norm": 2.9573776721954346, | |
| "learning_rate": 4.5e-06, | |
| "loss": 3.0235, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0006732193348592972, | |
| "grad_norm": 2.9073917865753174, | |
| "learning_rate": 5e-06, | |
| "loss": 3.0077, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0007405412683452268, | |
| "grad_norm": 2.82808256149292, | |
| "learning_rate": 4.99847706754774e-06, | |
| "loss": 2.8971, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0008078632018311566, | |
| "grad_norm": 3.302197217941284, | |
| "learning_rate": 4.993910125649561e-06, | |
| "loss": 3.0504, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0008751851353170863, | |
| "grad_norm": 3.039174795150757, | |
| "learning_rate": 4.986304738420684e-06, | |
| "loss": 3.0642, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.000942507068803016, | |
| "grad_norm": 3.3671231269836426, | |
| "learning_rate": 4.975670171853926e-06, | |
| "loss": 3.2195, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0010098290022889458, | |
| "grad_norm": 3.130558490753174, | |
| "learning_rate": 4.962019382530521e-06, | |
| "loss": 3.0515, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0010771509357748754, | |
| "grad_norm": 3.575861930847168, | |
| "learning_rate": 4.9453690018345144e-06, | |
| "loss": 3.1632, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0011444728692608052, | |
| "grad_norm": 3.495004892349243, | |
| "learning_rate": 4.925739315689991e-06, | |
| "loss": 3.1274, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.001211794802746735, | |
| "grad_norm": 3.904120445251465, | |
| "learning_rate": 4.903154239845798e-06, | |
| "loss": 3.3787, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0012791167362326645, | |
| "grad_norm": 4.107210636138916, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "loss": 3.2824, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0013464386697185943, | |
| "grad_norm": 3.9819908142089844, | |
| "learning_rate": 4.849231551964771e-06, | |
| "loss": 3.2564, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.001413760603204524, | |
| "grad_norm": 3.903653621673584, | |
| "learning_rate": 4.817959636416969e-06, | |
| "loss": 3.3043, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0014810825366904537, | |
| "grad_norm": 3.8290135860443115, | |
| "learning_rate": 4.783863644106502e-06, | |
| "loss": 3.4404, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0015484044701763835, | |
| "grad_norm": 4.112966537475586, | |
| "learning_rate": 4.746985115747918e-06, | |
| "loss": 3.3573, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0016157264036623133, | |
| "grad_norm": 4.385584831237793, | |
| "learning_rate": 4.707368982147318e-06, | |
| "loss": 3.4217, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0016830483371482428, | |
| "grad_norm": 4.105695724487305, | |
| "learning_rate": 4.665063509461098e-06, | |
| "loss": 3.7123, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0017503702706341726, | |
| "grad_norm": 4.301870346069336, | |
| "learning_rate": 4.620120240391065e-06, | |
| "loss": 3.52, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0018176922041201024, | |
| "grad_norm": 4.582427978515625, | |
| "learning_rate": 4.572593931387604e-06, | |
| "loss": 3.6086, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.001885014137606032, | |
| "grad_norm": 4.567037582397461, | |
| "learning_rate": 4.522542485937369e-06, | |
| "loss": 3.6933, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0019523360710919618, | |
| "grad_norm": 5.327396869659424, | |
| "learning_rate": 4.470026884016805e-06, | |
| "loss": 3.7651, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0020196580045778916, | |
| "grad_norm": 4.7465643882751465, | |
| "learning_rate": 4.415111107797445e-06, | |
| "loss": 3.9085, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002086979938063821, | |
| "grad_norm": 5.117805004119873, | |
| "learning_rate": 4.357862063693486e-06, | |
| "loss": 3.9753, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0021543018715497507, | |
| "grad_norm": 4.87061071395874, | |
| "learning_rate": 4.2983495008466285e-06, | |
| "loss": 3.8506, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0022216238050356807, | |
| "grad_norm": 5.614602088928223, | |
| "learning_rate": 4.236645926147493e-06, | |
| "loss": 4.0786, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0022889457385216103, | |
| "grad_norm": 5.901675701141357, | |
| "learning_rate": 4.172826515897146e-06, | |
| "loss": 4.2382, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.00235626767200754, | |
| "grad_norm": 5.645284175872803, | |
| "learning_rate": 4.106969024216348e-06, | |
| "loss": 4.1851, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.00242358960549347, | |
| "grad_norm": 6.641393184661865, | |
| "learning_rate": 4.039153688314146e-06, | |
| "loss": 4.067, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0024909115389793995, | |
| "grad_norm": 7.662880897521973, | |
| "learning_rate": 3.969463130731183e-06, | |
| "loss": 4.2374, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.002558233472465329, | |
| "grad_norm": 6.7443013191223145, | |
| "learning_rate": 3.897982258676867e-06, | |
| "loss": 4.727, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.002625555405951259, | |
| "grad_norm": 7.711762428283691, | |
| "learning_rate": 3.824798160583012e-06, | |
| "loss": 4.4554, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0026928773394371886, | |
| "grad_norm": 7.188761234283447, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 4.6155, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.002760199272923118, | |
| "grad_norm": 7.4346795082092285, | |
| "learning_rate": 3.6736789069647273e-06, | |
| "loss": 4.5334, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.002827521206409048, | |
| "grad_norm": 10.012019157409668, | |
| "learning_rate": 3.595927866972694e-06, | |
| "loss": 4.9844, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.002894843139894978, | |
| "grad_norm": 8.302594184875488, | |
| "learning_rate": 3.516841607689501e-06, | |
| "loss": 4.7557, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0029621650733809074, | |
| "grad_norm": 9.62204360961914, | |
| "learning_rate": 3.436516483539781e-06, | |
| "loss": 5.1486, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0030294870068668374, | |
| "grad_norm": 9.987607955932617, | |
| "learning_rate": 3.3550503583141726e-06, | |
| "loss": 5.2176, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.003096808940352767, | |
| "grad_norm": 10.805545806884766, | |
| "learning_rate": 3.272542485937369e-06, | |
| "loss": 5.6029, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0031641308738386965, | |
| "grad_norm": 12.249214172363281, | |
| "learning_rate": 3.189093389542498e-06, | |
| "loss": 6.0521, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0032314528073246265, | |
| "grad_norm": 15.126471519470215, | |
| "learning_rate": 3.1048047389991693e-06, | |
| "loss": 6.7284, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.003298774740810556, | |
| "grad_norm": 15.107416152954102, | |
| "learning_rate": 3.019779227044398e-06, | |
| "loss": 7.3697, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0033660966742964857, | |
| "grad_norm": 19.051593780517578, | |
| "learning_rate": 2.9341204441673267e-06, | |
| "loss": 9.7884, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0033660966742964857, | |
| "eval_loss": 3.452199935913086, | |
| "eval_runtime": 1979.4069, | |
| "eval_samples_per_second": 12.639, | |
| "eval_steps_per_second": 1.58, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0034334186077824157, | |
| "grad_norm": 1.2261101007461548, | |
| "learning_rate": 2.847932752400164e-06, | |
| "loss": 1.4156, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0035007405412683453, | |
| "grad_norm": 2.0353269577026367, | |
| "learning_rate": 2.761321158169134e-06, | |
| "loss": 2.1038, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.003568062474754275, | |
| "grad_norm": 2.1064200401306152, | |
| "learning_rate": 2.6743911843603134e-06, | |
| "loss": 2.0902, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.003635384408240205, | |
| "grad_norm": 2.1769261360168457, | |
| "learning_rate": 2.587248741756253e-06, | |
| "loss": 2.2087, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0037027063417261344, | |
| "grad_norm": 2.4015352725982666, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.2605, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.003770028275212064, | |
| "grad_norm": 2.4183402061462402, | |
| "learning_rate": 2.4127512582437486e-06, | |
| "loss": 2.3062, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.003837350208697994, | |
| "grad_norm": 2.6967923641204834, | |
| "learning_rate": 2.325608815639687e-06, | |
| "loss": 2.2717, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0039046721421839236, | |
| "grad_norm": 2.5622987747192383, | |
| "learning_rate": 2.238678841830867e-06, | |
| "loss": 2.3366, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.003971994075669854, | |
| "grad_norm": 2.803257703781128, | |
| "learning_rate": 2.1520672475998374e-06, | |
| "loss": 2.7368, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.004039316009155783, | |
| "grad_norm": 2.7448480129241943, | |
| "learning_rate": 2.0658795558326745e-06, | |
| "loss": 2.5, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.004106637942641713, | |
| "grad_norm": 2.9456703662872314, | |
| "learning_rate": 1.9802207729556023e-06, | |
| "loss": 2.5136, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.004173959876127642, | |
| "grad_norm": 2.9251086711883545, | |
| "learning_rate": 1.895195261000831e-06, | |
| "loss": 2.5065, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.004241281809613572, | |
| "grad_norm": 3.134636640548706, | |
| "learning_rate": 1.8109066104575023e-06, | |
| "loss": 2.2943, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0043086037430995015, | |
| "grad_norm": 2.8761112689971924, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "loss": 2.4636, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.004375925676585432, | |
| "grad_norm": 3.1227707862854004, | |
| "learning_rate": 1.6449496416858285e-06, | |
| "loss": 2.4784, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0044432476100713615, | |
| "grad_norm": 2.9461958408355713, | |
| "learning_rate": 1.56348351646022e-06, | |
| "loss": 2.5508, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.004510569543557291, | |
| "grad_norm": 3.04009747505188, | |
| "learning_rate": 1.4831583923105e-06, | |
| "loss": 2.4567, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.004577891477043221, | |
| "grad_norm": 3.5833160877227783, | |
| "learning_rate": 1.4040721330273063e-06, | |
| "loss": 2.79, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.00464521341052915, | |
| "grad_norm": 3.3486886024475098, | |
| "learning_rate": 1.3263210930352737e-06, | |
| "loss": 2.6876, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.00471253534401508, | |
| "grad_norm": 3.662419557571411, | |
| "learning_rate": 1.2500000000000007e-06, | |
| "loss": 2.6861, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.00477985727750101, | |
| "grad_norm": 3.3335697650909424, | |
| "learning_rate": 1.1752018394169882e-06, | |
| "loss": 2.564, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.00484717921098694, | |
| "grad_norm": 3.8791604042053223, | |
| "learning_rate": 1.1020177413231334e-06, | |
| "loss": 2.9028, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.004914501144472869, | |
| "grad_norm": 3.7736315727233887, | |
| "learning_rate": 1.0305368692688175e-06, | |
| "loss": 2.8806, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.004981823077958799, | |
| "grad_norm": 3.789177656173706, | |
| "learning_rate": 9.608463116858544e-07, | |
| "loss": 2.757, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0050491450114447285, | |
| "grad_norm": 3.818617582321167, | |
| "learning_rate": 8.930309757836517e-07, | |
| "loss": 2.4988, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.005116466944930658, | |
| "grad_norm": 3.7709665298461914, | |
| "learning_rate": 8.271734841028553e-07, | |
| "loss": 2.7545, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0051837888784165885, | |
| "grad_norm": 4.059813499450684, | |
| "learning_rate": 7.633540738525066e-07, | |
| "loss": 2.982, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.005251110811902518, | |
| "grad_norm": 3.7630608081817627, | |
| "learning_rate": 7.016504991533727e-07, | |
| "loss": 2.6344, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.005318432745388448, | |
| "grad_norm": 4.433750629425049, | |
| "learning_rate": 6.421379363065142e-07, | |
| "loss": 2.9538, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.005385754678874377, | |
| "grad_norm": 4.345767021179199, | |
| "learning_rate": 5.848888922025553e-07, | |
| "loss": 2.842, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.005453076612360307, | |
| "grad_norm": 4.756106376647949, | |
| "learning_rate": 5.299731159831953e-07, | |
| "loss": 3.3317, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.005520398545846236, | |
| "grad_norm": 5.476846694946289, | |
| "learning_rate": 4.774575140626317e-07, | |
| "loss": 3.3194, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.005587720479332166, | |
| "grad_norm": 5.1269683837890625, | |
| "learning_rate": 4.27406068612396e-07, | |
| "loss": 3.1785, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.005655042412818096, | |
| "grad_norm": 5.075381755828857, | |
| "learning_rate": 3.798797596089351e-07, | |
| "loss": 3.1507, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.005722364346304026, | |
| "grad_norm": 5.653160095214844, | |
| "learning_rate": 3.3493649053890325e-07, | |
| "loss": 3.3281, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.005789686279789956, | |
| "grad_norm": 5.514196395874023, | |
| "learning_rate": 2.9263101785268253e-07, | |
| "loss": 3.5453, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.005857008213275885, | |
| "grad_norm": 5.700414180755615, | |
| "learning_rate": 2.53014884252083e-07, | |
| "loss": 3.5347, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.005924330146761815, | |
| "grad_norm": 6.0169243812561035, | |
| "learning_rate": 2.1613635589349756e-07, | |
| "loss": 3.6269, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.005991652080247744, | |
| "grad_norm": 5.780705451965332, | |
| "learning_rate": 1.8204036358303173e-07, | |
| "loss": 3.4327, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.006058974013733675, | |
| "grad_norm": 7.222046852111816, | |
| "learning_rate": 1.507684480352292e-07, | |
| "loss": 3.8177, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.006126295947219604, | |
| "grad_norm": 7.44561767578125, | |
| "learning_rate": 1.223587092621162e-07, | |
| "loss": 4.2328, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.006193617880705534, | |
| "grad_norm": 7.6607747077941895, | |
| "learning_rate": 9.684576015420277e-08, | |
| "loss": 3.9552, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0062609398141914635, | |
| "grad_norm": 8.162426948547363, | |
| "learning_rate": 7.426068431000883e-08, | |
| "loss": 4.1219, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.006328261747677393, | |
| "grad_norm": 8.686196327209473, | |
| "learning_rate": 5.463099816548578e-08, | |
| "loss": 4.1694, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.006395583681163323, | |
| "grad_norm": 11.760191917419434, | |
| "learning_rate": 3.798061746947995e-08, | |
| "loss": 4.5217, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.006462905614649253, | |
| "grad_norm": 10.849732398986816, | |
| "learning_rate": 2.4329828146074096e-08, | |
| "loss": 4.6942, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.006530227548135183, | |
| "grad_norm": 10.977898597717285, | |
| "learning_rate": 1.3695261579316776e-08, | |
| "loss": 4.9961, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.006597549481621112, | |
| "grad_norm": 13.189175605773926, | |
| "learning_rate": 6.089874350439507e-09, | |
| "loss": 5.593, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.006664871415107042, | |
| "grad_norm": 16.70779800415039, | |
| "learning_rate": 1.5229324522605949e-09, | |
| "loss": 6.2964, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.006732193348592971, | |
| "grad_norm": 19.198680877685547, | |
| "learning_rate": 0.0, | |
| "loss": 8.8601, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006732193348592971, | |
| "eval_loss": 3.187575340270996, | |
| "eval_runtime": 1981.7, | |
| "eval_samples_per_second": 12.625, | |
| "eval_steps_per_second": 1.578, | |
| "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": 1.5160561466395853e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |