Instructions to use akahana/tinybert-javanese with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use akahana/tinybert-javanese with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="akahana/tinybert-javanese")# Load model directly from transformers import AutoTokenizer, AutoModelForMaskedLM tokenizer = AutoTokenizer.from_pretrained("akahana/tinybert-javanese") model = AutoModelForMaskedLM.from_pretrained("akahana/tinybert-javanese", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 30.0, | |
| "eval_steps": 500, | |
| "global_step": 71610, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04189359028068706, | |
| "grad_norm": 1.1499052047729492, | |
| "learning_rate": 4.993017734953219e-05, | |
| "loss": 9.973, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08378718056137412, | |
| "grad_norm": 1.1211209297180176, | |
| "learning_rate": 4.986035469906438e-05, | |
| "loss": 9.315, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.12568077084206117, | |
| "grad_norm": 1.0281990766525269, | |
| "learning_rate": 4.979053204859657e-05, | |
| "loss": 8.8239, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16757436112274823, | |
| "grad_norm": 1.1534616947174072, | |
| "learning_rate": 4.972070939812876e-05, | |
| "loss": 8.4254, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.20946795140343527, | |
| "grad_norm": 0.9906216859817505, | |
| "learning_rate": 4.9650886747660944e-05, | |
| "loss": 8.1225, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25136154168412234, | |
| "grad_norm": 0.9559820294380188, | |
| "learning_rate": 4.958106409719313e-05, | |
| "loss": 7.8557, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2932551319648094, | |
| "grad_norm": 0.7887147665023804, | |
| "learning_rate": 4.951124144672532e-05, | |
| "loss": 7.7282, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.33514872224549647, | |
| "grad_norm": 0.6530473232269287, | |
| "learning_rate": 4.944141879625751e-05, | |
| "loss": 7.6586, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3770423125261835, | |
| "grad_norm": 0.6546779870986938, | |
| "learning_rate": 4.9371596145789694e-05, | |
| "loss": 7.6438, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.41893590280687054, | |
| "grad_norm": 0.5660322308540344, | |
| "learning_rate": 4.9301773495321885e-05, | |
| "loss": 7.6237, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4608294930875576, | |
| "grad_norm": 0.6094719171524048, | |
| "learning_rate": 4.9231950844854076e-05, | |
| "loss": 7.5937, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5027230833682447, | |
| "grad_norm": 0.6101346015930176, | |
| "learning_rate": 4.916212819438626e-05, | |
| "loss": 7.5661, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.5446166736489317, | |
| "grad_norm": 0.5943477749824524, | |
| "learning_rate": 4.909230554391845e-05, | |
| "loss": 7.6007, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5865102639296188, | |
| "grad_norm": 0.6604776382446289, | |
| "learning_rate": 4.902248289345064e-05, | |
| "loss": 7.568, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.6284038542103059, | |
| "grad_norm": 0.6151777505874634, | |
| "learning_rate": 4.8952660242982826e-05, | |
| "loss": 7.5521, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6702974444909929, | |
| "grad_norm": 0.6381381750106812, | |
| "learning_rate": 4.888283759251501e-05, | |
| "loss": 7.5478, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.7121910347716799, | |
| "grad_norm": 0.6552081108093262, | |
| "learning_rate": 4.88130149420472e-05, | |
| "loss": 7.5293, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.754084625052367, | |
| "grad_norm": 0.6973659992218018, | |
| "learning_rate": 4.874319229157939e-05, | |
| "loss": 7.4983, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.795978215333054, | |
| "grad_norm": 0.8130584955215454, | |
| "learning_rate": 4.8673369641111576e-05, | |
| "loss": 7.5213, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.8378718056137411, | |
| "grad_norm": 0.8530446887016296, | |
| "learning_rate": 4.860354699064377e-05, | |
| "loss": 7.5105, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8797653958944281, | |
| "grad_norm": 0.8059477210044861, | |
| "learning_rate": 4.853372434017596e-05, | |
| "loss": 7.508, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.9216589861751152, | |
| "grad_norm": 0.7378331422805786, | |
| "learning_rate": 4.846390168970814e-05, | |
| "loss": 7.482, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9635525764558023, | |
| "grad_norm": 1.1823194026947021, | |
| "learning_rate": 4.839407903924033e-05, | |
| "loss": 7.4778, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.0054461667364893, | |
| "grad_norm": 0.8547298908233643, | |
| "learning_rate": 4.832425638877252e-05, | |
| "loss": 7.4529, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.0473397570171763, | |
| "grad_norm": 0.8535734415054321, | |
| "learning_rate": 4.825443373830471e-05, | |
| "loss": 7.4374, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0892333472978635, | |
| "grad_norm": 0.994597852230072, | |
| "learning_rate": 4.818461108783689e-05, | |
| "loss": 7.4306, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.1311269375785504, | |
| "grad_norm": 1.2056490182876587, | |
| "learning_rate": 4.8114788437369084e-05, | |
| "loss": 7.4322, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.1730205278592376, | |
| "grad_norm": 1.2451157569885254, | |
| "learning_rate": 4.8044965786901275e-05, | |
| "loss": 7.4205, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.2149141181399246, | |
| "grad_norm": 0.9964780211448669, | |
| "learning_rate": 4.797514313643346e-05, | |
| "loss": 7.4031, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.2568077084206117, | |
| "grad_norm": 0.8989804983139038, | |
| "learning_rate": 4.790532048596565e-05, | |
| "loss": 7.4043, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2987012987012987, | |
| "grad_norm": 1.1330469846725464, | |
| "learning_rate": 4.783549783549784e-05, | |
| "loss": 7.4031, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.3405948889819856, | |
| "grad_norm": 0.9531299471855164, | |
| "learning_rate": 4.7765675185030025e-05, | |
| "loss": 7.3866, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.3824884792626728, | |
| "grad_norm": 1.0342323780059814, | |
| "learning_rate": 4.7695852534562216e-05, | |
| "loss": 7.3477, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.42438206954336, | |
| "grad_norm": 1.0523111820220947, | |
| "learning_rate": 4.76260298840944e-05, | |
| "loss": 7.3726, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.466275659824047, | |
| "grad_norm": 1.298751711845398, | |
| "learning_rate": 4.755620723362659e-05, | |
| "loss": 7.3484, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.508169250104734, | |
| "grad_norm": 1.0065233707427979, | |
| "learning_rate": 4.7486384583158775e-05, | |
| "loss": 7.3567, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.550062840385421, | |
| "grad_norm": 1.2989579439163208, | |
| "learning_rate": 4.7416561932690966e-05, | |
| "loss": 7.3275, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.591956430666108, | |
| "grad_norm": 1.0343406200408936, | |
| "learning_rate": 4.734673928222316e-05, | |
| "loss": 7.3353, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.6338500209467952, | |
| "grad_norm": 0.9944115281105042, | |
| "learning_rate": 4.727691663175534e-05, | |
| "loss": 7.3304, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.6757436112274822, | |
| "grad_norm": 1.102974534034729, | |
| "learning_rate": 4.720709398128753e-05, | |
| "loss": 7.3128, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.7176372015081691, | |
| "grad_norm": 1.112282156944275, | |
| "learning_rate": 4.713727133081972e-05, | |
| "loss": 7.333, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.7595307917888563, | |
| "grad_norm": 1.2143328189849854, | |
| "learning_rate": 4.7067448680351914e-05, | |
| "loss": 7.3342, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.8014243820695435, | |
| "grad_norm": 1.1656922101974487, | |
| "learning_rate": 4.69976260298841e-05, | |
| "loss": 7.2995, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.8433179723502304, | |
| "grad_norm": 1.2085694074630737, | |
| "learning_rate": 4.692780337941628e-05, | |
| "loss": 7.2939, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.8852115626309174, | |
| "grad_norm": 1.1366217136383057, | |
| "learning_rate": 4.685798072894847e-05, | |
| "loss": 7.3079, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.9271051529116046, | |
| "grad_norm": 1.5368098020553589, | |
| "learning_rate": 4.678815807848066e-05, | |
| "loss": 7.2682, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.9689987431922917, | |
| "grad_norm": 1.1548763513565063, | |
| "learning_rate": 4.671833542801285e-05, | |
| "loss": 7.2708, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.0108923334729787, | |
| "grad_norm": 1.1510928869247437, | |
| "learning_rate": 4.664851277754504e-05, | |
| "loss": 7.2794, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.0527859237536656, | |
| "grad_norm": 1.103461503982544, | |
| "learning_rate": 4.6578690127077224e-05, | |
| "loss": 7.2612, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.0946795140343526, | |
| "grad_norm": 1.2767215967178345, | |
| "learning_rate": 4.6508867476609414e-05, | |
| "loss": 7.2436, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.13657310431504, | |
| "grad_norm": 1.3062710762023926, | |
| "learning_rate": 4.6439044826141605e-05, | |
| "loss": 7.2162, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.178466694595727, | |
| "grad_norm": 1.2461299896240234, | |
| "learning_rate": 4.6369222175673796e-05, | |
| "loss": 7.2387, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.220360284876414, | |
| "grad_norm": 1.5427358150482178, | |
| "learning_rate": 4.629939952520598e-05, | |
| "loss": 7.2004, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.262253875157101, | |
| "grad_norm": 1.390331506729126, | |
| "learning_rate": 4.6229576874738165e-05, | |
| "loss": 7.2403, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.3041474654377883, | |
| "grad_norm": 1.4087032079696655, | |
| "learning_rate": 4.6159754224270356e-05, | |
| "loss": 7.2157, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.346041055718475, | |
| "grad_norm": 1.2417359352111816, | |
| "learning_rate": 4.608993157380254e-05, | |
| "loss": 7.2277, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.387934645999162, | |
| "grad_norm": 1.4267281293869019, | |
| "learning_rate": 4.602010892333473e-05, | |
| "loss": 7.1999, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.429828236279849, | |
| "grad_norm": 1.3897684812545776, | |
| "learning_rate": 4.595028627286692e-05, | |
| "loss": 7.2155, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.471721826560536, | |
| "grad_norm": 1.326821208000183, | |
| "learning_rate": 4.5880463622399106e-05, | |
| "loss": 7.1705, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.5136154168412235, | |
| "grad_norm": 1.2585749626159668, | |
| "learning_rate": 4.58106409719313e-05, | |
| "loss": 7.1787, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.5555090071219104, | |
| "grad_norm": 1.4856244325637817, | |
| "learning_rate": 4.574081832146349e-05, | |
| "loss": 7.1923, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.5974025974025974, | |
| "grad_norm": 1.2883421182632446, | |
| "learning_rate": 4.567099567099568e-05, | |
| "loss": 7.1776, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.6392961876832843, | |
| "grad_norm": 1.4935518503189087, | |
| "learning_rate": 4.560117302052786e-05, | |
| "loss": 7.1711, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.6811897779639713, | |
| "grad_norm": 1.3920152187347412, | |
| "learning_rate": 4.553135037006005e-05, | |
| "loss": 7.1292, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.7230833682446587, | |
| "grad_norm": 1.2802495956420898, | |
| "learning_rate": 4.546152771959224e-05, | |
| "loss": 7.1558, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.7649769585253456, | |
| "grad_norm": 1.4111789464950562, | |
| "learning_rate": 4.5392403295629106e-05, | |
| "loss": 7.172, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.8068705488060326, | |
| "grad_norm": 1.6390964984893799, | |
| "learning_rate": 4.53225806451613e-05, | |
| "loss": 7.1263, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.84876413908672, | |
| "grad_norm": 1.4132812023162842, | |
| "learning_rate": 4.525275799469348e-05, | |
| "loss": 7.1259, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.890657729367407, | |
| "grad_norm": 1.4943978786468506, | |
| "learning_rate": 4.518293534422567e-05, | |
| "loss": 7.1252, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.932551319648094, | |
| "grad_norm": 1.3022414445877075, | |
| "learning_rate": 4.5113112693757856e-05, | |
| "loss": 7.107, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.974444909928781, | |
| "grad_norm": 1.4270446300506592, | |
| "learning_rate": 4.504329004329004e-05, | |
| "loss": 7.1278, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.016338500209468, | |
| "grad_norm": 1.3672137260437012, | |
| "learning_rate": 4.497346739282223e-05, | |
| "loss": 7.1282, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.058232090490155, | |
| "grad_norm": 1.955368995666504, | |
| "learning_rate": 4.490364474235442e-05, | |
| "loss": 7.1123, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.100125680770842, | |
| "grad_norm": 1.3990498781204224, | |
| "learning_rate": 4.483382209188661e-05, | |
| "loss": 7.1117, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.142019271051529, | |
| "grad_norm": 1.6294671297073364, | |
| "learning_rate": 4.47639994414188e-05, | |
| "loss": 7.0721, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.183912861332216, | |
| "grad_norm": 1.3939063549041748, | |
| "learning_rate": 4.469417679095099e-05, | |
| "loss": 7.0599, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.225806451612903, | |
| "grad_norm": 1.918155312538147, | |
| "learning_rate": 4.462435414048318e-05, | |
| "loss": 7.0759, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.2677000418935904, | |
| "grad_norm": 1.3072093725204468, | |
| "learning_rate": 4.455453149001536e-05, | |
| "loss": 7.0784, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.3095936321742774, | |
| "grad_norm": 1.3794573545455933, | |
| "learning_rate": 4.4484708839547554e-05, | |
| "loss": 7.0726, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.3514872224549643, | |
| "grad_norm": 1.4223533868789673, | |
| "learning_rate": 4.441488618907974e-05, | |
| "loss": 7.0513, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.3933808127356513, | |
| "grad_norm": 1.695520043373108, | |
| "learning_rate": 4.434506353861192e-05, | |
| "loss": 7.0586, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 3.4352744030163387, | |
| "grad_norm": 2.290275812149048, | |
| "learning_rate": 4.4275240888144113e-05, | |
| "loss": 7.0429, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 3.4771679932970256, | |
| "grad_norm": 1.5322943925857544, | |
| "learning_rate": 4.4205418237676304e-05, | |
| "loss": 7.0196, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 3.5190615835777126, | |
| "grad_norm": 1.4767639636993408, | |
| "learning_rate": 4.4135595587208495e-05, | |
| "loss": 7.0643, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 3.5609551738583995, | |
| "grad_norm": 1.4343881607055664, | |
| "learning_rate": 4.406577293674068e-05, | |
| "loss": 7.0153, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.602848764139087, | |
| "grad_norm": 1.7641897201538086, | |
| "learning_rate": 4.399664851277755e-05, | |
| "loss": 7.013, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 3.644742354419774, | |
| "grad_norm": 1.9688279628753662, | |
| "learning_rate": 4.392682586230973e-05, | |
| "loss": 6.9985, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.686635944700461, | |
| "grad_norm": 1.7434871196746826, | |
| "learning_rate": 4.385700321184192e-05, | |
| "loss": 7.0142, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 3.728529534981148, | |
| "grad_norm": 1.550470232963562, | |
| "learning_rate": 4.378718056137411e-05, | |
| "loss": 6.9975, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 3.7704231252618348, | |
| "grad_norm": 1.759869933128357, | |
| "learning_rate": 4.37173579109063e-05, | |
| "loss": 6.9821, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.812316715542522, | |
| "grad_norm": 2.052905797958374, | |
| "learning_rate": 4.364753526043849e-05, | |
| "loss": 6.9917, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 3.854210305823209, | |
| "grad_norm": 1.8859872817993164, | |
| "learning_rate": 4.357771260997068e-05, | |
| "loss": 6.9934, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.896103896103896, | |
| "grad_norm": 1.8349354267120361, | |
| "learning_rate": 4.3507889959502863e-05, | |
| "loss": 6.9861, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 3.937997486384583, | |
| "grad_norm": 2.519893169403076, | |
| "learning_rate": 4.3438067309035054e-05, | |
| "loss": 6.9611, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.97989107666527, | |
| "grad_norm": 1.506072759628296, | |
| "learning_rate": 4.336824465856724e-05, | |
| "loss": 6.9688, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.021784666945957, | |
| "grad_norm": 1.5342004299163818, | |
| "learning_rate": 4.329842200809943e-05, | |
| "loss": 6.9582, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.063678257226645, | |
| "grad_norm": 1.6476861238479614, | |
| "learning_rate": 4.3228599357631614e-05, | |
| "loss": 6.9712, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.105571847507331, | |
| "grad_norm": 2.112595319747925, | |
| "learning_rate": 4.3158776707163805e-05, | |
| "loss": 6.9568, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.147465437788019, | |
| "grad_norm": 2.390101194381714, | |
| "learning_rate": 4.3088954056695996e-05, | |
| "loss": 6.9501, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.189359028068705, | |
| "grad_norm": 1.900177240371704, | |
| "learning_rate": 4.301913140622818e-05, | |
| "loss": 6.935, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.231252618349393, | |
| "grad_norm": 1.7032443284988403, | |
| "learning_rate": 4.294930875576037e-05, | |
| "loss": 6.9343, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 4.27314620863008, | |
| "grad_norm": 1.8393847942352295, | |
| "learning_rate": 4.287948610529256e-05, | |
| "loss": 6.8927, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 4.3150397989107665, | |
| "grad_norm": 2.046727180480957, | |
| "learning_rate": 4.280966345482475e-05, | |
| "loss": 6.9156, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 4.356933389191454, | |
| "grad_norm": 1.832216501235962, | |
| "learning_rate": 4.273984080435694e-05, | |
| "loss": 6.89, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 4.39882697947214, | |
| "grad_norm": 1.8682448863983154, | |
| "learning_rate": 4.267001815388912e-05, | |
| "loss": 6.8995, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 4.440720569752828, | |
| "grad_norm": 2.0732340812683105, | |
| "learning_rate": 4.260089372992599e-05, | |
| "loss": 6.9094, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 4.482614160033515, | |
| "grad_norm": 1.6016206741333008, | |
| "learning_rate": 4.253107107945818e-05, | |
| "loss": 6.913, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 4.524507750314202, | |
| "grad_norm": 2.063062906265259, | |
| "learning_rate": 4.246124842899037e-05, | |
| "loss": 6.8943, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 4.566401340594889, | |
| "grad_norm": 1.9563026428222656, | |
| "learning_rate": 4.2391425778522555e-05, | |
| "loss": 6.8986, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 4.6082949308755765, | |
| "grad_norm": 1.8872498273849487, | |
| "learning_rate": 4.2321603128054746e-05, | |
| "loss": 6.8883, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 4.650188521156263, | |
| "grad_norm": 2.1376144886016846, | |
| "learning_rate": 4.225178047758693e-05, | |
| "loss": 6.8716, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 4.69208211143695, | |
| "grad_norm": 1.938679575920105, | |
| "learning_rate": 4.218195782711912e-05, | |
| "loss": 6.8836, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 4.733975701717637, | |
| "grad_norm": 1.9372957944869995, | |
| "learning_rate": 4.2112135176651305e-05, | |
| "loss": 6.8925, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 4.775869291998324, | |
| "grad_norm": 2.716827630996704, | |
| "learning_rate": 4.2042312526183496e-05, | |
| "loss": 6.8284, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 4.817762882279011, | |
| "grad_norm": 1.942700743675232, | |
| "learning_rate": 4.197248987571568e-05, | |
| "loss": 6.8753, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.859656472559698, | |
| "grad_norm": 2.026385545730591, | |
| "learning_rate": 4.190266722524787e-05, | |
| "loss": 6.8707, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 4.901550062840386, | |
| "grad_norm": 1.7594517469406128, | |
| "learning_rate": 4.183284457478006e-05, | |
| "loss": 6.8427, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 4.943443653121072, | |
| "grad_norm": 1.8161801099777222, | |
| "learning_rate": 4.176302192431225e-05, | |
| "loss": 6.8519, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 4.9853372434017595, | |
| "grad_norm": 2.6034481525421143, | |
| "learning_rate": 4.169319927384444e-05, | |
| "loss": 6.8448, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 5.027230833682447, | |
| "grad_norm": 1.93776535987854, | |
| "learning_rate": 4.162337662337663e-05, | |
| "loss": 6.8071, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.0691244239631335, | |
| "grad_norm": 2.0754964351654053, | |
| "learning_rate": 4.155355397290881e-05, | |
| "loss": 6.8386, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 5.111018014243821, | |
| "grad_norm": 2.0640342235565186, | |
| "learning_rate": 4.1483731322440996e-05, | |
| "loss": 6.8402, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 5.152911604524507, | |
| "grad_norm": 1.8218064308166504, | |
| "learning_rate": 4.141390867197319e-05, | |
| "loss": 6.8153, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 5.194805194805195, | |
| "grad_norm": 2.0181634426116943, | |
| "learning_rate": 4.134408602150538e-05, | |
| "loss": 6.8104, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 5.236698785085882, | |
| "grad_norm": 2.5224316120147705, | |
| "learning_rate": 4.127426337103757e-05, | |
| "loss": 6.8355, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.278592375366569, | |
| "grad_norm": 3.1008002758026123, | |
| "learning_rate": 4.120513894707444e-05, | |
| "loss": 6.8384, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 5.320485965647256, | |
| "grad_norm": 1.8872394561767578, | |
| "learning_rate": 4.113531629660662e-05, | |
| "loss": 6.8087, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 5.362379555927943, | |
| "grad_norm": 2.109281063079834, | |
| "learning_rate": 4.1065493646138805e-05, | |
| "loss": 6.8161, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 5.40427314620863, | |
| "grad_norm": 1.7881128787994385, | |
| "learning_rate": 4.0995670995670996e-05, | |
| "loss": 6.8215, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 5.446166736489317, | |
| "grad_norm": 2.5179624557495117, | |
| "learning_rate": 4.092584834520319e-05, | |
| "loss": 6.7883, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.488060326770004, | |
| "grad_norm": 2.4349751472473145, | |
| "learning_rate": 4.085602569473537e-05, | |
| "loss": 6.792, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 5.529953917050691, | |
| "grad_norm": 2.011018991470337, | |
| "learning_rate": 4.078620304426756e-05, | |
| "loss": 6.7846, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 5.571847507331379, | |
| "grad_norm": 2.519958019256592, | |
| "learning_rate": 4.071638039379975e-05, | |
| "loss": 6.7887, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 5.613741097612065, | |
| "grad_norm": 1.9241886138916016, | |
| "learning_rate": 4.064655774333194e-05, | |
| "loss": 6.7662, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 5.655634687892753, | |
| "grad_norm": 1.8995391130447388, | |
| "learning_rate": 4.057673509286413e-05, | |
| "loss": 6.7672, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 5.697528278173439, | |
| "grad_norm": 2.1511363983154297, | |
| "learning_rate": 4.050691244239632e-05, | |
| "loss": 6.7867, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 5.7394218684541265, | |
| "grad_norm": 1.8995012044906616, | |
| "learning_rate": 4.04370897919285e-05, | |
| "loss": 6.7563, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 5.781315458734814, | |
| "grad_norm": 1.83163321018219, | |
| "learning_rate": 4.036726714146069e-05, | |
| "loss": 6.7848, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 5.8232090490155, | |
| "grad_norm": 2.2616159915924072, | |
| "learning_rate": 4.029744449099288e-05, | |
| "loss": 6.7896, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 5.865102639296188, | |
| "grad_norm": 2.0548572540283203, | |
| "learning_rate": 4.0228320067029746e-05, | |
| "loss": 6.7633, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 5.906996229576874, | |
| "grad_norm": 2.4749302864074707, | |
| "learning_rate": 4.015849741656194e-05, | |
| "loss": 6.7267, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 5.948889819857562, | |
| "grad_norm": 1.906648874282837, | |
| "learning_rate": 4.008867476609413e-05, | |
| "loss": 6.7645, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 5.990783410138249, | |
| "grad_norm": 2.0839619636535645, | |
| "learning_rate": 4.001885211562631e-05, | |
| "loss": 6.8082, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 6.032677000418936, | |
| "grad_norm": 2.1202664375305176, | |
| "learning_rate": 3.9949029465158496e-05, | |
| "loss": 6.7625, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 6.074570590699623, | |
| "grad_norm": 1.988951563835144, | |
| "learning_rate": 3.987920681469069e-05, | |
| "loss": 6.7413, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.11646418098031, | |
| "grad_norm": 2.4327659606933594, | |
| "learning_rate": 3.980938416422287e-05, | |
| "loss": 6.7123, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 6.158357771260997, | |
| "grad_norm": 2.07710599899292, | |
| "learning_rate": 3.973956151375506e-05, | |
| "loss": 6.7316, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 6.200251361541684, | |
| "grad_norm": 1.9640876054763794, | |
| "learning_rate": 3.966973886328725e-05, | |
| "loss": 6.752, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 6.242144951822371, | |
| "grad_norm": 2.3012888431549072, | |
| "learning_rate": 3.959991621281944e-05, | |
| "loss": 6.7188, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 6.284038542103058, | |
| "grad_norm": 2.0262773036956787, | |
| "learning_rate": 3.953009356235163e-05, | |
| "loss": 6.7255, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.325932132383746, | |
| "grad_norm": 1.8689815998077393, | |
| "learning_rate": 3.946027091188382e-05, | |
| "loss": 6.7132, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 6.367825722664432, | |
| "grad_norm": 2.188612937927246, | |
| "learning_rate": 3.939044826141601e-05, | |
| "loss": 6.7407, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 6.4097193129451195, | |
| "grad_norm": 2.0168368816375732, | |
| "learning_rate": 3.9320625610948195e-05, | |
| "loss": 6.7132, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 6.451612903225806, | |
| "grad_norm": 2.496889352798462, | |
| "learning_rate": 3.925080296048038e-05, | |
| "loss": 6.7003, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 6.4935064935064934, | |
| "grad_norm": 2.1601486206054688, | |
| "learning_rate": 3.918098031001257e-05, | |
| "loss": 6.7056, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 6.535400083787181, | |
| "grad_norm": 2.300112009048462, | |
| "learning_rate": 3.9111157659544754e-05, | |
| "loss": 6.7314, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 6.577293674067867, | |
| "grad_norm": 2.321880578994751, | |
| "learning_rate": 3.9041335009076945e-05, | |
| "loss": 6.7166, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 6.619187264348555, | |
| "grad_norm": 2.029465913772583, | |
| "learning_rate": 3.8971512358609136e-05, | |
| "loss": 6.6908, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 6.661080854629242, | |
| "grad_norm": 2.258577585220337, | |
| "learning_rate": 3.890168970814133e-05, | |
| "loss": 6.7359, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 6.702974444909929, | |
| "grad_norm": 2.3579437732696533, | |
| "learning_rate": 3.883186705767351e-05, | |
| "loss": 6.7021, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 6.744868035190616, | |
| "grad_norm": 2.236828565597534, | |
| "learning_rate": 3.87620444072057e-05, | |
| "loss": 6.6897, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 6.786761625471303, | |
| "grad_norm": 2.6255593299865723, | |
| "learning_rate": 3.869222175673789e-05, | |
| "loss": 6.6899, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 6.82865521575199, | |
| "grad_norm": 2.297067880630493, | |
| "learning_rate": 3.862239910627008e-05, | |
| "loss": 6.7058, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 6.870548806032677, | |
| "grad_norm": 2.440605640411377, | |
| "learning_rate": 3.8553274682306945e-05, | |
| "loss": 6.6559, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 6.912442396313364, | |
| "grad_norm": 2.0427000522613525, | |
| "learning_rate": 3.848345203183913e-05, | |
| "loss": 6.6799, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 6.954335986594051, | |
| "grad_norm": 2.0323081016540527, | |
| "learning_rate": 3.841362938137132e-05, | |
| "loss": 6.6863, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 6.996229576874738, | |
| "grad_norm": 3.407731533050537, | |
| "learning_rate": 3.834380673090351e-05, | |
| "loss": 6.6767, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 7.038123167155425, | |
| "grad_norm": 2.112870931625366, | |
| "learning_rate": 3.8273984080435695e-05, | |
| "loss": 6.682, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 7.080016757436113, | |
| "grad_norm": 2.710810422897339, | |
| "learning_rate": 3.8204161429967886e-05, | |
| "loss": 6.7046, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 7.121910347716799, | |
| "grad_norm": 2.0754942893981934, | |
| "learning_rate": 3.813433877950007e-05, | |
| "loss": 6.6511, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 7.1638039379974865, | |
| "grad_norm": 3.1009552478790283, | |
| "learning_rate": 3.8064516129032254e-05, | |
| "loss": 6.666, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 7.205697528278174, | |
| "grad_norm": 2.1582441329956055, | |
| "learning_rate": 3.7994693478564445e-05, | |
| "loss": 6.6574, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 7.24759111855886, | |
| "grad_norm": 2.680147647857666, | |
| "learning_rate": 3.7924870828096636e-05, | |
| "loss": 6.6814, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 7.289484708839548, | |
| "grad_norm": 2.0264320373535156, | |
| "learning_rate": 3.785504817762883e-05, | |
| "loss": 6.668, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 7.331378299120234, | |
| "grad_norm": 2.032093048095703, | |
| "learning_rate": 3.778522552716101e-05, | |
| "loss": 6.6603, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 7.373271889400922, | |
| "grad_norm": 2.4837894439697266, | |
| "learning_rate": 3.77154028766932e-05, | |
| "loss": 6.6817, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 7.415165479681609, | |
| "grad_norm": 2.70166015625, | |
| "learning_rate": 3.764558022622539e-05, | |
| "loss": 6.6657, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 7.457059069962296, | |
| "grad_norm": 2.3508477210998535, | |
| "learning_rate": 3.757575757575758e-05, | |
| "loss": 6.6314, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 7.498952660242983, | |
| "grad_norm": 2.450437307357788, | |
| "learning_rate": 3.750593492528977e-05, | |
| "loss": 6.6551, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 7.5408462505236695, | |
| "grad_norm": 1.9939864873886108, | |
| "learning_rate": 3.743611227482195e-05, | |
| "loss": 6.6128, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 7.582739840804357, | |
| "grad_norm": 2.470285177230835, | |
| "learning_rate": 3.736628962435414e-05, | |
| "loss": 6.6126, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 7.624633431085044, | |
| "grad_norm": 2.5651469230651855, | |
| "learning_rate": 3.729646697388633e-05, | |
| "loss": 6.6694, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 7.666527021365731, | |
| "grad_norm": 2.361785650253296, | |
| "learning_rate": 3.722664432341852e-05, | |
| "loss": 6.6349, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 7.708420611646418, | |
| "grad_norm": 2.371994972229004, | |
| "learning_rate": 3.715682167295071e-05, | |
| "loss": 6.6483, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 7.750314201927106, | |
| "grad_norm": 2.862107038497925, | |
| "learning_rate": 3.708769724898758e-05, | |
| "loss": 6.634, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 7.792207792207792, | |
| "grad_norm": 2.815486192703247, | |
| "learning_rate": 3.701787459851976e-05, | |
| "loss": 6.6324, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 7.8341013824884795, | |
| "grad_norm": 1.930017352104187, | |
| "learning_rate": 3.6948051948051945e-05, | |
| "loss": 6.6275, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 7.875994972769166, | |
| "grad_norm": 3.1758625507354736, | |
| "learning_rate": 3.6878229297584136e-05, | |
| "loss": 6.6529, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 7.9178885630498534, | |
| "grad_norm": 2.1219429969787598, | |
| "learning_rate": 3.680840664711633e-05, | |
| "loss": 6.6085, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 7.95978215333054, | |
| "grad_norm": 2.1965785026550293, | |
| "learning_rate": 3.673858399664851e-05, | |
| "loss": 6.6206, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 8.001675743611228, | |
| "grad_norm": 2.489473581314087, | |
| "learning_rate": 3.66687613461807e-05, | |
| "loss": 6.6089, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 8.043569333891915, | |
| "grad_norm": 2.3411850929260254, | |
| "learning_rate": 3.659893869571289e-05, | |
| "loss": 6.6286, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 8.085462924172601, | |
| "grad_norm": 2.32071590423584, | |
| "learning_rate": 3.6529116045245084e-05, | |
| "loss": 6.5984, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 8.12735651445329, | |
| "grad_norm": 2.402956247329712, | |
| "learning_rate": 3.645929339477727e-05, | |
| "loss": 6.5952, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 8.169250104733976, | |
| "grad_norm": 2.6951029300689697, | |
| "learning_rate": 3.638947074430946e-05, | |
| "loss": 6.6106, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 8.211143695014663, | |
| "grad_norm": 2.807187080383301, | |
| "learning_rate": 3.6319648093841643e-05, | |
| "loss": 6.6109, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 8.253037285295349, | |
| "grad_norm": 2.798614025115967, | |
| "learning_rate": 3.624982544337383e-05, | |
| "loss": 6.6052, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 8.294930875576037, | |
| "grad_norm": 4.015589237213135, | |
| "learning_rate": 3.618000279290602e-05, | |
| "loss": 6.5995, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 8.336824465856724, | |
| "grad_norm": 2.6923959255218506, | |
| "learning_rate": 3.611018014243821e-05, | |
| "loss": 6.5855, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 8.37871805613741, | |
| "grad_norm": 2.112994909286499, | |
| "learning_rate": 3.6040357491970394e-05, | |
| "loss": 6.5968, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 8.420611646418099, | |
| "grad_norm": 2.8196451663970947, | |
| "learning_rate": 3.5970534841502585e-05, | |
| "loss": 6.5977, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 8.462505236698785, | |
| "grad_norm": 2.2421326637268066, | |
| "learning_rate": 3.5900712191034776e-05, | |
| "loss": 6.5846, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 8.504398826979472, | |
| "grad_norm": 2.634634256362915, | |
| "learning_rate": 3.583088954056697e-05, | |
| "loss": 6.5955, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 8.54629241726016, | |
| "grad_norm": 2.101125955581665, | |
| "learning_rate": 3.576106689009915e-05, | |
| "loss": 6.6013, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 8.588186007540846, | |
| "grad_norm": 2.719330072402954, | |
| "learning_rate": 3.569194246613601e-05, | |
| "loss": 6.5668, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 8.630079597821533, | |
| "grad_norm": 2.283790349960327, | |
| "learning_rate": 3.56221198156682e-05, | |
| "loss": 6.6107, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 8.671973188102221, | |
| "grad_norm": 2.1805171966552734, | |
| "learning_rate": 3.5552297165200393e-05, | |
| "loss": 6.5875, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 8.713866778382908, | |
| "grad_norm": 2.6632487773895264, | |
| "learning_rate": 3.5482474514732584e-05, | |
| "loss": 6.613, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 8.755760368663594, | |
| "grad_norm": 2.3296337127685547, | |
| "learning_rate": 3.541265186426477e-05, | |
| "loss": 6.5628, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 8.79765395894428, | |
| "grad_norm": 2.8429343700408936, | |
| "learning_rate": 3.534282921379696e-05, | |
| "loss": 6.5823, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 8.839547549224969, | |
| "grad_norm": 2.4361233711242676, | |
| "learning_rate": 3.527300656332915e-05, | |
| "loss": 6.5853, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 8.881441139505656, | |
| "grad_norm": 2.5633111000061035, | |
| "learning_rate": 3.5203183912861335e-05, | |
| "loss": 6.5979, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 8.923334729786342, | |
| "grad_norm": 2.350463628768921, | |
| "learning_rate": 3.513336126239352e-05, | |
| "loss": 6.5744, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 8.96522832006703, | |
| "grad_norm": 2.456291675567627, | |
| "learning_rate": 3.506353861192571e-05, | |
| "loss": 6.57, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 9.007121910347717, | |
| "grad_norm": 2.401036262512207, | |
| "learning_rate": 3.49937159614579e-05, | |
| "loss": 6.5614, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 9.049015500628403, | |
| "grad_norm": 2.5537233352661133, | |
| "learning_rate": 3.4923893310990085e-05, | |
| "loss": 6.5836, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 9.090909090909092, | |
| "grad_norm": 2.6386375427246094, | |
| "learning_rate": 3.4854070660522276e-05, | |
| "loss": 6.6178, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 9.132802681189778, | |
| "grad_norm": 2.508533477783203, | |
| "learning_rate": 3.478424801005447e-05, | |
| "loss": 6.5761, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 9.174696271470465, | |
| "grad_norm": 3.1510419845581055, | |
| "learning_rate": 3.471442535958665e-05, | |
| "loss": 6.558, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 9.216589861751151, | |
| "grad_norm": 2.6325526237487793, | |
| "learning_rate": 3.464460270911884e-05, | |
| "loss": 6.5661, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 9.25848345203184, | |
| "grad_norm": 2.9870827198028564, | |
| "learning_rate": 3.457478005865103e-05, | |
| "loss": 6.5392, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 9.300377042312526, | |
| "grad_norm": 2.4924209117889404, | |
| "learning_rate": 3.450495740818322e-05, | |
| "loss": 6.547, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 9.342270632593213, | |
| "grad_norm": 2.3227298259735107, | |
| "learning_rate": 3.44351347577154e-05, | |
| "loss": 6.5306, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 9.3841642228739, | |
| "grad_norm": 2.867182731628418, | |
| "learning_rate": 3.436531210724759e-05, | |
| "loss": 6.5628, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 9.426057813154587, | |
| "grad_norm": 2.2619149684906006, | |
| "learning_rate": 3.429548945677978e-05, | |
| "loss": 6.5192, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 9.467951403435274, | |
| "grad_norm": 2.232321262359619, | |
| "learning_rate": 3.422636503281665e-05, | |
| "loss": 6.56, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 9.509844993715962, | |
| "grad_norm": 2.4485862255096436, | |
| "learning_rate": 3.4156542382348835e-05, | |
| "loss": 6.557, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 9.551738583996649, | |
| "grad_norm": 2.4476943016052246, | |
| "learning_rate": 3.4086719731881026e-05, | |
| "loss": 6.5314, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 9.593632174277335, | |
| "grad_norm": 2.491731643676758, | |
| "learning_rate": 3.401689708141321e-05, | |
| "loss": 6.4952, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 9.635525764558023, | |
| "grad_norm": 2.6474783420562744, | |
| "learning_rate": 3.39470744309454e-05, | |
| "loss": 6.5499, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 9.67741935483871, | |
| "grad_norm": 2.5691514015197754, | |
| "learning_rate": 3.3877251780477585e-05, | |
| "loss": 6.5417, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 9.719312945119396, | |
| "grad_norm": 2.601832151412964, | |
| "learning_rate": 3.3807429130009776e-05, | |
| "loss": 6.5584, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 9.761206535400083, | |
| "grad_norm": 3.481239080429077, | |
| "learning_rate": 3.373760647954197e-05, | |
| "loss": 6.5403, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 9.803100125680771, | |
| "grad_norm": 3.0747485160827637, | |
| "learning_rate": 3.366778382907415e-05, | |
| "loss": 6.5751, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 9.844993715961458, | |
| "grad_norm": 2.2310988903045654, | |
| "learning_rate": 3.359796117860634e-05, | |
| "loss": 6.5046, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 9.886887306242144, | |
| "grad_norm": 2.4555273056030273, | |
| "learning_rate": 3.352813852813853e-05, | |
| "loss": 6.5544, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 9.928780896522833, | |
| "grad_norm": 3.1235666275024414, | |
| "learning_rate": 3.345831587767072e-05, | |
| "loss": 6.5396, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 9.970674486803519, | |
| "grad_norm": 2.2766611576080322, | |
| "learning_rate": 3.338849322720291e-05, | |
| "loss": 6.5542, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 10.012568077084206, | |
| "grad_norm": 3.0408995151519775, | |
| "learning_rate": 3.331867057673509e-05, | |
| "loss": 6.4978, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 10.054461667364894, | |
| "grad_norm": 2.8702831268310547, | |
| "learning_rate": 3.3248847926267283e-05, | |
| "loss": 6.5264, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 10.09635525764558, | |
| "grad_norm": 2.9117937088012695, | |
| "learning_rate": 3.317902527579947e-05, | |
| "loss": 6.5028, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 10.138248847926267, | |
| "grad_norm": 2.925631046295166, | |
| "learning_rate": 3.310920262533166e-05, | |
| "loss": 6.5143, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 10.180142438206955, | |
| "grad_norm": 2.6605536937713623, | |
| "learning_rate": 3.303937997486385e-05, | |
| "loss": 6.5394, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 10.222036028487642, | |
| "grad_norm": 2.31357479095459, | |
| "learning_rate": 3.2969557324396034e-05, | |
| "loss": 6.5224, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 10.263929618768328, | |
| "grad_norm": 2.6544747352600098, | |
| "learning_rate": 3.2899734673928225e-05, | |
| "loss": 6.5035, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 10.305823209049015, | |
| "grad_norm": 2.5945372581481934, | |
| "learning_rate": 3.2830610249965085e-05, | |
| "loss": 6.4977, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 10.347716799329703, | |
| "grad_norm": 3.120873212814331, | |
| "learning_rate": 3.2760787599497276e-05, | |
| "loss": 6.5399, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 10.38961038961039, | |
| "grad_norm": 2.705008029937744, | |
| "learning_rate": 3.269096494902947e-05, | |
| "loss": 6.4938, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 10.431503979891076, | |
| "grad_norm": 2.2395503520965576, | |
| "learning_rate": 3.262114229856166e-05, | |
| "loss": 6.4854, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 10.473397570171764, | |
| "grad_norm": 2.5891764163970947, | |
| "learning_rate": 3.255131964809384e-05, | |
| "loss": 6.5107, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 10.51529116045245, | |
| "grad_norm": 3.115931749343872, | |
| "learning_rate": 3.248149699762603e-05, | |
| "loss": 6.5389, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 10.557184750733137, | |
| "grad_norm": 2.264437675476074, | |
| "learning_rate": 3.2411674347158224e-05, | |
| "loss": 6.51, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 10.599078341013826, | |
| "grad_norm": 3.449631690979004, | |
| "learning_rate": 3.234185169669041e-05, | |
| "loss": 6.5161, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 10.640971931294512, | |
| "grad_norm": 2.478337526321411, | |
| "learning_rate": 3.227202904622259e-05, | |
| "loss": 6.5019, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 10.682865521575199, | |
| "grad_norm": 3.2756478786468506, | |
| "learning_rate": 3.2202206395754784e-05, | |
| "loss": 6.4869, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 10.724759111855885, | |
| "grad_norm": 2.7576985359191895, | |
| "learning_rate": 3.213238374528697e-05, | |
| "loss": 6.5206, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 10.766652702136573, | |
| "grad_norm": 2.200963020324707, | |
| "learning_rate": 3.206256109481916e-05, | |
| "loss": 6.48, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 10.80854629241726, | |
| "grad_norm": 2.7358744144439697, | |
| "learning_rate": 3.199273844435135e-05, | |
| "loss": 6.5126, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 10.850439882697946, | |
| "grad_norm": 2.7179319858551025, | |
| "learning_rate": 3.192291579388354e-05, | |
| "loss": 6.4699, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 10.892333472978635, | |
| "grad_norm": 2.811340808868408, | |
| "learning_rate": 3.1853093143415725e-05, | |
| "loss": 6.5056, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 10.934227063259321, | |
| "grad_norm": 3.010690450668335, | |
| "learning_rate": 3.1783270492947916e-05, | |
| "loss": 6.5103, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 10.976120653540008, | |
| "grad_norm": 3.213487148284912, | |
| "learning_rate": 3.171344784248011e-05, | |
| "loss": 6.4874, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 11.018014243820696, | |
| "grad_norm": 2.5710039138793945, | |
| "learning_rate": 3.164362519201229e-05, | |
| "loss": 6.4919, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 11.059907834101383, | |
| "grad_norm": 2.6933746337890625, | |
| "learning_rate": 3.1573802541544475e-05, | |
| "loss": 6.5284, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 11.101801424382069, | |
| "grad_norm": 3.775012254714966, | |
| "learning_rate": 3.1503979891076666e-05, | |
| "loss": 6.4894, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 11.143695014662757, | |
| "grad_norm": 3.2401301860809326, | |
| "learning_rate": 3.1434855467113534e-05, | |
| "loss": 6.4721, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 11.185588604943444, | |
| "grad_norm": 2.642794132232666, | |
| "learning_rate": 3.1365032816645725e-05, | |
| "loss": 6.4797, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 11.22748219522413, | |
| "grad_norm": 3.191567897796631, | |
| "learning_rate": 3.129521016617791e-05, | |
| "loss": 6.5022, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 11.269375785504817, | |
| "grad_norm": 2.816554307937622, | |
| "learning_rate": 3.12253875157101e-05, | |
| "loss": 6.4853, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 11.311269375785505, | |
| "grad_norm": 2.8666136264801025, | |
| "learning_rate": 3.1155564865242284e-05, | |
| "loss": 6.4839, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 11.353162966066192, | |
| "grad_norm": 2.9831254482269287, | |
| "learning_rate": 3.1085742214774475e-05, | |
| "loss": 6.5082, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 11.395056556346878, | |
| "grad_norm": 2.7065083980560303, | |
| "learning_rate": 3.101591956430666e-05, | |
| "loss": 6.4412, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 11.436950146627566, | |
| "grad_norm": 2.5580694675445557, | |
| "learning_rate": 3.094609691383885e-05, | |
| "loss": 6.4849, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 11.478843736908253, | |
| "grad_norm": 2.571390390396118, | |
| "learning_rate": 3.087627426337104e-05, | |
| "loss": 6.4689, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 11.52073732718894, | |
| "grad_norm": 2.835906982421875, | |
| "learning_rate": 3.0806451612903225e-05, | |
| "loss": 6.4887, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 11.562630917469628, | |
| "grad_norm": 3.1355161666870117, | |
| "learning_rate": 3.0736628962435416e-05, | |
| "loss": 6.4568, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 11.604524507750314, | |
| "grad_norm": 3.0155599117279053, | |
| "learning_rate": 3.066680631196761e-05, | |
| "loss": 6.4607, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 11.646418098031, | |
| "grad_norm": 2.6346957683563232, | |
| "learning_rate": 3.059698366149979e-05, | |
| "loss": 6.4706, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 11.688311688311689, | |
| "grad_norm": 2.4353625774383545, | |
| "learning_rate": 3.052716101103198e-05, | |
| "loss": 6.482, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 11.730205278592376, | |
| "grad_norm": 3.29835844039917, | |
| "learning_rate": 3.045733836056417e-05, | |
| "loss": 6.4625, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 11.772098868873062, | |
| "grad_norm": 2.233579158782959, | |
| "learning_rate": 3.038751571009636e-05, | |
| "loss": 6.4727, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 11.813992459153749, | |
| "grad_norm": 2.5708439350128174, | |
| "learning_rate": 3.0317693059628545e-05, | |
| "loss": 6.4751, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 11.855886049434437, | |
| "grad_norm": 2.29488205909729, | |
| "learning_rate": 3.0247870409160732e-05, | |
| "loss": 6.4599, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 11.897779639715123, | |
| "grad_norm": 2.858208179473877, | |
| "learning_rate": 3.0178047758692923e-05, | |
| "loss": 6.469, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 11.93967322999581, | |
| "grad_norm": 2.854923725128174, | |
| "learning_rate": 3.0108225108225107e-05, | |
| "loss": 6.4995, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 11.981566820276498, | |
| "grad_norm": 2.590484857559204, | |
| "learning_rate": 3.003910068426198e-05, | |
| "loss": 6.4508, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 12.023460410557185, | |
| "grad_norm": 3.3479676246643066, | |
| "learning_rate": 2.9969278033794163e-05, | |
| "loss": 6.4581, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 12.065354000837871, | |
| "grad_norm": 2.7855923175811768, | |
| "learning_rate": 2.9899455383326354e-05, | |
| "loss": 6.4744, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 12.10724759111856, | |
| "grad_norm": 3.2668962478637695, | |
| "learning_rate": 2.982963273285854e-05, | |
| "loss": 6.4731, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 12.149141181399246, | |
| "grad_norm": 2.850735664367676, | |
| "learning_rate": 2.9759810082390725e-05, | |
| "loss": 6.4788, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 12.191034771679933, | |
| "grad_norm": 2.9676952362060547, | |
| "learning_rate": 2.9689987431922916e-05, | |
| "loss": 6.4525, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 12.23292836196062, | |
| "grad_norm": 2.604408025741577, | |
| "learning_rate": 2.9620164781455107e-05, | |
| "loss": 6.4564, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 12.274821952241307, | |
| "grad_norm": 2.974653482437134, | |
| "learning_rate": 2.9550342130987295e-05, | |
| "loss": 6.463, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 12.316715542521994, | |
| "grad_norm": 3.372664213180542, | |
| "learning_rate": 2.9480519480519482e-05, | |
| "loss": 6.464, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 12.35860913280268, | |
| "grad_norm": 2.6891355514526367, | |
| "learning_rate": 2.941069683005167e-05, | |
| "loss": 6.4674, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 12.400502723083369, | |
| "grad_norm": 2.964113473892212, | |
| "learning_rate": 2.934087417958386e-05, | |
| "loss": 6.4539, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 12.442396313364055, | |
| "grad_norm": 2.7328097820281982, | |
| "learning_rate": 2.9271051529116045e-05, | |
| "loss": 6.4224, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 12.484289903644742, | |
| "grad_norm": 2.6205203533172607, | |
| "learning_rate": 2.9201228878648236e-05, | |
| "loss": 6.4266, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 12.52618349392543, | |
| "grad_norm": 3.681053400039673, | |
| "learning_rate": 2.9131406228180424e-05, | |
| "loss": 6.4549, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 12.568077084206116, | |
| "grad_norm": 2.9732627868652344, | |
| "learning_rate": 2.9061583577712608e-05, | |
| "loss": 6.4466, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 12.609970674486803, | |
| "grad_norm": 3.47816801071167, | |
| "learning_rate": 2.89917609272448e-05, | |
| "loss": 6.4408, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 12.651864264767491, | |
| "grad_norm": 2.70326566696167, | |
| "learning_rate": 2.892193827677699e-05, | |
| "loss": 6.4444, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 12.693757855048178, | |
| "grad_norm": 2.9219532012939453, | |
| "learning_rate": 2.8852115626309177e-05, | |
| "loss": 6.4183, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 12.735651445328864, | |
| "grad_norm": 2.8546571731567383, | |
| "learning_rate": 2.878229297584136e-05, | |
| "loss": 6.4399, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 12.777545035609553, | |
| "grad_norm": 2.95047926902771, | |
| "learning_rate": 2.8712470325373552e-05, | |
| "loss": 6.4396, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 12.819438625890239, | |
| "grad_norm": 3.397934675216675, | |
| "learning_rate": 2.8643345901410416e-05, | |
| "loss": 6.438, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 12.861332216170926, | |
| "grad_norm": 2.625852346420288, | |
| "learning_rate": 2.8573523250942607e-05, | |
| "loss": 6.4363, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 12.903225806451612, | |
| "grad_norm": 2.5299527645111084, | |
| "learning_rate": 2.85037006004748e-05, | |
| "loss": 6.3952, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 12.9451193967323, | |
| "grad_norm": 2.6445415019989014, | |
| "learning_rate": 2.8433877950006983e-05, | |
| "loss": 6.4559, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 12.987012987012987, | |
| "grad_norm": 2.9675769805908203, | |
| "learning_rate": 2.836405529953917e-05, | |
| "loss": 6.4447, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 13.028906577293673, | |
| "grad_norm": 2.607391119003296, | |
| "learning_rate": 2.829423264907136e-05, | |
| "loss": 6.446, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 13.070800167574362, | |
| "grad_norm": 3.196765661239624, | |
| "learning_rate": 2.8224409998603545e-05, | |
| "loss": 6.4336, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 13.112693757855048, | |
| "grad_norm": 5.778535842895508, | |
| "learning_rate": 2.8154587348135736e-05, | |
| "loss": 6.4339, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 13.154587348135735, | |
| "grad_norm": 3.0479419231414795, | |
| "learning_rate": 2.8084764697667927e-05, | |
| "loss": 6.4147, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 13.196480938416423, | |
| "grad_norm": 2.6787302494049072, | |
| "learning_rate": 2.8014942047200115e-05, | |
| "loss": 6.4312, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 13.23837452869711, | |
| "grad_norm": 2.7929670810699463, | |
| "learning_rate": 2.79451193967323e-05, | |
| "loss": 6.4224, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 13.280268118977796, | |
| "grad_norm": 2.722101926803589, | |
| "learning_rate": 2.787529674626449e-05, | |
| "loss": 6.4247, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 13.322161709258484, | |
| "grad_norm": 3.295348644256592, | |
| "learning_rate": 2.780547409579668e-05, | |
| "loss": 6.4435, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 13.36405529953917, | |
| "grad_norm": 2.5780696868896484, | |
| "learning_rate": 2.7735651445328865e-05, | |
| "loss": 6.406, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 13.405948889819857, | |
| "grad_norm": 2.955299139022827, | |
| "learning_rate": 2.7665828794861053e-05, | |
| "loss": 6.4633, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 13.447842480100544, | |
| "grad_norm": 3.8027708530426025, | |
| "learning_rate": 2.7596006144393244e-05, | |
| "loss": 6.4445, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 13.489736070381232, | |
| "grad_norm": 2.6895995140075684, | |
| "learning_rate": 2.7526183493925428e-05, | |
| "loss": 6.4015, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 13.531629660661919, | |
| "grad_norm": 2.6936516761779785, | |
| "learning_rate": 2.745636084345762e-05, | |
| "loss": 6.4211, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 13.573523250942605, | |
| "grad_norm": 2.948420763015747, | |
| "learning_rate": 2.738653819298981e-05, | |
| "loss": 6.4042, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 13.615416841223293, | |
| "grad_norm": 2.763885974884033, | |
| "learning_rate": 2.7316715542521997e-05, | |
| "loss": 6.393, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 13.65731043150398, | |
| "grad_norm": 3.1601672172546387, | |
| "learning_rate": 2.724759111855886e-05, | |
| "loss": 6.4398, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 13.699204021784666, | |
| "grad_norm": 2.4161715507507324, | |
| "learning_rate": 2.7177768468091052e-05, | |
| "loss": 6.401, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 13.741097612065355, | |
| "grad_norm": 3.0796055793762207, | |
| "learning_rate": 2.7107945817623236e-05, | |
| "loss": 6.4265, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 13.782991202346041, | |
| "grad_norm": 3.6223697662353516, | |
| "learning_rate": 2.7038123167155427e-05, | |
| "loss": 6.4075, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 13.824884792626728, | |
| "grad_norm": 2.6991615295410156, | |
| "learning_rate": 2.696830051668762e-05, | |
| "loss": 6.3912, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 13.866778382907416, | |
| "grad_norm": 3.1701860427856445, | |
| "learning_rate": 2.6898477866219803e-05, | |
| "loss": 6.4173, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 13.908671973188103, | |
| "grad_norm": 2.915432929992676, | |
| "learning_rate": 2.682865521575199e-05, | |
| "loss": 6.4179, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 13.950565563468789, | |
| "grad_norm": 3.155080795288086, | |
| "learning_rate": 2.675883256528418e-05, | |
| "loss": 6.3895, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 13.992459153749476, | |
| "grad_norm": 3.3861114978790283, | |
| "learning_rate": 2.6689009914816365e-05, | |
| "loss": 6.4279, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 14.034352744030164, | |
| "grad_norm": 3.301805019378662, | |
| "learning_rate": 2.6619187264348556e-05, | |
| "loss": 6.4072, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 14.07624633431085, | |
| "grad_norm": 3.305147171020508, | |
| "learning_rate": 2.6549364613880744e-05, | |
| "loss": 6.3949, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 14.118139924591537, | |
| "grad_norm": 2.7602477073669434, | |
| "learning_rate": 2.6479541963412935e-05, | |
| "loss": 6.4048, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 14.160033514872225, | |
| "grad_norm": 2.5257952213287354, | |
| "learning_rate": 2.640971931294512e-05, | |
| "loss": 6.4033, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 14.201927105152912, | |
| "grad_norm": 2.4649853706359863, | |
| "learning_rate": 2.633989666247731e-05, | |
| "loss": 6.374, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 14.243820695433598, | |
| "grad_norm": 2.7136335372924805, | |
| "learning_rate": 2.6270074012009497e-05, | |
| "loss": 6.3993, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 14.285714285714286, | |
| "grad_norm": 2.801712989807129, | |
| "learning_rate": 2.6200251361541685e-05, | |
| "loss": 6.4059, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 14.327607875994973, | |
| "grad_norm": 2.7054030895233154, | |
| "learning_rate": 2.6130428711073873e-05, | |
| "loss": 6.431, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 14.36950146627566, | |
| "grad_norm": 2.653932809829712, | |
| "learning_rate": 2.6060606060606063e-05, | |
| "loss": 6.4035, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 14.411395056556348, | |
| "grad_norm": 2.5450570583343506, | |
| "learning_rate": 2.5990783410138248e-05, | |
| "loss": 6.417, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 14.453288646837034, | |
| "grad_norm": 2.9578003883361816, | |
| "learning_rate": 2.592096075967044e-05, | |
| "loss": 6.4087, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 14.49518223711772, | |
| "grad_norm": 2.9408493041992188, | |
| "learning_rate": 2.5851836335707303e-05, | |
| "loss": 6.3936, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 14.537075827398407, | |
| "grad_norm": 2.756441116333008, | |
| "learning_rate": 2.5782013685239494e-05, | |
| "loss": 6.404, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 14.578969417679096, | |
| "grad_norm": 3.685004711151123, | |
| "learning_rate": 2.571219103477168e-05, | |
| "loss": 6.3932, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 14.620863007959782, | |
| "grad_norm": 2.670825719833374, | |
| "learning_rate": 2.5642368384303872e-05, | |
| "loss": 6.3839, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 14.662756598240469, | |
| "grad_norm": 3.0986082553863525, | |
| "learning_rate": 2.5572545733836056e-05, | |
| "loss": 6.3782, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 14.704650188521157, | |
| "grad_norm": 3.003432273864746, | |
| "learning_rate": 2.5502723083368247e-05, | |
| "loss": 6.3775, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 14.746543778801843, | |
| "grad_norm": 2.752516269683838, | |
| "learning_rate": 2.5432900432900435e-05, | |
| "loss": 6.3731, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 14.78843736908253, | |
| "grad_norm": 2.7697649002075195, | |
| "learning_rate": 2.536307778243262e-05, | |
| "loss": 6.3701, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 14.830330959363218, | |
| "grad_norm": 3.0245521068573, | |
| "learning_rate": 2.529325513196481e-05, | |
| "loss": 6.3916, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 14.872224549643905, | |
| "grad_norm": 3.1849350929260254, | |
| "learning_rate": 2.5223432481497e-05, | |
| "loss": 6.3993, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 14.914118139924591, | |
| "grad_norm": 3.6655123233795166, | |
| "learning_rate": 2.5153609831029185e-05, | |
| "loss": 6.3791, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 14.95601173020528, | |
| "grad_norm": 3.2252790927886963, | |
| "learning_rate": 2.5083787180561376e-05, | |
| "loss": 6.3865, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 14.997905320485966, | |
| "grad_norm": 2.8366169929504395, | |
| "learning_rate": 2.5013964530093564e-05, | |
| "loss": 6.3897, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 15.039798910766653, | |
| "grad_norm": 2.757725715637207, | |
| "learning_rate": 2.494414187962575e-05, | |
| "loss": 6.376, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 15.081692501047339, | |
| "grad_norm": 3.1640422344207764, | |
| "learning_rate": 2.4874319229157942e-05, | |
| "loss": 6.3796, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 15.123586091328027, | |
| "grad_norm": 2.849719285964966, | |
| "learning_rate": 2.480449657869013e-05, | |
| "loss": 6.3765, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 15.165479681608714, | |
| "grad_norm": 2.7223923206329346, | |
| "learning_rate": 2.4734673928222314e-05, | |
| "loss": 6.3953, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 15.2073732718894, | |
| "grad_norm": 3.173750162124634, | |
| "learning_rate": 2.4664851277754505e-05, | |
| "loss": 6.3724, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 15.249266862170089, | |
| "grad_norm": 3.054779529571533, | |
| "learning_rate": 2.4595028627286692e-05, | |
| "loss": 6.3764, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 15.291160452450775, | |
| "grad_norm": 3.277862071990967, | |
| "learning_rate": 2.4525205976818883e-05, | |
| "loss": 6.3583, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 15.333054042731462, | |
| "grad_norm": 2.9208297729492188, | |
| "learning_rate": 2.4456081552855748e-05, | |
| "loss": 6.3878, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 15.37494763301215, | |
| "grad_norm": 2.5356411933898926, | |
| "learning_rate": 2.4386258902387935e-05, | |
| "loss": 6.3705, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 15.416841223292836, | |
| "grad_norm": 2.8953468799591064, | |
| "learning_rate": 2.4316436251920126e-05, | |
| "loss": 6.3947, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 15.458734813573523, | |
| "grad_norm": 2.9166266918182373, | |
| "learning_rate": 2.424661360145231e-05, | |
| "loss": 6.3809, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 15.50062840385421, | |
| "grad_norm": 3.4554710388183594, | |
| "learning_rate": 2.41767909509845e-05, | |
| "loss": 6.3746, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 15.542521994134898, | |
| "grad_norm": 3.7208077907562256, | |
| "learning_rate": 2.410696830051669e-05, | |
| "loss": 6.3758, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 15.584415584415584, | |
| "grad_norm": 3.3161842823028564, | |
| "learning_rate": 2.4037145650048876e-05, | |
| "loss": 6.3744, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 15.62630917469627, | |
| "grad_norm": 2.4062047004699707, | |
| "learning_rate": 2.3967322999581064e-05, | |
| "loss": 6.381, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 15.668202764976959, | |
| "grad_norm": 3.1894476413726807, | |
| "learning_rate": 2.389750034911325e-05, | |
| "loss": 6.3895, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 15.710096355257646, | |
| "grad_norm": 2.9203104972839355, | |
| "learning_rate": 2.3827677698645442e-05, | |
| "loss": 6.363, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 15.751989945538332, | |
| "grad_norm": 3.000694513320923, | |
| "learning_rate": 2.375785504817763e-05, | |
| "loss": 6.3837, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 15.79388353581902, | |
| "grad_norm": 2.838684558868408, | |
| "learning_rate": 2.368803239770982e-05, | |
| "loss": 6.3859, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 15.835777126099707, | |
| "grad_norm": 2.648862361907959, | |
| "learning_rate": 2.3618209747242005e-05, | |
| "loss": 6.3411, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 15.877670716380393, | |
| "grad_norm": 3.5438232421875, | |
| "learning_rate": 2.3548387096774193e-05, | |
| "loss": 6.3627, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 15.91956430666108, | |
| "grad_norm": 2.8182501792907715, | |
| "learning_rate": 2.3478564446306384e-05, | |
| "loss": 6.3731, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 15.961457896941768, | |
| "grad_norm": 3.3253772258758545, | |
| "learning_rate": 2.340874179583857e-05, | |
| "loss": 6.396, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 16.003351487222456, | |
| "grad_norm": 3.668926954269409, | |
| "learning_rate": 2.3338919145370762e-05, | |
| "loss": 6.3437, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 16.045245077503143, | |
| "grad_norm": 3.028989315032959, | |
| "learning_rate": 2.3269096494902946e-05, | |
| "loss": 6.3837, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 16.08713866778383, | |
| "grad_norm": 3.220702648162842, | |
| "learning_rate": 2.3199273844435134e-05, | |
| "loss": 6.3609, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 16.129032258064516, | |
| "grad_norm": 3.1788036823272705, | |
| "learning_rate": 2.3129451193967325e-05, | |
| "loss": 6.3723, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 16.170925848345203, | |
| "grad_norm": 3.351151466369629, | |
| "learning_rate": 2.306032677000419e-05, | |
| "loss": 6.3731, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 16.21281943862589, | |
| "grad_norm": 2.933992862701416, | |
| "learning_rate": 2.299050411953638e-05, | |
| "loss": 6.3654, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 16.25471302890658, | |
| "grad_norm": 4.2123589515686035, | |
| "learning_rate": 2.2920681469068568e-05, | |
| "loss": 6.3364, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 16.296606619187266, | |
| "grad_norm": 2.9287397861480713, | |
| "learning_rate": 2.2850858818600755e-05, | |
| "loss": 6.3643, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 16.338500209467952, | |
| "grad_norm": 2.6518173217773438, | |
| "learning_rate": 2.2781036168132943e-05, | |
| "loss": 6.3538, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 16.38039379974864, | |
| "grad_norm": 3.490497589111328, | |
| "learning_rate": 2.271121351766513e-05, | |
| "loss": 6.365, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 16.422287390029325, | |
| "grad_norm": 3.090874195098877, | |
| "learning_rate": 2.264139086719732e-05, | |
| "loss": 6.3513, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 16.46418098031001, | |
| "grad_norm": 2.793083429336548, | |
| "learning_rate": 2.257156821672951e-05, | |
| "loss": 6.3815, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 16.506074570590698, | |
| "grad_norm": 2.656334638595581, | |
| "learning_rate": 2.2501745566261696e-05, | |
| "loss": 6.3677, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 16.547968160871388, | |
| "grad_norm": 2.950857162475586, | |
| "learning_rate": 2.2431922915793884e-05, | |
| "loss": 6.3601, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 16.589861751152075, | |
| "grad_norm": 2.948397636413574, | |
| "learning_rate": 2.236210026532607e-05, | |
| "loss": 6.3633, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 16.63175534143276, | |
| "grad_norm": 3.759934902191162, | |
| "learning_rate": 2.2292277614858262e-05, | |
| "loss": 6.3664, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 16.673648931713448, | |
| "grad_norm": 2.6607794761657715, | |
| "learning_rate": 2.222245496439045e-05, | |
| "loss": 6.3659, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 16.715542521994134, | |
| "grad_norm": 3.2569267749786377, | |
| "learning_rate": 2.2152632313922638e-05, | |
| "loss": 6.3477, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 16.75743611227482, | |
| "grad_norm": 3.1701977252960205, | |
| "learning_rate": 2.2082809663454825e-05, | |
| "loss": 6.3466, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 16.79932970255551, | |
| "grad_norm": 2.8855369091033936, | |
| "learning_rate": 2.2012987012987013e-05, | |
| "loss": 6.3774, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 16.841223292836197, | |
| "grad_norm": 2.8468215465545654, | |
| "learning_rate": 2.1943164362519204e-05, | |
| "loss": 6.3388, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 16.883116883116884, | |
| "grad_norm": 3.3314404487609863, | |
| "learning_rate": 2.187334171205139e-05, | |
| "loss": 6.3658, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 16.92501047339757, | |
| "grad_norm": 3.023106336593628, | |
| "learning_rate": 2.180351906158358e-05, | |
| "loss": 6.3443, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 16.966904063678257, | |
| "grad_norm": 3.2845230102539062, | |
| "learning_rate": 2.1733696411115766e-05, | |
| "loss": 6.3785, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 17.008797653958943, | |
| "grad_norm": 2.805790424346924, | |
| "learning_rate": 2.166457198715263e-05, | |
| "loss": 6.3792, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 17.05069124423963, | |
| "grad_norm": 2.893737554550171, | |
| "learning_rate": 2.159474933668482e-05, | |
| "loss": 6.3138, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 17.09258483452032, | |
| "grad_norm": 3.238863945007324, | |
| "learning_rate": 2.1525624912721686e-05, | |
| "loss": 6.3686, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 17.134478424801006, | |
| "grad_norm": 3.403582811355591, | |
| "learning_rate": 2.1455802262253877e-05, | |
| "loss": 6.3312, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 17.176372015081693, | |
| "grad_norm": 2.963287353515625, | |
| "learning_rate": 2.1385979611786064e-05, | |
| "loss": 6.3515, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 17.21826560536238, | |
| "grad_norm": 3.867340087890625, | |
| "learning_rate": 2.1316156961318255e-05, | |
| "loss": 6.3566, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 17.260159195643066, | |
| "grad_norm": 2.841190814971924, | |
| "learning_rate": 2.124633431085044e-05, | |
| "loss": 6.3308, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 17.302052785923753, | |
| "grad_norm": 2.872523307800293, | |
| "learning_rate": 2.1176511660382627e-05, | |
| "loss": 6.3433, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 17.34394637620444, | |
| "grad_norm": 3.156465530395508, | |
| "learning_rate": 2.1106689009914818e-05, | |
| "loss": 6.3779, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 17.38583996648513, | |
| "grad_norm": 3.5904667377471924, | |
| "learning_rate": 2.1036866359447005e-05, | |
| "loss": 6.3402, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 17.427733556765816, | |
| "grad_norm": 3.5753939151763916, | |
| "learning_rate": 2.0967043708979196e-05, | |
| "loss": 6.3572, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 17.469627147046502, | |
| "grad_norm": 3.129514217376709, | |
| "learning_rate": 2.089722105851138e-05, | |
| "loss": 6.3302, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 17.51152073732719, | |
| "grad_norm": 2.988732099533081, | |
| "learning_rate": 2.0827398408043568e-05, | |
| "loss": 6.3807, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 17.553414327607875, | |
| "grad_norm": 2.857875108718872, | |
| "learning_rate": 2.075757575757576e-05, | |
| "loss": 6.3519, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 17.59530791788856, | |
| "grad_norm": 4.023842811584473, | |
| "learning_rate": 2.0687753107107947e-05, | |
| "loss": 6.3467, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 17.63720150816925, | |
| "grad_norm": 3.049686908721924, | |
| "learning_rate": 2.0617930456640137e-05, | |
| "loss": 6.3306, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 17.679095098449938, | |
| "grad_norm": 3.3211073875427246, | |
| "learning_rate": 2.054810780617232e-05, | |
| "loss": 6.3611, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 17.720988688730625, | |
| "grad_norm": 3.064138174057007, | |
| "learning_rate": 2.047828515570451e-05, | |
| "loss": 6.3217, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 17.76288227901131, | |
| "grad_norm": 2.7812724113464355, | |
| "learning_rate": 2.04084625052367e-05, | |
| "loss": 6.3131, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 17.804775869291998, | |
| "grad_norm": 2.5516164302825928, | |
| "learning_rate": 2.0338639854768888e-05, | |
| "loss": 6.3428, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 17.846669459572684, | |
| "grad_norm": 2.9599711894989014, | |
| "learning_rate": 2.026881720430108e-05, | |
| "loss": 6.3545, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 17.88856304985337, | |
| "grad_norm": 2.8674137592315674, | |
| "learning_rate": 2.0198994553833263e-05, | |
| "loss": 6.3302, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 17.93045664013406, | |
| "grad_norm": 3.3227078914642334, | |
| "learning_rate": 2.012917190336545e-05, | |
| "loss": 6.3278, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 17.972350230414747, | |
| "grad_norm": 3.080399751663208, | |
| "learning_rate": 2.005934925289764e-05, | |
| "loss": 6.3206, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 18.014243820695434, | |
| "grad_norm": 4.004719257354736, | |
| "learning_rate": 1.998952660242983e-05, | |
| "loss": 6.3407, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 18.05613741097612, | |
| "grad_norm": 2.8186423778533936, | |
| "learning_rate": 1.991970395196202e-05, | |
| "loss": 6.3136, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 18.098031001256807, | |
| "grad_norm": 2.81748104095459, | |
| "learning_rate": 1.9849881301494204e-05, | |
| "loss": 6.3353, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 18.139924591537493, | |
| "grad_norm": 2.9991416931152344, | |
| "learning_rate": 1.9780058651026395e-05, | |
| "loss": 6.3194, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 18.181818181818183, | |
| "grad_norm": 3.4876794815063477, | |
| "learning_rate": 1.9710236000558583e-05, | |
| "loss": 6.3293, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 18.22371177209887, | |
| "grad_norm": 3.0756711959838867, | |
| "learning_rate": 1.964041335009077e-05, | |
| "loss": 6.341, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 18.265605362379556, | |
| "grad_norm": 3.171670436859131, | |
| "learning_rate": 1.9570590699622958e-05, | |
| "loss": 6.3075, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 18.307498952660243, | |
| "grad_norm": 3.3317439556121826, | |
| "learning_rate": 1.9500768049155145e-05, | |
| "loss": 6.3436, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 18.34939254294093, | |
| "grad_norm": 2.924349308013916, | |
| "learning_rate": 1.9430945398687336e-05, | |
| "loss": 6.3217, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 18.391286133221616, | |
| "grad_norm": 3.247955560684204, | |
| "learning_rate": 1.9361122748219524e-05, | |
| "loss": 6.3324, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 18.433179723502302, | |
| "grad_norm": 3.340263843536377, | |
| "learning_rate": 1.929130009775171e-05, | |
| "loss": 6.2993, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 18.475073313782993, | |
| "grad_norm": 2.973019599914551, | |
| "learning_rate": 1.92214774472839e-05, | |
| "loss": 6.3292, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 18.51696690406368, | |
| "grad_norm": 3.5055582523345947, | |
| "learning_rate": 1.9151654796816086e-05, | |
| "loss": 6.3175, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 18.558860494344366, | |
| "grad_norm": 2.9543776512145996, | |
| "learning_rate": 1.9081832146348277e-05, | |
| "loss": 6.3206, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 18.600754084625052, | |
| "grad_norm": 2.790940284729004, | |
| "learning_rate": 1.9012009495880465e-05, | |
| "loss": 6.3383, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 18.64264767490574, | |
| "grad_norm": 3.419908285140991, | |
| "learning_rate": 1.8942186845412653e-05, | |
| "loss": 6.3329, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 18.684541265186425, | |
| "grad_norm": 3.3396215438842773, | |
| "learning_rate": 1.887236419494484e-05, | |
| "loss": 6.312, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 18.726434855467115, | |
| "grad_norm": 2.6713643074035645, | |
| "learning_rate": 1.8802541544477028e-05, | |
| "loss": 6.315, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 18.7683284457478, | |
| "grad_norm": 3.2764880657196045, | |
| "learning_rate": 1.8733417120513895e-05, | |
| "loss": 6.3311, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 18.810222036028488, | |
| "grad_norm": 3.602581739425659, | |
| "learning_rate": 1.8663594470046083e-05, | |
| "loss": 6.327, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 18.852115626309175, | |
| "grad_norm": 3.052971124649048, | |
| "learning_rate": 1.8593771819578274e-05, | |
| "loss": 6.2911, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 18.89400921658986, | |
| "grad_norm": 3.0912699699401855, | |
| "learning_rate": 1.852394916911046e-05, | |
| "loss": 6.3057, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 18.935902806870548, | |
| "grad_norm": 2.631545305252075, | |
| "learning_rate": 1.845412651864265e-05, | |
| "loss": 6.3381, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 18.977796397151234, | |
| "grad_norm": 3.8213324546813965, | |
| "learning_rate": 1.8384303868174836e-05, | |
| "loss": 6.3123, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 19.019689987431924, | |
| "grad_norm": 3.3717353343963623, | |
| "learning_rate": 1.8314481217707024e-05, | |
| "loss": 6.3194, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 19.06158357771261, | |
| "grad_norm": 2.831409215927124, | |
| "learning_rate": 1.8244658567239215e-05, | |
| "loss": 6.3383, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 19.103477167993297, | |
| "grad_norm": 2.915093183517456, | |
| "learning_rate": 1.8174835916771403e-05, | |
| "loss": 6.3208, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 19.145370758273984, | |
| "grad_norm": 3.1236917972564697, | |
| "learning_rate": 1.810501326630359e-05, | |
| "loss": 6.3089, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 19.18726434855467, | |
| "grad_norm": 3.2876298427581787, | |
| "learning_rate": 1.8035190615835778e-05, | |
| "loss": 6.2975, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 19.229157938835357, | |
| "grad_norm": 2.6437103748321533, | |
| "learning_rate": 1.7965367965367965e-05, | |
| "loss": 6.3341, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 19.271051529116047, | |
| "grad_norm": 2.9252028465270996, | |
| "learning_rate": 1.7895545314900156e-05, | |
| "loss": 6.3404, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 19.312945119396733, | |
| "grad_norm": 3.4250340461730957, | |
| "learning_rate": 1.7825722664432344e-05, | |
| "loss": 6.3072, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 19.35483870967742, | |
| "grad_norm": 3.1287946701049805, | |
| "learning_rate": 1.775590001396453e-05, | |
| "loss": 6.3022, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 19.396732299958106, | |
| "grad_norm": 3.4577419757843018, | |
| "learning_rate": 1.76867755900014e-05, | |
| "loss": 6.2938, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 19.438625890238793, | |
| "grad_norm": 3.7131240367889404, | |
| "learning_rate": 1.7616952939533586e-05, | |
| "loss": 6.3088, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 19.48051948051948, | |
| "grad_norm": 3.6799802780151367, | |
| "learning_rate": 1.7547130289065774e-05, | |
| "loss": 6.3326, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 19.522413070800166, | |
| "grad_norm": 2.834351062774658, | |
| "learning_rate": 1.747730763859796e-05, | |
| "loss": 6.2952, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 19.564306661080856, | |
| "grad_norm": 3.0629451274871826, | |
| "learning_rate": 1.7407484988130152e-05, | |
| "loss": 6.3185, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 19.606200251361543, | |
| "grad_norm": 3.4801712036132812, | |
| "learning_rate": 1.733766233766234e-05, | |
| "loss": 6.3003, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 19.64809384164223, | |
| "grad_norm": 2.8250389099121094, | |
| "learning_rate": 1.7267839687194524e-05, | |
| "loss": 6.3033, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 19.689987431922916, | |
| "grad_norm": 3.5964672565460205, | |
| "learning_rate": 1.7198017036726715e-05, | |
| "loss": 6.293, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 19.731881022203602, | |
| "grad_norm": 2.7947146892547607, | |
| "learning_rate": 1.7128194386258903e-05, | |
| "loss": 6.2884, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 19.77377461248429, | |
| "grad_norm": 3.0473551750183105, | |
| "learning_rate": 1.7058371735791094e-05, | |
| "loss": 6.312, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 19.81566820276498, | |
| "grad_norm": 3.1810736656188965, | |
| "learning_rate": 1.698854908532328e-05, | |
| "loss": 6.3102, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 19.857561793045665, | |
| "grad_norm": 3.0046746730804443, | |
| "learning_rate": 1.6918726434855465e-05, | |
| "loss": 6.3115, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 19.89945538332635, | |
| "grad_norm": 2.6985220909118652, | |
| "learning_rate": 1.6848903784387656e-05, | |
| "loss": 6.3132, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 19.941348973607038, | |
| "grad_norm": 2.958906650543213, | |
| "learning_rate": 1.6779081133919844e-05, | |
| "loss": 6.3024, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 19.983242563887725, | |
| "grad_norm": 3.5484089851379395, | |
| "learning_rate": 1.6709258483452035e-05, | |
| "loss": 6.2989, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 20.02513615416841, | |
| "grad_norm": 4.328272342681885, | |
| "learning_rate": 1.6639435832984222e-05, | |
| "loss": 6.3162, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 20.067029744449098, | |
| "grad_norm": 3.0396926403045654, | |
| "learning_rate": 1.6569613182516407e-05, | |
| "loss": 6.3004, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 20.108923334729788, | |
| "grad_norm": 3.328972339630127, | |
| "learning_rate": 1.6499790532048598e-05, | |
| "loss": 6.2855, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 20.150816925010474, | |
| "grad_norm": 3.301114320755005, | |
| "learning_rate": 1.6429967881580785e-05, | |
| "loss": 6.2874, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 20.19271051529116, | |
| "grad_norm": 3.297041177749634, | |
| "learning_rate": 1.6360145231112976e-05, | |
| "loss": 6.3089, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 20.234604105571847, | |
| "grad_norm": 2.9122605323791504, | |
| "learning_rate": 1.6290322580645164e-05, | |
| "loss": 6.3157, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 20.276497695852534, | |
| "grad_norm": 2.8182084560394287, | |
| "learning_rate": 1.6220499930177348e-05, | |
| "loss": 6.3118, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 20.31839128613322, | |
| "grad_norm": 3.8560192584991455, | |
| "learning_rate": 1.615067727970954e-05, | |
| "loss": 6.2858, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 20.36028487641391, | |
| "grad_norm": 2.457240581512451, | |
| "learning_rate": 1.6080854629241726e-05, | |
| "loss": 6.3077, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 20.402178466694597, | |
| "grad_norm": 3.5376362800598145, | |
| "learning_rate": 1.6011031978773917e-05, | |
| "loss": 6.2892, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 20.444072056975283, | |
| "grad_norm": 3.3489222526550293, | |
| "learning_rate": 1.59412093283061e-05, | |
| "loss": 6.2973, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 20.48596564725597, | |
| "grad_norm": 3.600166082382202, | |
| "learning_rate": 1.587138667783829e-05, | |
| "loss": 6.31, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 20.527859237536656, | |
| "grad_norm": 3.255598783493042, | |
| "learning_rate": 1.580156402737048e-05, | |
| "loss": 6.2389, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 20.569752827817343, | |
| "grad_norm": 3.166994094848633, | |
| "learning_rate": 1.5731741376902668e-05, | |
| "loss": 6.303, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 20.61164641809803, | |
| "grad_norm": 3.615269184112549, | |
| "learning_rate": 1.566191872643486e-05, | |
| "loss": 6.281, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 20.65354000837872, | |
| "grad_norm": 3.1495063304901123, | |
| "learning_rate": 1.5592096075967043e-05, | |
| "loss": 6.2666, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 20.695433598659406, | |
| "grad_norm": 2.9170730113983154, | |
| "learning_rate": 1.552227342549923e-05, | |
| "loss": 6.2738, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 20.737327188940093, | |
| "grad_norm": 3.0922224521636963, | |
| "learning_rate": 1.545245077503142e-05, | |
| "loss": 6.2805, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 20.77922077922078, | |
| "grad_norm": 3.088012933731079, | |
| "learning_rate": 1.538262812456361e-05, | |
| "loss": 6.2906, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 20.821114369501466, | |
| "grad_norm": 2.939486503601074, | |
| "learning_rate": 1.53128054740958e-05, | |
| "loss": 6.2636, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 20.863007959782152, | |
| "grad_norm": 3.597949743270874, | |
| "learning_rate": 1.5242982823627986e-05, | |
| "loss": 6.2745, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 20.90490155006284, | |
| "grad_norm": 3.4760777950286865, | |
| "learning_rate": 1.5173160173160175e-05, | |
| "loss": 6.2702, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 20.94679514034353, | |
| "grad_norm": 3.04856014251709, | |
| "learning_rate": 1.5103337522692362e-05, | |
| "loss": 6.2841, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 20.988688730624215, | |
| "grad_norm": 2.849895477294922, | |
| "learning_rate": 1.503351487222455e-05, | |
| "loss": 6.2814, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 21.0305823209049, | |
| "grad_norm": 3.1246280670166016, | |
| "learning_rate": 1.496369222175674e-05, | |
| "loss": 6.2754, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 21.072475911185588, | |
| "grad_norm": 3.303846836090088, | |
| "learning_rate": 1.4894567797793605e-05, | |
| "loss": 6.2661, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 21.114369501466275, | |
| "grad_norm": 3.5818755626678467, | |
| "learning_rate": 1.4824745147325794e-05, | |
| "loss": 6.2804, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 21.15626309174696, | |
| "grad_norm": 3.0695786476135254, | |
| "learning_rate": 1.4754922496857982e-05, | |
| "loss": 6.284, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 21.19815668202765, | |
| "grad_norm": 3.6067614555358887, | |
| "learning_rate": 1.4685099846390168e-05, | |
| "loss": 6.2863, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 21.240050272308338, | |
| "grad_norm": 3.2230417728424072, | |
| "learning_rate": 1.4615277195922359e-05, | |
| "loss": 6.287, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 21.281943862589024, | |
| "grad_norm": 3.059466600418091, | |
| "learning_rate": 1.4545454545454545e-05, | |
| "loss": 6.2442, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 21.32383745286971, | |
| "grad_norm": 3.7770040035247803, | |
| "learning_rate": 1.4475631894986736e-05, | |
| "loss": 6.2612, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 21.365731043150397, | |
| "grad_norm": 3.3269879817962646, | |
| "learning_rate": 1.4405809244518923e-05, | |
| "loss": 6.2985, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 21.407624633431084, | |
| "grad_norm": 2.649940252304077, | |
| "learning_rate": 1.4335986594051109e-05, | |
| "loss": 6.2343, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 21.44951822371177, | |
| "grad_norm": 3.4042983055114746, | |
| "learning_rate": 1.42661639435833e-05, | |
| "loss": 6.2701, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 21.49141181399246, | |
| "grad_norm": 3.1958000659942627, | |
| "learning_rate": 1.4196341293115486e-05, | |
| "loss": 6.2866, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 21.533305404273147, | |
| "grad_norm": 3.6010313034057617, | |
| "learning_rate": 1.4126518642647677e-05, | |
| "loss": 6.2683, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 21.575198994553833, | |
| "grad_norm": 3.429414749145508, | |
| "learning_rate": 1.4056695992179864e-05, | |
| "loss": 6.2408, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 21.61709258483452, | |
| "grad_norm": 3.069561004638672, | |
| "learning_rate": 1.3986873341712054e-05, | |
| "loss": 6.2641, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 21.658986175115206, | |
| "grad_norm": 3.575247287750244, | |
| "learning_rate": 1.3917050691244241e-05, | |
| "loss": 6.2722, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 21.700879765395893, | |
| "grad_norm": 3.033505439758301, | |
| "learning_rate": 1.3847228040776427e-05, | |
| "loss": 6.2424, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 21.742773355676583, | |
| "grad_norm": 3.287740707397461, | |
| "learning_rate": 1.3777405390308618e-05, | |
| "loss": 6.2516, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 21.78466694595727, | |
| "grad_norm": 3.0363028049468994, | |
| "learning_rate": 1.3707582739840804e-05, | |
| "loss": 6.2641, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 21.826560536237956, | |
| "grad_norm": 3.1549689769744873, | |
| "learning_rate": 1.3637760089372995e-05, | |
| "loss": 6.2335, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 21.868454126518643, | |
| "grad_norm": 3.8512282371520996, | |
| "learning_rate": 1.3567937438905182e-05, | |
| "loss": 6.2729, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 21.91034771679933, | |
| "grad_norm": 4.0751824378967285, | |
| "learning_rate": 1.3498813014942047e-05, | |
| "loss": 6.2397, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 21.952241307080016, | |
| "grad_norm": 3.375235080718994, | |
| "learning_rate": 1.3428990364474236e-05, | |
| "loss": 6.2316, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 21.994134897360702, | |
| "grad_norm": 3.093156337738037, | |
| "learning_rate": 1.3359167714006423e-05, | |
| "loss": 6.2468, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 22.036028487641392, | |
| "grad_norm": 3.729182243347168, | |
| "learning_rate": 1.3289345063538614e-05, | |
| "loss": 6.2366, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 22.07792207792208, | |
| "grad_norm": 3.4075732231140137, | |
| "learning_rate": 1.32195224130708e-05, | |
| "loss": 6.2693, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 22.119815668202765, | |
| "grad_norm": 2.9553005695343018, | |
| "learning_rate": 1.3149699762602988e-05, | |
| "loss": 6.2592, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 22.16170925848345, | |
| "grad_norm": 3.094538688659668, | |
| "learning_rate": 1.3079877112135177e-05, | |
| "loss": 6.26, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 22.203602848764138, | |
| "grad_norm": 3.907914161682129, | |
| "learning_rate": 1.3010054461667365e-05, | |
| "loss": 6.2711, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 22.245496439044825, | |
| "grad_norm": 3.7182159423828125, | |
| "learning_rate": 1.2940231811199554e-05, | |
| "loss": 6.2713, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 22.287390029325515, | |
| "grad_norm": 2.8652303218841553, | |
| "learning_rate": 1.2870409160731741e-05, | |
| "loss": 6.2325, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 22.3292836196062, | |
| "grad_norm": 3.190359592437744, | |
| "learning_rate": 1.2800586510263929e-05, | |
| "loss": 6.2563, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 22.371177209886888, | |
| "grad_norm": 3.372394561767578, | |
| "learning_rate": 1.2730763859796118e-05, | |
| "loss": 6.2489, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 22.413070800167574, | |
| "grad_norm": 3.340397596359253, | |
| "learning_rate": 1.2660941209328306e-05, | |
| "loss": 6.2147, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 22.45496439044826, | |
| "grad_norm": 3.1127400398254395, | |
| "learning_rate": 1.2591118558860495e-05, | |
| "loss": 6.2588, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 22.496857980728947, | |
| "grad_norm": 4.315746307373047, | |
| "learning_rate": 1.2521295908392683e-05, | |
| "loss": 6.2641, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 22.538751571009634, | |
| "grad_norm": 3.204827070236206, | |
| "learning_rate": 1.2451473257924872e-05, | |
| "loss": 6.2506, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 22.580645161290324, | |
| "grad_norm": 3.653074026107788, | |
| "learning_rate": 1.238165060745706e-05, | |
| "loss": 6.2512, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 22.62253875157101, | |
| "grad_norm": 3.8693697452545166, | |
| "learning_rate": 1.2311827956989249e-05, | |
| "loss": 6.2515, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 22.664432341851697, | |
| "grad_norm": 3.9418985843658447, | |
| "learning_rate": 1.2242005306521436e-05, | |
| "loss": 6.2522, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 22.706325932132383, | |
| "grad_norm": 3.328951358795166, | |
| "learning_rate": 1.2172182656053624e-05, | |
| "loss": 6.2244, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 22.74821952241307, | |
| "grad_norm": 3.251552104949951, | |
| "learning_rate": 1.210305823209049e-05, | |
| "loss": 6.2413, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 22.790113112693756, | |
| "grad_norm": 3.0756313800811768, | |
| "learning_rate": 1.2033235581622679e-05, | |
| "loss": 6.2343, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 22.832006702974446, | |
| "grad_norm": 3.174830913543701, | |
| "learning_rate": 1.1963412931154867e-05, | |
| "loss": 6.2445, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 22.873900293255133, | |
| "grad_norm": 2.831454038619995, | |
| "learning_rate": 1.1893590280687056e-05, | |
| "loss": 6.2457, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 22.91579388353582, | |
| "grad_norm": 3.3783247470855713, | |
| "learning_rate": 1.1823767630219245e-05, | |
| "loss": 6.2202, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 22.957687473816506, | |
| "grad_norm": 3.4505226612091064, | |
| "learning_rate": 1.1753944979751431e-05, | |
| "loss": 6.2329, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 22.999581064097192, | |
| "grad_norm": 4.203530311584473, | |
| "learning_rate": 1.168412232928362e-05, | |
| "loss": 6.2464, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 23.04147465437788, | |
| "grad_norm": 3.295198678970337, | |
| "learning_rate": 1.1614299678815808e-05, | |
| "loss": 6.2163, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 23.083368244658566, | |
| "grad_norm": 3.6795082092285156, | |
| "learning_rate": 1.1544477028347997e-05, | |
| "loss": 6.2108, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 23.125261834939256, | |
| "grad_norm": 3.7577404975891113, | |
| "learning_rate": 1.1474654377880186e-05, | |
| "loss": 6.2406, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 23.167155425219942, | |
| "grad_norm": 4.524641036987305, | |
| "learning_rate": 1.1404831727412372e-05, | |
| "loss": 6.2449, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 23.20904901550063, | |
| "grad_norm": 3.3049490451812744, | |
| "learning_rate": 1.1335009076944561e-05, | |
| "loss": 6.202, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 23.250942605781315, | |
| "grad_norm": 3.6244115829467773, | |
| "learning_rate": 1.1265186426476749e-05, | |
| "loss": 6.2214, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 23.292836196062, | |
| "grad_norm": 3.1158556938171387, | |
| "learning_rate": 1.1195363776008938e-05, | |
| "loss": 6.2247, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 23.334729786342688, | |
| "grad_norm": 3.208771228790283, | |
| "learning_rate": 1.1125541125541126e-05, | |
| "loss": 6.2416, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 23.376623376623378, | |
| "grad_norm": 4.181106090545654, | |
| "learning_rate": 1.1055718475073313e-05, | |
| "loss": 6.2343, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 23.418516966904065, | |
| "grad_norm": 2.8972866535186768, | |
| "learning_rate": 1.0985895824605503e-05, | |
| "loss": 6.2186, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 23.46041055718475, | |
| "grad_norm": 3.1691384315490723, | |
| "learning_rate": 1.091607317413769e-05, | |
| "loss": 6.2328, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 23.502304147465438, | |
| "grad_norm": 3.214346408843994, | |
| "learning_rate": 1.084625052366988e-05, | |
| "loss": 6.2356, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 23.544197737746124, | |
| "grad_norm": 3.0547690391540527, | |
| "learning_rate": 1.0776427873202067e-05, | |
| "loss": 6.2245, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 23.58609132802681, | |
| "grad_norm": 3.6090760231018066, | |
| "learning_rate": 1.0707303449238935e-05, | |
| "loss": 6.2634, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 23.627984918307497, | |
| "grad_norm": 3.210068702697754, | |
| "learning_rate": 1.0637480798771122e-05, | |
| "loss": 6.2126, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 23.669878508588187, | |
| "grad_norm": 3.872507095336914, | |
| "learning_rate": 1.056765814830331e-05, | |
| "loss": 6.2286, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 23.711772098868874, | |
| "grad_norm": 4.503695011138916, | |
| "learning_rate": 1.0497835497835499e-05, | |
| "loss": 6.2156, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 23.75366568914956, | |
| "grad_norm": 3.963315486907959, | |
| "learning_rate": 1.0428012847367686e-05, | |
| "loss": 6.2247, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 23.795559279430247, | |
| "grad_norm": 3.4394917488098145, | |
| "learning_rate": 1.0358190196899876e-05, | |
| "loss": 6.234, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 23.837452869710933, | |
| "grad_norm": 3.403167724609375, | |
| "learning_rate": 1.0288367546432063e-05, | |
| "loss": 6.2045, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 23.87934645999162, | |
| "grad_norm": 2.8274378776550293, | |
| "learning_rate": 1.0218544895964251e-05, | |
| "loss": 6.2121, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 23.92124005027231, | |
| "grad_norm": 3.277188301086426, | |
| "learning_rate": 1.0148722245496438e-05, | |
| "loss": 6.222, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 23.963133640552996, | |
| "grad_norm": 3.0735063552856445, | |
| "learning_rate": 1.0078899595028628e-05, | |
| "loss": 6.2257, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 24.005027230833683, | |
| "grad_norm": 3.6680026054382324, | |
| "learning_rate": 1.0009076944560817e-05, | |
| "loss": 6.2131, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 24.04692082111437, | |
| "grad_norm": 3.134713888168335, | |
| "learning_rate": 9.939254294093005e-06, | |
| "loss": 6.2241, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 24.088814411395056, | |
| "grad_norm": 2.9466712474823, | |
| "learning_rate": 9.869431643625192e-06, | |
| "loss": 6.2158, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 24.130708001675742, | |
| "grad_norm": 3.468949794769287, | |
| "learning_rate": 9.79960899315738e-06, | |
| "loss": 6.1793, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 24.17260159195643, | |
| "grad_norm": 3.5487060546875, | |
| "learning_rate": 9.729786342689569e-06, | |
| "loss": 6.2218, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 24.21449518223712, | |
| "grad_norm": 4.345893383026123, | |
| "learning_rate": 9.659963692221758e-06, | |
| "loss": 6.2023, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 24.256388772517806, | |
| "grad_norm": 2.9016401767730713, | |
| "learning_rate": 9.590141041753946e-06, | |
| "loss": 6.23, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 24.298282362798492, | |
| "grad_norm": 4.17023229598999, | |
| "learning_rate": 9.520318391286135e-06, | |
| "loss": 6.2114, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 24.34017595307918, | |
| "grad_norm": 3.322115421295166, | |
| "learning_rate": 9.45049574081832e-06, | |
| "loss": 6.204, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 24.382069543359865, | |
| "grad_norm": 3.709805488586426, | |
| "learning_rate": 9.38067309035051e-06, | |
| "loss": 6.2087, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 24.42396313364055, | |
| "grad_norm": 3.225588798522949, | |
| "learning_rate": 9.311548666387376e-06, | |
| "loss": 6.2436, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 24.46585672392124, | |
| "grad_norm": 3.1229472160339355, | |
| "learning_rate": 9.241726015919565e-06, | |
| "loss": 6.2253, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 24.507750314201928, | |
| "grad_norm": 3.4445230960845947, | |
| "learning_rate": 9.171903365451753e-06, | |
| "loss": 6.2254, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 24.549643904482615, | |
| "grad_norm": 4.2796807289123535, | |
| "learning_rate": 9.102080714983942e-06, | |
| "loss": 6.221, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 24.5915374947633, | |
| "grad_norm": 3.2323966026306152, | |
| "learning_rate": 9.03225806451613e-06, | |
| "loss": 6.228, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 24.633431085043988, | |
| "grad_norm": 4.064596652984619, | |
| "learning_rate": 8.962435414048317e-06, | |
| "loss": 6.2363, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 24.675324675324674, | |
| "grad_norm": 3.068544864654541, | |
| "learning_rate": 8.893310990085183e-06, | |
| "loss": 6.2508, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 24.71721826560536, | |
| "grad_norm": 2.6201155185699463, | |
| "learning_rate": 8.823488339617372e-06, | |
| "loss": 6.2193, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 24.75911185588605, | |
| "grad_norm": 4.960629463195801, | |
| "learning_rate": 8.753665689149562e-06, | |
| "loss": 6.1999, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 24.801005446166737, | |
| "grad_norm": 3.191586971282959, | |
| "learning_rate": 8.683843038681749e-06, | |
| "loss": 6.2203, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 24.842899036447424, | |
| "grad_norm": 3.224745512008667, | |
| "learning_rate": 8.614020388213937e-06, | |
| "loss": 6.212, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 24.88479262672811, | |
| "grad_norm": 3.450741767883301, | |
| "learning_rate": 8.544197737746124e-06, | |
| "loss": 6.2386, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 24.926686217008797, | |
| "grad_norm": 4.297729969024658, | |
| "learning_rate": 8.474375087278313e-06, | |
| "loss": 6.2088, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 24.968579807289483, | |
| "grad_norm": 3.376110553741455, | |
| "learning_rate": 8.404552436810503e-06, | |
| "loss": 6.2176, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 25.010473397570173, | |
| "grad_norm": 3.0211358070373535, | |
| "learning_rate": 8.33472978634269e-06, | |
| "loss": 6.1906, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 25.05236698785086, | |
| "grad_norm": 2.8490803241729736, | |
| "learning_rate": 8.264907135874878e-06, | |
| "loss": 6.2, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 25.094260578131546, | |
| "grad_norm": 3.0233705043792725, | |
| "learning_rate": 8.195084485407065e-06, | |
| "loss": 6.1886, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 25.136154168412233, | |
| "grad_norm": 3.7582995891571045, | |
| "learning_rate": 8.125261834939255e-06, | |
| "loss": 6.2064, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 25.17804775869292, | |
| "grad_norm": 3.128079891204834, | |
| "learning_rate": 8.055439184471442e-06, | |
| "loss": 6.2264, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 25.219941348973606, | |
| "grad_norm": 3.1808972358703613, | |
| "learning_rate": 7.985616534003632e-06, | |
| "loss": 6.2149, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 25.261834939254292, | |
| "grad_norm": 3.2326996326446533, | |
| "learning_rate": 7.91579388353582e-06, | |
| "loss": 6.2142, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 25.303728529534983, | |
| "grad_norm": 3.267465114593506, | |
| "learning_rate": 7.845971233068007e-06, | |
| "loss": 6.2439, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 25.34562211981567, | |
| "grad_norm": 3.691075563430786, | |
| "learning_rate": 7.776148582600196e-06, | |
| "loss": 6.2178, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 25.387515710096356, | |
| "grad_norm": 3.290562152862549, | |
| "learning_rate": 7.706325932132383e-06, | |
| "loss": 6.2165, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 25.429409300377042, | |
| "grad_norm": 4.553886413574219, | |
| "learning_rate": 7.636503281664573e-06, | |
| "loss": 6.2165, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 25.47130289065773, | |
| "grad_norm": 4.013444423675537, | |
| "learning_rate": 7.566680631196761e-06, | |
| "loss": 6.2122, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 25.513196480938415, | |
| "grad_norm": 4.044810771942139, | |
| "learning_rate": 7.496857980728948e-06, | |
| "loss": 6.2533, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 25.555090071219105, | |
| "grad_norm": 3.788613796234131, | |
| "learning_rate": 7.427035330261137e-06, | |
| "loss": 6.2039, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 25.59698366149979, | |
| "grad_norm": 3.317281484603882, | |
| "learning_rate": 7.3572126797933255e-06, | |
| "loss": 6.2228, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 25.638877251780478, | |
| "grad_norm": 3.4238085746765137, | |
| "learning_rate": 7.287390029325514e-06, | |
| "loss": 6.1979, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 25.680770842061165, | |
| "grad_norm": 3.1558725833892822, | |
| "learning_rate": 7.217567378857702e-06, | |
| "loss": 6.2044, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 25.72266443234185, | |
| "grad_norm": 2.939328670501709, | |
| "learning_rate": 7.147744728389889e-06, | |
| "loss": 6.2312, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 25.764558022622538, | |
| "grad_norm": 4.0037455558776855, | |
| "learning_rate": 7.0779220779220775e-06, | |
| "loss": 6.228, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 25.806451612903224, | |
| "grad_norm": 4.4582343101501465, | |
| "learning_rate": 7.008099427454267e-06, | |
| "loss": 6.2132, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 25.848345203183914, | |
| "grad_norm": 3.006201982498169, | |
| "learning_rate": 6.938276776986455e-06, | |
| "loss": 6.2242, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 25.8902387934646, | |
| "grad_norm": 3.6898059844970703, | |
| "learning_rate": 6.8684541265186436e-06, | |
| "loss": 6.2134, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 25.932132383745287, | |
| "grad_norm": 3.3489785194396973, | |
| "learning_rate": 6.798631476050832e-06, | |
| "loss": 6.2042, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 25.974025974025974, | |
| "grad_norm": 3.2489922046661377, | |
| "learning_rate": 6.729507052087698e-06, | |
| "loss": 6.2212, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 26.01591956430666, | |
| "grad_norm": 4.022356033325195, | |
| "learning_rate": 6.659684401619885e-06, | |
| "loss": 6.2423, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 26.057813154587347, | |
| "grad_norm": 4.803937911987305, | |
| "learning_rate": 6.589861751152074e-06, | |
| "loss": 6.2319, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 26.099706744868037, | |
| "grad_norm": 3.7283337116241455, | |
| "learning_rate": 6.520039100684262e-06, | |
| "loss": 6.1924, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 26.141600335148723, | |
| "grad_norm": 3.817946672439575, | |
| "learning_rate": 6.450216450216451e-06, | |
| "loss": 6.2039, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 26.18349392542941, | |
| "grad_norm": 3.4621963500976562, | |
| "learning_rate": 6.380393799748639e-06, | |
| "loss": 6.214, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 26.225387515710096, | |
| "grad_norm": 4.458475112915039, | |
| "learning_rate": 6.310571149280827e-06, | |
| "loss": 6.2327, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 26.267281105990783, | |
| "grad_norm": 3.1324493885040283, | |
| "learning_rate": 6.240748498813015e-06, | |
| "loss": 6.2518, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 26.30917469627147, | |
| "grad_norm": 3.410626173019409, | |
| "learning_rate": 6.1709258483452034e-06, | |
| "loss": 6.2054, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 26.351068286552156, | |
| "grad_norm": 3.221602201461792, | |
| "learning_rate": 6.101103197877392e-06, | |
| "loss": 6.2297, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 26.392961876832846, | |
| "grad_norm": 3.1413893699645996, | |
| "learning_rate": 6.031280547409579e-06, | |
| "loss": 6.2134, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 26.434855467113533, | |
| "grad_norm": 3.3834433555603027, | |
| "learning_rate": 5.961457896941768e-06, | |
| "loss": 6.167, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 26.47674905739422, | |
| "grad_norm": 3.016921281814575, | |
| "learning_rate": 5.891635246473957e-06, | |
| "loss": 6.2181, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 26.518642647674906, | |
| "grad_norm": 3.4190244674682617, | |
| "learning_rate": 5.821812596006145e-06, | |
| "loss": 6.172, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 26.560536237955592, | |
| "grad_norm": 3.519742488861084, | |
| "learning_rate": 5.751989945538333e-06, | |
| "loss": 6.2144, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 26.60242982823628, | |
| "grad_norm": 3.083923101425171, | |
| "learning_rate": 5.682167295070521e-06, | |
| "loss": 6.204, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 26.64432341851697, | |
| "grad_norm": 3.8977878093719482, | |
| "learning_rate": 5.612344644602709e-06, | |
| "loss": 6.1759, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 26.686217008797655, | |
| "grad_norm": 3.5598249435424805, | |
| "learning_rate": 5.5425219941348974e-06, | |
| "loss": 6.2233, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 26.72811059907834, | |
| "grad_norm": 3.6333513259887695, | |
| "learning_rate": 5.472699343667086e-06, | |
| "loss": 6.2133, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 26.770004189359028, | |
| "grad_norm": 3.2468085289001465, | |
| "learning_rate": 5.402876693199274e-06, | |
| "loss": 6.2081, | |
| "step": 63900 | |
| }, | |
| { | |
| "epoch": 26.811897779639715, | |
| "grad_norm": 3.6896772384643555, | |
| "learning_rate": 5.333054042731463e-06, | |
| "loss": 6.1935, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 26.8537913699204, | |
| "grad_norm": 3.263144016265869, | |
| "learning_rate": 5.26323139226365e-06, | |
| "loss": 6.2127, | |
| "step": 64100 | |
| }, | |
| { | |
| "epoch": 26.895684960201088, | |
| "grad_norm": 3.2848362922668457, | |
| "learning_rate": 5.193408741795839e-06, | |
| "loss": 6.2074, | |
| "step": 64200 | |
| }, | |
| { | |
| "epoch": 26.937578550481778, | |
| "grad_norm": 3.675541639328003, | |
| "learning_rate": 5.123586091328027e-06, | |
| "loss": 6.2015, | |
| "step": 64300 | |
| }, | |
| { | |
| "epoch": 26.979472140762464, | |
| "grad_norm": 3.413780689239502, | |
| "learning_rate": 5.0537634408602155e-06, | |
| "loss": 6.2218, | |
| "step": 64400 | |
| }, | |
| { | |
| "epoch": 27.02136573104315, | |
| "grad_norm": 4.108157634735107, | |
| "learning_rate": 4.983940790392404e-06, | |
| "loss": 6.212, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 27.063259321323837, | |
| "grad_norm": 3.7690155506134033, | |
| "learning_rate": 4.9141181399245915e-06, | |
| "loss": 6.22, | |
| "step": 64600 | |
| }, | |
| { | |
| "epoch": 27.105152911604524, | |
| "grad_norm": 3.379786491394043, | |
| "learning_rate": 4.84429548945678e-06, | |
| "loss": 6.2334, | |
| "step": 64700 | |
| }, | |
| { | |
| "epoch": 27.14704650188521, | |
| "grad_norm": 3.5175390243530273, | |
| "learning_rate": 4.774472838988968e-06, | |
| "loss": 6.1932, | |
| "step": 64800 | |
| }, | |
| { | |
| "epoch": 27.1889400921659, | |
| "grad_norm": 2.8454129695892334, | |
| "learning_rate": 4.704650188521157e-06, | |
| "loss": 6.2087, | |
| "step": 64900 | |
| }, | |
| { | |
| "epoch": 27.230833682446587, | |
| "grad_norm": 3.4630961418151855, | |
| "learning_rate": 4.634827538053345e-06, | |
| "loss": 6.2142, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 27.272727272727273, | |
| "grad_norm": 3.339860677719116, | |
| "learning_rate": 4.565004887585533e-06, | |
| "loss": 6.1772, | |
| "step": 65100 | |
| }, | |
| { | |
| "epoch": 27.31462086300796, | |
| "grad_norm": 3.0743260383605957, | |
| "learning_rate": 4.495182237117721e-06, | |
| "loss": 6.2044, | |
| "step": 65200 | |
| }, | |
| { | |
| "epoch": 27.356514453288646, | |
| "grad_norm": 3.2576496601104736, | |
| "learning_rate": 4.4253595866499095e-06, | |
| "loss": 6.2191, | |
| "step": 65300 | |
| }, | |
| { | |
| "epoch": 27.398408043569333, | |
| "grad_norm": 3.326819896697998, | |
| "learning_rate": 4.355536936182097e-06, | |
| "loss": 6.1762, | |
| "step": 65400 | |
| }, | |
| { | |
| "epoch": 27.44030163385002, | |
| "grad_norm": 3.4447667598724365, | |
| "learning_rate": 4.285714285714286e-06, | |
| "loss": 6.1823, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 27.48219522413071, | |
| "grad_norm": 3.4771687984466553, | |
| "learning_rate": 4.215891635246475e-06, | |
| "loss": 6.228, | |
| "step": 65600 | |
| }, | |
| { | |
| "epoch": 27.524088814411396, | |
| "grad_norm": 3.3457424640655518, | |
| "learning_rate": 4.146068984778662e-06, | |
| "loss": 6.1651, | |
| "step": 65700 | |
| }, | |
| { | |
| "epoch": 27.565982404692082, | |
| "grad_norm": 3.006155490875244, | |
| "learning_rate": 4.076246334310851e-06, | |
| "loss": 6.2026, | |
| "step": 65800 | |
| }, | |
| { | |
| "epoch": 27.60787599497277, | |
| "grad_norm": 4.228708744049072, | |
| "learning_rate": 4.006423683843038e-06, | |
| "loss": 6.1923, | |
| "step": 65900 | |
| }, | |
| { | |
| "epoch": 27.649769585253456, | |
| "grad_norm": 3.4744226932525635, | |
| "learning_rate": 3.937299259879905e-06, | |
| "loss": 6.1891, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 27.691663175534142, | |
| "grad_norm": 3.8300633430480957, | |
| "learning_rate": 3.867476609412093e-06, | |
| "loss": 6.2237, | |
| "step": 66100 | |
| }, | |
| { | |
| "epoch": 27.733556765814832, | |
| "grad_norm": 2.9689528942108154, | |
| "learning_rate": 3.7976539589442818e-06, | |
| "loss": 6.217, | |
| "step": 66200 | |
| }, | |
| { | |
| "epoch": 27.77545035609552, | |
| "grad_norm": 3.1309947967529297, | |
| "learning_rate": 3.7278313084764698e-06, | |
| "loss": 6.2061, | |
| "step": 66300 | |
| }, | |
| { | |
| "epoch": 27.817343946376205, | |
| "grad_norm": 3.4571166038513184, | |
| "learning_rate": 3.658008658008658e-06, | |
| "loss": 6.1863, | |
| "step": 66400 | |
| }, | |
| { | |
| "epoch": 27.85923753665689, | |
| "grad_norm": 3.354229211807251, | |
| "learning_rate": 3.5881860075408466e-06, | |
| "loss": 6.1996, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 27.901131126937578, | |
| "grad_norm": 3.745568037033081, | |
| "learning_rate": 3.5183633570730346e-06, | |
| "loss": 6.1839, | |
| "step": 66600 | |
| }, | |
| { | |
| "epoch": 27.943024717218265, | |
| "grad_norm": 3.356715440750122, | |
| "learning_rate": 3.448540706605223e-06, | |
| "loss": 6.2048, | |
| "step": 66700 | |
| }, | |
| { | |
| "epoch": 27.98491830749895, | |
| "grad_norm": 2.964492082595825, | |
| "learning_rate": 3.378718056137411e-06, | |
| "loss": 6.228, | |
| "step": 66800 | |
| }, | |
| { | |
| "epoch": 28.02681189777964, | |
| "grad_norm": 3.336606502532959, | |
| "learning_rate": 3.3088954056695994e-06, | |
| "loss": 6.1953, | |
| "step": 66900 | |
| }, | |
| { | |
| "epoch": 28.068705488060328, | |
| "grad_norm": 3.264971971511841, | |
| "learning_rate": 3.239072755201788e-06, | |
| "loss": 6.1783, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 28.110599078341014, | |
| "grad_norm": 3.4968082904815674, | |
| "learning_rate": 3.169250104733976e-06, | |
| "loss": 6.2117, | |
| "step": 67100 | |
| }, | |
| { | |
| "epoch": 28.1524926686217, | |
| "grad_norm": 3.4082252979278564, | |
| "learning_rate": 3.099427454266164e-06, | |
| "loss": 6.2278, | |
| "step": 67200 | |
| }, | |
| { | |
| "epoch": 28.194386258902387, | |
| "grad_norm": 3.52056884765625, | |
| "learning_rate": 3.029604803798352e-06, | |
| "loss": 6.2037, | |
| "step": 67300 | |
| }, | |
| { | |
| "epoch": 28.236279849183074, | |
| "grad_norm": 3.6062779426574707, | |
| "learning_rate": 2.9597821533305406e-06, | |
| "loss": 6.1952, | |
| "step": 67400 | |
| }, | |
| { | |
| "epoch": 28.278173439463764, | |
| "grad_norm": 3.158705472946167, | |
| "learning_rate": 2.889959502862729e-06, | |
| "loss": 6.2067, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 28.32006702974445, | |
| "grad_norm": 3.6732075214385986, | |
| "learning_rate": 2.820136852394917e-06, | |
| "loss": 6.1752, | |
| "step": 67600 | |
| }, | |
| { | |
| "epoch": 28.361960620025137, | |
| "grad_norm": 2.842560291290283, | |
| "learning_rate": 2.7503142019271054e-06, | |
| "loss": 6.1823, | |
| "step": 67700 | |
| }, | |
| { | |
| "epoch": 28.403854210305823, | |
| "grad_norm": 3.412233591079712, | |
| "learning_rate": 2.6804915514592934e-06, | |
| "loss": 6.1997, | |
| "step": 67800 | |
| }, | |
| { | |
| "epoch": 28.44574780058651, | |
| "grad_norm": 2.8313143253326416, | |
| "learning_rate": 2.610668900991482e-06, | |
| "loss": 6.187, | |
| "step": 67900 | |
| }, | |
| { | |
| "epoch": 28.487641390867196, | |
| "grad_norm": 3.122307300567627, | |
| "learning_rate": 2.541544477028348e-06, | |
| "loss": 6.1977, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 28.529534981147883, | |
| "grad_norm": 3.4732697010040283, | |
| "learning_rate": 2.4717218265605365e-06, | |
| "loss": 6.1913, | |
| "step": 68100 | |
| }, | |
| { | |
| "epoch": 28.571428571428573, | |
| "grad_norm": 3.3936917781829834, | |
| "learning_rate": 2.4018991760927245e-06, | |
| "loss": 6.2216, | |
| "step": 68200 | |
| }, | |
| { | |
| "epoch": 28.61332216170926, | |
| "grad_norm": 3.2980170249938965, | |
| "learning_rate": 2.332076525624913e-06, | |
| "loss": 6.1989, | |
| "step": 68300 | |
| }, | |
| { | |
| "epoch": 28.655215751989946, | |
| "grad_norm": 4.099823951721191, | |
| "learning_rate": 2.2622538751571013e-06, | |
| "loss": 6.1755, | |
| "step": 68400 | |
| }, | |
| { | |
| "epoch": 28.697109342270632, | |
| "grad_norm": 3.7930960655212402, | |
| "learning_rate": 2.1924312246892893e-06, | |
| "loss": 6.2155, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 28.73900293255132, | |
| "grad_norm": 3.620065212249756, | |
| "learning_rate": 2.1226085742214777e-06, | |
| "loss": 6.2041, | |
| "step": 68600 | |
| }, | |
| { | |
| "epoch": 28.780896522832006, | |
| "grad_norm": 3.3451735973358154, | |
| "learning_rate": 2.0527859237536657e-06, | |
| "loss": 6.2095, | |
| "step": 68700 | |
| }, | |
| { | |
| "epoch": 28.822790113112696, | |
| "grad_norm": 4.007857799530029, | |
| "learning_rate": 1.982963273285854e-06, | |
| "loss": 6.2283, | |
| "step": 68800 | |
| }, | |
| { | |
| "epoch": 28.864683703393382, | |
| "grad_norm": 4.236888885498047, | |
| "learning_rate": 1.9131406228180425e-06, | |
| "loss": 6.1799, | |
| "step": 68900 | |
| }, | |
| { | |
| "epoch": 28.90657729367407, | |
| "grad_norm": 3.222273111343384, | |
| "learning_rate": 1.8433179723502305e-06, | |
| "loss": 6.2008, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 28.948470883954755, | |
| "grad_norm": 3.8649580478668213, | |
| "learning_rate": 1.7734953218824187e-06, | |
| "loss": 6.194, | |
| "step": 69100 | |
| }, | |
| { | |
| "epoch": 28.99036447423544, | |
| "grad_norm": 3.9630191326141357, | |
| "learning_rate": 1.7036726714146071e-06, | |
| "loss": 6.2083, | |
| "step": 69200 | |
| }, | |
| { | |
| "epoch": 29.032258064516128, | |
| "grad_norm": 3.9617035388946533, | |
| "learning_rate": 1.6338500209467953e-06, | |
| "loss": 6.2197, | |
| "step": 69300 | |
| }, | |
| { | |
| "epoch": 29.074151654796815, | |
| "grad_norm": 3.4647514820098877, | |
| "learning_rate": 1.5640273704789835e-06, | |
| "loss": 6.1919, | |
| "step": 69400 | |
| }, | |
| { | |
| "epoch": 29.116045245077505, | |
| "grad_norm": 3.7548468112945557, | |
| "learning_rate": 1.4942047200111717e-06, | |
| "loss": 6.171, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 29.15793883535819, | |
| "grad_norm": 3.4267735481262207, | |
| "learning_rate": 1.42438206954336e-06, | |
| "loss": 6.2031, | |
| "step": 69600 | |
| }, | |
| { | |
| "epoch": 29.199832425638878, | |
| "grad_norm": 3.166888952255249, | |
| "learning_rate": 1.3545594190755483e-06, | |
| "loss": 6.1937, | |
| "step": 69700 | |
| }, | |
| { | |
| "epoch": 29.241726015919564, | |
| "grad_norm": 2.9794344902038574, | |
| "learning_rate": 1.2847367686077363e-06, | |
| "loss": 6.2068, | |
| "step": 69800 | |
| }, | |
| { | |
| "epoch": 29.28361960620025, | |
| "grad_norm": 3.056293249130249, | |
| "learning_rate": 1.2149141181399247e-06, | |
| "loss": 6.1652, | |
| "step": 69900 | |
| }, | |
| { | |
| "epoch": 29.325513196480937, | |
| "grad_norm": 3.851149320602417, | |
| "learning_rate": 1.145789694176791e-06, | |
| "loss": 6.2129, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 29.367406786761624, | |
| "grad_norm": 3.670929193496704, | |
| "learning_rate": 1.0759670437089792e-06, | |
| "loss": 6.2068, | |
| "step": 70100 | |
| }, | |
| { | |
| "epoch": 29.409300377042314, | |
| "grad_norm": 3.3581252098083496, | |
| "learning_rate": 1.0061443932411674e-06, | |
| "loss": 6.2208, | |
| "step": 70200 | |
| }, | |
| { | |
| "epoch": 29.451193967323, | |
| "grad_norm": 3.7551257610321045, | |
| "learning_rate": 9.363217427733557e-07, | |
| "loss": 6.179, | |
| "step": 70300 | |
| }, | |
| { | |
| "epoch": 29.493087557603687, | |
| "grad_norm": 2.9767682552337646, | |
| "learning_rate": 8.664990923055439e-07, | |
| "loss": 6.2157, | |
| "step": 70400 | |
| }, | |
| { | |
| "epoch": 29.534981147884373, | |
| "grad_norm": 3.3218774795532227, | |
| "learning_rate": 7.966764418377322e-07, | |
| "loss": 6.1773, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 29.57687473816506, | |
| "grad_norm": 4.360437870025635, | |
| "learning_rate": 7.268537913699204e-07, | |
| "loss": 6.1985, | |
| "step": 70600 | |
| }, | |
| { | |
| "epoch": 29.618768328445746, | |
| "grad_norm": 3.544264078140259, | |
| "learning_rate": 6.570311409021087e-07, | |
| "loss": 6.1973, | |
| "step": 70700 | |
| }, | |
| { | |
| "epoch": 29.660661918726436, | |
| "grad_norm": 3.7416069507598877, | |
| "learning_rate": 5.872084904342969e-07, | |
| "loss": 6.2031, | |
| "step": 70800 | |
| }, | |
| { | |
| "epoch": 29.702555509007123, | |
| "grad_norm": 3.0346035957336426, | |
| "learning_rate": 5.173858399664851e-07, | |
| "loss": 6.2123, | |
| "step": 70900 | |
| }, | |
| { | |
| "epoch": 29.74444909928781, | |
| "grad_norm": 3.2308425903320312, | |
| "learning_rate": 4.4756318949867344e-07, | |
| "loss": 6.2106, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 29.786342689568496, | |
| "grad_norm": 3.0109570026397705, | |
| "learning_rate": 3.7774053903086163e-07, | |
| "loss": 6.213, | |
| "step": 71100 | |
| }, | |
| { | |
| "epoch": 29.828236279849182, | |
| "grad_norm": 3.733609199523926, | |
| "learning_rate": 3.079178885630499e-07, | |
| "loss": 6.1984, | |
| "step": 71200 | |
| }, | |
| { | |
| "epoch": 29.87012987012987, | |
| "grad_norm": 3.5430541038513184, | |
| "learning_rate": 2.3809523809523814e-07, | |
| "loss": 6.183, | |
| "step": 71300 | |
| }, | |
| { | |
| "epoch": 29.912023460410555, | |
| "grad_norm": 3.1964950561523438, | |
| "learning_rate": 1.6827258762742634e-07, | |
| "loss": 6.1817, | |
| "step": 71400 | |
| }, | |
| { | |
| "epoch": 29.953917050691246, | |
| "grad_norm": 3.6197755336761475, | |
| "learning_rate": 9.844993715961458e-08, | |
| "loss": 6.1907, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 29.995810640971932, | |
| "grad_norm": 3.035473346710205, | |
| "learning_rate": 2.862728669180282e-08, | |
| "loss": 6.1677, | |
| "step": 71600 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "step": 71610, | |
| "total_flos": 781486986700800.0, | |
| "train_loss": 6.528281962779993, | |
| "train_runtime": 3948.1737, | |
| "train_samples_per_second": 580.354, | |
| "train_steps_per_second": 18.137 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 71610, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 5000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 781486986700800.0, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |