| { | |
| "best_global_step": 25817, | |
| "best_metric": 0.7780532248284215, | |
| "best_model_checkpoint": "./sbert_large_sentiment/checkpoint-25817", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 25817, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.007746833481814309, | |
| "grad_norm": 7.575475215911865, | |
| "learning_rate": 2.5690679060160085e-07, | |
| "loss": 1.0936, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.015493666963628618, | |
| "grad_norm": 5.179758548736572, | |
| "learning_rate": 5.151045701006972e-07, | |
| "loss": 1.0065, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.023240500445442926, | |
| "grad_norm": 5.142735958099365, | |
| "learning_rate": 7.733023495997935e-07, | |
| "loss": 0.9436, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.030987333927257235, | |
| "grad_norm": 6.526517868041992, | |
| "learning_rate": 1.0315001290988898e-06, | |
| "loss": 0.8957, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.03873416740907154, | |
| "grad_norm": 8.164371490478516, | |
| "learning_rate": 1.2896979085979861e-06, | |
| "loss": 0.7752, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.04648100089088585, | |
| "grad_norm": 9.465094566345215, | |
| "learning_rate": 1.5478956880970825e-06, | |
| "loss": 0.7205, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.05422783437270016, | |
| "grad_norm": 7.948523998260498, | |
| "learning_rate": 1.806093467596179e-06, | |
| "loss": 0.6522, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.06197466785451447, | |
| "grad_norm": 7.701857089996338, | |
| "learning_rate": 2.064291247095275e-06, | |
| "loss": 0.6041, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.06972150133632878, | |
| "grad_norm": 10.935823440551758, | |
| "learning_rate": 2.3224890265943715e-06, | |
| "loss": 0.5922, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.07746833481814308, | |
| "grad_norm": 6.648813247680664, | |
| "learning_rate": 2.5806868060934675e-06, | |
| "loss": 0.6016, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.08521516829995739, | |
| "grad_norm": 9.691781997680664, | |
| "learning_rate": 2.838884585592564e-06, | |
| "loss": 0.5578, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.0929620017817717, | |
| "grad_norm": 4.378147125244141, | |
| "learning_rate": 3.0970823650916602e-06, | |
| "loss": 0.5731, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.10070883526358601, | |
| "grad_norm": 10.45535945892334, | |
| "learning_rate": 3.3552801445907566e-06, | |
| "loss": 0.5728, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.10845566874540032, | |
| "grad_norm": 9.362472534179688, | |
| "learning_rate": 3.613477924089853e-06, | |
| "loss": 0.5498, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.11620250222721462, | |
| "grad_norm": 15.068540573120117, | |
| "learning_rate": 3.87167570358895e-06, | |
| "loss": 0.5497, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.12394933570902894, | |
| "grad_norm": 6.796774387359619, | |
| "learning_rate": 4.129873483088046e-06, | |
| "loss": 0.5392, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.13169616919084323, | |
| "grad_norm": 10.11823844909668, | |
| "learning_rate": 4.388071262587142e-06, | |
| "loss": 0.5499, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.13944300267265755, | |
| "grad_norm": 9.95432186126709, | |
| "learning_rate": 4.646269042086238e-06, | |
| "loss": 0.5475, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.14718983615447187, | |
| "grad_norm": 22.89798355102539, | |
| "learning_rate": 4.9044668215853345e-06, | |
| "loss": 0.5188, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.15493666963628616, | |
| "grad_norm": 3.961637496948242, | |
| "learning_rate": 5.162664601084431e-06, | |
| "loss": 0.5502, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.16268350311810048, | |
| "grad_norm": 3.6492996215820312, | |
| "learning_rate": 5.420862380583527e-06, | |
| "loss": 0.5352, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.17043033659991477, | |
| "grad_norm": 5.891486644744873, | |
| "learning_rate": 5.679060160082624e-06, | |
| "loss": 0.5126, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.1781771700817291, | |
| "grad_norm": 10.750714302062988, | |
| "learning_rate": 5.93725793958172e-06, | |
| "loss": 0.5269, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.1859240035635434, | |
| "grad_norm": 7.711459636688232, | |
| "learning_rate": 6.1954557190808165e-06, | |
| "loss": 0.5504, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.1936708370453577, | |
| "grad_norm": 11.42162036895752, | |
| "learning_rate": 6.453653498579913e-06, | |
| "loss": 0.4935, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.20141767052717202, | |
| "grad_norm": 8.477349281311035, | |
| "learning_rate": 6.711851278079009e-06, | |
| "loss": 0.5252, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.20916450400898634, | |
| "grad_norm": 8.771768569946289, | |
| "learning_rate": 6.970049057578106e-06, | |
| "loss": 0.4995, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.21691133749080063, | |
| "grad_norm": 12.874269485473633, | |
| "learning_rate": 7.228246837077202e-06, | |
| "loss": 0.5089, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.22465817097261495, | |
| "grad_norm": 10.64578628540039, | |
| "learning_rate": 7.4864446165762984e-06, | |
| "loss": 0.5227, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.23240500445442924, | |
| "grad_norm": 5.0879950523376465, | |
| "learning_rate": 7.744642396075395e-06, | |
| "loss": 0.4903, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.24015183793624356, | |
| "grad_norm": 7.7417402267456055, | |
| "learning_rate": 8.00284017557449e-06, | |
| "loss": 0.5321, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.24789867141805788, | |
| "grad_norm": 16.59149932861328, | |
| "learning_rate": 8.261037955073588e-06, | |
| "loss": 0.5125, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.2556455048998722, | |
| "grad_norm": 10.474128723144531, | |
| "learning_rate": 8.519235734572683e-06, | |
| "loss": 0.5244, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.26339233838168646, | |
| "grad_norm": 7.808370113372803, | |
| "learning_rate": 8.777433514071779e-06, | |
| "loss": 0.5206, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.2711391718635008, | |
| "grad_norm": 14.291008949279785, | |
| "learning_rate": 9.035631293570876e-06, | |
| "loss": 0.5635, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.2788860053453151, | |
| "grad_norm": 4.823723793029785, | |
| "learning_rate": 9.293829073069971e-06, | |
| "loss": 0.5186, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.2866328388271294, | |
| "grad_norm": 8.087597846984863, | |
| "learning_rate": 9.552026852569069e-06, | |
| "loss": 0.5106, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.29437967230894374, | |
| "grad_norm": 19.820837020874023, | |
| "learning_rate": 9.810224632068164e-06, | |
| "loss": 0.5069, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.302126505790758, | |
| "grad_norm": 7.311481475830078, | |
| "learning_rate": 9.992396528226095e-06, | |
| "loss": 0.5305, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.3098733392725723, | |
| "grad_norm": 8.929943084716797, | |
| "learning_rate": 9.963704181909475e-06, | |
| "loss": 0.5463, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.31762017275438664, | |
| "grad_norm": 9.268566131591797, | |
| "learning_rate": 9.935011835592857e-06, | |
| "loss": 0.5328, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.32536700623620096, | |
| "grad_norm": 9.394050598144531, | |
| "learning_rate": 9.906319489276236e-06, | |
| "loss": 0.5097, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.3331138397180153, | |
| "grad_norm": 6.343189716339111, | |
| "learning_rate": 9.877627142959616e-06, | |
| "loss": 0.5063, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.34086067319982954, | |
| "grad_norm": 8.755598068237305, | |
| "learning_rate": 9.848934796642996e-06, | |
| "loss": 0.5202, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.34860750668164386, | |
| "grad_norm": 13.985782623291016, | |
| "learning_rate": 9.820242450326375e-06, | |
| "loss": 0.5019, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.3563543401634582, | |
| "grad_norm": 7.063718318939209, | |
| "learning_rate": 9.791550104009757e-06, | |
| "loss": 0.4844, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.3641011736452725, | |
| "grad_norm": 15.77436351776123, | |
| "learning_rate": 9.762857757693137e-06, | |
| "loss": 0.5007, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.3718480071270868, | |
| "grad_norm": 8.492551803588867, | |
| "learning_rate": 9.734165411376516e-06, | |
| "loss": 0.4676, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.3795948406089011, | |
| "grad_norm": 13.73719596862793, | |
| "learning_rate": 9.705473065059896e-06, | |
| "loss": 0.5103, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.3873416740907154, | |
| "grad_norm": 3.9511256217956543, | |
| "learning_rate": 9.676780718743276e-06, | |
| "loss": 0.4936, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.3950885075725297, | |
| "grad_norm": 7.649303436279297, | |
| "learning_rate": 9.648088372426655e-06, | |
| "loss": 0.4997, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.40283534105434404, | |
| "grad_norm": 12.8671236038208, | |
| "learning_rate": 9.619396026110037e-06, | |
| "loss": 0.5093, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.41058217453615836, | |
| "grad_norm": 9.342642784118652, | |
| "learning_rate": 9.590703679793416e-06, | |
| "loss": 0.4908, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.4183290080179727, | |
| "grad_norm": 10.612739562988281, | |
| "learning_rate": 9.562011333476796e-06, | |
| "loss": 0.5141, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.42607584149978694, | |
| "grad_norm": 5.249755382537842, | |
| "learning_rate": 9.533318987160176e-06, | |
| "loss": 0.4738, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.43382267498160126, | |
| "grad_norm": 15.617774963378906, | |
| "learning_rate": 9.504626640843556e-06, | |
| "loss": 0.4782, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.4415695084634156, | |
| "grad_norm": 20.442564010620117, | |
| "learning_rate": 9.475934294526937e-06, | |
| "loss": 0.4772, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.4493163419452299, | |
| "grad_norm": 5.345826148986816, | |
| "learning_rate": 9.447241948210317e-06, | |
| "loss": 0.4764, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.4570631754270442, | |
| "grad_norm": 21.79194450378418, | |
| "learning_rate": 9.418549601893696e-06, | |
| "loss": 0.4892, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.4648100089088585, | |
| "grad_norm": 15.640652656555176, | |
| "learning_rate": 9.389857255577074e-06, | |
| "loss": 0.5199, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.4725568423906728, | |
| "grad_norm": 6.19819974899292, | |
| "learning_rate": 9.361164909260456e-06, | |
| "loss": 0.4474, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.4803036758724871, | |
| "grad_norm": 7.241565704345703, | |
| "learning_rate": 9.332472562943835e-06, | |
| "loss": 0.4672, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.48805050935430144, | |
| "grad_norm": 9.49963092803955, | |
| "learning_rate": 9.303780216627215e-06, | |
| "loss": 0.4701, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.49579734283611576, | |
| "grad_norm": 16.092571258544922, | |
| "learning_rate": 9.275087870310595e-06, | |
| "loss": 0.4628, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.50354417631793, | |
| "grad_norm": 10.93955135345459, | |
| "learning_rate": 9.246395523993975e-06, | |
| "loss": 0.474, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.5112910097997444, | |
| "grad_norm": 9.083719253540039, | |
| "learning_rate": 9.217703177677354e-06, | |
| "loss": 0.4771, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.5190378432815587, | |
| "grad_norm": 9.085740089416504, | |
| "learning_rate": 9.189010831360736e-06, | |
| "loss": 0.4803, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.5267846767633729, | |
| "grad_norm": 17.334964752197266, | |
| "learning_rate": 9.160318485044115e-06, | |
| "loss": 0.478, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.5345315102451873, | |
| "grad_norm": 4.976545810699463, | |
| "learning_rate": 9.131626138727495e-06, | |
| "loss": 0.5116, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.5422783437270016, | |
| "grad_norm": 8.329999923706055, | |
| "learning_rate": 9.102933792410875e-06, | |
| "loss": 0.4831, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.5500251772088159, | |
| "grad_norm": 10.104872703552246, | |
| "learning_rate": 9.074241446094255e-06, | |
| "loss": 0.4905, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.5577720106906302, | |
| "grad_norm": 8.581650733947754, | |
| "learning_rate": 9.045549099777636e-06, | |
| "loss": 0.4656, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.5655188441724445, | |
| "grad_norm": 8.831694602966309, | |
| "learning_rate": 9.016856753461016e-06, | |
| "loss": 0.455, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.5732656776542588, | |
| "grad_norm": 12.80400276184082, | |
| "learning_rate": 8.988164407144395e-06, | |
| "loss": 0.4556, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.5810125111360731, | |
| "grad_norm": 10.323575019836426, | |
| "learning_rate": 8.959472060827775e-06, | |
| "loss": 0.4952, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.5887593446178875, | |
| "grad_norm": 8.978657722473145, | |
| "learning_rate": 8.930779714511155e-06, | |
| "loss": 0.4774, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.5965061780997017, | |
| "grad_norm": 9.755346298217773, | |
| "learning_rate": 8.902087368194534e-06, | |
| "loss": 0.4607, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.604253011581516, | |
| "grad_norm": 4.952932357788086, | |
| "learning_rate": 8.873395021877916e-06, | |
| "loss": 0.4931, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.6119998450633304, | |
| "grad_norm": 14.65958309173584, | |
| "learning_rate": 8.844702675561296e-06, | |
| "loss": 0.4723, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.6197466785451446, | |
| "grad_norm": 12.291789054870605, | |
| "learning_rate": 8.816010329244675e-06, | |
| "loss": 0.4916, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.627493512026959, | |
| "grad_norm": 6.109698295593262, | |
| "learning_rate": 8.787317982928055e-06, | |
| "loss": 0.4749, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.6352403455087733, | |
| "grad_norm": 9.7838134765625, | |
| "learning_rate": 8.758625636611435e-06, | |
| "loss": 0.4621, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.6429871789905875, | |
| "grad_norm": 6.760039806365967, | |
| "learning_rate": 8.729933290294814e-06, | |
| "loss": 0.4806, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.6507340124724019, | |
| "grad_norm": 10.97524642944336, | |
| "learning_rate": 8.701240943978194e-06, | |
| "loss": 0.4591, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.6584808459542162, | |
| "grad_norm": 9.859977722167969, | |
| "learning_rate": 8.672548597661574e-06, | |
| "loss": 0.4742, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.6662276794360306, | |
| "grad_norm": 11.13088321685791, | |
| "learning_rate": 8.643856251344953e-06, | |
| "loss": 0.4593, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.6739745129178448, | |
| "grad_norm": 8.611045837402344, | |
| "learning_rate": 8.615163905028335e-06, | |
| "loss": 0.4711, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.6817213463996591, | |
| "grad_norm": 9.801880836486816, | |
| "learning_rate": 8.586471558711715e-06, | |
| "loss": 0.4639, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.6894681798814735, | |
| "grad_norm": 6.184256553649902, | |
| "learning_rate": 8.557779212395094e-06, | |
| "loss": 0.4921, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.6972150133632877, | |
| "grad_norm": 9.72716999053955, | |
| "learning_rate": 8.529086866078474e-06, | |
| "loss": 0.4829, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.7049618468451021, | |
| "grad_norm": 7.580606937408447, | |
| "learning_rate": 8.500394519761854e-06, | |
| "loss": 0.4638, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.7127086803269164, | |
| "grad_norm": 12.59068775177002, | |
| "learning_rate": 8.471702173445235e-06, | |
| "loss": 0.4818, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.7204555138087306, | |
| "grad_norm": 5.276298522949219, | |
| "learning_rate": 8.443009827128615e-06, | |
| "loss": 0.4736, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.728202347290545, | |
| "grad_norm": 17.732887268066406, | |
| "learning_rate": 8.414317480811995e-06, | |
| "loss": 0.4515, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.7359491807723593, | |
| "grad_norm": 6.198581695556641, | |
| "learning_rate": 8.385625134495374e-06, | |
| "loss": 0.4766, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.7436960142541736, | |
| "grad_norm": 5.835597038269043, | |
| "learning_rate": 8.356932788178754e-06, | |
| "loss": 0.4506, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.7514428477359879, | |
| "grad_norm": 10.405342102050781, | |
| "learning_rate": 8.328240441862134e-06, | |
| "loss": 0.4703, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.7591896812178022, | |
| "grad_norm": 10.531950950622559, | |
| "learning_rate": 8.299548095545515e-06, | |
| "loss": 0.471, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.7669365146996165, | |
| "grad_norm": 11.877711296081543, | |
| "learning_rate": 8.270855749228895e-06, | |
| "loss": 0.4678, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.7746833481814308, | |
| "grad_norm": 7.640944957733154, | |
| "learning_rate": 8.242163402912274e-06, | |
| "loss": 0.4662, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.7824301816632452, | |
| "grad_norm": 17.27083969116211, | |
| "learning_rate": 8.213471056595652e-06, | |
| "loss": 0.4871, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.7901770151450594, | |
| "grad_norm": 3.0645809173583984, | |
| "learning_rate": 8.184778710279034e-06, | |
| "loss": 0.4574, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.7979238486268737, | |
| "grad_norm": 4.7715935707092285, | |
| "learning_rate": 8.156086363962414e-06, | |
| "loss": 0.4433, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.8056706821086881, | |
| "grad_norm": 4.5974202156066895, | |
| "learning_rate": 8.127394017645793e-06, | |
| "loss": 0.4439, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.8134175155905023, | |
| "grad_norm": 13.890436172485352, | |
| "learning_rate": 8.098701671329173e-06, | |
| "loss": 0.4354, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.8211643490723167, | |
| "grad_norm": 27.995084762573242, | |
| "learning_rate": 8.070009325012553e-06, | |
| "loss": 0.4115, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.828911182554131, | |
| "grad_norm": 2.1986517906188965, | |
| "learning_rate": 8.041316978695934e-06, | |
| "loss": 0.4243, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.8366580160359454, | |
| "grad_norm": 17.675518035888672, | |
| "learning_rate": 8.012624632379314e-06, | |
| "loss": 0.495, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.8444048495177596, | |
| "grad_norm": 7.1354146003723145, | |
| "learning_rate": 7.983932286062693e-06, | |
| "loss": 0.4688, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.8521516829995739, | |
| "grad_norm": 5.76608943939209, | |
| "learning_rate": 7.955239939746073e-06, | |
| "loss": 0.4364, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.8598985164813883, | |
| "grad_norm": 15.63326358795166, | |
| "learning_rate": 7.926547593429453e-06, | |
| "loss": 0.4496, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.8676453499632025, | |
| "grad_norm": 10.311728477478027, | |
| "learning_rate": 7.897855247112833e-06, | |
| "loss": 0.4649, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.8753921834450169, | |
| "grad_norm": 5.432675838470459, | |
| "learning_rate": 7.869162900796214e-06, | |
| "loss": 0.4307, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.8831390169268312, | |
| "grad_norm": 6.74233341217041, | |
| "learning_rate": 7.840470554479594e-06, | |
| "loss": 0.4629, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.8908858504086454, | |
| "grad_norm": 9.045225143432617, | |
| "learning_rate": 7.811778208162973e-06, | |
| "loss": 0.4742, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.8986326838904598, | |
| "grad_norm": 16.038272857666016, | |
| "learning_rate": 7.783085861846353e-06, | |
| "loss": 0.4335, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.9063795173722741, | |
| "grad_norm": 11.987189292907715, | |
| "learning_rate": 7.754393515529733e-06, | |
| "loss": 0.4534, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.9141263508540884, | |
| "grad_norm": 2.1488113403320312, | |
| "learning_rate": 7.725701169213114e-06, | |
| "loss": 0.4566, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.9218731843359027, | |
| "grad_norm": 0.6424712538719177, | |
| "learning_rate": 7.697008822896494e-06, | |
| "loss": 0.4397, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.929620017817717, | |
| "grad_norm": 9.486328125, | |
| "learning_rate": 7.668316476579874e-06, | |
| "loss": 0.4564, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.9373668512995313, | |
| "grad_norm": 9.135612487792969, | |
| "learning_rate": 7.639624130263253e-06, | |
| "loss": 0.4608, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.9451136847813456, | |
| "grad_norm": 9.029398918151855, | |
| "learning_rate": 7.610931783946632e-06, | |
| "loss": 0.4596, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.95286051826316, | |
| "grad_norm": 5.138970851898193, | |
| "learning_rate": 7.582239437630013e-06, | |
| "loss": 0.4445, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.9606073517449742, | |
| "grad_norm": 8.201314926147461, | |
| "learning_rate": 7.5535470913133925e-06, | |
| "loss": 0.4514, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.9683541852267885, | |
| "grad_norm": 5.1406145095825195, | |
| "learning_rate": 7.524854744996772e-06, | |
| "loss": 0.4546, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.9761010187086029, | |
| "grad_norm": 7.5569353103637695, | |
| "learning_rate": 7.496162398680153e-06, | |
| "loss": 0.4465, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.9838478521904171, | |
| "grad_norm": 5.765118598937988, | |
| "learning_rate": 7.4674700523635324e-06, | |
| "loss": 0.4355, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.9915946856722315, | |
| "grad_norm": 7.945804595947266, | |
| "learning_rate": 7.438777706046912e-06, | |
| "loss": 0.4694, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.9993415191540458, | |
| "grad_norm": 10.627592086791992, | |
| "learning_rate": 7.410085359730293e-06, | |
| "loss": 0.5065, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8090112858948102, | |
| "eval_loss": 0.4515499174594879, | |
| "eval_macro_f1": 0.7780532248284215, | |
| "eval_runtime": 57.5502, | |
| "eval_samples_per_second": 398.765, | |
| "eval_steps_per_second": 49.852, | |
| "step": 25817 | |
| } | |
| ], | |
| "logging_steps": 200, | |
| "max_steps": 77451, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.811880764833306e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |