| { |
| "results": { |
| "unseen_lanes": { |
| "accuracy": 0.9329816940738442, |
| "random_baseline": 0.2137126998901743, |
| "n": 3223, |
| "by_n_routes": { |
| "2": { |
| "acc": 0.9279279279279279, |
| "n": 111 |
| }, |
| "4": { |
| "acc": 0.9437652811735942, |
| "n": 818 |
| }, |
| "5": { |
| "acc": 0.9487179487179487, |
| "n": 624 |
| }, |
| "8": { |
| "acc": 0.9341317365269461, |
| "n": 334 |
| }, |
| "3": { |
| "acc": 0.9074074074074074, |
| "n": 324 |
| }, |
| "7": { |
| "acc": 0.907928388746803, |
| "n": 391 |
| }, |
| "6": { |
| "acc": 0.9395085066162571, |
| "n": 529 |
| }, |
| "9": { |
| "acc": 0.8913043478260869, |
| "n": 92 |
| } |
| }, |
| "by_difficulty": { |
| "easy": { |
| "acc": 0.9458927693064437, |
| "n": 2033 |
| }, |
| "hard": { |
| "acc": 0.9109243697478991, |
| "n": 1190 |
| } |
| }, |
| "by_mode": { |
| "topical": { |
| "acc": 0.8886666666666667, |
| "n": 1500 |
| }, |
| "mixed": { |
| "acc": 0.9803921568627451, |
| "n": 1275 |
| }, |
| "axis": { |
| "acc": 0.9464285714285714, |
| "n": 448 |
| } |
| } |
| }, |
| "unseen_domain": { |
| "accuracy": 0.7009111845399019, |
| "random_baseline": 0.25206113185236045, |
| "n": 9987, |
| "by_n_routes": { |
| "5": { |
| "acc": 0.6979293544457978, |
| "n": 1642 |
| }, |
| "4": { |
| "acc": 0.7129848229342327, |
| "n": 2965 |
| }, |
| "3": { |
| "acc": 0.7323481116584565, |
| "n": 1827 |
| }, |
| "2": { |
| "acc": 0.8065573770491803, |
| "n": 915 |
| }, |
| "6": { |
| "acc": 0.6463547334058759, |
| "n": 919 |
| }, |
| "7": { |
| "acc": 0.6275252525252525, |
| "n": 792 |
| }, |
| "8": { |
| "acc": 0.6140350877192983, |
| "n": 855 |
| }, |
| "9": { |
| "acc": 0.6666666666666666, |
| "n": 72 |
| } |
| }, |
| "by_difficulty": { |
| "easy": { |
| "acc": 0.6996331153293986, |
| "n": 6269 |
| }, |
| "hard": { |
| "acc": 0.7030661646046261, |
| "n": 3718 |
| } |
| }, |
| "by_mode": { |
| "topical": { |
| "acc": 0.7009111845399019, |
| "n": 9987 |
| } |
| } |
| }, |
| "unseen_axis": { |
| "accuracy": 0.6001931434089812, |
| "random_baseline": 0.23578575644002975, |
| "n": 12426, |
| "by_n_routes": { |
| "4": { |
| "acc": 0.5033030553261767, |
| "n": 4844 |
| }, |
| "5": { |
| "acc": 0.6116119725928597, |
| "n": 2773 |
| }, |
| "3": { |
| "acc": 0.6335429769392034, |
| "n": 2385 |
| }, |
| "6": { |
| "acc": 0.7689713322091062, |
| "n": 1186 |
| }, |
| "7": { |
| "acc": 0.7330508474576272, |
| "n": 944 |
| }, |
| "8": { |
| "acc": 0.7108843537414966, |
| "n": 294 |
| } |
| }, |
| "by_difficulty": { |
| "hard": { |
| "acc": 0.5900483481063659, |
| "n": 4964 |
| }, |
| "easy": { |
| "acc": 0.6069418386491557, |
| "n": 7462 |
| } |
| }, |
| "by_mode": { |
| "axis": { |
| "acc": 0.6001931434089812, |
| "n": 12426 |
| } |
| } |
| }, |
| "hard": { |
| "accuracy": 0.9109243697478991, |
| "random_baseline": 0.21085734293717487, |
| "n": 1190, |
| "by_n_routes": { |
| "4": { |
| "acc": 0.9256505576208178, |
| "n": 269 |
| }, |
| "5": { |
| "acc": 0.9203187250996016, |
| "n": 251 |
| }, |
| "8": { |
| "acc": 0.9448818897637795, |
| "n": 127 |
| }, |
| "3": { |
| "acc": 0.8571428571428571, |
| "n": 126 |
| }, |
| "7": { |
| "acc": 0.89171974522293, |
| "n": 157 |
| }, |
| "6": { |
| "acc": 0.9095477386934674, |
| "n": 199 |
| }, |
| "2": { |
| "acc": 0.9117647058823529, |
| "n": 34 |
| }, |
| "9": { |
| "acc": 0.8888888888888888, |
| "n": 27 |
| } |
| }, |
| "by_difficulty": { |
| "hard": { |
| "acc": 0.9109243697478991, |
| "n": 1190 |
| } |
| }, |
| "by_mode": { |
| "topical": { |
| "acc": 0.8462929475587704, |
| "n": 553 |
| }, |
| "mixed": { |
| "acc": 0.9765957446808511, |
| "n": 470 |
| }, |
| "axis": { |
| "acc": 0.9401197604790419, |
| "n": 167 |
| } |
| } |
| } |
| }, |
| "base_model": "/home/ahmed-wasfy/Documents/projects/arabic-llm-from-scratch-training/Nawah-BERT-6M-e3-FINAL", |
| "formulation": "routing_head", |
| "log_history": [ |
| { |
| "loss": 1.4586485290527345, |
| "grad_norm": 9.760659217834473, |
| "learning_rate": 0.0001194, |
| "epoch": 0.030902348578491966, |
| "step": 200 |
| }, |
| { |
| "loss": 1.0262959289550782, |
| "grad_norm": 10.573455810546875, |
| "learning_rate": 0.0002394, |
| "epoch": 0.06180469715698393, |
| "step": 400 |
| }, |
| { |
| "loss": 0.7875787353515625, |
| "grad_norm": 8.561141967773438, |
| "learning_rate": 0.0002999797249000742, |
| "epoch": 0.09270704573547589, |
| "step": 600 |
| }, |
| { |
| "loss": 0.6901941680908203, |
| "grad_norm": 11.628190040588379, |
| "learning_rate": 0.0002998150920740215, |
| "epoch": 0.12360939431396786, |
| "step": 800 |
| }, |
| { |
| "loss": 0.5956812667846679, |
| "grad_norm": 6.421826362609863, |
| "learning_rate": 0.00029948518059649235, |
| "epoch": 0.15451174289245984, |
| "step": 1000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.5233871340751648, |
| "eval_unseen_lanes_accuracy": 0.8091839900713621, |
| "eval_unseen_lanes_runtime": 0.2565, |
| "eval_unseen_lanes_samples_per_second": 12565.863, |
| "eval_unseen_lanes_steps_per_second": 198.839, |
| "epoch": 0.15451174289245984, |
| "step": 1000 |
| }, |
| { |
| "eval_unseen_domain_loss": 0.98743736743927, |
| "eval_unseen_domain_accuracy": 0.6317212376088915, |
| "eval_unseen_domain_runtime": 0.5442, |
| "eval_unseen_domain_samples_per_second": 18352.837, |
| "eval_unseen_domain_steps_per_second": 288.515, |
| "epoch": 0.15451174289245984, |
| "step": 1000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.176316499710083, |
| "eval_unseen_axis_accuracy": 0.5215676806695638, |
| "eval_unseen_axis_runtime": 0.6167, |
| "eval_unseen_axis_samples_per_second": 20148.724, |
| "eval_unseen_axis_steps_per_second": 316.192, |
| "epoch": 0.15451174289245984, |
| "step": 1000 |
| }, |
| { |
| "eval_hard_loss": 0.5815076231956482, |
| "eval_hard_accuracy": 0.7857142857142857, |
| "eval_hard_runtime": 0.1608, |
| "eval_hard_samples_per_second": 7400.389, |
| "eval_hard_steps_per_second": 118.157, |
| "epoch": 0.15451174289245984, |
| "step": 1000 |
| }, |
| { |
| "loss": 0.5344510269165039, |
| "grad_norm": 13.737728118896484, |
| "learning_rate": 0.0002989903544316465, |
| "epoch": 0.18541409147095178, |
| "step": 1200 |
| }, |
| { |
| "loss": 0.48971908569335937, |
| "grad_norm": 13.952028274536133, |
| "learning_rate": 0.00029833115948043487, |
| "epoch": 0.21631644004944375, |
| "step": 1400 |
| }, |
| { |
| "loss": 0.464974365234375, |
| "grad_norm": 5.948032379150391, |
| "learning_rate": 0.0002975083229783524, |
| "epoch": 0.24721878862793573, |
| "step": 1600 |
| }, |
| { |
| "loss": 0.4640715789794922, |
| "grad_norm": 7.1836748123168945, |
| "learning_rate": 0.00029652275269313814, |
| "epoch": 0.27812113720642767, |
| "step": 1800 |
| }, |
| { |
| "loss": 0.4463813018798828, |
| "grad_norm": 6.2541351318359375, |
| "learning_rate": 0.00029537553592331084, |
| "epoch": 0.30902348578491967, |
| "step": 2000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.3902101218700409, |
| "eval_unseen_lanes_accuracy": 0.8644120384734719, |
| "eval_unseen_lanes_runtime": 0.2597, |
| "eval_unseen_lanes_samples_per_second": 12409.924, |
| "eval_unseen_lanes_steps_per_second": 196.372, |
| "epoch": 0.30902348578491967, |
| "step": 2000 |
| }, |
| { |
| "eval_unseen_domain_loss": 0.9627552032470703, |
| "eval_unseen_domain_accuracy": 0.6365274857314509, |
| "eval_unseen_domain_runtime": 0.537, |
| "eval_unseen_domain_samples_per_second": 18597.421, |
| "eval_unseen_domain_steps_per_second": 292.36, |
| "epoch": 0.30902348578491967, |
| "step": 2000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.1707075834274292, |
| "eval_unseen_axis_accuracy": 0.554160630935136, |
| "eval_unseen_axis_runtime": 0.6407, |
| "eval_unseen_axis_samples_per_second": 19393.76, |
| "eval_unseen_axis_steps_per_second": 304.344, |
| "epoch": 0.30902348578491967, |
| "step": 2000 |
| }, |
| { |
| "eval_hard_loss": 0.45382222533226013, |
| "eval_hard_accuracy": 0.8436974789915966, |
| "eval_hard_runtime": 0.1554, |
| "eval_hard_samples_per_second": 7656.751, |
| "eval_hard_steps_per_second": 122.251, |
| "epoch": 0.30902348578491967, |
| "step": 2000 |
| }, |
| { |
| "loss": 0.4215016555786133, |
| "grad_norm": 8.966357231140137, |
| "learning_rate": 0.00029406793829864144, |
| "epoch": 0.3399258343634116, |
| "step": 2200 |
| }, |
| { |
| "loss": 0.4028751754760742, |
| "grad_norm": 12.1080961227417, |
| "learning_rate": 0.0002926014023838882, |
| "epoch": 0.37082818294190356, |
| "step": 2400 |
| }, |
| { |
| "loss": 0.3866802978515625, |
| "grad_norm": 12.563167572021484, |
| "learning_rate": 0.00029097754608733376, |
| "epoch": 0.40173053152039556, |
| "step": 2600 |
| }, |
| { |
| "loss": 0.38334075927734373, |
| "grad_norm": 6.601385593414307, |
| "learning_rate": 0.00028919816087588017, |
| "epoch": 0.4326328800988875, |
| "step": 2800 |
| }, |
| { |
| "loss": 0.3662459564208984, |
| "grad_norm": 9.042179107666016, |
| "learning_rate": 0.0002872652097986712, |
| "epoch": 0.4635352286773795, |
| "step": 3000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.353814959526062, |
| "eval_unseen_lanes_accuracy": 0.879304995345951, |
| "eval_unseen_lanes_runtime": 0.263, |
| "eval_unseen_lanes_samples_per_second": 12254.541, |
| "eval_unseen_lanes_steps_per_second": 193.913, |
| "epoch": 0.4635352286773795, |
| "step": 3000 |
| }, |
| { |
| "eval_unseen_domain_loss": 0.9891335368156433, |
| "eval_unseen_domain_accuracy": 0.6471412836687694, |
| "eval_unseen_domain_runtime": 0.5289, |
| "eval_unseen_domain_samples_per_second": 18884.131, |
| "eval_unseen_domain_steps_per_second": 296.867, |
| "epoch": 0.4635352286773795, |
| "step": 3000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.210868239402771, |
| "eval_unseen_axis_accuracy": 0.5622887493964268, |
| "eval_unseen_axis_runtime": 0.6483, |
| "eval_unseen_axis_samples_per_second": 19166.765, |
| "eval_unseen_axis_steps_per_second": 300.782, |
| "epoch": 0.4635352286773795, |
| "step": 3000 |
| }, |
| { |
| "eval_hard_loss": 0.40300533175468445, |
| "eval_hard_accuracy": 0.8596638655462185, |
| "eval_hard_runtime": 0.1621, |
| "eval_hard_samples_per_second": 7340.691, |
| "eval_hard_steps_per_second": 117.204, |
| "epoch": 0.4635352286773795, |
| "step": 3000 |
| }, |
| { |
| "loss": 0.362061882019043, |
| "grad_norm": 15.32834529876709, |
| "learning_rate": 0.00028518082532142156, |
| "epoch": 0.49443757725587145, |
| "step": 3200 |
| }, |
| { |
| "loss": 0.3505752182006836, |
| "grad_norm": 6.457422256469727, |
| "learning_rate": 0.00028294730697384356, |
| "epoch": 0.5253399258343634, |
| "step": 3400 |
| }, |
| { |
| "loss": 0.36617053985595704, |
| "grad_norm": 8.091940879821777, |
| "learning_rate": 0.000280567118812765, |
| "epoch": 0.5562422744128553, |
| "step": 3600 |
| }, |
| { |
| "loss": 0.3390579986572266, |
| "grad_norm": 5.5310492515563965, |
| "learning_rate": 0.00027804288670373826, |
| "epoch": 0.5871446229913473, |
| "step": 3800 |
| }, |
| { |
| "loss": 0.3409170913696289, |
| "grad_norm": 4.008467674255371, |
| "learning_rate": 0.0002753773954241395, |
| "epoch": 0.6180469715698393, |
| "step": 4000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.31256136298179626, |
| "eval_unseen_lanes_accuracy": 0.8935774123487434, |
| "eval_unseen_lanes_runtime": 0.2695, |
| "eval_unseen_lanes_samples_per_second": 11957.459, |
| "eval_unseen_lanes_steps_per_second": 189.212, |
| "epoch": 0.6180469715698393, |
| "step": 4000 |
| }, |
| { |
| "eval_unseen_domain_loss": 0.9226401448249817, |
| "eval_unseen_domain_accuracy": 0.6455392009612496, |
| "eval_unseen_domain_runtime": 0.5426, |
| "eval_unseen_domain_samples_per_second": 18407.339, |
| "eval_unseen_domain_steps_per_second": 289.371, |
| "epoch": 0.6180469715698393, |
| "step": 4000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.1576690673828125, |
| "eval_unseen_axis_accuracy": 0.5476420408820216, |
| "eval_unseen_axis_runtime": 0.6322, |
| "eval_unseen_axis_samples_per_second": 19653.69, |
| "eval_unseen_axis_steps_per_second": 308.423, |
| "epoch": 0.6180469715698393, |
| "step": 4000 |
| }, |
| { |
| "eval_hard_loss": 0.3740597665309906, |
| "eval_hard_accuracy": 0.8756302521008403, |
| "eval_hard_runtime": 0.156, |
| "eval_hard_samples_per_second": 7626.695, |
| "eval_hard_steps_per_second": 121.771, |
| "epoch": 0.6180469715698393, |
| "step": 4000 |
| }, |
| { |
| "loss": 0.33510574340820315, |
| "grad_norm": 5.052737712860107, |
| "learning_rate": 0.0002725735855909527, |
| "epoch": 0.6489493201483313, |
| "step": 4200 |
| }, |
| { |
| "loss": 0.3274649429321289, |
| "grad_norm": 4.481323719024658, |
| "learning_rate": 0.00026963455041662936, |
| "epoch": 0.6798516687268232, |
| "step": 4400 |
| }, |
| { |
| "loss": 0.3164741325378418, |
| "grad_norm": 10.086233139038086, |
| "learning_rate": 0.0002665635322966019, |
| "epoch": 0.7107540173053152, |
| "step": 4600 |
| }, |
| { |
| "loss": 0.31062273025512693, |
| "grad_norm": 8.942065238952637, |
| "learning_rate": 0.00026336391923221597, |
| "epoch": 0.7416563658838071, |
| "step": 4800 |
| }, |
| { |
| "loss": 0.30892057418823243, |
| "grad_norm": 5.766172409057617, |
| "learning_rate": 0.00026003924109302743, |
| "epoch": 0.7725587144622992, |
| "step": 5000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.3219285011291504, |
| "eval_unseen_lanes_accuracy": 0.8923363326093702, |
| "eval_unseen_lanes_runtime": 0.2565, |
| "eval_unseen_lanes_samples_per_second": 12567.451, |
| "eval_unseen_lanes_steps_per_second": 198.864, |
| "epoch": 0.7725587144622992, |
| "step": 5000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.126468300819397, |
| "eval_unseen_domain_accuracy": 0.6296185040552719, |
| "eval_unseen_domain_runtime": 0.5379, |
| "eval_unseen_domain_samples_per_second": 18568.213, |
| "eval_unseen_domain_steps_per_second": 291.9, |
| "epoch": 0.7725587144622992, |
| "step": 5000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.4002909660339355, |
| "eval_unseen_axis_accuracy": 0.5616449380331563, |
| "eval_unseen_axis_runtime": 0.6183, |
| "eval_unseen_axis_samples_per_second": 20097.832, |
| "eval_unseen_axis_steps_per_second": 315.393, |
| "epoch": 0.7725587144622992, |
| "step": 5000 |
| }, |
| { |
| "eval_hard_loss": 0.407728374004364, |
| "eval_hard_accuracy": 0.8714285714285714, |
| "eval_hard_runtime": 0.1627, |
| "eval_hard_samples_per_second": 7314.571, |
| "eval_hard_steps_per_second": 116.787, |
| "epoch": 0.7725587144622992, |
| "step": 5000 |
| }, |
| { |
| "loss": 0.29508913040161133, |
| "grad_norm": 12.423507690429688, |
| "learning_rate": 0.0002565931657225882, |
| "epoch": 0.8034610630407911, |
| "step": 5200 |
| }, |
| { |
| "loss": 0.3014940643310547, |
| "grad_norm": 5.474195957183838, |
| "learning_rate": 0.0002530294948920165, |
| "epoch": 0.8343634116192831, |
| "step": 5400 |
| }, |
| { |
| "loss": 0.30600976943969727, |
| "grad_norm": 6.128206253051758, |
| "learning_rate": 0.0002493521601058157, |
| "epoch": 0.865265760197775, |
| "step": 5600 |
| }, |
| { |
| "loss": 0.29127384185791017, |
| "grad_norm": 7.009108543395996, |
| "learning_rate": 0.0002455652182645697, |
| "epoch": 0.896168108776267, |
| "step": 5800 |
| }, |
| { |
| "loss": 0.2868840217590332, |
| "grad_norm": 4.6414289474487305, |
| "learning_rate": 0.00024167284718929817, |
| "epoch": 0.927070457354759, |
| "step": 6000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.2860090136528015, |
| "eval_unseen_lanes_accuracy": 0.9097114489605957, |
| "eval_unseen_lanes_runtime": 0.263, |
| "eval_unseen_lanes_samples_per_second": 12253.064, |
| "eval_unseen_lanes_steps_per_second": 193.89, |
| "epoch": 0.927070457354759, |
| "step": 6000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.0391852855682373, |
| "eval_unseen_domain_accuracy": 0.6249123861019326, |
| "eval_unseen_domain_runtime": 0.5278, |
| "eval_unseen_domain_samples_per_second": 18923.382, |
| "eval_unseen_domain_steps_per_second": 297.484, |
| "epoch": 0.927070457354759, |
| "step": 6000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.4318047761917114, |
| "eval_unseen_axis_accuracy": 0.5544825366167713, |
| "eval_unseen_axis_runtime": 0.6292, |
| "eval_unseen_axis_samples_per_second": 19750.198, |
| "eval_unseen_axis_steps_per_second": 309.938, |
| "epoch": 0.927070457354759, |
| "step": 6000 |
| }, |
| { |
| "eval_hard_loss": 0.3728649616241455, |
| "eval_hard_accuracy": 0.8815126050420168, |
| "eval_hard_runtime": 0.1566, |
| "eval_hard_samples_per_second": 7599.279, |
| "eval_hard_steps_per_second": 121.333, |
| "epoch": 0.927070457354759, |
| "step": 6000 |
| }, |
| { |
| "loss": 0.29034774780273437, |
| "grad_norm": 4.269179344177246, |
| "learning_rate": 0.000237679341012411, |
| "epoch": 0.957972805933251, |
| "step": 6200 |
| }, |
| { |
| "loss": 0.27252685546875, |
| "grad_norm": 3.099332809448242, |
| "learning_rate": 0.0002335891054403457, |
| "epoch": 0.9888751545117429, |
| "step": 6400 |
| }, |
| { |
| "loss": 0.2299000358581543, |
| "grad_norm": 2.502373695373535, |
| "learning_rate": 0.00022940665289311417, |
| "epoch": 1.019777503090235, |
| "step": 6600 |
| }, |
| { |
| "loss": 0.19722644805908204, |
| "grad_norm": 7.622342586517334, |
| "learning_rate": 0.00022513659752612167, |
| "epoch": 1.0506798516687268, |
| "step": 6800 |
| }, |
| { |
| "loss": 0.18992324829101562, |
| "grad_norm": 1.0242284536361694, |
| "learning_rate": 0.00022078365013974898, |
| "epoch": 1.0815822002472189, |
| "step": 7000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.3706987798213959, |
| "eval_unseen_lanes_accuracy": 0.9121936084393423, |
| "eval_unseen_lanes_runtime": 0.2633, |
| "eval_unseen_lanes_samples_per_second": 12241.369, |
| "eval_unseen_lanes_steps_per_second": 193.705, |
| "epoch": 1.0815822002472189, |
| "step": 7000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.5655317306518555, |
| "eval_unseen_domain_accuracy": 0.6401321718233703, |
| "eval_unseen_domain_runtime": 0.5232, |
| "eval_unseen_domain_samples_per_second": 19088.449, |
| "eval_unseen_domain_steps_per_second": 300.079, |
| "epoch": 1.0815822002472189, |
| "step": 7000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.2009739875793457, |
| "eval_unseen_axis_accuracy": 0.569853532914856, |
| "eval_unseen_axis_runtime": 0.6295, |
| "eval_unseen_axis_samples_per_second": 19738.798, |
| "eval_unseen_axis_steps_per_second": 309.759, |
| "epoch": 1.0815822002472189, |
| "step": 7000 |
| }, |
| { |
| "eval_hard_loss": 0.4946191608905792, |
| "eval_hard_accuracy": 0.8890756302521008, |
| "eval_hard_runtime": 0.1578, |
| "eval_hard_samples_per_second": 7540.271, |
| "eval_hard_steps_per_second": 120.391, |
| "epoch": 1.0815822002472189, |
| "step": 7000 |
| }, |
| { |
| "loss": 0.19767265319824218, |
| "grad_norm": 14.961465835571289, |
| "learning_rate": 0.00021635261298231433, |
| "epoch": 1.1124845488257107, |
| "step": 7200 |
| }, |
| { |
| "loss": 0.1902615547180176, |
| "grad_norm": 11.509106636047363, |
| "learning_rate": 0.0002118483744521485, |
| "epoch": 1.1433868974042027, |
| "step": 7400 |
| }, |
| { |
| "loss": 0.1893865203857422, |
| "grad_norm": 6.442070007324219, |
| "learning_rate": 0.00020727590370462763, |
| "epoch": 1.1742892459826946, |
| "step": 7600 |
| }, |
| { |
| "loss": 0.19014299392700196, |
| "grad_norm": 6.42080020904541, |
| "learning_rate": 0.00020264024517011316, |
| "epoch": 1.2051915945611866, |
| "step": 7800 |
| }, |
| { |
| "loss": 0.19989730834960937, |
| "grad_norm": 2.8540380001068115, |
| "learning_rate": 0.00019794651298884748, |
| "epoch": 1.2360939431396787, |
| "step": 8000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.24737592041492462, |
| "eval_unseen_lanes_accuracy": 0.9236735960285448, |
| "eval_unseen_lanes_runtime": 0.259, |
| "eval_unseen_lanes_samples_per_second": 12442.489, |
| "eval_unseen_lanes_steps_per_second": 196.887, |
| "epoch": 1.2360939431396787, |
| "step": 8000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.0949585437774658, |
| "eval_unseen_domain_accuracy": 0.6586562531290678, |
| "eval_unseen_domain_runtime": 0.5385, |
| "eval_unseen_domain_samples_per_second": 18544.793, |
| "eval_unseen_domain_steps_per_second": 291.532, |
| "epoch": 1.2360939431396787, |
| "step": 8000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.4516733884811401, |
| "eval_unseen_axis_accuracy": 0.5766940286496056, |
| "eval_unseen_axis_runtime": 0.6236, |
| "eval_unseen_axis_samples_per_second": 19926.265, |
| "eval_unseen_axis_steps_per_second": 312.701, |
| "epoch": 1.2360939431396787, |
| "step": 8000 |
| }, |
| { |
| "eval_hard_loss": 0.331650972366333, |
| "eval_hard_accuracy": 0.8957983193277311, |
| "eval_hard_runtime": 0.1588, |
| "eval_hard_samples_per_second": 7494.912, |
| "eval_hard_steps_per_second": 119.667, |
| "epoch": 1.2360939431396787, |
| "step": 8000 |
| }, |
| { |
| "loss": 0.20300497055053712, |
| "grad_norm": 5.189218997955322, |
| "learning_rate": 0.0001931998853689441, |
| "epoch": 1.2669962917181705, |
| "step": 8200 |
| }, |
| { |
| "loss": 0.17553037643432617, |
| "grad_norm": 3.2245094776153564, |
| "learning_rate": 0.00018840559887369732, |
| "epoch": 1.2978986402966626, |
| "step": 8400 |
| }, |
| { |
| "loss": 0.19114152908325197, |
| "grad_norm": 5.41471529006958, |
| "learning_rate": 0.0001835689426445131, |
| "epoch": 1.3288009888751544, |
| "step": 8600 |
| }, |
| { |
| "loss": 0.17892492294311524, |
| "grad_norm": 3.311156988143921, |
| "learning_rate": 0.00017869525256583495, |
| "epoch": 1.3597033374536465, |
| "step": 8800 |
| }, |
| { |
| "loss": 0.1781780242919922, |
| "grad_norm": 1.3695971965789795, |
| "learning_rate": 0.00017378990537850246, |
| "epoch": 1.3906056860321385, |
| "step": 9000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.27402055263519287, |
| "eval_unseen_lanes_accuracy": 0.9156065777226187, |
| "eval_unseen_lanes_runtime": 0.2681, |
| "eval_unseen_lanes_samples_per_second": 12022.189, |
| "eval_unseen_lanes_steps_per_second": 190.236, |
| "epoch": 1.3906056860321385, |
| "step": 9000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.082552194595337, |
| "eval_unseen_domain_accuracy": 0.6728747371583058, |
| "eval_unseen_domain_runtime": 0.5492, |
| "eval_unseen_domain_samples_per_second": 18185.53, |
| "eval_unseen_domain_steps_per_second": 285.884, |
| "epoch": 1.3906056860321385, |
| "step": 9000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.6101419925689697, |
| "eval_unseen_axis_accuracy": 0.5588282633188476, |
| "eval_unseen_axis_runtime": 0.6124, |
| "eval_unseen_axis_samples_per_second": 20291.969, |
| "eval_unseen_axis_steps_per_second": 318.44, |
| "epoch": 1.3906056860321385, |
| "step": 9000 |
| }, |
| { |
| "eval_hard_loss": 0.3630846440792084, |
| "eval_hard_accuracy": 0.892436974789916, |
| "eval_hard_runtime": 0.1629, |
| "eval_hard_samples_per_second": 7303.301, |
| "eval_hard_steps_per_second": 116.607, |
| "epoch": 1.3906056860321385, |
| "step": 9000 |
| }, |
| { |
| "loss": 0.1790849494934082, |
| "grad_norm": 1.0810199975967407, |
| "learning_rate": 0.00016885831274803545, |
| "epoch": 1.4215080346106304, |
| "step": 9200 |
| }, |
| { |
| "loss": 0.18830131530761718, |
| "grad_norm": 6.382857799530029, |
| "learning_rate": 0.00016390591529438968, |
| "epoch": 1.4524103831891224, |
| "step": 9400 |
| }, |
| { |
| "loss": 0.19004894256591798, |
| "grad_norm": 4.801127910614014, |
| "learning_rate": 0.00015893817658976849, |
| "epoch": 1.4833127317676142, |
| "step": 9600 |
| }, |
| { |
| "loss": 0.18906343460083008, |
| "grad_norm": 5.73111629486084, |
| "learning_rate": 0.00015396057713111408, |
| "epoch": 1.5142150803461063, |
| "step": 9800 |
| }, |
| { |
| "loss": 0.16564352035522462, |
| "grad_norm": 14.169121742248535, |
| "learning_rate": 0.00014897860829392618, |
| "epoch": 1.5451174289245984, |
| "step": 10000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.3004416823387146, |
| "eval_unseen_lanes_accuracy": 0.9174681973316786, |
| "eval_unseen_lanes_runtime": 0.2645, |
| "eval_unseen_lanes_samples_per_second": 12186.875, |
| "eval_unseen_lanes_steps_per_second": 192.842, |
| "epoch": 1.5451174289245984, |
| "step": 10000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.3140738010406494, |
| "eval_unseen_domain_accuracy": 0.6836887954340642, |
| "eval_unseen_domain_runtime": 0.5226, |
| "eval_unseen_domain_samples_per_second": 19111.476, |
| "eval_unseen_domain_steps_per_second": 300.441, |
| "epoch": 1.5451174289245984, |
| "step": 10000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.10270094871521, |
| "eval_unseen_axis_accuracy": 0.577015934331241, |
| "eval_unseen_axis_runtime": 0.6091, |
| "eval_unseen_axis_samples_per_second": 20400.548, |
| "eval_unseen_axis_steps_per_second": 320.144, |
| "epoch": 1.5451174289245984, |
| "step": 10000 |
| }, |
| { |
| "eval_hard_loss": 0.386659175157547, |
| "eval_hard_accuracy": 0.8941176470588236, |
| "eval_hard_runtime": 0.1575, |
| "eval_hard_samples_per_second": 7557.259, |
| "eval_hard_steps_per_second": 120.662, |
| "epoch": 1.5451174289245984, |
| "step": 10000 |
| }, |
| { |
| "loss": 0.17348814010620117, |
| "grad_norm": 3.8261921405792236, |
| "learning_rate": 0.00014399776627408, |
| "epoch": 1.5760197775030902, |
| "step": 10200 |
| }, |
| { |
| "loss": 0.17659198760986328, |
| "grad_norm": 1.6768690347671509, |
| "learning_rate": 0.00013902354602432623, |
| "epoch": 1.6069221260815822, |
| "step": 10400 |
| }, |
| { |
| "loss": 0.17375629425048827, |
| "grad_norm": 7.939939975738525, |
| "learning_rate": 0.00013406143519216157, |
| "epoch": 1.637824474660074, |
| "step": 10600 |
| }, |
| { |
| "loss": 0.17606712341308595, |
| "grad_norm": 3.530085802078247, |
| "learning_rate": 0.0001291169080657601, |
| "epoch": 1.6687268232385661, |
| "step": 10800 |
| }, |
| { |
| "loss": 0.15471673965454102, |
| "grad_norm": 8.739562034606934, |
| "learning_rate": 0.00012419541953464147, |
| "epoch": 1.6996291718170582, |
| "step": 11000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.23692560195922852, |
| "eval_unseen_lanes_accuracy": 0.9236735960285448, |
| "eval_unseen_lanes_runtime": 0.2785, |
| "eval_unseen_lanes_samples_per_second": 11574.333, |
| "eval_unseen_lanes_steps_per_second": 183.15, |
| "epoch": 1.6996291718170582, |
| "step": 11000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.0458940267562866, |
| "eval_unseen_domain_accuracy": 0.6836887954340642, |
| "eval_unseen_domain_runtime": 0.536, |
| "eval_unseen_domain_samples_per_second": 18632.711, |
| "eval_unseen_domain_steps_per_second": 292.914, |
| "epoch": 1.6996291718170582, |
| "step": 11000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.5432320833206177, |
| "eval_unseen_axis_accuracy": 0.5790278448414614, |
| "eval_unseen_axis_runtime": 0.6443, |
| "eval_unseen_axis_samples_per_second": 19286.547, |
| "eval_unseen_axis_steps_per_second": 302.662, |
| "epoch": 1.6996291718170582, |
| "step": 11000 |
| }, |
| { |
| "eval_hard_loss": 0.3232015073299408, |
| "eval_hard_accuracy": 0.9016806722689076, |
| "eval_hard_runtime": 0.1601, |
| "eval_hard_samples_per_second": 7433.475, |
| "eval_hard_steps_per_second": 118.686, |
| "epoch": 1.6996291718170582, |
| "step": 11000 |
| }, |
| { |
| "loss": 0.16208927154541017, |
| "grad_norm": 9.356490135192871, |
| "learning_rate": 0.00011930239907174065, |
| "epoch": 1.73053152039555, |
| "step": 11200 |
| }, |
| { |
| "loss": 0.15787601470947266, |
| "grad_norm": 4.05722188949585, |
| "learning_rate": 0.00011444324474351796, |
| "epoch": 1.7614338689740419, |
| "step": 11400 |
| }, |
| { |
| "loss": 0.165930118560791, |
| "grad_norm": 1.2218636274337769, |
| "learning_rate": 0.00010962331725471657, |
| "epoch": 1.792336217552534, |
| "step": 11600 |
| }, |
| { |
| "loss": 0.16691875457763672, |
| "grad_norm": 5.362765789031982, |
| "learning_rate": 0.00010484793403433862, |
| "epoch": 1.823238566131026, |
| "step": 11800 |
| }, |
| { |
| "loss": 0.15950778961181641, |
| "grad_norm": 4.139005661010742, |
| "learning_rate": 0.00010012236336936396, |
| "epoch": 1.854140914709518, |
| "step": 12000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.23127563297748566, |
| "eval_unseen_lanes_accuracy": 0.9239838659633881, |
| "eval_unseen_lanes_runtime": 0.2522, |
| "eval_unseen_lanes_samples_per_second": 12777.842, |
| "eval_unseen_lanes_steps_per_second": 202.194, |
| "epoch": 1.854140914709518, |
| "step": 12000 |
| }, |
| { |
| "eval_unseen_domain_loss": 0.9452027082443237, |
| "eval_unseen_domain_accuracy": 0.6982076699709623, |
| "eval_unseen_domain_runtime": 0.5298, |
| "eval_unseen_domain_samples_per_second": 18849.027, |
| "eval_unseen_domain_steps_per_second": 296.315, |
| "epoch": 1.854140914709518, |
| "step": 12000 |
| }, |
| { |
| "eval_unseen_axis_loss": 1.4740277528762817, |
| "eval_unseen_axis_accuracy": 0.5853854820537583, |
| "eval_unseen_axis_runtime": 0.6167, |
| "eval_unseen_axis_samples_per_second": 20149.292, |
| "eval_unseen_axis_steps_per_second": 316.201, |
| "epoch": 1.854140914709518, |
| "step": 12000 |
| }, |
| { |
| "eval_hard_loss": 0.3201354742050171, |
| "eval_hard_accuracy": 0.8941176470588236, |
| "eval_hard_runtime": 0.1574, |
| "eval_hard_samples_per_second": 7559.64, |
| "eval_hard_steps_per_second": 120.7, |
| "epoch": 1.854140914709518, |
| "step": 12000 |
| }, |
| { |
| "loss": 0.1575588130950928, |
| "grad_norm": 2.324737071990967, |
| "learning_rate": 9.545181859268278e-05, |
| "epoch": 1.8850432632880099, |
| "step": 12200 |
| }, |
| { |
| "loss": 0.15390984535217286, |
| "grad_norm": 8.164409637451172, |
| "learning_rate": 9.084145233165512e-05, |
| "epoch": 1.9159456118665017, |
| "step": 12400 |
| }, |
| { |
| "loss": 0.1659238624572754, |
| "grad_norm": 4.467927932739258, |
| "learning_rate": 8.629635082364105e-05, |
| "epoch": 1.9468479604449938, |
| "step": 12600 |
| }, |
| { |
| "loss": 0.16303443908691406, |
| "grad_norm": 0.978862464427948, |
| "learning_rate": 8.182152830477452e-05, |
| "epoch": 1.9777503090234858, |
| "step": 12800 |
| }, |
| { |
| "loss": 0.12827521324157715, |
| "grad_norm": 9.180611610412598, |
| "learning_rate": 7.74219214781692e-05, |
| "epoch": 2.008652657601978, |
| "step": 13000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.29380783438682556, |
| "eval_unseen_lanes_accuracy": 0.9277071051815079, |
| "eval_unseen_lanes_runtime": 0.2597, |
| "eval_unseen_lanes_samples_per_second": 12409.628, |
| "eval_unseen_lanes_steps_per_second": 196.367, |
| "epoch": 2.008652657601978, |
| "step": 13000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.2944514751434326, |
| "eval_unseen_domain_accuracy": 0.7021127465705417, |
| "eval_unseen_domain_runtime": 0.5274, |
| "eval_unseen_domain_samples_per_second": 18936.119, |
| "eval_unseen_domain_steps_per_second": 297.684, |
| "epoch": 2.008652657601978, |
| "step": 13000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.235142469406128, |
| "eval_unseen_axis_accuracy": 0.5928697891517786, |
| "eval_unseen_axis_runtime": 0.6267, |
| "eval_unseen_axis_samples_per_second": 19826.464, |
| "eval_unseen_axis_steps_per_second": 311.135, |
| "epoch": 2.008652657601978, |
| "step": 13000 |
| }, |
| { |
| "eval_hard_loss": 0.39865994453430176, |
| "eval_hard_accuracy": 0.9042016806722689, |
| "eval_hard_runtime": 0.1598, |
| "eval_hard_samples_per_second": 7446.484, |
| "eval_hard_steps_per_second": 118.893, |
| "epoch": 2.008652657601978, |
| "step": 13000 |
| }, |
| { |
| "loss": 0.09030275344848633, |
| "grad_norm": 4.3157548904418945, |
| "learning_rate": 7.310238406766051e-05, |
| "epoch": 2.03955500618047, |
| "step": 13200 |
| }, |
| { |
| "loss": 0.0979629898071289, |
| "grad_norm": 5.349492073059082, |
| "learning_rate": 6.886768146309235e-05, |
| "epoch": 2.0704573547589615, |
| "step": 13400 |
| }, |
| { |
| "loss": 0.087772216796875, |
| "grad_norm": 3.7547731399536133, |
| "learning_rate": 6.472248546305456e-05, |
| "epoch": 2.1013597033374536, |
| "step": 13600 |
| }, |
| { |
| "loss": 0.08256999015808106, |
| "grad_norm": 1.9062801599502563, |
| "learning_rate": 6.067136912087232e-05, |
| "epoch": 2.1322620519159456, |
| "step": 13800 |
| }, |
| { |
| "loss": 0.09442170143127442, |
| "grad_norm": 4.407528877258301, |
| "learning_rate": 5.671880169953349e-05, |
| "epoch": 2.1631644004944377, |
| "step": 14000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.29582127928733826, |
| "eval_unseen_lanes_accuracy": 0.9298789947254111, |
| "eval_unseen_lanes_runtime": 0.2712, |
| "eval_unseen_lanes_samples_per_second": 11885.213, |
| "eval_unseen_lanes_steps_per_second": 188.069, |
| "epoch": 2.1631644004944377, |
| "step": 14000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.4203895330429077, |
| "eval_unseen_domain_accuracy": 0.6918994693101032, |
| "eval_unseen_domain_runtime": 0.5351, |
| "eval_unseen_domain_samples_per_second": 18662.663, |
| "eval_unseen_domain_steps_per_second": 293.385, |
| "epoch": 2.1631644004944377, |
| "step": 14000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.1755688190460205, |
| "eval_unseen_axis_accuracy": 0.594479317559955, |
| "eval_unseen_axis_runtime": 0.6099, |
| "eval_unseen_axis_samples_per_second": 20374.27, |
| "eval_unseen_axis_steps_per_second": 319.731, |
| "epoch": 2.1631644004944377, |
| "step": 14000 |
| }, |
| { |
| "eval_hard_loss": 0.40849950909614563, |
| "eval_hard_accuracy": 0.907563025210084, |
| "eval_hard_runtime": 0.1592, |
| "eval_hard_samples_per_second": 7474.619, |
| "eval_hard_steps_per_second": 119.343, |
| "epoch": 2.1631644004944377, |
| "step": 14000 |
| }, |
| { |
| "loss": 0.09402852058410645, |
| "grad_norm": 6.6268205642700195, |
| "learning_rate": 5.28691437411185e-05, |
| "epoch": 2.1940667490729293, |
| "step": 14200 |
| }, |
| { |
| "loss": 0.0864114761352539, |
| "grad_norm": 6.456910610198975, |
| "learning_rate": 4.912664225617335e-05, |
| "epoch": 2.2249690976514214, |
| "step": 14400 |
| }, |
| { |
| "loss": 0.0929531478881836, |
| "grad_norm": 2.4680755138397217, |
| "learning_rate": 4.549542603833318e-05, |
| "epoch": 2.2558714462299134, |
| "step": 14600 |
| }, |
| { |
| "loss": 0.0913322925567627, |
| "grad_norm": 1.958465337753296, |
| "learning_rate": 4.197950110936387e-05, |
| "epoch": 2.2867737948084055, |
| "step": 14800 |
| }, |
| { |
| "loss": 0.07560197353363036, |
| "grad_norm": 7.140666961669922, |
| "learning_rate": 3.8582746299648135e-05, |
| "epoch": 2.3176761433868975, |
| "step": 15000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.31193897128105164, |
| "eval_unseen_lanes_accuracy": 0.932671424139001, |
| "eval_unseen_lanes_runtime": 0.2695, |
| "eval_unseen_lanes_samples_per_second": 11957.861, |
| "eval_unseen_lanes_steps_per_second": 189.218, |
| "epoch": 2.3176761433868975, |
| "step": 15000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.5022051334381104, |
| "eval_unseen_domain_accuracy": 0.7045158706318214, |
| "eval_unseen_domain_runtime": 0.5152, |
| "eval_unseen_domain_samples_per_second": 19385.086, |
| "eval_unseen_domain_steps_per_second": 304.742, |
| "epoch": 2.3176761433868975, |
| "step": 15000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.4407944679260254, |
| "eval_unseen_axis_accuracy": 0.598020280057943, |
| "eval_unseen_axis_runtime": 0.6255, |
| "eval_unseen_axis_samples_per_second": 19864.27, |
| "eval_unseen_axis_steps_per_second": 311.728, |
| "epoch": 2.3176761433868975, |
| "step": 15000 |
| }, |
| { |
| "eval_hard_loss": 0.4339127838611603, |
| "eval_hard_accuracy": 0.9084033613445378, |
| "eval_hard_runtime": 0.1603, |
| "eval_hard_samples_per_second": 7424.243, |
| "eval_hard_steps_per_second": 118.538, |
| "epoch": 2.3176761433868975, |
| "step": 15000 |
| }, |
| { |
| "loss": 0.09124773979187012, |
| "grad_norm": 1.9079341888427734, |
| "learning_rate": 3.5308908968992e-05, |
| "epoch": 2.348578491965389, |
| "step": 15200 |
| }, |
| { |
| "loss": 0.0789580774307251, |
| "grad_norm": 3.3277528285980225, |
| "learning_rate": 3.2161600872470804e-05, |
| "epoch": 2.379480840543881, |
| "step": 15400 |
| }, |
| { |
| "loss": 0.0785712718963623, |
| "grad_norm": 8.918529510498047, |
| "learning_rate": 2.9144294175877418e-05, |
| "epoch": 2.4103831891223733, |
| "step": 15600 |
| }, |
| { |
| "loss": 0.08375604629516602, |
| "grad_norm": 1.9613817930221558, |
| "learning_rate": 2.6260317625168052e-05, |
| "epoch": 2.4412855377008653, |
| "step": 15800 |
| }, |
| { |
| "loss": 0.0875228214263916, |
| "grad_norm": 0.23354528844356537, |
| "learning_rate": 2.351285287413069e-05, |
| "epoch": 2.4721878862793574, |
| "step": 16000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.27757585048675537, |
| "eval_unseen_lanes_accuracy": 0.9332919640086875, |
| "eval_unseen_lanes_runtime": 0.269, |
| "eval_unseen_lanes_samples_per_second": 11979.34, |
| "eval_unseen_lanes_steps_per_second": 189.558, |
| "epoch": 2.4721878862793574, |
| "step": 16000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.3426191806793213, |
| "eval_unseen_domain_accuracy": 0.7027135275858616, |
| "eval_unseen_domain_runtime": 0.5238, |
| "eval_unseen_domain_samples_per_second": 19065.435, |
| "eval_unseen_domain_steps_per_second": 299.717, |
| "epoch": 2.4721878862793574, |
| "step": 16000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.2119669914245605, |
| "eval_unseen_axis_accuracy": 0.5920650249476903, |
| "eval_unseen_axis_runtime": 0.6271, |
| "eval_unseen_axis_samples_per_second": 19813.628, |
| "eval_unseen_axis_steps_per_second": 310.933, |
| "epoch": 2.4721878862793574, |
| "step": 16000 |
| }, |
| { |
| "eval_hard_loss": 0.3819628953933716, |
| "eval_hard_accuracy": 0.9109243697478991, |
| "eval_hard_runtime": 0.1609, |
| "eval_hard_samples_per_second": 7396.846, |
| "eval_hard_steps_per_second": 118.101, |
| "epoch": 2.4721878862793574, |
| "step": 16000 |
| }, |
| { |
| "loss": 0.07829498767852783, |
| "grad_norm": 1.8762905597686768, |
| "learning_rate": 2.0904930974328466e-05, |
| "epoch": 2.503090234857849, |
| "step": 16200 |
| }, |
| { |
| "loss": 0.08220289230346679, |
| "grad_norm": 7.849063396453857, |
| "learning_rate": 1.8439429031190515e-05, |
| "epoch": 2.533992583436341, |
| "step": 16400 |
| }, |
| { |
| "loss": 0.08165282249450684, |
| "grad_norm": 0.32895785570144653, |
| "learning_rate": 1.6119067029938165e-05, |
| "epoch": 2.564894932014833, |
| "step": 16600 |
| }, |
| { |
| "loss": 0.08682233810424805, |
| "grad_norm": 7.921652793884277, |
| "learning_rate": 1.3946404834849312e-05, |
| "epoch": 2.595797280593325, |
| "step": 16800 |
| }, |
| { |
| "loss": 0.08006702423095703, |
| "grad_norm": 8.002998352050781, |
| "learning_rate": 1.192383936517129e-05, |
| "epoch": 2.626699629171817, |
| "step": 17000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.2994181513786316, |
| "eval_unseen_lanes_accuracy": 0.9329816940738442, |
| "eval_unseen_lanes_runtime": 0.2673, |
| "eval_unseen_lanes_samples_per_second": 12058.253, |
| "eval_unseen_lanes_steps_per_second": 190.807, |
| "epoch": 2.626699629171817, |
| "step": 17000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.4774185419082642, |
| "eval_unseen_domain_accuracy": 0.7055171723240212, |
| "eval_unseen_domain_runtime": 0.5317, |
| "eval_unseen_domain_samples_per_second": 18784.163, |
| "eval_unseen_domain_steps_per_second": 295.295, |
| "epoch": 2.626699629171817, |
| "step": 17000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.3520829677581787, |
| "eval_unseen_axis_accuracy": 0.5997907613069371, |
| "eval_unseen_axis_runtime": 0.6139, |
| "eval_unseen_axis_samples_per_second": 20241.068, |
| "eval_unseen_axis_steps_per_second": 317.641, |
| "epoch": 2.626699629171817, |
| "step": 17000 |
| }, |
| { |
| "eval_hard_loss": 0.4058825969696045, |
| "eval_hard_accuracy": 0.9117647058823529, |
| "eval_hard_runtime": 0.1604, |
| "eval_hard_samples_per_second": 7416.818, |
| "eval_hard_steps_per_second": 118.42, |
| "epoch": 2.626699629171817, |
| "step": 17000 |
| }, |
| { |
| "loss": 0.0812064266204834, |
| "grad_norm": 3.427633285522461, |
| "learning_rate": 1.005360195079697e-05, |
| "epoch": 2.657601977750309, |
| "step": 17200 |
| }, |
| { |
| "loss": 0.07186338424682617, |
| "grad_norm": 2.308717727661133, |
| "learning_rate": 8.33775587062237e-06, |
| "epoch": 2.688504326328801, |
| "step": 17400 |
| }, |
| { |
| "loss": 0.08242843627929687, |
| "grad_norm": 4.111525058746338, |
| "learning_rate": 6.778194076301119e-06, |
| "epoch": 2.719406674907293, |
| "step": 17600 |
| }, |
| { |
| "loss": 0.07691887378692627, |
| "grad_norm": 10.234223365783691, |
| "learning_rate": 5.376637103906678e-06, |
| "epoch": 2.750309023485785, |
| "step": 17800 |
| }, |
| { |
| "loss": 0.05823780059814453, |
| "grad_norm": 1.5334542989730835, |
| "learning_rate": 4.134631175806541e-06, |
| "epoch": 2.781211372064277, |
| "step": 18000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.29644232988357544, |
| "eval_unseen_lanes_accuracy": 0.9329816940738442, |
| "eval_unseen_lanes_runtime": 0.262, |
| "eval_unseen_lanes_samples_per_second": 12300.706, |
| "eval_unseen_lanes_steps_per_second": 194.643, |
| "epoch": 2.781211372064277, |
| "step": 18000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.4905004501342773, |
| "eval_unseen_domain_accuracy": 0.7009111845399019, |
| "eval_unseen_domain_runtime": 0.555, |
| "eval_unseen_domain_samples_per_second": 17993.704, |
| "eval_unseen_domain_steps_per_second": 282.869, |
| "epoch": 2.781211372064277, |
| "step": 18000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.323622465133667, |
| "eval_unseen_axis_accuracy": 0.6001931434089812, |
| "eval_unseen_axis_runtime": 0.6254, |
| "eval_unseen_axis_samples_per_second": 19869.586, |
| "eval_unseen_axis_steps_per_second": 311.811, |
| "epoch": 2.781211372064277, |
| "step": 18000 |
| }, |
| { |
| "eval_hard_loss": 0.40079689025878906, |
| "eval_hard_accuracy": 0.9109243697478991, |
| "eval_hard_runtime": 0.1574, |
| "eval_hard_samples_per_second": 7559.33, |
| "eval_hard_steps_per_second": 120.695, |
| "epoch": 2.781211372064277, |
| "step": 18000 |
| }, |
| { |
| "loss": 0.08730937957763672, |
| "grad_norm": 0.42866209149360657, |
| "learning_rate": 3.0535464948426957e-06, |
| "epoch": 2.8121137206427687, |
| "step": 18200 |
| }, |
| { |
| "loss": 0.07673277854919433, |
| "grad_norm": 9.0943021774292, |
| "learning_rate": 2.134575732699473e-06, |
| "epoch": 2.8430160692212607, |
| "step": 18400 |
| }, |
| { |
| "loss": 0.0871958827972412, |
| "grad_norm": 8.693181037902832, |
| "learning_rate": 1.3787327141271065e-06, |
| "epoch": 2.8739184177997528, |
| "step": 18600 |
| }, |
| { |
| "loss": 0.08861166954040528, |
| "grad_norm": 1.7328418493270874, |
| "learning_rate": 7.868512984726017e-07, |
| "epoch": 2.904820766378245, |
| "step": 18800 |
| }, |
| { |
| "loss": 0.07439324855804444, |
| "grad_norm": 7.242162227630615, |
| "learning_rate": 3.595844597511666e-07, |
| "epoch": 2.935723114956737, |
| "step": 19000 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.2913786470890045, |
| "eval_unseen_lanes_accuracy": 0.9332919640086875, |
| "eval_unseen_lanes_runtime": 0.2558, |
| "eval_unseen_lanes_samples_per_second": 12600.204, |
| "eval_unseen_lanes_steps_per_second": 199.383, |
| "epoch": 2.935723114956737, |
| "step": 19000 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.4577001333236694, |
| "eval_unseen_domain_accuracy": 0.7011114448783419, |
| "eval_unseen_domain_runtime": 0.5411, |
| "eval_unseen_domain_samples_per_second": 18457.163, |
| "eval_unseen_domain_steps_per_second": 290.155, |
| "epoch": 2.935723114956737, |
| "step": 19000 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.2998874187469482, |
| "eval_unseen_axis_accuracy": 0.5989859971028488, |
| "eval_unseen_axis_runtime": 0.6285, |
| "eval_unseen_axis_samples_per_second": 19770.314, |
| "eval_unseen_axis_steps_per_second": 310.254, |
| "epoch": 2.935723114956737, |
| "step": 19000 |
| }, |
| { |
| "eval_hard_loss": 0.39369863271713257, |
| "eval_hard_accuracy": 0.9109243697478991, |
| "eval_hard_runtime": 0.1559, |
| "eval_hard_samples_per_second": 7631.639, |
| "eval_hard_steps_per_second": 121.85, |
| "epoch": 2.935723114956737, |
| "step": 19000 |
| }, |
| { |
| "loss": 0.07134580135345459, |
| "grad_norm": 0.48950284719467163, |
| "learning_rate": 9.740356627385149e-08, |
| "epoch": 2.9666254635352285, |
| "step": 19200 |
| }, |
| { |
| "loss": 0.07458440780639648, |
| "grad_norm": 4.208376407623291, |
| "learning_rate": 5.978606257561746e-10, |
| "epoch": 2.9975278121137205, |
| "step": 19400 |
| }, |
| { |
| "eval_unseen_lanes_loss": 0.2913234233856201, |
| "eval_unseen_lanes_accuracy": 0.9339125038783742, |
| "eval_unseen_lanes_runtime": 0.2524, |
| "eval_unseen_lanes_samples_per_second": 12768.441, |
| "eval_unseen_lanes_steps_per_second": 202.045, |
| "epoch": 3.0, |
| "step": 19416 |
| }, |
| { |
| "eval_unseen_domain_loss": 1.4579074382781982, |
| "eval_unseen_domain_accuracy": 0.7015119655552218, |
| "eval_unseen_domain_runtime": 0.5345, |
| "eval_unseen_domain_samples_per_second": 18684.464, |
| "eval_unseen_domain_steps_per_second": 293.728, |
| "epoch": 3.0, |
| "step": 19416 |
| }, |
| { |
| "eval_unseen_axis_loss": 2.300905466079712, |
| "eval_unseen_axis_accuracy": 0.599388379204893, |
| "eval_unseen_axis_runtime": 0.6288, |
| "eval_unseen_axis_samples_per_second": 19762.884, |
| "eval_unseen_axis_steps_per_second": 310.137, |
| "epoch": 3.0, |
| "step": 19416 |
| }, |
| { |
| "eval_hard_loss": 0.39358070492744446, |
| "eval_hard_accuracy": 0.9117647058823529, |
| "eval_hard_runtime": 0.1501, |
| "eval_hard_samples_per_second": 7930.316, |
| "eval_hard_steps_per_second": 126.618, |
| "epoch": 3.0, |
| "step": 19416 |
| }, |
| { |
| "train_runtime": 170.3986, |
| "train_samples_per_second": 3646.104, |
| "train_steps_per_second": 113.945, |
| "total_flos": 0.0, |
| "train_loss": 0.2352386224242852, |
| "epoch": 3.0, |
| "step": 19416 |
| } |
| ] |
| } |
