| { | |
| "best_metric": 0.021864905953407288, | |
| "best_model_checkpoint": "TASK2/model1\\checkpoint-60236", | |
| "epoch": 2.0, | |
| "global_step": 60236, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.983398632047281e-05, | |
| "loss": 0.3368, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9667972640945615e-05, | |
| "loss": 0.0613, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9501958961418424e-05, | |
| "loss": 0.0552, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.933594528189123e-05, | |
| "loss": 0.038, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9169931602364037e-05, | |
| "loss": 0.0505, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9003917922836842e-05, | |
| "loss": 0.0558, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.883790424330965e-05, | |
| "loss": 0.0377, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.867189056378246e-05, | |
| "loss": 0.0452, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8505876884255264e-05, | |
| "loss": 0.0395, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.833986320472807e-05, | |
| "loss": 0.0446, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8173849525200878e-05, | |
| "loss": 0.048, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8007835845673686e-05, | |
| "loss": 0.0345, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.784182216614649e-05, | |
| "loss": 0.0436, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.76758084866193e-05, | |
| "loss": 0.0486, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7509794807092108e-05, | |
| "loss": 0.0529, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7343781127564913e-05, | |
| "loss": 0.03, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7177767448037718e-05, | |
| "loss": 0.0445, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7011753768510527e-05, | |
| "loss": 0.0265, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6845740088983335e-05, | |
| "loss": 0.0506, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.667972640945614e-05, | |
| "loss": 0.0476, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.651371272992895e-05, | |
| "loss": 0.0385, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6347699050401754e-05, | |
| "loss": 0.0424, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.6181685370874562e-05, | |
| "loss": 0.0259, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.6015671691347367e-05, | |
| "loss": 0.0382, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.5849658011820176e-05, | |
| "loss": 0.0246, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.568364433229298e-05, | |
| "loss": 0.0211, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.551763065276579e-05, | |
| "loss": 0.0374, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.5351616973238597e-05, | |
| "loss": 0.0263, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5185603293711404e-05, | |
| "loss": 0.0396, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.501958961418421e-05, | |
| "loss": 0.0275, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4853575934657016e-05, | |
| "loss": 0.0251, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4687562255129823e-05, | |
| "loss": 0.0368, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4521548575602631e-05, | |
| "loss": 0.0322, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4355534896075438e-05, | |
| "loss": 0.0285, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4189521216548245e-05, | |
| "loss": 0.0298, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4023507537021053e-05, | |
| "loss": 0.0425, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3857493857493858e-05, | |
| "loss": 0.0307, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.3691480177966665e-05, | |
| "loss": 0.0233, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.3525466498439472e-05, | |
| "loss": 0.0246, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.335945281891228e-05, | |
| "loss": 0.0315, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3193439139385087e-05, | |
| "loss": 0.0315, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.3027425459857894e-05, | |
| "loss": 0.0329, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.2861411780330699e-05, | |
| "loss": 0.0258, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.2695398100803507e-05, | |
| "loss": 0.0274, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2529384421276314e-05, | |
| "loss": 0.0273, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.236337074174912e-05, | |
| "loss": 0.0309, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.219735706222193e-05, | |
| "loss": 0.0255, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2031343382694736e-05, | |
| "loss": 0.0331, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1865329703167543e-05, | |
| "loss": 0.028, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1699316023640348e-05, | |
| "loss": 0.0273, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1533302344113154e-05, | |
| "loss": 0.0222, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1367288664585963e-05, | |
| "loss": 0.0308, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.120127498505877e-05, | |
| "loss": 0.0363, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1035261305531576e-05, | |
| "loss": 0.0324, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0869247626004385e-05, | |
| "loss": 0.0221, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.070323394647719e-05, | |
| "loss": 0.0199, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0537220266949997e-05, | |
| "loss": 0.0256, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0371206587422803e-05, | |
| "loss": 0.0294, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0205192907895612e-05, | |
| "loss": 0.0254, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0039179228368419e-05, | |
| "loss": 0.0314, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9955094252722309, | |
| "eval_loss": 0.023077525198459625, | |
| "eval_runtime": 2150.2467, | |
| "eval_samples_per_second": 28.273, | |
| "eval_steps_per_second": 3.534, | |
| "step": 30118 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.873165548841225e-06, | |
| "loss": 0.0142, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.707151869314032e-06, | |
| "loss": 0.0214, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.541138189786839e-06, | |
| "loss": 0.0092, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.375124510259647e-06, | |
| "loss": 0.0124, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.209110830732452e-06, | |
| "loss": 0.0151, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.043097151205261e-06, | |
| "loss": 0.012, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.877083471678068e-06, | |
| "loss": 0.017, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.711069792150874e-06, | |
| "loss": 0.0205, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.545056112623681e-06, | |
| "loss": 0.0205, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.379042433096488e-06, | |
| "loss": 0.0205, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.213028753569295e-06, | |
| "loss": 0.0123, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.047015074042101e-06, | |
| "loss": 0.0149, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.881001394514908e-06, | |
| "loss": 0.0142, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.714987714987717e-06, | |
| "loss": 0.0148, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.5489740354605225e-06, | |
| "loss": 0.0093, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.382960355933329e-06, | |
| "loss": 0.0111, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.216946676406137e-06, | |
| "loss": 0.0163, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.050932996878943e-06, | |
| "loss": 0.0198, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.88491931735175e-06, | |
| "loss": 0.0146, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.718905637824558e-06, | |
| "loss": 0.0252, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.552891958297364e-06, | |
| "loss": 0.0242, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.3868782787701715e-06, | |
| "loss": 0.0096, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.220864599242978e-06, | |
| "loss": 0.0193, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.054850919715784e-06, | |
| "loss": 0.0143, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.888837240188592e-06, | |
| "loss": 0.0156, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.722823560661399e-06, | |
| "loss": 0.0129, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.556809881134206e-06, | |
| "loss": 0.0151, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.390796201607013e-06, | |
| "loss": 0.0063, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.2247825220798196e-06, | |
| "loss": 0.0137, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.058768842552627e-06, | |
| "loss": 0.0108, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.892755163025434e-06, | |
| "loss": 0.0212, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.726741483498241e-06, | |
| "loss": 0.0111, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.560727803971048e-06, | |
| "loss": 0.011, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.394714124443854e-06, | |
| "loss": 0.007, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.228700444916661e-06, | |
| "loss": 0.0117, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.0626867653894685e-06, | |
| "loss": 0.0078, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.896673085862275e-06, | |
| "loss": 0.0058, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.730659406335082e-06, | |
| "loss": 0.0133, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.5646457268078892e-06, | |
| "loss": 0.0097, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.398632047280696e-06, | |
| "loss": 0.0111, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.232618367753503e-06, | |
| "loss": 0.016, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.06660468822631e-06, | |
| "loss": 0.0147, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.9005910086991167e-06, | |
| "loss": 0.0112, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.7345773291719243e-06, | |
| "loss": 0.0077, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.568563649644731e-06, | |
| "loss": 0.0123, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.4025499701175378e-06, | |
| "loss": 0.0102, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.236536290590345e-06, | |
| "loss": 0.0109, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0705226110631517e-06, | |
| "loss": 0.01, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9045089315359589e-06, | |
| "loss": 0.0161, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.7384952520087656e-06, | |
| "loss": 0.0104, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.5724815724815726e-06, | |
| "loss": 0.008, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.4064678929543795e-06, | |
| "loss": 0.0101, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.2404542134271865e-06, | |
| "loss": 0.0093, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.0744405338999935e-06, | |
| "loss": 0.0155, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.084268543728004e-07, | |
| "loss": 0.0109, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.424131748456073e-07, | |
| "loss": 0.008, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.763994953184143e-07, | |
| "loss": 0.0108, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.103858157912213e-07, | |
| "loss": 0.0155, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.4437213626402814e-07, | |
| "loss": 0.0079, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.835845673683512e-08, | |
| "loss": 0.0113, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9959042010724742, | |
| "eval_loss": 0.021864905953407288, | |
| "eval_runtime": 2302.6637, | |
| "eval_samples_per_second": 26.402, | |
| "eval_steps_per_second": 3.301, | |
| "step": 60236 | |
| } | |
| ], | |
| "max_steps": 60236, | |
| "num_train_epochs": 2, | |
| "total_flos": 6.384188864533709e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |