| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.262571103526735, | |
| "eval_steps": 5000, | |
| "global_step": 55048, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11376564277588168, | |
| "grad_norm": 4.343299865722656, | |
| "learning_rate": 7.960000000000001e-05, | |
| "loss": 4.3757, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22753128555176336, | |
| "grad_norm": 3.4980385303497314, | |
| "learning_rate": 0.0001596, | |
| "loss": 1.7997, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.3412969283276451, | |
| "grad_norm": 3.350770950317383, | |
| "learning_rate": 0.0002396, | |
| "loss": 1.4879, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.4550625711035267, | |
| "grad_norm": 3.362269163131714, | |
| "learning_rate": 0.0003196, | |
| "loss": 1.3834, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.5688282138794084, | |
| "grad_norm": 2.424733877182007, | |
| "learning_rate": 0.0003996, | |
| "loss": 1.3351, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5688282138794084, | |
| "eval_accuracy": 0.668216, | |
| "eval_loss": 1.3343349695205688, | |
| "eval_runtime": 15.5029, | |
| "eval_samples_per_second": 16125.98, | |
| "eval_steps_per_second": 31.542, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6825938566552902, | |
| "grad_norm": 2.601680040359497, | |
| "learning_rate": 0.00047960000000000006, | |
| "loss": 1.3097, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.7963594994311718, | |
| "grad_norm": 1.9989418983459473, | |
| "learning_rate": 0.0005596, | |
| "loss": 1.2866, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.9101251422070534, | |
| "grad_norm": 1.8081185817718506, | |
| "learning_rate": 0.0006396, | |
| "loss": 1.2689, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.023890784982935, | |
| "grad_norm": 1.6405588388442993, | |
| "learning_rate": 0.00071952, | |
| "loss": 1.2487, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.1376564277588168, | |
| "grad_norm": 1.22613525390625, | |
| "learning_rate": 0.00079952, | |
| "loss": 1.2165, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.1376564277588168, | |
| "eval_accuracy": 0.686196, | |
| "eval_loss": 1.2481120824813843, | |
| "eval_runtime": 15.8446, | |
| "eval_samples_per_second": 15778.257, | |
| "eval_steps_per_second": 30.862, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.2514220705346986, | |
| "grad_norm": 1.2695167064666748, | |
| "learning_rate": 0.0007996786565611985, | |
| "loss": 1.2046, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.36518771331058, | |
| "grad_norm": 1.2521305084228516, | |
| "learning_rate": 0.0007987086748436788, | |
| "loss": 1.1849, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.4789533560864618, | |
| "grad_norm": 1.19619619846344, | |
| "learning_rate": 0.0007970896788508052, | |
| "loss": 1.1534, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.5927189988623436, | |
| "grad_norm": 1.0483107566833496, | |
| "learning_rate": 0.0007948275336376884, | |
| "loss": 1.1312, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.7064846416382253, | |
| "grad_norm": 1.3618515729904175, | |
| "learning_rate": 0.0007919213896323948, | |
| "loss": 1.112, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.7064846416382253, | |
| "eval_accuracy": 0.716556, | |
| "eval_loss": 1.1176625490188599, | |
| "eval_runtime": 15.2386, | |
| "eval_samples_per_second": 16405.688, | |
| "eval_steps_per_second": 32.09, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.820250284414107, | |
| "grad_norm": 0.9307771325111389, | |
| "learning_rate": 0.0007883817747762077, | |
| "loss": 1.0986, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.9340159271899886, | |
| "grad_norm": 0.9189246296882629, | |
| "learning_rate": 0.0007842073597303121, | |
| "loss": 1.0847, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.04778156996587, | |
| "grad_norm": 0.8016377687454224, | |
| "learning_rate": 0.0007794081581686037, | |
| "loss": 1.0506, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.161547212741752, | |
| "grad_norm": 0.9774219989776611, | |
| "learning_rate": 0.0007739919744091065, | |
| "loss": 1.0158, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.2753128555176336, | |
| "grad_norm": 0.8449124693870544, | |
| "learning_rate": 0.0007679676160878387, | |
| "loss": 1.0138, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.2753128555176336, | |
| "eval_accuracy": 0.732636, | |
| "eval_loss": 1.052935242652893, | |
| "eval_runtime": 15.2399, | |
| "eval_samples_per_second": 16404.275, | |
| "eval_steps_per_second": 32.087, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.3890784982935154, | |
| "grad_norm": 0.9211858510971069, | |
| "learning_rate": 0.0007613448798360993, | |
| "loss": 1.0113, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.502844141069397, | |
| "grad_norm": 0.7870326042175293, | |
| "learning_rate": 0.0007541345353494786, | |
| "loss": 1.0024, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.616609783845279, | |
| "grad_norm": 0.8683303594589233, | |
| "learning_rate": 0.0007463483078745015, | |
| "loss": 1.0032, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.73037542662116, | |
| "grad_norm": 1.031267523765564, | |
| "learning_rate": 0.000738007485475254, | |
| "loss": 0.9961, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.8441410693970424, | |
| "grad_norm": 0.8440726399421692, | |
| "learning_rate": 0.0007291089356699791, | |
| "loss": 0.9909, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.8441410693970424, | |
| "eval_accuracy": 0.741524, | |
| "eval_loss": 1.0143921375274658, | |
| "eval_runtime": 14.9299, | |
| "eval_samples_per_second": 16744.96, | |
| "eval_steps_per_second": 32.753, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.9579067121729237, | |
| "grad_norm": 0.708281934261322, | |
| "learning_rate": 0.0007196848947861554, | |
| "loss": 0.9832, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 3.0716723549488054, | |
| "grad_norm": 0.692997395992279, | |
| "learning_rate": 0.000709742030952583, | |
| "loss": 0.9383, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 3.185437997724687, | |
| "grad_norm": 0.8190609216690063, | |
| "learning_rate": 0.0006992857783851634, | |
| "loss": 0.9193, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 3.299203640500569, | |
| "grad_norm": 0.7791016697883606, | |
| "learning_rate": 0.0006883428362373026, | |
| "loss": 0.9197, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 3.4129692832764507, | |
| "grad_norm": 0.6834008693695068, | |
| "learning_rate": 0.0006769309995941914, | |
| "loss": 0.9236, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 3.4129692832764507, | |
| "eval_accuracy": 0.748324, | |
| "eval_loss": 0.9886829257011414, | |
| "eval_runtime": 14.8544, | |
| "eval_samples_per_second": 16830.07, | |
| "eval_steps_per_second": 32.92, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 3.526734926052332, | |
| "grad_norm": 0.7945353388786316, | |
| "learning_rate": 0.0006650809067991791, | |
| "loss": 0.9259, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 3.640500568828214, | |
| "grad_norm": 0.7771942019462585, | |
| "learning_rate": 0.000652788107427868, | |
| "loss": 0.924, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 3.7542662116040955, | |
| "grad_norm": 0.7232080101966858, | |
| "learning_rate": 0.0006400842315977677, | |
| "loss": 0.9149, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 3.868031854379977, | |
| "grad_norm": 0.6129056215286255, | |
| "learning_rate": 0.0006270032202430253, | |
| "loss": 0.9142, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 3.981797497155859, | |
| "grad_norm": 0.7053471803665161, | |
| "learning_rate": 0.0006135401606551002, | |
| "loss": 0.914, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.981797497155859, | |
| "eval_accuracy": 0.754936, | |
| "eval_loss": 0.9585933089256287, | |
| "eval_runtime": 15.1971, | |
| "eval_samples_per_second": 16450.463, | |
| "eval_steps_per_second": 32.177, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 4.09556313993174, | |
| "grad_norm": 0.7185536623001099, | |
| "learning_rate": 0.0005997438247807972, | |
| "loss": 0.8508, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 4.2093287827076225, | |
| "grad_norm": 0.7729761600494385, | |
| "learning_rate": 0.0005856090312640852, | |
| "loss": 0.8434, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 4.323094425483504, | |
| "grad_norm": 0.8405170440673828, | |
| "learning_rate": 0.0005711724058927512, | |
| "loss": 0.8442, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 4.436860068259386, | |
| "grad_norm": 0.6555745005607605, | |
| "learning_rate": 0.0005564574250751392, | |
| "loss": 0.852, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 4.550625711035267, | |
| "grad_norm": 0.6500961184501648, | |
| "learning_rate": 0.0005415031062964693, | |
| "loss": 0.849, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 4.550625711035267, | |
| "eval_accuracy": 0.758092, | |
| "eval_loss": 0.9504426121711731, | |
| "eval_runtime": 14.8523, | |
| "eval_samples_per_second": 16832.413, | |
| "eval_steps_per_second": 32.924, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 4.664391353811149, | |
| "grad_norm": 0.6345399618148804, | |
| "learning_rate": 0.0005263038333083039, | |
| "loss": 0.8493, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 4.778156996587031, | |
| "grad_norm": 0.7285176515579224, | |
| "learning_rate": 0.0005108991688044689, | |
| "loss": 0.8505, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 4.891922639362912, | |
| "grad_norm": 0.7443712949752808, | |
| "learning_rate": 0.000495314163389589, | |
| "loss": 0.8515, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 5.005688282138794, | |
| "grad_norm": 0.6659076809883118, | |
| "learning_rate": 0.0004795899698565036, | |
| "loss": 0.8403, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 5.1194539249146755, | |
| "grad_norm": 0.7172214984893799, | |
| "learning_rate": 0.00046372068282238195, | |
| "loss": 0.7614, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 5.1194539249146755, | |
| "eval_accuracy": 0.761088, | |
| "eval_loss": 0.9495302438735962, | |
| "eval_runtime": 15.3396, | |
| "eval_samples_per_second": 16297.69, | |
| "eval_steps_per_second": 31.878, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 5.233219567690558, | |
| "grad_norm": 0.8739346861839294, | |
| "learning_rate": 0.0004477477751198958, | |
| "loss": 0.767, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 5.346985210466439, | |
| "grad_norm": 0.8742613196372986, | |
| "learning_rate": 0.0004316972214137623, | |
| "loss": 0.7723, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 5.460750853242321, | |
| "grad_norm": 0.7512331604957581, | |
| "learning_rate": 0.00041559512263430705, | |
| "loss": 0.7738, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 5.5745164960182025, | |
| "grad_norm": 0.7592815160751343, | |
| "learning_rate": 0.00039948379493191056, | |
| "loss": 0.7735, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 5.688282138794084, | |
| "grad_norm": 0.6751989722251892, | |
| "learning_rate": 0.00038335718753151784, | |
| "loss": 0.7726, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 5.688282138794084, | |
| "eval_accuracy": 0.763736, | |
| "eval_loss": 0.9361330270767212, | |
| "eval_runtime": 17.5361, | |
| "eval_samples_per_second": 14256.306, | |
| "eval_steps_per_second": 27.885, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 5.802047781569966, | |
| "grad_norm": 0.7502247095108032, | |
| "learning_rate": 0.0003672737214802269, | |
| "loss": 0.7735, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 5.915813424345847, | |
| "grad_norm": 0.7880488038063049, | |
| "learning_rate": 0.000351227356466713, | |
| "loss": 0.7736, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 6.0295790671217295, | |
| "grad_norm": 0.7424056529998779, | |
| "learning_rate": 0.0003352762228480271, | |
| "loss": 0.7455, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 6.143344709897611, | |
| "grad_norm": 0.8073525428771973, | |
| "learning_rate": 0.0003194143300116524, | |
| "loss": 0.6797, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 6.257110352673493, | |
| "grad_norm": 0.7268177270889282, | |
| "learning_rate": 0.00030369914003658996, | |
| "loss": 0.6867, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 6.257110352673493, | |
| "eval_accuracy": 0.763196, | |
| "eval_loss": 0.9648858308792114, | |
| "eval_runtime": 15.5355, | |
| "eval_samples_per_second": 16092.147, | |
| "eval_steps_per_second": 31.476, | |
| "step": 55000 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 87900, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 5000, | |
| "total_flos": 2.354556845700649e+18, | |
| "train_batch_size": 512, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |