| { | |
| "best_metric": 0.9456379413604736, | |
| "best_model_checkpoint": "data/Gemma-2-2B_task-3_180-samples_config-2_full/checkpoint-76", | |
| "epoch": 16.0, | |
| "eval_steps": 500, | |
| "global_step": 136, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11764705882352941, | |
| "grad_norm": 0.43580687046051025, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.3992, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 0.46208158135414124, | |
| "learning_rate": 5e-06, | |
| "loss": 1.4171, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 0.43559974431991577, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4244, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.7058823529411765, | |
| "grad_norm": 0.40987280011177063, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.3588, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "grad_norm": 0.3987618684768677, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3981, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "eval_loss": 1.3785014152526855, | |
| "eval_runtime": 11.296, | |
| "eval_samples_per_second": 3.187, | |
| "eval_steps_per_second": 3.187, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.1764705882352942, | |
| "grad_norm": 0.4353489577770233, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.4094, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.4117647058823528, | |
| "grad_norm": 0.3717004060745239, | |
| "learning_rate": 3e-05, | |
| "loss": 1.3371, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.6470588235294117, | |
| "grad_norm": 0.3343724012374878, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.307, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.8823529411764706, | |
| "grad_norm": 0.3171851933002472, | |
| "learning_rate": 4e-05, | |
| "loss": 1.3072, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.2429279088974, | |
| "eval_runtime": 11.2872, | |
| "eval_samples_per_second": 3.189, | |
| "eval_steps_per_second": 3.189, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 2.1176470588235294, | |
| "grad_norm": 0.25433221459388733, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.2631, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 2.3529411764705883, | |
| "grad_norm": 0.22598546743392944, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1956, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.588235294117647, | |
| "grad_norm": 0.24507024884223938, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 1.1856, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.8235294117647056, | |
| "grad_norm": 0.2662090063095093, | |
| "learning_rate": 6e-05, | |
| "loss": 1.1985, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.9411764705882355, | |
| "eval_loss": 1.1441607475280762, | |
| "eval_runtime": 11.2937, | |
| "eval_samples_per_second": 3.188, | |
| "eval_steps_per_second": 3.188, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 3.0588235294117645, | |
| "grad_norm": 0.25578731298446655, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 1.1262, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 3.2941176470588234, | |
| "grad_norm": 0.22478534281253815, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0913, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 3.5294117647058822, | |
| "grad_norm": 0.255871057510376, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 1.0533, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.764705882352941, | |
| "grad_norm": 0.4229178726673126, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0783, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.23409676551818848, | |
| "learning_rate": 8.5e-05, | |
| "loss": 0.9971, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.0311872959136963, | |
| "eval_runtime": 11.3103, | |
| "eval_samples_per_second": 3.183, | |
| "eval_steps_per_second": 3.183, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 4.235294117647059, | |
| "grad_norm": 0.22290097177028656, | |
| "learning_rate": 9e-05, | |
| "loss": 0.9928, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 4.470588235294118, | |
| "grad_norm": 0.19320537149906158, | |
| "learning_rate": 9.5e-05, | |
| "loss": 0.9715, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 4.705882352941177, | |
| "grad_norm": 0.15621793270111084, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9689, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 4.9411764705882355, | |
| "grad_norm": 0.17081919312477112, | |
| "learning_rate": 9.999238475781957e-05, | |
| "loss": 0.9268, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 4.9411764705882355, | |
| "eval_loss": 0.9882259368896484, | |
| "eval_runtime": 11.3101, | |
| "eval_samples_per_second": 3.183, | |
| "eval_steps_per_second": 3.183, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 5.176470588235294, | |
| "grad_norm": 0.20606370270252228, | |
| "learning_rate": 9.99695413509548e-05, | |
| "loss": 0.9067, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 5.411764705882353, | |
| "grad_norm": 0.14960861206054688, | |
| "learning_rate": 9.99314767377287e-05, | |
| "loss": 0.9299, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 5.647058823529412, | |
| "grad_norm": 0.14424344897270203, | |
| "learning_rate": 9.987820251299122e-05, | |
| "loss": 0.9017, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 5.882352941176471, | |
| "grad_norm": 0.15432150661945343, | |
| "learning_rate": 9.980973490458728e-05, | |
| "loss": 0.9442, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 0.9652664661407471, | |
| "eval_runtime": 11.2872, | |
| "eval_samples_per_second": 3.189, | |
| "eval_steps_per_second": 3.189, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 6.117647058823529, | |
| "grad_norm": 0.14191652834415436, | |
| "learning_rate": 9.972609476841367e-05, | |
| "loss": 0.8809, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 6.352941176470588, | |
| "grad_norm": 0.14149774610996246, | |
| "learning_rate": 9.962730758206611e-05, | |
| "loss": 0.8682, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 6.588235294117647, | |
| "grad_norm": 0.15507805347442627, | |
| "learning_rate": 9.951340343707852e-05, | |
| "loss": 0.893, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 6.823529411764706, | |
| "grad_norm": 0.14243623614311218, | |
| "learning_rate": 9.938441702975689e-05, | |
| "loss": 0.9253, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 6.9411764705882355, | |
| "eval_loss": 0.9536962509155273, | |
| "eval_runtime": 11.2984, | |
| "eval_samples_per_second": 3.186, | |
| "eval_steps_per_second": 3.186, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 7.0588235294117645, | |
| "grad_norm": 0.14463533461093903, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 0.8901, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 7.294117647058823, | |
| "grad_norm": 0.15607739984989166, | |
| "learning_rate": 9.908135917238321e-05, | |
| "loss": 0.873, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 7.529411764705882, | |
| "grad_norm": 0.16006982326507568, | |
| "learning_rate": 9.890738003669029e-05, | |
| "loss": 0.8446, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 7.764705882352941, | |
| "grad_norm": 0.1541738659143448, | |
| "learning_rate": 9.871850323926177e-05, | |
| "loss": 0.8343, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.18918535113334656, | |
| "learning_rate": 9.851478631379982e-05, | |
| "loss": 0.8684, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 0.9479315876960754, | |
| "eval_runtime": 11.3029, | |
| "eval_samples_per_second": 3.185, | |
| "eval_steps_per_second": 3.185, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 8.235294117647058, | |
| "grad_norm": 0.1631922721862793, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 0.8289, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 8.470588235294118, | |
| "grad_norm": 0.17181901633739471, | |
| "learning_rate": 9.806308479691595e-05, | |
| "loss": 0.8583, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 8.705882352941176, | |
| "grad_norm": 0.17497360706329346, | |
| "learning_rate": 9.781523779815179e-05, | |
| "loss": 0.8415, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 8.941176470588236, | |
| "grad_norm": 0.19706816971302032, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.8043, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 8.941176470588236, | |
| "eval_loss": 0.9456379413604736, | |
| "eval_runtime": 11.2924, | |
| "eval_samples_per_second": 3.188, | |
| "eval_steps_per_second": 3.188, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 9.176470588235293, | |
| "grad_norm": 0.1770046502351761, | |
| "learning_rate": 9.727592877996585e-05, | |
| "loss": 0.8124, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 9.411764705882353, | |
| "grad_norm": 0.18705680966377258, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 0.811, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 9.647058823529411, | |
| "grad_norm": 0.20000538229942322, | |
| "learning_rate": 9.667902132486009e-05, | |
| "loss": 0.8088, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 9.882352941176471, | |
| "grad_norm": 0.22094132006168365, | |
| "learning_rate": 9.635919272833938e-05, | |
| "loss": 0.7924, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 0.9502108097076416, | |
| "eval_runtime": 11.2942, | |
| "eval_samples_per_second": 3.187, | |
| "eval_steps_per_second": 3.187, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 10.117647058823529, | |
| "grad_norm": 0.22133776545524597, | |
| "learning_rate": 9.602524267262203e-05, | |
| "loss": 0.7704, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 10.352941176470589, | |
| "grad_norm": 0.23369313776493073, | |
| "learning_rate": 9.567727288213005e-05, | |
| "loss": 0.7875, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 10.588235294117647, | |
| "grad_norm": 0.2591407895088196, | |
| "learning_rate": 9.53153893518325e-05, | |
| "loss": 0.7537, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 10.823529411764707, | |
| "grad_norm": 0.27133455872535706, | |
| "learning_rate": 9.493970231495835e-05, | |
| "loss": 0.7535, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 10.941176470588236, | |
| "eval_loss": 0.9591252207756042, | |
| "eval_runtime": 11.2879, | |
| "eval_samples_per_second": 3.189, | |
| "eval_steps_per_second": 3.189, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 11.058823529411764, | |
| "grad_norm": 0.2570657730102539, | |
| "learning_rate": 9.45503262094184e-05, | |
| "loss": 0.7897, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 11.294117647058824, | |
| "grad_norm": 0.2605714499950409, | |
| "learning_rate": 9.414737964294636e-05, | |
| "loss": 0.7747, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 11.529411764705882, | |
| "grad_norm": 0.4368426501750946, | |
| "learning_rate": 9.373098535696979e-05, | |
| "loss": 0.7026, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 11.764705882352942, | |
| "grad_norm": 0.3336321711540222, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 0.7311, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 0.34348052740097046, | |
| "learning_rate": 9.285836503510562e-05, | |
| "loss": 0.694, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_loss": 0.986273467540741, | |
| "eval_runtime": 11.2808, | |
| "eval_samples_per_second": 3.191, | |
| "eval_steps_per_second": 3.191, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 12.235294117647058, | |
| "grad_norm": 0.42941805720329285, | |
| "learning_rate": 9.24024048078213e-05, | |
| "loss": 0.6778, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 12.470588235294118, | |
| "grad_norm": 0.4540957808494568, | |
| "learning_rate": 9.193352839727121e-05, | |
| "loss": 0.6747, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 12.705882352941176, | |
| "grad_norm": 0.45547229051589966, | |
| "learning_rate": 9.145187862775209e-05, | |
| "loss": 0.6948, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 12.941176470588236, | |
| "grad_norm": 0.5169070959091187, | |
| "learning_rate": 9.09576022144496e-05, | |
| "loss": 0.6881, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 12.941176470588236, | |
| "eval_loss": 0.9993712902069092, | |
| "eval_runtime": 11.2809, | |
| "eval_samples_per_second": 3.191, | |
| "eval_steps_per_second": 3.191, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 13.176470588235293, | |
| "grad_norm": 0.45654669404029846, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.6667, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 13.411764705882353, | |
| "grad_norm": 0.5385121703147888, | |
| "learning_rate": 8.993177550236464e-05, | |
| "loss": 0.5996, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 13.647058823529411, | |
| "grad_norm": 0.5320378541946411, | |
| "learning_rate": 8.940053768033609e-05, | |
| "loss": 0.6385, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 13.882352941176471, | |
| "grad_norm": 0.5594288110733032, | |
| "learning_rate": 8.885729807284856e-05, | |
| "loss": 0.6566, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_loss": 1.0534077882766724, | |
| "eval_runtime": 11.2778, | |
| "eval_samples_per_second": 3.192, | |
| "eval_steps_per_second": 3.192, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 14.117647058823529, | |
| "grad_norm": 0.4900166690349579, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 0.6437, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 14.352941176470589, | |
| "grad_norm": 0.7026910781860352, | |
| "learning_rate": 8.773547901113862e-05, | |
| "loss": 0.538, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 14.588235294117647, | |
| "grad_norm": 0.6666274666786194, | |
| "learning_rate": 8.715724127386972e-05, | |
| "loss": 0.6209, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 14.823529411764707, | |
| "grad_norm": 0.7703569531440735, | |
| "learning_rate": 8.656768508095853e-05, | |
| "loss": 0.5597, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 14.941176470588236, | |
| "eval_loss": 1.111732006072998, | |
| "eval_runtime": 11.2952, | |
| "eval_samples_per_second": 3.187, | |
| "eval_steps_per_second": 3.187, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 15.058823529411764, | |
| "grad_norm": 0.6548103094100952, | |
| "learning_rate": 8.596699001693255e-05, | |
| "loss": 0.6079, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 15.294117647058824, | |
| "grad_norm": 0.7128686904907227, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 0.543, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 15.529411764705882, | |
| "grad_norm": 0.7173427939414978, | |
| "learning_rate": 8.473291852294987e-05, | |
| "loss": 0.5608, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 15.764705882352942, | |
| "grad_norm": 0.7026836276054382, | |
| "learning_rate": 8.409991800312493e-05, | |
| "loss": 0.5659, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "grad_norm": 0.7972676157951355, | |
| "learning_rate": 8.345653031794292e-05, | |
| "loss": 0.497, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_loss": 1.1691418886184692, | |
| "eval_runtime": 11.309, | |
| "eval_samples_per_second": 3.183, | |
| "eval_steps_per_second": 3.183, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "step": 136, | |
| "total_flos": 5.868075337383936e+16, | |
| "train_loss": 0.886881584630293, | |
| "train_runtime": 2470.3442, | |
| "train_samples_per_second": 2.753, | |
| "train_steps_per_second": 0.162 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 400, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 7, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.868075337383936e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |