| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.999133333333333, |
| "eval_steps": 500, |
| "global_step": 30000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0016666666666666668, |
| "grad_norm": 6.296432018280029, |
| "learning_rate": 2.9400000000000002e-06, |
| "loss": 9.36233642578125, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0033333333333333335, |
| "grad_norm": 2.045384407043457, |
| "learning_rate": 5.940000000000001e-06, |
| "loss": 6.69770751953125, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.005, |
| "grad_norm": 3.558886766433716, |
| "learning_rate": 8.939999999999999e-06, |
| "loss": 5.728419189453125, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.006666666666666667, |
| "grad_norm": 3.144193649291992, |
| "learning_rate": 1.1940000000000001e-05, |
| "loss": 5.055216064453125, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.008333333333333333, |
| "grad_norm": 2.3523426055908203, |
| "learning_rate": 1.4940000000000001e-05, |
| "loss": 4.644241027832031, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 2.9722487926483154, |
| "learning_rate": 1.794e-05, |
| "loss": 4.292540283203125, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.011666666666666667, |
| "grad_norm": 2.6705331802368164, |
| "learning_rate": 2.094e-05, |
| "loss": 3.987220458984375, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.013333333333333334, |
| "grad_norm": 2.841386318206787, |
| "learning_rate": 2.394e-05, |
| "loss": 3.8409832763671874, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.0002333333333333, |
| "grad_norm": 2.705993890762329, |
| "learning_rate": 2.6940000000000003e-05, |
| "loss": 3.7106192016601565, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.0019, |
| "grad_norm": 1.8092248439788818, |
| "learning_rate": 2.994e-05, |
| "loss": 3.5095693969726565, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.0035666666666667, |
| "grad_norm": 1.9855880737304688, |
| "learning_rate": 2.9999795775354255e-05, |
| "loss": 3.3909701538085937, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.0052333333333334, |
| "grad_norm": 1.4056228399276733, |
| "learning_rate": 2.999916635078951e-05, |
| "loss": 3.2903216552734373, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.0069, |
| "grad_norm": 1.8909028768539429, |
| "learning_rate": 2.9998111659093307e-05, |
| "loss": 3.177409362792969, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.0085666666666666, |
| "grad_norm": 1.9932523965835571, |
| "learning_rate": 2.999663173016903e-05, |
| "loss": 3.1271542358398436, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.0102333333333333, |
| "grad_norm": 1.831948161125183, |
| "learning_rate": 2.9994726605976675e-05, |
| "loss": 3.0346929931640627, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.0119, |
| "grad_norm": 1.6663613319396973, |
| "learning_rate": 2.9992396340531704e-05, |
| "loss": 2.9763214111328127, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.0135666666666667, |
| "grad_norm": 1.478272795677185, |
| "learning_rate": 2.998964099990346e-05, |
| "loss": 2.957916564941406, |
| "step": 850 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.7375118732452393, |
| "learning_rate": 2.998646066221335e-05, |
| "loss": 2.8912548828125, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.5892730951309204, |
| "learning_rate": 2.9982855417632584e-05, |
| "loss": 2.7499594116210937, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.5600744485855103, |
| "learning_rate": 2.9978825368379653e-05, |
| "loss": 2.7288330078125, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.6788824796676636, |
| "learning_rate": 2.9974370628717398e-05, |
| "loss": 2.713819885253906, |
| "step": 1050 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.7233264446258545, |
| "learning_rate": 2.9969491324949794e-05, |
| "loss": 2.6783953857421876, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.5754250288009644, |
| "learning_rate": 2.9964187595418376e-05, |
| "loss": 2.6226959228515625, |
| "step": 1150 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.48687744140625, |
| "learning_rate": 2.9958459590498285e-05, |
| "loss": 2.5779833984375, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.7073004245758057, |
| "learning_rate": 2.9952307472594024e-05, |
| "loss": 2.5385700988769533, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 2.0446057319641113, |
| "learning_rate": 2.994573141613486e-05, |
| "loss": 2.52190673828125, |
| "step": 1300 |
| }, |
| { |
| "epoch": 3.0004333333333335, |
| "grad_norm": 1.5144495964050293, |
| "learning_rate": 2.9938731607569865e-05, |
| "loss": 2.5131741333007813, |
| "step": 1350 |
| }, |
| { |
| "epoch": 3.0021, |
| "grad_norm": 1.5741095542907715, |
| "learning_rate": 2.993130824536264e-05, |
| "loss": 2.440640411376953, |
| "step": 1400 |
| }, |
| { |
| "epoch": 3.0037666666666665, |
| "grad_norm": 1.5569274425506592, |
| "learning_rate": 2.992346153998567e-05, |
| "loss": 2.392763214111328, |
| "step": 1450 |
| }, |
| { |
| "epoch": 3.0054333333333334, |
| "grad_norm": 1.5848331451416016, |
| "learning_rate": 2.991519171391439e-05, |
| "loss": 2.3313829040527345, |
| "step": 1500 |
| }, |
| { |
| "epoch": 3.0071, |
| "grad_norm": 1.8444602489471436, |
| "learning_rate": 2.9906499001620838e-05, |
| "loss": 2.3318301391601564, |
| "step": 1550 |
| }, |
| { |
| "epoch": 3.008766666666667, |
| "grad_norm": 1.575213074684143, |
| "learning_rate": 2.9897383649567032e-05, |
| "loss": 2.304534912109375, |
| "step": 1600 |
| }, |
| { |
| "epoch": 3.0104333333333333, |
| "grad_norm": 1.579645037651062, |
| "learning_rate": 2.9887845916197987e-05, |
| "loss": 2.236707305908203, |
| "step": 1650 |
| }, |
| { |
| "epoch": 3.0121, |
| "grad_norm": 1.7910284996032715, |
| "learning_rate": 2.9877886071934353e-05, |
| "loss": 2.208928985595703, |
| "step": 1700 |
| }, |
| { |
| "epoch": 3.0137666666666667, |
| "grad_norm": 1.4495810270309448, |
| "learning_rate": 2.986750439916479e-05, |
| "loss": 2.2188836669921876, |
| "step": 1750 |
| }, |
| { |
| "epoch": 4.000566666666667, |
| "grad_norm": 1.7691011428833008, |
| "learning_rate": 2.9856701192237937e-05, |
| "loss": 2.2033157348632812, |
| "step": 1800 |
| }, |
| { |
| "epoch": 4.002233333333334, |
| "grad_norm": 1.9443449974060059, |
| "learning_rate": 2.9845476757454072e-05, |
| "loss": 2.041663818359375, |
| "step": 1850 |
| }, |
| { |
| "epoch": 4.0039, |
| "grad_norm": 1.7102004289627075, |
| "learning_rate": 2.9833831413056427e-05, |
| "loss": 2.014143829345703, |
| "step": 1900 |
| }, |
| { |
| "epoch": 4.005566666666667, |
| "grad_norm": 1.77164626121521, |
| "learning_rate": 2.9821765489222165e-05, |
| "loss": 1.9876158142089844, |
| "step": 1950 |
| }, |
| { |
| "epoch": 4.007233333333334, |
| "grad_norm": 2.660857677459717, |
| "learning_rate": 2.9809279328053023e-05, |
| "loss": 1.9998252868652344, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.8511937856674194, |
| "learning_rate": 2.97963732835656e-05, |
| "loss": 1.9002688598632813, |
| "step": 2050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.6527299880981445, |
| "learning_rate": 2.978304772168134e-05, |
| "loss": 1.8423493957519532, |
| "step": 2100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.9586758613586426, |
| "learning_rate": 2.9769303020216133e-05, |
| "loss": 1.8361962890625, |
| "step": 2150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.9420219659805298, |
| "learning_rate": 2.975513956886963e-05, |
| "loss": 1.7866915893554687, |
| "step": 2200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.9848837852478027, |
| "learning_rate": 2.9740557769214157e-05, |
| "loss": 1.895875244140625, |
| "step": 2250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.8065719604492188, |
| "learning_rate": 2.9725558034683384e-05, |
| "loss": 1.966142120361328, |
| "step": 2300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.592037558555603, |
| "learning_rate": 2.9710140790560537e-05, |
| "loss": 1.9299343872070311, |
| "step": 2350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.7908258438110352, |
| "learning_rate": 2.9694306473966394e-05, |
| "loss": 1.9318330383300781, |
| "step": 2400 |
| }, |
| { |
| "epoch": 2.000166666666667, |
| "grad_norm": 1.6185426712036133, |
| "learning_rate": 2.9678055533846865e-05, |
| "loss": 1.8744639587402343, |
| "step": 2450 |
| }, |
| { |
| "epoch": 2.0018333333333334, |
| "grad_norm": 1.7479583024978638, |
| "learning_rate": 2.966138843096027e-05, |
| "loss": 1.509072265625, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 2.1598591804504395, |
| "learning_rate": 2.9644305637864277e-05, |
| "loss": 1.3206808471679687, |
| "step": 2550 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 2.0368692874908447, |
| "learning_rate": 2.962680763890249e-05, |
| "loss": 1.4926580810546874, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 2.4042108058929443, |
| "learning_rate": 2.960889493019075e-05, |
| "loss": 1.4857977294921876, |
| "step": 2650 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 2.3829689025878906, |
| "learning_rate": 2.9590568019603022e-05, |
| "loss": 1.4448892211914062, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 2.0621273517608643, |
| "learning_rate": 2.9571827426757036e-05, |
| "loss": 1.4954299926757812, |
| "step": 2750 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 2.3212335109710693, |
| "learning_rate": 2.955267368299953e-05, |
| "loss": 1.5473933410644531, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.9204355478286743, |
| "learning_rate": 2.95331073313912e-05, |
| "loss": 1.5146267700195313, |
| "step": 2850 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.509225606918335, |
| "learning_rate": 2.9513128926691287e-05, |
| "loss": 1.5232659912109374, |
| "step": 2900 |
| }, |
| { |
| "epoch": 2.000166666666667, |
| "grad_norm": 1.8049054145812988, |
| "learning_rate": 2.9492739035341867e-05, |
| "loss": 1.4681259155273438, |
| "step": 2950 |
| }, |
| { |
| "epoch": 2.0018333333333334, |
| "grad_norm": 1.975081443786621, |
| "learning_rate": 2.9471938235451777e-05, |
| "loss": 1.0566283416748048, |
| "step": 3000 |
| }, |
| { |
| "epoch": 2.0035, |
| "grad_norm": 2.3022708892822266, |
| "learning_rate": 2.945072711678023e-05, |
| "loss": 1.32686767578125, |
| "step": 3050 |
| }, |
| { |
| "epoch": 2.0051666666666668, |
| "grad_norm": 2.0772008895874023, |
| "learning_rate": 2.9429106280720074e-05, |
| "loss": 1.314195556640625, |
| "step": 3100 |
| }, |
| { |
| "epoch": 2.0068333333333332, |
| "grad_norm": 2.195183515548706, |
| "learning_rate": 2.9407076340280788e-05, |
| "loss": 1.3024288940429687, |
| "step": 3150 |
| }, |
| { |
| "epoch": 2.0085, |
| "grad_norm": 2.197946548461914, |
| "learning_rate": 2.9384637920071054e-05, |
| "loss": 1.4395762634277345, |
| "step": 3200 |
| }, |
| { |
| "epoch": 2.0101666666666667, |
| "grad_norm": 2.1107280254364014, |
| "learning_rate": 2.9361791656281066e-05, |
| "loss": 1.5124710083007813, |
| "step": 3250 |
| }, |
| { |
| "epoch": 2.011833333333333, |
| "grad_norm": 1.8990710973739624, |
| "learning_rate": 2.93385381966645e-05, |
| "loss": 1.459885711669922, |
| "step": 3300 |
| }, |
| { |
| "epoch": 2.0135, |
| "grad_norm": 2.0525221824645996, |
| "learning_rate": 2.9314878200520135e-05, |
| "loss": 1.4967599487304688, |
| "step": 3350 |
| }, |
| { |
| "epoch": 3.0002, |
| "grad_norm": 1.8729641437530518, |
| "learning_rate": 2.929081233867316e-05, |
| "loss": 1.4636871337890625, |
| "step": 3400 |
| }, |
| { |
| "epoch": 3.0018666666666665, |
| "grad_norm": 2.0232698917388916, |
| "learning_rate": 2.9266341293456167e-05, |
| "loss": 1.2386469268798828, |
| "step": 3450 |
| }, |
| { |
| "epoch": 3.0035333333333334, |
| "grad_norm": 2.249340534210205, |
| "learning_rate": 2.9241465758689805e-05, |
| "loss": 1.2618608856201172, |
| "step": 3500 |
| }, |
| { |
| "epoch": 3.0052, |
| "grad_norm": 2.1609978675842285, |
| "learning_rate": 2.921618643966308e-05, |
| "loss": 1.2303255462646485, |
| "step": 3550 |
| }, |
| { |
| "epoch": 3.006866666666667, |
| "grad_norm": 2.3875741958618164, |
| "learning_rate": 2.9190504053113403e-05, |
| "loss": 1.2375845336914062, |
| "step": 3600 |
| }, |
| { |
| "epoch": 3.0085333333333333, |
| "grad_norm": 1.87738835811615, |
| "learning_rate": 2.9164419327206226e-05, |
| "loss": 1.2702031707763672, |
| "step": 3650 |
| }, |
| { |
| "epoch": 3.0102, |
| "grad_norm": 2.0321192741394043, |
| "learning_rate": 2.913793300151442e-05, |
| "loss": 1.267023239135742, |
| "step": 3700 |
| }, |
| { |
| "epoch": 3.0118666666666667, |
| "grad_norm": 1.972650408744812, |
| "learning_rate": 2.9111045826997314e-05, |
| "loss": 1.2457857513427735, |
| "step": 3750 |
| }, |
| { |
| "epoch": 3.013533333333333, |
| "grad_norm": 1.905271053314209, |
| "learning_rate": 2.9083758565979378e-05, |
| "loss": 1.2576905059814454, |
| "step": 3800 |
| }, |
| { |
| "epoch": 4.000333333333334, |
| "grad_norm": 1.6564491987228394, |
| "learning_rate": 2.9056071992128622e-05, |
| "loss": 1.1955221557617188, |
| "step": 3850 |
| }, |
| { |
| "epoch": 4.002, |
| "grad_norm": 2.0704731941223145, |
| "learning_rate": 2.902798689043466e-05, |
| "loss": 0.7375415802001953, |
| "step": 3900 |
| }, |
| { |
| "epoch": 4.003666666666667, |
| "grad_norm": 1.9580872058868408, |
| "learning_rate": 2.899950405718646e-05, |
| "loss": 0.8625458526611328, |
| "step": 3950 |
| }, |
| { |
| "epoch": 4.005333333333334, |
| "grad_norm": 1.8181036710739136, |
| "learning_rate": 2.8970624299949755e-05, |
| "loss": 0.8049311065673828, |
| "step": 4000 |
| }, |
| { |
| "epoch": 4.007, |
| "grad_norm": 1.8069491386413574, |
| "learning_rate": 2.894134843754415e-05, |
| "loss": 0.7935343170166016, |
| "step": 4050 |
| }, |
| { |
| "epoch": 4.008666666666667, |
| "grad_norm": 2.0135111808776855, |
| "learning_rate": 2.8911677300019924e-05, |
| "loss": 0.9039474487304687, |
| "step": 4100 |
| }, |
| { |
| "epoch": 4.0103333333333335, |
| "grad_norm": 1.8082574605941772, |
| "learning_rate": 2.8881611728634453e-05, |
| "loss": 0.9203302764892578, |
| "step": 4150 |
| }, |
| { |
| "epoch": 4.012, |
| "grad_norm": 2.180549144744873, |
| "learning_rate": 2.8851152575828398e-05, |
| "loss": 0.9213037109375, |
| "step": 4200 |
| }, |
| { |
| "epoch": 4.0136666666666665, |
| "grad_norm": 1.984939455986023, |
| "learning_rate": 2.8820300705201538e-05, |
| "loss": 0.9233769226074219, |
| "step": 4250 |
| }, |
| { |
| "epoch": 5.0005, |
| "grad_norm": 2.5990309715270996, |
| "learning_rate": 2.8789056991488248e-05, |
| "loss": 0.9673636627197265, |
| "step": 4300 |
| }, |
| { |
| "epoch": 5.002166666666667, |
| "grad_norm": 2.5070834159851074, |
| "learning_rate": 2.8757422320532728e-05, |
| "loss": 1.1629231262207032, |
| "step": 4350 |
| }, |
| { |
| "epoch": 5.003833333333334, |
| "grad_norm": 2.3747425079345703, |
| "learning_rate": 2.872539758926388e-05, |
| "loss": 1.2130209350585937, |
| "step": 4400 |
| }, |
| { |
| "epoch": 5.0055, |
| "grad_norm": 2.0789525508880615, |
| "learning_rate": 2.8692983705669872e-05, |
| "loss": 1.1065919494628906, |
| "step": 4450 |
| }, |
| { |
| "epoch": 5.0071666666666665, |
| "grad_norm": 2.5845859050750732, |
| "learning_rate": 2.86601815887724e-05, |
| "loss": 1.1318372344970704, |
| "step": 4500 |
| }, |
| { |
| "epoch": 5.0088333333333335, |
| "grad_norm": 2.2027413845062256, |
| "learning_rate": 2.8626992168600624e-05, |
| "loss": 1.1491417694091797, |
| "step": 4550 |
| }, |
| { |
| "epoch": 5.0105, |
| "grad_norm": 2.105492353439331, |
| "learning_rate": 2.8593416386164818e-05, |
| "loss": 1.1174581909179688, |
| "step": 4600 |
| }, |
| { |
| "epoch": 5.012166666666666, |
| "grad_norm": 2.147904872894287, |
| "learning_rate": 2.8559455193429664e-05, |
| "loss": 1.1449388885498046, |
| "step": 4650 |
| }, |
| { |
| "epoch": 5.013833333333333, |
| "grad_norm": 1.890125036239624, |
| "learning_rate": 2.8525109553287275e-05, |
| "loss": 1.125641860961914, |
| "step": 4700 |
| }, |
| { |
| "epoch": 6.000633333333333, |
| "grad_norm": 1.946161150932312, |
| "learning_rate": 2.84903804395299e-05, |
| "loss": 1.1542523193359375, |
| "step": 4750 |
| }, |
| { |
| "epoch": 6.0023, |
| "grad_norm": 1.9561102390289307, |
| "learning_rate": 2.8455268836822295e-05, |
| "loss": 1.0541219329833984, |
| "step": 4800 |
| }, |
| { |
| "epoch": 6.003966666666667, |
| "grad_norm": 1.9987578392028809, |
| "learning_rate": 2.8419775740673836e-05, |
| "loss": 1.0218131256103515, |
| "step": 4850 |
| }, |
| { |
| "epoch": 6.005633333333333, |
| "grad_norm": 1.905269980430603, |
| "learning_rate": 2.8383902157410255e-05, |
| "loss": 1.003708953857422, |
| "step": 4900 |
| }, |
| { |
| "epoch": 6.0073, |
| "grad_norm": 2.036243438720703, |
| "learning_rate": 2.8347649104145137e-05, |
| "loss": 0.9822654724121094, |
| "step": 4950 |
| }, |
| { |
| "epoch": 6.008966666666667, |
| "grad_norm": 1.8789093494415283, |
| "learning_rate": 2.8311017608751073e-05, |
| "loss": 1.018622512817383, |
| "step": 5000 |
| }, |
| { |
| "epoch": 6.010633333333334, |
| "grad_norm": 2.332475185394287, |
| "learning_rate": 2.8274008709830523e-05, |
| "loss": 1.049513397216797, |
| "step": 5050 |
| }, |
| { |
| "epoch": 6.0123, |
| "grad_norm": 1.8751661777496338, |
| "learning_rate": 2.823662345668634e-05, |
| "loss": 1.068065185546875, |
| "step": 5100 |
| }, |
| { |
| "epoch": 6.013966666666667, |
| "grad_norm": 2.1109721660614014, |
| "learning_rate": 2.8198862909292073e-05, |
| "loss": 1.0169496154785156, |
| "step": 5150 |
| }, |
| { |
| "epoch": 7.000866666666667, |
| "grad_norm": 1.8698688745498657, |
| "learning_rate": 2.8160728138261865e-05, |
| "loss": 1.0039041900634766, |
| "step": 5200 |
| }, |
| { |
| "epoch": 7.002533333333333, |
| "grad_norm": 1.8243681192398071, |
| "learning_rate": 2.8122220224820113e-05, |
| "loss": 0.9678915405273437, |
| "step": 5250 |
| }, |
| { |
| "epoch": 7.0042, |
| "grad_norm": 1.8842891454696655, |
| "learning_rate": 2.8083340260770828e-05, |
| "loss": 0.9515978240966797, |
| "step": 5300 |
| }, |
| { |
| "epoch": 7.005866666666667, |
| "grad_norm": 2.006537675857544, |
| "learning_rate": 2.8044089348466655e-05, |
| "loss": 1.051205825805664, |
| "step": 5350 |
| }, |
| { |
| "epoch": 7.007533333333333, |
| "grad_norm": 2.143925666809082, |
| "learning_rate": 2.8004468600777642e-05, |
| "loss": 0.9571745300292969, |
| "step": 5400 |
| }, |
| { |
| "epoch": 7.0092, |
| "grad_norm": 1.9070833921432495, |
| "learning_rate": 2.7964479141059662e-05, |
| "loss": 0.9707305145263672, |
| "step": 5450 |
| }, |
| { |
| "epoch": 7.010866666666667, |
| "grad_norm": 1.8321707248687744, |
| "learning_rate": 2.7924122103122588e-05, |
| "loss": 0.9359377288818359, |
| "step": 5500 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.9077509641647339, |
| "learning_rate": 2.788339863119813e-05, |
| "loss": 0.6973258972167968, |
| "step": 5550 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.3570711612701416, |
| "learning_rate": 2.7842309879907394e-05, |
| "loss": 0.6986883544921875, |
| "step": 5600 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.8015130758285522, |
| "learning_rate": 2.7800857014228154e-05, |
| "loss": 0.6809234619140625, |
| "step": 5650 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.9509077072143555, |
| "learning_rate": 2.7759041209461817e-05, |
| "loss": 0.6638854217529296, |
| "step": 5700 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.7216967344284058, |
| "learning_rate": 2.771686365120009e-05, |
| "loss": 0.5996527862548828, |
| "step": 5750 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.4293559789657593, |
| "learning_rate": 2.7674325535291387e-05, |
| "loss": 0.6288544845581054, |
| "step": 5800 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.4647794961929321, |
| "learning_rate": 2.7631428067806903e-05, |
| "loss": 0.6884970855712891, |
| "step": 5850 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.9533915519714355, |
| "learning_rate": 2.7588172465006436e-05, |
| "loss": 0.7003134918212891, |
| "step": 5900 |
| }, |
| { |
| "epoch": 2.000166666666667, |
| "grad_norm": 1.5402069091796875, |
| "learning_rate": 2.754455995330388e-05, |
| "loss": 0.6689606475830078, |
| "step": 5950 |
| }, |
| { |
| "epoch": 2.0018333333333334, |
| "grad_norm": 1.5695101022720337, |
| "learning_rate": 2.750059176923248e-05, |
| "loss": 0.4137184524536133, |
| "step": 6000 |
| }, |
| { |
| "epoch": 2.0035, |
| "grad_norm": 1.9652310609817505, |
| "learning_rate": 2.745626915940975e-05, |
| "loss": 0.46977787017822265, |
| "step": 6050 |
| }, |
| { |
| "epoch": 2.0051666666666668, |
| "grad_norm": 1.7700436115264893, |
| "learning_rate": 2.7411593380502133e-05, |
| "loss": 0.463070182800293, |
| "step": 6100 |
| }, |
| { |
| "epoch": 2.0068333333333332, |
| "grad_norm": 1.8935976028442383, |
| "learning_rate": 2.7366565699189398e-05, |
| "loss": 0.44758522033691406, |
| "step": 6150 |
| }, |
| { |
| "epoch": 2.0085, |
| "grad_norm": 1.3832303285598755, |
| "learning_rate": 2.732118739212868e-05, |
| "loss": 0.4535627746582031, |
| "step": 6200 |
| }, |
| { |
| "epoch": 2.0101666666666667, |
| "grad_norm": 1.6829535961151123, |
| "learning_rate": 2.7275459745918316e-05, |
| "loss": 0.48570785522460935, |
| "step": 6250 |
| }, |
| { |
| "epoch": 2.011833333333333, |
| "grad_norm": 1.8293315172195435, |
| "learning_rate": 2.722938405706136e-05, |
| "loss": 0.5032914352416992, |
| "step": 6300 |
| }, |
| { |
| "epoch": 2.0135, |
| "grad_norm": 1.7647825479507446, |
| "learning_rate": 2.7182961631928812e-05, |
| "loss": 0.5367673110961914, |
| "step": 6350 |
| }, |
| { |
| "epoch": 3.0002, |
| "grad_norm": 1.6019330024719238, |
| "learning_rate": 2.7136193786722592e-05, |
| "loss": 0.5786269378662109, |
| "step": 6400 |
| }, |
| { |
| "epoch": 3.0018666666666665, |
| "grad_norm": 1.6498947143554688, |
| "learning_rate": 2.7089081847438228e-05, |
| "loss": 0.6523352813720703, |
| "step": 6450 |
| }, |
| { |
| "epoch": 3.0035333333333334, |
| "grad_norm": 2.055866241455078, |
| "learning_rate": 2.7041627149827236e-05, |
| "loss": 0.6170930862426758, |
| "step": 6500 |
| }, |
| { |
| "epoch": 3.0052, |
| "grad_norm": 2.074732542037964, |
| "learning_rate": 2.6993831039359267e-05, |
| "loss": 0.6060370635986329, |
| "step": 6550 |
| }, |
| { |
| "epoch": 3.006866666666667, |
| "grad_norm": 2.256472587585449, |
| "learning_rate": 2.694569487118396e-05, |
| "loss": 0.6324485778808594, |
| "step": 6600 |
| }, |
| { |
| "epoch": 3.0085333333333333, |
| "grad_norm": 1.6386387348175049, |
| "learning_rate": 2.6897220010092508e-05, |
| "loss": 0.6060771179199219, |
| "step": 6650 |
| }, |
| { |
| "epoch": 3.0102, |
| "grad_norm": 1.8649691343307495, |
| "learning_rate": 2.6848407830478967e-05, |
| "loss": 0.5979516220092773, |
| "step": 6700 |
| }, |
| { |
| "epoch": 3.0118666666666667, |
| "grad_norm": 1.7192860841751099, |
| "learning_rate": 2.6799259716301293e-05, |
| "loss": 0.6279150390625, |
| "step": 6750 |
| }, |
| { |
| "epoch": 3.013533333333333, |
| "grad_norm": 1.6655539274215698, |
| "learning_rate": 2.6749777061042103e-05, |
| "loss": 0.625608139038086, |
| "step": 6800 |
| }, |
| { |
| "epoch": 4.000333333333334, |
| "grad_norm": 1.068650484085083, |
| "learning_rate": 2.6699961267669162e-05, |
| "loss": 0.5783480072021484, |
| "step": 6850 |
| }, |
| { |
| "epoch": 4.002, |
| "grad_norm": 1.2487739324569702, |
| "learning_rate": 2.6649813748595595e-05, |
| "loss": 0.2357199287414551, |
| "step": 6900 |
| }, |
| { |
| "epoch": 4.003666666666667, |
| "grad_norm": 1.3043678998947144, |
| "learning_rate": 2.659933592563987e-05, |
| "loss": 0.2731271743774414, |
| "step": 6950 |
| }, |
| { |
| "epoch": 4.005333333333334, |
| "grad_norm": 1.223210334777832, |
| "learning_rate": 2.654852922998545e-05, |
| "loss": 0.22788461685180664, |
| "step": 7000 |
| }, |
| { |
| "epoch": 4.007, |
| "grad_norm": 1.1483008861541748, |
| "learning_rate": 2.649739510214025e-05, |
| "loss": 0.22346155166625978, |
| "step": 7050 |
| }, |
| { |
| "epoch": 4.008666666666667, |
| "grad_norm": 1.2172214984893799, |
| "learning_rate": 2.6445934991895764e-05, |
| "loss": 0.23918920516967773, |
| "step": 7100 |
| }, |
| { |
| "epoch": 4.0103333333333335, |
| "grad_norm": 1.1359659433364868, |
| "learning_rate": 2.639415035828598e-05, |
| "loss": 0.24985692977905274, |
| "step": 7150 |
| }, |
| { |
| "epoch": 4.012, |
| "grad_norm": 1.475920557975769, |
| "learning_rate": 2.6342042669546e-05, |
| "loss": 0.2808365821838379, |
| "step": 7200 |
| }, |
| { |
| "epoch": 4.0136666666666665, |
| "grad_norm": 1.2670186758041382, |
| "learning_rate": 2.628961340307043e-05, |
| "loss": 0.2782126426696777, |
| "step": 7250 |
| }, |
| { |
| "epoch": 5.0005, |
| "grad_norm": 1.717673659324646, |
| "learning_rate": 2.6236864045371453e-05, |
| "loss": 0.36194786071777346, |
| "step": 7300 |
| }, |
| { |
| "epoch": 5.002166666666667, |
| "grad_norm": 1.797800898551941, |
| "learning_rate": 2.618379609203674e-05, |
| "loss": 0.6739380645751953, |
| "step": 7350 |
| }, |
| { |
| "epoch": 5.003833333333334, |
| "grad_norm": 2.0240073204040527, |
| "learning_rate": 2.613041104768699e-05, |
| "loss": 0.6977305603027344, |
| "step": 7400 |
| }, |
| { |
| "epoch": 5.0055, |
| "grad_norm": 1.6391247510910034, |
| "learning_rate": 2.6076710425933307e-05, |
| "loss": 0.6162394332885742, |
| "step": 7450 |
| }, |
| { |
| "epoch": 5.0071666666666665, |
| "grad_norm": 2.932299852371216, |
| "learning_rate": 2.6022695749334263e-05, |
| "loss": 0.6486746215820313, |
| "step": 7500 |
| }, |
| { |
| "epoch": 5.0088333333333335, |
| "grad_norm": 2.05151629447937, |
| "learning_rate": 2.5968368549352748e-05, |
| "loss": 0.6304220581054687, |
| "step": 7550 |
| }, |
| { |
| "epoch": 5.0105, |
| "grad_norm": 1.9714049100875854, |
| "learning_rate": 2.5913730366312535e-05, |
| "loss": 0.6141956710815429, |
| "step": 7600 |
| }, |
| { |
| "epoch": 5.012166666666666, |
| "grad_norm": 2.1039156913757324, |
| "learning_rate": 2.5858782749354603e-05, |
| "loss": 0.6495011901855469, |
| "step": 7650 |
| }, |
| { |
| "epoch": 5.013833333333333, |
| "grad_norm": 1.6918100118637085, |
| "learning_rate": 2.5803527256393243e-05, |
| "loss": 0.63647216796875, |
| "step": 7700 |
| }, |
| { |
| "epoch": 6.000633333333333, |
| "grad_norm": 1.6919469833374023, |
| "learning_rate": 2.5747965454071845e-05, |
| "loss": 0.671699447631836, |
| "step": 7750 |
| }, |
| { |
| "epoch": 6.0023, |
| "grad_norm": 1.7445021867752075, |
| "learning_rate": 2.569209891771852e-05, |
| "loss": 0.6194692611694336, |
| "step": 7800 |
| }, |
| { |
| "epoch": 6.003966666666667, |
| "grad_norm": 1.7923142910003662, |
| "learning_rate": 2.5635929231301407e-05, |
| "loss": 0.5799810409545898, |
| "step": 7850 |
| }, |
| { |
| "epoch": 6.005633333333333, |
| "grad_norm": 1.78199303150177, |
| "learning_rate": 2.5579457987383783e-05, |
| "loss": 0.577097282409668, |
| "step": 7900 |
| }, |
| { |
| "epoch": 6.0073, |
| "grad_norm": 1.3529242277145386, |
| "learning_rate": 2.5522686787078885e-05, |
| "loss": 0.5641065979003906, |
| "step": 7950 |
| }, |
| { |
| "epoch": 6.008966666666667, |
| "grad_norm": 1.547831416130066, |
| "learning_rate": 2.546561724000454e-05, |
| "loss": 0.5749254608154297, |
| "step": 8000 |
| }, |
| { |
| "epoch": 6.010633333333334, |
| "grad_norm": 1.794567584991455, |
| "learning_rate": 2.5408250964237518e-05, |
| "loss": 0.6101033782958984, |
| "step": 8050 |
| }, |
| { |
| "epoch": 6.0123, |
| "grad_norm": 1.645218849182129, |
| "learning_rate": 2.5350589586267645e-05, |
| "loss": 0.6382519149780274, |
| "step": 8100 |
| }, |
| { |
| "epoch": 6.013966666666667, |
| "grad_norm": 1.7477233409881592, |
| "learning_rate": 2.52926347409517e-05, |
| "loss": 0.5900593185424805, |
| "step": 8150 |
| }, |
| { |
| "epoch": 7.000866666666667, |
| "grad_norm": 1.5820660591125488, |
| "learning_rate": 2.5234388071467064e-05, |
| "loss": 0.5954227828979493, |
| "step": 8200 |
| }, |
| { |
| "epoch": 7.002533333333333, |
| "grad_norm": 1.5947580337524414, |
| "learning_rate": 2.517585122926512e-05, |
| "loss": 0.5810695266723633, |
| "step": 8250 |
| }, |
| { |
| "epoch": 7.0042, |
| "grad_norm": 1.4849404096603394, |
| "learning_rate": 2.511702587402444e-05, |
| "loss": 0.5662033843994141, |
| "step": 8300 |
| }, |
| { |
| "epoch": 7.005866666666667, |
| "grad_norm": 1.6737949848175049, |
| "learning_rate": 2.5057913673603722e-05, |
| "loss": 0.6508497619628906, |
| "step": 8350 |
| }, |
| { |
| "epoch": 7.007533333333333, |
| "grad_norm": 1.9604586362838745, |
| "learning_rate": 2.4998516303994502e-05, |
| "loss": 0.5735390090942383, |
| "step": 8400 |
| }, |
| { |
| "epoch": 7.0092, |
| "grad_norm": 1.8022760152816772, |
| "learning_rate": 2.493883544927365e-05, |
| "loss": 0.5821025848388672, |
| "step": 8450 |
| }, |
| { |
| "epoch": 7.010866666666667, |
| "grad_norm": 1.723585605621338, |
| "learning_rate": 2.4878872801555596e-05, |
| "loss": 0.5521516799926758, |
| "step": 8500 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.7751977443695068, |
| "learning_rate": 2.4818630060944365e-05, |
| "loss": 0.40423309326171875, |
| "step": 8550 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.3075405359268188, |
| "learning_rate": 2.4758108935485397e-05, |
| "loss": 0.41454471588134767, |
| "step": 8600 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.5956394672393799, |
| "learning_rate": 2.4697311141117076e-05, |
| "loss": 0.4045108413696289, |
| "step": 8650 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 2.0534520149230957, |
| "learning_rate": 2.4636238401622122e-05, |
| "loss": 0.3971108627319336, |
| "step": 8700 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.553052544593811, |
| "learning_rate": 2.457489244857867e-05, |
| "loss": 0.3435604476928711, |
| "step": 8750 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.145978569984436, |
| "learning_rate": 2.451327502131123e-05, |
| "loss": 0.36573856353759765, |
| "step": 8800 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.2501201629638672, |
| "learning_rate": 2.4451387866841323e-05, |
| "loss": 0.4070841598510742, |
| "step": 8850 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.8662524223327637, |
| "learning_rate": 2.438923273983799e-05, |
| "loss": 0.4094845962524414, |
| "step": 8900 |
| }, |
| { |
| "epoch": 2.000166666666667, |
| "grad_norm": 1.2666409015655518, |
| "learning_rate": 2.4326811402568e-05, |
| "loss": 0.3822321319580078, |
| "step": 8950 |
| }, |
| { |
| "epoch": 2.0018333333333334, |
| "grad_norm": 1.274243712425232, |
| "learning_rate": 2.4264125624845918e-05, |
| "loss": 0.17029172897338868, |
| "step": 9000 |
| }, |
| { |
| "epoch": 2.0035, |
| "grad_norm": 1.71302330493927, |
| "learning_rate": 2.420117718398391e-05, |
| "loss": 0.189314022064209, |
| "step": 9050 |
| }, |
| { |
| "epoch": 2.0051666666666668, |
| "grad_norm": 1.406741738319397, |
| "learning_rate": 2.4137967864741365e-05, |
| "loss": 0.18286312103271485, |
| "step": 9100 |
| }, |
| { |
| "epoch": 2.0068333333333332, |
| "grad_norm": 1.4597975015640259, |
| "learning_rate": 2.4074499459274273e-05, |
| "loss": 0.1649922752380371, |
| "step": 9150 |
| }, |
| { |
| "epoch": 2.0085, |
| "grad_norm": 0.8979786038398743, |
| "learning_rate": 2.4010773767084428e-05, |
| "loss": 0.15474456787109375, |
| "step": 9200 |
| }, |
| { |
| "epoch": 2.0101666666666667, |
| "grad_norm": 1.3335304260253906, |
| "learning_rate": 2.3946792594968398e-05, |
| "loss": 0.17460403442382813, |
| "step": 9250 |
| }, |
| { |
| "epoch": 2.011833333333333, |
| "grad_norm": 1.6577045917510986, |
| "learning_rate": 2.3882557756966304e-05, |
| "loss": 0.20299341201782226, |
| "step": 9300 |
| }, |
| { |
| "epoch": 2.0135, |
| "grad_norm": 1.227666974067688, |
| "learning_rate": 2.3818071074310386e-05, |
| "loss": 0.21543956756591798, |
| "step": 9350 |
| }, |
| { |
| "epoch": 3.0002, |
| "grad_norm": 1.5215626955032349, |
| "learning_rate": 2.3753334375373355e-05, |
| "loss": 0.27363029479980466, |
| "step": 9400 |
| }, |
| { |
| "epoch": 3.0018666666666665, |
| "grad_norm": 1.530497431755066, |
| "learning_rate": 2.3688349495616578e-05, |
| "loss": 0.407540397644043, |
| "step": 9450 |
| }, |
| { |
| "epoch": 3.0035333333333334, |
| "grad_norm": 2.1815409660339355, |
| "learning_rate": 2.3623118277538013e-05, |
| "loss": 0.38198047637939453, |
| "step": 9500 |
| }, |
| { |
| "epoch": 3.0052, |
| "grad_norm": 1.9969043731689453, |
| "learning_rate": 2.355764257061998e-05, |
| "loss": 0.370845947265625, |
| "step": 9550 |
| }, |
| { |
| "epoch": 3.006866666666667, |
| "grad_norm": 2.477900981903076, |
| "learning_rate": 2.3491924231276722e-05, |
| "loss": 0.3945932388305664, |
| "step": 9600 |
| }, |
| { |
| "epoch": 3.0085333333333333, |
| "grad_norm": 1.5564457178115845, |
| "learning_rate": 2.3425965122801773e-05, |
| "loss": 0.3718429183959961, |
| "step": 9650 |
| }, |
| { |
| "epoch": 3.0102, |
| "grad_norm": 1.9686049222946167, |
| "learning_rate": 2.3359767115315122e-05, |
| "loss": 0.3622703552246094, |
| "step": 9700 |
| }, |
| { |
| "epoch": 3.0118666666666667, |
| "grad_norm": 1.5884681940078735, |
| "learning_rate": 2.3293332085710204e-05, |
| "loss": 0.3964054870605469, |
| "step": 9750 |
| }, |
| { |
| "epoch": 3.013533333333333, |
| "grad_norm": 1.6316170692443848, |
| "learning_rate": 2.3226661917600657e-05, |
| "loss": 0.392611083984375, |
| "step": 9800 |
| }, |
| { |
| "epoch": 4.000333333333334, |
| "grad_norm": 0.867811381816864, |
| "learning_rate": 2.3159758501266963e-05, |
| "loss": 0.34639438629150393, |
| "step": 9850 |
| }, |
| { |
| "epoch": 4.002, |
| "grad_norm": 1.0062099695205688, |
| "learning_rate": 2.3092623733602794e-05, |
| "loss": 0.09693001747131348, |
| "step": 9900 |
| }, |
| { |
| "epoch": 4.003666666666667, |
| "grad_norm": 1.0147392749786377, |
| "learning_rate": 2.3025259518061277e-05, |
| "loss": 0.11223688125610351, |
| "step": 9950 |
| }, |
| { |
| "epoch": 4.005333333333334, |
| "grad_norm": 0.8574983477592468, |
| "learning_rate": 2.295766776460101e-05, |
| "loss": 0.09009716987609863, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.335, |
| "grad_norm": 1.9024637937545776, |
| "learning_rate": 2.2889850389631912e-05, |
| "loss": 2.2911944580078125, |
| "step": 10050 |
| }, |
| { |
| "epoch": 0.33666666666666667, |
| "grad_norm": 2.425832748413086, |
| "learning_rate": 2.2821809315960874e-05, |
| "loss": 2.136341400146484, |
| "step": 10100 |
| }, |
| { |
| "epoch": 0.3383333333333333, |
| "grad_norm": 1.8301362991333008, |
| "learning_rate": 2.2753546472737268e-05, |
| "loss": 2.104422302246094, |
| "step": 10150 |
| }, |
| { |
| "epoch": 0.34, |
| "grad_norm": 1.5538557767868042, |
| "learning_rate": 2.2685063795398218e-05, |
| "loss": 2.0370709228515627, |
| "step": 10200 |
| }, |
| { |
| "epoch": 0.3416666666666667, |
| "grad_norm": 1.5662953853607178, |
| "learning_rate": 2.2616363225613756e-05, |
| "loss": 2.068159637451172, |
| "step": 10250 |
| }, |
| { |
| "epoch": 0.3433333333333333, |
| "grad_norm": 1.5763871669769287, |
| "learning_rate": 2.2547446711231746e-05, |
| "loss": 2.0447808837890626, |
| "step": 10300 |
| }, |
| { |
| "epoch": 0.345, |
| "grad_norm": 1.388518214225769, |
| "learning_rate": 2.2479700900763443e-05, |
| "loss": 2.0605596923828124, |
| "step": 10350 |
| }, |
| { |
| "epoch": 0.3466666666666667, |
| "grad_norm": 1.4810057878494263, |
| "learning_rate": 2.2410362586529476e-05, |
| "loss": 2.0104673767089842, |
| "step": 10400 |
| }, |
| { |
| "epoch": 0.34833333333333333, |
| "grad_norm": 1.6268818378448486, |
| "learning_rate": 2.234081416837596e-05, |
| "loss": 1.985343475341797, |
| "step": 10450 |
| }, |
| { |
| "epoch": 0.35, |
| "grad_norm": 1.4944297075271606, |
| "learning_rate": 2.2271057618189578e-05, |
| "loss": 1.9991339111328126, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.3516666666666667, |
| "grad_norm": 1.4048922061920166, |
| "learning_rate": 2.220109491375811e-05, |
| "loss": 1.9501966857910156, |
| "step": 10550 |
| }, |
| { |
| "epoch": 0.35333333333333333, |
| "grad_norm": 1.6254700422286987, |
| "learning_rate": 2.2130928038714378e-05, |
| "loss": 1.93730224609375, |
| "step": 10600 |
| }, |
| { |
| "epoch": 0.355, |
| "grad_norm": 1.4676744937896729, |
| "learning_rate": 2.2060558982479992e-05, |
| "loss": 1.9180142211914062, |
| "step": 10650 |
| }, |
| { |
| "epoch": 0.3566666666666667, |
| "grad_norm": 1.3778892755508423, |
| "learning_rate": 2.1989989740208944e-05, |
| "loss": 1.9412496948242188, |
| "step": 10700 |
| }, |
| { |
| "epoch": 0.35833333333333334, |
| "grad_norm": 1.4417188167572021, |
| "learning_rate": 2.1919222312731048e-05, |
| "loss": 1.8980422973632813, |
| "step": 10750 |
| }, |
| { |
| "epoch": 0.36, |
| "grad_norm": 1.4003877639770508, |
| "learning_rate": 2.184825870649522e-05, |
| "loss": 1.8576702880859375, |
| "step": 10800 |
| }, |
| { |
| "epoch": 0.3616666666666667, |
| "grad_norm": 1.3203614950180054, |
| "learning_rate": 2.177710093351255e-05, |
| "loss": 1.8865699768066406, |
| "step": 10850 |
| }, |
| { |
| "epoch": 0.36333333333333334, |
| "grad_norm": 1.4510948657989502, |
| "learning_rate": 2.170575101129931e-05, |
| "loss": 1.8597201538085937, |
| "step": 10900 |
| }, |
| { |
| "epoch": 0.365, |
| "grad_norm": 1.339342713356018, |
| "learning_rate": 2.1634210962819703e-05, |
| "loss": 1.8870362854003906, |
| "step": 10950 |
| }, |
| { |
| "epoch": 0.36666666666666664, |
| "grad_norm": 1.309877634048462, |
| "learning_rate": 2.1562482816428542e-05, |
| "loss": 1.8619964599609375, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.36833333333333335, |
| "grad_norm": 1.2446576356887817, |
| "learning_rate": 2.1490568605813725e-05, |
| "loss": 1.8535054016113282, |
| "step": 11050 |
| }, |
| { |
| "epoch": 0.37, |
| "grad_norm": 1.4111701250076294, |
| "learning_rate": 2.1418470369938573e-05, |
| "loss": 1.8347019958496094, |
| "step": 11100 |
| }, |
| { |
| "epoch": 0.37166666666666665, |
| "grad_norm": 1.4939404726028442, |
| "learning_rate": 2.134619015298402e-05, |
| "loss": 1.8600994873046874, |
| "step": 11150 |
| }, |
| { |
| "epoch": 0.37333333333333335, |
| "grad_norm": 1.4104790687561035, |
| "learning_rate": 2.1273730004290677e-05, |
| "loss": 1.8550985717773438, |
| "step": 11200 |
| }, |
| { |
| "epoch": 0.375, |
| "grad_norm": 1.2873270511627197, |
| "learning_rate": 2.120109197830068e-05, |
| "loss": 1.8232687377929688, |
| "step": 11250 |
| }, |
| { |
| "epoch": 0.37666666666666665, |
| "grad_norm": 1.2587337493896484, |
| "learning_rate": 2.1128278134499493e-05, |
| "loss": 1.7981582641601563, |
| "step": 11300 |
| }, |
| { |
| "epoch": 0.37833333333333335, |
| "grad_norm": 1.4026145935058594, |
| "learning_rate": 2.1055290537357482e-05, |
| "loss": 1.800906219482422, |
| "step": 11350 |
| }, |
| { |
| "epoch": 0.38, |
| "grad_norm": 1.3134163618087769, |
| "learning_rate": 2.0982131256271392e-05, |
| "loss": 1.8477032470703125, |
| "step": 11400 |
| }, |
| { |
| "epoch": 0.38166666666666665, |
| "grad_norm": 1.2536565065383911, |
| "learning_rate": 2.0908802365505685e-05, |
| "loss": 1.8300701904296874, |
| "step": 11450 |
| }, |
| { |
| "epoch": 0.38333333333333336, |
| "grad_norm": 1.2861531972885132, |
| "learning_rate": 2.0835305944133705e-05, |
| "loss": 1.8158242797851563, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.385, |
| "grad_norm": 1.2281335592269897, |
| "learning_rate": 2.076164407597875e-05, |
| "loss": 1.8379869079589843, |
| "step": 11550 |
| }, |
| { |
| "epoch": 0.38666666666666666, |
| "grad_norm": 1.3885990381240845, |
| "learning_rate": 2.0687818849554986e-05, |
| "loss": 1.7722903442382814, |
| "step": 11600 |
| }, |
| { |
| "epoch": 0.3883333333333333, |
| "grad_norm": 1.2439100742340088, |
| "learning_rate": 2.0613832358008226e-05, |
| "loss": 1.8434024047851563, |
| "step": 11650 |
| }, |
| { |
| "epoch": 0.39, |
| "grad_norm": 1.2999515533447266, |
| "learning_rate": 2.053968669905659e-05, |
| "loss": 1.7893605041503906, |
| "step": 11700 |
| }, |
| { |
| "epoch": 0.39166666666666666, |
| "grad_norm": 1.1664999723434448, |
| "learning_rate": 2.046538397493103e-05, |
| "loss": 1.7857025146484375, |
| "step": 11750 |
| }, |
| { |
| "epoch": 0.3933333333333333, |
| "grad_norm": 1.2600429058074951, |
| "learning_rate": 2.0390926292315704e-05, |
| "loss": 1.8158566284179687, |
| "step": 11800 |
| }, |
| { |
| "epoch": 0.395, |
| "grad_norm": 1.2101469039916992, |
| "learning_rate": 2.03163157622883e-05, |
| "loss": 1.7761709594726562, |
| "step": 11850 |
| }, |
| { |
| "epoch": 0.39666666666666667, |
| "grad_norm": 1.3131691217422485, |
| "learning_rate": 2.0241554500260102e-05, |
| "loss": 1.7952947998046875, |
| "step": 11900 |
| }, |
| { |
| "epoch": 0.3983333333333333, |
| "grad_norm": 1.2902121543884277, |
| "learning_rate": 2.0166644625916092e-05, |
| "loss": 1.7892556762695313, |
| "step": 11950 |
| }, |
| { |
| "epoch": 0.4, |
| "grad_norm": 1.1372772455215454, |
| "learning_rate": 2.0091588263154793e-05, |
| "loss": 1.7282296752929687, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.40166666666666667, |
| "grad_norm": 1.213036298751831, |
| "learning_rate": 2.0016387540028093e-05, |
| "loss": 1.7450872802734374, |
| "step": 12050 |
| }, |
| { |
| "epoch": 0.4033333333333333, |
| "grad_norm": 1.2814973592758179, |
| "learning_rate": 1.9941044588680866e-05, |
| "loss": 1.7453732299804687, |
| "step": 12100 |
| }, |
| { |
| "epoch": 0.405, |
| "grad_norm": 1.076398253440857, |
| "learning_rate": 1.9865561545290553e-05, |
| "loss": 1.7421353149414063, |
| "step": 12150 |
| }, |
| { |
| "epoch": 0.4066666666666667, |
| "grad_norm": 1.2121100425720215, |
| "learning_rate": 1.9789940550006595e-05, |
| "loss": 1.72051025390625, |
| "step": 12200 |
| }, |
| { |
| "epoch": 0.4083333333333333, |
| "grad_norm": 1.2553529739379883, |
| "learning_rate": 1.971418374688972e-05, |
| "loss": 1.7336581420898438, |
| "step": 12250 |
| }, |
| { |
| "epoch": 0.41, |
| "grad_norm": 1.2400550842285156, |
| "learning_rate": 1.9638293283851205e-05, |
| "loss": 1.7676593017578126, |
| "step": 12300 |
| }, |
| { |
| "epoch": 0.4116666666666667, |
| "grad_norm": 1.231232762336731, |
| "learning_rate": 1.9562271312591927e-05, |
| "loss": 1.7237821960449218, |
| "step": 12350 |
| }, |
| { |
| "epoch": 0.41333333333333333, |
| "grad_norm": 1.1945688724517822, |
| "learning_rate": 1.9486119988541394e-05, |
| "loss": 1.732925567626953, |
| "step": 12400 |
| }, |
| { |
| "epoch": 0.415, |
| "grad_norm": 1.1468254327774048, |
| "learning_rate": 1.9409841470796603e-05, |
| "loss": 1.6542854309082031, |
| "step": 12450 |
| }, |
| { |
| "epoch": 0.4166666666666667, |
| "grad_norm": 1.2474437952041626, |
| "learning_rate": 1.933343792206085e-05, |
| "loss": 1.698963623046875, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.41833333333333333, |
| "grad_norm": 1.2392055988311768, |
| "learning_rate": 1.9256911508582395e-05, |
| "loss": 1.7225572204589843, |
| "step": 12550 |
| }, |
| { |
| "epoch": 0.42, |
| "grad_norm": 1.131912350654602, |
| "learning_rate": 1.9180264400093046e-05, |
| "loss": 1.7018638610839845, |
| "step": 12600 |
| }, |
| { |
| "epoch": 0.4216666666666667, |
| "grad_norm": 1.1391109228134155, |
| "learning_rate": 1.9103498769746638e-05, |
| "loss": 1.7238023376464844, |
| "step": 12650 |
| }, |
| { |
| "epoch": 0.42333333333333334, |
| "grad_norm": 1.255763053894043, |
| "learning_rate": 1.902661679405744e-05, |
| "loss": 1.698165283203125, |
| "step": 12700 |
| }, |
| { |
| "epoch": 0.425, |
| "grad_norm": 1.1407846212387085, |
| "learning_rate": 1.895116168037029e-05, |
| "loss": 1.7422465515136718, |
| "step": 12750 |
| }, |
| { |
| "epoch": 0.4266666666666667, |
| "grad_norm": 1.241137981414795, |
| "learning_rate": 1.887405577490636e-05, |
| "loss": 1.6755677795410155, |
| "step": 12800 |
| }, |
| { |
| "epoch": 0.42833333333333334, |
| "grad_norm": 1.2269986867904663, |
| "learning_rate": 1.8796840029432137e-05, |
| "loss": 1.66367431640625, |
| "step": 12850 |
| }, |
| { |
| "epoch": 0.43, |
| "grad_norm": 1.1101669073104858, |
| "learning_rate": 1.871951663322385e-05, |
| "loss": 1.7193191528320313, |
| "step": 12900 |
| }, |
| { |
| "epoch": 0.43166666666666664, |
| "grad_norm": 1.2445309162139893, |
| "learning_rate": 1.8642087778609944e-05, |
| "loss": 1.6630491638183593, |
| "step": 12950 |
| }, |
| { |
| "epoch": 0.43333333333333335, |
| "grad_norm": 1.3862735033035278, |
| "learning_rate": 1.856455566090888e-05, |
| "loss": 1.698961944580078, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.435, |
| "grad_norm": 1.0933082103729248, |
| "learning_rate": 1.8486922478366923e-05, |
| "loss": 1.7113790893554688, |
| "step": 13050 |
| }, |
| { |
| "epoch": 0.43666666666666665, |
| "grad_norm": 1.1887568235397339, |
| "learning_rate": 1.8409190432095788e-05, |
| "loss": 1.659815216064453, |
| "step": 13100 |
| }, |
| { |
| "epoch": 0.43833333333333335, |
| "grad_norm": 1.2273800373077393, |
| "learning_rate": 1.833136172601025e-05, |
| "loss": 1.701916961669922, |
| "step": 13150 |
| }, |
| { |
| "epoch": 0.44, |
| "grad_norm": 1.1295883655548096, |
| "learning_rate": 1.825343856676566e-05, |
| "loss": 1.7027659606933594, |
| "step": 13200 |
| }, |
| { |
| "epoch": 0.44166666666666665, |
| "grad_norm": 1.234627366065979, |
| "learning_rate": 1.8175423163695356e-05, |
| "loss": 1.6670001220703126, |
| "step": 13250 |
| }, |
| { |
| "epoch": 0.44333333333333336, |
| "grad_norm": 1.1243114471435547, |
| "learning_rate": 1.8097317728748052e-05, |
| "loss": 1.6672360229492187, |
| "step": 13300 |
| }, |
| { |
| "epoch": 0.445, |
| "grad_norm": 1.2413411140441895, |
| "learning_rate": 1.801912447642511e-05, |
| "loss": 1.6659980773925782, |
| "step": 13350 |
| }, |
| { |
| "epoch": 0.44666666666666666, |
| "grad_norm": 1.2024633884429932, |
| "learning_rate": 1.7940845623717748e-05, |
| "loss": 1.6188279724121093, |
| "step": 13400 |
| }, |
| { |
| "epoch": 0.4483333333333333, |
| "grad_norm": 1.215316891670227, |
| "learning_rate": 1.7862483390044192e-05, |
| "loss": 1.627742919921875, |
| "step": 13450 |
| }, |
| { |
| "epoch": 0.45, |
| "grad_norm": 1.278456687927246, |
| "learning_rate": 1.7784039997186736e-05, |
| "loss": 1.6489530944824218, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.45166666666666666, |
| "grad_norm": 1.2035270929336548, |
| "learning_rate": 1.7705517669228773e-05, |
| "loss": 1.6360804748535156, |
| "step": 13550 |
| }, |
| { |
| "epoch": 0.4533333333333333, |
| "grad_norm": 1.1333221197128296, |
| "learning_rate": 1.7626918632491714e-05, |
| "loss": 1.657493133544922, |
| "step": 13600 |
| }, |
| { |
| "epoch": 0.455, |
| "grad_norm": 1.0751276016235352, |
| "learning_rate": 1.754824511547186e-05, |
| "loss": 1.636985321044922, |
| "step": 13650 |
| }, |
| { |
| "epoch": 0.45666666666666667, |
| "grad_norm": 1.622294306755066, |
| "learning_rate": 1.746949934877725e-05, |
| "loss": 1.6701731872558594, |
| "step": 13700 |
| }, |
| { |
| "epoch": 0.4583333333333333, |
| "grad_norm": 1.2836705446243286, |
| "learning_rate": 1.7390683565064386e-05, |
| "loss": 1.6268263244628907, |
| "step": 13750 |
| }, |
| { |
| "epoch": 0.46, |
| "grad_norm": 1.2211285829544067, |
| "learning_rate": 1.7311799998974947e-05, |
| "loss": 1.618380126953125, |
| "step": 13800 |
| }, |
| { |
| "epoch": 0.46166666666666667, |
| "grad_norm": 1.1164684295654297, |
| "learning_rate": 1.723285088707242e-05, |
| "loss": 1.6643341064453125, |
| "step": 13850 |
| }, |
| { |
| "epoch": 0.4633333333333333, |
| "grad_norm": 1.0905755758285522, |
| "learning_rate": 1.7153838467778698e-05, |
| "loss": 1.6010598754882812, |
| "step": 13900 |
| }, |
| { |
| "epoch": 0.465, |
| "grad_norm": 1.0284701585769653, |
| "learning_rate": 1.707476498131062e-05, |
| "loss": 1.6055384826660157, |
| "step": 13950 |
| }, |
| { |
| "epoch": 0.4666666666666667, |
| "grad_norm": 1.1286269426345825, |
| "learning_rate": 1.699563266961643e-05, |
| "loss": 1.6123344421386718, |
| "step": 14000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.572548747062683, |
| "learning_rate": 1.6916443776312232e-05, |
| "loss": 1.6946664428710938, |
| "step": 14050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.4154376983642578, |
| "learning_rate": 1.6837200546618386e-05, |
| "loss": 1.6747663879394532, |
| "step": 14100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.5416319370269775, |
| "learning_rate": 1.6757905227295814e-05, |
| "loss": 1.6729904174804688, |
| "step": 14150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.6411312818527222, |
| "learning_rate": 1.6678560066582336e-05, |
| "loss": 1.6301991271972656, |
| "step": 14200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.6199766397476196, |
| "learning_rate": 1.65991673141289e-05, |
| "loss": 1.5882893371582032, |
| "step": 14250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.5008357763290405, |
| "learning_rate": 1.6519729220935818e-05, |
| "loss": 1.6250067138671875, |
| "step": 14300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.6415075063705444, |
| "learning_rate": 1.6440248039288936e-05, |
| "loss": 1.6562266540527344, |
| "step": 14350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.5272722244262695, |
| "learning_rate": 1.6360726022695772e-05, |
| "loss": 1.7231492614746093, |
| "step": 14400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 1.362077236175537, |
| "learning_rate": 1.6281165425821623e-05, |
| "loss": 1.7879104614257812, |
| "step": 14450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.7108640670776367, |
| "learning_rate": 1.620156850442564e-05, |
| "loss": 1.7281192016601563, |
| "step": 14500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.938968300819397, |
| "learning_rate": 1.612193751529688e-05, |
| "loss": 1.3673304748535156, |
| "step": 14550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.8007022142410278, |
| "learning_rate": 1.6042274716190307e-05, |
| "loss": 1.3540971374511719, |
| "step": 14600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 2.0127193927764893, |
| "learning_rate": 1.596258236576279e-05, |
| "loss": 1.4364506530761718, |
| "step": 14650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.9128210544586182, |
| "learning_rate": 1.588286272350905e-05, |
| "loss": 1.3906419372558594, |
| "step": 14700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.8449937105178833, |
| "learning_rate": 1.5803118049697622e-05, |
| "loss": 1.362480926513672, |
| "step": 14750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.9162813425064087, |
| "learning_rate": 1.5724946162719046e-05, |
| "loss": 1.4156272888183594, |
| "step": 14800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 2.2521564960479736, |
| "learning_rate": 1.56467545354907e-05, |
| "loss": 1.4028970336914062, |
| "step": 14850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.6907557249069214, |
| "learning_rate": 1.5566949021822878e-05, |
| "loss": 1.614343719482422, |
| "step": 14900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 1.3894444704055786, |
| "learning_rate": 1.5487127433609407e-05, |
| "loss": 1.702283172607422, |
| "step": 14950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.8563153743743896, |
| "learning_rate": 1.5407292034009256e-05, |
| "loss": 1.6103950500488282, |
| "step": 15000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 1.842654824256897, |
| "learning_rate": 1.5327445086572982e-05, |
| "loss": 1.4466780090332032, |
| "step": 15050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.6292502880096436, |
| "learning_rate": 1.5247588855178562e-05, |
| "loss": 1.3934080505371094, |
| "step": 15100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 2.048772096633911, |
| "learning_rate": 1.5167725603967189e-05, |
| "loss": 1.4326779174804687, |
| "step": 15150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.892514705657959, |
| "learning_rate": 1.5087857597279092e-05, |
| "loss": 1.3785092163085937, |
| "step": 15200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.91106379032135, |
| "learning_rate": 1.5007987099589327e-05, |
| "loss": 1.3833732604980469, |
| "step": 15250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.9060615301132202, |
| "learning_rate": 1.4928116375443582e-05, |
| "loss": 1.4064006042480468, |
| "step": 15300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 2.072082757949829, |
| "learning_rate": 1.4848247689393963e-05, |
| "loss": 1.361354217529297, |
| "step": 15350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.6465338468551636, |
| "learning_rate": 1.4768383305934785e-05, |
| "loss": 1.6327732849121093, |
| "step": 15400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.6932364702224731, |
| "learning_rate": 1.4688525489438387e-05, |
| "loss": 1.5731205749511719, |
| "step": 15450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.85811185836792, |
| "learning_rate": 1.4608676504090895e-05, |
| "loss": 1.3331919860839845, |
| "step": 15500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.8710098266601562, |
| "learning_rate": 1.4528838613828076e-05, |
| "loss": 0.9880081176757812, |
| "step": 15550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 2.09991192817688, |
| "learning_rate": 1.4449014082271099e-05, |
| "loss": 1.0064217376708984, |
| "step": 15600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 1.9291355609893799, |
| "learning_rate": 1.4369205172662382e-05, |
| "loss": 1.0313744354248047, |
| "step": 15650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 2.0349481105804443, |
| "learning_rate": 1.4289414147801419e-05, |
| "loss": 0.9832716369628907, |
| "step": 15700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 2.0009899139404297, |
| "learning_rate": 1.4209643269980633e-05, |
| "loss": 0.9983782196044921, |
| "step": 15750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 2.009942054748535, |
| "learning_rate": 1.4129894800921207e-05, |
| "loss": 1.0713680267333985, |
| "step": 15800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 2.0135700702667236, |
| "learning_rate": 1.405017100170899e-05, |
| "loss": 1.0285359191894532, |
| "step": 15850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 1.6245927810668945, |
| "learning_rate": 1.3970474132730356e-05, |
| "loss": 1.4042190551757812, |
| "step": 15900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 1.8125025033950806, |
| "learning_rate": 1.3890806453608157e-05, |
| "loss": 1.3763958740234374, |
| "step": 15950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 1.988704800605774, |
| "learning_rate": 1.3811170223137613e-05, |
| "loss": 1.4478152465820313, |
| "step": 16000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.6693629026412964, |
| "learning_rate": 1.3731567699222302e-05, |
| "loss": 1.8988229370117187, |
| "step": 16050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.6327314376831055, |
| "learning_rate": 1.365200113881011e-05, |
| "loss": 1.7711756896972657, |
| "step": 16100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.6666102409362793, |
| "learning_rate": 1.3572472797829294e-05, |
| "loss": 1.6638182067871095, |
| "step": 16150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.7380634546279907, |
| "learning_rate": 1.3492984931124453e-05, |
| "loss": 1.6675144958496093, |
| "step": 16200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.6629725694656372, |
| "learning_rate": 1.341353979239265e-05, |
| "loss": 1.617051239013672, |
| "step": 16250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.7792505025863647, |
| "learning_rate": 1.3334139634119483e-05, |
| "loss": 1.550360107421875, |
| "step": 16300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.6665221452713013, |
| "learning_rate": 1.3254786707515244e-05, |
| "loss": 1.5528482055664063, |
| "step": 16350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.690185785293579, |
| "learning_rate": 1.3175483262451064e-05, |
| "loss": 1.6005775451660156, |
| "step": 16400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 1.4981169700622559, |
| "learning_rate": 1.309623154739515e-05, |
| "loss": 1.6941712951660157, |
| "step": 16450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.5367906093597412, |
| "learning_rate": 1.3017033809349013e-05, |
| "loss": 1.6436634826660157, |
| "step": 16500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.8300843238830566, |
| "learning_rate": 1.2937892293783772e-05, |
| "loss": 1.196645278930664, |
| "step": 16550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.5175204277038574, |
| "learning_rate": 1.2858809244576494e-05, |
| "loss": 1.1753823852539063, |
| "step": 16600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.5690001249313354, |
| "learning_rate": 1.277978690394656e-05, |
| "loss": 1.1949736022949218, |
| "step": 16650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.6250964403152466, |
| "learning_rate": 1.2700827512392096e-05, |
| "loss": 1.1929642486572265, |
| "step": 16700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.5922327041625977, |
| "learning_rate": 1.2621933308626441e-05, |
| "loss": 1.1697836303710938, |
| "step": 16750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.7106561660766602, |
| "learning_rate": 1.2544682389876441e-05, |
| "loss": 1.2012087249755858, |
| "step": 16800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.7163861989974976, |
| "learning_rate": 1.2465923855290036e-05, |
| "loss": 1.2091240692138672, |
| "step": 16850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.6691616773605347, |
| "learning_rate": 1.2387237168649554e-05, |
| "loss": 1.314822998046875, |
| "step": 16900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 1.534340262413025, |
| "learning_rate": 1.2308624560936433e-05, |
| "loss": 1.409350128173828, |
| "step": 16950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.8180577754974365, |
| "learning_rate": 1.223008826103174e-05, |
| "loss": 1.37287109375, |
| "step": 17000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 1.8621281385421753, |
| "learning_rate": 1.2151630495653046e-05, |
| "loss": 1.2171533203125, |
| "step": 17050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.6037293672561646, |
| "learning_rate": 1.2073253489291228e-05, |
| "loss": 1.1620811462402343, |
| "step": 17100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 1.8049744367599487, |
| "learning_rate": 1.1994959464147432e-05, |
| "loss": 1.1776678466796875, |
| "step": 17150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.7445180416107178, |
| "learning_rate": 1.1916750640070056e-05, |
| "loss": 1.1673489379882813, |
| "step": 17200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.6527574062347412, |
| "learning_rate": 1.1838629234491833e-05, |
| "loss": 1.1704206848144532, |
| "step": 17250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.7367023229599, |
| "learning_rate": 1.1760597462366926e-05, |
| "loss": 1.203247299194336, |
| "step": 17300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 1.7566173076629639, |
| "learning_rate": 1.1682657536108159e-05, |
| "loss": 1.119371337890625, |
| "step": 17350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.5711489915847778, |
| "learning_rate": 1.160481166552426e-05, |
| "loss": 1.3711380004882812, |
| "step": 17400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.6498943567276, |
| "learning_rate": 1.1527062057757245e-05, |
| "loss": 1.3604568481445312, |
| "step": 17450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.685180902481079, |
| "learning_rate": 1.14494109172198e-05, |
| "loss": 1.1471074676513673, |
| "step": 17500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.5319304466247559, |
| "learning_rate": 1.1371860445532802e-05, |
| "loss": 0.8557035827636719, |
| "step": 17550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 1.7613927125930786, |
| "learning_rate": 1.1294412841462897e-05, |
| "loss": 0.8202173614501953, |
| "step": 17600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 1.6619144678115845, |
| "learning_rate": 1.121707030086014e-05, |
| "loss": 0.8459001159667969, |
| "step": 17650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 1.6346546411514282, |
| "learning_rate": 1.1139835016595775e-05, |
| "loss": 0.8195448303222657, |
| "step": 17700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 1.7639893293380737, |
| "learning_rate": 1.1062709178500023e-05, |
| "loss": 0.815630111694336, |
| "step": 17750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 1.736759901046753, |
| "learning_rate": 1.0985694973300004e-05, |
| "loss": 0.8889206695556641, |
| "step": 17800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 1.7116466760635376, |
| "learning_rate": 1.0908794584557749e-05, |
| "loss": 0.8401109313964844, |
| "step": 17850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 1.6388331651687622, |
| "learning_rate": 1.0832010192608289e-05, |
| "loss": 1.072819595336914, |
| "step": 17900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 1.7291067838668823, |
| "learning_rate": 1.075534397449782e-05, |
| "loss": 1.0581070709228515, |
| "step": 17950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 2.015186309814453, |
| "learning_rate": 1.0678798103921991e-05, |
| "loss": 1.088599395751953, |
| "step": 18000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.7369530200958252, |
| "learning_rate": 1.0602374751164265e-05, |
| "loss": 1.2541021728515624, |
| "step": 18050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.7617831230163574, |
| "learning_rate": 1.0526076083034415e-05, |
| "loss": 1.2021905517578124, |
| "step": 18100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.7245622873306274, |
| "learning_rate": 1.0449904262807043e-05, |
| "loss": 1.1469745635986328, |
| "step": 18150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.8544925451278687, |
| "learning_rate": 1.0373861450160277e-05, |
| "loss": 1.1247216033935548, |
| "step": 18200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.7715084552764893, |
| "learning_rate": 1.0297949801114521e-05, |
| "loss": 1.1083958435058594, |
| "step": 18250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.976885199546814, |
| "learning_rate": 1.0222171467971355e-05, |
| "loss": 1.0610853576660155, |
| "step": 18300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.7652332782745361, |
| "learning_rate": 1.0146528599252468e-05, |
| "loss": 1.0758203125, |
| "step": 18350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 1.835084319114685, |
| "learning_rate": 1.0071023339638775e-05, |
| "loss": 1.1495576477050782, |
| "step": 18400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 1.7060741186141968, |
| "learning_rate": 9.995657829909594e-06, |
| "loss": 1.2754519653320313, |
| "step": 18450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.588921308517456, |
| "learning_rate": 9.920434206881963e-06, |
| "loss": 1.1839189910888672, |
| "step": 18500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.7923060655593872, |
| "learning_rate": 9.845354603350045e-06, |
| "loss": 0.5976836395263672, |
| "step": 18550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.5783723592758179, |
| "learning_rate": 9.770421148024648e-06, |
| "loss": 0.6248928451538086, |
| "step": 18600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.575974464416504, |
| "learning_rate": 9.695635965472885e-06, |
| "loss": 0.629935417175293, |
| "step": 18650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.6231439113616943, |
| "learning_rate": 9.621001176057947e-06, |
| "loss": 0.6451435852050781, |
| "step": 18700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.5992910861968994, |
| "learning_rate": 9.546518895878952e-06, |
| "loss": 0.6247399520874023, |
| "step": 18750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.6657859086990356, |
| "learning_rate": 9.473676260970532e-06, |
| "loss": 0.6678472137451172, |
| "step": 18800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.9037035703659058, |
| "learning_rate": 9.399502175012888e-06, |
| "loss": 0.6817168426513672, |
| "step": 18850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.8003382682800293, |
| "learning_rate": 9.32548687838987e-06, |
| "loss": 0.753537826538086, |
| "step": 18900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 1.7780545949935913, |
| "learning_rate": 9.251632469636292e-06, |
| "loss": 0.8301016235351563, |
| "step": 18950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.8590174913406372, |
| "learning_rate": 9.177941042725378e-06, |
| "loss": 0.8636976623535156, |
| "step": 19000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 2.1247830390930176, |
| "learning_rate": 9.104414687009359e-06, |
| "loss": 0.6943458557128906, |
| "step": 19050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.5333961248397827, |
| "learning_rate": 9.031055487160248e-06, |
| "loss": 0.677508544921875, |
| "step": 19100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 1.7279947996139526, |
| "learning_rate": 8.957865523110738e-06, |
| "loss": 0.6879003143310547, |
| "step": 19150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.7829846143722534, |
| "learning_rate": 8.884846869995237e-06, |
| "loss": 0.6784191131591797, |
| "step": 19200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.758758783340454, |
| "learning_rate": 8.812001598091012e-06, |
| "loss": 0.6750822448730469, |
| "step": 19250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.7838960886001587, |
| "learning_rate": 8.739331772759506e-06, |
| "loss": 0.716082763671875, |
| "step": 19300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 1.786886215209961, |
| "learning_rate": 8.66683945438776e-06, |
| "loss": 0.6563807678222656, |
| "step": 19350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.738071322441101, |
| "learning_rate": 8.594526698330037e-06, |
| "loss": 0.9007662200927734, |
| "step": 19400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.8483673334121704, |
| "learning_rate": 8.522395554849496e-06, |
| "loss": 0.9248552703857422, |
| "step": 19450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.4636003971099854, |
| "learning_rate": 8.450448069060103e-06, |
| "loss": 0.7119151306152344, |
| "step": 19500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.5235882997512817, |
| "learning_rate": 8.378686280868617e-06, |
| "loss": 0.43539573669433596, |
| "step": 19550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 1.765830397605896, |
| "learning_rate": 8.30711222491678e-06, |
| "loss": 0.4487119674682617, |
| "step": 19600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 1.503596305847168, |
| "learning_rate": 8.235727930523598e-06, |
| "loss": 0.44962635040283205, |
| "step": 19650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 1.5412774085998535, |
| "learning_rate": 8.164535421627828e-06, |
| "loss": 0.4337549591064453, |
| "step": 19700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 1.6753467321395874, |
| "learning_rate": 8.093536716730584e-06, |
| "loss": 0.43847557067871096, |
| "step": 19750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 1.709405541419983, |
| "learning_rate": 8.022733828838124e-06, |
| "loss": 0.49284374237060546, |
| "step": 19800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 1.593888759613037, |
| "learning_rate": 7.952128765404746e-06, |
| "loss": 0.46281536102294923, |
| "step": 19850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 1.4703963994979858, |
| "learning_rate": 7.88172352827589e-06, |
| "loss": 0.6160999298095703, |
| "step": 19900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 1.6609055995941162, |
| "learning_rate": 7.81152011363138e-06, |
| "loss": 0.6013171768188477, |
| "step": 19950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 2.0141754150390625, |
| "learning_rate": 7.741520511928819e-06, |
| "loss": 0.6710720825195312, |
| "step": 20000 |
| }, |
| { |
| "epoch": 5.003166666666667, |
| "grad_norm": 1.949172854423523, |
| "learning_rate": 7.671726707847171e-06, |
| "loss": 1.0902951049804688, |
| "step": 20050 |
| }, |
| { |
| "epoch": 5.004833333333333, |
| "grad_norm": 1.7889505624771118, |
| "learning_rate": 7.602140680230471e-06, |
| "loss": 1.0368961334228515, |
| "step": 20100 |
| }, |
| { |
| "epoch": 5.0065, |
| "grad_norm": 1.857659935951233, |
| "learning_rate": 7.5327644020317124e-06, |
| "loss": 1.07505859375, |
| "step": 20150 |
| }, |
| { |
| "epoch": 5.008166666666667, |
| "grad_norm": 1.8462475538253784, |
| "learning_rate": 7.4635998402569545e-06, |
| "loss": 1.0229257202148438, |
| "step": 20200 |
| }, |
| { |
| "epoch": 5.009833333333333, |
| "grad_norm": 1.9917975664138794, |
| "learning_rate": 7.394648955909486e-06, |
| "loss": 1.0416568756103515, |
| "step": 20250 |
| }, |
| { |
| "epoch": 5.0115, |
| "grad_norm": 1.6674444675445557, |
| "learning_rate": 7.327286283161366e-06, |
| "loss": 1.0210507202148438, |
| "step": 20300 |
| }, |
| { |
| "epoch": 5.013166666666667, |
| "grad_norm": 1.5405360460281372, |
| "learning_rate": 7.2587642417068355e-06, |
| "loss": 1.0572975921630858, |
| "step": 20350 |
| }, |
| { |
| "epoch": 5.014833333333334, |
| "grad_norm": 1.8200913667678833, |
| "learning_rate": 7.190461685325534e-06, |
| "loss": 1.0956484985351562, |
| "step": 20400 |
| }, |
| { |
| "epoch": 5.0165, |
| "grad_norm": 2.1134114265441895, |
| "learning_rate": 7.122380550580563e-06, |
| "loss": 1.2368092346191406, |
| "step": 20450 |
| }, |
| { |
| "epoch": 6.001533333333334, |
| "grad_norm": 1.9816176891326904, |
| "learning_rate": 7.054522767757137e-06, |
| "loss": 1.5210466003417968, |
| "step": 20500 |
| }, |
| { |
| "epoch": 6.0032, |
| "grad_norm": 1.9438225030899048, |
| "learning_rate": 6.986890260807822e-06, |
| "loss": 1.4293966674804688, |
| "step": 20550 |
| }, |
| { |
| "epoch": 6.004866666666667, |
| "grad_norm": 1.842617392539978, |
| "learning_rate": 6.919484947298004e-06, |
| "loss": 1.4090545654296875, |
| "step": 20600 |
| }, |
| { |
| "epoch": 6.0065333333333335, |
| "grad_norm": 2.061250686645508, |
| "learning_rate": 6.852308738351514e-06, |
| "loss": 1.41453125, |
| "step": 20650 |
| }, |
| { |
| "epoch": 6.0082, |
| "grad_norm": 1.9555222988128662, |
| "learning_rate": 6.7853635385964585e-06, |
| "loss": 1.3674409484863281, |
| "step": 20700 |
| }, |
| { |
| "epoch": 6.0098666666666665, |
| "grad_norm": 2.1003122329711914, |
| "learning_rate": 6.718651246111195e-06, |
| "loss": 1.3919035339355468, |
| "step": 20750 |
| }, |
| { |
| "epoch": 6.011533333333333, |
| "grad_norm": 1.9350862503051758, |
| "learning_rate": 6.652173752370525e-06, |
| "loss": 1.3142074584960937, |
| "step": 20800 |
| }, |
| { |
| "epoch": 6.0132, |
| "grad_norm": 1.7350631952285767, |
| "learning_rate": 6.585932942192071e-06, |
| "loss": 1.3737037658691407, |
| "step": 20850 |
| }, |
| { |
| "epoch": 6.014866666666666, |
| "grad_norm": 1.7135426998138428, |
| "learning_rate": 6.519930693682841e-06, |
| "loss": 1.4159625244140626, |
| "step": 20900 |
| }, |
| { |
| "epoch": 7.0002, |
| "grad_norm": 1.8795244693756104, |
| "learning_rate": 6.454168878185959e-06, |
| "loss": 1.4865910339355468, |
| "step": 20950 |
| }, |
| { |
| "epoch": 7.0018666666666665, |
| "grad_norm": 1.576035499572754, |
| "learning_rate": 6.388649360227622e-06, |
| "loss": 1.3686183166503907, |
| "step": 21000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.685487985610962, |
| "learning_rate": 6.323373997464244e-06, |
| "loss": 0.8170787811279296, |
| "step": 21050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.7664726972579956, |
| "learning_rate": 6.25834464062976e-06, |
| "loss": 0.8275099945068359, |
| "step": 21100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.848008155822754, |
| "learning_rate": 6.193563133483182e-06, |
| "loss": 0.8038805389404297, |
| "step": 21150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.8948545455932617, |
| "learning_rate": 6.129031312756302e-06, |
| "loss": 0.7818323516845703, |
| "step": 21200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 1.9466619491577148, |
| "learning_rate": 6.064751008101625e-06, |
| "loss": 0.7761798095703125, |
| "step": 21250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 2.0793747901916504, |
| "learning_rate": 6.000724042040485e-06, |
| "loss": 0.7409461975097656, |
| "step": 21300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 1.97606360912323, |
| "learning_rate": 5.936952229911398e-06, |
| "loss": 0.7577394866943359, |
| "step": 21350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.0533604621887207, |
| "learning_rate": 5.873437379818551e-06, |
| "loss": 0.8345895385742188, |
| "step": 21400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 1.9522418975830078, |
| "learning_rate": 5.810181292580571e-06, |
| "loss": 0.9583216094970703, |
| "step": 21450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.4512640237808228, |
| "learning_rate": 5.747185761679443e-06, |
| "loss": 0.8717018127441406, |
| "step": 21500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.619828701019287, |
| "learning_rate": 5.684452573209695e-06, |
| "loss": 0.32352745056152343, |
| "step": 21550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.3838049173355103, |
| "learning_rate": 5.621983505827709e-06, |
| "loss": 0.31983285903930664, |
| "step": 21600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.41144859790802, |
| "learning_rate": 5.5597803307013315e-06, |
| "loss": 0.31932796478271486, |
| "step": 21650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.4198278188705444, |
| "learning_rate": 5.497844811459626e-06, |
| "loss": 0.32736438751220703, |
| "step": 21700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.3595664501190186, |
| "learning_rate": 5.4361787041429076e-06, |
| "loss": 0.30974163055419923, |
| "step": 21750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.3194063901901245, |
| "learning_rate": 5.376008987449838e-06, |
| "loss": 0.344722785949707, |
| "step": 21800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.8237543106079102, |
| "learning_rate": 5.31488146647034e-06, |
| "loss": 0.3528789138793945, |
| "step": 21850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.607699990272522, |
| "learning_rate": 5.2540285449239175e-06, |
| "loss": 0.39773021697998046, |
| "step": 21900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 1.5691676139831543, |
| "learning_rate": 5.193451948156284e-06, |
| "loss": 0.4504950714111328, |
| "step": 21950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.8552980422973633, |
| "learning_rate": 5.133153393678602e-06, |
| "loss": 0.5431232452392578, |
| "step": 22000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 1.9186702966690063, |
| "learning_rate": 5.073134591118771e-06, |
| "loss": 0.42785770416259766, |
| "step": 22050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.3061662912368774, |
| "learning_rate": 5.013397242172963e-06, |
| "loss": 0.4046816635131836, |
| "step": 22100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 1.4843345880508423, |
| "learning_rate": 4.95394304055738e-06, |
| "loss": 0.41539897918701174, |
| "step": 22150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.51164710521698, |
| "learning_rate": 4.8947736719602194e-06, |
| "loss": 0.40053470611572267, |
| "step": 22200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.4673335552215576, |
| "learning_rate": 4.835890813993903e-06, |
| "loss": 0.39697242736816407, |
| "step": 22250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.4424810409545898, |
| "learning_rate": 4.777296136147484e-06, |
| "loss": 0.42418296813964845, |
| "step": 22300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 1.4194953441619873, |
| "learning_rate": 4.71899129973933e-06, |
| "loss": 0.3895756912231445, |
| "step": 22350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.515943169593811, |
| "learning_rate": 4.660977957870013e-06, |
| "loss": 0.5933302307128906, |
| "step": 22400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.757616400718689, |
| "learning_rate": 4.603257755375458e-06, |
| "loss": 0.6306729125976562, |
| "step": 22450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.4781798124313354, |
| "learning_rate": 4.545832328780266e-06, |
| "loss": 0.4712517547607422, |
| "step": 22500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.182904601097107, |
| "learning_rate": 4.488703306251368e-06, |
| "loss": 0.24859485626220704, |
| "step": 22550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 1.5798511505126953, |
| "learning_rate": 4.431872307551807e-06, |
| "loss": 0.2671176719665527, |
| "step": 22600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 1.18501877784729, |
| "learning_rate": 4.3753409439948675e-06, |
| "loss": 0.2593963623046875, |
| "step": 22650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 1.144731044769287, |
| "learning_rate": 4.319110818398342e-06, |
| "loss": 0.24566781997680665, |
| "step": 22700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 1.4454432725906372, |
| "learning_rate": 4.263183525039118e-06, |
| "loss": 0.2538939094543457, |
| "step": 22750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 1.2419778108596802, |
| "learning_rate": 4.207560649607959e-06, |
| "loss": 0.2871647071838379, |
| "step": 22800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 1.2202904224395752, |
| "learning_rate": 4.152243769164566e-06, |
| "loss": 0.2689621734619141, |
| "step": 22850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 1.1901906728744507, |
| "learning_rate": 4.097234452092839e-06, |
| "loss": 0.36949249267578127, |
| "step": 22900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 1.2119687795639038, |
| "learning_rate": 4.042534258056423e-06, |
| "loss": 0.35492183685302736, |
| "step": 22950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 1.8483269214630127, |
| "learning_rate": 3.988144737954481e-06, |
| "loss": 0.46745311737060546, |
| "step": 23000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 2.0833899974823, |
| "learning_rate": 3.934067433877738e-06, |
| "loss": 0.7630504608154297, |
| "step": 23050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 2.058403253555298, |
| "learning_rate": 3.880303879064737e-06, |
| "loss": 0.7528865814208985, |
| "step": 23100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 2.207179307937622, |
| "learning_rate": 3.826855597858376e-06, |
| "loss": 0.7208507537841797, |
| "step": 23150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 2.088376998901367, |
| "learning_rate": 3.7737241056626865e-06, |
| "loss": 0.7016201019287109, |
| "step": 23200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 2.3134944438934326, |
| "learning_rate": 3.720910908899886e-06, |
| "loss": 0.7238885498046875, |
| "step": 23250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 2.1278038024902344, |
| "learning_rate": 3.6684175049676287e-06, |
| "loss": 0.6847061920166015, |
| "step": 23300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 2.9893929958343506, |
| "learning_rate": 3.6162453821965983e-06, |
| "loss": 0.7510233306884766, |
| "step": 23350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.13517689704895, |
| "learning_rate": 3.5643960198082636e-06, |
| "loss": 1.0430319976806641, |
| "step": 23400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 2.2010929584503174, |
| "learning_rate": 3.512870887872963e-06, |
| "loss": 1.0488350677490235, |
| "step": 23450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.5543965101242065, |
| "learning_rate": 3.4616714472682326e-06, |
| "loss": 0.9235614776611328, |
| "step": 23500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.5966523885726929, |
| "learning_rate": 3.4107991496373624e-06, |
| "loss": 0.27849832534790037, |
| "step": 23550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.2076386213302612, |
| "learning_rate": 3.3602554373482532e-06, |
| "loss": 0.2752775955200195, |
| "step": 23600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.2504218816757202, |
| "learning_rate": 3.3100417434525134e-06, |
| "loss": 0.27129779815673827, |
| "step": 23650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.3227320909500122, |
| "learning_rate": 3.261153879385321e-06, |
| "loss": 0.2785273551940918, |
| "step": 23700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.3441181182861328, |
| "learning_rate": 3.211597813036285e-06, |
| "loss": 0.28865924835205076, |
| "step": 23750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.1878491640090942, |
| "learning_rate": 3.162375979928301e-06, |
| "loss": 0.3303158950805664, |
| "step": 23800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.7056766748428345, |
| "learning_rate": 3.1134897756341128e-06, |
| "loss": 0.49927947998046873, |
| "step": 23850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.5071324110031128, |
| "learning_rate": 3.0649405862104772e-06, |
| "loss": 0.6785525512695313, |
| "step": 23900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 2.0096499919891357, |
| "learning_rate": 3.0167297881588686e-06, |
| "loss": 0.6695662689208984, |
| "step": 23950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.6812424659729004, |
| "learning_rate": 2.968858748386437e-06, |
| "loss": 0.6377016830444336, |
| "step": 24000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 1.3275223970413208, |
| "learning_rate": 2.9213288241672647e-06, |
| "loss": 0.36499958038330077, |
| "step": 24050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.76749587059021, |
| "learning_rate": 2.8741413631038877e-06, |
| "loss": 0.3513433837890625, |
| "step": 24100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 1.3912204504013062, |
| "learning_rate": 2.8272977030890675e-06, |
| "loss": 0.35687347412109377, |
| "step": 24150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.466273546218872, |
| "learning_rate": 2.7807991722678895e-06, |
| "loss": 0.35906410217285156, |
| "step": 24200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.5847675800323486, |
| "learning_rate": 2.734647089000072e-06, |
| "loss": 0.34384849548339846, |
| "step": 24250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.3546370267868042, |
| "learning_rate": 2.688842761822607e-06, |
| "loss": 0.4250868988037109, |
| "step": 24300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 1.9650839567184448, |
| "learning_rate": 2.6433874894126654e-06, |
| "loss": 0.5334797668457031, |
| "step": 24350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.6792277097702026, |
| "learning_rate": 2.5982825605507595e-06, |
| "loss": 0.6985765075683594, |
| "step": 24400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.343593955039978, |
| "learning_rate": 2.553529254084207e-06, |
| "loss": 0.7399945068359375, |
| "step": 24450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.3937770128250122, |
| "learning_rate": 2.509128838890883e-06, |
| "loss": 0.5113412094116211, |
| "step": 24500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.1141469478607178, |
| "learning_rate": 2.465082573843225e-06, |
| "loss": 0.23568450927734375, |
| "step": 24550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 1.1517999172210693, |
| "learning_rate": 2.421391707772566e-06, |
| "loss": 0.25807512283325196, |
| "step": 24600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 1.0975669622421265, |
| "learning_rate": 2.3780574794337005e-06, |
| "loss": 0.2481801414489746, |
| "step": 24650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 1.232245683670044, |
| "learning_rate": 2.335081117469777e-06, |
| "loss": 0.25898883819580076, |
| "step": 24700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 1.7666044235229492, |
| "learning_rate": 2.2924638403774597e-06, |
| "loss": 0.26850635528564454, |
| "step": 24750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 1.3763225078582764, |
| "learning_rate": 2.250206856472383e-06, |
| "loss": 0.31158803939819335, |
| "step": 24800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 1.831871747970581, |
| "learning_rate": 2.2083113638548874e-06, |
| "loss": 0.6100301742553711, |
| "step": 24850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 1.8685821294784546, |
| "learning_rate": 2.1667785503760577e-06, |
| "loss": 0.6580564117431641, |
| "step": 24900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 1.8616738319396973, |
| "learning_rate": 2.1256095936040294e-06, |
| "loss": 0.677899169921875, |
| "step": 24950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 2.0279905796051025, |
| "learning_rate": 2.0848056607906236e-06, |
| "loss": 0.5570061111450195, |
| "step": 25000 |
| }, |
| { |
| "epoch": 5.003166666666667, |
| "grad_norm": 1.8605475425720215, |
| "learning_rate": 2.0443679088382326e-06, |
| "loss": 0.7983790588378906, |
| "step": 25050 |
| }, |
| { |
| "epoch": 5.004833333333333, |
| "grad_norm": 1.5200269222259521, |
| "learning_rate": 2.0042974842670273e-06, |
| "loss": 0.8496729278564453, |
| "step": 25100 |
| }, |
| { |
| "epoch": 5.0065, |
| "grad_norm": 1.671268343925476, |
| "learning_rate": 1.9645955231824477e-06, |
| "loss": 0.8661923217773437, |
| "step": 25150 |
| }, |
| { |
| "epoch": 5.008166666666667, |
| "grad_norm": 1.4824550151824951, |
| "learning_rate": 1.926046169477867e-06, |
| "loss": 0.8316961669921875, |
| "step": 25200 |
| }, |
| { |
| "epoch": 5.009833333333333, |
| "grad_norm": 1.636396884918213, |
| "learning_rate": 1.8870770769158634e-06, |
| "loss": 0.8602359008789062, |
| "step": 25250 |
| }, |
| { |
| "epoch": 5.0115, |
| "grad_norm": 1.7281094789505005, |
| "learning_rate": 1.8484797713576217e-06, |
| "loss": 0.8439288330078125, |
| "step": 25300 |
| }, |
| { |
| "epoch": 5.013166666666667, |
| "grad_norm": 1.910600185394287, |
| "learning_rate": 1.8102553471416584e-06, |
| "loss": 0.9095307922363282, |
| "step": 25350 |
| }, |
| { |
| "epoch": 5.014833333333334, |
| "grad_norm": 1.7145682573318481, |
| "learning_rate": 1.772404888034283e-06, |
| "loss": 0.8995892333984375, |
| "step": 25400 |
| }, |
| { |
| "epoch": 5.0165, |
| "grad_norm": 4.496703624725342, |
| "learning_rate": 1.7349294671988892e-06, |
| "loss": 1.0104307556152343, |
| "step": 25450 |
| }, |
| { |
| "epoch": 6.001533333333334, |
| "grad_norm": 2.30837082862854, |
| "learning_rate": 1.6978301471655071e-06, |
| "loss": 1.1597572326660157, |
| "step": 25500 |
| }, |
| { |
| "epoch": 6.0032, |
| "grad_norm": 2.291196584701538, |
| "learning_rate": 1.6611079798006862e-06, |
| "loss": 1.1664737701416015, |
| "step": 25550 |
| }, |
| { |
| "epoch": 6.004866666666667, |
| "grad_norm": 2.1547529697418213, |
| "learning_rate": 1.6247640062776853e-06, |
| "loss": 1.171448745727539, |
| "step": 25600 |
| }, |
| { |
| "epoch": 6.0065333333333335, |
| "grad_norm": 2.3163843154907227, |
| "learning_rate": 1.5887992570469213e-06, |
| "loss": 1.124134292602539, |
| "step": 25650 |
| }, |
| { |
| "epoch": 6.0082, |
| "grad_norm": 2.36687970161438, |
| "learning_rate": 1.5532147518067913e-06, |
| "loss": 1.1288784790039061, |
| "step": 25700 |
| }, |
| { |
| "epoch": 6.0098666666666665, |
| "grad_norm": 2.190530776977539, |
| "learning_rate": 1.5180114994747251e-06, |
| "loss": 1.1292955780029297, |
| "step": 25750 |
| }, |
| { |
| "epoch": 6.011533333333333, |
| "grad_norm": 2.3975346088409424, |
| "learning_rate": 1.483190498158606e-06, |
| "loss": 1.1214398193359374, |
| "step": 25800 |
| }, |
| { |
| "epoch": 6.0132, |
| "grad_norm": 2.2489449977874756, |
| "learning_rate": 1.4487527351284557e-06, |
| "loss": 1.15248779296875, |
| "step": 25850 |
| }, |
| { |
| "epoch": 6.014866666666666, |
| "grad_norm": 2.1485483646392822, |
| "learning_rate": 1.414699186788459e-06, |
| "loss": 1.1434088134765625, |
| "step": 25900 |
| }, |
| { |
| "epoch": 7.0002, |
| "grad_norm": 2.2149617671966553, |
| "learning_rate": 1.3810308186492588e-06, |
| "loss": 1.2384030151367187, |
| "step": 25950 |
| }, |
| { |
| "epoch": 7.0018666666666665, |
| "grad_norm": 1.8158613443374634, |
| "learning_rate": 1.3477485853005988e-06, |
| "loss": 1.1393611907958985, |
| "step": 26000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.8052189350128174, |
| "learning_rate": 1.3148534303842473e-06, |
| "loss": 0.6258907699584961, |
| "step": 26050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.859277606010437, |
| "learning_rate": 1.2823462865672564e-06, |
| "loss": 0.6151887893676757, |
| "step": 26100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 1.9823532104492188, |
| "learning_rate": 1.2502280755154994e-06, |
| "loss": 0.5917576980590821, |
| "step": 26150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 1.922063946723938, |
| "learning_rate": 1.2184997078675525e-06, |
| "loss": 0.5746277236938476, |
| "step": 26200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 2.231234073638916, |
| "learning_rate": 1.187162083208867e-06, |
| "loss": 0.597679443359375, |
| "step": 26250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 2.0167667865753174, |
| "learning_rate": 1.1562160900462793e-06, |
| "loss": 0.5644707489013672, |
| "step": 26300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 2.836120128631592, |
| "learning_rate": 1.1256626057827961e-06, |
| "loss": 0.6299059677124024, |
| "step": 26350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.149437665939331, |
| "learning_rate": 1.0955024966927418e-06, |
| "loss": 0.9067845153808594, |
| "step": 26400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 2.1338465213775635, |
| "learning_rate": 1.065736617897175e-06, |
| "loss": 0.9151818084716797, |
| "step": 26450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.5150409936904907, |
| "learning_rate": 1.0363658133396608e-06, |
| "loss": 0.8027437591552734, |
| "step": 26500 |
| }, |
| { |
| "epoch": 2.002066666666667, |
| "grad_norm": 1.4865727424621582, |
| "learning_rate": 1.0073909157623346e-06, |
| "loss": 0.2347201919555664, |
| "step": 26550 |
| }, |
| { |
| "epoch": 2.0037333333333334, |
| "grad_norm": 1.0282039642333984, |
| "learning_rate": 9.788127466822905e-07, |
| "loss": 0.21864686965942381, |
| "step": 26600 |
| }, |
| { |
| "epoch": 2.0054, |
| "grad_norm": 1.0267033576965332, |
| "learning_rate": 9.506321163682908e-07, |
| "loss": 0.21448984146118164, |
| "step": 26650 |
| }, |
| { |
| "epoch": 2.0070666666666668, |
| "grad_norm": 1.1616917848587036, |
| "learning_rate": 9.23401560845214e-07, |
| "loss": 0.22048572540283204, |
| "step": 26700 |
| }, |
| { |
| "epoch": 2.0087333333333333, |
| "grad_norm": 1.1124871969223022, |
| "learning_rate": 8.960104036060512e-07, |
| "loss": 0.22924829483032227, |
| "step": 26750 |
| }, |
| { |
| "epoch": 2.0104, |
| "grad_norm": 1.0723026990890503, |
| "learning_rate": 8.690191328044084e-07, |
| "loss": 0.26663469314575194, |
| "step": 26800 |
| }, |
| { |
| "epoch": 2.0120666666666667, |
| "grad_norm": 1.8677761554718018, |
| "learning_rate": 8.42428513716168e-07, |
| "loss": 0.42245777130126955, |
| "step": 26850 |
| }, |
| { |
| "epoch": 2.013733333333333, |
| "grad_norm": 1.3421262502670288, |
| "learning_rate": 8.162393002576496e-07, |
| "loss": 0.5923210144042969, |
| "step": 26900 |
| }, |
| { |
| "epoch": 2.0154, |
| "grad_norm": 2.0107803344726562, |
| "learning_rate": 7.904522349642312e-07, |
| "loss": 0.5859238433837891, |
| "step": 26950 |
| }, |
| { |
| "epoch": 3.0007333333333333, |
| "grad_norm": 1.6371034383773804, |
| "learning_rate": 7.650680489693051e-07, |
| "loss": 0.557980842590332, |
| "step": 27000 |
| }, |
| { |
| "epoch": 3.0024, |
| "grad_norm": 1.261396050453186, |
| "learning_rate": 7.400874619835291e-07, |
| "loss": 0.3162060737609863, |
| "step": 27050 |
| }, |
| { |
| "epoch": 3.0040666666666667, |
| "grad_norm": 1.6194435358047485, |
| "learning_rate": 7.155111822744376e-07, |
| "loss": 0.29564001083374025, |
| "step": 27100 |
| }, |
| { |
| "epoch": 3.005733333333333, |
| "grad_norm": 1.2328189611434937, |
| "learning_rate": 6.913399066463582e-07, |
| "loss": 0.29919641494750976, |
| "step": 27150 |
| }, |
| { |
| "epoch": 3.0074, |
| "grad_norm": 1.320834994316101, |
| "learning_rate": 6.675743204206458e-07, |
| "loss": 0.2997510528564453, |
| "step": 27200 |
| }, |
| { |
| "epoch": 3.0090666666666666, |
| "grad_norm": 1.3888412714004517, |
| "learning_rate": 6.442150974162642e-07, |
| "loss": 0.28582775115966796, |
| "step": 27250 |
| }, |
| { |
| "epoch": 3.0107333333333335, |
| "grad_norm": 1.2989741563796997, |
| "learning_rate": 6.212628999306719e-07, |
| "loss": 0.362674560546875, |
| "step": 27300 |
| }, |
| { |
| "epoch": 3.0124, |
| "grad_norm": 1.8883610963821411, |
| "learning_rate": 5.987183787210437e-07, |
| "loss": 0.4638848876953125, |
| "step": 27350 |
| }, |
| { |
| "epoch": 3.0140666666666664, |
| "grad_norm": 1.7672468423843384, |
| "learning_rate": 5.765821729858322e-07, |
| "loss": 0.6259365463256836, |
| "step": 27400 |
| }, |
| { |
| "epoch": 3.0157333333333334, |
| "grad_norm": 1.3275811672210693, |
| "learning_rate": 5.548549103466338e-07, |
| "loss": 0.6669663238525391, |
| "step": 27450 |
| }, |
| { |
| "epoch": 4.000933333333333, |
| "grad_norm": 1.4393675327301025, |
| "learning_rate": 5.335372068303929e-07, |
| "loss": 0.460877799987793, |
| "step": 27500 |
| }, |
| { |
| "epoch": 4.0026, |
| "grad_norm": 1.035534381866455, |
| "learning_rate": 5.126296668519426e-07, |
| "loss": 0.20778966903686524, |
| "step": 27550 |
| }, |
| { |
| "epoch": 4.004266666666667, |
| "grad_norm": 1.0742510557174683, |
| "learning_rate": 4.921328831968668e-07, |
| "loss": 0.22145992279052734, |
| "step": 27600 |
| }, |
| { |
| "epoch": 4.005933333333333, |
| "grad_norm": 0.9139225482940674, |
| "learning_rate": 4.7204743700468734e-07, |
| "loss": 0.21092445373535157, |
| "step": 27650 |
| }, |
| { |
| "epoch": 4.0076, |
| "grad_norm": 1.1079626083374023, |
| "learning_rate": 4.5237389775239445e-07, |
| "loss": 0.21827945709228516, |
| "step": 27700 |
| }, |
| { |
| "epoch": 4.009266666666667, |
| "grad_norm": 1.6898112297058105, |
| "learning_rate": 4.331128232382925e-07, |
| "loss": 0.22777963638305665, |
| "step": 27750 |
| }, |
| { |
| "epoch": 4.010933333333333, |
| "grad_norm": 1.336200475692749, |
| "learning_rate": 4.142647595661969e-07, |
| "loss": 0.27035619735717775, |
| "step": 27800 |
| }, |
| { |
| "epoch": 4.0126, |
| "grad_norm": 2.4315943717956543, |
| "learning_rate": 3.958302411299369e-07, |
| "loss": 0.5600542068481446, |
| "step": 27850 |
| }, |
| { |
| "epoch": 4.014266666666667, |
| "grad_norm": 2.0830955505371094, |
| "learning_rate": 3.778097905982092e-07, |
| "loss": 0.6118479156494141, |
| "step": 27900 |
| }, |
| { |
| "epoch": 4.015933333333333, |
| "grad_norm": 2.0048952102661133, |
| "learning_rate": 3.60203918899768e-07, |
| "loss": 0.6278486251831055, |
| "step": 27950 |
| }, |
| { |
| "epoch": 5.0015, |
| "grad_norm": 1.8044838905334473, |
| "learning_rate": 3.430131252089236e-07, |
| "loss": 0.5059378433227539, |
| "step": 28000 |
| }, |
| { |
| "epoch": 5.003166666666667, |
| "grad_norm": 2.2389776706695557, |
| "learning_rate": 3.2623789693140006e-07, |
| "loss": 0.7381629180908204, |
| "step": 28050 |
| }, |
| { |
| "epoch": 5.004833333333333, |
| "grad_norm": 1.718327522277832, |
| "learning_rate": 3.0987870969051147e-07, |
| "loss": 0.7961085510253906, |
| "step": 28100 |
| }, |
| { |
| "epoch": 5.0065, |
| "grad_norm": 1.874211072921753, |
| "learning_rate": 2.939360273136776e-07, |
| "loss": 0.8100373077392579, |
| "step": 28150 |
| }, |
| { |
| "epoch": 5.008166666666667, |
| "grad_norm": 1.6292507648468018, |
| "learning_rate": 2.7871672728501106e-07, |
| "loss": 0.7771754455566406, |
| "step": 28200 |
| }, |
| { |
| "epoch": 5.009833333333333, |
| "grad_norm": 1.8408050537109375, |
| "learning_rate": 2.636000466905697e-07, |
| "loss": 0.8050576782226563, |
| "step": 28250 |
| }, |
| { |
| "epoch": 5.0115, |
| "grad_norm": 1.880056619644165, |
| "learning_rate": 2.489011830860577e-07, |
| "loss": 0.7910971069335937, |
| "step": 28300 |
| }, |
| { |
| "epoch": 5.013166666666667, |
| "grad_norm": 2.0853724479675293, |
| "learning_rate": 2.3462055322420572e-07, |
| "loss": 0.8597761535644531, |
| "step": 28350 |
| }, |
| { |
| "epoch": 5.014833333333334, |
| "grad_norm": 1.9814640283584595, |
| "learning_rate": 2.2075856199968493e-07, |
| "loss": 0.8509133911132812, |
| "step": 28400 |
| }, |
| { |
| "epoch": 5.0165, |
| "grad_norm": 5.22918176651001, |
| "learning_rate": 2.0731560243761937e-07, |
| "loss": 0.9620509338378906, |
| "step": 28450 |
| }, |
| { |
| "epoch": 6.001533333333334, |
| "grad_norm": 2.324591875076294, |
| "learning_rate": 1.9429205568245012e-07, |
| "loss": 1.0737631225585937, |
| "step": 28500 |
| }, |
| { |
| "epoch": 6.0032, |
| "grad_norm": 2.495436668395996, |
| "learning_rate": 1.816882909871237e-07, |
| "loss": 1.0971829223632812, |
| "step": 28550 |
| }, |
| { |
| "epoch": 6.004866666666667, |
| "grad_norm": 2.3265445232391357, |
| "learning_rate": 1.6950466570262223e-07, |
| "loss": 1.1054248046875, |
| "step": 28600 |
| }, |
| { |
| "epoch": 6.0065333333333335, |
| "grad_norm": 2.473598003387451, |
| "learning_rate": 1.5774152526783826e-07, |
| "loss": 1.0576546478271485, |
| "step": 28650 |
| }, |
| { |
| "epoch": 6.0082, |
| "grad_norm": 2.548072576522827, |
| "learning_rate": 1.463992031997724e-07, |
| "loss": 1.0655419921875, |
| "step": 28700 |
| }, |
| { |
| "epoch": 6.0098666666666665, |
| "grad_norm": 2.308596134185791, |
| "learning_rate": 1.354780210840828e-07, |
| "loss": 1.0685891723632812, |
| "step": 28750 |
| }, |
| { |
| "epoch": 6.011533333333333, |
| "grad_norm": 2.5529487133026123, |
| "learning_rate": 1.2497828856596228e-07, |
| "loss": 1.0625331115722656, |
| "step": 28800 |
| }, |
| { |
| "epoch": 6.0132, |
| "grad_norm": 2.527147054672241, |
| "learning_rate": 1.1490030334136703e-07, |
| "loss": 1.0997956848144532, |
| "step": 28850 |
| }, |
| { |
| "epoch": 6.014866666666666, |
| "grad_norm": 2.388906240463257, |
| "learning_rate": 1.0524435114856679e-07, |
| "loss": 1.0955790710449218, |
| "step": 28900 |
| }, |
| { |
| "epoch": 7.0002, |
| "grad_norm": 2.370323657989502, |
| "learning_rate": 9.601070576005123e-08, |
| "loss": 1.1930422210693359, |
| "step": 28950 |
| }, |
| { |
| "epoch": 7.0018666666666665, |
| "grad_norm": 1.9059492349624634, |
| "learning_rate": 8.719962897475952e-08, |
| "loss": 1.0841987609863282, |
| "step": 29000 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 2.007089138031006, |
| "learning_rate": 7.881137061066967e-08, |
| "loss": 0.64043701171875, |
| "step": 29050 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 2.127192974090576, |
| "learning_rate": 7.084616849770242e-08, |
| "loss": 0.6587284088134766, |
| "step": 29100 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 2.225581645965576, |
| "learning_rate": 6.330424847098504e-08, |
| "loss": 0.6331925964355469, |
| "step": 29150 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 2.0980887413024902, |
| "learning_rate": 5.6185824364449744e-08, |
| "loss": 0.6076303100585938, |
| "step": 29200 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 2.4637224674224854, |
| "learning_rate": 4.9491098004765256e-08, |
| "loss": 0.6287211227416992, |
| "step": 29250 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.9836453199386597, |
| "learning_rate": 4.322025920561967e-08, |
| "loss": 0.575589942932129, |
| "step": 29300 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 2.6764419078826904, |
| "learning_rate": 3.7373485762336454e-08, |
| "loss": 0.6231599807739258, |
| "step": 29350 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.39729905128479, |
| "learning_rate": 3.19509434468318e-08, |
| "loss": 0.8767373657226563, |
| "step": 29400 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 2.4006640911102295, |
| "learning_rate": 2.6952786002921725e-08, |
| "loss": 0.8967770385742188, |
| "step": 29450 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.45648992061615, |
| "learning_rate": 2.2379155141950547e-08, |
| "loss": 0.7814482879638672, |
| "step": 29500 |
| }, |
| { |
| "epoch": 1.0016666666666667, |
| "grad_norm": 1.808127522468567, |
| "learning_rate": 1.8230180538784115e-08, |
| "loss": 0.5795806884765625, |
| "step": 29550 |
| }, |
| { |
| "epoch": 1.0033333333333334, |
| "grad_norm": 1.9930639266967773, |
| "learning_rate": 1.4505979828129401e-08, |
| "loss": 0.6081678390502929, |
| "step": 29600 |
| }, |
| { |
| "epoch": 1.005, |
| "grad_norm": 2.1188387870788574, |
| "learning_rate": 1.1206658601197185e-08, |
| "loss": 0.595140495300293, |
| "step": 29650 |
| }, |
| { |
| "epoch": 1.0066666666666666, |
| "grad_norm": 2.0174291133880615, |
| "learning_rate": 8.332310402714427e-09, |
| "loss": 0.579749412536621, |
| "step": 29700 |
| }, |
| { |
| "epoch": 1.0083333333333333, |
| "grad_norm": 2.3788022994995117, |
| "learning_rate": 5.883016728259749e-09, |
| "loss": 0.6074051284790039, |
| "step": 29750 |
| }, |
| { |
| "epoch": 1.01, |
| "grad_norm": 1.9448168277740479, |
| "learning_rate": 3.858847021970258e-09, |
| "loss": 0.5630842208862304, |
| "step": 29800 |
| }, |
| { |
| "epoch": 1.0116666666666667, |
| "grad_norm": 2.65925669670105, |
| "learning_rate": 2.25985867455647e-09, |
| "loss": 0.6157372665405273, |
| "step": 29850 |
| }, |
| { |
| "epoch": 1.0133333333333334, |
| "grad_norm": 2.4260473251342773, |
| "learning_rate": 1.086097021681942e-09, |
| "loss": 0.8739725494384766, |
| "step": 29900 |
| }, |
| { |
| "epoch": 1.015, |
| "grad_norm": 2.43849515914917, |
| "learning_rate": 3.375953426842937e-10, |
| "loss": 0.8965128326416015, |
| "step": 29950 |
| }, |
| { |
| "epoch": 2.0004, |
| "grad_norm": 1.4195460081100464, |
| "learning_rate": 1.437485961597318e-11, |
| "loss": 0.7812672424316406, |
| "step": 30000 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 30000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 9223372036854775807, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.704080406538674e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
