| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.999915498738517, | |
| "global_step": 51770, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1583011583011583e-06, | |
| "loss": 2.5038, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00028957528957528956, | |
| "loss": 1.5196, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005791505791505791, | |
| "loss": 1.297, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0005999696654176312, | |
| "loss": 1.2658, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0005998690722049872, | |
| "loss": 1.2455, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0005996980613784548, | |
| "loss": 1.2319, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0005994566730961414, | |
| "loss": 1.2214, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005991449640427416, | |
| "loss": 1.213, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005987630074162269, | |
| "loss": 1.2059, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005983108929106564, | |
| "loss": 1.2019, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005977887266951138, | |
| "loss": 1.1952, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0005971966313887766, | |
| "loss": 1.1878, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0005965347460321212, | |
| "loss": 1.1876, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0005958032260542726, | |
| "loss": 1.1812, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0005950022432365049, | |
| "loss": 1.1786, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0005941319856719031, | |
| "loss": 1.1741, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0005931926577211924, | |
| "loss": 1.1714, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0005921844799647499, | |
| "loss": 1.1685, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0005911076891508052, | |
| "loss": 1.166, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0005899625381398457, | |
| "loss": 1.1635, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0005887492958452381, | |
| "loss": 1.1602, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_alliteration_score": 0.33734156646083846, | |
| "eval_harmonic_meter_score": 0.08910893104383713, | |
| "eval_harmonic_rhyme_score": 0.18416322432066148, | |
| "eval_meter_score": 0.27888669912946146, | |
| "eval_rhyme_score": 0.6019525602542674, | |
| "eval_runtime": 1534.0921, | |
| "eval_samples_per_second": 1.76, | |
| "eval_steps_per_second": 0.22, | |
| "step": 5177 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0005874682471700796, | |
| "loss": 1.1552, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0005861196929402952, | |
| "loss": 1.1417, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0005847039498339947, | |
| "loss": 1.1433, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0005832213503071088, | |
| "loss": 1.1418, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0005816722425153186, | |
| "loss": 1.1408, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0005800569902322985, | |
| "loss": 1.1394, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0005783759727642932, | |
| "loss": 1.1364, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0005766295848610451, | |
| "loss": 1.1355, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0005748182366230962, | |
| "loss": 1.1337, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0005729423534054853, | |
| "loss": 1.1325, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0005710023757178627, | |
| "loss": 1.1299, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.000568998759121046, | |
| "loss": 1.1285, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0005669319741200425, | |
| "loss": 1.1278, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0005648025060535602, | |
| "loss": 1.1261, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0005626108549800381, | |
| "loss": 1.1245, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0005603575355602176, | |
| "loss": 1.123, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0005580430769362867, | |
| "loss": 1.123, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0005556680226076214, | |
| "loss": 1.1201, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0005532329303031583, | |
| "loss": 1.12, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0005507383718504232, | |
| "loss": 1.1176, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0005481849330412508, | |
| "loss": 1.1157, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_alliteration_score": 0.38736706413148564, | |
| "eval_harmonic_meter_score": 0.0895880700551104, | |
| "eval_harmonic_rhyme_score": 0.41931891077613487, | |
| "eval_meter_score": 0.3107654652536241, | |
| "eval_rhyme_score": 0.7829217375053705, | |
| "eval_runtime": 1335.1994, | |
| "eval_samples_per_second": 2.022, | |
| "eval_steps_per_second": 0.253, | |
| "step": 10354 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.000545573213494224, | |
| "loss": 1.1078, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0005429038265138671, | |
| "loss": 1.1014, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0005401773989466244, | |
| "loss": 1.1022, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0005373945710336596, | |
| "loss": 1.1014, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0005345559962605089, | |
| "loss": 1.1008, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0005316623412036252, | |
| "loss": 1.0992, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.000528714285373846, | |
| "loss": 1.0989, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0005257125210568268, | |
| "loss": 1.0979, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.0005226577531504722, | |
| "loss": 1.0984, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.0005195506989994064, | |
| "loss": 1.0965, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0005163920882265211, | |
| "loss": 1.0963, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0005131826625616392, | |
| "loss": 1.0953, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.0005099231756673361, | |
| "loss": 1.0929, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0005066143929619589, | |
| "loss": 1.0941, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.000503257091439885, | |
| "loss": 1.0925, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0004998520594890613, | |
| "loss": 1.0895, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00049640009670587, | |
| "loss": 1.09, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0004929020137073603, | |
| "loss": 1.0894, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0004893586319408926, | |
| "loss": 1.0892, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.0004857707834912409, | |
| "loss": 1.0881, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0004821393108851951, | |
| "loss": 1.0867, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_alliteration_score": 0.3956901936289819, | |
| "eval_harmonic_meter_score": 0.10651791396667262, | |
| "eval_harmonic_rhyme_score": 0.4171799492877077, | |
| "eval_meter_score": 0.3202445754833885, | |
| "eval_rhyme_score": 0.7817402156606966, | |
| "eval_runtime": 1356.2806, | |
| "eval_samples_per_second": 1.991, | |
| "eval_steps_per_second": 0.249, | |
| "step": 15531 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.0004784650668937127, | |
| "loss": 1.0717, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.0004747489143316642, | |
| "loss": 1.0692, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.0004709917258552203, | |
| "loss": 1.071, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00046719438375692797, | |
| "loss": 1.0713, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.0004633577797585233, | |
| "loss": 1.0712, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 0.0004594828148015305, | |
| "loss": 1.0705, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.00045557039883569595, | |
| "loss": 1.0694, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 0.0004516214506053063, | |
| "loss": 1.0682, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 0.000447636897433442, | |
| "loss": 1.07, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.000443617675004216, | |
| "loss": 1.0672, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 0.00043956472714304834, | |
| "loss": 1.0667, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.0004354790055950309, | |
| "loss": 1.0662, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 0.0004313614698014302, | |
| "loss": 1.0656, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00042721308667438394, | |
| "loss": 1.0649, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00042303483036984366, | |
| "loss": 1.0636, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00041882768205881495, | |
| "loss": 1.0629, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.00041459262969695184, | |
| "loss": 1.0627, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.0004103306677925571, | |
| "loss": 1.0611, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.00040604279717304357, | |
| "loss": 1.0595, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0004017300247499127, | |
| "loss": 1.0609, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_alliteration_score": 0.41442581461562794, | |
| "eval_harmonic_meter_score": 0.0987295348952947, | |
| "eval_harmonic_rhyme_score": 0.4578505943539395, | |
| "eval_meter_score": 0.3220205905401726, | |
| "eval_rhyme_score": 0.8007234223820661, | |
| "eval_runtime": 1328.0652, | |
| "eval_samples_per_second": 2.033, | |
| "eval_steps_per_second": 0.255, | |
| "step": 20708 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.00039739336328230323, | |
| "loss": 1.0365, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.00039303383113916687, | |
| "loss": 1.0371, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 0.00038865245206012774, | |
| "loss": 1.0398, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00038425025491507883, | |
| "loss": 1.0413, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.0003798282734625755, | |
| "loss": 1.0416, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.0003753875461070794, | |
| "loss": 1.0409, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.0003709291156551129, | |
| "loss": 1.0413, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.0003664540290703784, | |
| "loss": 1.0401, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 0.00036196333722790264, | |
| "loss": 1.0395, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.00035745809466726145, | |
| "loss": 1.038, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.0003529393593449451, | |
| "loss": 1.0382, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 0.00034840819238591994, | |
| "loss": 1.0386, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.0003438656578344473, | |
| "loss": 1.0371, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.0003393128224042155, | |
| "loss": 1.0361, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 0.0003347507552278469, | |
| "loss": 1.0355, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 0.00033018052760583447, | |
| "loss": 1.0334, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.0003256032127549717, | |
| "loss": 1.0341, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 0.0003210198855563304, | |
| "loss": 1.0338, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 0.00031643162230284954, | |
| "loss": 1.0315, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.00031183950044659135, | |
| "loss": 1.0321, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 0.0003072445983457252, | |
| "loss": 1.0313, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_alliteration_score": 0.41847826086956524, | |
| "eval_harmonic_meter_score": 0.10429340770685562, | |
| "eval_harmonic_rhyme_score": 0.44835226139123785, | |
| "eval_meter_score": 0.31689277852422454, | |
| "eval_rhyme_score": 0.7882407888398814, | |
| "eval_runtime": 1355.1583, | |
| "eval_samples_per_second": 1.992, | |
| "eval_steps_per_second": 0.249, | |
| "step": 25885 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 0.0003026479950112996, | |
| "loss": 1.0191, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.00029805076985386, | |
| "loss": 1.0078, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 0.00029345400242997323, | |
| "loss": 1.0091, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 0.0002888587721887175, | |
| "loss": 1.0091, | |
| "step": 26750 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.0002842661582181979, | |
| "loss": 1.01, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 0.000279677238992146, | |
| "loss": 1.009, | |
| "step": 27250 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 0.00027509309211666463, | |
| "loss": 1.0085, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 0.0002705147940771754, | |
| "loss": 1.0078, | |
| "step": 27750 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 0.0002659434199856307, | |
| "loss": 1.0075, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 0.0002613800433280466, | |
| "loss": 1.0069, | |
| "step": 28250 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 0.0002568257357124192, | |
| "loss": 1.0053, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 0.0002522815666170804, | |
| "loss": 1.0038, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 0.00024774860313955555, | |
| "loss": 1.0052, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 0.00024322790974597822, | |
| "loss": 1.003, | |
| "step": 29250 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.00023872054802112475, | |
| "loss": 1.0028, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 0.00023422757641912385, | |
| "loss": 1.0033, | |
| "step": 29750 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 0.0002297500500149027, | |
| "loss": 1.001, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 0.00022528902025642543, | |
| "loss": 1.0013, | |
| "step": 30250 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 0.00022084553471778432, | |
| "loss": 0.9993, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 0.00021642063685319983, | |
| "loss": 0.9969, | |
| "step": 30750 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 0.00021201536575198834, | |
| "loss": 0.9967, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_alliteration_score": 0.4239095315024233, | |
| "eval_harmonic_meter_score": 0.10471658916077409, | |
| "eval_harmonic_rhyme_score": 0.5099131421399749, | |
| "eval_meter_score": 0.3276790822599601, | |
| "eval_rhyme_score": 0.8216395530060676, | |
| "eval_runtime": 1328.5938, | |
| "eval_samples_per_second": 2.032, | |
| "eval_steps_per_second": 0.254, | |
| "step": 31062 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.00020763075589455592, | |
| "loss": 0.98, | |
| "step": 31250 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 0.00020326783690947226, | |
| "loss": 0.9726, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00019892763333168628, | |
| "loss": 0.9741, | |
| "step": 31750 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 0.000194611164361936, | |
| "loss": 0.9746, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 0.0001903194436274124, | |
| "loss": 0.9738, | |
| "step": 32250 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 0.0001860534789437309, | |
| "loss": 0.9742, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 0.00018181427207826875, | |
| "loss": 0.9748, | |
| "step": 32750 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.0001776028185149218, | |
| "loss": 0.9728, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 0.00017342010722033724, | |
| "loss": 0.9748, | |
| "step": 33250 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 0.00016926712041167666, | |
| "loss": 0.9698, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00016514483332596397, | |
| "loss": 0.9703, | |
| "step": 33750 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 0.000161054213991073, | |
| "loss": 0.9726, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 0.00015699622299840705, | |
| "loss": 0.9695, | |
| "step": 34250 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 0.00015297181327732549, | |
| "loss": 0.9697, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 0.00014898192987136932, | |
| "loss": 0.9681, | |
| "step": 34750 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 0.000145027509716339, | |
| "loss": 0.9671, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 0.0001411094814202753, | |
| "loss": 0.9667, | |
| "step": 35250 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 0.00013722876504539635, | |
| "loss": 0.9644, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 0.00013338627189204153, | |
| "loss": 0.9647, | |
| "step": 35750 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 0.0001295829042846731, | |
| "loss": 0.9636, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_alliteration_score": 0.439117929050815, | |
| "eval_harmonic_meter_score": 0.1013845208427906, | |
| "eval_harmonic_rhyme_score": 0.5635069892236091, | |
| "eval_meter_score": 0.32354736135648304, | |
| "eval_rhyme_score": 0.8347036594862961, | |
| "eval_runtime": 1288.6194, | |
| "eval_samples_per_second": 2.095, | |
| "eval_steps_per_second": 0.262, | |
| "step": 36239 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.00012581955535998448, | |
| "loss": 0.9639, | |
| "step": 36250 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.0001220971088571674, | |
| "loss": 0.9378, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 0.00011841643891038518, | |
| "loss": 0.9388, | |
| "step": 36750 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 0.00011477840984350193, | |
| "loss": 0.9395, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 0.00011118387596711477, | |
| "loss": 0.9403, | |
| "step": 37250 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 0.00010763368137793809, | |
| "loss": 0.9399, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 0.00010412865976058613, | |
| "loss": 0.939, | |
| "step": 37750 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 0.00010066963419180093, | |
| "loss": 0.9391, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 9.725741694717035e-05, | |
| "loss": 0.9372, | |
| "step": 38250 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 9.389280931038336e-05, | |
| "loss": 0.9398, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 9.057660138506682e-05, | |
| "loss": 0.9387, | |
| "step": 38750 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 8.730957190924632e-05, | |
| "loss": 0.9378, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 8.409248807247727e-05, | |
| "loss": 0.9378, | |
| "step": 39250 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 8.092610533568725e-05, | |
| "loss": 0.9366, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 7.781116725377309e-05, | |
| "loss": 0.935, | |
| "step": 39750 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 7.474840530099277e-05, | |
| "loss": 0.9356, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 7.173853869919559e-05, | |
| "loss": 0.9341, | |
| "step": 40250 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 6.878227424892822e-05, | |
| "loss": 0.936, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 6.588030616345898e-05, | |
| "loss": 0.9338, | |
| "step": 40750 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 6.303331590575642e-05, | |
| "loss": 0.9329, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 6.0241972028463316e-05, | |
| "loss": 0.9333, | |
| "step": 41250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_alliteration_score": 0.42669584245076586, | |
| "eval_harmonic_meter_score": 0.10211813060163026, | |
| "eval_harmonic_rhyme_score": 0.5177186665744951, | |
| "eval_meter_score": 0.323707550822104, | |
| "eval_rhyme_score": 0.8264104348934438, | |
| "eval_runtime": 1297.8514, | |
| "eval_samples_per_second": 2.08, | |
| "eval_steps_per_second": 0.26, | |
| "step": 41416 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 5.7506930016901755e-05, | |
| "loss": 0.9268, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 5.4828832135146994e-05, | |
| "loss": 0.9115, | |
| "step": 41750 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 5.2208307275205774e-05, | |
| "loss": 0.9122, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 4.9645970809335146e-05, | |
| "loss": 0.9119, | |
| "step": 42250 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 4.7142424445535695e-05, | |
| "loss": 0.912, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 4.4698256086254156e-05, | |
| "loss": 0.9114, | |
| "step": 42750 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 4.231403969032698e-05, | |
| "loss": 0.9132, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 3.999033513819922e-05, | |
| "loss": 0.9138, | |
| "step": 43250 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 3.772768810044874e-05, | |
| "loss": 0.9122, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 3.552662990964793e-05, | |
| "loss": 0.9115, | |
| "step": 43750 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 3.338767743559162e-05, | |
| "loss": 0.9135, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 3.131133296392159e-05, | |
| "loss": 0.9113, | |
| "step": 44250 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 2.929808407817651e-05, | |
| "loss": 0.9122, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 2.734840354529305e-05, | |
| "loss": 0.9112, | |
| "step": 44750 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 2.5462749204587507e-05, | |
| "loss": 0.9105, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 2.3641563860241965e-05, | |
| "loss": 0.9095, | |
| "step": 45250 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 2.1885275177322048e-05, | |
| "loss": 0.9109, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 2.019429558134873e-05, | |
| "loss": 0.9095, | |
| "step": 45750 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 1.856902216144962e-05, | |
| "loss": 0.9089, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 1.7009836577111302e-05, | |
| "loss": 0.9091, | |
| "step": 46250 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 1.551710496855515e-05, | |
| "loss": 0.9084, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_alliteration_score": 0.43871778755499685, | |
| "eval_harmonic_meter_score": 0.10228823262732262, | |
| "eval_harmonic_rhyme_score": 0.5493860049384648, | |
| "eval_meter_score": 0.32683853787547473, | |
| "eval_rhyme_score": 0.8411704041762098, | |
| "eval_runtime": 1292.8693, | |
| "eval_samples_per_second": 2.088, | |
| "eval_steps_per_second": 0.261, | |
| "step": 46593 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.4091177870757209e-05, | |
| "loss": 0.9044, | |
| "step": 46750 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 1.2732390131132907e-05, | |
| "loss": 0.8983, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 1.1441060830905591e-05, | |
| "loss": 0.8971, | |
| "step": 47250 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 1.0217493210177418e-05, | |
| "loss": 0.8978, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 9.061974596719934e-06, | |
| "loss": 0.8991, | |
| "step": 47750 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 7.974776338501631e-06, | |
| "loss": 0.8969, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 6.956153739967863e-06, | |
| "loss": 0.9, | |
| "step": 48250 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 6.0063460020883915e-06, | |
| "loss": 0.898, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 5.125576166185996e-06, | |
| "loss": 0.8977, | |
| "step": 48750 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 4.314051061560497e-06, | |
| "loss": 0.8983, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 3.571961256919276e-06, | |
| "loss": 0.897, | |
| "step": 49250 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.8994810156265035e-06, | |
| "loss": 0.8978, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 2.2967682547812782e-06, | |
| "loss": 0.8968, | |
| "step": 49750 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 1.7639645081341524e-06, | |
| "loss": 0.8982, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 1.3011948928511873e-06, | |
| "loss": 0.8971, | |
| "step": 50250 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 9.085680801330208e-07, | |
| "loss": 0.8975, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 5.861762696956151e-07, | |
| "loss": 0.897, | |
| "step": 50750 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 3.340951681194082e-07, | |
| "loss": 0.8984, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 1.523839710711683e-07, | |
| "loss": 0.897, | |
| "step": 51250 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 4.108534940331365e-08, | |
| "loss": 0.8969, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 2.2543913346106945e-10, | |
| "loss": 0.8982, | |
| "step": 51750 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_alliteration_score": 0.4247429105640386, | |
| "eval_harmonic_meter_score": 0.11325822463626577, | |
| "eval_harmonic_rhyme_score": 0.5430495530851162, | |
| "eval_meter_score": 0.33602397160908126, | |
| "eval_rhyme_score": 0.8321207205854919, | |
| "eval_runtime": 1301.4663, | |
| "eval_samples_per_second": 2.075, | |
| "eval_steps_per_second": 0.26, | |
| "step": 51770 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 51770, | |
| "total_flos": 5.976175248261775e+17, | |
| "train_loss": 0.575801368235897, | |
| "train_runtime": 18250.6753, | |
| "train_samples_per_second": 363.114, | |
| "train_steps_per_second": 2.837 | |
| } | |
| ], | |
| "max_steps": 51770, | |
| "num_train_epochs": 10, | |
| "total_flos": 5.976175248261775e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |