| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.999903428295509, | |
| "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": 2.3104, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00028957528957528956, | |
| "loss": 1.4338, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005791505791505791, | |
| "loss": 1.2416, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0005999696654176312, | |
| "loss": 1.2139, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0005998690722049872, | |
| "loss": 1.1936, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0005996980613784548, | |
| "loss": 1.1791, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0005994566730961414, | |
| "loss": 1.1672, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005991449640427416, | |
| "loss": 1.1574, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005987630074162269, | |
| "loss": 1.1486, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005983108929106564, | |
| "loss": 1.1427, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005977887266951138, | |
| "loss": 1.1356, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0005971966313887766, | |
| "loss": 1.1272, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0005965347460321212, | |
| "loss": 1.1254, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0005958032260542726, | |
| "loss": 1.118, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0005950022432365049, | |
| "loss": 1.1146, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0005941319856719031, | |
| "loss": 1.1097, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0005931926577211924, | |
| "loss": 1.1063, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0005921844799647499, | |
| "loss": 1.1027, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0005911076891508052, | |
| "loss": 1.0999, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0005899625381398457, | |
| "loss": 1.0966, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0005887492958452381, | |
| "loss": 1.0931, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_alliteration_score": 0.3998726520216492, | |
| "eval_harmonic_meter_score": 0.1010308935212264, | |
| "eval_harmonic_rhyme_score": 0.35292231999777424, | |
| "eval_meter_score": 0.3077031352625674, | |
| "eval_rhyme_score": 0.7291149689204202, | |
| "eval_runtime": 3230.3537, | |
| "eval_samples_per_second": 0.836, | |
| "eval_steps_per_second": 0.026, | |
| "step": 5177 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0005874682471700796, | |
| "loss": 1.0865, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0005861196929402952, | |
| "loss": 1.0684, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0005847039498339947, | |
| "loss": 1.0705, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0005832213503071088, | |
| "loss": 1.0694, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0005816722425153186, | |
| "loss": 1.0678, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0005800569902322985, | |
| "loss": 1.0669, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0005783759727642932, | |
| "loss": 1.0634, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0005766295848610451, | |
| "loss": 1.062, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0005748182366230962, | |
| "loss": 1.0602, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0005729423534054853, | |
| "loss": 1.0587, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0005710023757178627, | |
| "loss": 1.0564, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.000568998759121046, | |
| "loss": 1.0547, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0005669319741200425, | |
| "loss": 1.0536, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0005648025060535602, | |
| "loss": 1.0517, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0005626108549800381, | |
| "loss": 1.0498, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0005603575355602176, | |
| "loss": 1.0482, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0005580430769362867, | |
| "loss": 1.0479, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0005556680226076214, | |
| "loss": 1.0446, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0005532329303031583, | |
| "loss": 1.0444, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0005507383718504232, | |
| "loss": 1.0416, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0005481849330412508, | |
| "loss": 1.0392, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_alliteration_score": 0.43183984747378457, | |
| "eval_harmonic_meter_score": 0.09296526825958593, | |
| "eval_harmonic_rhyme_score": 0.5441505816703797, | |
| "eval_meter_score": 0.30327833996094755, | |
| "eval_rhyme_score": 0.8429996906126559, | |
| "eval_runtime": 2339.9147, | |
| "eval_samples_per_second": 1.154, | |
| "eval_steps_per_second": 0.036, | |
| "step": 10354 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.000545573213494224, | |
| "loss": 1.0256, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0005429038265138671, | |
| "loss": 1.0149, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0005401773989466244, | |
| "loss": 1.0166, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0005373945710336596, | |
| "loss": 1.0157, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0005345559962605089, | |
| "loss": 1.0155, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0005316623412036252, | |
| "loss": 1.0135, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.000528714285373846, | |
| "loss": 1.0134, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0005257125210568268, | |
| "loss": 1.0118, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.0005226577531504722, | |
| "loss": 1.012, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.0005195506989994064, | |
| "loss": 1.0101, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0005163920882265211, | |
| "loss": 1.0097, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0005131826625616392, | |
| "loss": 1.0085, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.0005099231756673361, | |
| "loss": 1.0052, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0005066143929619589, | |
| "loss": 1.0064, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.000503257091439885, | |
| "loss": 1.0044, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0004998520594890613, | |
| "loss": 1.0015, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00049640009670587, | |
| "loss": 1.0014, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0004929020137073603, | |
| "loss": 1.0006, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0004893586319408926, | |
| "loss": 0.9997, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.0004857707834912409, | |
| "loss": 0.9991, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0004821393108851951, | |
| "loss": 0.9969, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_alliteration_score": 0.40482822655524603, | |
| "eval_harmonic_meter_score": 0.10445053009647749, | |
| "eval_harmonic_rhyme_score": 0.49816467657522634, | |
| "eval_meter_score": 0.3098783806475492, | |
| "eval_rhyme_score": 0.8195044525640993, | |
| "eval_runtime": 2332.5068, | |
| "eval_samples_per_second": 1.158, | |
| "eval_steps_per_second": 0.036, | |
| "step": 15531 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.0004784650668937127, | |
| "loss": 0.9695, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.0004747489143316642, | |
| "loss": 0.9666, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.0004709917258552203, | |
| "loss": 0.9684, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00046719438375692797, | |
| "loss": 0.9701, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.0004633577797585233, | |
| "loss": 0.9693, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 0.0004594828148015305, | |
| "loss": 0.9689, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.00045557039883569595, | |
| "loss": 0.9675, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 0.0004516214506053063, | |
| "loss": 0.966, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 0.000447636897433442, | |
| "loss": 0.9664, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.000443617675004216, | |
| "loss": 0.9645, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 0.00043956472714304834, | |
| "loss": 0.9635, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.0004354790055950309, | |
| "loss": 0.9621, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 0.0004313614698014302, | |
| "loss": 0.962, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00042721308667438394, | |
| "loss": 0.9609, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00042303483036984366, | |
| "loss": 0.9596, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00041882768205881495, | |
| "loss": 0.9578, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.00041459262969695184, | |
| "loss": 0.9568, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.0004103306677925571, | |
| "loss": 0.9552, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.00040604279717304357, | |
| "loss": 0.9534, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0004017300247499127, | |
| "loss": 0.9541, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_alliteration_score": 0.43839452395768513, | |
| "eval_harmonic_meter_score": 0.12086540507848588, | |
| "eval_harmonic_rhyme_score": 0.4986143276050798, | |
| "eval_meter_score": 0.338880767934961, | |
| "eval_rhyme_score": 0.8207791218734892, | |
| "eval_runtime": 2519.5282, | |
| "eval_samples_per_second": 1.072, | |
| "eval_steps_per_second": 0.034, | |
| "step": 20708 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.00039739336328230323, | |
| "loss": 0.9477, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.00039303383113916687, | |
| "loss": 0.9123, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 0.00038865245206012774, | |
| "loss": 0.9168, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00038425025491507883, | |
| "loss": 0.9193, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.0003798282734625755, | |
| "loss": 0.9201, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.0003753875461070794, | |
| "loss": 0.9188, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.0003709291156551129, | |
| "loss": 0.9192, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.0003664540290703784, | |
| "loss": 0.9173, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 0.00036196333722790264, | |
| "loss": 0.9166, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.00035745809466726145, | |
| "loss": 0.9145, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.0003529393593449451, | |
| "loss": 0.9144, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 0.00034840819238591994, | |
| "loss": 0.9139, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.0003438656578344473, | |
| "loss": 0.9126, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.0003393128224042155, | |
| "loss": 0.9119, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 0.0003347507552278469, | |
| "loss": 0.9099, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 0.00033018052760583447, | |
| "loss": 0.9072, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.0003256032127549717, | |
| "loss": 0.908, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 0.0003210198855563304, | |
| "loss": 0.9063, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 0.00031643162230284954, | |
| "loss": 0.9036, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.00031183950044659135, | |
| "loss": 0.9039, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 0.0003072445983457252, | |
| "loss": 0.9023, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_alliteration_score": 0.432449105490438, | |
| "eval_harmonic_meter_score": 0.11469247564220626, | |
| "eval_harmonic_rhyme_score": 0.5565104008090618, | |
| "eval_meter_score": 0.3233335011377191, | |
| "eval_rhyme_score": 0.8467553072173599, | |
| "eval_runtime": 2231.152, | |
| "eval_samples_per_second": 1.21, | |
| "eval_steps_per_second": 0.038, | |
| "step": 25885 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 0.0003026479950112996, | |
| "loss": 0.8805, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.00029805076985386, | |
| "loss": 0.8592, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 0.00029345400242997323, | |
| "loss": 0.8613, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 0.0002888587721887175, | |
| "loss": 0.8621, | |
| "step": 26750 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.0002842661582181979, | |
| "loss": 0.8628, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 0.000279677238992146, | |
| "loss": 0.8614, | |
| "step": 27250 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 0.00027509309211666463, | |
| "loss": 0.8617, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 0.0002705147940771754, | |
| "loss": 0.8606, | |
| "step": 27750 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 0.0002659434199856307, | |
| "loss": 0.8597, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 0.0002613800433280466, | |
| "loss": 0.8589, | |
| "step": 28250 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 0.0002568257357124192, | |
| "loss": 0.856, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 0.0002522815666170804, | |
| "loss": 0.8542, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 0.00024774860313955555, | |
| "loss": 0.8544, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 0.00024322790974597822, | |
| "loss": 0.8517, | |
| "step": 29250 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.00023872054802112475, | |
| "loss": 0.8522, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 0.00023422757641912385, | |
| "loss": 0.8512, | |
| "step": 29750 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 0.0002297500500149027, | |
| "loss": 0.8495, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 0.00022528902025642543, | |
| "loss": 0.8473, | |
| "step": 30250 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 0.00022084553471778432, | |
| "loss": 0.8451, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 0.00021642063685319983, | |
| "loss": 0.8414, | |
| "step": 30750 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 0.00021201536575198834, | |
| "loss": 0.8411, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_alliteration_score": 0.43164362519201227, | |
| "eval_harmonic_meter_score": 0.10449656298407087, | |
| "eval_harmonic_rhyme_score": 0.5587655117502281, | |
| "eval_meter_score": 0.3140831187981907, | |
| "eval_rhyme_score": 0.8511610758760736, | |
| "eval_runtime": 2100.1558, | |
| "eval_samples_per_second": 1.286, | |
| "eval_steps_per_second": 0.04, | |
| "step": 31062 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.00020763075589455592, | |
| "loss": 0.8075, | |
| "step": 31250 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 0.00020326783690947226, | |
| "loss": 0.796, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00019892763333168628, | |
| "loss": 0.7985, | |
| "step": 31750 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 0.000194611164361936, | |
| "loss": 0.7995, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 0.0001903194436274124, | |
| "loss": 0.7987, | |
| "step": 32250 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 0.0001860534789437309, | |
| "loss": 0.7977, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 0.00018181427207826875, | |
| "loss": 0.799, | |
| "step": 32750 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.0001776028185149218, | |
| "loss": 0.797, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 0.00017342010722033724, | |
| "loss": 0.799, | |
| "step": 33250 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 0.00016926712041167666, | |
| "loss": 0.7938, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00016514483332596397, | |
| "loss": 0.7928, | |
| "step": 33750 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 0.000161054213991073, | |
| "loss": 0.7948, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 0.00015699622299840705, | |
| "loss": 0.7913, | |
| "step": 34250 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 0.00015297181327732549, | |
| "loss": 0.7911, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 0.00014898192987136932, | |
| "loss": 0.7883, | |
| "step": 34750 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 0.000145027509716339, | |
| "loss": 0.7865, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 0.0001411094814202753, | |
| "loss": 0.7863, | |
| "step": 35250 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 0.00013722876504539635, | |
| "loss": 0.7826, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 0.00013338627189204153, | |
| "loss": 0.7829, | |
| "step": 35750 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 0.0001295829042846731, | |
| "loss": 0.7816, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_alliteration_score": 0.44621513944223107, | |
| "eval_harmonic_meter_score": 0.11710967524898745, | |
| "eval_harmonic_rhyme_score": 0.5515777130298052, | |
| "eval_meter_score": 0.33484803934701896, | |
| "eval_rhyme_score": 0.8477574870320731, | |
| "eval_runtime": 2122.0667, | |
| "eval_samples_per_second": 1.272, | |
| "eval_steps_per_second": 0.04, | |
| "step": 36239 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.00012581955535998448, | |
| "loss": 0.7792, | |
| "step": 36250 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.0001220971088571674, | |
| "loss": 0.734, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 0.00011841643891038518, | |
| "loss": 0.7362, | |
| "step": 36750 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 0.00011477840984350193, | |
| "loss": 0.7386, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 0.00011118387596711477, | |
| "loss": 0.7384, | |
| "step": 37250 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 0.00010763368137793809, | |
| "loss": 0.7364, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 0.00010412865976058613, | |
| "loss": 0.7367, | |
| "step": 37750 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 0.00010066963419180093, | |
| "loss": 0.7365, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 9.725741694717035e-05, | |
| "loss": 0.7343, | |
| "step": 38250 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 9.389280931038336e-05, | |
| "loss": 0.737, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 9.057660138506682e-05, | |
| "loss": 0.7345, | |
| "step": 38750 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 8.730957190924632e-05, | |
| "loss": 0.7332, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 8.409248807247727e-05, | |
| "loss": 0.7335, | |
| "step": 39250 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 8.092610533568725e-05, | |
| "loss": 0.7302, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 7.781116725377309e-05, | |
| "loss": 0.7291, | |
| "step": 39750 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 7.474840530099277e-05, | |
| "loss": 0.7303, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 7.173853869919559e-05, | |
| "loss": 0.7288, | |
| "step": 40250 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 6.878227424892822e-05, | |
| "loss": 0.7283, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 6.588030616345898e-05, | |
| "loss": 0.726, | |
| "step": 40750 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 6.303331590575642e-05, | |
| "loss": 0.7246, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 6.0241972028463316e-05, | |
| "loss": 0.7243, | |
| "step": 41250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_alliteration_score": 0.43950617283950616, | |
| "eval_harmonic_meter_score": 0.11078968787875117, | |
| "eval_harmonic_rhyme_score": 0.5599690603568903, | |
| "eval_meter_score": 0.3230506264598415, | |
| "eval_rhyme_score": 0.850570153117891, | |
| "eval_runtime": 2165.4131, | |
| "eval_samples_per_second": 1.247, | |
| "eval_steps_per_second": 0.039, | |
| "step": 41416 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 5.7506930016901755e-05, | |
| "loss": 0.7133, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 5.4828832135146994e-05, | |
| "loss": 0.6873, | |
| "step": 41750 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 5.2208307275205774e-05, | |
| "loss": 0.6887, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 4.9645970809335146e-05, | |
| "loss": 0.6883, | |
| "step": 42250 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 4.7142424445535695e-05, | |
| "loss": 0.6887, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 4.4698256086254156e-05, | |
| "loss": 0.688, | |
| "step": 42750 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 4.231403969032698e-05, | |
| "loss": 0.6895, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 3.999033513819922e-05, | |
| "loss": 0.6901, | |
| "step": 43250 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 3.772768810044874e-05, | |
| "loss": 0.6893, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 3.552662990964793e-05, | |
| "loss": 0.6873, | |
| "step": 43750 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 3.338767743559162e-05, | |
| "loss": 0.6895, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 3.131133296392159e-05, | |
| "loss": 0.6868, | |
| "step": 44250 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 2.929808407817651e-05, | |
| "loss": 0.6875, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 2.734840354529305e-05, | |
| "loss": 0.687, | |
| "step": 44750 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 2.5462749204587507e-05, | |
| "loss": 0.6864, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 2.3641563860241965e-05, | |
| "loss": 0.6841, | |
| "step": 45250 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 2.1885275177322048e-05, | |
| "loss": 0.6865, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 2.019429558134873e-05, | |
| "loss": 0.685, | |
| "step": 45750 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 1.856902216144962e-05, | |
| "loss": 0.6834, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 1.7009836577111302e-05, | |
| "loss": 0.6827, | |
| "step": 46250 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 1.551710496855515e-05, | |
| "loss": 0.6828, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_alliteration_score": 0.44789762340036565, | |
| "eval_harmonic_meter_score": 0.11082478811479961, | |
| "eval_harmonic_rhyme_score": 0.5453593958718314, | |
| "eval_meter_score": 0.328679747890422, | |
| "eval_rhyme_score": 0.8485504968592449, | |
| "eval_runtime": 2181.3379, | |
| "eval_samples_per_second": 1.238, | |
| "eval_steps_per_second": 0.039, | |
| "step": 46593 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.4091177870757209e-05, | |
| "loss": 0.673, | |
| "step": 46750 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 1.2732390131132907e-05, | |
| "loss": 0.6644, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 1.1441060830905591e-05, | |
| "loss": 0.6632, | |
| "step": 47250 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 1.0217493210177418e-05, | |
| "loss": 0.6641, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 9.061974596719934e-06, | |
| "loss": 0.6654, | |
| "step": 47750 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 7.974776338501631e-06, | |
| "loss": 0.6633, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 6.956153739967863e-06, | |
| "loss": 0.6656, | |
| "step": 48250 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 6.0063460020883915e-06, | |
| "loss": 0.664, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 5.125576166185996e-06, | |
| "loss": 0.6638, | |
| "step": 48750 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 4.314051061560497e-06, | |
| "loss": 0.6645, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 3.571961256919276e-06, | |
| "loss": 0.6624, | |
| "step": 49250 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.8994810156265035e-06, | |
| "loss": 0.663, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 2.2967682547812782e-06, | |
| "loss": 0.6631, | |
| "step": 49750 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 1.7639645081341524e-06, | |
| "loss": 0.663, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 1.3011948928511873e-06, | |
| "loss": 0.6641, | |
| "step": 50250 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 9.085680801330208e-07, | |
| "loss": 0.6634, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 5.861762696956151e-07, | |
| "loss": 0.6629, | |
| "step": 50750 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 3.340951681194082e-07, | |
| "loss": 0.6652, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 1.523839710711683e-07, | |
| "loss": 0.663, | |
| "step": 51250 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 4.108534940331365e-08, | |
| "loss": 0.6623, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 2.2543913346106945e-10, | |
| "loss": 0.6634, | |
| "step": 51750 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_alliteration_score": 0.4406211936662607, | |
| "eval_harmonic_meter_score": 0.11891243404089967, | |
| "eval_harmonic_rhyme_score": 0.5297086635780841, | |
| "eval_meter_score": 0.33801080805418304, | |
| "eval_rhyme_score": 0.8432321768961265, | |
| "eval_runtime": 2127.802, | |
| "eval_samples_per_second": 1.269, | |
| "eval_steps_per_second": 0.04, | |
| "step": 51770 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 51770, | |
| "total_flos": 1.310953664443056e+18, | |
| "train_loss": 0.8821613311353097, | |
| "train_runtime": 68769.7958, | |
| "train_samples_per_second": 96.366, | |
| "train_steps_per_second": 0.753 | |
| } | |
| ], | |
| "max_steps": 51770, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.310953664443056e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |