Source symbols: src/transformers/optimization.py
Deterministic, value-free symbol metadata extracted from public source. Implementations and literal values are not copied into this page.
- Revision: main@9a0fe3f5dd36ffe1888133f09eb03f1eb14b8a6e · 2026-07-20T12:47:48Z
- Declarations: 51
- Referenced calls: 63
Declared symbols
_get_constant_lambda— def at line 35get_constant_schedule— def at line 39; referencesLambdaLRget_reduce_on_plateau_schedule— def at line 56; referencesReduceLROnPlateau_get_constant_schedule_with_warmup_lr_lambda— def at line 74; referencesfloat,maxget_constant_schedule_with_warmup— def at line 80; referencespartial,LambdaLR_get_linear_schedule_with_warmup_lr_lambda— def at line 101; referencesfloat,maxget_linear_schedule_with_warmup— def at line 107; referencespartial,LambdaLR_get_cosine_schedule_with_warmup_lr_lambda— def at line 134; referencesfloat,max,cosget_cosine_schedule_with_warmup— def at line 143; referencespartial,LambdaLR_get_cosine_with_hard_restarts_schedule_with_warmup_lr_lambda— def at line 177; referencesfloat,max,cosget_cosine_with_hard_restarts_schedule_with_warmup— def at line 188; referencespartial,LambdaLR_get_polynomial_decay_schedule_with_warmup_lr_lambda— def at line 221; referencesfloat,maxget_polynomial_decay_schedule_with_warmup— def at line 242; referencesnot,ValueError,partial,LambdaLR_get_inverse_sqrt_schedule_lr_lambda— def at line 288; referencesfloat,max,sqrtget_inverse_sqrt_schedule— def at line 296; referencespartial,LambdaLR_get_cosine_schedule_with_warmup_lr_lambda— def at line 326; referencesfloat,max,cosget_cosine_with_min_lr_schedule_with_warmup— def at line 337; referencesValueError,partial,LambdaLR_get_cosine_with_min_lr_schedule_with_warmup_lr_rate_lambda— def at line 389; referencesfloat,max,cos,ValueError,partial,LambdaLR,sqrt,warn,append,len,pop,copyget_cosine_with_min_lr_schedule_with_warmup_lr_rate— def at line 414; referencesValueError,partial,LambdaLR,float,max,cos,sqrt,warn,append,len,pop,copy_get_wsd_scheduler_lambda— def at line 470; referencesfloat,max,cos,sqrt,ValueError,warn,partial,LambdaLR,append,len,pop,copyget_wsd_schedule— def at line 508; referencesValueError,warn,partial,LambdaLR,append,len,pop,copyStreamingAverage— class at line 581; referencesappend,len,pop,copy__init__— def at line 592; referencesappend,len,pop,copystreamavg— def at line 597; referencesappend,len,pop,copystate_dict— def at line 608; referencescopyload_state_dict— def at line 615; referencesget,copyGreedyLR— class at line 621; referencesValueError,isinstance,TypeError,len,list,copy,float,StreamingAverage,streamavg,is_better,_increase_lr,_reduce_lr,enumerate,max,print,_reset,min,state_dict,get,load_state_dict__init__— def at line 678; referencesValueError,isinstance,TypeError,len,list,copy,float,StreamingAverage,streamavg,is_better,_increase_lr,_reduce_lr,enumerate,max,print,_reset,minstep— def at line 749; referencesfloat,streamavg,is_better,_increase_lr,_reduce_lr,enumerate,max,print,_reset,min,StreamingAverageis_better— def at line 795; referencesenumerate,float,max,print,_reset,min,StreamingAverage_reduce_lr— def at line 807; referencesenumerate,float,max,print,_reset,min,StreamingAverage_increase_lr— def at line 826; referencesenumerate,float,min,print,StreamingAverage_reset— def at line 838; referencesenumerate,float,StreamingAverage,printget_last_lr— def at line 855state_dict— def at line 859load_state_dict— def at line 892; referencesget,StreamingAverage,zip,GreedyLRget_greedy_schedule— def at line 928; referencesGreedyLRget_scheduler— def at line 960; referencesSchedulerType,isinstancescheduler_hook— def at line 1003; referencesstep,register_post_accumulate_grad_hook,LayerWiseDummyScheduler,schedule_funcAdafactor— class at line 1057; referencesValueError__init__— def at line 1142; referencesValueError_get_lr— def at line 1174; referencesmin,sqrt,max,len,norm,numel,mean,rsqrt_,unsqueeze,rsqrt,mul,no_grad,closure_get_options— def at line 1185; referenceslen,norm,numel,mean,rsqrt_,unsqueeze,rsqrt,mul,no_grad,closure_rms— def at line 1191; referencesnorm,numel,mean,rsqrt_,unsqueeze,rsqrt,mul,no_grad,closure_approx_sq_grad— def at line 1195; referencesmean,rsqrt_,unsqueeze,rsqrt,mul,no_grad,closurestep— def at line 1203; referencesclosureAdafactorSchedule— class at line 1297; referencessuper,__init__,_get_lr,len__init__— def at line 1305; referencessuperlr_lambda— def at line 1306; referencessuper,__init__get_lr— def at line 1315; references_get_lr,lenget_adafactor_schedule— def at line 1327; referencesAdafactorSchedule
File-level call references
get_logger— line 32LambdaLR— line 53ReduceLROnPlateau— line 71float— line 76max— line 76partial— line 97cos— line 140not— line 274ValueError— line 275sqrt— line 292warn— line 557append— line 599len— line 602pop— line 603copy— line 611get— line 616isinstance— line 698TypeError— line 699list— line 722StreamingAverage— line 747streamavg— line 761is_better— line 776_increase_lr— line 785_reduce_lr— line 789enumerate— line 809print— line 816_reset— line 824min— line 829state_dict— line 888load_state_dict— line 921zip— line 924GreedyLR— line 941SchedulerType— line 985get_scheduler— line 995step— line 1006register_post_accumulate_grad_hook— line 1010LayerWiseDummyScheduler— line 1012schedule_func— line 1015super— line 1171__init__— line 1171norm— line 1192numel— line 1192mean— line 1198rsqrt_— line 1198unsqueeze— line 1198rsqrt— line 1199mul— line 1200no_grad— line 1202closure— line 1213RuntimeError— line 1223_get_options— line 1228zeros_like— line 1235zeros— line 1237_rms— line 1257_get_lr— line 1258pow— line 1260mul_— line 1266add_— line 1266_approx_sq_grad— line 1270div_— line 1278clamp_— line 1278copy_— line 1292AdafactorSchedule— line 1342
Source: src/transformers/optimization.py