^jQ)$dZddlZddlmZddlZddlmZddlmZm Z m Z m Z m Z ddl mZdZefd Zd Zd Zd ZGd dZgdfdZgddfdZgdfdZgdfdZddej2j4dej2j4fdZy)zM Model / state_dict utils Hacked together by / Copyright 2020 Ross Wightman Ndeepcopy)FrozenBatchNorm2d)BatchNormAct2dSyncBatchNormActFrozenBatchNormAct2dfreeze_batch_norm_2dunfreeze_batch_norm_2d)ModelEmact|trt|jSt |drt|j St |drt|j S|S)Nmodule _orig_mod) isinstancer unwrap_modelemahasattrrr)models [/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/timm/utils/model.pyrrsR%"EII&& 5( # - - UK (0 0Lc.||jSN) state_dict)r unwrap_fns rget_state_dictrs U  & & ((rcrtj|jgddzjS)zA calculate average channel square mean of output activations raxisr)torchmeanitemrinputoutputs ravg_sq_ch_meanr( s, ::fkkyk1Q6 7 < < >>rcltj|jgdjSz> calculate average channel variance of output activations rr r"r#varr$r%s r avg_ch_varr-&' ::fjjij0 1 6 6 88rcltj|jgdjSr*r+r%s ravg_ch_var_residualr0,r.rc"eZdZdZdZdZdZy)ActivationStatsHookaIterates through each of `model`'s modules and matches modules using unix pattern matching based on `hook_fn_locs` and registers `hook_fn` to the module if there is a match. Arguments: model (nn.Module): model from which we will extract the activation stats hook_fn_locs (List[str]): List of `hook_fn` locations based on Unix type string matching with the name of model's modules. hook_fns (List[Callable]): List of hook functions to be registered at every module in `layer_names`. Inspiration from https://docs.fast.ai/callback.hook.html. Refer to https://gist.github.com/amaarora/6e56942fcb46e67ba203f3009b30d950 for an example on how to plot Signal Propagation Plots using `ActivationStatsHook`. c||_||_||_t|t|k7r t dt d|D|_t||D]\}}|j||y)Nz_Please provide `hook_fns` for each `hook_fn_locs`, their lengths are different.c38K|]}|jgfywr)__name__).0hook_fns r z/ActivationStatsHook.__init__..KsIW7++R0Is) r hook_fn_locshook_fnslen ValueErrordictstatszip register_hook)selfrr9r: hook_fn_locr7s r__init__zActivationStatsHook.__init__Dsx (  | H -./ /III $' h$? 5 K   {G 4 5rcfd}|S)Ncj|||}jjj|yr)r>r5append)rr&r'outr7rAs rappend_activation_statszAActivationStatsHook._create_hook..append_activation_statsPs.&%0C JJw'' ( / / 4r)rAr7rHs`` r _create_hookz ActivationStatsHook._create_hookOs 5'&rc|jjD]<\}}tj||s|j|j |>yr)r named_modulesfnmatchregister_forward_hookrJ)rArBr7namers rr@z!ActivationStatsHook.register_hookVsN JJ446 ELD&??45  ( ():):7)C D ErN)r5 __module__ __qualname____doc__rCrJr@rIrrr2r22s" 5'Err2)rrTcttjdd|}t|||}||}|jS)aExtract average square channel mean and variance of activations during forward pass to plot Signal Propagation Plots (SPP). Paper: https://arxiv.org/abs/2101.08692 Example Usage: https://gist.github.com/amaarora/6e56942fcb46e67ba203f3009b30d950 gg?)r9r:)r"normalr2r>)rr9r: input_shapexhook_s rextract_spp_statsr[]s7  R[)A u<( SD aA ::rTfreezec|dvsJdt|tjjjj tjjjj ttfr tdt|tr|g}|}|Dcgc]}|j|}}t|s"tt|j\}}t||D]\}}|j!D]}|dk(rdnd|_|s,d}|dk(rt%|} t|tjjjj tjjjj ttfs|||| t'|} t|t(t*fs|||| ycc}w) a9 Freeze or unfreeze parameters of the specified modules and those of all their hierarchical descendants. This is done in place. Args: root_module (nn.Module, optional): Root module relative to which the `submodules` are referenced. submodules (list[str]): List of modules for which the parameters will be (un)frozen. They are to be provided as named modules relative to the root module (accessible via `root_module.named_modules()`). An empty list means that the whole root module will be (un)frozen. Defaults to [] include_bn_running_stats (bool): Whether to also (un)freeze the running statistics of batch norm 2d layers. Defaults to `True`. mode (bool): Whether to freeze ("freeze") or unfreeze ("unfreeze"). Defaults to `"freeze"`. )r\unfreezez,`mode` must be one of "freeze" or "unfreeze"zYou have provided a batch norm layer as the `root module`. Please use `timm.utils.model.freeze_batch_norm_2d` or `timm.utils.model.unfreeze_batch_norm_2d` instead.r\FTc|jdd}t|dkDr(|j|dj|d|y|j||y)N.r r)rsplitr; get_submodule add_module)rrO submodulesplits r_add_submodulez(_freeze_unfreeze.._add_submodulesP C+u:>((q2==eAh R%%dI6rN)rr"nnmodules batchnorm BatchNorm2d SyncBatchNormrrAssertionErrorstrrbr;listr?named_children parameters requires_gradr r rr) root_module submodulesinclude_bn_running_statsmoderLmnprfress r_freeze_unfreezerzos ) )Y+YY )+ HH   & & 2 2 HH   & & 4 4     lm m*c" \ M8BC1+++A.CJC z?$(k.H.H.J)K$L! zM:.81 BA'+x'7eTAO B # 7x*1-a((22>>((22@@&( " #;37-Q/a"35I!JK";37?8 Ds Gc"t|||dy)a Freeze parameters of the specified modules and those of all their hierarchical descendants. This is done in place. Args: root_module (nn.Module): Root module relative to which `submodules` are referenced. submodules (list[str]): List of modules for which the parameters will be frozen. They are to be provided as named modules relative to the root module (accessible via `root_module.named_modules()`). An empty list means that the whole root module will be frozen. Defaults to `[]`. include_bn_running_stats (bool): Whether to also freeze the running statistics of `BatchNorm2d` and `SyncBatchNorm` layers. These will be converted to `FrozenBatchNorm2d` in place. Hint: During fine tuning, it's good practice to freeze batch norm stats. And note that these are different to the affine parameters which are just normal PyTorch parameters. Defaults to `True`. Hint: If you want to freeze batch norm ONLY, use `timm.utils.model.freeze_batch_norm_2d`. Examples:: >>> model = timm.create_model('resnet18') >>> # Freeze up to and including layer2 >>> submodules = [n for n, _ in model.named_children()] >>> print(submodules) ['conv1', 'bn1', 'act1', 'maxpool', 'layer1', 'layer2', 'layer3', 'layer4', 'global_pool', 'fc'] >>> freeze(model, submodules[:submodules.index('layer2') + 1]) >>> # Check for yourself that it works as expected >>> print(model.layer2[0].conv1.weight.requires_grad) False >>> print(model.layer3[0].conv1.weight.requires_grad) True >>> # Unfreeze >>> unfreeze(model) r\rtruNrzrrrsrts rr\r\s@[*G_fnorc"t|||dy)a Unfreeze parameters of the specified modules and those of all their hierarchical descendants. This is done in place. Args: root_module (nn.Module): Root module relative to which `submodules` are referenced. submodules (list[str]): List of submodules for which the parameters will be (un)frozen. They are to be provided as named modules relative to the root module (accessible via `root_module.named_modules()`). An empty list means that the whole root module will be unfrozen. Defaults to `[]`. include_bn_running_stats (bool): Whether to also unfreeze the running statistics of `FrozenBatchNorm2d` layers. These will be converted to `BatchNorm2d` in place. Defaults to `True`. See example in docstring for `freeze`. r^r|Nr}r~s rr^r^s[*G_fpqrrreturnc<|s t|}fd||S)Nc|jD]n\}}t|drt|||jn9t|dr|j nt|dr|j |py)Nfusereparameterizeswitch_to_deploy)rorsetattrrrr)rv child_namechild_fuses rrz#reparameterize_model.._fusesq!"!1!1!3  Juf%:uzz|4 01$$& 23&&( %L rr)rinplacers @rreparameterize_modelrs"  %L Lr)F)rRrMcopyrr"torchvision.ops.miscr timm.layersrrrr r model_emar rrr(r-r0r2r[rzr\r^rgModulerrIrrrs 211 %1)? 9 9 (E(E^% $.0$U]C8L$& pF&($r"588??r