jddlZddlmZddlZddlZddlmZddlmZgdZej j j Z dej dej fdZdej dej fd Z dd ejd edzd edzd eedzdeddf dZ ddedededededeejf dZd edeej deejzdejdeeeej ff dZd ejdedej dedef dZdS)N)Any)nn)download_from_url)load_pretrained_paramsconv_sequence_ptset_device_and_dtypeexport_model_to_onnx _copy_tensor_bf16_to_float32_CompiledModulexreturncN|SN)clonedetachr s ]/var/www/html/Carbon-Document/venv/lib/python3.11/site-packages/doctr/models/utils/pytorch.pyr r s 7799    cX|jtjkr|n|Sr)dtypetorchbfloat16floatrs rr r !s"5>11177999q8rmodel path_or_url hash_prefix ignore_keyskwargsc |tjddStj|rt |f|dd|n|}t j|d}tdkrpD]}||vr||| |d \}} tfd || zDrtd d | d |zdS| |dS)a"Load a set of parameters onto a model >>> from doctr.models import load_pretrained_params >>> load_pretrained_params(model, "https://yoursource.com/yourcheckpoint-yourhash.pt") Args: model: the PyTorch model to be loaded path_or_url: the path or URL to the model parameters (checkpoint) hash_prefix: first characters of SHA256 expected hash ignore_keys: list of weights to be ignored from the state_dict **kwargs: additional arguments to be passed to `doctr.utils.data.download_from_url` Nzz)load_pretrained_params..Ls(LLq #LLLLLLrz5Unable to load state_dict, due to non-matching keys. zUnexpected keys: z Missing keys: ) loggingwarning validatorsurlrrloadlenpopload_state_dictany ValueError) rrrrr archive_path state_dictkey missing_keysunexpected_keyss ` rrr&sT&VWWW >+ & & +`;X``Y_``` Lu===J3{#3#3a#7#7 $ $Cj  s###(-(=(=jQV(=(W(W% o LLLL\O-KLLL L L HUoUU|UUV    j)))))rF in_channels out_channelsrelubnc |d| |d<tj||fi|g}|r'|tj||r(|tjd|S)aBuilds a convolutional-based layer sequence >>> from torch.nn import Sequential >>> from doctr.models import conv_sequence >>> module = Sequential(conv_sequence(3, 32, True, True, kernel_size=3)) Args: in_channels: number of input channels out_channels: number of output channels relu: whether ReLU should be used bn: should a batch normalization layer be added **kwargs: additional arguments to be passed to the convolutional layer Returns: list of layers biasT)inplace)getrConv2dappend BatchNorm2dReLU)r:r;r<r=rconv_seqs rrrVs0ZZB//F6N!#; !O!O!O!O PH 6|44555 /---... OrbatchesdevicercR|fd|DfS)aWSet the device and dtype of a model and its batches >>> import torch >>> from torch import nn >>> from doctr.models.utils import set_device_and_dtype >>> model = nn.Sequential(nn.Linear(8, 8), nn.ReLU(), nn.Linear(8, 4)) >>> batches = [torch.rand(8) for _ in range(2)] >>> model, batches = set_device_and_dtype(model, batches, device="cuda", dtype=torch.float16) Args: model: the model to be set batches: the batches to be set device: the device to be used dtype: the dtype to be used Returns: the model and batches set rHrc>g|]}|S)rJto)r(batchrHrs r z(set_device_and_dtype..s+1l1l1l[`%((&PU(2V2V1l1l1lrrL)rrGrHrs ``rrr{s:* 8868 / /1l1l1l1l1ldk1l1l1l llr model_name dummy_inputc tjj|||dfdgdgddiddiddddd |tjd |d|dS) aJExport model to ONNX format. >>> import torch >>> from doctr.models.classification import resnet18 >>> from doctr.models.utils import export_model_to_onnx >>> model = resnet18(pretrained=True) >>> export_model_to_onnx(model, "my_model", dummy_input=torch.randn(1, 3, 32, 32)) Args: model: the PyTorch model to be exported model_name: the name for the exported model dummy_input: the dummy input to the model kwargs: additional arguments to be passed to torch.onnx.export Returns: the path to the exported model z.onnxinputlogitsr batch_size)rSrTTF) input_names output_names dynamic_axes export_paramsdynamoverbosezModel exported to )ronnxexportr+info)rrPrQrs rr r s$ J   IZ !<0Q rlsE  ......   -*:ELU\9 99999#"$( -*-* 9-*t-*t-*cT! -*  -*  -*-*-*-*f """" " "  "  ")_ """"Jm mel+m585<5GmPUP[m 3U\" "#mmmm0   s  ad il      r