`ix NdZddlZddlZddlmZddlmZmZddl Z ddl m Z ddl m Z ddl mZdd lmZdd lmZmZmZdd lmZmZmZmZd d lmZejeZeedGddeZeedGddeZ eedGddeZ!dZ"dZ#Gdde j$Z%Gdde j$Z&Gdde j$Z'dAd"e j(d#e)d$e*d%e j(fd&Z+Gd'd(e j$Z,Gd)d*e j$Z-Gd+d,e j$Z.Gd-d.e j$Z/Gd/d0e j$Z0Gd1d2e j$Z1Gd3d4e j$Z2Gd5d6eZ3Gd7d8e j$Z4eGd9d:eZ5eGd;dd?e5Z7gd@Z8dS)BzPyTorch Donut Swin Transformer model. This implementation is identical to a regular Swin Transformer, without final layer norm on top of the final hidden states.N) dataclass)OptionalUnion)nn)ACT2FN)GradientCheckpointingLayer)PreTrainedModel) find_pruneable_heads_and_indicesmeshgridprune_linear_layer) ModelOutputauto_docstringlogging torch_int)DonutSwinConfigzS DonutSwin encoder's outputs, with potential hidden states and attentions. ) custom_introceZdZUdZdZeejed<dZ ee ejdfed<dZ ee ejdfed<dZ ee ejdfed<dS)DonutSwinEncoderOutputa reshaped_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each stage) of shape `(batch_size, hidden_size, height, width)`. Hidden-states of the model at the output of each layer plus the initial embedding outputs reshaped to include the spatial dimensions. Nlast_hidden_state. hidden_states attentionsreshaped_hidden_states) __name__ __module__ __qualname____doc__rrtorch FloatTensor__annotations__rtuplerr/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/donut/modeling_donut_swin.pyrr's6:x 12999=AM8E%"3S"89:AAA:>Ju0#567>>>FJHU5+