`iBdZddlZddlZddlmZddlmZmZmZm Z ddl Z ddl m cm Zddl m Z ddlmZddlmZdd lmZmZmZdd lmZmZdd lmZmZmZmZdd lm Z m!Z!m"Z"m#Z#m$Z$m%Z%d dl&m'Z'm(Z(m)Z)e$j*e+Z,dZ-dZ.dZ/dcdZ0de j1de j1fdZ2ee!dGdde Z3ee!dGdde Z4ee!Gdde Z5Gdd e j6Z7Gd!d"e j6Z8Gd#d$e j6Z9Gd%d&e j6Z:Gd'd(e j6Z;Gd)d*e j6Z<Gd+d,e j6Z=Gd-d.e j6Z>Gd/d0e j6Z?Gd1d2eZ@Gd3d4e j6ZAGd5d6e j6ZBGd7d8e j6ZCGd9d:e j6ZD ddde j1d?e j1d@ee j1dAeEdBeEdCee j1fdDZFGdEdFe j6ZGGdGdHe j6ZHGdIdJe j6ZIGdKdLe j6ZJGdMdNe j6ZKGdOdPeZLGdQdRe j6ZMGdSdTe j6ZNe!GdUdVeZOGdWdXeOZPe!dYGdZd[eOZQe!Gd\d]eOZRe!Gd^d_eOZSe!Gd`daeOZTgdbZUdS)ezPyTorch CLAP model.N) dataclass)AnyCallableOptionalUnion)nn)ACT2FN)GradientCheckpointingLayer)BaseModelOutputBaseModelOutputWithPooling,BaseModelOutputWithPoolingAndCrossAttentions)ALL_ATTENTION_FUNCTIONSPreTrainedModel)apply_chunking_to_forward find_pruneable_heads_and_indicesmeshgridprune_linear_layer) ModelOutputauto_docstringcan_return_tuplefilter_out_non_signature_kwargslogging torch_int)ClapAudioConfig ClapConfigClapTextConfigc|j\}}}|dddddddfdd|d}||||z|}|S)ae Interpolate data in time domain. This is used to compensate the resolution reduction in downsampling of a CNN. Args: hidden_states (`torch.FloatTensor` of shape (batch_size, time_length, classes_num)): Input hidden states ratio (`int`): The ratio of the length of the output to the length of the input. Nr)shaperepeatreshape) hidden_statesratio batch_size time_length classes_num upsampleds z/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/clap/modeling_clap.py interpolater*+sg.;-@*ZkaaaD!!!m,33Aq%CCI!!*kE.A;OOI c|j\}}}}||||z|||z||}|ddddddd|||}|S)aR Returns the resized hidden states. The output shape should be `(batch_size * num_windows, window_size, window_size, num_channels)` Args: hidden_states (`torch.FloatTensor` of shape `(batch_size, height, width, num_channels)`): Input hidden states window_size (`int`): Window size rrr r viewpermute contiguous)r# window_sizer%heightwidth num_channelswindowss r)window_partitionr:<s/<.A+J|!&&Fk);8Lk[gM##Aq!Q155@@BBGGKYdfrssG Nr+c|jd}|d||z||z|||}|ddddddd|||}|S)a Merges windows to produce higher resolution features. Args: windows (`torch.FloatTensor` of shape `(num_windows * batch_size, window_size, window_size, num_channels)`): Input windows window_size (`int`): Window size height (`int`): Height of the resized audio width (`int`): Width of the resized audio r0rrr r-r.r/r1)r9r5r6r7r8s r)window_reverser<Qsx=$Lll2v4e{6JKYdfrssGooaAq!Q//::<<AA"feUabbG Nr+c||}tj|d||z|z}||zS)a Replace non-padding symbols with their position numbers. Position numbers begin at padding_idx+1. Padding symbols are ignored. This is modified from fairseq's `utils.make_positions`. Args: x: torch.Tensor x: Returns: torch.Tensor rdim)neinttorchcumsumtype_aslong) input_ids padding_idxpast_key_values_lengthmaskincremental_indicess r)"create_position_ids_from_input_idsrKesg << $ $ ( ( * *D <!444<