export model

游雁

2023-02-14 1d4ab65c8bfebaecbcb0eec0064bae9a321cad75

 funasr/models/predictor/cif.py

@@ -68,7 +68,8 @@
            mask_2 = torch.cat([ones_t, mask], dim=1)

            mask = mask_2 - mask_1

            tail_threshold = mask * tail_threshold

            alphas = torch.cat([alphas, tail_threshold], dim=1)

            alphas = torch.cat([alphas, zeros_t], dim=1)

            alphas = torch.add(alphas, tail_threshold)

        else:

            tail_threshold = torch.tensor([tail_threshold], dtype=alphas.dtype).to(alphas.device)

            tail_threshold = torch.reshape(tail_threshold, (1, 1))

@@ -208,7 +209,8 @@
            mask_2 = torch.cat([ones_t, mask], dim=1)

            mask = mask_2 - mask_1

            tail_threshold = mask * tail_threshold

            alphas = torch.cat([alphas, tail_threshold], dim=1)

            alphas = torch.cat([alphas, zeros_t], dim=1)

            alphas = torch.add(alphas, tail_threshold)

        else:

            tail_threshold = torch.tensor([tail_threshold], dtype=alphas.dtype).to(alphas.device)

            tail_threshold = torch.reshape(tail_threshold, (1, 1))

@@ -542,9 +544,8 @@
            token_num_int = torch.max(token_num).type(torch.int32).item()

            acoustic_embeds = acoustic_embeds[:, :token_num_int, :]

        return acoustic_embeds, token_num, alphas, cif_peak, token_num2

    
    def get_upsample_timestamp(self, hidden, target_label=None, mask=None, ignore_id=-1, mask_chunk_predictor=None,

                target_label_length=None, token_num=None):



    def get_upsample_timestamp(self, hidden, mask=None, token_num=None):

        h = hidden

        b = hidden.shape[0]

        context = h.transpose(1, 2)

@@ -596,7 +597,8 @@
            mask_2 = torch.cat([ones_t, mask], dim=1)

            mask = mask_2 - mask_1

            tail_threshold = mask * tail_threshold

            alphas = torch.cat([alphas, tail_threshold], dim=1)

            alphas = torch.cat([alphas, zeros_t], dim=1)

            alphas = torch.add(alphas, tail_threshold)

        else:

            tail_threshold = torch.tensor([tail_threshold], dtype=alphas.dtype).to(alphas.device)

            tail_threshold = torch.reshape(tail_threshold, (1, 1))

@@ -654,4 +656,4 @@


        predictor_alignments = index_div_bool_zeros_count_tile_out

        predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype)

        return predictor_alignments.detach(), predictor_alignments_length.detach()
        return predictor_alignments.detach(), predictor_alignments_length.detach()