Dev gzf exp (#1664)

zhifu gao

2024-04-26 8fdc372c81ac6b0913353e3a8096593f67f31232

 funasr/datasets/sense_voice_datasets/datasets.py

@@ -1,3 +1,5 @@
import logging

import torch
import random

@@ -46,6 +48,8 @@
        self.float_pad_value = float_pad_value
        self.sos = kwargs.get("sos", "<|startoftranscript|>")
        self.eos = kwargs.get("eos", "<|endoftext|>")
        self.batch_size = kwargs.get("batch_size")
        self.batch_type = kwargs.get("batch_type")

    def get_source_len(self, index):
        item = self.index_ds[index]
@@ -95,8 +99,9 @@
        target_mask = (
            [0] * (prompt_ids_len) + [1] * (target_ids_len) + [1]
        )  # [sos, task, lid, text, eos]: [0, 0, 1, 1, 1]
        target_mask_lengths = len(target_mask)
        target_mask = torch.tensor(target_mask, dtype=torch.float32)

        target_mask_lengths = torch.tensor([target_mask_lengths], dtype=torch.int32)
        return {
            "speech": speech[0, :, :],
            "speech_lengths": speech_lengths,
@@ -124,4 +129,28 @@
                outputs[key] = torch.nn.utils.rnn.pad_sequence(
                    data_list, batch_first=True, padding_value=pad_value
                )

        if self.batch_type != "example":
            for i in range(3):
                outputs = self._filter_badcase(outputs)

        return outputs

    def _filter_badcase(self, outputs, i=0):
        b, t, _ = outputs["speech"].shape
        if b * t > self.batch_size:
            beg = torch.randint(0, 2, ()).item()
            logging.info(
                f"Warning, b * t: {b * t} > {self.batch_size}, drop half data {i}th, beg:{beg}"
            )
            for key, data_list in outputs.items():
                outputs[key] = outputs[key][beg : beg + b : 2]

            speech_lengths_max = outputs["speech_lengths_max"].max().item()
            outputs["speech"] = outputs["speech"][:, :speech_lengths_max, :]
            text_lengths_max = outputs["text_lengths"].max().item()
            outputs["text"] = outputs["text"][:, :text_lengths_max]
            target_mask_lengths_max = outputs["target_mask_lengths_max"].max().item()
            outputs["target_mask"] = outputs["target_mask"][:, :target_mask_lengths_max]

        return outputs