From ae49b2a8e1bc676e6014d8a12ebeec947b655e3e Mon Sep 17 00:00:00 2001
From: 莫拉古 <61447879+yechaoying@users.noreply.github.com>
Date: 星期五, 29 十一月 2024 09:55:43 +0800
Subject: [PATCH] 变量名写错了 (#2249)
---
funasr/models/paraformer/cif_predictor.py | 644 ++++++++++++++++++++++++++++++++++++++++-----------------
1 files changed, 447 insertions(+), 197 deletions(-)
diff --git a/funasr/models/paraformer/cif_predictor.py b/funasr/models/paraformer/cif_predictor.py
index 383d9ca..24145cd 100644
--- a/funasr/models/paraformer/cif_predictor.py
+++ b/funasr/models/paraformer/cif_predictor.py
@@ -1,67 +1,91 @@
+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+# MIT License (https://opensource.org/licenses/MIT)
+
import torch
-from torch import nn
-from torch import Tensor
import logging
import numpy as np
-from funasr.train_utils.device_funcs import to_device
-from funasr.models.transformer.utils.nets_utils import make_pad_mask
-from funasr.models.scama.utils import sequence_mask
-from typing import Optional, Tuple
from funasr.register import tables
+from funasr.train_utils.device_funcs import to_device
+from funasr.models.transformer.utils.nets_utils import make_pad_mask
+from torch.cuda.amp import autocast
+
@tables.register("predictor_classes", "CifPredictor")
-class CifPredictor(nn.Module):
- def __init__(self, idim, l_order, r_order, threshold=1.0, dropout=0.1, smooth_factor=1.0, noise_threshold=0, tail_threshold=0.45):
+class CifPredictor(torch.nn.Module):
+ def __init__(
+ self,
+ idim,
+ l_order,
+ r_order,
+ threshold=1.0,
+ dropout=0.1,
+ smooth_factor=1.0,
+ noise_threshold=0,
+ tail_threshold=0.45,
+ ):
super().__init__()
- self.pad = nn.ConstantPad1d((l_order, r_order), 0)
- self.cif_conv1d = nn.Conv1d(idim, idim, l_order + r_order + 1, groups=idim)
- self.cif_output = nn.Linear(idim, 1)
+ self.pad = torch.nn.ConstantPad1d((l_order, r_order), 0)
+ self.cif_conv1d = torch.nn.Conv1d(idim, idim, l_order + r_order + 1, groups=idim)
+ self.cif_output = torch.nn.Linear(idim, 1)
self.dropout = torch.nn.Dropout(p=dropout)
self.threshold = threshold
self.smooth_factor = smooth_factor
self.noise_threshold = noise_threshold
self.tail_threshold = tail_threshold
- def forward(self, hidden, target_label=None, mask=None, ignore_id=-1, mask_chunk_predictor=None,
- target_label_length=None):
- h = hidden
- context = h.transpose(1, 2)
- queries = self.pad(context)
- memory = self.cif_conv1d(queries)
- output = memory + context
- output = self.dropout(output)
- output = output.transpose(1, 2)
- output = torch.relu(output)
- output = self.cif_output(output)
- alphas = torch.sigmoid(output)
- alphas = torch.nn.functional.relu(alphas * self.smooth_factor - self.noise_threshold)
- if mask is not None:
- mask = mask.transpose(-1, -2).float()
- alphas = alphas * mask
- if mask_chunk_predictor is not None:
- alphas = alphas * mask_chunk_predictor
- alphas = alphas.squeeze(-1)
- mask = mask.squeeze(-1)
- if target_label_length is not None:
- target_length = target_label_length
- elif target_label is not None:
- target_length = (target_label != ignore_id).float().sum(-1)
- else:
- target_length = None
- token_num = alphas.sum(-1)
- if target_length is not None:
- alphas *= (target_length / token_num)[:, None].repeat(1, alphas.size(1))
- elif self.tail_threshold > 0.0:
- hidden, alphas, token_num = self.tail_process_fn(hidden, alphas, token_num, mask=mask)
-
- acoustic_embeds, cif_peak = cif(hidden, alphas, self.threshold)
-
- if target_length is None and self.tail_threshold > 0.0:
- token_num_int = torch.max(token_num).type(torch.int32).item()
- acoustic_embeds = acoustic_embeds[:, :token_num_int, :]
-
+ def forward(
+ self,
+ hidden,
+ target_label=None,
+ mask=None,
+ ignore_id=-1,
+ mask_chunk_predictor=None,
+ target_label_length=None,
+ ):
+
+ with autocast(False):
+ h = hidden
+ context = h.transpose(1, 2)
+ queries = self.pad(context)
+ memory = self.cif_conv1d(queries)
+ output = memory + context
+ output = self.dropout(output)
+ output = output.transpose(1, 2)
+ output = torch.relu(output)
+ output = self.cif_output(output)
+ alphas = torch.sigmoid(output)
+ alphas = torch.nn.functional.relu(alphas * self.smooth_factor - self.noise_threshold)
+ if mask is not None:
+ mask = mask.transpose(-1, -2).float()
+ alphas = alphas * mask
+ if mask_chunk_predictor is not None:
+ alphas = alphas * mask_chunk_predictor
+ alphas = alphas.squeeze(-1)
+ mask = mask.squeeze(-1)
+ if target_label_length is not None:
+ target_length = target_label_length
+ elif target_label is not None:
+ target_length = (target_label != ignore_id).float().sum(-1)
+ else:
+ target_length = None
+ token_num = alphas.sum(-1)
+ if target_length is not None:
+ alphas *= (target_length / token_num)[:, None].repeat(1, alphas.size(1))
+ elif self.tail_threshold > 0.0:
+ hidden, alphas, token_num = self.tail_process_fn(
+ hidden, alphas, token_num, mask=mask
+ )
+
+ acoustic_embeds, cif_peak = cif(hidden, alphas, self.threshold)
+
+ if target_length is None and self.tail_threshold > 0.0:
+ 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
def tail_process_fn(self, hidden, alphas, token_num=None, mask=None):
@@ -87,10 +111,9 @@
return hidden, alphas, token_num_floor
-
- def gen_frame_alignments(self,
- alphas: torch.Tensor = None,
- encoder_sequence_length: torch.Tensor = None):
+ def gen_frame_alignments(
+ self, alphas: torch.Tensor = None, encoder_sequence_length: torch.Tensor = None
+ ):
batch_size, maximum_length = alphas.size()
int_type = torch.int32
@@ -113,11 +136,15 @@
index_div = torch.floor(torch.true_divide(alphas_cumsum, index)).type(int_type)
index_div_bool_zeros = index_div.eq(0)
index_div_bool_zeros_count = torch.sum(index_div_bool_zeros, dim=-1) + 1
- index_div_bool_zeros_count = torch.clamp(index_div_bool_zeros_count, 0, encoder_sequence_length.max())
+ index_div_bool_zeros_count = torch.clamp(
+ index_div_bool_zeros_count, 0, encoder_sequence_length.max()
+ )
token_num_mask = (~make_pad_mask(token_num, maxlen=max_token_num)).to(token_num.device)
index_div_bool_zeros_count *= token_num_mask
- index_div_bool_zeros_count_tile = index_div_bool_zeros_count[:, :, None].repeat(1, 1, maximum_length)
+ index_div_bool_zeros_count_tile = index_div_bool_zeros_count[:, :, None].repeat(
+ 1, 1, maximum_length
+ )
ones = torch.ones_like(index_div_bool_zeros_count_tile)
zeros = torch.zeros_like(index_div_bool_zeros_count_tile)
ones = torch.cumsum(ones, dim=2)
@@ -128,34 +155,41 @@
index_div_bool_zeros_count_tile = 1 - index_div_bool_zeros_count_tile_bool.type(int_type)
index_div_bool_zeros_count_tile_out = torch.sum(index_div_bool_zeros_count_tile, dim=1)
index_div_bool_zeros_count_tile_out = index_div_bool_zeros_count_tile_out.type(int_type)
- predictor_mask = (~make_pad_mask(encoder_sequence_length, maxlen=encoder_sequence_length.max())).type(
- int_type).to(encoder_sequence_length.device)
+ predictor_mask = (
+ (~make_pad_mask(encoder_sequence_length, maxlen=encoder_sequence_length.max()))
+ .type(int_type)
+ .to(encoder_sequence_length.device)
+ )
index_div_bool_zeros_count_tile_out = index_div_bool_zeros_count_tile_out * predictor_mask
predictor_alignments = index_div_bool_zeros_count_tile_out
- predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype)
+ predictor_alignments_length = predictor_alignments.sum(-1).type(
+ encoder_sequence_length.dtype
+ )
return predictor_alignments.detach(), predictor_alignments_length.detach()
-@tables.register("predictor_classes", "CifPredictorV2")
-class CifPredictorV2(nn.Module):
- def __init__(self,
- idim,
- l_order,
- r_order,
- threshold=1.0,
- dropout=0.1,
- smooth_factor=1.0,
- noise_threshold=0,
- tail_threshold=0.0,
- tf2torch_tensor_name_prefix_torch="predictor",
- tf2torch_tensor_name_prefix_tf="seq2seq/cif",
- tail_mask=True,
- ):
- super(CifPredictorV2, self).__init__()
- self.pad = nn.ConstantPad1d((l_order, r_order), 0)
- self.cif_conv1d = nn.Conv1d(idim, idim, l_order + r_order + 1)
- self.cif_output = nn.Linear(idim, 1)
+@tables.register("predictor_classes", "CifPredictorV2")
+class CifPredictorV2(torch.nn.Module):
+ def __init__(
+ self,
+ idim,
+ l_order,
+ r_order,
+ threshold=1.0,
+ dropout=0.1,
+ smooth_factor=1.0,
+ noise_threshold=0,
+ tail_threshold=0.0,
+ tf2torch_tensor_name_prefix_torch="predictor",
+ tf2torch_tensor_name_prefix_tf="seq2seq/cif",
+ tail_mask=True,
+ ):
+ super().__init__()
+
+ self.pad = torch.nn.ConstantPad1d((l_order, r_order), 0)
+ self.cif_conv1d = torch.nn.Conv1d(idim, idim, l_order + r_order + 1)
+ self.cif_output = torch.nn.Linear(idim, 1)
self.dropout = torch.nn.Dropout(p=dropout)
self.threshold = threshold
self.smooth_factor = smooth_factor
@@ -165,47 +199,61 @@
self.tf2torch_tensor_name_prefix_tf = tf2torch_tensor_name_prefix_tf
self.tail_mask = tail_mask
- def forward(self, hidden, target_label=None, mask=None, ignore_id=-1, mask_chunk_predictor=None,
- target_label_length=None):
- h = hidden
- context = h.transpose(1, 2)
- queries = self.pad(context)
- output = torch.relu(self.cif_conv1d(queries))
- output = output.transpose(1, 2)
+ def forward(
+ self,
+ hidden,
+ target_label=None,
+ mask=None,
+ ignore_id=-1,
+ mask_chunk_predictor=None,
+ target_label_length=None,
+ ):
- output = self.cif_output(output)
- alphas = torch.sigmoid(output)
- alphas = torch.nn.functional.relu(alphas * self.smooth_factor - self.noise_threshold)
- if mask is not None:
- mask = mask.transpose(-1, -2).float()
- alphas = alphas * mask
- if mask_chunk_predictor is not None:
- alphas = alphas * mask_chunk_predictor
- alphas = alphas.squeeze(-1)
- mask = mask.squeeze(-1)
- if target_label_length is not None:
- target_length = target_label_length
- elif target_label is not None:
- target_length = (target_label != ignore_id).float().sum(-1)
- else:
- target_length = None
- token_num = alphas.sum(-1)
- if target_length is not None:
- alphas *= (target_length / token_num)[:, None].repeat(1, alphas.size(1))
- elif self.tail_threshold > 0.0:
- if self.tail_mask:
- hidden, alphas, token_num = self.tail_process_fn(hidden, alphas, token_num, mask=mask)
+ with autocast(False):
+ h = hidden
+ context = h.transpose(1, 2)
+ queries = self.pad(context)
+ output = torch.relu(self.cif_conv1d(queries))
+ output = output.transpose(1, 2)
+
+ output = self.cif_output(output)
+ alphas = torch.sigmoid(output)
+ alphas = torch.nn.functional.relu(alphas * self.smooth_factor - self.noise_threshold)
+ if mask is not None:
+ mask = mask.transpose(-1, -2).float()
+ alphas = alphas * mask
+ if mask_chunk_predictor is not None:
+ alphas = alphas * mask_chunk_predictor
+ alphas = alphas.squeeze(-1)
+ mask = mask.squeeze(-1)
+ if target_label_length is not None:
+ target_length = target_label_length.squeeze(-1)
+ elif target_label is not None:
+ target_length = (target_label != ignore_id).float().sum(-1)
else:
- hidden, alphas, token_num = self.tail_process_fn(hidden, alphas, token_num, mask=None)
+ target_length = None
+ token_num = alphas.sum(-1)
+ if target_length is not None:
+ alphas *= (target_length / token_num)[:, None].repeat(1, alphas.size(1))
+ elif self.tail_threshold > 0.0:
+ if self.tail_mask:
+ hidden, alphas, token_num = self.tail_process_fn(
+ hidden, alphas, token_num, mask=mask
+ )
+ else:
+ hidden, alphas, token_num = self.tail_process_fn(
+ hidden, alphas, token_num, mask=None
+ )
- acoustic_embeds, cif_peak = cif(hidden, alphas, self.threshold)
- if target_length is None and self.tail_threshold > 0.0:
- token_num_int = torch.max(token_num).type(torch.int32).item()
- acoustic_embeds = acoustic_embeds[:, :token_num_int, :]
+ acoustic_embeds, cif_peak = cif(hidden, alphas, self.threshold)
+ if target_length is None and self.tail_threshold > 0.0:
+ 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
- def forward_chunk(self, hidden, cache=None):
+ def forward_chunk(self, hidden, cache=None, **kwargs):
+ is_final = kwargs.get("is_final", False)
batch_size, len_time, hidden_size = hidden.shape
h = hidden
context = h.transpose(1, 2)
@@ -225,15 +273,15 @@
cache_hiddens = []
if cache is not None and "chunk_size" in cache:
- alphas[:, :cache["chunk_size"][0]] = 0.0
- if "is_final" in cache and not cache["is_final"]:
- alphas[:, sum(cache["chunk_size"][:2]):] = 0.0
+ alphas[:, : cache["chunk_size"][0]] = 0.0
+ if not is_final:
+ alphas[:, sum(cache["chunk_size"][:2]) :] = 0.0
if cache is not None and "cif_alphas" in cache and "cif_hidden" in cache:
cache["cif_hidden"] = to_device(cache["cif_hidden"], device=hidden.device)
cache["cif_alphas"] = to_device(cache["cif_alphas"], device=alphas.device)
hidden = torch.cat((cache["cif_hidden"], hidden), dim=1)
alphas = torch.cat((cache["cif_alphas"], alphas), dim=1)
- if cache is not None and "is_final" in cache and cache["is_final"]:
+ if cache is not None and is_final:
tail_hidden = torch.zeros((batch_size, 1, hidden_size), device=hidden.device)
tail_alphas = torch.tensor([[self.tail_threshold]], device=alphas.device)
tail_alphas = torch.tile(tail_alphas, (batch_size, 1))
@@ -277,10 +325,12 @@
max_token_len = max(token_length)
if max_token_len == 0:
- return hidden, torch.stack(token_length, 0)
+ return hidden, torch.stack(token_length, 0), None, None
list_ls = []
for b in range(batch_size):
- pad_frames = torch.zeros((max_token_len - token_length[b], hidden_size), device=alphas.device)
+ pad_frames = torch.zeros(
+ (max_token_len - token_length[b], hidden_size), device=alphas.device
+ )
if token_length[b] == 0:
list_ls.append(pad_frames)
else:
@@ -291,8 +341,7 @@
cache["cif_alphas"] = torch.unsqueeze(cache["cif_alphas"], axis=0)
cache["cif_hidden"] = torch.stack(cache_hiddens, axis=0)
cache["cif_hidden"] = torch.unsqueeze(cache["cif_hidden"], axis=0)
- return torch.stack(list_ls, 0), torch.stack(token_length, 0)
-
+ return torch.stack(list_ls, 0), torch.stack(token_length, 0), None, None
def tail_process_fn(self, hidden, alphas, token_num=None, mask=None):
b, t, d = hidden.size()
@@ -320,9 +369,9 @@
return hidden, alphas, token_num_floor
- def gen_frame_alignments(self,
- alphas: torch.Tensor = None,
- encoder_sequence_length: torch.Tensor = None):
+ def gen_frame_alignments(
+ self, alphas: torch.Tensor = None, encoder_sequence_length: torch.Tensor = None
+ ):
batch_size, maximum_length = alphas.size()
int_type = torch.int32
@@ -345,11 +394,15 @@
index_div = torch.floor(torch.true_divide(alphas_cumsum, index)).type(int_type)
index_div_bool_zeros = index_div.eq(0)
index_div_bool_zeros_count = torch.sum(index_div_bool_zeros, dim=-1) + 1
- index_div_bool_zeros_count = torch.clamp(index_div_bool_zeros_count, 0, encoder_sequence_length.max())
+ index_div_bool_zeros_count = torch.clamp(
+ index_div_bool_zeros_count, 0, encoder_sequence_length.max()
+ )
token_num_mask = (~make_pad_mask(token_num, maxlen=max_token_num)).to(token_num.device)
index_div_bool_zeros_count *= token_num_mask
- index_div_bool_zeros_count_tile = index_div_bool_zeros_count[:, :, None].repeat(1, 1, maximum_length)
+ index_div_bool_zeros_count_tile = index_div_bool_zeros_count[:, :, None].repeat(
+ 1, 1, maximum_length
+ )
ones = torch.ones_like(index_div_bool_zeros_count_tile)
zeros = torch.zeros_like(index_div_bool_zeros_count_tile)
ones = torch.cumsum(ones, dim=2)
@@ -360,77 +413,205 @@
index_div_bool_zeros_count_tile = 1 - index_div_bool_zeros_count_tile_bool.type(int_type)
index_div_bool_zeros_count_tile_out = torch.sum(index_div_bool_zeros_count_tile, dim=1)
index_div_bool_zeros_count_tile_out = index_div_bool_zeros_count_tile_out.type(int_type)
- predictor_mask = (~make_pad_mask(encoder_sequence_length, maxlen=encoder_sequence_length.max())).type(
- int_type).to(encoder_sequence_length.device)
+ predictor_mask = (
+ (~make_pad_mask(encoder_sequence_length, maxlen=encoder_sequence_length.max()))
+ .type(int_type)
+ .to(encoder_sequence_length.device)
+ )
index_div_bool_zeros_count_tile_out = index_div_bool_zeros_count_tile_out * predictor_mask
predictor_alignments = index_div_bool_zeros_count_tile_out
- predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype)
+ predictor_alignments_length = predictor_alignments.sum(-1).type(
+ encoder_sequence_length.dtype
+ )
return predictor_alignments.detach(), predictor_alignments_length.detach()
- def gen_tf2torch_map_dict(self):
-
- tensor_name_prefix_torch = self.tf2torch_tensor_name_prefix_torch
- tensor_name_prefix_tf = self.tf2torch_tensor_name_prefix_tf
- map_dict_local = {
- ## predictor
- "{}.cif_conv1d.weight".format(tensor_name_prefix_torch):
- {"name": "{}/conv1d/kernel".format(tensor_name_prefix_tf),
- "squeeze": None,
- "transpose": (2, 1, 0),
- }, # (256,256,3),(3,256,256)
- "{}.cif_conv1d.bias".format(tensor_name_prefix_torch):
- {"name": "{}/conv1d/bias".format(tensor_name_prefix_tf),
- "squeeze": None,
- "transpose": None,
- }, # (256,),(256,)
- "{}.cif_output.weight".format(tensor_name_prefix_torch):
- {"name": "{}/conv1d_1/kernel".format(tensor_name_prefix_tf),
- "squeeze": 0,
- "transpose": (1, 0),
- }, # (1,256),(1,256,1)
- "{}.cif_output.bias".format(tensor_name_prefix_torch):
- {"name": "{}/conv1d_1/bias".format(tensor_name_prefix_tf),
- "squeeze": None,
- "transpose": None,
- }, # (1,),(1,)
- }
- return map_dict_local
- def convert_tf2torch(self,
- var_dict_tf,
- var_dict_torch,
- ):
- map_dict = self.gen_tf2torch_map_dict()
- var_dict_torch_update = dict()
- for name in sorted(var_dict_torch.keys(), reverse=False):
- names = name.split('.')
- if names[0] == self.tf2torch_tensor_name_prefix_torch:
- name_tf = map_dict[name]["name"]
- data_tf = var_dict_tf[name_tf]
- if map_dict[name]["squeeze"] is not None:
- data_tf = np.squeeze(data_tf, axis=map_dict[name]["squeeze"])
- if map_dict[name]["transpose"] is not None:
- data_tf = np.transpose(data_tf, map_dict[name]["transpose"])
- data_tf = torch.from_numpy(data_tf).type(torch.float32).to("cpu")
- assert var_dict_torch[name].size() == data_tf.size(), "{}, {}, {} != {}".format(name, name_tf,
- var_dict_torch[
- name].size(),
- data_tf.size())
- var_dict_torch_update[name] = data_tf
- logging.info(
- "torch tensor: {}, {}, loading from tf tensor: {}, {}".format(name, data_tf.size(), name_tf,
- var_dict_tf[name_tf].shape))
-
- return var_dict_torch_update
+@tables.register("predictor_classes", "CifPredictorV2Export")
+class CifPredictorV2Export(torch.nn.Module):
+ def __init__(self, model, **kwargs):
+ super().__init__()
+
+ self.pad = model.pad
+ self.cif_conv1d = model.cif_conv1d
+ self.cif_output = model.cif_output
+ self.threshold = model.threshold
+ self.smooth_factor = model.smooth_factor
+ self.noise_threshold = model.noise_threshold
+ self.tail_threshold = model.tail_threshold
+
+ def forward(
+ self,
+ hidden: torch.Tensor,
+ mask: torch.Tensor,
+ ):
+ alphas, token_num = self.forward_cnn(hidden, mask)
+ mask = mask.transpose(-1, -2).float()
+ mask = mask.squeeze(-1)
+ hidden, alphas, token_num = self.tail_process_fn(hidden, alphas, mask=mask)
+ acoustic_embeds, cif_peak = cif_export(hidden, alphas, self.threshold)
+
+ return acoustic_embeds, token_num, alphas, cif_peak
+
+ def forward_cnn(
+ self,
+ hidden: torch.Tensor,
+ mask: torch.Tensor,
+ ):
+ h = hidden
+ context = h.transpose(1, 2)
+ queries = self.pad(context)
+ output = torch.relu(self.cif_conv1d(queries))
+ output = output.transpose(1, 2)
+
+ output = self.cif_output(output)
+ alphas = torch.sigmoid(output)
+ alphas = torch.nn.functional.relu(alphas * self.smooth_factor - self.noise_threshold)
+ mask = mask.transpose(-1, -2).float()
+ alphas = alphas * mask
+ alphas = alphas.squeeze(-1)
+ token_num = alphas.sum(-1)
+
+ return alphas, token_num
+
+ def tail_process_fn(self, hidden, alphas, token_num=None, mask=None):
+ b, t, d = hidden.size()
+ tail_threshold = self.tail_threshold
+
+ zeros_t = torch.zeros((b, 1), dtype=torch.float32, device=alphas.device)
+ ones_t = torch.ones_like(zeros_t)
+
+ mask_1 = torch.cat([mask, zeros_t], dim=1)
+ mask_2 = torch.cat([ones_t, mask], dim=1)
+ mask = mask_2 - mask_1
+ tail_threshold = mask * tail_threshold
+ alphas = torch.cat([alphas, zeros_t], dim=1)
+ alphas = torch.add(alphas, tail_threshold)
+
+ zeros = torch.zeros((b, 1, d), dtype=hidden.dtype).to(hidden.device)
+ hidden = torch.cat([hidden, zeros], dim=1)
+ token_num = alphas.sum(dim=-1)
+ token_num_floor = torch.floor(token_num)
+
+ return hidden, alphas, token_num_floor
-class mae_loss(nn.Module):
+@torch.jit.script
+def cif_v1_export(hidden, alphas, threshold: float):
+ device = hidden.device
+ dtype = hidden.dtype
+ batch_size, len_time, hidden_size = hidden.size()
+ threshold = torch.tensor([threshold], dtype=alphas.dtype).to(alphas.device)
+
+ frames = torch.zeros(batch_size, len_time, hidden_size, dtype=dtype, device=device)
+ fires = torch.zeros(batch_size, len_time, dtype=dtype, device=device)
+
+ # prefix_sum = torch.cumsum(alphas, dim=1)
+ prefix_sum = torch.cumsum(alphas, dim=1, dtype=torch.float64).to(
+ torch.float32
+ ) # cumsum precision degradation cause wrong result in extreme
+ prefix_sum_floor = torch.floor(prefix_sum)
+ dislocation_prefix_sum = torch.roll(prefix_sum, 1, dims=1)
+ dislocation_prefix_sum_floor = torch.floor(dislocation_prefix_sum)
+
+ dislocation_prefix_sum_floor[:, 0] = 0
+ dislocation_diff = prefix_sum_floor - dislocation_prefix_sum_floor
+
+ fire_idxs = dislocation_diff > 0
+ fires[fire_idxs] = 1
+ fires = fires + prefix_sum - prefix_sum_floor
+
+ # prefix_sum_hidden = torch.cumsum(alphas.unsqueeze(-1).tile((1, 1, hidden_size)) * hidden, dim=1)
+ prefix_sum_hidden = torch.cumsum(alphas.unsqueeze(-1).tile((1, 1, hidden_size)) * hidden, dim=1)
+ frames = prefix_sum_hidden[fire_idxs]
+ shift_frames = torch.roll(frames, 1, dims=0)
+
+ batch_len = fire_idxs.sum(1)
+ batch_idxs = torch.cumsum(batch_len, dim=0)
+ shift_batch_idxs = torch.roll(batch_idxs, 1, dims=0)
+ shift_batch_idxs[0] = 0
+ shift_frames[shift_batch_idxs] = 0
+
+ remains = fires - torch.floor(fires)
+ # remain_frames = remains[fire_idxs].unsqueeze(-1).tile((1, hidden_size)) * hidden[fire_idxs]
+ remain_frames = remains[fire_idxs].unsqueeze(-1).tile((1, hidden_size)) * hidden[fire_idxs]
+
+ shift_remain_frames = torch.roll(remain_frames, 1, dims=0)
+ shift_remain_frames[shift_batch_idxs] = 0
+
+ frames = frames - shift_frames + shift_remain_frames - remain_frames
+
+ # max_label_len = batch_len.max()
+ max_label_len = alphas.sum(dim=-1)
+ max_label_len = torch.floor(max_label_len).max().to(dtype=torch.int64)
+
+ # frame_fires = torch.zeros(batch_size, max_label_len, hidden_size, dtype=dtype, device=device)
+ frame_fires = torch.zeros(batch_size, max_label_len, hidden_size, dtype=dtype, device=device)
+ indices = torch.arange(max_label_len, device=device).expand(batch_size, -1)
+ frame_fires_idxs = indices < batch_len.unsqueeze(1)
+ frame_fires[frame_fires_idxs] = frames
+ return frame_fires, fires
+
+
+@torch.jit.script
+def cif_export(hidden, alphas, threshold: float):
+ batch_size, len_time, hidden_size = hidden.size()
+ threshold = torch.tensor([threshold], dtype=alphas.dtype).to(alphas.device)
+
+ # loop varss
+ integrate = torch.zeros([batch_size], dtype=alphas.dtype, device=hidden.device)
+ frame = torch.zeros([batch_size, hidden_size], dtype=hidden.dtype, device=hidden.device)
+ # intermediate vars along time
+ list_fires = []
+ list_frames = []
+
+ for t in range(len_time):
+ alpha = alphas[:, t]
+ distribution_completion = (
+ torch.ones([batch_size], dtype=alphas.dtype, device=hidden.device) - integrate
+ )
+
+ integrate += alpha
+ list_fires.append(integrate)
+
+ fire_place = integrate >= threshold
+ integrate = torch.where(
+ fire_place,
+ integrate - torch.ones([batch_size], dtype=alphas.dtype, device=hidden.device),
+ integrate,
+ )
+ cur = torch.where(fire_place, distribution_completion, alpha)
+ remainds = alpha - cur
+
+ frame += cur[:, None] * hidden[:, t, :]
+ list_frames.append(frame)
+ frame = torch.where(
+ fire_place[:, None].repeat(1, hidden_size), remainds[:, None] * hidden[:, t, :], frame
+ )
+
+ fires = torch.stack(list_fires, 1)
+ frames = torch.stack(list_frames, 1)
+
+ fire_idxs = fires >= threshold
+ frame_fires = torch.zeros_like(hidden)
+ max_label_len = frames[0, fire_idxs[0]].size(0)
+ for b in range(batch_size):
+ frame_fire = frames[b, fire_idxs[b]]
+ frame_len = frame_fire.size(0)
+ frame_fires[b, :frame_len, :] = frame_fire
+
+ if frame_len >= max_label_len:
+ max_label_len = frame_len
+ frame_fires = frame_fires[:, :max_label_len, :]
+ return frame_fires, fires
+
+
+class mae_loss(torch.nn.Module):
def __init__(self, normalize_length=False):
super(mae_loss, self).__init__()
self.normalize_length = normalize_length
- self.criterion = torch.nn.L1Loss(reduction='sum')
+ self.criterion = torch.nn.L1Loss(reduction="sum")
def forward(self, token_length, pre_token_length):
loss_token_normalizer = token_length.size(0)
@@ -459,19 +640,17 @@
list_fires.append(integrate)
fire_place = integrate >= threshold
- integrate = torch.where(fire_place,
- integrate - torch.ones([batch_size], device=hidden.device),
- integrate)
- cur = torch.where(fire_place,
- distribution_completion,
- alpha)
+ integrate = torch.where(
+ fire_place, integrate - torch.ones([batch_size], device=hidden.device), integrate
+ )
+ cur = torch.where(fire_place, distribution_completion, alpha)
remainds = alpha - cur
frame += cur[:, None] * hidden[:, t, :]
list_frames.append(frame)
- frame = torch.where(fire_place[:, None].repeat(1, hidden_size),
- remainds[:, None] * hidden[:, t, :],
- frame)
+ frame = torch.where(
+ fire_place[:, None].repeat(1, hidden_size), remainds[:, None] * hidden[:, t, :], frame
+ )
fires = torch.stack(list_fires, 1)
frames = torch.stack(list_frames, 1)
@@ -484,6 +663,76 @@
pad_l = torch.zeros([max_label_len - l.size(0), hidden_size], device=hidden.device)
list_ls.append(torch.cat([l, pad_l], 0))
return torch.stack(list_ls, 0), fires
+
+
+def cif_wo_hidden_v1(alphas, threshold, return_fire_idxs=False):
+ batch_size, len_time = alphas.size()
+ device = alphas.device
+ dtype = alphas.dtype
+
+ threshold = torch.tensor([threshold], dtype=alphas.dtype).to(alphas.device)
+
+ fires = torch.zeros(batch_size, len_time, dtype=dtype, device=device)
+
+ # prefix_sum = torch.cumsum(alphas, dim=1)
+ prefix_sum = torch.cumsum(alphas, dim=1, dtype=torch.float64).to(
+ torch.float32
+ ) # cumsum precision degradation cause wrong result in extreme
+ prefix_sum_floor = torch.floor(prefix_sum)
+ dislocation_prefix_sum = torch.roll(prefix_sum, 1, dims=1)
+ dislocation_prefix_sum_floor = torch.floor(dislocation_prefix_sum)
+
+ dislocation_prefix_sum_floor[:, 0] = 0
+ dislocation_diff = prefix_sum_floor - dislocation_prefix_sum_floor
+
+ fire_idxs = dislocation_diff > 0
+ fires[fire_idxs] = 1
+ fires = fires + prefix_sum - prefix_sum_floor
+ if return_fire_idxs:
+ return fires, fire_idxs
+ return fires
+
+
+def cif_v1(hidden, alphas, threshold):
+ fires, fire_idxs = cif_wo_hidden_v1(alphas, threshold, return_fire_idxs=True)
+
+ device = hidden.device
+ dtype = hidden.dtype
+ batch_size, len_time, hidden_size = hidden.size()
+ # frames = torch.zeros(batch_size, len_time, hidden_size, dtype=dtype, device=device)
+ # prefix_sum_hidden = torch.cumsum(alphas.unsqueeze(-1).tile((1, 1, hidden_size)) * hidden, dim=1)
+ frames = torch.zeros(batch_size, len_time, hidden_size, dtype=dtype, device=device)
+ prefix_sum_hidden = torch.cumsum(alphas.unsqueeze(-1).tile((1, 1, hidden_size)) * hidden, dim=1)
+
+ frames = prefix_sum_hidden[fire_idxs]
+ shift_frames = torch.roll(frames, 1, dims=0)
+
+ batch_len = fire_idxs.sum(1)
+ batch_idxs = torch.cumsum(batch_len, dim=0)
+ shift_batch_idxs = torch.roll(batch_idxs, 1, dims=0)
+ shift_batch_idxs[0] = 0
+ shift_frames[shift_batch_idxs] = 0
+
+ remains = fires - torch.floor(fires)
+ # remain_frames = remains[fire_idxs].unsqueeze(-1).tile((1, hidden_size)) * hidden[fire_idxs]
+ remain_frames = remains[fire_idxs].unsqueeze(-1).tile((1, hidden_size)) * hidden[fire_idxs]
+
+ shift_remain_frames = torch.roll(remain_frames, 1, dims=0)
+ shift_remain_frames[shift_batch_idxs] = 0
+
+ frames = frames - shift_frames + shift_remain_frames - remain_frames
+
+ # max_label_len = batch_len.max()
+ max_label_len = (
+ torch.round(alphas.sum(-1)).int().max()
+ ) # torch.round to calculate the max length
+
+ # frame_fires = torch.zeros(batch_size, max_label_len, hidden_size, dtype=dtype, device=device)
+ frame_fires = torch.zeros(batch_size, max_label_len, hidden_size, dtype=dtype, device=device)
+ indices = torch.arange(max_label_len, device=device).expand(batch_size, -1)
+ frame_fires_idxs = indices < batch_len.unsqueeze(1)
+ frame_fires[frame_fires_idxs] = frames
+ return frame_fires, fires
def cif_wo_hidden(alphas, threshold):
@@ -501,10 +750,11 @@
list_fires.append(integrate)
fire_place = integrate >= threshold
- integrate = torch.where(fire_place,
- integrate - torch.ones([batch_size], device=alphas.device)*threshold,
- integrate)
+ integrate = torch.where(
+ fire_place,
+ integrate - torch.ones([batch_size], device=alphas.device) * threshold,
+ integrate,
+ )
fires = torch.stack(list_fires, 1)
return fires
-
--
Gitblit v1.9.1